Archives of Pediatrics and Adolescent Medicine

2007 Archives of Pediatrics and Adolescent Medicine

doi: 10.1001/archpedi.161.7.629

The Relationship Between Self-injurious Behavior and Suicide in a Young Adult Population Self-injurious behavior starts during adolescence and may be an attempt to cope with psychological distress. In this study of 2875 students from 2 universities, 10.1% reported self-injurious behavior, three-fourths of whom reported this behavior occurring more than once. Those with such behavior were nearly 4-fold more likely to also report suicidality and nearly 10-fold more likely to have attempted suicide. Individuals with both self-injurious behaviors and suicidality were more likely to have been abused and to have an eating disorder than students without such behaviors. This study suggests that while self-injurious behavior may be a coping mechanism employed to avoid suicide, it may also serve as a harbinger of all forms of suicidality in some individuals and its presence should trigger assessment for suicide risk. See Article Child Care and the Well-being of Children Twelve million children spend a significant amount of their time each year in some type of child care, which has become a fact of life for many families in our society. Nevertheless, parents often express concerns about the potential harmful effects of care on their children. This review article summarizes what is known about the age of entry into child care, amount and type of care, and its quality and discusses how each of these affects children's social, physical, and cognitive development. The review provides pediatricians with important information on how to guide families with their child care decisions and how to care for children at home. See Article Pay for Performance Alone Cannot Drive Quality Despite the rapid growth of pay-for-performance programs in the United States, evidence of their impact on quality of care is limited. This study of 44 pediatric practices sought to determine whether coupling pay for performance with other interventions would accelerate improvement in the care of 13 380 children with asthma in those practices. The program rewarded practices with up to a 7% increase in their fee schedule, practice-level performance thresholds, and building improvement capacity for participating in the collaborative, achieving network. During the 3-year study period, the percentage of the network population of children with asthma who received optimal care increased from 4% to 88% and the proportion of children receiving influenza vaccine nearly tripled. This study demonstrated that pay for performance, when coupled with robust approaches for quality improvement, can accelerate transformation among pediatric health care providers. See Article Effect of Regulatory Warnings on Antidepressant Prescribing for Children and Adolescents Agencies in both the United Kingdom (UK) and the United States have issued warnings on the increased risk of suicide associated with the use of antidepressants in children and adolescents. Little is known about the effects of these warnings and accompanying publicity on the prescribing of antidepressants for pediatric patients. This study used data of more than 400 000 Tennessee Medicaid patients per month to examine trends in antidepressant prescribing to patients aged 2 to 17 years after the appearance of agency warnings compared with the 24 months before. There was a 33% decrease in the number of new users of antidepressants in the 21 months following the UK warning, most pronounced for nonfluoxetine antidepressants in which initiation of use decreased by 54%. In contrast, new users of fluoxetine increased by 60%. The public health effect of these changes in prescribing patterns is unknown. View LargeDownload See Article

Cuttler, Leona;Kenney, Genevieve M.

2007 Archives of Pediatrics & Adolescent Medicine

doi: 10.1001/archpedi.161.7.630pmid: 17606824

One in 4 US children has health insurance through public, government-sponsored programs—Medicaid or the State Children's Health Insurance Program (SCHIP). These public programs are important not only because of the vast number of children they cover but also because children who lack health insurance have worse access to care than those with either public or private health insurance. Public programs also disproportionately serve children with special health care needs. Furthermore, children's health care facilities depend heavily on Medicaid and SCHIP patients and the accompanying reimbursements to maintain programs and services, including programs that also benefit privately insured children. Therefore, Medicaid and SCHIP policies have a tremendous impact on health care delivery to US children, shaping the scope and quality of health care as well as the nature of pediatric practice. This article reviews key aspects of SCHIP, a program whose future is at a crossroads, focusing on issues that are important to pediatricians and others who deliver care to children. What is the goal of schip? The goal of SCHIP is to provide health insurance coverage for uninsured, lower-income children whose family incomes are too high to qualify for Medicaid.1,2 To limit federal outlays and allow state flexibility in SCHIP, the federal government provides states with capped grants that offset the bulk of costs, while states design their SCHIP programs within broad federal rules. Created in 1997, SCHIP was authorized for 10 years and allotted $40 billion in federal funds for 1998 to 2007. If it is to continue, SCHIP must be reauthorized by Congress in 2007. How many children receive health insurance from schip? During 2005, SCHIP provided health insurance to 6 million children over the course of a year, many of whom would otherwise be uninsured.3,4 Around that same period, Medicaid covered 28 million children. Together, these public programs provide insurance coverage to one-quarter of US children. How is schip structured among states? Each state has the option of using SCHIP funds to allow eligible children into the state Medicaid program (ie, a Medicaid expansion), to create a separate SCHIP insurance program, or to create a combined approach (Table). States choosing the Medicaid expansion model must provide full Medicaid benefits to all enrolled children and cannot cap enrollment if allocated SCHIP funds are expended. By contrast, states choosing separate SCHIP programs can create a different set of benefits and may limit enrollment based on availability of funds. Regardless of the type of SCHIP program selected, Medicaid-eligible children must be enrolled in the state Medicaid program. Table. View LargeDownload State-by-State Data for State Children's Health Insurance Program Structure and Percentage of Uninsured Childrena There is great variability among states in eligibility criteria, benefits, premiums, and co-pays under SCHIP. The SCHIP income eligibility thresholds range from 140% of the federal poverty level (FPL) in North Dakota to 350% of the FPL in New Jersey. Overall, 41 states established SCHIP eligibility at or above the congressional target of 200% of the FPL (ie, $33 200 for a family of 3).5 Some states have used waivers to expand SCHIP beyond low-income children to include pregnant women, parents of SCHIP-insured children, and/or childless adults. These expansions are controversial; opponents believe that SCHIP funds should be used to provide insurance coverage for more children, whereas proponents argue that expanding family coverage benefits children as well. What does schip cover? There is variability among states' SCHIP benefits. The SCHIP programs that are Medicaid expansions cover the same services as Medicaid. Separate SCHIP programs have some mandatory guidelines (including provision of preventive well-child care without family cost sharing) but generally have fewer benefits and more family cost sharing than those using Medicaid expansions.3 How is schip financed? The State Children's Health Insurance Program is jointly financed by the federal and state governments through a matching funds program, a system analogous to Medicaid. The proportion paid by the federal government differs among states, with federal contributions provided as finite block grants. Each state's annual allotment is based on a formula that considers the state's share of low-income and uninsured children.6 States have 3 years to spend the allotment, after which unexpended funds can be redistributed by the federal government. Both the formula and the redistribution process are controversial. To encourage state participation, there is greater federal matching for SCHIP spending than for Medicaid spending. How is schip different from medicaid? Although Medicaid and SCHIP are both public programs that provide health insurance to low-income children and are both financed by a combination of federal and state funds, there are major differences in their structure and scope. Different family incomes: By design, children eligible for SCHIP have incomes that exceed Medicaid eligibility for the same age group in each state. Coverage under Medicaid is mandated for children aged 6 years and younger with family income up to 133% of the FPL and for children aged 6 to 18 years with family income up to 100% of the FPL. However, income levels for SCHIP are at state option. Entitlement program vs block grant: Medicaid is an open-ended entitlement program, meaning that every child who meets eligibility criteria can enroll in the program.7 Federal funds for Medicaid are guaranteed with no preset limits. By contrast, federal funds for SCHIP are provided as capped block grants to states. Therefore, SCHIP does not guarantee eligibility for individual children; states with separate SCHIP programs may cap enrollment if SCHIP funds are depleted. If SCHIP is reauthorized with fewer funds than needed to cover current enrollment, states using the Medicaid model could continue to enroll eligible children in Medicaid, drawing on Medicaid matching funds to cover those children; however, states with separate SCHIP models would have to rely exclusively on state funds, which might lead them to limit enrollment. Different levels of federal matching funds: Although determined on a state-by-state basis, the federal government's share of SCHIP spending is enhanced compared with Medicaid (on average, 70% vs 57%, respectively). Scope of coverage: Medicaid law requires a broad range of benefits, including early and periodic screening, diagnosis, and treatment, which provide children with screening and treatment services in a relatively uniform manner across states (although some exceptions to this uniformity may be possible under a waiver process and the Deficit Reduction Act of 2005). State SCHIP programs using Medicaid expansion models must provide SCHIP enrollees with these same benefits. However, states with separate SCHIP programs, although subject to broad guidelines, do not require the full benefits inherent in Medicaid. Benefits vary among these states, but generally, separate SCHIP programs have fewer benefits and additional family cost sharing compared with Medicaid. Program size: The State Children's Health Insurance Program plays a key role in providing insurance coverage to children who would otherwise be uninsured. However, SCHIP covers many fewer children than Medicaid (6 million children vs 28 million children, respectively, in 2005).3 Together, federal and state spending on children under SCHIP and Medicaid programs were $7 billion and $52 billion, respectively, in 2005.3 Whereas the majority of SCHIP spending is for children, only 16% of Medicaid spending is for children (even though 48% of Medicaid recipients are children).7 This is because Medicaid also supports health care coverage for elderly and other sick and low-income adults, and their per capita health expenses are much higher than those of covered children. When total spending is considered (including for both children and adults), SCHIP spending is about 2% of Medicaid spending.7 To what extent do pediatricians participate in schip? Adequate physician participation is key to ensuring that enrolled children have access to services. A national survey conducted by the American Academy of Pediatrics in 2000 found that participation by pediatricians in SCHIP and Medicaid was high overall (89%).8,9 However, only two-thirds of physicians accepted all SCHIP and Medicaid patients, with office-based primary care pediatricians less likely to accept patients than those in safety net settings. Physician participation in SCHIP and Medicaid varied markedly among states, and states with the lower quartile of Medicaid payments had substantially lower physician participation rates. Similar data relating physician participation in Medicaid to reimbursement have been documented by others.10-15 Less is known about physician reimbursement under SCHIP alone and about whether variation in reimbursement rates affect participation, particularly because there is heavy reliance on managed care under SCHIP. However, SCHIP fees for states that use Medicaid expansion models are the same as Medicaid fees and are thus substantially lower than Medicare fees for equivalent services. In addition to low payment, administrative burden and capitated managed care may also interfere with physicians' willingness to accept publicly covered patients.8,9 Other questions are the location of physicians who accept SCHIP and Medicaid as well as their proximity to low-income children. These issues raise questions about potential access of SCHIP- and Medicaid-covered children to needed services. How effective is schip? The State Children's Health Insurance Program is widely viewed as a success.1 Together with Medicaid, SCHIP has helped to reduce the number of uninsured low-income children by about one-third, falling from 23% in 1997 to 15% in 2004.3 To date, there is little conclusive evidence of “crowd out” (ie, few families dropped private coverage to enroll children in SCHIP). The State Children's Health Insurance Program has improved the likelihood that a child has a medical home.1,16 Aggregate data are difficult to assess owing to differences among state SCHIP programs, but individual programs have demonstrated improvements in access,1,17,18 quality of care,18 and reduction in ethnic disparities.18 A national study indicated improvements in several access measures following expansions under SCHIP.19 What challenges facing schip can affect child health care and pediatrics? SCHIP Reauthorization There is broad support for reauthorizing SCHIP, which is generally viewed as a successful program. However, the content and scope of a reauthorized SCHIP program are controversial—particularly regarding overall funding, eligibility criteria, and formulas for fund allocations and reallocations among states. Funding Many states already face shortfalls in SCHIP funds. If SCHIP is reauthorized at its current funding level, this would not be adequate to maintain current enrollment levels; more than a million children could lose coverage. In the coming years, it is estimated that more than $12 billion in additional federal funds over the next 5 years (and an additional $30 billion over 10 years) is needed to maintain current SCHIP enrollment levels.20,21 Eligibility There is controversy about whether SCHIP funds should be used only to cover low-income children or whether coverage should be expanded to parents of eligible children, higher-income children, pregnant women, and/or low-income childless adults. Formula for Allocation and Reallocation of Federal SCHIP Funds to States The formula for distribution of federal SCHIP funds across the states is controversial because of concerns about the quality of the data used to derive the formula, the fact that state allotments decrease as the number of low-income uninsured persons declines, and current rules that allow states to retain their allotments for 3 years regardless of current needs. Perhaps even more controversial are methods for reallocating unexpended SCHIP funds from states with surpluses to others facing shortfalls. The complex financing of SCHIP has led to uncertainties among states in predicting available SCHIP funds. Whereas states can all benefit from SCHIP and can pull together to maintain and enhance the program, the formula issue potentially divides states. Medicaid The success of SCHIP rests in part on its function of building on Medicaid by covering children who are ineligible for Medicaid. In this sense, many argue that continued SCHIP success depends on maintaining or strengthening Medicaid. However, a reauthorized SCHIP program will require additional funds to maintain current coverage, and it might be suggested that these funds come from cuts in Medicaid. Indeed, the success of SCHIP has raised questions about whether Medicaid should also have SCHIP-like features such as caps on funding, greater family cost sharing, and/or fewer mandatory benefits. Such measures might reduce costs while providing coverage. However, children with poor health status are more likely to be covered by Medicaid than SCHIP, and SCHIP's capacity to improve their health status may not match the more extensive Medicaid benefits. Also, cost sharing results in reduced use of essential health services, particularly for low-income families. Further, SCHIP has shown that “caps on federal Medicaid spending could leave more children without coverage and that the imprecision of any funding formulas would lead to poorly targeted distribution of funds for states.”7 Quality of Health Care Delivery Most states have reported on at least 1 of 4 quality measures for their SCHIP programs.22 However, inconsistencies in measurement and reporting remain, and these 4 measures are very limited in scope. Further, there is no required reporting for measures of inpatient care. In other areas of health (eg, Medicare), there is a growing impetus to report health care indices and assess the quality of care. Child health programs have been late to enter the quality-reporting and analysis arena in part because reliable and valid measures for children were lacking. However, progress has been made in the development of such measures. The reauthorization of SCHIP provides an opportunity to incorporate well-designed quality assessment in a model child health program. Millions of US Children Remain Uninsured Currently, according to the US census, 9 million US children do not have health insurance. Close to three-quarters of these children are eligible for Medicaid or SCHIP but are not enrolled.23 Thus, reducing the uninsured problem among US children hinges on enrolling more eligible children in both Medicaid and SCHIP. Expanding public coverage will require adequate federal funding in both programs and policy changes that address enrollment barriers.24-26 Back to top Article Information Correspondence: Dr Leona Cuttler, The Rainbow Center for Child Health Policy, Rainbow Babies and Children's Hospital, Case Western Reserve University, 11100 Euclid Ave, Room 737, Cleveland, OH 44106 ([email protected]). Author Contributions:Study concept and design: Cuttler and Kenney. Acquisition of data: Cuttler. Drafting of the manuscript: Cuttler and Kenney. Critical revision of the manuscript for important intellectual content: Cuttler and Kenney. Study supervision: Cuttler. Financial Disclosure: None reported. Funding/Support: This work was supported in part by an award from the Robert Wood Johnson Foundation (Dr Cuttler) and the Rainbow Babies and Children's Hospital Board of Trustees (Dr Cuttler). Additional Contributions: We thank Justin Yee, BA, of the Urban Institute for providing helpful research assistance. References 1. Kenney GYee J SCHIP at a crossroads: experience to date and challenges ahead. Health Aff (Millwood) 2007;26 (2) 356- 369PubMedGoogle ScholarCrossref 2. Kenney GChang DI The State Children's Health Insurance Program: successes, shortcomings, and challenges. Health Aff 2004;23 (5) 51- 62PubMedGoogle ScholarCrossref 3. Kaiser Family Foundation, A decade of SCHIP experience and issues for reauthorization. http://www.kff.org/medicaid/upload/7574-2.pdfAccessed January 20, 2007 4. Kenney GCook A Coverage Patterns Among SCHIP-Eligible Children and Their Parents: Health Policy Online Brief 15. washington, dc Urban Institute2007; 5. Ross DCCox LMarks CKaiser Family Foundation, Resuming the path to health coverage for children and parents: a 50 state update on eligibility rules, enrollment and renewal procedures, and cost-sharing practices in Medicaid and SCHIP in 2006. http://www.kff.org/medicaid/upload/7608.pdfAccessed January 20, 2007 6. Herz EJFernandez BPeterson C State Children's Health Insurance Program (SCHIP): a brief overview. http://www.law.umaryland.edu/marshall/crsreports/crsdocuments/RL3047303232005.pdfAccessed July 20, 2006 7. Kaiser Family Foundation, Health coverage for low-income population: a comparison of Medicaid and SCHIP. http://www.kff.org/medicaid/upload/7488.pdfAccessed January 14, 2007 8. Data raise concerns about Medicaid access. AAP News 2001;18 (4) 143Google Scholar 9. Bucciarelli RL The effect of Medicaid participation by private and safety net pediatricians on incremental expansion of coverage for children. Pediatrics 2003;112 (2) 416PubMedGoogle ScholarCrossref 10. Berman SDolins JTang S-FYudkowsky BK Factors that influence the willingness of private primary care pediatricians to accept more Medicaid patients. Pediatrics 2002;110 (2, pt 1) 239- 248PubMedGoogle ScholarCrossref 11. Zuckerman SMcFeeters JCunningham PNichols L Changes in Medicaid physician fees, 1998-2003: implications for physician participation. Health Aff (Millwood) 2004;Suppl Web ExclusivesW4-374- W4-384PubMeddoi: 10.1377/hlthaff.w4.374Google Scholar 12. Skaggs DLClemens SMVitale MGFemino JDKay RM Access to orthopedic care for children with Medicaid vs private insurance in California. Pediatrics 2001;107 (6) 1405- 1408PubMedGoogle ScholarCrossref 13. Tang S-FAmerican Academy of Pediatrics, Medicaid reimbursement survey, 2001: 50 states and the District of Columbia. http://www.aap.org/research/medreimPDF01/all_states.PDFAccessed April 9, 2007 14. Wang ECChoe MCMeara JGKoempel JA Inequality of access to surgical specialty health care: why children with government-funded insurance have less access than those with private insurance in southern California. Pediatrics 2004;114 (5) e584- e590PubMedGoogle ScholarCrossref 15. Cunningham PMay JH Medicaid Patients Increasingly Concentrated Among Physicians: Tracking Report No. 16. Washington, DC Center for Studying Health System Change2006; 16. Quinn ARosenbach M Beyond coverage: SCHIP makes strides toward providing a usual source of care to low-income children. http://www.mathematica-mpr.com/publications/PDFs/schipstrides.pdfAccessed January 14, 2007 17. Damiano PCWillard JCMomany ETChowdhury J The impact of the Iowa S-SCHIP program on access, health status, and the family environment. Ambul Pediatr 2003;3 (5) 263- 269PubMedGoogle ScholarCrossref 18. Shone LPDick AWKlein JDZwanziger JSzilagyi PG Reduction in racial and ethnic disparities after enrollment in the State Children's Health Insurance Program. Pediatrics 2005;115 (6) e697- e705PubMedGoogle ScholarCrossref 19. Davidoff AKenney GDubay L Effects of the State Children's Health Insurance Program expansions on children with chronic health conditions. Pediatrics 2005;116 (1) e34- e42PubMedGoogle ScholarCrossref 20. Broaddus MPark E Freezing SCHIP Funding in Coming Years Would Reverse Recent Gains in Children's Health Coverage. Washington, DC Center on Budget and Policy Priorities2006; 21. Peterson CL SCHIP Original Allotments: Description and Analysis. Washington, DC Congressional Research Service2006; 22. Day SKatz ARosenbach M Improving performance measurement in the State Children's Health Insurance Program. http://www.mathematica-mpr.com/publications/PDFs/performmeasure.pdfAccessed January 14, 2007 23. Dubay LHolahan JCook A The uninsured and the affordability of health insurance coverage. Health Aff (Millwood) 2007;26 (1) w22- w30PubMedGoogle ScholarCrossref 24. Dorn SKenney G Automatically Enrolling Eligible Children and Families into Medicaid and SCHIP: Opportunities, Obstacles, and Options for Federal Policymakers. New York, NY Commonwealth Fund2006; 25. Summer LMann C Instability of Public Health Insurance Coverage for Children and Their Families: Causes, Consequences, and Remedies. New York, NY Commonwealth Fund2006; 26. Haley JKenney G Low-income uninsured children with special health care needs: why aren't they enrolled in public health insurance programs? Pediatrics 2007;119 (1) 60- 68PubMedGoogle ScholarCrossref

Whitlock, Janis;Knox, Kerry L.

2007 Archives of Pediatrics & Adolescent Medicine

doi: 10.1001/archpedi.161.7.634pmid: 17606825

Abstract Objective To test the hypothesis that self-injurious behavior (SIB) signals an attempt to cope with psychological distress that may co-occur or lead to suicidal behaviors in individuals experiencing more duress than they can effectively mitigate. Design Analysis of a cross-sectional data set of college-age students. Setting Two universities in the northeastern United States in the spring of 2005. Participants A random sample of 8300 students was invited to participate in a Web-based survey; 3069 (37.0%) responded. Cases in which a majority of the responses were missing or in which SIB or suicide status was indeterminable were omitted, resulting in 2875 usable cases. Exposure Self-injurious behavior. Main Outcome Measures Main outcome was suicidality; adjusted odds ratios (AORs) for suicidality by SIB status when demographic characteristics, history of trauma, distress, informal help-seeking, and attraction to life are considered. Results One quarter of the sample reported SIB, suicidality, or both; 40.3% of those reporting SIB also report suicidality. Self-injurious behavior status was predictive of suicidality when controlling for demographic variables (AOR, 6.2; 95% confidence interval [CI], 4.9-7.8). Addition of trauma and distress variables attenuated this relationship (AOR, 3.7; 95% CI, 2.7-4.9). Compared with respondents reporting only suicidality, those also reporting SIB were more likely to report suicide ideation (AOR, 2.8; 95% CI, 2.0-3.8), plan (AOR, 5.6; 95% CI, 3.9-7.9), gesture (AOR, 7.3; 95% CI, 3.4-15.8), and attempt (AOR, 9.6; 95% CI, 5.4-17.1). Lifetime SIB frequency exhibits a curvilinear relationship to suicidality. Conclusions Since it is well established that SIB is not a suicidal gesture, many clinicians assume that suicide assessment is unnecessary. Our findings suggest that the presence of SIB should trigger suicide assessment. Self-injurious behavior (SIB) is defined1 as self-inflicted destruction of the body for purposes not socially sanctioned and without suicidal intent. Typically associated with clinical populations, there are few epidemiological studies of SIB in community populations. Extant studies are limited by small or potentially biased samples. Available evidence suggests that approximately 4% of the general adult population and 21% of clinical populations report at least occasional SIB. Estimates of SIB prevalence in college and high school students range from 12% to 38%.2-5 A recent representative study of college students, using the same data on which these analyses are based, showed a 17% lifetime prevalence.6 Several researchers have postulated that SIB is a mechanism used to compensate for inadequate affect regulation in situations perceived as stressful.7,8 Although primarily derived from clinical populations, the affect-regulation theory helps to explain SIB in community populations as well, since many report it as a method of coping with unwanted negative emotion.9,10 If so, individuals vulnerable to SIB may also be at heightened risk of suicidality when trauma or psychological distress overwhelms their capacity to cope effectively. Most clinical and community studies show an average age of onset in mid to late adolescence followed by a decline in early adulthood.1,4,11 In high school and college students, between 34% and 45% of individuals with SIB indicate that they also experience suicidal ideation.6,12 While there is consistent evidence that SIB and suicide co-occur,13-17 the nature of this relationship is less clear. Self-injurious behavior and suicide appear to share several important correlates, including depression, alcohol or substance abuse, psychological pain, cognitive constriction, and dysregulation of the serotonin and noradrenergic systems.18-27 However, SIB and attempted and completed suicide are widely recognized to exhibit key differences in motivation, lethality, hopelessness, intent to die, and attraction to life even when an individual displays both forms of behavior.6,13,14 Two distinct models dominate conceptualization of the relationship between SIB and suicidal behaviors. One model views SIB as part of a constellation of suicidal behaviors.28 The other model views individuals who deliberately injure themselves and those who are suicidal as 2 different populations.1,13,29,30 Typically, the latter model is used to argue that SIB is most commonly used as a way to regulate negative affect and to avoid suicide. We posit an alternative to both of these models. We hypothesize that while individual SIB acts are rarely, if ever, undertaken with suicidal intent, SIB signals an attempt to cope with psychological distress that may co-occur or lead to suicidal behaviors in individuals experiencing more duress than they can effectively mitigate. If so, suicidal behaviors would be likely to either co-exist or evolve over time if SIB begins to fail as a functional coping mechanism. Consistent with this, we expected that (1) SIB status would predict suicidality independent of demographic characteristics associated with either, (2) SIB respondents who were also suicidal would exhibit higher levels of conditions known to be associated with distress and fewer protective factors than SIB only or suicidality-only individuals, (3) SIB frequency would bear a positive linear relationship to suicidality, and (4) SIB status would significantly predict all forms of suicidal behaviors rather than solely ideation. These hypotheses were tested using combined data from 2 college student populations. Methods Sample Participants were drawn from a random sample of 8300 undergraduate and graduate students (33.7% of the total combined population) from 2 northeastern universities. All were sent a postcard inviting them to participate in a Web-based survey in the spring of 2005. Soon after, each received a personalized e-mail with a link to the survey. A total of 3069 (37.0%) students completed the survey. Cases in which a majority of the responses were missing or in which SIB or suicide status was indeterminable were omitted (n = 194), resulting in 2875 (34.6%) cases retained for analysis. Sample demographics were largely representative of the overall student population, although there were significantly more women in the sample population than in the population from which they were drawn (56.3% vs 47.6%). Of these, 490 (17.0%) had practiced SIB and 423 (14.7%) reported suicidality (715 unduplicated responses). Study design and questionnaire The survey was administered on a secure Internet server, requiring 10 to 25 minutes to complete. The Web-based survey allowed for complex skip patterns viewable only by those for whom the questions were relevant. The survey also allowed students to immediately make the screen go blank if they were interrupted or feared being observed. Links to local resources were placed on the bottom of every page and a “distraction” toggle allowed anyone who needed a break to see an unrelated Web page. The study was approved by the Committee for Human Subjects at both institutions. All students provided online assent before taking the survey and were free to discontinue participation at any time by closing their Web browser. The survey consisted of 4 broad conceptual domains: (1) sociodemographic characteristics, (2) mental health indicators, (3) risk and protective factors, and (4) help-seeking history and preferences. There was a mix of epidemiological and psychological survey items. Multiple existing scales were reviewed and, where possible, the survey contained validated items. The survey was field tested with 25 students, 13 of whom were known to be self-injurious. Measures for which rates are well documented, such as lifetime prevalence of suicidality, were consistent with other available data.31,32 Examination of discriminate and convergent validity in between variable analyses within the survey also showed predictable relationship patterns. For example, the Attraction to Life Scale was inversely correlated with the K-6 scale (r = −0.64) and positively associated with life satisfaction (r = 0.67) at P<.001. Although too numerous to report here (validity and reliability of this tool is the subject of a future article), there were no unexpected correlations in any of the discriminate and convergent validity tests. Assessment of sib All respondents received an initial screening question for SIB: “Have you ever done any of the following with the intention of hurting yourself?” This was followed by a list of 16 SIB behaviors identified through examination of existing SIB surveys,33 a review of existing literature, and ongoing interviews with mental health providers and self-injurers. A later question asked respondents who indicated having practiced SIB whether they had done so “to practice suicide” or “to commit suicide.” Fourteen observations were omitted from the SIB category for purposes of analyses, since, by definition, SIB is an act undertaken without suicidal intent. Assessment of suicidality Lifetime suicidality was measured using a binary response item34 that asked, “Have you ever seriously considered suicide or attempted suicide?” Respondents who answered affirmatively were asked to select any of 8 statements that applied to them. For purposes of these analyses, these statements were clustered into the following 4 categories: (1) ideation (“I thought seriously about it”), (2) plan (“I had a general plan but did not carry it out ”; “I had a method but did not carry it out”), (3) gesture (“I wrote a suicide note but did not leave it where it could be found”; “I wrote a suicide note and did leave it where it could be found”), and (4) attempt (“I made a serious attempt but no medical intervention occurred”; “I made a serious attempt that received medical attention”). Respondents with multiple responses were placed into only 1 of these categories based on the most serious of their response selections, since understanding lethality may be a critical discriminating factor among self-injurious individuals. Respondents could also select the statement, “Although I considered suicide, I was not that serious about it.” This statement was not categorized, but was used independently to examine whether SIB status affected the selection of this response. Demographic correlates Demographic characteristics and known or putative conditions comorbid with SIB, suicidality, or both were included in the analyses, including gender, age, race/ethnicity, and sexual orientation. Following US census codes, race/ethnicity codes included non-Hispanic black, non-Hispanic white, and Hispanic. An Asian/Asian American category was included as well. The “other” category included American Indian/Alaskan Native, Middle Eastern or East Indian, Native Hawaiian or Pacific Islander, and biracial/ethnic or multiracial/ethnic. These were collapsed into 4 broader categories: Caucasian, black, Asian/Asian American, and other. Gender included 3 options: male, female, and transgendered/nongendered; only 2 respondents selected the last category. Sexual orientation included 4 response options: straight, gay or lesbian, bisexual, and questioning; all that applied could be selected. For these analyses, the 51 respondents who chose 2 or more sexual orientations were categorized as “questioning.” Indicators of trauma and distress Respondent reports of several risk factors included eating disorders; history of sexual, emotional, or physical abuse; and psychological and physical distress. Physical distress was measured using a binary variable reflecting the presence of 4 Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition35 characteristics of disordered eating, which was coded positively if respondents indicated that they had ever repeatedly: (1) severely restricted eating, (2) binged or purged, (3) over-exercised to lose or manage weight, or (4) used laxatives to lose or manage weight. Psychological distress in the past 30 days was assessed using the K-6 scale36,37 (Chronbach α = 0.78). Presence or absence of abuse history was measured using 3 questions developed for this study: “Have you ever been in a physically abusive relationship (including family relationships, romantic relationships, acquaintances, or friendships)?”, “Have you ever experienced sexual touching or penetration against your will?”, and “Have you ever been in a relationship that was emotionally abusive (including family relationships, romantic relationships, acquaintances, or friendships)?” Protective factors Two protective factors were included in these analyses: attraction to life and informal help-seeking. The Attraction to Life Scale was taken from the Multi-Attitude Suicide Tendency Scale.38 Four items with the highest factor loading were selected from the original 7-item scale. All 4 items loaded above 0.7 in the present study and showed acceptable reliability (Chronbach α = 0.77). The informal help-seeking variable was derived from the question: “Who do you feel comfortable getting help from when you feel anxious, sad, or depressed?” Respondents were presented with 17 options and asked to select all that applied. These were then totaled to create the informal help-seeking variable. Statistical analyses All analyses were conducted using SPSS version 13.0 (SPSS Inc, Chicago, Illinois). Descriptive statistics and crude and adjusted odds ratios (AORs) with 95% confidence intervals (CIs) were used to examine the relationship between SIB, suicidality, and correlates. Both bivariate and multinomial logistic regression analyses were used. To examine the extent to which SIB status predicted suicidality, the first analysis used binary logistic regression to examine the relationship between demographic characteristics, trauma and distress indicators, and protective factors and suicidality (coded as absent or present). The second analysis sought to differentiate respondents with SIB only from respondents who reported both SIB and suicidality. Using the group reporting SIB only as the reference group, multinomial logistic regression was then used to examine 3 SIB and suicide-related categories: SIB only, SIB and suicidality, and suicidality only. The analysis examined the extent to which these groups differed from the SIB-only group on demographic characteristics, trauma and distress indicators, and protective factors. The last analysis was intended to determine whether suicidal SIB respondents differed from suicidal non-SIB respondents in terms of specific suicidal characteristics or lethality. To accomplish this, logistic regression analysis was conducted between SIB status (SIB or no SIB) and 4 binary-coded measures of suicidality: ideation, plan, gesture, and attempt. Since demographic characteristics of the population were known, all logistic regression analyses were weighted to control for gender differences in the sample and the population and to equalize differences in response rates in each university. Univariate statistics reported in Table 1 were not weighted. Table 1. View LargeDownload Unweighted Univariate Statistics for Demographic and Trauma Variables Used in Analyses a Results Study population Overall, the sample contained more women than men and 73.0% of the entire group was between the ages of 18 and 24 years. Two thirds of the sample (66.4%) was Caucasian, with Asian/Asian American being the next most represented ethnic/racial category (17.1%). Respondents identifying as heterosexual accounted for 92.5% of the total sample, with 2.2% identifying as gay or lesbian, 2.9% identifying as bisexual, and 2.6% indicating that they were questioning their sexual orientation. As shown in Table 1, 715 (24.9%) respondents ever reported SIB, suicidality, or both. Of those reporting SIB, suicidality, or both, most (40.8%; 10.2% of the total sample) practiced only SIB, 27.4% (6.9% of the total sample) reported SIB and suicidality, and 31.7% (7.9% of the total sample) reported just suicidality. Although not shown in Table 1, when SIB only is broken down by reported lifetime frequency 117 (23.9%) report single incidents, 227 (46.5%) report 2 to 10 incidents, 78 (15.9%) report 11 to 50 incidents, and 42 (8.6%) report more than 50 incidents; in 24 (4.9%) cases, SIB frequency was unknown. Table 1 shows a pattern in keeping with our prediction that individuals reporting no SIB or suicide report lower levels of trauma and distress than those reporting SIB, suicidality, or both. It also shows that individuals reporting both SIB and suicide also report higher levels of trauma and psychological and physical distress than SIB only, suicide only, and neither SIB nor suicide before all variables are taken into account. Relationship between sib and suicidality The second analysis tested our first hypothesis, that SIB status would predict suicide status even when demographic variables were controlled. As shown in Table 2, SIB was strongly predictive of suicidality. Analyses that examine the relationship between reported lifetime SIB frequency and suicide suggest that the relationship to suicidality increases as SIB activity increases until respondents report more than 50 SIB incidents. Adjusted odds ratios for demographics comparing individuals reporting any suicidality showed that suicidal individuals were more likely to be black and to report their sexual orientation as bisexual. They were also more likely to exhibit heightened psychological distress in the last 30 days (a score higher than 13 is considered an indication of psychological distress), to report a greater lifetime prevalence of eating disorders, and to report a history of emotional and sexual trauma. They were also less likely to report informal help-seeking and attraction to life. Table 2. View LargeDownload Logistic Regression of Suicide Status on Demographics, Indicators of Trauma and Distress, Protective Factors, and Self-injury Status a We also predicted that trauma, distress variables, and protective factors would attenuate this relationship by accounting for some of the variance observed. To test this, variables were entered in blocks with demographic characteristics entered first, followed by trauma variables and distress variables. The final block entered the 2 protective factors. Entry of demographic variables had no effect on the relationship between self-injury and suicide (AOR, 6.2; 95% CI, 4.9-7.8). Addition of trauma and psychological and physical distress variables significantly attenuated the relationship between SIB status and suicidality (AOR, 3.7; 95% CI, 2.7-4.9). As shown in the final model, addition of the protective factors weakened the relationship between SIB and suicide only modestly (AOR, 3.4; 95% CI, 2.5-4.6). Close examination of differences between SIB-only respondents and those reporting any suicidality (not shown) were consistent with the hypothesis that respondents reporting both SIB and suicide would report more history of trauma, more psychological and physical distress, and fewer protective factors. Compared with SIB-only respondents, those reporting SIB and suicidality cited higher rates of sexual abuse (AOR, 2.9; 95% CI, 1.4-5.4), emotional abuse (AOR, 1.9; 95% CI, 1.1-3.1), and disordered eating (AOR, 1.8; 95% CI, 1.1-2.9). They also reported less informal help-seeking (AOR, 0.8; 95% CI, 0.7-0.9) and attraction to life (AOR, 0.7; 95% CI, 0.5-0.9). Examination of differences between SIB only and suicide only showed that those reporting suicidality only were significantly more likely to be black (AOR, 5.4; 95% CI, 1.6-17.9) or Asian/Asian American (AOR, 2.7; 95% CI, 1.5-4.7) than Caucasian. They were also significantly more likely to be older than 24 years than between 18 and 20 years (AOR, 2.3; 95% CI, 1.37-4.0) and report less attraction to life (AOR, 0.7; 95% CI, 0.6-0.9). The last hypothesis examined the extent to which SIB and suicide overlap for some individuals. We hypothesized that, among respondents reporting suicidality, those also reporting SIB would be equally likely to report all forms of suicidal behavior, not solely ideation. Adjusted odds ratios (Table 3) support this hypothesis and show that SIB status significantly predicts suicide ideation, plan, gesture, and attempt. Indeed, the strength of the AORs increased as the reported suicide-linked behaviors became more serious and, therefore, potentially more lethal. Examination of differences in the statement “Although I considered suicide, I was not that serious about it” when controlling for all other demographic variables showed no difference between SIB and non-SIB respondents. Although not shown in the table, results also showed that women were 2.2 times (95% CI, 1.2-3.4) more likely to report attempting suicide than men. In comparison to students identifying as straight, students reporting as gay or lesbian were 4.2 times (95% CI, 1.2-14.1) more likely to report attempting suicide, while students identifying as bisexual or questioning were more likely to report planning suicide (AOR, 4.0; 95% CI, 2.1-7.6). These findings are consistent with existing research.39,40 Table 3. View LargeDownload Bivariate Logistic Regression of Demographic and Self-injury Status on Suicidality a Comment We hypothesized that SIB signals a coping strategy to deal with psychological distress that may co-occur or lead to suicidal behaviors in individuals experiencing more duress than they can ultimately effectively mitigate. Consistent with our hypotheses, this study showed that SIB was a strong predictor of suicidality, that individuals who evidenced SIB and suicidality were significantly more likely to score higher on trauma and distress variables and lower on protective factors than those exhibiting SIB only, and that the risk of suicidality increased as SIB frequency increased. We also found that a reported history of SIB predicted all forms of suicidal behavior, not solely ideation. Assuming that the temporal sequence is as we hypothesize here, namely, that SIB precedes or co-occurs with suicide, these findings suggest that, in individuals using SIB as a means of coping with undesired affect, suicide may become a viable consideration if psychological duress overwhelms their capacity to functionally cope using SIB or other methods, such as substance use. Finding that the association between SIB incidents and suicide peaks at 11 to 50 incidents (after which the risk declines) invites several possible interpretations. The one most consistent with our hypotheses suggests that SIB, alone or in addition to other mechanisms, effectively mitigates sustained or sporadic distress for enduring periods among some individuals. The fact that most (60.0%) of those reporting SIB evidenced no suicidality at all supports this theory and helps to explain why so many individuals in the study population using SIB remain undetected by informal and formal support systems.6,12 An alternative explanation for the curvilinear relationship between SIB frequency and suicidality is that high levels of SIB include individuals for whom SIB becomes habitual, compulsive, and initiated in response to stimuli not directly linked to current affective state. Although not explored here, the trend raises questions with clinical implications worthy of further investigation. Our finding that SIB predicted all forms of suicidality and that the magnitude of the association increased as the seriousness of the suicidality increased is consistent with Joiner’s41 theory that engagement in SIB may inadvertently embolden and prepare individuals for more lethal suicide-related behaviors than those who do not engage in SIB. However, because we cannot discern temporal sequence of SIB relative to suicide in these analyses, the applicability of Joiner's theory to this data are limited. Our findings do, however, point to the need for effective means of distinguishing deliberately self-injurious individuals likely to exhibit suicidal behavior from those unlikely to exhibit suicidality. This study is not without limitations. Reliance on data from 2 universities and a less than ideal response rate suggests the possibility of systematic bias among nonrespondents. Nevertheless, the response rate in this study was higher than reported for national surveys conducted on college campuses.42 Moreover, demographic characteristics of the population were known; we were able to use weighted analyses that may compensate for any systematic bias. Comparison to the results reported from the 2005 National College Health Assessment study demonstrates that our sample was more diverse than the national sample, containing more graduate students, international students, men, and minority students.43 Finally, reliance on single-item measures may not capture experiences of interest with a high degree of specificity, and these analyses did not differentiate between a number of potentially important temporal issues such as age at onset and cessation for SIB and suicidality. Self-injurious behavior is present at concerning levels among community adolescents and young adults. Since it is well established that SIB is not a suicidal gesture in and of itself, many clinicians assume that suicide assessment is unnecessary. Our study suggests that, while SIB may serve as a functional, if maladaptive, coping mechanism used to avoid suicide, it may also serve as a harbinger of all forms of suicidality in a subset of individuals. Until clinical tools capable of differentiating levels of risk of suicidality or serious physical harm in patients who exhibit SIB are developed, our findings suggest that the presence of SIB should trigger suicide assessment. The variance accounted for in the link between SIB and suicide by trauma and distress variables also suggests that presence of SIB should trigger psychological assessment and referral. Back to top Article Information Correspondence: Janis Whitlock, MPH, PhD, Family Life Development Center, Beebe Hall, Cornell University, Ithaca, NY 14853 ([email protected]). Accepted for Publication: October 19, 2006. Author Contributions: Dr Whitlock has had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Whitlock and Knox. Acquisition of data: Whitlock. Analysis and interpretation of data: Whitlock and Knox. Drafting of the manuscript: Whitlock and Knox. Critical revision of the manuscript for important intellectual content: Whitlock and Knox. Statistical analysis: Whitlock and Knox. Obtained funding: Whitlock. Administrative, technical, and material support: Whitlock. Study supervision: Whitlock. Financial Disclosure: None reported. Funding/Support: This study was supported by the Seed and Innovation grant awarded by the School of Human Ecology, Cornell University (Dr Whitlock). Analyses were supported, in part, by grant MH055317 of the National Institutes of Mental Health–funded P20 Developing Center for Public Health and Population-Based Approaches to Suicide Prevention (Dr Knox). Disclaimer: The statements and opinions expressed are those of the authors and are not a reflection of the institutions that funded this study. Additional Contributions: John Eckenrode, PhD, and Wendy Nilsen, PhD, provided substantive contributions to early drafts of the manuscript; Amanda Purington, BA, Brian Lukoff, BA, Daniel Silverman, MD, Gina Baral, MPH, and John Kolligian, PhD, made this study possible. References 1. Favazza ARConterio K Female habitual self-mutilators. Acta Psychiatr Scand 1989;79283- 289PubMedGoogle ScholarCrossref 2. Gratz KL Measurement of deliberate self-harm: preliminary data on the Deliberate Self-Harm Inventory. J Psychopathol Behav Assess 2001;23253- 263Google ScholarCrossref 3. Muehlenkamp JJGutierrez PMOsman ABarrios LC Validation of the Positive and Negative Suicide Ideation (PANSI) inventory in a diverse sample of young adults. J Clin Psychol 2005;61431- 445PubMedGoogle ScholarCrossref 4. Stanley BGameroff MJMichalsen BAMann JJ Are suicide attempters who self-mutilate a unique population? Am J Psychiatry 2001;158427- 432PubMedGoogle ScholarCrossref 5. Kokaliari E Deliberate Self-injury: An Investigation of the Prevalence and Psychosocial Meanings in a Non-clinical Female College Population. Northampton, Mass Smith College School for Social Work2005; 6. Whitlock JLEckenrode JESilverman D Self-injurious behavior in a college population. Pediatrics 2006;1171939- 1948PubMedGoogle ScholarCrossref 7. Esposito CSpirito ABoergers JDonaldson D Affective, behavioral, and cognitive functioning in adolescents with multiple suicide attempts. Suicide Life Threat Behav 2003;33389- 399PubMedGoogle ScholarCrossref 8. Chapman ALGratz KLBrown MZ Solving the puzzle of deliberate self-harm: the experimental avoidance model. Behav Res Ther 2006;44371- 394PubMedGoogle ScholarCrossref 9. Ross SHeath N A study of the frequency of self-mutilation in a community sample of adolescents. J Youth Adolesc 2002;3166- 77Google ScholarCrossref 10. Klonsky ED The functions of deliberate self-injury: a review of the evidence. Clin Psychol Rev 2007;27 ((2)) 226- 239PubMedGoogle ScholarCrossref 11. Briere JGil E Self-mutilation in clinical and general population samples: prevalence, correlates, and functions. Am J Orthopsychiatry 1998;68609- 620PubMedGoogle ScholarCrossref 12. Hawton KRodham KEvans EWeatherall R Deliberate self-harm in adolescents: self report survey in schools in England. BMJ 2002;3251207- 1211PubMedGoogle ScholarCrossref 13. Linehan MM Suicidal people: one population or two? Ann N Y Acad Sci 1986;48716- 33PubMedGoogle ScholarCrossref 14. Brown MComtois KALineham MM Reasons for suicide attempts and nonsuicidal self-injury in women with borderline personality disorder. J Abnorm Psychol 2002;111198- 202PubMedGoogle ScholarCrossref 15. Nock MKJoiner TEGordon KHLoyd-Richardson EPrinstein M Non-suicidal self-injury among adolescents: diagnostic correlates and relation to suicide attempts. Psychiatry Res 2006;14465- 72PubMedGoogle ScholarCrossref 16. Guertin TLloyd-Richardson ESpirito ADonaldson DBoergers J Self-mutilative behavior in adolescents who attempt suicide by overdose. J Am Acad Child Adolesc Psychiatry 2001;401062- 1069PubMedGoogle ScholarCrossref 17. Muehlenkamp JJGutierrez PM An investigation of differences between self-injurious behavior and suicide attempts in a sample of adolescents. Suicide Life Threat Behav 2004;3412- 24PubMedGoogle ScholarCrossref 18. Bennett SCoggan CAdams P Problematising depression: young people, mental health and suicidal behaviours. Soc Sci Med 2003;57289- 299PubMedGoogle ScholarCrossref 19. Linehan MMArmstrong HESuarez AAllmon DHeard HL Cognitive-behavioral treatment of chronically parasuicidal borderline patients. Arch Gen Psychiatry 1991;481060- 1064PubMedGoogle ScholarCrossref 20. Linehan MMTutek DAHeard HLHE A Interpersonal outcome of cognitive behavioral treatment for chronically suicidal borderline patients. Am J Psychiatry 1994;1511771- 1776PubMedGoogle Scholar 21. Salkovskis PMAtha CStorer D Cognitive-behavioural problem solving in the treatment of patients who repeatedly attempt suicide: a controlled trial. Br J Psychiatry 1990;157871- 876PubMedGoogle ScholarCrossref 22. Upadhyaya AKConwell YDuberstein PRDenning DCox C Attempted suicide in older depressed patients: effect of cognitive functioning. Am J Geriatr Psychiatry 1999;7317- 320PubMedGoogle Scholar 23. Andrus JKFleming DWHeumann MAWassell JTHopkins DDGordon J Surveillance of attempted suicide among adolescents in Oregon, 1988. Am J Public Health 1991;811067- 1069PubMedGoogle ScholarCrossref 24. Mann JJWaternaux CHaas GLMalone KM Toward a clinical model of suicidal behavior in psychiatric patients. Am J Psychiatry 1999;156181- 189PubMedGoogle Scholar 25. Mann JJ Searching for triggers of suicidal behavior. Am J Psychiatry 2004;161395- 397PubMedGoogle ScholarCrossref 26. Oquendo MAMalone KMEllis SPSackeim HAMann JJ Inadequacy of antidepressant treatment for patients with major depression who are at risk for suicidal behavior. Am J Psychiatry 1999;156190- 194PubMedGoogle Scholar 27. Yates TM The developmental psychopathology of self-injurious behavior: compensatory regulation in posttraumatic adaptation. Clin Psychol Rev 2004;2435- 74PubMedGoogle ScholarCrossref 28. Skegg K Self-harm. Lancet 2005;3661471- 1483PubMedGoogle ScholarCrossref 29. Linehan MMMaris RWedCanetto SSedSara Sed Behavioral treatments of suicidal behaviors: definitional obfuscation and treatment outcomes. Review of Suicidology. New York, NY Guilford Press2000;84- 111Google Scholar 30. Muehlenkamp JJ Self-injurious behavior as a separate clinical syndrome. Am J Orthopsychiatry 2005;75324- 333PubMedGoogle ScholarCrossref 31. Kessler RCBorges GWalters EE Prevalence of and risk factors for lifetime suicide attempts in the National Comorbidity Survey. Arch Gen Psychiatry 1999;56617- 626PubMedGoogle ScholarCrossref 32. National Center for Health Statistics, Table 46: death rates for suicide, according to sex, race, Hispanic origin, and age: selected years 1950-2003 Health, United States, 2005: with chartbook on trends in the health of Americans http://www.cdc.gov/nchs/data/hus/hus05.pdf#046Accessed December 11, 2006Google Scholar 33. Gutierrez PMOsman ABarrios FXKopper BA Development and initial validation of the Self-Harm Behavior Survey. J Pers Assess 2001;77475- 490PubMedGoogle ScholarCrossref 34. Savin-Williams RCReam GL Suicide attempts among sexual-minority male youth. J Clin Child Adolesc Psychol 2003;32509- 522PubMedGoogle ScholarCrossref 35. American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC American Psychiatric Association1994; 36. Kessler RCMcGonagle KAZhao S et al. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: results from the National Comorbidity Survey. Arch Gen Psychiatry 1994;518- 19PubMedGoogle ScholarCrossref 37. Kessler RCBarker PRColpe LJ et al. Screening for serious mental illness in the general population. Arch Gen Psychiatry 2003;60184- 189PubMedGoogle ScholarCrossref 38. Orbach IMilstein IHar-Even DApter ATyano SWelizur A A Multi-Attitude Suicide Tendency Scale for adolescents. Psychol Assess 1991;3398- 404Google ScholarCrossref 39. Muehrer P Suicide and sexual orientation: a critical summary of recent research and directions for future research. Suicide Life Threat Behav 1995;2572- 81PubMedGoogle Scholar 40. Remafedi GFrench SStory M The relationship between suicide risk and sexual orientation: results from a population-based study. Am J Public Health 1998;8857- 60PubMedGoogle ScholarCrossref 41. Joiner TE Why People Die by Suicide. Cambridge, Mass Harvard University Press2006; 42. American College Health Association,; National College Health Assessment, Reference group data report, fall 2003 http://www.acha.org/projects_programs/NCHA_docs/ACHA-NCHA_Reference_Group_Report_Fall2003.pdf 43. American College Health Association, American College Health Association: National College Health Assessment: Spring 2003 reference group report. J Am Coll Health 2005;53199- 210PubMedGoogle Scholar

2007 Archives of Pediatrics & Adolescent Medicine

doi: 10.1001/archpedi.161.7.64010.1001/archpedi.161.4.356

Error in Text. In the article titled “Understanding Autism: Parents and Pediatricians in Historical Perspective” by Silverman and Brosco published in the April issue of the Archives (2007;161[4]:392-398), an error occurred on page 394. In the first paragraph of the second column, the fourth and fifth sentences should have read as follows: “Parents are guaranteed a say in the review process: CAN maintains a scientific review committee comprising scientific degree–holding parents of children with autism; this review committee ranks projects after an initial review by a scientific advisory group (written communication, Therese Finazzo, January 5, 2006). The National Alliance for Autism Research maintains a similar 2-tiered system (written communication, Alycia Halladay, PhD, December 27, 2005).”

Brunner, Romuald;Parzer, Peter;Haffner, Johann;Steen, Rainer;Roos, Jeanette;Klett, Martin;Resch, Franz

2007 Archives of Pediatrics and Adolescent Medicine

doi: 10.1001/archpedi.161.7.641pmid: 17606826

Abstract Objective To determine the prevalence and the associated psychological and social factors of occasional and repetitive deliberate self-harming behavior in adolescents. Design Cross-sectional self-report survey. Setting One hundred twenty-one schools in Germany. Participants A representative sample of 5759 ninth-grade students was studied between 2004 and 2005. Outcome Measures Deliberate self-harm (DSH) and suicidal behaviors, emotional and behavioral problems (Youth Self-Report), living standard, family composition, parental conflict and illness, school type and performance, relationship to peers, bullying, body satisfaction and dieting, media consumption, smoking, and alcohol and drug use. Results Occasional forms of DSH within the previous year were reported by 10.9% of the ninth-grade students. Four percent of the students reported repetitive forms of DSH. Suicidal behavior was strongly associated with repetitive DSH, an association that held for both subtypes of DSH. The findings also indicated that social background factors were important concomitants of occasional DSH but were not related to an increased likelihood of repetitive DSH. Symptoms of depression/anxiety and delinquent/aggressive behavior were associated with self-harming behavior in both adolescent girls and boys. Conclusions The data suggest that there is a link between social factors and occasional DSH and, especially in repetitive DSH, that there is a strong association between DSH and suicidal behavior as well as DSH and emotional and behavioral problems. These findings indicate a different pathway in the development of DSH in adolescents. The results support a need to investigate the possible neurobiological underpinnings of DSH within a longitudinal model to enhance the knowledge of this poorly understood behavior. Recent epidemiological studies have demonstrated high rates of deliberate self-harm (DSH) in adolescents.1,2 Deliberate self-harm is defined as the intentional injuring of one's body without apparent suicidal intent.3 Repetition of DSH is also frequent in adolescents and includes a wide range of behaviors (eg, cutting and burning). Approximately 6% to 7% of school students reported having self-harmed in the previous year.1,2 Although DSH is serious and may precede suicide,4,5 few patients who deliberately self-harm are referred to professional institutions.1 Common indicators that have been found to be associated with repeated self-harm in adolescents include personality disturbances, depression, alcohol and drug use, troubled relationships with peers and/or family members, poor school performance, and chronic psychosocial and behavioral problems.1 There is growing evidence that DSH has become more prevalent in adolescents during the past decade and a half.6 A high association between self-harming behavior and a diagnosis of borderline personality disorder has been shown in adult7 as well as adolescent8 patients. Likewise, the proportion of young self-harming inpatients with depressive, bipolar, or substance abuse disorders has increased during the past several years.9 High rates of DSH have been reported from 2 large school surveys using anonymous self-report questionnaires.1,2 At least 1 episode of DSH during the previous 12 months was reported by 6% to 7% of the surveyed 15-year-old students. Approximately two-thirds of adolescents who recounted episodes of self-harm were engaged in self-injurious behavior (ie, cutting), whereas the other third reported poisoning mainly due to overdosing on medication.1,2 Adolescent girls had significantly higher rates (3-7 times) of episodes of self-harm than adolescent boys. Both studies documented an association between a history of DSH and a broad range of psychiatric symptoms, such as low self-esteem, drug abuse, and, especially in adolescent girls, depression, anxiety, and impulsivity. In this study, we focused on self-injurious behaviors without suicidal intent (ie, cutting and burning). Empirical data from previous clinical10 as well as nonclinical11 studies have revealed the following motives of adolescents who engage in DSH: to get relief from distress, to escape from a situation, and to show how desperate they are feeling.12 Single or occasionally performed self-mutilating acts may be interpreted as time-limited testing or imitation behavior1 during the period of adolescence, whereas repetitiously performed acts serve primarily as methods of affect regulation in the context of further psychiatric diagnoses. It may be assumed that a single episode of DSH can have a qualitatively different meaning than repeated episodes of DSH. The primary purpose of our study was to determine the prevalence of 2 different types of DSH in adolescents: occasional and repetitive. Second, we aimed to investigate the relationship between DSH and a wide range of internalizing (ie, withdrawal, somatic complaints, and anxiety/depression), externalizing (ie, delinquency and aggression), and suicidal (ie, suicidal ideation, suicide intention, and suicide attempts) behaviors as well as social background factors and risk-taking behaviors. Methods Sample and procedures The Heidelberg School Study was used to determine the prevalence and psychological concomitants of self-destructive behavior and other forms of risk behavior in adolescents. Enrollment in this study took place in cooperation with the Heidelberg Public Health Service between October 2004 and January 2005. The primary sampling pool included all schools that had a ninth grade in the Rhein-Neckar district. This area is typical for geographically mixed populations in Germany and is representative of the distribution of types of schools and parental socioeconomic statuses.13 The German school system consists of 4 school types characterized by the academic ability of the students and different types of graduation: Gymnasium, 8 years of school after 4 years of elementary school, terminating with the general qualification for university entrance; Realschule, 6 years of school after 4 years of elementary school, terminating with a secondary-school level-I certificate; Hauptschule, 9 years of elementary school; and Förderschule, a school for special educational needs (for children who have lower intellectual abilities but are literate, in general). Overall, 116 of 121 schools agreed to participate. Five schools declined to participate without giving a reason. All students in ninth-grade classes were invited to take part in the study (N = 6842). The analyzed data represent 95.8% of the approached students and 85.2% of the entire population ninth-grade students in this school district. The Figure shows the number of eligible and actual participants in this study and reasons for noninclusion. Figure. View LargeDownload Flowchart of investigated subjects. Informed consent for participation was obtained from the contacted adolescents. Their parents or guardians were notified about the study by letter about 4 weeks before the study's commencement. The adolescents' anonymity and voluntary status were ensured before they agreed to participate in the study. Data collectors, who had been trained in test administration by the researchers, administered the questionnaire to all subjects during a regular class period. A teacher was always present in the classroom because of legal reasons, but he or she did not intervene. In the case of the adolescents in special education classes, the survey personnel assisted subjects who had problems filling out the questionnaire and offered them additional time to complete it. Only in a few cases was this support necessary. The study was approved by the ethics committee of the faculty of medicine at the University of Heidelberg. Assessment A self-report booklet, which included the Youth Self-Report and 53 additional items, was administered to all participating adolescents. A pilot test of this comprehensive questionnaire was completed in 2 school classes. Demographic information included age, sex, and nationality. The additional items in the self-report booklet included school situation, family background, relationships with peers, media consumption, and substance intake during the past 6 months. Pertinent parts of the German version14 of the Schedule for Affective Disorders and Schizophrenia for School-Age Children15 were administered to assess the prevalence and frequency of self-harm and suicidal behavior. The frequency of DSH was assessed by the following response options: never, 1 to 3 times a year, and 4 times or more a year. The frequency of 1 to 3 times a year was defined as occasional DSH, and the frequency of 4 times or more a year was defined as repetitive DSH. Lifetime experiences with suicide ideation, plans, and attempts were queried. To assess a broad range of emotional and behavioral problems potentially associated with self-harm, the German version16 of the Youth Self-Report17 was administered, a self-report version of the Child Behavior Checklist.18 Statistical analysis The frequency of self-mutilation was assessed as a multinomial variable with the following 3 categories: no DSH, occasional DSH, and repetitive DSH. The relationship of the explanatory variables with self-mutilation was analyzed using the multinomial logistic regression analysis. The multinomial logistic regression analysis is a generalization of the logistic regression analysis for binary response variables to nominal response variables with more than 2 categories. It can be thought of as simultaneously estimating logistic regressions for each possible pair of categories. Only n − 1 equations have to be estimated for a response variable with n categories, as the coefficients for the other pairs can be calculated from these equations. In practice, one category is selected as the reference category and all others are compared with it. We used no DSH as the reference category. A multivariate multinomial logistic regression, including all explanatory variables and the interactions of each variable with sex, contained too many colinearities to provide interpretable results. Therefore, we carried out a multinomial regression analysis for each explanatory variable, with self-mutilation as the response variable, the explanatory variable, sex, and the interaction of the explanatory variable with sex. Then, we calculated a multinomial regression, including all explanatory variables, but only with those interactions with sex that were significant in the bivariate regressions. Each interaction that was not significant in the context of the multivariate regression was also removed. The resultant model is the one we report. To find the most important explanatory variables, we carried out a stepwise backward regression procedure to minimize the Akaike information criterion. The Akaike information criterion is a measurement of goodness of fit. The model with the smallest Akaike information criterion can be considered the best-fitting model. As in all large surveys with many variables, the accumulation of missing values was a problem. From the 42 variables used in the multivariate analysis, 10 were complete, 21 had less than 1% of values missing, and the remaining 11 had less than 5% of values missing. Complete data were available for 78.8% of the participants; 99.7% had missing values in up to 7 variables. The remaining 17 participants had missing values in 8 to 12 variables. The analysis of only complete cases with the multivariate regressions would have resulted in a substantially reduced and unrepresentative sample. To overcome this problem, we used multiple imputations. Missing values were replaced by imputed values using the multivariate imputation by chained equations algorithm.19 This procedure was repeated 10 times to create 10 complete data sets. The multivariate regression analysis was carried out with each data set and the results were combined.20 The bivariate regressions used the original data set, as missing values were not a problem (both self-mutilation and sex had no missing values). Statistical significance was set at P<.05; all tests were likelihood ratio tests. All of the analyses were conducted using the statistical computer software program Stata, version 9.2 (Stata Corp, College Station, Texas). Results The sample consisted of 5759 ninth-grade students who completed assessment forms. The mean age of the participating adolescents was 14.9 years (SD, 0.73); 2752 (49.8%) were female. The occasional form of DSH within the previous year was reported by 630 (10.9%) students, whereas 229 (4.0%) students reported the repetitive form of DSH; 14.8% of the adolescents with occasional DSH and 27.1% with repetitive DSH were receiving psychological treatment. Table 1 presents the sample classified into 3 groups (no, occasional, and repetitive DSH) on the basis of the adolescents' reported history of DSH. For each group, the table reports sociodemographic variables, school performance, familial background, media consumption, substance intake, suicidal behavior, body-related issues, and emotional and behavioral problems. Table 1. View LargeDownload Characteristics of All Potential Predictors of Deliberate Self-harma Most participants were German (88.7%). The proportion of different nationalities was typical for the youth population in Germany. The distribution of school types among the participating adolescents was also typical for Germany.13 Of all participants, 75.4% were living with both parents. With respect to suicidal behavior, suicidal thoughts were reported by 14.4% of the adolescents. A life-long history of 1 or more suicide attempts was reported by approximately 8% of the students. Considering each explanatory variable separately, we found each variable (except for body mass index [calculated as weight in kilograms divided by height in meters squared] and number of friends) to have a significant association with DSH. We found that an individual's sex had a significant interaction with school performance, smoking cigarettes, parental marital problems, social problems, and delinquent behavior. In the multivariate model only, school performance (likelihood ratio, χ22 = 7.18; P = .03) and smoking cigarettes (likelihood ratio, χ26 = 15.33; P = .02) were still significant, whereas parental marital problems (likelihood ratio, χ24 = 7.95; P = .09), social problems (likelihood ratio, χ22 = 2.18; P = .34), and delinquent behavior (likelihood ratio, χ22 = 1.41; P = .5) no longer had sex-specific effects. Thus, only the interactions of sex with school performance and with smoking cigarettes were included in the final model. Table 2 shows the adjusted odds ratios for the odds of occasional or repetitive forms of DSH vs no DSH for the final model. The occurrence (risk) of repetitive DSH was more than 2 times higher among German adolescents in comparison with adolescents from other countries. There was also evidence that adolescents with a low body mass index were at high risk to be engaged in repetitive DSH; the lower the body mass index, the higher the risk of repetitive DSH. Adolescent girls with low school performance were more likely to be engaged in occasional acts of DSH. For both sexes, there was an increase in risk of occasional DSH in adolescents with lower academic achievement. Table 2. View LargeDownload Adjusted ORs a for the Odds of Occasional or Repetitive DSH vs No DSH for the Final Model The findings may indicate that social factors (eg, school type, academic achievement, and health problems of parents and siblings) are important concomitants in the onset of occasional DSH, whereas these factors did not display any role (heightened risk) in the case of repetitive DSH. There was no evidence that other kinds of living circumstance (eg, family composition, housing problems, or financial problems) were associated with either type of DSH. Adolescent girls who smoked demonstrated a comparably high rate of risk (approximately 2-3 times higher) for both types of DSH. There was no significant association between smoking and DSH in adolescent boys. Generally, more adolescent girls than adolescent boys were smoking in this age group. These findings may indicate that smoking displays a different meaning for adolescent girls and may also be a sign of having problems. The amount of alcohol intake or use of drugs (eg, analgesics, tranquilizers, or barbiturates) was not related to any type of DSH. Adolescents who reported occasional consumption of illicit drugs demonstrated an elevated risk for occasional DSH, whereas adolescents with a frequent consumption of illicit drugs showed no heightened risk for either occasional or repetitive DSH. This finding may indicate that youths with a high frequency of drug consumption regulate their emotions or tension by drugs rather than by self-mutilating acts. The strongest risk for being engaged in DSH was the occurrence of suicidal behavior, especially suicidal ideation. There is evidence to suggest a qualitative difference between occasional and repetitive forms of DSH. Adolescents who reported that they sometimes had suicidal thoughts demonstrated a 3-fold higher risk of occasional DSH, whereas the risk of repetitive DSH was increased 7-fold. Adolescents who reported a frequent occurrence of suicidal ideation showed an 18-fold risk of being engaged in repetitive DSH, in contrast to an approximately 2-fold greater risk for occasional DSH. A history of suicide attempts increased the risk for the occurrence of DSH. A history of more than 1 suicide attempt increased the risk for repetitive DSH by 6-fold and occasional DSH by 3-fold. With respect to body-related issues, adolescents who perceived themselves to be overweight demonstrated a 3-fold greater risk for repetitive DSH. A combination of this distorted body image and a low body mass index—the core symptoms of eating disorders—increased the risk for DSH, especially for repetitive DSH. Regarding family background factors, health problems in the family as well as problems with siblings were associated with a slightly increased risk of occasional DSH. Symptoms of anxiety and depression, as measured by the Youth Self-Report, were associated with an increased risk of both types of DSH. Delinquent behavior was also positively linked to both types of DSH, whereas aggressive behavior was related only to occasional DSH. Comment In this study, occasional forms of DSH within the previous year were reported by 14.9% of the school students. Four percent demonstrated repetitive forms of DSH. The prevalence rate of 18.9% for all forms of DSH was higher than in previous studies conducted among students in England (6.9%)1 and Australia (6.2%).2 This discrepancy might be explained by the different cultural background factors in Germany, or it might possibly be an indicator for the increased incidence of DSH. Congruent findings of the prevalence of repetitive DSH can be found. For example, 3.7% of the adolescent sample from England reported multiple acts of DSH, which is very similar to the prevalence rate in our study (4.0%). Contrary to the finding of a 3- to 7-fold higher prevalence rate in adolescent girls compared with adolescent boys in both previous studies, we found a doubled prevalence rate of DSH in adolescent girls. The results of this study suggest that social factors like school-related (eg, school type and poor academic achievement) and family-related (eg, health problems of parents and/or siblings) variables especially demonstrated a strong association with occasional DSH. However, these factors did not show any association with repetitive forms of DSH. Psychological factors seem to be more strongly associated with repetitive DSH than with occasional DSH. Body image problems as well as the self-perception of having problems were significantly associated only with repetitive DSH, whereas suicidal behavior (ie, suicidal ideation and suicidal attempts) was associated with both forms of DSH (though much more strongly with repetitive DSH). In summary, our results indicate that different influencing factors may be present in the development of the 2 types of DSH. Against this background, the differentiation between occasional and repetitive DSH is reasonable, because there is clinical evidence that repetitive forms of DSH are those predominantly linked to psychiatric syndromes.9 Our results also clearly demonstrate that a greater amount of internalizing problems like anxiety and depressive symptoms, as well as externalizing problems like delinquent behavior, is closely related to both subtypes of DSH. This co-occurrence of both categories of symptoms may indicate disturbances in personality development. Aggressive behavior also showed an influence on DSH, though a significant association was revealed only with occasional DSH. An unexpected finding was that no sex-specific association between the investigated variables and both subtypes of DSH could be found in this study, with the exception of smoking and low academic performance in adolescent girls. These latter findings may indicate that smoking has a different meaning for adolescent girls than adolescent boys and may also be a sign of having other problems. Previous studies have pointed to a relationship between smoking and suicidal phenomena as well as poor body image and suicidal phenomena among adolescent girls.21,22 Overall, the results of this study support the view that DSH can be linked to emotional and behavioral problems in adolescents. Previous findings of an association of DSH with awareness of recent self-harm in peers has led to the postulation of a modeling effect in accordance with instances of self-harm among adolescent psychiatric patients.23 Because adolescents with psychological problems may be more prone to undertake self-mutilating acts, the opinion that DSH among adolescents may be because they view it as “fashionable” could not be supported by this study. The overall interpretation of our results is limited because of the cross-sectional design of the study. We do not know whether the psychological symptoms are causes or consequences of DSH. A possible causal relationship can be investigated only in a longitudinally designed study. This drawback may be alleviated by the fact that the investigation took place at onset of DSH, as there is growing evidence from clinical and nonclinical samples that DSH typically begins in early to middle adolescence.10 An advantage of the cross-sectional design is its strict anonymity, which heightened the acceptance of all involved parties and was a prerequisite of the school authorities. Furthermore, we can postulate that the investigated social background features represent more stable factors, thus making it unlikely that they are consequences of DSH. Several factors strengthen the validity of the findings reported in this study. First, our sample was representative of ninth-grade students in Germany. Second, we used a clear definition of type and frequency of DSH to differentiate it from other self-destructive acts. Third, adolescents' reports may be of enhanced validity, as this survey was based on a self-rating conducted anonymously. The high prevalence rate of DSH as well as its strong link to suicidal behavior and emotional and behavioral problems may serve as a forewarning to school counselors and public health authorities. Along with former studies,1,2 this study provides strong evidence that adolescents with DSH rarely seek professional help. The results of this survey highlight the importance of an awareness of DSH so that interventions can be properly targeted. Recent studies in the adult clinical population have confirmed a strong connection between self-harm and subsequent suicide.24 It has been assumed that in many cases self-mutilating acts represent a transient period of distress, whereas in other cases it is an important indicator of psychiatric disturbances.4 Only a longitudinal study of long duration will be able to provide answers as to whether occasional occurrences of DSH are precursors of repetitive DSH and under which conditions a remission or transition into repetitive DSH might occur. Another question of particular interest is whether repetitive DSH also occurs periodically or represents an early indicator of severe personality disturbances like borderline personality disorder and whether it is linked to other severe psychiatric conditions like depressive, bipolar, and substance use disorders.9 It has been posited that the pathways from emotional, behavioral, and social factors to DSH are influenced by complex and insufficiently known interactions among constitutional factors and external factors, such as stressors, trauma, family pathology, and cultural factors.24 Future research on DSH in adolescents therefore should address aspects of possible biological vulnerabilities (including genetic markers) that could illuminate differences between the subtypes of DSH. Back to top Article Information Correspondence: Romuald Brunner, MD, Department of Child and Adolescent Psychiatry, University of Heidelberg, Blumenstrasse 8, Heidelberg 69115, Germany ([email protected]). Author Contributions:Study concept and design: Brunner, Parzer, Haffner, Steen, Roos, Klett, and Resch. Acquisition of data: Parzer, Haffner, Steen, and Klett. Analysis and interpretation of data: Brunner, Parzer, and Haffner. Drafting of the manuscript: Brunner and Parzer. Critical revision of the manuscript for important intellectual content: Haffner, Steen, Roos, Klett, and Resch. Statistical analysis: Parzer. Obtained funding: Haffner, Steen, Klett, and Resch. Administrative, technical, and material support: Brunner, Parzer, Haffner, Steen, Roos, Klett, and Resch. Study supervision: Brunner, Haffner, Roos, and Resch. Accepted for Publication: March 22, 2007. Financial Disclosure: None reported. Funding/Support: This research was supported by the Lautenschlaeger Foundation, the Public Health Department of Heidelberg, and the faculty of medicine at the University of Heidelberg. Role of the Sponsor: None of the sponsoring organizations had any involvement in the design or conducting of the study; in the collection, management, analysis, or interpretation of data; or the preparation, review, or approval of the manuscript. Additional Contributions: We thank the teachers and survey personnel for their support. References 1. Hawton KRodham KEvans EWeatherall R Deliberate self harm in adolescents: self report survey in schools in England. BMJ 2002;325 (7374) 1207- 1211PubMedGoogle ScholarCrossref 2. De Leo DHeller TS Who are the kids who self-harm? an Australian self-report school survey. Med J Aust 2004;181 (3) 140- 144PubMedGoogle Scholar 3. Pattison EMKahan J The deliberate self-harm syndrome. Am J Psychiatry 1983;140 (7) 867- 872PubMedGoogle Scholar 4. Brent DA The aftercare of adolescents with deliberate self-harm. J Child Psychol Psychiatry 1997;38 (3) 277- 286PubMedGoogle ScholarCrossref 5. Olfson MGameroff MJMarcus SCGreenberg TShaffer D Emergency treatment of young people following deliberate self-harm. Arch Gen Psychiatry 2005;62 (10) 1122- 1128PubMedGoogle ScholarCrossref 6. Hawton KHarriss LHall SSimkin SBale EBond A Deliberate self-harm in Oxford, 1990-2000: a time of change in patient characteristics. Psychol Med 2003;33 (6) 987- 995PubMedGoogle ScholarCrossref 7. Linehan MM Cognitive-Behavioral Treatment of Borderline Personality Disorder. New York, NY Guilford Press1993; 8. Brunner RParzer PResch F Dissoziative Symptome und traumatische Lebensereignisse bei Jugendlichen mit einer Borderline-Störung. Persönlichkeitsstörungen 2001;54- 12Google Scholar 9. Olfson MGameroff MJMarcus SCGreenberg TShaffer D National trends in hospitalization of youth with intentional self-inflicted injuries. Am J Psychiatry 2005;162 (7) 1328- 1335PubMedGoogle ScholarCrossref 10. Nixon MKCloutier PFAggarwal S Affect regulation and addictive aspects of repetitive self-injury in hospitalized adolescents. J Am Acad Child Adolesc Psychiatry 2002;41 (11) 1333- 1341PubMedGoogle ScholarCrossref 11. Evans EHawton KRodham K Factors associated with suicidal phenomena in adolescents: a systematic review of population-based studies. Clin Psychol Rev 2004;24 (8) 957- 979PubMedGoogle ScholarCrossref 12. Rodham KHawton KEvans E Reasons for deliberate self-harm: comparison of self-poisoners and self-cutters in a community sample of adolescents. J Am Acad Child Adolesc Psychiatry 2004;43 (1) 80- 87PubMedGoogle ScholarCrossref 13. Statistisches Bundesamt, Bildung und Kultur: Allgemeinbildende Schulen, Schuljahr 2004/2005. Wiesbaden, Germany Statistisches Bundesamt2006; 14. Delmo CWeiffenbach OGabriel MPoustka F Kiddie-Sads-Present and Lifetime Version (K-SADS-PL): Auflage der Deutsche Forschungsversion. 3rd ed. Frankfurt, Germany Klinik für Psychiatrie und Psychotherapie des Kindes, Jugendalters der Universität Frankfurt2000; 15. Kaufman JBirhamer BBrent D et al. Schedule for Affective Disorders and Schizophrenia for School-Age Children, Present and Lifetime Version (K-SADS-PL): initial reliability and validity data. J Am Acad Child Adolesc Psychiatry 1997;36 (7) 980- 988PubMedGoogle ScholarCrossref 16. Döpfner MPlück JBölte SLenz KMelchers PHeim K Youth Self-Report: German Adaptation of the Youth Self-Report (YSR). 2nd ed. Cologne, Germany University of Cologne1998; 17. Achenbach TM Manual for the Youth Self-Report and 1991 Profile. Burlington, VT University of Vermont, Department of Psychology1991; 18. Achenbach TM Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington, VT University of Vermont, Department of Psychology1991; 19. van Buuren SOudshoorn K Flexible Multivariate Imputation by MICE. Leiden, the Netherlands TNO Prevention and Health1999;TNO Report PG/VGZ/99.054 20. Rubin DB Multiple Imputation for Nonresponse in Surveys. New York, NY Wiley1987; 21. Breslau NSchultz LRJohnson EOPeterson ELDavis GC Smoking and the risk of suicidal behavior: a prospective study of a community sample. Arch Gen Psychiatry 2005;62 (3) 328- 334PubMedGoogle ScholarCrossref 22. Eaton DKLowry RBrener NDGaluska DACrosby AE Associations of body mass index and perceived weight with suicide ideation and suicide attempts among US high school students. Arch Pediatr Adolesc Med 2005;159 (6) 513- 519PubMedGoogle ScholarCrossref 23. Taiminen TJKallio-Soukainen KNokso-Koivisto HKaljonen AHelenius H Contagion of deliberate self-harm among adolescent inpatients. J Am Acad Child Adolesc Psychiatry 1998;37 (2) 211- 217PubMedGoogle ScholarCrossref 24. Skegg K Self-harm. Lancet 2005;366 (9495) 1471- 1483PubMedGoogle ScholarCrossref

Mandel, Keith E.;Kotagal, Uma R.

2007 Archives of Pediatrics & Adolescent Medicine

doi: 10.1001/archpedi.161.7.650pmid: 17606827

Abstract Objective To determine whether aligning design characteristics of a pay-for-performance program with objectives of an asthma improvement collaborative builds improvement capability and accelerates improvement. Design Interrupted time series analysis of the impact of pay for performance on results of an asthma improvement collaborative. Setting Forty-four pediatric practices within greater Cincinnati. Participants Forty-four pediatric practices with 13 380 children with asthma. Interventions The pay-for-performance program rewarded practices for participating in the collaborative, achieving network- and practice-level performance thresholds, and building improvement capability. Pay for performance was coupled with additional improvement interventions related to the collaborative. Outcome Measures Flu shot percentage, controller medication percentage for children with persistent asthma, and written self-management plan percentage. Results The pay-for-performance program provided each practice with the potential to earn a 7% fee schedule increase. Three practices earned a 2% increase, 13 earned a 4% increase, 2 earned a 5% increase, 14 earned a 6% increase, and 11 earned a 7% increase. Between October 1, 2003, and November 30, 2006, the percentage of the network asthma population receiving “perfect care” increased from 4% to 88%. The percentage of the network asthma population receiving the influenza vaccine increased from 22% to 41%, and then to 62% during the prior 3 flu seasons. Conclusion Linking design characteristics of a pay-for-performance program to a collaborative focused on improving care for a defined population, building improvement capability, and driving system changes at the provider level resulted in substantive and sustainable improvement. Despite the rapid growth of pay-for-performance programs across the United States,1-3 evidence regarding their effect on quality of care is limited.4-11 Even in instances in which pay-for-performance programs have been linked to measurable improvement, attribution is problematic.12-14 Although “guiding principles” exist,8,9,15-20 the lack of evidence regarding effective design characteristics for pay-for-performance programs remains a significant concern. Although ideal aspects of pay-for-performance programs remain elusive, we hypothesized that aligning pay-for-performance program design characteristics with the primary objectives of a large- scale asthma improvement collaborative and coupling pay for performance with other interventions would enhance improvement capability and accelerate improvement, within and across primary care practices. This approach was based on the contention that pay for performance should be viewed as a catalyst to accelerate sustainable transformation at the provider level and that an overdependence on pay for performance alone to drive quality should be avoided. Based on results achieved, key pay-for-performance program design principles will be reviewed to inform the national dialog among providers, payers, and employers. Methods The Physician-Hospital Organization (PHO) affiliated with Cincinnati Children's Hospital Medical Center launched an asthma improvement collaborative in October 2003, impacting more than 13 000 children with asthma across 44 primary care practices (165 physicians) within greater Cincinnati, representing approximately 35% of the region's pediatric asthma population. The primary care practices are organized as an independent practice association. The PHO elected to focus on asthma because the prevalence is high, care is usually managed by primary care practices, and extensive literature exists regarding the positive impact of improvement interventions on process and outcome measures. The aim of the asthma initiative is to improve evidence-based care, thus reducing asthma-related emergency department/urgent care visits, admissions, office visits because of acute symptoms, missed school days, missed workdays, and daytime and evening symptoms. The initiative is also designed to build improvement capability and redesign care delivery within primary care practices, thus supporting sustainable systems for future improvement. The PHO approached Anthem Blue Cross and Blue Shield in Ohio (Anthem) in early 2004 to recruit support for an asthma pay-for-performance program. Anthem provides coverage to the highest percentage of the commercially insured population in greater Cincinnati.21 Anthem agreed to fund the pay-for-performance program and allowed the PHO to design the program. Primary objectives of the asthma pay-for-performance program were to reward measurable improvements in asthma care achieved at the network and practice level for the all-payer population, accelerate practice engagement in improvement work, support the business case for quality improvement, obtain experience designing and administering pay-for-performance programs, and influence the design of future pay-for-performance programs initiated by payers and employers. Recognizing the importance of coupling pay for performance with a comprehensive approach to quality improvement, the asthma collaborative included the following strategies: multidisciplinary quality leadership teams at each practice (ie, physician, nurse or medical assistant, and office manager); concurrent data collection at the encounter through use of an asthma decision support tool; all-payer asthma population identification based on chart review confirmation of administrative data obtained from practices, Cincinnati Children's Hospital Medical Center, and Anthem; Web-based asthma registry with real-time reporting capabilities, including network-, practice-, and patient-level data for the process and outcome measures of focus; transparent comparative practice data using tabular, bar chart, and statistical process control formats (discussed at monthly independent practice association board meetings and shared with practices via the PHO Web site and direct mail); practice workflow redesign based on principles of high reliability22 (ie, process redesign to reduce missed opportunities to capture data on, and address, key aspects of care at the patient encounter); patient self-management collaborative; flu shot improvement collaborative; and multiple network meetings and conference calls to promote communication and collaboration among practices. The asthma pay-for-performance program consisted of 3 reward levels (Figure), with practices having the potential to earn a 7% fee schedule increase. Single and all-payer data from the PHO asthma registry were used to calculate network and practice-specific performance, respectively. Because the improvement collaborative has maintained a significant focus on engaging all levels of practice personnel and redesigning practice workflow, we elected to frame the pay-for-performance initiative as a practice reward program; thus, a physician-level incentive was not included. In addition, because of challenges measuring performance at the individual physician level,23 the primary focus of the improvement collaborative has been network- and practice-level performance. Figure. View LargeDownload Asthma pay-for-performance program: conceptual model. Anthem indicates Anthem Blue Cross and Blue Shield in Ohio. First-level reward (pay for participation) The first-level reward (2% fee schedule increase) was designed to recognize practices for committing to the asthma improvement collaborative objectives and for devoting significant time and effort among physicians, nurses, office manager, and other staff. All practices received the first-level reward, regardless of practice-specific performance. Second-level reward (pay for network performance) The second-level reward was a network-level incentive de signed to accelerate practice engagement and promote communication and collaboration among practices. Communication and collaboration were deemed highly important to accelerate the spread of successful interventions across practices, particularly those related to improving reliability. The flu shot percentage and controller medication percentage for children with asthma insured by Anthem were selected as the network-level process measures because of provider relevancy and the ability of Anthem to query claims data and compare results with the PHO asthma registry data. The network flu shot threshold of 30% represented a 36% increase relative to network performance for the 2003-2004 flu season (22%). The network controller medication threshold was set at 70%, the 2004 national average reported by the National Committee for Quality Assurance.24 Although using an all-payer asthma population denominator was preferred, calculating network-level performance based on the Anthem asthma population denominator was deemed important to recruiting payer support for the pay-for-performance program and to documenting value to payer leadership. Failure of the network to meet both thresholds would have resulted in no further fee schedule increase to any practice, regardless of practice-specific performance. Network performance exceeded thresholds for both measures, with all practices receiving another 2% fee schedule increase, once again regardless of practice-specific performance. Third-level reward (pay for improvement capability, pay for practice performance, and pay for population-based improvement) The third level was designed to reward individual practices for outstanding performance for 3 process measures relative to the all-payer asthma population; however, each practice had to first meet designated eligibility criteria. The eligibility criteria were designed to address improvement capability and sustainability within each practice. Practices were required to develop their asthma registry by conducting chart reviews to confirm the asthma diagnosis and active status of patients identified via administrative data obtained from the practice, Cincinnati Children's Hospital Medical Center, and Anthem. Practices also had to incorporate concurrent data collection into workflow with a high level of reliability by capturing designated process and outcome data, per the standardized decision support tool, on at least 85% of the confirmed all-payer asthma population within 15 months from project inception. Requiring practices to achieve the 85% threshold was essential to ensuring that adequate data were available to measure clinical performance relative to the all-payer asthma population denominator. The third level included 3 process measures with high thresholds: flu shot percentage (50% threshold), controller medication percentage (75% threshold), and written self-management plan percentage (80% threshold). If eligibility criteria were met, practices earned a 1% fee schedule increase for each measure for which the threshold was achieved. Network- and practice-specific performance were assessed as of December 31, 2004, with the fee schedule increases effective from May 1, 2004, through December 31, 2005, for the first-level reward, and from March 1, 2005, through December 31, 2005, for the second- and third-level rewards. The fee schedule increases applied to all services billed across all Anthem-covered lives (under both fully insured and self-insured products) receiving care at these practices (ie, not limited to patients with asthma or to asthma-related services). Although the asthma pay-for-performance program concluded on December 31, 2005, Anthem subsequently initiated a communitywide pediatric pay-for-performance program in early 2006, rewarding practice performance on 5 measures, 2 of which were asthma related (ie, flu shot percentage and controller medication percentage). Results The distribution of rewards earned by 43 practices according to fee schedule increase was as follows: those with a fee schedule increase of 2%, 3 (7%); an increase of 4%, 13 (30%); an increase of 5%, 2 (5%); an increase of 6%, 14 (33%); and an increase of 7%, 11 (26%) (percentages do not total 100 because of rounding). One practice was ineligible for the asthma pay-for-performance program because of a separate contractual relationship with Anthem. The 3 practices with a 2% fee schedule increase were deemed ineligible for further rewards because of failure to meet independent practice association board–designated requirements regarding level of participation in the asthma improvement initiative. Regarding the second-level reward, network performance exceeded thresholds for both measures: 54% for the flu shot measure (30% target) and 90% for the controller medication measure (70% target). Among the 40 practices considered for the third-level reward, 40 (100%) completed requirements relative to establishing an asthma registry and 27 (68%) captured the key process and outcome data on at least 85% of the all-payer asthma population. Among the 27 practices meeting both eligibility criteria for the third-level reward, 26 (96%) achieved the 75% threshold for the controller medication measure, 19 (70%) achieved the 80% threshold for the written self-management plan measure, and 18 (67%) achieved the 50% threshold for the flu shot measure. Although the asthma pay-for-performance program concluded on December 31, 2005, the improvement collaborative has continued. The following results reflect network and practice performance relative to the process measures of focus: between October 1, 2003, and November 30, 2006, the cumulative percentage of the network all-payer asthma population receiving “perfect care” increased from 4% to 88%, with 18 of 44 practices (41%) achieving a perfect care percentage of 95% or greater (perfect care is a composite measure reflecting patients with severity classified based on the National Heart, Lung, and Blood Institute guideline criteria,25 a written self-management plan, and controller medications [if classified with persistent asthma]); and the percentage of the network all-payer asthma population receiving the influenza vaccine increased from 22% at baseline (2003-2004 season [September 1 through March 31]) to 41% for the 2004-2005 season, to 62% for the 2005-2006 season, with 7 of 44 practices (16%) achieving an influenza vaccination percentage of 80% or greater for the 2005-2006 season. Although not described in this article, improvement in outcome measures related to admissions, emergency department visits, urgent care visits, office visits because of acute symptoms, missed school days, missed workdays, and parent or patient confidence in managing the condition has also been documented. Comment Aligning pay for performance with the asthma improvement collaborative has resulted in high perfect care and influenza vaccination percentages for the network all-payer asthma population, a higher level of improvement capability among practices, and substantial progress toward system redesign. Based on this experience, we suggest that the following pay-for-performance design principles are highly effective in supporting provider efforts to improve care. Provider is used to denote a practice, hospital, or other site of care, not a physician. Group is used to denote performance across multiple practices, hospitals, or other sites of care. A physician-level incentive was not included in the pay-for-performance program, nor is it included in the following recommendations. Although the pay-for-performance design principles were applied in the context of process measures, they could also be linked to outcome measures (Table). Table. View LargeDownload Key Pay-for-Performance Program Design Principlesa The first recommendation is to allocate a portion of pay-for-performance funds to reward all providers for committing to, and investing resources toward, improvement efforts, regardless of provider-specific performance. There is concern that pay-for-performance programs tend to reward preexisting high performers26 and may be demotivating to providers with the greatest opportunity to improve, particularly if performance thresholds are viewed as unobtainable. Promoting provider engagement in quality improvement efforts, and sustaining commitment, by awarding a portion of pay-for-performance funds to all providers, regardless of provider-specific performance, may help address these concerns.27 The second recommendation is to reward all providers for achieving group-level performance thresholds (regardless of provider-specific performance) before rewarding provider-specific performance. We found that the group-level incentive had a powerful effect in promoting shared learning and the spread of successful interventions across providers, pushing early adopters to even higher performance levels to increase the likelihood that the group-level thresholds would be achieved, accelerating engagement of providers in the improvement initiative, and maintaining focus on improving care to the aggregate population across practices. Transparency of comparative provider data was particularly helpful in maximizing the impact of the group-level incentive. Use of a group-level incentive does not require that providers be organized as an independent practice association or PHO (eg, a group-level incentive could be created among providers participating in an improvement collaborative or among providers grouped based on geographical boundaries or other factors). Recognizing the variable success of improvement collaboratives,28,29 use of a group-level incentive may be an important factor to consider; the availability of a centralized Web-based registry or regional health information organization to assemble data across provider sites is also essential in this regard. Although extensively used to alter behavior at the physician or single provider site level, we are not aware of any pay-for-performance programs linked to group-level performance. The third recommendation is to calculate group- and provider-level performance using an all-payer population denominator. Linking financial rewards to all-payer population performance is consistent with how providers collect and analyze data for quality improvement and how registries are established for populations with chronic conditions. An all-payer population focus also promotes efforts to address equity differences. Concerns regarding adverse payer mix at the group and provider level would need to be addressed; however, these concerns should not preclude use of the all-payer population denominator. One potential option would be to adjust group- and provider-level targets based on the Medicaid/uninsured payer mix percentage. Although desirable in terms of increasing the reward pool, use of an all-payer population denominator does not require pay-for-performance involvement by multiple payers; support from even a single large commercial or governmental payer can have a powerful effect in driving providers to focus on improving all-payer population-based measures. This approach would likely necessitate use of provider data to calculate performance. A major benefit of using provider data is that the dialogue is more likely to remain focused on improvement work that needs to be accomplished, vs extended debate over the validity of payer claims data30; use of provider data also affords the opportunity to include reward measures that cannot be tracked via payer claims data (eg, written self-management plans and quality-of-life measures). This approach necessitates that providers have access to robust data tracking and reporting systems; providers will also need to closely monitor data quality and be prepared to address concerns raised by payers and employers. The fourth recommendation is to require providers to pursue evidence-based interventions that build improvement capability and sustainability before rewarding provider-specific performance. In addition to rewarding clinical performance, pay-for-performance programs can be leveraged to enhance improvement capability and promote sustainability among providers. This objective was accomplished by requiring practices to meet “eligibility criteria” to qualify for third-level rewards. Using the quality framework of Donabedian,31 identifying the all-payer asthma population and creating an electronic registry represent “structural” aspects; redesigning practice workflow through the use of high-reliability change concepts represents “process” aspects. Although evidence is not yet available, it is reasonable to expect that the long-term return on investment to patients, payers, and employers will be enhanced by incorporating improvement capability and sustainability into the design of pay-for-performance programs. The fifth recommendation is to allocate a portion of pay-for-performance funds to reward outstanding provider-specific performance. Although it may be argued that this portion of pay-for-performance funds will be predominantly earned by preexisting high performers,26 rewarding providers for achieving high quality remains important. Aggressive performance thresholds should be established so that even preexisting high performers are encouraged to further improve care. By aligning design characteristics of the pay-for-performance program with a collaborative focused on improving processes and outcomes of care for a condition-specific population, building improvement capability, and driving system changes at the provider level, we have established a framework for achieving more substantive and sustainable improvement. Overdependence on pay for performance to drive improved quality is likely a suboptimal approach with questionable long-term viability; rather, pay for performance, when coupled with robust approaches to quality improvement, can be a catalyst to accelerate sustainable transformation among providers. Correspondence: Keith E. Mandel, MD, Physician-Hospital Organization, Cincinnati Children's Hospital Medical Center, 3333 Burnet Ave, Mail Location Code 9011, Cincinnati, OH 45229-3039 ([email protected]). Accepted for Publication: January 4, 2007. Author Contributions: Dr Mandel had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Mandel and Kotagal. Acquisition of data: Mandel. Analysis and interpretation of data: Mandel. Drafting of the manuscript: Mandel and Kotagal. Critical revision of the manuscript for important intellectual content: Mandel and Kotagal. Statistical analysis: Mandel. Administrative, technical, and material support: Mandel. Study supervision: Mandel and Kotagal. Financial Disclosure: None reported. Additional Contributions: We would like to acknowledge the primary care practices and leadership of Ohio Valley Primary Care Associates, LLC, and Tri State Child Health Services, Inc, the leadership of Cincinnati Children's Hospital Medical Center, and the PHO quality improvement team members, for all their contributions and support. We also thank Barry Malinowski, MD, and colleagues at Anthem Blue Cross and Blue Shield of Ohio for their support of the pay-for-performance program. References 1. Rosenthal MBFernandopulle RSong HRLandon B Paying for quality: providers' incentives for quality improvement. Health Aff (Millwood) 2004;23 (2) 127- 141Google ScholarCrossref 2. Baker GBCarter B Pay for Performance Issues and Trends: Key Findings From the Med-Vantage 2005 National P4P Survey. San Francisco, Calif Med-Vantage Inc2005; 3. Rosenthal MBLandon BENormand STFrank RGEpstein AM Pay for performance in commercial HMOs. N Engl J Med 2006;355 (18) 1895- 1902PubMedGoogle ScholarCrossref 4. Dudley RA Pay-for-performance research: how to learn what clinicians and policy makers need to know. JAMA 2005;294 (14) 1821- 1823PubMedGoogle ScholarCrossref 5. Dudley RAFrolich ARobinowitz DLTalavera JABroadhead PLuft HS Strategies to Support Quality-Based Purchasing: A Review of the Evidence. Rockville, Md Agency for Healthcare Research and Quality2004;Technical Review 10; AHRQ publication 04-0057 6. Kane RLJohnson PETown RJButler M Economic Incentives for Preventive Care. Rockville, Md Agency for Healthcare Research and Quality2004;Evidence Report/Technology Assessment 101; AHRQ publication 04-E024-2 7. Casalino LGillies RRShortell SM et al. External incentives, information technology, and organized processes to improve health care quality for patients with chronic diseases. JAMA 2003;289 (4) 434- 441PubMedGoogle ScholarCrossref 8. Committee on Redesigning Health Insurance Performance Measures, Payment, and Performance Improvement Programs, Institute of Medicine, Rewarding Provider Performance: Aligning Incentives in Medicare. Washington, DC National Academy Press2006; 9. Sorbero MEDamberg CLShaw R et al. Assessment of Pay for PerformanceOptions for Medicare Physician Services: Final Report. Santa Monica, Calif Rand Health2006; 10. Freed GLUren RL Pay for performance: an overview for pediatrics. J Pediatr 2006;149 (1) 120- 124PubMedGoogle ScholarCrossref 11. Petersen LALeChauncy DWUrech TDaw CSookanan S Does pay for performance improve the quality of health care? Ann Intern Med 2006;145 (4) 265- 272PubMedGoogle ScholarCrossref 12. Chassin MR Does paying for performance improve the quality of health care? Med Care Res Rev 2006;63 (1(suppl)) 122S- 125SPubMedGoogle ScholarCrossref 13. Levin-Scherz JDeVita NTimbie J Impact of pay-for-performance contracts and network registry on diabetes and asthma HEDIS® measures in an integrated delivery network. Med Care Res Rev 2006;63 (1) (suppl)14S- 28SPubMedGoogle ScholarCrossref 14. Town RWholey DRKralewski JDowd B Assessing the influence of incentives on physicians and medical groups. Med Care Res Rev 2004;61 (3(suppl)) 80S- 118SPubMedGoogle ScholarCrossref 15. Wu HWedNishimi RYedKizer KWed Pay for Performance Programs: Guiding Principles and Design Strategies. Washington, DC National Quality Forum2005; 16. Dudley RARosenthal MB Pay for Performance: A Decision Guide for Purchasers. Rockville, Md Agency for Healthcare Research and Quality2006;AHRQ publication 06-0047 17. American Medical Association, Pay for performance/physician reimbursement methodologies. http://www.ama-assn.org/ama/pub/category/print/14416.htmlAccessed July 17, 2006 18. American College of Physicians, Linking Physician Payments to Quality Care. Philadelphia, Pa American College of Physicians2005;Position paper 19. American Healthways Inc, Outcomes-based compensation: pay for performance design principles. http://www.rewardingquality.comAccessed December 29, 2006 20. Porter METeisberg EO Redefining Health Care: Creating Value-Based Competition on Results. Boston, Mass Harvard Business School Press2006; 21. HealthLeaders Research, Ohio and Kentucky Health Plan Data, July 1, 2004 22. Nolan TResar RHaraden CGriffin F Improving the reliability of health care. http://www.ihi.org/IHI/Results/WhitePapers/ImprovingtheReliabilityofHealthCare.htmAccessed December 28, 2006 23. Landon BENormand STBlumenthal DDaley J Physician clinical performance assessment: prospects and barriers. JAMA 2003;290 (9) 1183- 1189PubMedGoogle ScholarCrossref 24. National Committee for Quality Assurance, The state of health care quality: 2005. http://www.ncqa.org/Accessed July 18, 2006 25. National Heart, Lung, and Blood Institute, Guidelines for the diagnosis and management of asthma. http://www.nhlbi.nih.gov/guidelines/asthma/asthgdln.pdfAccessed July 18, 2006 26. Rosenthal MBFrank RGLi ZEpstein AM Early experience with pay for performance: from concept to practice. JAMA 2005;294 (14) 1788- 1793PubMedGoogle ScholarCrossref 27. Birkmeyer NJBirkmeyer JD Strategies for improving surgical quality: should payers reward excellence or effort? N Engl J Med 2006;354 (8) 864- 870PubMedGoogle ScholarCrossref 28. Mittman BS Creating the evidence base for quality improvement collaboratives. Ann Intern Med 2004;140 (11) 897- 901PubMedGoogle ScholarCrossref 29. Homer CJForbes PHorvitz LPeterson LEWypij DHeinrich P Impact of a quality improvement program on care and outcomes for children with asthma. Arch Pediatr Adolesc Med 2005;159 (5) 464- 469PubMedGoogle ScholarCrossref 30. Lee THMeyer GSBrennan TA A middle ground on public accountability. N Engl J Med 2004;350 (23) 2409- 2412PubMedGoogle ScholarCrossref 31. Donabedian A Evaluating the quality of medical care. Milbank Q 2005;83 (4) 691- 729PubMedGoogle ScholarCrossref

2007 Archives of Pediatrics & Adolescent Medicine

doi: 10.1001/archpedi.161.7.655

In the article titled “Off-label Drug Use in Hospitalized Children” by Shah et al, published in the March issue of the Archives (2007;161[3]:282-290), the value given in Table 2 (page 285) in the “Characteristic” column, “Age, %” subheading, “29 d to 1 y” row in the “PHIS No.” column, should have been 99 614. Online versions of this article on the Archives of Pediatrics & Adolescent Medicine Web site were corrected on May 8, 2007.

Porter, Todd R.;Crane, Lori A.;Dickinson, L. Miriam;Gannon, Jason;Drisko, Jodi;DiGuiseppi, Carolyn

2007 Archives of Pediatrics & Adolescent Medicine

doi: 10.1001/archpedi.161.7.656pmid: 17606828

Abstract Objective To describe parent opinions about when typical children can engage in activities unsupervised. Design Telephone survey combined with the Behavioral Risk Factor Surveillance System. Setting Colorado. Participants Nine hundred forty-five households with children aged 1 to 14 years. Main Exposures Family and household characteristics and caregiver behaviors. Main Outcome Measures Mean ages at which the caregiver believes typical children can bathe without an adult present, cross busy streets without holding hands, and bicycle in busy streets unsupervised. Results For bathing, the mean age was 6.6 (range, 2-15) years; mean ages were 1.0 year older among Hispanic white parents (95% confidence interval [CI], 0.5-1.4 years) and 0.8 year younger among parents whose child rode with an impaired driver in the past month (95% CI, −1.5 to −0.1 years). For street crossing, mean age was 9.0 (range, 3-16) years; mean ages were 1.2 years older among Hispanic white parents (95% CI, 0.6-1.8 years), 0.7 year older in single-parent households (95% CI, 0.1-1.3 years), and 0.3 year younger among parents whose child rode with a speeding driver in the past month (95% CI, −0.5 to 0.0 year). For bicycling, mean age was 12.2 (range, 6-21) years; mean ages were 1.5 years younger in households with a risky drinker (95% CI, −2.5 to −0.5 years) and 0.5 year younger among parents whose child rode with a speeding driver in the past month (95% CI, −0.9 to −0.1 year). Conclusions Parent opinions about when adult supervision is unnecessary varied with parent behavior and family and household characteristics. Differential supervision may partially explain reports of lower child injury rates among Hispanic and less educated families. Identification of parent and household factors associated with supervision practices might help pediatricians target counseling about age-appropriate supervision. Parental supervision has been recognized as vital to childhood injury prevention by national organizations devoted to improving the welfare of children. The American Academy of Pediatrics, Safe Kids Worldwide, the National Highway Traffic Safety Administration, and the Consumer Product Safety Commission have created numerous guidelines regarding supervision of children, including those participating in water,1-7 pedestrian,6-8 and pedal-cyclist activities.7,9,10 However, these organizations do not all agree on the appropriate age for a child to be left unsupervised in any of these activities. The American Academy of Pediatrics recommends that children 6 years and younger need adult supervision for crossing streets,7 whereas Safe Kids Worldwide recommends this supervision for children 10 years and younger.8 Many recommendations do not specify an age when supervision is no longer required. The American Academy of Pediatrics recommends that “young children should ride [bicycles] only with adult supervision.”9 Similarly, the Consumer Product Safety Commission recommends against leaving “young children” alone in the bath,1 whereas the American Academy of Pediatrics specifies 5 years as the youngest age for unsupervised bathing.5 The variability and lack of specificity are partly due to the recognition that individual differences affect the child's attainment of the cognitive and motor skills necessary to successfully complete each task and partly because the literature on when children are, on average, able to accomplish various tasks is limited.11 Numerous studies have attempted to define the relationship between supervision and subsequent childhood injury.12-22 Morrongiello et al15 have shown that mothers who reported leaving children unsupervised at younger ages had children who experienced more injuries. Parental supervisory behavior may act as a mediator between child behavior and a hazardous environment.14,17,18,23 How the parent chooses to modify this mixture of one-on-one supervision with environmental safeguards is influenced by the attributes of the parent,14-16,18,21,24-27 the child,15,17,18,21,24,26 and the environment.12,14,20,25 The overall effectiveness of parental supervisory behavior, however, depends on how it mediates the dynamic of the child's behavior and interaction with the environment. The effect of parents' attributes on their beliefs about the appropriate age to leave children unsupervised has not been well characterized. Parents who take risks with their own behavior may similarly allow more risk taking in their children, by freeing them from adult supervision at younger ages. Differences in beliefs about parental supervision may mediate previously reported differences in injury risk by race, ethnicity, and immigration status.28-31 The aims of our study were to describe parent opinions about the ages at which a typical child can be allowed to bathe in a tub or ride a bicycle on a busy street without direct adult supervision or to cross a busy street without holding hands, and to examine how characteristics of the household or parent may influence this opinion. Methods Data collection and sampling We performed a cross-sectional analysis using the Colorado 2004 linked Behavioral Risk Factor Surveillance System (BRFSS)–Child Health Survey (CHS). The Colorado Department of Public Health and Environment (including J.G. and J.D.), in consultation with investigators from the University of Colorado School of Medicine (including L.A.C. and C.D.) and local and state agencies, constructed the 124-item CHS as an add-on to the Colorado 2004 BRFSS. The BRFSS collects annual prevalence data on adult risk behaviors and preventive health practices using random-digit dialing and computer-assisted telephone interviewing.32 The goal of the CHS was to collect similar data on children aged 1 to 14 years. Using standard BRFSS sampling methods,32 a representative sample of Colorado households with telephones was selected for the initial BRFSS. The response rate (completed interviews from among eligible households in the sample) for the Colorado 2004 BRFSS, computed in accordance with standards defined by the Council of American Survey Research Organizations,33 was 62.7%.34 Among all states, the median 2004 BRFSS response rate was 52.7% (range, 32.2%-66.6%).34 Households completing the BRFSS telephone questionnaire that reported children aged 1 to 14 years living in the home were then asked to complete the additional CHS telephone survey 1 week later. When more than 1 eligible child resided in the house, 1 child was randomly selected as the survey subject. Whenever possible, the adult caregiver with the most knowledge of the child was selected as the respondent. The Council of American Survey Research Organizations response rate for the CHS was 76.8%. The overall response rate for the BRFSS-CHS was therefore 48.2%. We excluded households in which the respondent declined to link BRFSS and CHS data. Of 997 CHS surveys completed in 2004, 980 respondents (98.3%) agreed to data linkage. Dependent (outcome) variables The 3 primary outcomes measured by the CHS were the ages at which the caregiver believed that a typical child is able to bathe without an adult in the room, to ride a bicycle in a busy street without an adult, and to cross a busy street without holding hands (Table 1). Questions on bathtub and pedestrian supervision were taken from the unpublished Injury Control and Risk Survey 2, phase 2 (Karin Mack, PhD, National Center for Injury Prevention and Control, Centers for Disease Control and Prevention; written communication; May 1, 2006). These questions have not been validated against observed behavior, but their face validity is supported by similarities to the validated supervision instrument devised by Morrongiello and Corbett.35 Table 1. View LargeDownload Child Supervision Questions, Colorado Child Health Survey, 2004 Independent variables Numerous factors that may be associated with parent opinion about the study outcomes have been identified,12,14-16,18,20,21,24-27 and those characteristics included in the linked data set were considered for analysis. We chose caregiver risky and protective behaviors to model themes of conscientiousness and locus of control devised by Morrongiello and colleagues.18,35 These included whether the child rode in a car in the past 30 days with an adult or friend who had just consumed at least 1 alcoholic beverage or used marijuana or another illicit drug; whether the child always used a motor vehicle occupant restraint; whether a firearm was kept in the house; and the frequency with which the child rode in a car with someone who drove at least 10 miles per hour above the speed limit within the past 30 days, categorized as occurring 0, 1 to 4, or 5 or more times and treated as linear for these 3 categories. We also assessed risky drinking (>2 drinks per day for men and >1 drink per day for women) in the past 30 days, and any history of suicidal ideation (ie, seriously thought about trying to hurt oneself in a way that might have resulted in death) in the past year in the BRFSS respondent (not necessarily the caregiver). Parent- and self-reported safety behaviors agree moderately well or better with observed behavior, demonstrating a modest social desirability bias.36-39 Self-reported alcohol use in the BRFSS is highly correlated with per capita alcohol sales.40 The BRFSS questions on suicidal ideation and gun ownership have high test-retest reliability.41 Household and family characteristics included the child's race and ethnicity, number of children younger than 18 years at home, single vs multiple caregivers in the household, the caregiver's relation to the child, annual household income, urban vs rural residence, the child's insurance status, and the highest educational level obtained by any adult in the household. Missing data Of 997 completed surveys, 813 (81.5%) had complete data for every variable and an additional 108 (10.8%) were missing data for only 1 study variable. Cases missing responses for questions on unsupervised bathing (1.2% of completed surveys), crossing a busy street (1.4% of completed surveys), and cycling on a busy street (1.4% of completed surveys) were excluded from the relevant analyses. For independent variables in which less than 1% of the total sample had missing values, responses were assumed to be missing at random; these cases (1.0%) were dropped.42 We eliminated 25 cases (2.5%) in which responses to questions on income, educational level, and race/ethnicity were all missing. All other cases without analyzable responses for race/ethnicity (1.5%) were coded as “other/unspecified race.” For all other variables, we imputed the value of missing responses as the median or the modal category.43 These variables included annual household income (4.4% “don't know”; 1.4% “refused”), insurance coverage (0.4% “don't know”), the child riding with a speeding driver in the past 30 days (3.2% “don't know”; 0.2% “refused”), and keeping a firearm in the home (0.4% “don't know”; 1.0% “refused”). Our final data set included 945 CHS surveys (94.8% of all completed CHS surveys). This data set was similar to the 2004 Colorado population44 in terms of race/ethnicity of children aged 1 to 14 years (eg, 67.0% of respondents vs 65.7% in the Colorado population being non-Hispanic white), annual household income (29.5% vs 31.3% earning ≥$75 000; 19.4% vs 20.5% earning <$25 000), having at least a high school degree (92.9% vs 88.7%), and urban residence (81.4% vs 84.5%). Data analysis Weighted Colorado population estimates and means were calculated for all independent variables. For each outcome, the weighted percentage was calculated for the full sample of 945 respondents. Outcome means and medians were calculated after excluding parent responses of “never” (eg, child should never be allowed to bathe without an adult present). For crude and adjusted analyses, we assigned all “never” responses to 21 years of age (indicating adulthood). Because of the inconsistency and lack of specificity in national recommendations for the ages at which children can be free from supervision, we were unable to define specific ages to differentiate adequate vs inadequate supervision. Moreover, categorizing parent responses using a cut point could imply that extreme age values are adequate even if inappropriate (eg, that a typical 15-year-old should bathe with an adult in the room). Instead, we constructed separate multiple linear regression models for each of the 3 primary outcomes, accounting for survey weights and strata, to calculate the mean ages at which surveyed parents indicated that typical children could be free of supervision. For each outcome, we first analyzed its association with each household, family, and behavioral characteristic. Independent variables related to any of the 3 outcomes at P < .15 were retained, and the rest were omitted from further analysis. We then modeled each dependent outcome with these initially retained variables, using a backward elimination approach, removing the variable with the highest P value at each step, examining coefficients to ensure any change in the coefficients was less than 20%, and retaining in the final regression model those variables with P<.15. For each variable, estimated mean differences from the age specified as the reference value, with 95% confidence intervals, are reported. Human subjects protection Telephone consent was obtained at the time of each call. The institutional review board of the Colorado Department of Public Health and Environment approved the BRFSS and CHS instruments, and the University of Colorado Multiple Institutional Review Board approved the data analysis plan. Results Table 2 shows unweighted frequencies from the linked BRFSS-CHS and weighted Colorado population estimates for each independent variable. The majority of children were non-Hispanic white, insured, and lived with both parents; about half of households surveyed had an income of $50000 or more per year and included an adult with at least a college degree. Table 2. View LargeDownload Characteristics of Study Households, Colorado Population, 2004 Bathing without an adult in the room The mean and median ages reported by parents to be appropriate for a child to bathe without an adult in the room were 6.7 and 6.0 years, respectively (range, 2-15 years). Crude mean ages, overall P values, and adjusted mean age differences (in years) with 95% confidence intervals are presented in Table 3. Parents of children who had ridden with an impaired driver in the past 30 days, on average indicated that a child could be allowed to bathe alone at an age 9.6 months younger than parents of children who had not. Younger ages for bathing alone were also associated with other risky behaviors in crude analyses, but none were retained in the adjusted model. An older age for allowing a child to bathe alone was reported by Hispanic compared with non-Hispanic white parents and by parents in households in which the highest educational level was less than a college degree compared with those with a college degree or higher; differences for Hispanic white parents and those with some college or less than a high school degree were statistically significant. A younger age for allowing a child to bathe alone was reported by black compared with white parents, by parents with annual household incomes of less than $75 000 compared with those with higher incomes, and by those whose child lacked insurance coverage vs those whose child did not, although the differences were modest and, in most cases, may have been due to chance. Table 3. View LargeDownload Association of Parent and Household Characteristics With Parent Opinions About the Appropriate Ages at Which Adult Supervision Is Unnecessary Crossing a busy street without holding hands Parents reported that a typical child could safely cross a busy street without holding hands at the mean and median ages of 9.0 (range, 3-16) years. The more often the child had ridden with a speeding driver in the past 30 days, the younger the age at which parents believed a child could be allowed to cross a busy street without holding hands, with a reduction of 3.6 months in age with each categorical increase in frequency (Table 3); this difference was marginally significant. Compared with non-Hispanic white parents, parents in all other racial/ethnic groups believed a child should be older (by 0.7-1.2 years) before being allowed to cross the street without holding hands, as did parents of children living in single- vs multiple-caregiver households. Older ages were also reported by parents whose children did not always use passenger restraints in the car compared with those who did, and rural vs urban parents, although these differences may have been due to chance. Riding a bicycle on a busy street without an adult The mean and median ages reported by parents to be appropriate for a child to bicycle on a busy street without an adult were 12.2 and 12.0 (range, 6-21) years, respectively. Caregivers in households with a risky drinker reported an age that was 1.5 years younger than caregivers in other households. As with pedestrian activity, the more often the child had ridden with a speeding driver in the past 30 days, the younger the age at which parents believed a typical child should be allowed to bicycle on a busy street without an adult, with a reduction of 6 months in age with each categorical increase in frequency. Caregivers with an annual household income of less than $25 000 reported an age 1.4 years older than that reported by caregivers with incomes of $75 000 or higher (Table 3). Parent opinions about bicycling on a busy street unsupervised were not associated with race/ethnicity or educational level. Comment We found a wide distribution in the ages at which caregivers report that a typical child should be allowed to undertake certain activities unsupervised. Parent opinions about appropriate ages were associated with caregiver behaviors and family and household characteristics. A wide range of risky behaviors were associated with younger crude mean ages for unsupervised activity. In contrast, Hispanic ethnicity, low levels of income and household education, living in a single caregiver household, lack of insurance coverage, and rural residence were all associated with higher crude mean ages for freedom from supervision. Several of these relationships changed in adjusted analyses, reflecting correlations among the risky behaviors and the household and family characteristics. Applying our results for family and household characteristics to previous studies may shed light on previously observed, unexplained differences in pediatric unintentional injury rates. Nationally representative samples have shown lower unintentional injury rates in Hispanic compared with non-Hispanic white children that were not explained by health insurance status or poverty28,30,31 and lower injury risk in children of low-income immigrant compared with US-born mothers that was not explained by health care accessibility, parenting style, or assistance with parenting.29 Overpeck and colleagues30 also found that lack of health insurance and poverty did not explain lower unintentional child injury rates in households having less than a high school education. Perhaps the older suggested ages for supervision reported in our study by Hispanic parents and parents in households with low educational levels indicate differences in the culture of these household environments that could partly explain the observed lower unintentional injury rates, especially given that the home environment is the most common location of unintentional injuries to children.28,31,45-48 Schwebel et al29 suggested that interventions targeted toward at-risk children might be informed by identification of protective factors exhibited by lower-risk families, such as Hispanic or immigrant families. Differences by ethnicity and educational level could also reflect differences in real or perceived dangers of the neighborhoods in which the families reside, eg, crime or traffic danger, such that supervision until an older age is appropriate. Morrongiello and colleagues17,18 conducted observational studies and surveys to elucidate the relationship between supervision and childhood injury. Although their findings may not be generalizable to all populations, the studies have shown that attributes of both the parent and child influence the extent of supervision reported, with more conscientious caregivers likelier to keep the child in view, and mothers of 4- to 5-year-olds less likely to supervise than mothers of 2- to 3-year-olds. In our study, caregivers who reported risky behaviors also reported lower ages for freedom from supervision, consistent with the validated construct of conscientiousness (or lack of it) and locus of control (fatalism) devised by Morrongiello and Corbett.35 If direct supervision is a protective factor in unintentional pediatric injury, then parents who display risky behaviors (lack of conscientiousness) would be expected to have children with higher rates of unintentional injuries.14-16 Unfortunately, we were not able to measure unintentional injury rates in our study sample. The association of younger reported ages for allowing unsupervised activities among parents who report risky behavior may also be explained by a lack of awareness or understanding by the caregivers of what is safe and appropriate behavior for themselves and their children. The wide range of reported ages for unsupervised activities suggests substantial caregiver uncertainty. This uncertainty is not surprising given the lack of specificity and agreement among national organizations regarding what ages are appropriate and safe for unsupervised pedestrian, bicycling, and bathing activities.1-11 Under these circumstances, considerable variation in caregiver understanding of when children can safely participate in various activities alone is expected. Our study was limited by the use of parent-reported opinions about a hypothetical situation for a typical child, as opposed to their actual behavior in real-life situations. However, similarities of our variables to the constructs of Morrongiello et al,17 who produced more accurate supervision assessments, support the validity of our results. In addition, our study used a population-based sample that collected broader sociodemographic information in the context of parent opinion on supervision for these activities. A second limitation is the possibility of selection bias given the combined response rate for the BRFSS-CHS of 48.2% and the elimination of additional cases with missing data. Despite these issues, our final study sample was similar in race/ethnicity, annual income, educational level, and urban residence to 2004 statewide census data, suggesting that it was a reasonable representation of the state population. Our results may also have been biased by residual confounding from unmeasured covariates, although we were able to examine and control for a variety of sociodemographic characteristics in our analyses. However, we lacked data on positive attributes, such as social support or environmental protective factors, that are likely to influence parent opinions about child supervision. Finally, there is the potential for bias from our use of median imputation for selected missing responses. Zhou and colleagues43 have compared median imputation to multiple imputation and found no significant differences in the observed results in samples as large as ours was. However, the median imputation method may reduce standard errors, resulting in smaller P values than with multiple imputation. Conclusions Adequate child supervision is likely to be the most effective defense against many childhood unintentional injuries. However, the wide range of appropriate ages for supervision reported in our study reflects the variability in supervision recommendations among organizations that parents trust as sources of information, which may be due at least in part to difficulty in defining what constitutes supervisory neglect.22 A lack of self-efficacy among child health care providers on how to educate parents on appropriate supervision practices may also affect parent reports of appropriate ages for supervision. Further research should apply validated supervision measures35 to population-based studies to gain insight into how attributes of the individual parent and household environment, as well as those of clinicians, influence supervision practices and ultimately unintentional injury risk. Research should also clarify the ages at which a child is developmentally capable of successfully participating in activities unsupervised that carry a high potential for moderate to severe unintentional injury. In the meantime, clinicians should highlight the importance of appropriate supervision by assessing the parent's understanding of the appropriate ages for freedom from supervision, the safety of the physical and social environment to which the child is exposed, and the value placed on direct supervision vs environmental buffers (such as safety gates and pool fences) and counseling parents appropriately in response to this assessment. Back to top Article Information Correspondence: Carolyn DiGuiseppi, MD, MPH, PhD, Department of Preventive Medicine and Biometrics, University of Colorado School of Medicine, 4200 E 9th Ave, Campus Box B119, Denver, CO 80262 ([email protected]). Accepted for Publication: January 18, 2007. Author Contributions: Dr Porter had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Porter and DiGuiseppi. Acquisition of data: Drisko. Analysis and interpretation of data: Porter, Crane, Dickinson, Gannon, and DiGuiseppi. Drafting of the manuscript: Porter and DiGuiseppi. Critical revision of the manuscript for important intellectual content: Porter, Crane, Dickinson, Gannon, Drisko, and DiGuiseppi. Obtained funding: Drisko and DiGuiseppi. Administrative, technical, and material support: Gannon and Drisko. Study supervision: DiGuiseppi. Financial Disclosure: None reported. Funding/Support: This study was supported by grant R49-CCR811509 from the Centers for Disease Control and Prevention and grant D33HP02610 for Preventive Medicine Residencies from the Health Resources and Services Administration (Dr Porter). Disclaimer: The contents are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention or the Health Resources and Services Administration. References 1. US Consumer Product Safety Commission, Consumer product safety alert: prevent child in-home drowning deaths. CPSC Document 5013 http://www.cpsc.gov/cpscpub/pubs/drown.htmlAccessed October 19, 2005Google Scholar 2. US Consumer Product Safety Commission, Backyard pool: always supervise children, safety commission warns. CPSC Document 5097 http://www.cpsc.gov/cpscpub/pubs/5097.htmlAccessed October 19, 2005Google Scholar 3. Safe Kids Worldwide, Preventing accidental injury: safety tips: water and drowning safety. http://www.safekids.org/tips/tips_water.htmlAccessed October 19, 2005 4. Brenner RA Prevention of drowning in infants, children, and adolescents. Pediatrics 2003;112 (2) 440- 445PubMedGoogle ScholarCrossref 5. American Academy of Pediatrics Committee on Injury, Violence, and Poison Prevention, Prevention of drowning in infants, children, and adolescents. Pediatrics 2003;112 (2) 437- 439PubMedGoogle ScholarCrossref 6. American Academy of Pediatrics Committee on Injury and Poison Prevention, Injury Prevention and Control for Children and Youth. 3rd ed. Elk Grove Village, IL American Academy of Pediatrics1997; 7. American Academy of Pediatrics, The Injury Prevention Program: age-related safety sheets, 6 years old. http://www.aap.org/family/tippmain.htmAccessed April 7, 2007 8. Safe Kids Worldwide, Preventing accidental injury: safety tips: pedestrian safety. http://www.safekids.org/tips/tips_ped.htmAccessed October 19, 2005 9. Safe Kids Worldwide, Preventing accidental injury: safety tips: bike safety. http://www.safekids.org/tips/tips_bike.htmAccessed October 19, 2005 10. American Academy of Pediatrics, The Injury Prevention Program: safe bicycling starts early. http://www.aap.org/family/bicycle.htmAccessed August 24, 2005 11. Centers for Disease Control and Prevention, National Center for Injury Prevention and Control CDC Injury Research Agenda. Atlanta, GA Centers for Disease Control and Prevention2002; 12. Winn DGAgran PFCastillo DN Pedestrian injuries to children younger than 5 years of age. Pediatrics 1991;88 (4) 776- 782PubMedGoogle Scholar 13. Wills KEChristoffel KKLavigne JV et al. Kids ’N’ Cars Research Team, Patterns and correlates of supervision in child pedestrian injury. J Pediatr Psychol 1997;22 (1) 89- 104PubMedGoogle ScholarCrossref 14. Morrongiello BAOndejko LLittlejohn A Understanding toddlers' in-home injuries, II: examining parental strategies, and their efficacy, for managing child injury risk. J Pediatr Psychol 2004;29 (6) 433- 446PubMedGoogle ScholarCrossref 15. Morrongiello BAOndejko LLittlejohn A Understanding toddlers' in-home injuries, I: context, correlates, and determinants. J Pediatr Psychol 2004;29 (6) 415- 431PubMedGoogle ScholarCrossref 16. Morrongiello BAHouse K Measuring parent attributes and supervision behaviors relevant to child injury risk: examining the usefulness of questionnaire measures. Inj Prev 2004;10 (2) 114- 118PubMedGoogle ScholarCrossref 17. Morrongiello BACorbett MMcCourt MJohnston N Understanding unintentional injury-risk in young children, I: the nature and scope of caregiver supervision of children at home. J Pediatr Psychol 2006;31 (6) 529- 539PubMedGoogle ScholarCrossref 18. Morrongiello BACorbett MMcCourt MJohnston N Understanding unintentional injury risk in young children, II: the contribution of caregiver supervision, child attributes, and parent attributes. J Pediatr Psychol 2006;31 (6) 540- 551PubMedGoogle ScholarCrossref 19. Landen MGBauer UKohn M Inadequate supervision as a cause of injury deaths among young children in Alaska and Louisiana. Pediatrics 2003;111 (2) 328- 331PubMedGoogle ScholarCrossref 20. Gärling AGärling T Mothers' supervision and perception of young children's risk of unintentional injury in the home. J Pediatr Psychol 1993;18 (1) 105- 114PubMedGoogle ScholarCrossref 21. Dal Santo JAGoodman RMGlik DJackson K Childhood unintentional injuries: factors predicting injury risk among preschoolers. J Pediatr Psychol 2004;29 (4) 273- 283PubMedGoogle ScholarCrossref 22. Coohey C Defining and classifying supervisory neglect. Child Maltreat 2003;8 (2) 145- 156PubMedGoogle ScholarCrossref 23. Saluja GBrenner RMorrongiello BAHaynie DRivera MCheng TL The role of supervision in child injury risk: definition, conceptual and measurement issues. Inj Control Saf Promot 2004;11 (1) 17- 22PubMedGoogle ScholarCrossref 24. Schwebel DCBounds ML The role of parents and temperament on children's estimation of physical ability: links to unintentional injury prevention. J Pediatr Psychol 2003;28 (7) 505- 516PubMedGoogle ScholarCrossref 25. Rivara FPBergman ABDrake C Parental attitudes and practices toward children as pedestrians. Pediatrics 1989;84 (6) 1017- 1021PubMedGoogle Scholar 26. Dunne RGAsher KNRivara FP Behavior and parental expectations of child pedestrians. Pediatrics 1992;89 (3) 486- 490PubMedGoogle Scholar 27. Coohey C Home alone and other inadequately supervised children. Child Welfare 1998;77 (3) 291- 310PubMedGoogle Scholar 28. Simon TDBublitz CHambidge SJ External causes of pediatric injury-related emergency department visits in the United States. Acad Emerg Med 2004;11 (10) 1042- 1048PubMedGoogle ScholarCrossref 29. Schwebel DCBrezausek CMRamey CTRamey SL Injury risk among children of low-income US-born and immigrant mothers. Health Psychol 2005;24 (5) 501- 507PubMedGoogle ScholarCrossref 30. Overpeck MDJones DHTrumble ACScheidt PCBijur PE Socioeconomic and racial/ethnic factors affecting non-fatal medically attended injury rates in US children. Inj Prev 1997;3 (4) 272- 276PubMedGoogle ScholarCrossref 31. Hambidge SJDavidson AJGonzales RSteiner JF Epidemiology of pediatric injury-related primary care office visits in the United States. Pediatrics 2002;109 (4) 559- 565PubMedGoogle ScholarCrossref 32. Behavioral Risk Factor Surveillance System: operational and user's guide, version 3.0. March4 2005;http://www.cdc.gov/brfss/pdf/userguide.pdfAccessed November 10, 2005 33. Council of American Survey Research Organizations, Home page. http://www.casro.orgAccessed April 26, 2006 34. National Center for Chronic Disease Prevention and Health Promotion, Behavioral Risk Factor Surveillance System summary data quality report. June15 2005;http://ftp.cdc.gov/pub/Data/Brfss/2004SummaryDataQualityReport.pdfAccessed January 15, 2007 35. Morrongiello BACorbett M The Parent Supervision Attributes Profile Questionnaire: a measure of supervision relevant to children's risk of unintentional injury. Inj Prev 2006;12 (1) 19- 23PubMedGoogle ScholarCrossref 36. Robertson ASRivara FPEbel BELymp JFChristakis DA Validation of parent self reported home safety practices. Inj Prev 2005;11 (4) 209- 212PubMedGoogle ScholarCrossref 37. Nelson DE Validity of self reported data on injury prevention behavior: lessons from observational and self reported surveys of safety belt use in the US. Inj Prev 1996;2 (1) 67- 69PubMedGoogle ScholarCrossref 38. Hatfield PMStaresinic AGSorkness CAPeterson NMSchirmer JKatcher ML Validating self reported home safety practices in a culturally diverse non–inner city population. Inj Prev 2006;12 (1) 52- 57PubMedGoogle ScholarCrossref 39. Centers for Disease Control, Comparison of observed and self-reported seat belt use rates: United States. MMWR Morb Mortal Wkly Rep 1988;37 (36) 349- 351PubMedGoogle Scholar 40. Smith PFRemington PLWilliamson DFAnda RF A comparison of alcohol sales data with survey data on self-reported alcohol use in 21 states. Am J Public Health 1990;80 (3) 309- 312PubMedGoogle ScholarCrossref 41. Koziol-McLain JBrand DMorgan DLeff MLowenstein SR Measuring injury risk factors: question reliability in a statewide sample. Inj Prev 2000;6 (2) 148- 150PubMedGoogle ScholarCrossref 42. Howell DC Treatment of missing data. December23 2002;http://www.uvm.edu/~dhowell/StatPages/More_Stuff/Missing_Data/Missing.htmlAccessed May 10, 2005 43. Zhou XHEckert GJTierney WM Multiple imputation in public health research. Stat Med 2001;20 (9-10) 1541- 1549PubMedGoogle ScholarCrossref 44. Colorado Department of Public Health and Environment, 2004 Colorado population and statistics. http://www.cdphe.state.co.us/hs/vs/Accessed November 17, 2006 45. Runyan CWPerkis DMarshall SW et al. Unintentional injuries in the home in the United States, II: morbidity. Am J Prev Med 2005;28 (1) 80- 87PubMedGoogle ScholarCrossref 46. Runyan CWCasteel CPerkis D et al. Unintentional injuries in the home in the United States, I: mortality. Am J Prev Med 2005;28 (1) 73- 79PubMedGoogle ScholarCrossref 47. Phelan KJKhoury JKalkwarf HLanphear B Residential injuries in U.S. children and adolescents. Public Health Rep 2005;120 (1) 63- 70PubMedGoogle Scholar 48. Nagaraja JMenkedick JPhelan KJAshley PZhang XLanphear BP Deaths from residential injuries in US children and adolescents, 1985-1997. Pediatrics 2005;116 (2) 454- 461PubMedGoogle ScholarCrossref

Blume, Heidi K.;Loch, Christian M.;Li, Christopher I.

2007 Archives of Pediatrics and Adolescent Medicine

doi: 10.1001/archpedi.161.7.663pmid: 17606829

Abstract Objective To investigate the association between maternal socioeconomic status and the risk of encephalopathy in full-term newborns. Design Population-based case-control study. Setting Washington State births from 1994 through 2002 recorded in the linked Washington State Birth Registry and Comprehensive Hospital Abstract Reporting System. Participants Cases (n = 1060) were singleton full-term newborns with Comprehensive Hospital Abstract Reporting System International Classification of Diseases, Ninth Revision diagnoses of seizures, birth asphyxia, central nervous system dysfunction, or cerebral irritability. Control cases (n = 5330) were singleton full-term newborns selected from the same database. Main Exposures Socioeconomic status was defined by median income of the census tract of the mother's residence, number of years of maternal educational achievement, or maternal insurance status. Main Outcome Measures Odds ratios estimating the risk of encephalopathy associated with disadvantaged socioeconomic status were calculated in 3 separate analyses using multivariate adjusted logistic regression. Results Newborns of mothers living in neighborhoods in which residents have a low median income were at increased risk of encephalopathy compared with newborns in neighborhoods in which residents have a median income more than 3 times the poverty level (adjusted odds ratio, 1.9; 95% confidence interval, 1.5-2.3). There was also a trend for increasing risk of encephalopathy associated with decreasing neighborhood income (P<.001). Newborns of mothers with less than 12 years of educational achievement had a higher risk of encephalopathy compared with newborns of mothers with more than 16 years of educational achievement (adjusted odds ratio, 1.7; 95% confidence interval, 1.3-2.3). Newborns of mothers receiving public insurance also had a higher risk of encephalopathy compared with newborns of mothers who have commercial insurance (adjusted odds ratio, 1.4; 95% confidence interval, 1.2-1.7). Conclusion Disadvantaged socioeconomic status was independently associated with an increased risk of encephalopathy in full-term newborns. These findings suggest that a mother's socioeconomic status may influence the risk of encephalopathy for her full-term newborn. Disadvantaged socioeconomic status (SES) has been associated with many indicators of poor health in children including higher risks of infant mortality,1-3 preterm birth,4,5 and learning disabilities.6 Neonatal encephalopathy is “a clinically defined syndrome of disturbed neurological function in the earliest days of life in the term infant, manifested by difficulty initiating and maintaining respiration, depression of tone and reflexes, subnormal level of consciousness, and (often) seizures.”7(p1325) One to 4 newborns per 1000 live births have symptoms consistent with neonatal encephalopathy,8 and most newborns with moderate to severe neonatal encephalopathy have profound developmental disabilities or die in the first year of life.9 Given the association of disadvantaged SES with neonatal morbidity and mortality and with other forms of neurological disease in adults and children,1,3,10 it might be expected that full-term newborns of mothers with low SES could also be at higher risk of encephalopathy in the perinatal period. However, to our knowledge, there have been few studies in this area. One Australian study found that full-term newborns of mothers with low SES had an approximately 3-fold increased risk of neonatal encephalopathy compared with newborns of mothers with high SES.11 This study used maternal employment and health insurance status to determine SES. Given the limited range of these variables, it was difficult to determine whether there was a trend in the risk of neonatal encephalopathy associated with SES or if the association existed only for certain occupations. The authors also cautioned that the mechanism of action of their measures of SES on the risk of neonatal encephalopathy could vary in different populations and required further investigation. The relationship between SES and cerebral palsy, one potential outcome of neonatal encephalopathy, has been studied more than the relationship between SES and neonatal encephalopathy. However, this relationship is complicated because preterm infants are at higher risk of cerebral palsy than full-term infants and low SES is associated with preterm birth. In addition, cerebral palsy is diagnosed in the postnatal period and, therefore, may be caused or exacerbated by postnatal factors. Studies of the association between cerebral palsy and SES have yielded conflicting results. Several investigators have found an association between SES and cerebral palsy,12,13 but others have not found a significant association.14,15 Recent studies of children with cerebral palsy suggest that the association between low SES and congenital cerebral palsy may be stronger in children of normal birth weight than in children of low birth weight.12,13,15 Given that disadvantaged SES has been associated with premature birth and infant mortality16 and that factors associated with the risk of premature birth such as inflammatory states, maternal infection, or poor prenatal care have also been associated with neonatal encephalopathy,17-19 we hypothesized that disadvantaged SES could also be associated with encephalopathy in full-term newborns. We examined the relationship between SES as measured by median neighborhood income of the mother's residence at the time of childbirth, maternal educational achievement, or insurance status, and the risk of encephalopathy in full-term newborns in Washington State from January 1994 through December 2002. Methods Data source We conducted a population-based case-control study linking information from the Washington State Birth Registry to the Comprehensive Hospital Abstract Reporting System (CHARS) for births that occurred in Washington State from January 1994 through December 2002. The Birth Registry contains information recorded on the birth certificate for every birth in Washington State. CHARS is a database created by the Washington State Department of Health that includes ICD-9International Classification of Diseases, Ninth Revision (ICD-9) discharge diagnosis codes and other administrative information for all hospitalizations in nonfederal hospitals in Washington State, including those for both mothers and their newborns. The CHARS records of mothers and newborns are linked to each infant's Washington State birth certificate data using unique identifiers. During the study period, 89% of all singleton births in Washington State were linked to CHARS records (n = 586 118). The information was deidentified before receipt by the authors. Institutional review board approval from the Washington State Department of Health and the University of Washington, Seattle, for use of these data was received before the conduct of the study. Participants Case infants (n = 1060) were singleton full-term (≥37 weeks' gestational age) newborns maintained in the birth hospital or admitted to the hospital within 2 days of birth whose ICD-9 discharge diagnose codes included severe birth asphyxia; birth asphyxia with neurological involvement (768.5), unspecified birth asphyxia in live born infant (768.9), newborn convulsions (779.0), convulsions (780.3), other and unspecified cerebral irritability in the newborn (779.1), or cerebral depression, coma, and other abnormal cerebral signs; and central nervous system dysfunction in newborn, not otherwise specified (779.2) in the CHARS database. We believed that most infants with these diagnoses would have evidence of neurological dysfunction in the perinatal period given the definitions of these diagnosis codes. Control infants (n = 5330) were singleton full-term (≥37 weeks' gestational age) newborns selected from the same database. Five control infants were randomly selected for each infant with encephalopathy and were matched only by birth year. All subjects included in the final analyses had maternal information recorded in the CHARS database. All infants with congenital anomalies (225 cases and 373 controls) or drug withdrawal syndrome (20 cases and 8 controls) were excluded from the analysis. All infants whose gestational age was unknown or recorded as more than 45 weeks were also excluded (24 cases and 73 controls). Definition of ses We used 3 variables to measure SES, including median income of the neighborhood of the mother's residence at the time of the infant's birth, number of years of maternal educational achievement, and maternal health insurance status. We used the median income of the census tract of the mother's residence as listed on the birth certificate to define neighborhood income. We then categorized the median neighborhood income based on the US poverty level for a family of 4 (2 adults and 2 children) for each year from 1994 through 2002. The US poverty level for a family of 4 ranged from an annual income of $15 029 in 1994 to $18 244 in 2002.20 Subjects without a maternal residence listed or those whose residence could not be linked to a census tract were excluded from the final neighborhood income analysis (96 cases and 355 controls were excluded). Maternal educational achievement was defined as the number of years of schooling as reported on the birth certificate and was categorized based on the number of years completed, as follows: less than 12 years, 12 years, 13 to 15 years, or greater than or equal to 16 years. Subjects who did not have maternal educational achievement listed on the birth certificate were excluded from analysis of maternal educational achievement (103 cases and 475 controls were excluded). Maternal insurance status was defined as the primary payer listed in the CHARS database. Maternal insurance was divided into 2 groups: public insurance (primary payer listed as Medicaid or Medicare) and private insurance (primary payer listed as commercial insurance, health maintenance organization, or health care services contractor). Two small groups of mothers with self-pay (14 cases and 92 controls) or other insurance (16 cases and 73 controls) were excluded from the final analyses by insurance status. Statistical analysis Odds ratios were calculated using univariate and multivariate logistic regression. We performed separate analyses to determine the relationship between the 3 measures of SES and the risk of encephalopathy. Confounders were selected before analysis based on the likelihood of confounding given results of our univariate analyses and previous studies11,17,21 that indicated that these variables may have an effect on the risk of encephalopathy and mother's SES. The confounders in our models included maternal age, parity, maternal race/ethnicity, marital status, presence of preeclampsia, and birth year. Maternal race/ethnicity was recorded on the birth certificate and was usually reported by the mother, although hospital staff may also record this information. The original adjusted odds ratios (AORs) were compared with the results obtained when the intermediate variables of timing of prenatal care, exposure to intrapartum fever or chorioamnionitis, smoking, and urban vs rural residence were individually added to the model to determine the effect these intermediate variables might have on the relationship between encephalopathy and low SES. A change of more than 10% was predetermined to indicate a significant alteration of the odds ratio. To evaluate trends in our exposures of interest, we used logistic regression to estimate the change in the risk of encephalopathy associated with 1 unit of change in neighborhood income or educational achievement levels. For neighborhood income, we defined the unit of change as 1 poverty level, the dollar figure assigned to each year to define poverty level (eg, $15 029 for 1994 or $17 960 for 2001) to ensure that each unit was roughly equivalent from year to year. For educational achievement, the unit of change was 1 year of maternal educational achievement. Results The distribution of maternal age, race/ethnicity, marital status, rural residence, and reported smoking were similar between the encephalopathy and control groups. Mothers of case infants were more likely to be nulliparous, to have had later entry into prenatal care, and to have preeclampsia than mothers of controls. Infants with encephalopathy were more likely to be male or have low birth weight than control infants. Infants exposed to intrapartum fever or chorioamnionitis also had a higher risk of encephalopathy than unexposed infants (Table 1). Table 1. View LargeDownload Demographic Factors and Univariate Risks for Encephalopathy Residence in a low-income neighborhood, with median income less than 2 times the poverty level, was associated with a 1.9-fold (95% confidence interval [CI], 1.5-2.3) increased risk of encephalopathy when compared with residence in neighborhoods with a median income more than 3 times the poverty level (Table 2). Using a continuous variable for neighborhood income, we identified a linear trend in the risk of encephalopathy associated with neighborhood income. For each poverty-level unit increase in median neighborhood income, the risk of encephalopathy decreased by 24% (AOR, 0.76; 95% CI, 0.69-0.84; P for trend <.001). Table 2. View LargeDownload Measures of SES and Risk of Encephalopathy The association between neighborhood income and encephalopathy persisted in the subgroups of infants diagnosed with seizures or birth asphyxia and was strongest in the group of infants diagnosed with severe birth asphyxia (Table 2). Addition of timing of prenatal care, smoking, or exposure to intrapartum fever or chorioamnionitis to the model or removal of preeclampsia from the model altered the original AORs by less than 5%, indicating that these intermediate variables were likely not responsible for the association between neighborhood income and risk of encephalopathy. Fewer years of maternal educational achievement were also associated with an increased risk of encephalopathy. Compared with infants of mothers with 16 years of educational achievement or more, risk of encephalopathy was increased 1.3-fold (95% CI, 1.0-1.6) in infants of mothers with 13 to 15 years of educational achievement, 1.6-fold (95% CI, 1.1-1.6) in infants of mothers with 12 years of educational achievement, and 1.7-fold (95% CI, 1.3-2.3) in infants of mothers with less than 12 years of educational achievement (Table 2). We also identified a linear trend, such that higher maternal educational achievement was associated with a decreased risk of encephalopathy. For each additional year of educational achievement, the risk of encephalopathy decreased by 6% (AOR, 0.94; 95% CI, 0.92-0.97; P<.001). This relationship seemed to persist in the groups of infants diagnosed as having seizures or asphyxia but was only significant in infants with seizures whose mothers had 12 years of educational achievement or infants with any asphyxia and mothers with 12 years of educational achievement or less (Table 2). Maternal public health insurance, compared with private insurance, was also associated with an increased risk of encephalopathy (AOR, 1.4; 95% CI, 1.2-1.7), seizures (AOR, 1.3; 95% CI, 1.0-1.6), and asphyxia (AOR, 1.3; 95% CI, 1.0-1.6) (Table 2). There was less than a 5% change in the AORs when the intermediate variables (timing of prenatal care, exposure to intrapartum fever or chorioamnionitis, smoking, or urban vs rural residence) were individually added to the final models for educational achievement and insurance status, indicating that these intermediate variables are likely not responsible for the association between maternal educational achievement or insurance and risk of encephalopathy. In addition to this finding, we did not find evidence of effect modification of the relationship between SES and encephalopathy by parity or maternal age. In a full model including all 3 indicators of SES (neighborhood income, maternal educational achievement, and maternal health insurance) and the preselected confounders, we found that the risk of encephalopathy associated with neighborhood income and maternal insurance decreased by less than 10% (range, 1%-7.5%) with the addition of 1 or both of the other SES factors. This indicates that neighborhood income and maternal insurance status were equally predictive of disease and were relatively independent of each other. However, the odds ratios associated with maternal educational achievement decreased by 23% when both income and maternal insurance were added to the model. Adding educational achievement to the income or maternal insurance models changed these odds ratios by less than 3.5% (range, 0.4%-3.4%). This indicates that the relationship between educational achievement and encephalopathy may be influenced by other factors associated with neighborhood income or insurance status. Comment Disadvantaged SES was associated with an increased risk of encephalopathy in full-term newborns in Washington State. This association was significant whether neighborhood income, maternal educational achievement, or insurance status was used as a measure of SES. Of these 3 measures, low median neighborhood income was most robustly associated with the risk of encephalopathy. However, we found that all of these components of SES are important and that each measure of SES evaluated seemed to have an effect on the risk of encephalopathy in full-term newborns. Socioeconomic status can be difficult to measure or to define consistently and accurately. Several studies of the relationship between SES and encephalopathy or cerebral palsy have used self-reported measures, which are prone to bias, or general measures such as insurance status or maternal employment. We used the report of mother's residence from the birth certificate combined with the median income of the census tract of the mother's residence from census data to generate our neighborhood income variable. These data were less susceptible to reporting bias than self-report of income or other measures. This measure also provides a representation of the neighborhood environment of each participant, which may be relevant in different ways than individual income alone. One limitation to this approach is that we were unable to identify a census tract for some control and case infants, so there may have been some nondifferential bias if the mothers of infants with encephalopathy and missing neighborhood income data were somehow different than the mothers of controls with missing neighborhood income data. Our data for maternal insurance status was also unbiased but provided only 2 broad categories. Maternal educational achievement was determined by self-report on the birth certificate and was, therefore, most prone to reporting bias of the 3 measures used in this study. Another limitation of this study was the use of ICD-9 codes to identify newborns with encephalopathy and our inability to verify these diagnoses by medical record review. We were unable to conduct medical record reviews because all information was deidentified and the infants were hospitalized at many institutions throughout Washington State. Despite these limitations, to our knowledge, this is one of the largest studies to investigate the association between encephalopathy in the term newborn and maternal SES and to provide evidence that disadvantaged antenatal SES is a risk factor for neonatal encephalopathy. The link between SES and infant mortality and between SES and prematurity has been firmly established in many earlier studies.2,3,22 This study provides evidence that low SES is also associated with adverse neonatal neurological outcomes in full-term newborns. We found this association to hold true for 3 different classifications of SES and, in most cases, for diagnoses of any encephalopathy, neonatal seizures, or birth asphyxia. However, the underlying mechanisms that account for these associations remain unclear, although disparities in infant health have existed for many years and many theories have been advanced to explain the relationship between low SES and poor perinatal health. The disparities between neonatal outcomes for the rich and the poor continue despite medical advances in prenatal and perinatal care, and prenatal programs such as the Women, Infants and Children Program and others that were designed to improve maternal and child health by improving access to medical care, maternal nutrition, and support for vulnerable women and their children.23 The association between disadvantaged SES and encephalopathy in full-term newborns in this study could not be attributed to differences in prenatal care, race/ethnicity, urban vs rural residence, exposure to intrapartum fever or chorioamnionitis, or rate of preeclampsia. Other factors such as social stress, perceived disparities, neighborhood violence, and social support systems may influence perinatal health24; however, we were unable to measure the effect of these factors on the risk of encephalopathy. There is also growing evidence that chronic stress may lead to alterations in maternal physiology such as altered vascular responses and stress hormone production.24 While there is little firm evidence that links these changes to premature birth or neonatal outcomes, it is possible that physiological changes induced by chronic stressors could influence neonatal outcomes and may influence the risk of neonatal encephalopathy. Conclusions Indicators of disadvantaged SES such as residence in a low-income neighborhood at the time of birth, lack of advanced maternal educational achievement, or maternal public insurance were associated with increased risk of encephalopathy in full-term newborns in Washington State. This association does not seem to be mediated by timing of prenatal care, race/ethnicity, smoking, exposure to intrapartum fever or chorioamnionitis, or preeclampsia. While the mechanisms behind the relationship between SES and the risk of neonatal encephalopathy remain unclear, these findings suggest that maternal SES may influence the risk of encephalopathy in newborns. Further investigation is necessary to attempt to identify factors such as stress, perceived disparity, differential infection rates, toxic exposures, or other characteristics that may lead to higher risk of encephalopathy in populations with disadvantaged SES. Once identified, these factors may provide targets for interventions to improve neonatal outcomes and reduce the risk of neonatal encephalopathy in vulnerable populations. Correspondence: Heidi K. Blume, MD, MPH, Division of Pediatric Neurology, Children's Hospital and Regional Medical Center, University of Washington, 4800 Sandpoint Way, Mailstop B-5552, Seattle, WA 98105 ([email protected]). Accepted for Publication: January 25, 2007. Author Contributions:Study concept and design: Blume. Acquisition of data: Blume. Analysis and interpretation of data: Blume, Loch, and Li. Drafting of the manuscript: Blume. Critical revision of the manuscript for important intellectual content: Blume, Loch, and Li. Statistical analysis: Blume and Loch. Administrative, technical, and material support: Blume. Study supervision: Li. Financial Disclosure: None reported. References 1. Turrell GMengersen K Socioeconomic status and infant mortality in Australia: a national study of small urban areas, 1985-89. Soc Sci Med 2000;50 (9) 1209- 1225PubMedGoogle ScholarCrossref 2. Bambang SSpencer NJLogan SGill L Cause-specific perinatal death rates, birth weight and deprivation in the West Midlands, 1991-93. Child Care Health Dev 2000;26 (1) 73- 82PubMedGoogle ScholarCrossref 3. Wise PH The anatomy of a disparity in infant mortality. Annu Rev Public Health January 2003;24341- 362PubMedGoogle ScholarCrossref 4. Lumley J The epidemiology of preterm birth. Baillieres Clin Obstet Gynaecol 1993;7 (3) 477- 498PubMedGoogle ScholarCrossref 5. Spencer NJLogan SGill L Trends and social patterning of birthweight in Sheffield, 1985-94. Arch Dis Child Fetal Neonatal Ed 1999;81 (2) F138- F140PubMedGoogle ScholarCrossref 6. Resnick MBGueorguieva RVCarter RL et al. The impact of low birth weight, perinatal conditions, and sociodemographic factors on educational outcome in kindergarten. Pediatrics 1999;104 (6) e74http://pediatrics.aappublications.org/cgi/content/full/104/6/e74Accessed April 16, 2007Google ScholarCrossref 7. Nelson KBLeviton A How much of neonatal encephalopathy is due to birth asphyxia? Am J Dis Child 1991;145 (11) 1325- 1331PubMedGoogle ScholarCrossref 8. American College of Obstetricians and Gynecologists, Task Force on Neonatal Encephalopathy and Cerebral Palsy; American Academy of Pediatrics, Neonatal Encephalopathy and Cerebral Palsy: Defining the Pathogenesis and Pathophysiology. Washington, DC American College of Obstetricians and Gynecologists2003; 9. Volpe J Neurology of the Newborn. 4th ed. Philadelphia, PA WB Saunders Co2001; 10. Qureshi AISuri MFSaad MHopkins LN Educational attainment and risk of stroke and myocardial infarction. Med Sci Monit 2003;9 (11) CR466- CR473PubMedGoogle Scholar 11. Badawi NKurinczuk JJKeogh JM et al. Antepartum risk factors for newborn encephalopathy: the Western Australian case-control study. BMJ 1998;317 (7172) 1549- 1553PubMedGoogle ScholarCrossref 12. Dolk HPattenden SJohnson A Cerebral palsy, low birthweight and socio-economic deprivation: inequalities in a major cause of childhood disability. Paediatr Perinat Epidemiol 2001;15 (4) 359- 363PubMedGoogle ScholarCrossref 13. Dowding VMBarry C Cerebral palsy: social class differences in prevalence in relation to birthweight and severity of disability. J Epidemiol Community Health 1990;44 (3) 191- 195PubMedGoogle ScholarCrossref 14. Nelson KBEllenberg JH Antecedents of cerebral palsy: multivariate analysis of risk. N Engl J Med 1986;315 (2) 81- 86PubMedGoogle ScholarCrossref 15. Sundrum RLogan SWallace ASpencer N Cerebral palsy and socioeconomic status: a retrospective cohort study. Arch Dis Child 2005;90 (1) 15- 18PubMedGoogle ScholarCrossref 16. Rutter DRQuine L Inequalities in pregnancy outcome: a review of psychosocial and behavioural mediators. Soc Sci Med 1990;30 (5) 553- 568PubMedGoogle ScholarCrossref 17. Badawi NKurinczuk JJKeogh JM et al. Intrapartum risk factors for newborn encephalopathy: the Western Australian case-control study. BMJ 1998;317 (7172) 1554- 1558PubMedGoogle ScholarCrossref 18. Petrova ADemissie KRhoads GGSmulian JCMarcella SAnanth CV Association of maternal fever during labor with neonatal and infant morbidity and mortality. Obstet Gynecol 2001;98 (1) 20- 27PubMedGoogle ScholarCrossref 19. Shalak LFPerlman JM Infection markers and early signs of neonatal encephalopathy in the term infant. Ment Retard Dev Disabil Res Rev 2002;8 (1) 14- 19PubMedGoogle ScholarCrossref 20. US Census Bureau,http://www.census.gov/hhes/www/poverty/threshld.htmlAccessed January 15, 2005 21. Wu YWEscobar GJGrether JKCroen LAGreene JDNewman TB Chorioamnionitis and cerebral palsy in term and near-term infants. JAMA 2003;290 (20) 2677- 2684PubMedGoogle ScholarCrossref 22. Wen SWSmith GYang QWalker M Epidemiology of preterm birth and neonatal outcome. Semin Fetal Neonatal Med 2004;9 (6) 429- 435PubMedGoogle ScholarCrossref 23. Wise PH The transformation of child health in the United States. Health Aff (Millwood) 2004;23 (5) 9- 25PubMedGoogle ScholarCrossref 24. Rich-Edwards JWGrizzard TA Psychosocial stress and neuroendocrine mechanisms in preterm delivery. Am J Obstet Gynecol 2005;192 (5(suppl)) S30- S35PubMedGoogle ScholarCrossref

Bradley, Robert H.;Vandell, Deborah Lowe

2007 Archives of Pediatrics and Adolescent Medicine

doi: 10.1001/archpedi.161.7.669pmid: 17606830

Abstract Objective To evaluate studies of child care with specific attention to the impact of age at entry and amount, quality, and type of care on children's adaptive functioning. Data Sources MEDLINE, PsychINFO, and the SAGE Full-Text Collection. Study Selection The review considers correlational and experimental research conducted throughout the world that includes an adequate description of the type of care provided. Main Exposures Amount, quality, and type of child care. Main Outcome Measures Language, cognitive and social competence, achievement, behavioral problems, relationships with parents, communicable illnesses, and asthma. Results Children who began care early in life and were in care 30 or more hours a week were at increased risk for stress-related behavioral problems. Elevated risk was more likely if they had difficulties interacting with peers or had insensitive parents. Children in day care centers had higher language scores and early school achievement, especially if they came from disadvantaged backgrounds and the centers offered high-quality care. Attending arrangements with 6 or more children increased the likelihood of communicable illnesses and ear infections, albeit those illnesses had no long-term adverse consequences. Conclusions Child care is a multidimensional phenomenon. Guidance on when to place a child in nonparental care and what kind of care to use is complicated because of the multiplicity of sometimes offsetting effects on children. Child care experiences interact with experiences at home and the child's own characteristics, and research indicates that the quality of child care matters. In 2002, the US Bureau of the Census estimated that 11 596 000 children younger than 5 years routinely spend a significant amount of time in nonparental care.1 Child care arrangements vary widely in quality, amount, and type.1,2 Consequently, concerned parents and public officials wonder about the impact of child care. Several recent large-scale studies,2-6 in conjunction with almost 4 decades of smaller-scale studies, offer more authoritative answers to key questions about the quality, amount, and type of child care children receive. Providing definitive information about child care is challenging because most research is correlational, with researchers identifying statistical associations rather than causal effects. In only a few studies have researchers used random assignment to study nonparental care. Even those studies had significant limitations in terms of the questions posed or samples used.4,7-10 In the absence of random assignment, researchers have included family and child measures as covariates to minimize selection bias. Only recently have researchers incorporated extensive covariates.11 Although using multiple covariates reduces the likelihood of drawing erroneous conclusions about child care, this strategy can yield conservative estimates of effects.12 Another problem with the child care literature is that less formal and unregulated settings are underrepresented in most studies.2 Their omission has likely resulted in conservative estimates of the effects of child care quality on children's development.11,13 Another challenge is that child care is a complex phenomenon that varies along multiple dimensions (eg, amount, type, and quality), making it difficult to isolate the effects of just 1 dimension. Recent research has attempted to segregate the impact of each dimension by controlling for the other ones, but it is not possible to fully disentangle the dimensions of care yet still provide meaningful estimates of the effects of each dimension. The most comprehensive study about this topic is the Study of Early Child Care (SECC), which is funded by the National Institute of Child Health and Human Development (NICHD).14,15 This prospective, longitudinal study includes detailed assessments of child care (amount and timing, quality, and types of care settings) for more than 1200 children who have been studied from birth through middle childhood. Children initially resided near 10 research sites but now reside in 38 states. This review aims to provide some context to research on child care. It summarizes what is known about 3 aspects of nonparental care (age at entry and amount of care, type of care, and quality of care) and how each affects children's social, physical, and cognitive development. To identify studies of child care appropriate for this review, we searched MEDLINE, PsychINFO, and the SAGE Full-Text Collection. We then selected studies that focused on correlational and experimental research conducted throughout the world that included an adequate description of the type of care provided. Age at entry and amount of child care As the number of women entering the workforce increased rapidly in the 1970s, so did the number of children entering child care. Researchers, policymakers, and the public at large began to ask questions regarding the possible negative consequences of nonparental care, especially large amounts of care beginning in the first year of life.16,17 It is not easy to determine the effects of beginning care early in life or spending lots of time in care because children who enter care early also tend to spend large amounts of time there. Data from the SECC show that 84% of children experience nonmaternal care on a routine basis by age 12 months.18 Of this group, 72% enter care by age 4 months. At first entry, infants are in care for 29 hours a week on average. Once initiated, the amount of time spent in care on a weekly basis remains essentially stable throughout early childhood. However, nonparental care is anything but a consistent experience for many young children. On average, 34% of children younger than 3 years experience multiple arrangements, with 44% of 3- and 4-year-olds experiencing multiple arrangements.19 In summary, natural confounding occurs among the ages at which children begin child care, the amount of care, and the number of arrangements, making it difficult to isolate the effects of any 1 aspect of the child care experience. Granting this limitation, evidence has indicated that spending large amounts of time in child care from early infancy can have negative consequences. Attachment Some child care studies published in the 1980s and 1990s provided evidence of elevated rates of attachment insecurity for children who began full-time care early in life,20,21 but none of these studies controlled for child care quality, and most did not control for family factors. In the SECC,22,23 when quality and type of care were controlled for along with family background, children exposed to large amounts of care were at increased risk for attachment insecurity only if their mothers were highly insensitive. Mother-Child Interactions Although some studies published in the 1980s and 1990s reported the amount of early care to be related to more negative mother-infant interactions,24,25 others found positive effects26,27 or no effects.28 Much of this initial research was limited because of small sample sizes, few controls for family background, and reliance on a single time of measurement. The SECC29 addressed these concerns: mother-child interactions were observed at 6, 15, 24, and 36 months for more than 1000 families. These observations were evaluated by trained raters who were blinded to children's child care backgrounds. More hours in child care were linked to less maternal sensitivity and less positive engagement between the child and the mother, controlling for quality and type of child care, family income, maternal educational level, marital or partner status, maternal depression, maternal separation anxiety, child sex, child temperament, and ethnicity. In a follow-up report30 at 54 months and during the first grade, a higher number of hours in care was associated with less maternal sensitivity and less positive engagement in white children but greater maternal sensitivity and more positive engagement in African American and Latino children. Behavioral Problems Another area of investigation concerns the link between time in nonparental care and children's behavioral problems.31 The NICHD Early Child Care Research Network has reported a series of carefully controlled analyses showing that amount of time in nonparental care is associated with poor peer interactions and adjustment problems from the age of 2 years to the end of kindergarten.32-34 In follow-up analyses at grades 3 and 5, relations between hours and behavioral problems were no longer statistically significant, suggesting a fade-out effect.35,36 To more precisely delineate the relation between amount of care and behavioral problems, the SECC controlled for quality of interactions with caregivers and mothers, type of care, and family background.32 These controls resulted in modest reductions in the hours effect, but the effects continued to be statistically significant. Cortisol Levels and Peer Relationships Several studies37-39 implicate the role played by peer relationships in the observed association between amount of care and behavioral maladjustment. Salivary cortisol levels tend to increase from midmorning to midafternoon on days when children are in child care but not on days when they stay at home.39,40 Cortisol levels increased across the day, especially for children who had difficulty regulating negative emotions and behavior,36 who were more fearful,38 who were less involved in peer play, and who were less socially competent.37 Rises in cortisol levels occurred more in toddlers and preschoolers than infants and school-aged children.38-40 These findings suggest that toddlers and preschoolers who are learning to negotiate with peers may experience group settings as stressful. Although less socially competent children exhibit greater increases in cortisol levels in peer group activities,41-43 temperamental differences in children also appear to play a role.37,42,44 Some evidence has suggested that high-quality child care might reduce the stress induced by spending long hours in care,39,45 but much remains unclear regarding whether changes in the organization of child care programs or in caregivers' efforts might result in less stressful environments for children.46 Overall, research indicates that being in care for 30 or more hours a week is associated with small but statistically significant increases in behavioral problems. Children who had 45 or more hours of care per week from ages 3 to 54 months had the most behavioral problems as kindergartners. Interestingly, in a study performed in Japan (where the average quality of care is higher), spending many hours in care was not associated with increased behavioral maladjustment.47 Social Competence Evidence has shown that spending time in nonparental care increases children's social knowledge and skills.48,49 Morales and Bridges50 found that children with experience in child care were more accomplished at entertaining themselves and managing challenges. A French study51 showed that children who spent time in child care manifested more self-confidence, were more outgoing, and showed less distress in new situations. Although research provides some evidence that nonparental care may enhance social competence, the evidence is mixed.52-54 In the SECC, analyses performed with careful controlling for home experience and child characteristics indicated that although nonparental care was associated with more positive and skilled peer play in child care, it was not associated with improved competence in peer play in a laboratory situation or as rated by parents.34 Language and Achievement Findings are mixed with respect to amount (and timing) of child care and children's cognitive, language, and academic performance. Analyses conducted by the NICHD Early Child Care Research Network11,34,36,55,56 found no relation between amount or timing and cognitive and language measures from infancy through the fifth grade, controlling for other aspects of care and family factors. Other researchers, however, have detected amount or timing effects with some conditions but only at some ages or only for some income or ethnic groups.57-60 Communicable Illness Not surprisingly, spending time in child care, especially in care with large numbers of other children, increases the likelihood that children will be exposed to common pathogens and experience more bouts of common communicable illnesses.61-65 Studies show an increased prevalence of diarrheal illness especially in the first 2 years of life for children who attend child care. However, by the age of 3 years little difference was found between children in nonparental care and those reared at home.61-63 Studies62,64,65 also show an increased prevalence of upper respiratory tract infections and otitis media in children who attend child care, with some evidence suggesting that experience in child care during the first 3 years of life might afford some immunity to colds as children reach elementary school. The major factor that contributed to respiratory infections was the number of other children present in the child care arrangement; number of hours of care was insignificant.61,62,64,65 Asthma Research on the relation between asthma and experience in child care has been inconsistent. Some studies66,67 show that child care is associated with an increased likelihood of asthma symptoms, but these studies may reflect transient wheeze more than persistent asthma,66 especially for children with a family history of atopy.68 One of the studies66 that showed an increased risk was conducted in Norway, where children do not enter care until after the age of 1 year. Other studies69-72 showed no relation or a decreased likelihood of asthma for children who attended child care, consistent with arguments that respiratory syncytial virus infections help promote the TH1 phenotype that is protective against atopic asthma.73-75 Recent unpublished findings from the SECC also indicated that time in care before the age of 1 year was associated with a decreased likelihood of late-onset asthma. Quality of child care Established Standards The American Public Health Association (APHA) and the American Academy of Pediatrics (AAP) have established standards for assuring quality of out-of-home child care.76 These standards, often referred to as structural quality, address such issues as age-based caregiver-child ratios, group size, health and safety practices, and qualifications and continuing education for child care providers. Consistent with recommendations made by the APHA and AAP, research demonstrates that high caregiver-child ratios, small group size, and well-trained caregivers result in higher-quality care.5,77 In settings where child-adult ratios were lower, caregivers spent less time managing children and children were less apathetic and distressed78; caregivers were more stimulating, responsive, and supportive.2,78-80 Caregivers were also more responsive, more socially stimulating, and less restrictive when fewer children were in the group.2,78,79,81 Caregivers tend to be more stimulating and supportive, organize materials better, and provide more age-appropriate experiences when they have more education and child-related training.2,79-83 These findings suggest that higher structural quality increases the likelihood of higher process quality conceptualized as supportive interactions with caregivers, positive interactions with peers, and opportunities for cognitively stimulating play. Consistent with this argument, the SECC84 observed positive developmental outcomes, controlling for maternal educational level and parenting quality, when children attended centers that were in compliance with the APHA's and AAP's recommended guidelines. Children who attended centers that met child-adult ratio standards displayed fewer behavioral problems and more positive social behaviors.11 Similar relations emerged between structural or caregiver characteristics and child developmental outcomes in child care homes.81 In their review of child care studies, the Committee on Family and Work Policies of the National Academy of Sciences concluded that when process quality was higher and adult-child ratios were higher, children appeared happier and more securely attached to caregivers in care settings.77 When child-adult ratios were higher and caregivers were more sensitive and positive, children appeared more prosocial and positively engaged with peers. Children were also rated as more cognitively competent in child care settings that offered more opportunities for art, blocks, and dramatic play and in settings where caregivers had college degrees and more early-childhood training. Twenty-three studies were cited by the National Academy of Sciences77 as finding relations between process quality and children's cognitive and social-emotional development, after controlling for family and child background factors. Since the National Academy of Sciences report was prepared, other investigators85-87 have also found higher-quality care to be associated with better cognitive performance and fewer behavioral problems. Long-term Adaptive Functioning High-quality care appears to be associated with long-term adaptive functioning as well. In the SECC,36,55,56,88 caregiver behavior predicted children's performance on standardized cognitive and language assessments through fifth grade, controlling for amount and type of care and an extensive list of family covariates. In the Cost, Quality, and Outcomes Study,3 a prospective longitudinal study of 579 children who attended 151 centers in 4 states, child care quality predicted cognitive, language, and social development during the early grade-school years. Children who had closer relationships with their preschool teachers were more sociable in kindergarten, controlling for earlier child adjustment and family factors. Children who were enrolled in higher-quality child care displayed better math skills during kindergarten and second grade. In addition, children who had closer relationships with their caregivers at the age of 4 years were reported by their second-grade teachers to be more socially competent with peers, controlling for family factors and previous child functioning.89 Stress Reduction More positive child care environments appear to reduce stress in children.90 Children who attended high-quality child care homes showed decreases in cortisol levels from morning to afternoon, whereas children in low-quality child care homes showed increases. This rise is the opposite of the typical pattern for circadian rhythm of cortisol but similar to a rise across the workday that has been recorded in adult executives who were under high pressure. Health Outcomes To date, few studies have investigated the relations between quality of child care and children's physical health. Results from the SECC indicated links between group size and communicable illnesses: children who attended centers with more than 6 children in a group had more bouts of common communicable illnesses.62,63 Other studies64-66 have obtained similar results. In one of the few experimental studies directed at hygiene practices, Kotch90 reported that staff from 60 randomly selected centers were trained in personal hygiene and environmental sanitation practices. They increased their hand washing, sanitary food handling, disinfection of diapering areas, and use of step cans for diaper disposal and reduced their preparation of food in diapering areas. The result was fewer cases of diarrhea. Child Care Quality and Family Risk Although not every study has observed significant associations between child care quality and child well-being, most have, especially those that measured quality on more than 1 occasion. The effects tend to be modest and more often observed for outcomes such as language and academic achievement.91 In some domains, poor-quality care appears to function as a risk factor. In the SECC, low-quality care, coupled with low maternal sensitivity, was associated with infant-mother attachment insecurity.23 In other cases, high-quality child care served as a protective factor for children who were otherwise at risk. For example, children of depressed mothers appeared to be more positively engaged with their mothers when they attended higher-quality child care.31 In analyses of school readiness, receptive language, and expressive language, higher-quality child care was found to buffer young children from the negative effects of poverty.92 A major limitation of studies on child care quality is that the poorest-quality child care settings have been difficult to observe; thus, the effect of quality may well be underestimated. Such an underestimation is particularly problematic given that quality estimates derived from the SECC suggest that child care quality is not high. Positive caregiving was rated as “not characteristic” in 60% of the settings that were observed, and only 10% were rated as “excellent.”2 To better understand the effects of quality, the factors that are examined as part of quality must be expanded. The quality of the peer environment in child care may be especially important to consider given the negative impact of poor peer relations on stress.39,93 The specific strategies caregivers use to promote children's social skills and to handle children's noncompliance and aggression may help to explain the effects associated with quantity of care. Type of child care Characteristics of Caregivers and Styles of Care The advantages of having children in small, informal settings where the caregiver-child ratios are more favorable vs having them in centers where there tends to be more professionalism have long been debated. Clarke-Stewart94 found that child care centers typically had better educated caregivers with more professional orientation, larger group sizes, more time spent in “lessons” and structured activities, and more child-oriented materials. In contrast, family child care homes typically devoted more time to free exploration, casual learning, and watching television. Similar findings were obtained in the SECC.2,79 Centers had larger group sizes and higher child-adult ratios but more stimulating environments and caregivers with more training. Cognitive and Social Outcomes Consistent with their greater educational focus, children who attended centers scored higher on cognitive assessments, controlling for family demographics and parenting. Children in center-based care were also more competent with strangers and more independent of mothers in a laboratory playroom. Least advanced were children with caregivers in their own homes. Children with more experience in center-based care obtained higher cognitive and language scores from the age of 2 years to third grade, controlling for family background and the quality and amount of child care.56 Loeb et al85 conducted a study of 451 children in 3 sites. Children were observed in centers and child care homes and when being cared for by family members after their parents began working, in connection with Temporary Assistance to Needy Families. Children who attended centers obtained higher cognitive and school readiness scores, controlling for family background and previous child performance. Participation in centers was unrelated to behavioral problems. Health Outcomes Because centers generally have large numbers of children, the likelihood of contracting communicable illnesses and otitis media is greater than in care where 6 or fewer other children are present.60-64,95-98 However, time spent in child care centers appears to be associated with no long-term health consequences and may even provide some protection against atopic diseases.68,69 Low-Income Children Experimental and quasi-experimental evidence indicates that high-quality center-based care confers social and academic benefits for children at risk for school failure, with evidence indicating that the benefits last into adulthood. Not only do children who receive high-quality care in centers do better in school, but they are also more likely to go to college, to avoid teenage pregnancy and criminality, and not to require governmental assistance.7-9,99-103 Three recent nationwide studies9,10,103 of Head Start families indicate that center-based care of even modest quality can confer social and cognitive benefits to low-income children. However, the effects do not appear to be as widespread as the effects of high-quality day care centers, and it is not yet clear whether the effects will be as long lasting or apply to all children. For example, results from the Head Start Impact Study10 provide little evidence of positive impact on children from Spanish-speaking families. An effort is under way, under the auspices of the Institution for Education Sciences at the US Department of Education, to determine whether carefully constructed preschool curricula can be brought to scale with good success. Efforts to provide prekindergarten programs for all children whose parents seek it are being mounted in many states as well. A significant number of these efforts are also being evaluated to determine effectiveness. These state and federal efforts should clarify the impact of moderate-quality to high-quality center-based preschool programs. All in all, day care centers appear to accelerate children's language and cognitive development, and for low-income children, time spent in centers of moderate to high quality appears to provide an even wider array of advantages. However, there seems to be a trade-off regarding type and quality of care. In centers, more children are present but caregivers have more professional training; therefore, these children are more likely to have structured stimulation but perhaps less likely to receive sensitive individual care. Family Influences In the United States, there are deeply held beliefs about the primacy of the family. Such beliefs are buttressed by psychological theories that see parents as having the principal role in shaping children's futures.104 As children have begun spending large amounts of time in nonparental care, worries have emerged about whether child care might supplant the family. As part of the SECC, information about the families was collected during face-to-face contacts at ages 1, 6, 15, 24, 36, and 54 months as well as during the first, third, and fifth grades. Information was collected about family structure and size, parental employment, income, psychosocial characteristics of the mother and father, attitudes and beliefs about parenting and child care, depression, and quality of parenting in observed interactions with the child. Physical characteristics of the home, including the opportunities it provides for social and cognitive enrichment, were assessed. In a series of analyses through grade 5, family factors were found to be consistently stronger predictors of children's cognitive, language, social-emotional, and behavioral outcomes. Family effects, on average, were 2 to 3 times as large as the effects of child care. In addition, similar-size family effects were obtained for children with extensive child care experience and those with less or no child care experience.105 In summary, the impact of the family on child behavior and development does not seem to have abated as a consequence of children spending substantial time away from their parents. That is not to say, however, that family influences may not wane as children get older and spend increasing amounts of time with the media and persons outside the family. Conclusions Although much progress has been made in understanding the effects of early child care, further research is clearly needed. Despite the uncertainties that remain, it is probably fair to conclude that it matters where children get their care, when they start care, how much care they get during infancy and early childhood, and whether the care they get meets standards of quality. Children who spend many hours in care beginning in infancy appear to be at increased risk for stress-related behavioral problems. Behavioral problems are more likely if the child struggles when interacting with peers and if parents do not provide high-quality care. Day care centers appear to confer some advantage to language development and achievement, especially for poor children and if the centers offer high-quality care. Children in arrangements that meet the standards established by the APHA and AAP tend to receive better care and to have better developmental outcomes. However, attending arrangements with 6 or more children increases the likelihood of communicable illnesses and ear infections, albeit those illnesses appear to have no long-term adverse consequences. Pediatricians can help provide guidance on child care decisions. In addition, they can provide perspective on the issues, including reminding parents that what children experience at home accounts for substantially more variance in child outcomes than does what they experience in child care.105 Thus, parents need to remain vigilant for signs of stress, ready to spend time with their children in productive and enjoyable activities, and cogniz