Archives of General Psychiatry

Harris, James C.

doi: 10.1001/archpsyc.64.4.398pmid: 17404116

I love you and need you always, I know I am extremely difficult at times due to my absorbtion (sic) in my work. . . . If you decide you want to be free, I consent but I earnestly believe we can have our best lives together. —Norman to Mary Rockwell, undated, 19501(p376) “I pray thee, then,/Write me as one that loves his fellow-men.” —Abou Ben Adhem, poem read at Norman Rockwell's funeral2(p59) When interviewed for a magazine profile in 1960, Norman Rockwell (1894-1978) emphatically defined himself as a genre painter. To make sure his interviewer got it right, he spelled it out for her, “That's spelled g-e-n-r-e.”1(p432) For more than 60 years, beginning with his first Saturday Evening Post cover, Boy With Baby Carriage, published on May 20, 1916, his anecdotal vignettes chronicled American life and values. In all, Rockwell completed 322 covers for the Post over nearly half a century. He represented America's proud patriotic strength and its democratic principles during 2 World Wars, and when the civil rights movement began, he documented the injustice of bigotry and the consequences of racial hatred. Throughout, as a keen observer of human nature, he depicted with a wry wit and a real sense of humor the hopes and struggles of growing up in America. Early in his career, as a matter of artistic technique, he was advised to invite the viewer into his illustrations. His success in engaging viewers through their active imaginations was widely appreciated. Adored by his viewing public and frequently scorned by avant-garde critics, Rockwell always hoped for more positive critical recognition. Despite his art training and familiarity with modern art (Picasso was one of his favorite artists), overall he did not receive such recognition. A re-evaluation of his work is under way,3 initiated by a traveling exhibition of selected art works from the Norman Rockwell Museum in Stockbridge, Mass. View LargeDownload Norman Rockwell (1894-1978), American. Cover image: The Art Critic, 1955. Oil on canvas, 39.5 × 36.26 in. ©1955 SEPS: Licensed by Curtis Publishing Co, Indianapolis, Ind. All rights reserved (http://www.curtispublishing.com). Rockwell married Irene O’Connor in 1916; the marriage ended in 1930. Irene remarried 5 days after the divorce that she had demanded was finalized. Afterwards, Rockwell rarely spoke of her. Filled with self-doubt when his fame and wealth were not enough to sustain his marriage, he immersed himself in his work, his lifelong pattern when stressed. Later that year, he married Mary Barstow, a 22-year-old school teacher 14 years his junior whom he met on a blind date. She gave birth to their 3 sons (Jarvis, Tom, and Peter) during the first 6 years of their marriage. With the onset of World War II, Rockwell turned his talents to the war effort. To help America understand what our troops were experiencing, he introduced GI Willie Gillis on Post covers. Gillis, an everyman filled with quiet strength and determination, appeared on 11 covers over the next 5 years, first as an Army recruit and finally as a college student, apparently supported by the GI Bill. Rockwell's most important contribution to the war effort was his Four Freedoms paintings based on freedoms proclaimed in President Franklin Delano Roosevelt's State of the Union address on January 6, 1941. Roosevelt hoped for a secure new world based on freedom of speech and expression, freedom of worship, freedom from want (economic security), and freedom from fear (a reduction in armaments so that no one nation has the means to commit an act of aggression against another). Freedom from Fear (Figure 1) shows a mother tucking sleeping children safely into bed while the father warmly looks down on them. The father holds a newspaper in his left hand with headlines that describe the Battle of Britain, reporting on the bombing of London (September 7, 1940, and May 16, 1941) by the German Luftwaffe. The Four Freedoms were completed in 6 months, and the Post published them on successive weeks beginning February 20, 1943; each was accompanied by an essay. The poet Stephen Vincent Benét wrote the essay on Freedom from Fear. These paintings became enduring national symbols4 that explained to Americans why they were fighting (N. and T. DeRobertis, oral communication, December 2006). The images were the centerpiece for a traveling war bond drive that raised more than $130 million. His Freedom of Speech was purchased for the permanent collection at the Metropolitan Museum of Art. Figure 1. View LargeDownload Norman Rockwell (1894-1978), American. Freedom from Fear, 1943. Oil on canvas, 45.75 × 35.5 in. ©1943 SEPS: Licensed by Curtis Publishing Co, Indianapolis, Ind. All rights reserved (http://www.curtispublishing.com). After the war, America gradually returned to a time of normalcy. However, Rockwell was dealing with new fears and anxieties. By 1948, Mary's alcoholism, mood swings, and depression were apparent to him, their family, and their friends. It became clear that she required treatment as she was sneaking off to a neighbor's home for a drink or reaching for a flask when driving her mother-in-law home.1(p353) Rockwell was loyal to his wife but was increasingly concerned about her unreliability in representing him. “No one could be sure if she would have her drinking under control, or if she would talk coherently at all.”1(p380) He put Mary first and was distressed when she raised questions about a divorce (epigraph) while ill. In the summer of 1952, she began treatment at the Austen Riggs Center in Stockbridge, Mass, with its director, Dr Robert Knight. In 1953, Mary drove weekly from Vermont to the Austen Riggs Center for treatment. However, after 2 automobile accidents, her driving license was revoked. When the severity of her condition was made clear to him, Rockwell arranged to sell their home in Vermont and move permanently to Stockbridge in December 1953. In August 1952, Rockwell sought therapy for depression himself. His therapist was Erik Erikson, himself a painter in his earlier years. Erikson was concerned that Rockwell might be at risk for suicide. Rockwell asked his son to remove his gun from the studio. Rockwell used his art to work through his emotions; Erikson noted that, “His happiness is determined by the progress of his work.”1 In the summer of 1954, Rockwell started The Art Critic (cover), published in the Post on April 16, 1955. His son Jarvis, an aspiring art student, was his model for the young art critic, and Mary modeled for the woman whose pendant is being examined with a magnifying glass as the older men holding tall glasses in the adjacent painting glare at him with scorn and disapproval. This painting seems to reflect 2 common attitudes toward Rockwell's work, sentimental approval by the general public for his innocent themes and condescending dismissal by the “art community.” Critic Clement Greenberg referred to these attitudes as avant-garde and kitsch,5 with Rockwell falling in the second category. Rockwell's methods reveal a greater depth of knowledge about art than generally appreciated. For The Art Critic, although Mary was the model for the woman, Rockwell based the composition on a portrait by Frans Hals, Portrait of a Woman, and a sketch by Peter Paul Rubens of his first wife, Isabella Brant.1 The attitude of the woman can be traced in 20 preparatory photographs and oil paintings. In them, Rockwell teased out Mary's most evocative expressions, photographing each facial characteristic until he found the visual humor he sought. At first, she overreacts to the young man who stares at her; then she is more attentive and disapproving; and finally she appears delighted, even flirtatious. Rockwell privately referred to this painting as a sexual joke; it was one that Jarvis did not appreciate.1(p401) Initially, the adjacent painting was a landscape, but it gradually shifted to become 3 Dutch elders who seem appalled at the young man's impudence. The image of the elders is a parody of Frans Hals's A Banquet of the Officers of the St George Militia Company and Rembrandt's The Syndics of the Clothmakers' Guild (but 3 judges are shown instead of 6).1 While Rockwell worked on The Art Critic, his son Tom was accidentally stabbed in a freak fencing accident.1 Under considerable family stress, Rockwell was admitted for a “month's rest” to St Luke's Hospital. In June 1955, the Austen Riggs psychiatrist referred Mary to the Institute of Living in Hartford, Conn, where she had the first of her treatments with electroconvulsive therapy for depression, remaining there for 6 months. For Rockwell, her illness was reminiscent of that of his first wife, who died in 1934, 4 years after their divorce, an apparent suicide by drowning (in the bathtub), following a 2-year hospitalization at McLean Hospital in Boston. Mary was treated with electroconvulsive therapy again in 1958. Then, on August 25, 1959, “They were enjoying being together . . . ,” Jarvis recalled, but she seemed “sluggish at lunch.” After lunch, Mary took a nap.6 When Rockwell went upstairs to wake her, he found her dead. “Heart failure” was the diagnosis on the death certificate, but the family initially assumed it was suicide. Some suggested that she poisoned herself with pills; however, her daughter-in-law indicated that there was no note and that no medications were missing. Later, Rockwell wrote to friends, “She was so unhappy, try to think of it as a release for her.”1(p427) Mary was 51 years old when she died. She was buried in the cemetery adjacent to their first home in Stockbridge. Friends wrote encouraging letters to Rockwell: “Mary knew you had given her much to live on in your life together and felt you to be a pillar in her often troubled life.”1(p428) It was support from his friends and his sense of irony that saved him from despair. The April 16, 1960, Post cover (Figure 2) seemed to many to be a tribute to Mary. It is the only one of his covers that shows a stained glass window. Rockwell paints himself hard at work completing an angel. Figure 2. View LargeDownload Norman Rockwell (1894-1978), American. Stained Glass, 1960. Oil on canvas, 46 × 43 in. ©1960 SEPS: Licensed by Curtis Publishing Co, Indianapolis, Ind. All rights reserved (http://www.curtispublishing.com). Norman Rockwell remarried on October 25, 1961, and Erikson attended the ceremony. It was a happy marriage, and Rockwell continued to paint until the last 2 years of his life. When Norman Rockwell died in 1978, his funeral service began with his favorite poem, Abou Ben Adhem. The poem (epigraph) reflects Rockwell's life philosophy about love for others. Many would agree that, most of all, this democratic, American g-e-n-r-e artist’s paintings do indeed reflect his love for his fellow-men. References 1. Claridge L Norman Rockwell: A Life. New York, NY: The Modern Library;2001; 2. Hunt L Selected Writings. New York, NY: Routledge;2003; 3. Hennessey MHKnutson A Norman Rockwell: Pictures for the American People. New York, NY: Harry N. Abrams, Inc;1999; 4. Murray SMcCabe J Norman Rockwell's Four Freedoms: Images That Inspire a Nation. Stockbridge, Mass Berkshire House;1993; 5. Greenberg C Art and Culture: Critical Essays. Boston, Mass: Beacon Press; 1989; 6. Rockwell N My Adventures as an Illustrator. New York: Harry N. Abrams, Inc;1988;

Westover, Arthur N.;McBride, Susan;Haley, Robert W.

doi: 10.1001/archpsyc.64.4.495pmid: 17404126

Abstract Context The abuse of stimulant drugs is increasing in the western United States. Although numerous case reports and animal studies suggest a link with stroke, epidemiologic studies have yielded conflicting results. Objective To test the hypothesis that young adults who abuse amphetamines or cocaine are at a higher risk of stroke. Design, Setting, and Participants Using a cross-sectional design and from a quality indicators' database of 3 148 165 discharges from Texas hospitals, we estimated the secular trends from January 1, 2000, to December 31, 2003, in the abuse of various drugs and of strokes. We developed separate logistic regression models of risk factors for hemorrhagic (n = 937) and ischemic (n = 998) stroke discharges of persons aged 18 to 44 years in 2003, and for mortality risk in patients with stroke. Main Outcome Measure Incidence of stroke using definitions from the Agency for Healthcare Research and Quality's stroke mortality Inpatient Quality Indicator. Results From 2000 to 2003, the rate of increase was greatest for abuse of amphetamines, followed by cannabis and cocaine. The rate of strokes also increased, particularly among amphetamine abusers. In 812 247 discharges in 2003, amphetamine abuse was associated with hemorrhagic stroke (adjusted odds ratio [OR], 4.95; 95% confidence interval [CI], 3.24-7.55), but not with ischemic stroke; cocaine abuse was associated with hemorrhagic (OR, 2.33; 95% CI, 1.74-3.11) and ischemic (OR, 2.03; 95% CI, 1.48-2.79) stroke. Amphetamine, but not cocaine, abuse was associated with a higher risk of death after hemorrhagic stroke (OR, 2.63; 95% CI, 1.07-6.50). Conclusion Increases in stimulant drug abuse may increase the rate of hospital admissions for strokes and stroke-related mortality. Evidence has been accumulating for 2 decades supporting a link between abuse of stimulant drugs and strokes in young people.1-8 Human imaging and postmortem examination, as well as laboratory animal models, suggest that stimulant drugs, such as cocaine and amphetamines, might produce strokes by direct effects on the cerebral circulation, including elevated blood pressure, vasculitis, and cerebral vasospasm.7,9-18 The support for a link between stimulant abuse and strokes in humans has come primarily from case reports and case series of strokes in young people who have abused these drugs. Such reports led the Food and Drug Administration to issue a public health advisory on the stimulant phenylpropanolamine and a request for companies to remove it from all products19 and to ban ephedra from over-the-counter products in February 2004.20 The ban was struck down in April 200521 and then reinstated by the Tenth Circuit US Court of Appeals in August 2006.22 Despite the many case reports, the causal link has remained controversial because of the possibility of coincidental co-occurrence from the high prevalence of drug abuse and methodologic weaknesses and conflicting findings of the 5 epidemiologic case-control studies and 1 cross-sectional study that have examined the association.23-28 Concerns over medical complications from cocaine abuse have recently been amplified by police and news reports of a rapid increase in abuse of methamphetamine, illegally produced in backyard drug laboratories from pseudoephedrine and illegally imported from Mexico. To address the problem longitudinally in a large population-representative sample that would address many of the shortcomings of prior studies, we analyzed a database of all patients hospitalized from January 1, 2000, to December 31, 2003, in Texas hospitals covered by a state quality-of-care reporting law. Methods Statewide hospital database The Texas Health Care Information Council (THCIC) oversees the mandatory reporting of a standardized International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)–coded discharge database for generation of quality indicators from all state-licensed hospitals, except small rural hospitals exempted by the state statute. These data have been provided to the Agency for Healthcare Research and Quality's Healthcare Cost and Utilization Project for public reporting of quality indicators since 1999. Compared with 501 Texas hospitals listed in the American Hospital Association Guide in 2003, THCIC averaged 434 hospitals reporting during 2003 (86.6%), which included approximately 91.3% (57 277/62 726) of hospital beds and 95.1% (2 439 043/2 565 861) of admissions in the state (excluding Department of Veterans Affairs and military hospitals that do not provide admissions' information to either the THCIC or the American Hospital Association).29,30 We analyzed the annual databases from 2000 to 2003, which together contained coded records for 10 834 435 hospital discharges during this 4-year period. In the target age group of 18 to 44 years, the THCIC database contained 751 568 discharges in 2000, 777 997 in 2001, 806 353 in 2002, and 812 247 in 2003. Because the system generates publicly reported quality indicators, THCIC electronically edits records submitted by hospitals for consistency and conformity to its data collection procedures. Records failing the edits are returned to the hospital for correction, and then resubmitted. The THCIC then builds the database that each submitting hospital reviews, corrects, and certifies. The THCIC database contains the following 11 diagnosis fields: 1 admitting diagnosis, 1 principle discharge diagnosis, 8 secondary discharge diagnoses, and 1 external cause of injury field. The completeness of documentation of comorbidities for mortality adjustment in state discharge data from record systems providing 9 diagnosis fields (sensitivity, 0.926) has been shown to be virtually the same as that from record systems providing 25 discharge fields (sensitivity, 0.933).31 In the THCIC database, all 9 diagnosis fields were used in 13.5% of records in 2000 and 20.6% of records in 2003, and an average of 4.4 codes per record were recorded in 2000 and 4.9 in 2003. Case definition of acute stroke Acute strokes were defined using the ICD-9-CM codes in the Agency for Healthcare Research and Quality's Inpatient Quality Indicator 17 (Table 1).32 To define strokes, we searched in only the principal discharge diagnosis field, which has been shown to maximize the sensitivity and specificity of stroke ascertainment.33 There were 1887 all-type strokes in 2000, 2097 in 2001, 2133 in 2002, and 2252 in 2003. Strokes were further categorized as hemorrhagic, ischemic, or “other” only in the 2003 data, the only year in which the distinction was accurately coded by the hospitals. Before a coding rule change by the Centers for Medicare and Medicaid Services in October 2002, hospitals were not required to make the distinction.34 In 2003, 937 hemorrhagic and 998 ischemic strokes occurred. Analyses of secular trends during the 4 years were performed for all-type strokes combined. Definitions of risk factors We defined all conditions that might predispose to stroke (Table 1)23 from the codes in all available diagnosis fields to maximize the completeness of ascertainment. We defined abuse of a given drug by the ICD-9-CM codes for either abuse or dependence. To capture those most likely to have a close temporal relationship between stroke and substance abuse, patients with diagnoses coded as “in remission” were not categorized as substance abusers for our analysis. Nonspecific codes of drug abuse or dependence not indicating the specific drug (codes 304.8, 304.9, and 305.9) were not used. To avoid statistical disclosure for subjects with sensitive diagnoses, the THCIC database provides age in 5 broad categories (eg, 18-44 years) and does not release sex. Other coded measures used in the analysis included in-hospital death and the patients' metropolitan statistical area, defined by the US Census Bureau35 and used as a measure of urbanity. The cause of in-hospital death was not available. Patients' race was categorized as Asian, black, white, Hispanic, or “other” (comprising patients with missing race data or those classified as American Indians, Eskimos, Aleutians, or other race); the other category accounted for 6.1% of patients. Statistical analyses The annual prevalence rates of abuse of specific drugs were defined as the number of discharges with the recorded drug abuse per 100 discharges (with the standard error of a proportion). The percentage change in drug abuse prevalence was the prevalence rate in the given year minus the prevalence rate in the baseline year 2000, divided by the prevalence rate in 2000 (with the standard error of a percentage change36). The incidence rate of acute stroke was the number of discharges with codes meeting the Agency for Healthcare Research and Quality's-Inpatient Quality Indicator definition for stroke per 100 000 discharges (with the standard error of a proportion). The significance of the change in rates during the 4 years was tested with the Cochrane-Armitage test for trend, which tests the null hypothesis that during the 4 years the slope of the prevalence was 0 (not changing over time). The adjusted odds ratios (ORs) for the association of risk factors with acute hemorrhagic or ischemic stroke and their 95% confidence intervals (CIs) were obtained by multivariate logistic regression. The multivariate population-attributable risk percentage was calculated from the adjusted ORs and the risk factor prevalences by the method of Bruzzi et al.37 Statistical analyses were performed with SAS statistical software, version 9.1, for Windows (SAS Institute Inc, Cary, NC). Results In patients hospitalized in Texas hospitals, cocaine was reported to be the second most frequently abused drug, after alcohol, and amphetamines were the fifth most frequent (Figure 1A). While the rates of abuse of alcohol and hallucinogens did not increase during the 4 years, the rates of abuse of cocaine, cannabis, opioids, and amphetamines increased significantly (Figure 1A). Of these increases, the rate of increase was greatest for amphetamines (Figure 1B). The prevalence of amphetamine abuse in 2003 was higher in hospitals in rural (non–metropolitan statistical area) counties than in suburban or urban counties (OR, 1.40; 95% CI, 1.26-1.55; P<.001); the prevalence of cocaine abuse was lower in rural counties than in urban or suburban counties (OR, 0.72; 95% CI, 0.67-0.76; P<.001). The incidence rate of stroke among amphetamine abusers, cannabis abusers, and hospitalized patients without any associated abuse of alcohol, cocaine, cannabis, amphetamines, opioids, or hallucinogens trended upward from 2000 to 2003 (Figure 1C). The rate of increase was greatest in stroke associated with amphetamine abuse (Figure 1D). Multivariate logistic regression models from 2003 identified different patterns of association with hemorrhagic and ischemic stroke (Table 2). Amphetamine abuse was strongly associated with hemorrhagic stroke, but not with ischemic stroke. The strength of its association with hemorrhagic stroke was more than twice that of cocaine or tobacco use, but less than that of cerebrovascular anomalies, intracranial tumors, and hypertension. Combinations of abused drugs did not significantly contribute to the logistic regression models. Alcohol abuse has been associated with hemorrhagic stroke in prior studies,38 and trended toward significance in this model. Atrial fibrillation or flutter was a strong risk factor for ischemic stroke in univariate analysis (OR, 1.72; 95% CI, 1.04-2.85) (eTable), but it did not remain significant in the multivariate model because of strong collinearity with the “miscellaneous cardiac” variable (malignant neoplasm of the heart, acquired mural thrombus following myocardial infarction, heart valve disorder, prosthetic heart valve, and atrial septal defect). After controlling for amphetamine and cocaine abuse, a summary indicator of “any illicit drug use” was not independently associated with hemorrhagic stroke (OR, 1.42; 95% CI, 0.93-2.18). The associations between amphetamine abuse and hemorrhagic stroke (P=.14 by the Breslow-Day test for homogeneity) and cocaine abuse and hemorrhagic (P=.55) or ischemic (P=.37) stroke did not vary by race. Among persons aged 18 to 44 years in 2003, in-hospital death occurred in 3763 admissions (0.46%). An analysis of in-hospital death in the 2003 data showed that hemorrhagic stroke carried a much higher risk of death than that in all other hospitalized patients (OR, 58.3; 95% CI, 49.6-68.5; P<.001); in contrast, the increased mortality risk from ischemic stroke was much less (OR, 11.7; 95% CI, 8.8-15.6; P<.001). In patients with hemorrhagic strokes, only amphetamine abuse, coagulation defects, and hypertension were strong independent predictors of in-hospital death; in contrast, in patients with ischemic stroke, only acute myocardial infarction was significantly associated with death (Table 3). Repeating the analysis in patients with all types of strokes combined showed that the increased risk of death from amphetamine abuse (OR, 3.92; 95% CI, 1.79-8.59; P<.001) was greater than the increased risk from coagulation defects (OR, 3.06; 95% CI, 1.89-4.95; P<.001), and 3 times higher than that from hypertension (OR, 1.29; 95% CI, 0.97-1.73; P = .08). If the associations are causal and unbiased, in 2003 in Texas, 14.4% of hemorrhagic strokes and 14.4% of ischemic strokes in hospitals were accounted for by abuse of drugs, including amphetamines, cocaine, cannabis, and tobacco (Figure 2). Comment Controlling for other risk factors, we found that amphetamine abuse was associated with twice the risk of hemorrhagic stroke as cocaine abuse. In contrast, amphetamine abuse was not associated with increased risk of ischemic stroke, while cocaine abuse was associated with an increased risk. Amphetamine abuse, but not cocaine abuse, was associated with increased risk of death after a hemorrhagic stroke. The public health implications of these findings are heightened by growing news accounts suggesting a recent increase in methamphetamine abuse, particularly in the southwestern, western, and midwestern states.39,40 This concern was supported by our finding that, among hospitalized patients in Texas from 2000 to 2003, the rate of amphetamine abuse was increasing faster than that of any other drug, including cocaine, and the rate of strokes among amphetamine abusers was increasing faster than the rate of strokes among abusers of any other drug. Animal studies support the biological plausibility of a causal link between cocaine and amphetamine abuse and strokes. Intravenous methamphetamine, in rhesus monkeys, causes microhemorrhaging, thrombosis, infarction, poor vascular filling, and fragmentation of small arterioles and capillary beds.12,13 Methamphetamine also has been shown to exacerbate ischemic brain injury in mice.41 Cocaine causes vasoconstriction11 and disruption of cerebrovascular autoregulation in the presence of increased blood pressure.42 Past controlled epidemiologic studies in humans, beset by the difficulties of studying the medical effects of drug abuse, have not convincingly supported the link between stimulant use and stroke. Of the 4 studies providing affirmative evidence, all used the case-control design. Kaku and Lowenstein23 linked drug abuse in general to stroke, but did not separate out stimulants from nonstimulants. Petitti et al,24 using rigorous means to document exposure and outcome measures in a managed care population of women, found a strong association between stimulant use and stroke, but had relatively few strokes and did not separately quantify the effects of cocaine and amphetamines on hemorrhagic and ischemic stroke. Kernan et al27 found no overall association between phenylpropanolamine—a stimulant in many over-the-counter products—and hemorrhagic stroke, but in subgroup analysis found a strong association in women but not in men. Reanalysis of the same case-control series by Morgenstern et al28 found no overall association of ephedra-containing products with stroke, but subgroup analysis found a nonsignificant trend toward association in the group using the highest daily dose. These 2 studies21,43 proved controversial. Two studies by Qureshi et al25,26 reported the lack of association between crack cocaine use and stroke. A case-control study,25 which excluded approximately half of the stroke cases for lack of clear drug abuse history, found paradoxical protective effects for known stroke risk factors, including smoking, alcohol abuse, and diabetes mellitus. An analysis of the Third National Health and Nutrition Examination Survey found no association between cocaine use and self-reported history of nonfatal stroke.26 To our knowledge, no study has assessed the link between amphetamine abuse and stroke in the context of the recent increase in methamphetamine abuse in the southwestern, western, and midwestern states. Our finding of an increasing secular trend in the prevalence rate of amphetamine abuse and the incidence rate of amphetamine-associated strokes in a hospital patient population can be explained by either an increase in amphetamine abusers in the community or an increasing intensity of use leading to more complications. National prevalence surveys of drug abuse show that the rate of methamphetamine abusers is highest in the western, southwestern, and midwestern states,40 but apparently did not increase during the early years of this decade.44 This suggests that the increased rate in our hospital population is because of the increased intensity of methamphetamine use. This interpretation is supported by 2 recent reports. First, among methamphetamine abusers, the percentage meeting abuse or dependence criteria of illicit drugs during the past 12 months has increased precipitously (from 27.5% in 2002 to 59.3% in 2004).45 Second, the American Association of Poison Control Centers Toxic Exposure Surveillance System46 reported a statistically significant increase in total methamphetamine-related deaths (from 13 to 23) from 2002 to 2003; deaths from cocaine abuse were virtually unchanged (from 52 to 53), and heroin-related deaths decreased (from 40 to 23). Although case reports47-50 have suggested a link between cannabis use and stroke, to our knowledge, this is the first controlled epidemiologic study to report a significantly increased OR for this link. Human immunodeficiency virus infection was associated with ischemic stroke in univariate analysis, but did not remain significant after controlling for the other known causes. This is consistent with prior studies51-54 of human immunodeficiency virus and stroke. A strength of our study is that it was done in a database representative of hospitalized conditions in all but the smallest rural hospitals in Texas. Because approximately 80% of patients who experience strokes are hospitalized,24,55 and approximately 80% of deaths within 30 days after hospital admission take place before discharge,56 our study represents the state's stroke population reasonably well. Our findings confirmed the popular view that methamphetamine abuse is more common in rural populations, whereas cocaine abuse is more common in urban populations. The large size of the THCIC database provides unusually ample statistical power to test associations with hemorrhagic and ischemic strokes. The THCIC database, although subject to misclassification of measurements, has several characteristics that are likely to have maximized the accuracy of the data. Specifically, these include state-mandated public reporting overseen by a state agency, a standardized coding protocol, electronic auditing, and rigorous edit criteria required by law, and the fact that the number of discharge diagnoses collected did not limit the sensitivity of condition reporting.31 Whereas the exact temporal link between last drug abuse and stroke was not quantified in the database, by not counting diagnoses of substance abuse in remission, we narrowed the measure of drug abuse to active drug users. Because case reports8,57,58 have documented strokes delayed by several days to months after last use, our measures of active drug abuse should be sufficient for demonstrating epidemiologic links. In some cases, a secondary diagnosis may be the consequence of stroke rather than a risk factor. For example, while atrial fibrillation is a known risk factor for ischemic stroke,59 it also can be a rare complication of ischemic stroke.60 In this study, we were unable to distinguish primary from secondary or recurrent strokes; consequently, the incidence rates and population-attributable risk percentages refer to all strokes. Still, the major concern is misclassification of variables in a database of ICD-9-CM–coded discharge diagnoses. The occurrence of acute strokes has been shown to be accurately captured by ICD-9-CM principal discharge diagnosis in 2 validation studies.33,61 The specificity of drug abuse histories is thought to be high when it is recorded. In a multihospital study,62 when measured against urine toxicology test results, the sensitivity of self-report of cocaine use was 72%. A similar study63 in an obstetrical unit found self-report sensitivities of 58% to 70% for different drugs of abuse. One third of trauma patients in the US National Trauma Data Bank received a urine toxicology screen.64 The sensitivity of major comorbid conditions in hospital ICD-9-CM codes is generally greater than 65%.65 The prevalence rates of amphetamine and cocaine abuse in our study were similar to those measured in the National Survey of Drug Abuse,66 the California study67 of women in labor, and the study by Petitti et al.24 Misclassification of drug abuse history is an unavoidable hazard of studies of drug abuse. More to the point, however, is whether and how much such misclassification biases estimates of its associations with stroke and similar outcomes to which it predisposes. Latkin et al68 found, in a study of human immunodeficiency virus, that failure of subjects to disclose drug abuse was strongly related to a measured scale of social desirability concerns, but the level of social desirability concerns was not associated with the rate of human immunodeficiency virus infection, thus indicating a nondifferential information bias. Magder et al69 directly examined whether incomplete reporting of cocaine abuse biased logistic regression analyses of the association of cocaine abuse and stroke. Both studies concluded that the point estimates of the ORs are not seriously affected, while their CIs are broadened. If misclassification of drug abuse history behaves similarly in Texas hospitals, the ORs of our logistic regression models should be relatively unaffected by misclassification of drug abuse histories, and the broadening of their CIs is offset by the large sample sizes available for analysis. Finding strong associations of stroke with certain drugs of abuse, such as amphetamines, cocaine, and tobacco, but not with others, including opioids and hallucinogens, adds further confidence to the associations. Correspondence: Robert W. Haley, MD, Division of Epidemiology and Preventive Medicine, Department of Internal Medicine, The University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, TX 75390-8874 ([email protected]). Submitted for Publication: June 14, 2006; final revision received August 24, 2006; accepted September 5, 2006. Financial Disclosure: None reported. Funding/Support: This study was supported in part by grant R25 MH68338-01 from the National Institute of Mental Health. Role of the Sponsor: The funding body had no role in data extraction and analyses, in the writing of the manuscript, or in the decision to submit the manuscript for publication. Additional Information: The online-only eTable is available. Acknowledgment: We thank John Rush, MD, for reviewing the final manuscript and making helpful suggestions; Nicole Johnson, RN, MBA, and Sylvia Cook, MA, for important background information about the database; and John Hedl, PhD, for providing the THCIC database. References 1. Sloan MAKittner SJFeeser BRGardner JEpstein AWozniak MAWityk RJStern BJPrice TRMacko RFJohnson CJEarley CJBuchholz D Illicit drug-associated ischemic stroke in the Baltimore-Washington Young Stroke Study. Neurology 1998;501688- 1693PubMedGoogle ScholarCrossref 2. Levine SRWelch KM Cocaine and stroke. Stroke 1988;19779- 783PubMedGoogle ScholarCrossref 3. Buxton NMcConachie NS Amphetamine abuse and intracranial haemorrhage. J R Soc Med 2000;93472- 477PubMedGoogle Scholar 4. Lichtenfeld PJRubin DBFeldman RS Subarachnoid hemorrhage precipitated by cocaine snorting. Arch Neurol 1984;41223- 224PubMedGoogle ScholarCrossref 5. Levine SRBrust JCFutrell NHo KLBlake DMillikan CHBrass LMFayad PSchultz LRSelwa JFWelch KMA Cerebrovascular complications of the use of the “crack” form of alkaloidal cocaine. N Engl J Med 1990;323699- 704PubMedGoogle ScholarCrossref 6. Kibayashi KMastri ARHirsch CS Cocaine induced intracerebral hemorrhage: analysis of predisposing factors and mechanisms causing hemorrhagic strokes. Hum Pathol 1995;26659- 663PubMedGoogle ScholarCrossref 7. Delaney PEstes M Intracranial hemorrhage with amphetamine abuse. Neurology 1980;301125- 1128PubMedGoogle ScholarCrossref 8. Yen DJWang SJJu THChen CCLiao KKFuh JLHu HH Stroke associated with methamphetamine inhalation. Eur Neurol 1994;3416- 22PubMedGoogle ScholarCrossref 9. Citron BPHalpern MMcCarron MLundberg GDMcCormick RPincus IJTatter DHaverback BJ Necrotizing angiitis associated with drug abuse. N Engl J Med 1970;2831003- 1011PubMedGoogle ScholarCrossref 10. Nolte KBBrass LMFletterick CF Intracranial hemorrhage associated with cocaine abuse: a prospective autopsy study. Neurology 1996;461291- 1296PubMedGoogle ScholarCrossref 11. Nunez BDMiao LRoss JNNunez MMBaim DSCarrozza JP JrMorgan JP Effects of cocaine on carotid vascular reactivity in swine after balloon vascular injury. Stroke 1994;25631- 638PubMedGoogle ScholarCrossref 12. Rumbaugh CLBergeron RTScanlan RLTeal JSSegall HDFang HCMcCormick R Cerebral vascular changes secondary to amphetamine abuse in the experimental animal. Radiology 1971;101345- 351PubMedGoogle Scholar 13. Rumbaugh CLFang HCHiggins REBergeron RTSegall HDTeal JS Cerebral microvascular injury in experimental drug abuse. Invest Radiol 1976;11282- 294PubMedGoogle ScholarCrossref 14. Rumbaugh CLBergeron RTFang HCMcCormick R Cerebral angiographic changes in the drug abuse patient. Radiology 1971;101335- 344PubMedGoogle Scholar 15. Bostwick DG Amphetamine induced cerebral vasculitis. Hum Pathol 1981;121031- 1033PubMedGoogle ScholarCrossref 16. Harrington HHeller HADawson DCaplan LRumbaugh C Intracerebral hemorrhage and oral amphetamine. Arch Neurol 1983;40503- 507PubMedGoogle ScholarCrossref 17. Brust JC Vasculitis owing to substance abuse. Neurol Clin 1997;15945- 957PubMedGoogle ScholarCrossref 18. Konzen JPLevine SRGarcia JH Vasospasm and thrombus formation as possible mechanisms of stroke related to alkaloidal cocaine. Stroke 1995;261114- 1118PubMedGoogle ScholarCrossref 19. Center for Drug Evaluation and Research, US Food and Drug Administration, Phenylpropanolamine (PPA) information page. http://www.fda.gov/cder/drug/infopage/ppa/default.htmAugust 21, 2006 20. US Food and Drug Administration, Final rule declaring dietary supplements containing ephedrine alkaloids adulterated because they present an unreasonable risk. http://fr.cos.com/cgi-bin/getRec?id=20040211a160August 21, 2006 21. Gardiner H Judge's decision lifts ban on sale of Ephedra in Utah. http://www.nytimes.com/2005/04/15/health/15ephedra.html?ex=1169269200&en=a22ef4322147998b&ei=5070January 18, 2007 22. NPI Center, Appeals court rules for FDA in Ephedra ban: overturns lower court finding in Nutraceutical v von Eschenbach. http://www.ck10.uscourts.gov/opinions/05/05-4151.pdfJanuary 18, 2007 23. Kaku DALowenstein DH Emergence of recreational drug abuse as a major risk factor for stroke in young adults. Ann Intern Med 1990;113821- 827PubMedGoogle ScholarCrossref 24. Petitti DBSidney SQuesenberry CBernstein A Stroke and cocaine or amphetamine use. Epidemiology 1998;9596- 600PubMedGoogle ScholarCrossref 25. Qureshi AIAkbar MSCzander ESafdar KJanssen RSFrankel MR Crack cocaine use and stroke in young patients. Neurology 1997;48341- 345PubMedGoogle ScholarCrossref 26. Qureshi AISuri MFGuterman LRHopkins LN Cocaine use and the likelihood of nonfatal myocardial infarction and stroke: data from the Third National Health and Nutrition Examination Survey. Circulation 2001;103502- 506PubMedGoogle ScholarCrossref 27. Kernan WNViscoli CMBrass LMBroderick JPBrott TFeldmann EMorganstern LBWilterdink JLHorwitz RI Phenylpropanolamine and the risk of hemorrhagic stroke. N Engl J Med 2000;3431826- 1832PubMedGoogle ScholarCrossref 28. Morgenstern LBViscoli CMKernan WNBrass LMBroderick JPFeldmann EWilterdink JLBrott THorwitz RI Use of Ephedra-containing products and risk for hemorrhagic stroke. Neurology 2003;60132- 135PubMedGoogle ScholarCrossref 29. American Hospital Association, AHA Guide to the Health Care Field. Chicago, Ill: American Hospital Association; 2004; 30. Texas Department of State Health Services, Center for Health Statistics, Texas Hospital Inpatient Discharge Public Use Data file, calendar years 2000 to 2003. http://www.dshs.state.tx.us/thcic/Hospitals/HospitalData.shtmJanuary 17, 2007 31. Romano PSMark DH Bias in the coding of hospital discharge data and its implications for quality assessment. Med Care 1994;3281- 90PubMedGoogle ScholarCrossref 32. Guide to Inpatient Quality Indicators: quality of care in hospitals—volume, mortality, and utilization. http://www.qualityindicators.ahrq.gov/iqi_download.htmAugust 21, 2006 33. Broderick JBrott TKothari RMiller RKhoury JPancioli AGebel JMills DMinneci LShukla R The Greater Cincinnati/Northern Kentucky Stroke Study: preliminary first-ever and total incidence rates of stroke among blacks. Stroke 1998;29415- 421PubMedGoogle ScholarCrossref 34. Centers for Medicare and Medicaid Services, Medicare program: changes to the hospital inpatient prospective payment systems and fiscal year 2003 rates. http://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=2002_register&docid=page+49981-50030.pdfAugust 21, 2006 35. US Census Bureau, Current lists of metropolitan and micropolitan statistical areas and definitions. http://www.census.gov/population/www/estimates/metrodef.html August 21, 2006 36. US Census Bureau, Percent changes: American Community Survey. http://www.census.gov/acs/www/Downloads/ACS/PercChg.pdfJanuary 18, 2007 37. Bruzzi PGreen SBByar DPBrinton LASchairer C Estimating the population attributable risk for multiple risk factors using case-control data. Am J Epidemiol 1985;122904- 914PubMedGoogle Scholar 38. O’Connor ADRusyniak DEBruno A Cerebrovascular and cardiovascular complications of alcohol and sympathomimetic drug abuse. Med Clin North Am 2005;891343- 1358PubMedGoogle ScholarCrossref 39. National Association of Counties, The Meth epidemic in America: two surveys of US counties: the criminal effect of Meth on communities: the impact of Meth on children. http://www.naco.org/Content/ContentGroups/Publications1/Press_Releases/Documents/NACo-MethSurvey.pdfAugust 21, 2006 40. Community Epidemiology Work Group, Epidemiologic Trends in Drug Abuse. http://www.drugabuse.gov/PDF/CEWG/AdvReport605.pdfJanuary 18, 2007 41. Wang YHayashi TChang CFChiang YHTsao LISu TPBorlongan CLin SZ Methamphetamine potentiates ischemia/reperfusion insults after transient middle cerebral artery ligation. Stroke 2001;32775- 782PubMedGoogle ScholarCrossref 42. Kelley PASharkey JPhilip RRitchie IM Acute cocaine alters cerebrovascular autoregulation in the rat neocortex. Brain Res Bull 1993;31581- 585PubMedGoogle ScholarCrossref 43. Stier BGHennekens CH Phenylpropanolamine and hemorrhagic stroke in the Hemorrhagic Stroke Project: a reappraisal in the context of science, the Food and Drug Administration, and the law. Ann Epidemiol 2006;1649- 52PubMedGoogle ScholarCrossref 44. Substance Abuse and Mental Health Services Administration, Overview of findings from the 2004 National Survey on Drug Use and Health. http://www.oas.samhsa.gov/nsduh/2k4nsduh/2k4overview/2k4overview.pdfJanuary 18, 2007 45. Substance Abuse and Mental Health Services Administration, The NSDUH report: methamphetamine use, abuse, and dependence: 2002, 2003, and 2004. http://www.oas.samhsa.gov/2k5/meth/meth.pdf August 21, 2006 46. Watson WALitovitz TLKlein-Schwartz WRodgers GC JrYouniss JReid NRouse WGRembert RSBorys D 2003 annual report of the American Association of Poison Control Centers Toxic Exposure Surveillance System. Am J Emerg Med 2004;22335- 404PubMedGoogle ScholarCrossref 47. Mateo IPinedo AGomez-Beldarrain MBasterretxea JMGarcia-Monco JC Recurrent stroke associated with cannabis use. J Neurol Neurosurg Psychiatry 2005;76435- 437PubMedGoogle ScholarCrossref 48. Geller TLoftis LBrink DS Cerebellar infarction in adolescent males associated with acute marijuana use. Pediatrics. 2004;113:e365-e370.http://pediatrics.aappublications.org/cgi/content/full/113/4/e365. January 18, 2007 49. Finsterer JChristian PWolfgang K Occipital stroke shortly after cannabis consumption. Clin Neurol Neurosurg 2004;106305- 308PubMedGoogle ScholarCrossref 50. Mesec ARot UGrad A Cerebrovascular disease associated with marijuana abuse: a case report. Cerebrovasc Dis 2001;11284- 285PubMedGoogle ScholarCrossref 51. Cole JWPinto ANHebel JRBuchholz DWEarley CJJohnson CJMacko RFPrice TRSloan MAStern BJWityk RJWozniak MAKittner SJ Acquired immunodeficiency syndrome and the risk of stroke. Stroke 2004;3551- 56PubMedGoogle ScholarCrossref 52. Qureshi AIJanssen RSKaron JMWeissman JPAkbar MSSafdar KFrankel MR Human immunodeficiency virus infection and stroke in young patients. Arch Neurol 1997;541150- 1153PubMedGoogle ScholarCrossref 53. Patel VBSacoor ZFrancis PBill PLBhigjee AIConnolly C Ischemic stroke in young HIV-positive patients in Kwazulu-Natal, South Africa. Neurology 2005;65759- 761PubMedGoogle ScholarCrossref 54. Connor MDLammie GABell JEWarlow CPSimmonds PBrettle RD Cerebral infarction in adult AIDS patients: observations from the Edinburgh HIV Autopsy Cohort. Stroke 2000;312117- 2126PubMedGoogle ScholarCrossref 55. Piriyawat PSmajsova MSmith MAPallegar SAl-Wabil AGarcia NMRisser JMMoye LAMorgenstern LB Comparison of active and passive surveillance for cerebrovascular disease: the Brain Attack Surveillance in Corpus Christi (BASIC) Project. Am J Epidemiol 2002;1561062- 1069PubMedGoogle ScholarCrossref 56. Jencks SFWilliams DKKay TL Assessing hospital-associated deaths from discharge data: the role of length of stay and comorbidities. JAMA 1988;2602240- 2246PubMedGoogle ScholarCrossref 57. Ohta KMori MYoritaka AOkamoto KKishida S Delayed ischemic stroke associated with methamphetamine use. J Emerg Med 2005;28165- 167PubMedGoogle ScholarCrossref 58. Rothrock JFRubenstein RLyden PD Ischemic stroke associated with methamphetamine inhalation. Neurology 1988;38589- 592PubMedGoogle ScholarCrossref 59. Wolf PAAbbott RDKannel WB Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke 1991;22983- 988PubMedGoogle ScholarCrossref 60. Weimar CRoth MPZillessen GGlahn JWimmer MLBusse OHaberl RLDiener HCGerman Stroke Data Bank Collaborators, Complications following acute ischemic stroke. Eur Neurol 2002;48133- 140PubMedGoogle ScholarCrossref 61. Heckbert SRKooperberg CSafford MMPsaty BMHsia JMcTiernan AGaziano JMFrishman WHCurb JD Comparison of self-report, hospital discharge codes, and adjudication of cardiovascular events in the Women's Health Initiative. Am J Epidemiol 2004;1601152- 1158PubMedGoogle ScholarCrossref 62. Hollander JETodd KHGreen GHeilpern KLKarras DJSinger AJBrogan GXFunk JPStrahan JB Chest pain associated with cocaine: an assessment of prevalence in suburban and urban emergency departments. Ann Emerg Med 1995;26671- 676PubMedGoogle ScholarCrossref 63. Ostrea EM JrKnapp DKTannenbaum LOstrea ARRomero ASalari VAger J Estimates of illicit drug use during pregnancy by maternal interview, hair analysis, and meconium analysis. J Pediatr 2001;138344- 348PubMedGoogle ScholarCrossref 64. Kon AAPretzlaff RKMarcin JP The association of race and ethnicity with rates of drug and alcohol testing among US trauma patients. Health Policy 2004;69159- 167Google ScholarCrossref 65. Romano PS Asking too much of administrative data? J Am Coll Surg 2003;196337- 338PubMedGoogle ScholarCrossref 66. Substance Abuse and Mental Health Services Administration, 2004 National Survey on Drug Use and Health: detailed tables. http://www.oas.samhsa.gov/NSDUH/2k4nsduh/2k4tabs/Sect7peTabs66to129.htm#tab7.101b August 21, 2006 67. Vega WAKolody BHwang JNoble A Prevalence and magnitude of perinatal substance exposures in California. N Engl J Med 1993;329850- 854PubMedGoogle ScholarCrossref 68. Latkin CAVlahov DAnthony JC Socially desirable responding and self-reported HIV infection risk behaviors among intravenous drug users. Addiction 1993;88517- 526PubMedGoogle ScholarCrossref 69. Magder LSSloan MADuh SHAbate JFKittner SJ Utilization of multiple imperfect assessments of the dependent variable in a logistic regression analysis. Stat Med 2000;1999- 111PubMedGoogle ScholarCrossref

Nemeroff, Charles B.;Kalali, Amir;Keller, Martin B.;Charney, Dennis S.;Lenderts, Susan E.;Cascade, Elisa F.;Stephenson, Hugo;Schatzberg, Alan F.

doi: 10.1001/archpsyc.64.4.466pmid: 17404123

Abstract Context IMS Health Inc data presented by the Food and Drug Administration (FDA) on September 13 and 14, 2004, at a joint meeting of the Center for Drug Evaluation and Research's Psychopharmacologic Drugs Advisory Committee and the FDA's Pediatric Advisory Committee suggested that the number of children and teenagers who were prescribed antidepressants continued to increase in 2004, despite widespread publicity surrounding 2 FDA advisories regarding the potential for pediatric suicidality with selective serotonin reuptake inhibitor use. These results are contradictory to findings from the Medco Health Solutions, Inc, March 2004 analysis of pharmacy benefit claims and a separate subsequent analysis conducted by NDC Health using dispensing data from March 31, 2004, through June 30, 2005. Objectives To investigate the contradictory findings and provide additional analyses on the prescribing trends of antidepressants across age groups and physician specialties in the United States. Design Retail pharmacy prescription data and physician audit data were obtained from Verispan, a joint venture between Quintiles Transnational and McKesson. In addition to examining prescribing trends, a joinpoint regression analysis was conducted to identify the timing for significant changes in prescription use. Results The analyses suggest that the number of children and teenagers who were prescribed antidepressants has decreased significantly (P = .02) in the wake of widespread publicity surrounding the FDA public health advisories. Another impact of the advisories seems to be a shift in care from “generalists” to psychiatric specialists when it comes to prescribing antidepressants to patients younger than 18 years. Finally, the analyses highlight a slight shift in prescribing toward the non–selective serotonin reuptake inhibitor bupropion hydrochloride, even though it carries the same FDA “black box” warning as the selective serotonin reuptake inhibitors. Conclusions The effect on antidepressant prescribing volume observed in our analysis of the Verispan data parallels earlier findings reported by Medco Health Solutions, Inc, and NDC Health that the FDA actions have had a significant effect on the prescribing of antidepressants to children and adolescents. Together, these findings underline the importance of presenting a fair balance within the media due to the significant reach of this channel among prescribing physicians. In response to a US Food and Drug Administration (FDA) request for clarification of findings from a review of 3 pediatric studies of paroxetine hydrochloride, GlaxoSmithKline, Research Triangle Park, NC, submitted a report in May 2003 to the FDA and the Medicines and Healthcare Products Regulatory Agency of Great Britain describing the relationship between paroxetine use and pediatric suicidality, defined as suicidal thoughts and/or behavior among pediatric patients.1 Specifically, the report suggested an increase in risk of events termed possibly suicide related and suicide attempts for paroxetine-treated patients compared with placebo-treated patients. At the same time, a pooled analysis from all 3 studies failed to demonstrate significantly greater efficacy vs placebo, attributed at least in part to unusually high placebo response rates.2 Following the data analysis, the FDA requested pediatric suicidality data from all clinical trials from all antidepressant manufacturers, including paroxetine hydrochloride (Paxil; GlaxoSmithKline), citalopram hydrobromide (Celexa; Forest Laboratories, Inc, New York), fluvoxamine maleate (Luvox; Solvay Pharmaceuticals, Marietta, Ga), mirtazapine (Remeron; Organon, Roseland, NJ), nefazodone (Serzone; Bristol-Myers Squibb, New York, NY), sertraline hydrochloride (Zoloft; Pfizer Inc, New York), fluoxetine (Prozac; Eli Lilly and Co, Indianapolis, Ind), venlafaxine (Effexor; Wyeth, Madison, NJ), and bupropion hydrochloride (Wellbutrin; GlaxoSmithKline). Based on the FDA's preliminary analysis, the agency issued a public health advisory on October 27, 2003, to call attention to the health care professional reports of the occurrence of suicidality in clinical trials for antidepressants in pediatric and adolescent patients with depression. In addition, the agency scheduled a hearing in February 2004 to review data analyzed to date. At this hearing, which received a great deal of media attention, the committee discussed plans for a more comprehensive analysis and advised the FDA to issue a more substantial warning related to the potential adverse effects of the selective serotonin reuptake inhibitors (SSRIs) and other newer antidepressants (before the completion of the more comprehensive analysis). On March 22, 2004, the FDA issued another public health advisory asking manufacturers of SSRIs and newer antidepressants to include a warning statement in product labeling that recommends monitoring in adult and pediatric patients for the emergence of suicidality. On September 13 and 14, 2004, the Center for Drug Evaluation and Research's Psychopharmacologic Drugs Advisory Committee and the FDA's Pediatric Advisory Committee met again to discuss the analysis of the pediatric suicidality data based on the reclassification of events performed by Columbia University, New York. Key conclusions resulting from this meeting included the following: The data from these studies suggest that antidepressants increase the risk of suicidality in pediatric patients. Although there was some variability in findings, the panel was unable to conclude that any single agent was free of risk. The committee supports a “black box” warning for all antidepressants for pediatric use. Shortly after the September meetings, Medco Health Solutions, Inc, Franklin Lakes, NJ, at the request of the New York Times, conducted an analysis of its pharmacy benefit claims and found that the number of teenagers and children prescribed antidepressants had decreased by 18% during the past year.3 This finding contradicts information presented by the FDA based on data from IMS Health Inc, Fairfield, Conn. The FDA officials told the committee that prescriptions continued to increase in 2004, despite the confusion regarding antidepressant safety. An analysis of prescription-dispensing data provided by NDC Health, Atlanta, from March 31, 2004, through June 30, 2005, supports the Medco Health Solutions, Inc, finding and demonstrates that prescriptions for antidepressants in patients 18 years and younger have decreased by approximately 20% in the aftermath of the FDA public health advisory.4 Given that the data released by Medco Health Solutions, Inc, and now NDC Health are in stark contrast to the information released by the FDA in 2004, there is a clear need for a more comprehensive analysis of the impact of the publicity of the pediatric suicidality data on physician practice patterns. Herein, we use a combination of prescription data (based on approximately 55% of all US retail pharmacy claims—private payers, Medicaid, and cash) and physician survey data to provide additional insight on the impact of the pediatric suicidality findings. Methods We obtained retail pharmacy prescription data from Verispan, Yardley, Pa, a joint venture between Quintiles Transnational and McKesson, San Francisco, Calif. The Verispan data capture more than 1.4 billion patient-centric prescriptions per year, nearly half of all prescription activity in the United States. This data set includes prescriptions from a variety of retail channels (eg, national retail chains and mass merchandisers) from a near census of US pharmacies. The Verispan retail pharmacy database also captures information from all payer types, including cash. The average payer mix for antidepressants from January 2002 through December 2005 was as follows: 77.6% third-party private payers, 12.7% Medicaid, and 9.7% cash. In addition to payer type, because the data are patient-centric, information on patient age is available for every prescription from mid-2002 to the present (obtaining age data before that time frame is complex because of changes in the Verispan data). To highlight the impact of the suicidality data on antidepressant use, we did the following: (1) focused our analysis on the June 2000 to March 2005 time frame to clearly depict the prescribing effect; (2) analyzed total antidepressant prescriptions as a moving quarterly total, a time series aggregate over 3 months, to “smooth out” some of the monthly variation observed in the data; and (3) created a “suicide index” to normalize for the differences in prescription volume between the younger than 18, 18 to 25, and 26 years and older age categories. Specifically, the suicide index assigns a baseline value of 100 to the initial time point in each of the age series. Subsequent data points are presented relative to this baseline value: numbers greater than 100 signify an increase in market volume compared with baseline, while values less than 100 are indicative of market contraction. To test for significance among observed prescription trends, we performed a joinpoint regression analysis to determine average monthly percentage change in antidepressant prescribing. The analysis was performed using computer software (Joinpoint Software, version 3; National Cancer Institute, Bethesda, Md). This software performs a series of Monte Carlo permutation-based tests to test for the significance of the number of joinpoints in the data (ie, points at which trends change), first testing for 0 joinpoints and then up to as many as 3 joinpoints.5 The software program used (Joinpoint Regression Program) fits a joinpoint model to the trend data, and has been used to analyze antidepressant prescription data over time in England.6 The models were based on linear regression, with the log monthly prescription volume as the dependent variable and the month as the independent variable. Finally, to further explore the effect of the reported suicidality data on practice patterns, we also examined data from Verispan's Physician Drug and Diagnosis Audit. The Physician Drug and Diagnosis Audit provides a national-level disease and treatment database on a survey of approximately 3400 office-based physicians across 29 specialties. Information collected by the audit is projected by region and specialty to provide a representative view of physician practice in the United States. The Physician Drug and Diagnosis Audit data were analyzed to better understand changes in specialty physician treatment of patients younger than 18 years for depression and the antidepressants prescribed. Results Figure 1 presents total antidepressant prescriptions as a moving quarterly total from quarter August 2000 to quarter March 2005. As seen in Figure 1, the growth in total antidepressant prescriptions seemed to begin to slow around the timing of the public health advisories in October 2003 and March 2004 about pediatric suicidality. However, by the time the FDA panel met to discuss the data findings in September 2004, the proceedings from the meeting had little effect on the market—behavior had already been altered. In Figure 2, we present the moving quarterly total antidepressant prescriptions for 3 different age groups: those younger than 18 years, those 18 to 25 years, and those 26 years and older. Data for each age group in Figure 2 are presented as an index relative to the baseline measure of total prescriptions to facilitate comparisons across groups. As seen in Figure 2, in contrast to the 18 to 25 and 26 years and older age groups, the younger than 18 years market seems to be seasonal, with a decrease in prescribing volume during the summer months. In addition to highlighting the seasonal nature of the younger than 18 years market, the data in Figure 2 also suggest that the 2 public health advisories issued in October 2003 and March 2004 likely caused significant deviation from the historical trend of antidepressant use in patients younger than 18 years. In comparison, the effect on patients 26 years and older was much more subtle: the market seemed to decelerate in growth, but did not actually contract. The effect on the population aged 18 to 25 years was in between these 2 findings: a contraction of the market occurred, but not to the extent observed in the younger than 18 years age group. To test the significance of these observed market trends and to identify timing for market changes, we conducted a joinpoint analysis on each of the 3 age groups. As seen in Figure 3, the joinpoint analysis of prescription volume among patients younger than 18 years demonstrated that the number of prescriptions increased by a monthly average of 0.79% from April 2002 to February 2004 (95% confidence interval [CI], 0.45%-1.13%; P<.001 for test of the null hypothesis that monthly percentage change is 0). After February 2004, there was a decrease in the number of prescriptions by a monthly average of 4.23% (95% CI, −8.44% to 0.18%; P = .06), although the 95% CI for a joinpoint at February 2004 was wide (June 2002-April 2004). Although the CI is wide, we attribute the variation to the seasonality of the data and a natural decrease in antidepressant prescription volume in the summer months. By July 2004 (95% CI, August 2003-January 2005), the market began to stabilize such that there was no significant change in prescribing trends from July 2004 to March 2005 (95% CI, −1.16% to 1.48%; P = .92). In other words, the market effect may have potentially occurred before the FDA advisory panel in September 2004, when the results of the pediatric suicidality analysis were fully presented. In the 18- to 25-year age group, prescription volume increased at a monthly average of 0.48% (95% CI, 0.27%-0.70%; P<.001) from April 2002 to March 2004 (95% CI, April 2003-June 2004). Following the public health advisory in March 2004, prescription trends changed significantly in this age group to reflect an average monthly decrease of 0.72% (95% CI, −1.29% to −0.14%; P = .02). With respect to the 26 years and older age group, although the market seemed to slow in growth following the October 2003 and March 2004 communications, the effect on prescribing trends was not statistically significant (P = .98). To obtain a better understanding of the impact of the reported pediatric suicidality data on care patterns in depressed patients younger than 18 years, we compared physician specialty mix and antidepressant market share before and after March 2004. Figure 4 includes a comparison of specialty mix, and the Table provides insight into the class of antidepressant prescribed. As seen in Figure 4, there has been a shift in the providers of care of depressed patients younger than 18 years away from primary care and other providers to psychiatrists. In quarter February 2004, psychiatrists accounted for 44% of patient care for those younger than 18 years. In contrast, psychiatrists cared for 63% of pediatric/adolescent depressed patients in quarter February 2005. Classes of antidepressants and other agents prescribed to treat depression in the younger than 18 years population have also changed after the black box warning (Table). Although fluoxetine is the only antidepressant indicated for use in pediatric patients, there has been an increase in prescribing of non-SSRIs (eg, bupropion, tricyclic agents, and gabapentin) to patients younger than 18 years that is not observed in the 18 years and older population. While bupropion and tricyclic agents have demonstrated efficacy in treating depression in adults, neither of these products (nor gabapentin) is approved for use in depression in pediatric patients, and the tricyclic agents are associated with greater adverse effects than the SSRIs, and are lethal in overdose. Comment The level of effect on antidepressant prescribing volume observed in our analysis of the Verispan data supports the findings reported by Medco Health Solutions, Inc, and NDC Health, suggesting that the number of children and teenagers prescribed antidepressants has decreased dramatically since the October 2003 and March 2004 FDA-issued public health advisories that reported risks of suicidality with the use of antidepressants. According to Medco Health Solutions, Inc, the number of patients younger than 18 years prescribed antidepressants decreased sharply by 18% in the first quarter of 2004 and by an additional 5% in the second quarter of 2004.3,11 The review by Medco Health Solutions, Inc, included the 10.1 million pediatric patients covered under its pharmacy benefit management program. Our analysis of the Verispan data, which capture approximately half of all prescriptions in the United States, including private insurance, Medicaid, and cash prescriptions, shows only a 5% decrease in the first quarter of 2004, with an additional 11% decrease in the second quarter of 2004. The difference in the magnitude of the effect on prescribing volume between the Medco Health Solutions, Inc, data (private payers only) and our analysis of the Verispan data (all payers) suggests that the prescribing behavior may have been slower to change for nonprivately insured individuals (eg, those with Medicaid insurance) than for the privately insured. Although the Medco Health Solutions, Inc, NDC Health, and Verispan analyses suggest that the public health advisories regarding the safety of antidepressants have had a direct effect on physician practice patterns, the FDA reported that it did not observe a decline in pediatric antidepressant prescribing during the September 2004 hearings. Rather, FDA officials asserted that the use of antidepressants by children and teenagers was still increasing. Using data provided by IMS Health Inc, the FDA found that pediatric antidepressant prescriptions continued to increase by 7% in 2004.3 Agency officials reported that the March 2004 advisory had no effect on prescription trends, with the number of prescriptions for antidepressants given to children and teenagers growing by almost 8% in the first half of 2004.12 Whereas the Verispan, Medco Health Solutions, Inc, and NDC Health data are patient-centric, with age information readily available for each prescription captured, the IMS Health Inc data are primarily based on a survey of drug use by pharmacies, with no collection of patient-specific information; it remains unclear how the FDA distinguished between adult and pediatric prescriptions in its analysis.3 Because of data limitations associated with the IMS Health Inc data, it is more likely that our analysis of Verispan data and the Medco Health Solutions, Inc, and NDC Health analyses more clearly depict reality—the FDA actions have had a significant effect on the prescribing of antidepressants to children and adolescents. In addition to a decrease in prescribing of antidepressants to individuals younger than 18 years, the FDA actions have also resulted in a shift of care from generalists to psychiatrists. Although the number of depressed individuals younger than 18 years is small relative to the broader population, anecdotal evidence suggests overdemand for specialist services and, as a result, longer than historically observed waiting times for appointments. Finally, our analysis shows a slight shift in prescribing toward bupropion (a non-SSRI), which could stem in large part from physicians attributing the increased risk of suicidality primarily to SSRIs, even though bupropion is also labeled with a black box warning. Interestingly, we did not see any difference in trends within the SSRI class with respect to dosage or product selection, despite the fact that fluoxetine is the only SSRI formally approved by the FDA for the treatment of depression in children. This is perhaps a function of the fact that although some drugs demonstrated a weak association with suicidal signals, the FDA concluded that all drugs in the class carry the same black box warning.13 US psychiatrists, as represented by the American Psychiatric Association and the American Academy of Child and Adolescent Psychiatry, have expressed concern that “the FDA action may limit access to necessary, appropriate, and effective treatment for children and adolescents with depression, anxiety, and other psychiatric disorders.”14(p1) This is especially interesting given that a previous preliminary study by the American College of Neuropsychopharmacology Task Force on SSRIs and Suicidal Behavior in Youth found no increase in suicidality among young patients taking SSRIs and other effective new-generation antidepressants,10 and this has been confirmed and extended in their final report.15 Although we are not able to comment on whether the observed decrease in prescription volume is appropriate, our analyses allow for the conclusion that the FDA's actions have had an effect on prescribing volume for patients younger than 18 years, the specialty mix of physicians treating patients younger than 18 years with antidepressants, and the types of medications used in treating depression. In the current media environment in which safety concerns may be intensified because of several recent product recalls (eg, rofecoxib [Vioxx]; Merck & Co, Whitehouse Station, NJ, and natalizumab [Tysabri]; Biogen Idec, Cambridge, Mass), physician organizations (eg, the American Medical Association, the American Psychiatric Association, and the American Academy of Child and Adolescent Psychiatry) are concerned that the proved benefits of SSRI antidepressants may be underemphasized in discussions of potential risks and, as a result, there will be a decrease in access to appropriate treatment for children and adolescents.14,16 The FDA recently released results from an analysis that evaluated adult suicide and ideation data. The findings were mostly positive, and suggested that antidepressant drugs do not exacerbate suicidal thoughts in patients 30 years and older, but that the suicide thoughts/ideation seen in the pediatric data extends in young adults up to age 25 years.17 To date, these data results (both positive and negative) have received considerably less media attention in comparison with the release of the pediatric suicidality data. Recognizing that the results of the adult analysis were only public as of December 13, 2006, it remains to be seen if and how these findings will impact prescribing in both the 18 to 25 years population and the 26 years and older population. It is evident, however, that there is need for additional exploration into the relationship between FDA action, media reaction, and physician behavior change to ensure that dissemination of drug safety information does not interfere with appropriate access to care. Correspondence: Amir Kalali, MD, Quintiles, 10201 Wateridge Cir, San Diego, CA 92121 ([email protected]). Submitted for Publication: January 25, 2006; final revision received May 5, 2006; accepted May 25, 2006. Financial Disclosure: Dr Nemeroff has received grants from or performed research for the American Foundation for Suicide Prevention, AstraZeneca, Bristol-Myers Squibb, Forest Laboratories, Inc, Janssen Pharmaceutica, NARSAD: The Mental Health Research Association, the National Institute of Mental Health, Pfizer Pharmaceuticals, and Wyeth-Ayerst Laboratories; has been a consultant to Abbott Laboratories, Acadia Pharmaceuticals, Bristol-Myers Squibb, Corcept Therapeutics, Cypress Bioscience, Cyberonics, Eli Lilly and Co, Entrepreneur's Fund, Forest Laboratories, Inc, GlaxoSmithKline, i3 DLN, Janssen Pharmaceutica, Lundbeck, Otsuka America Pharmaceutical, Inc, Pfizer Pharmaceuticals, Quintiles Transnational, UCB Pharma, and Wyeth-Ayerst Laboratories; has been on the speakers bureau for Abbott Laboratories, GlaxoSmithKline, Janssen Pharmaceutica, and Pfizer Pharmaceuticals; is a stockholder in Acadia Pharmaceuticals, Corcept Therapeutics, Cypress Bioscience, and NovaDel Pharma Inc; is on the board of directors of the American Foundation for Suicide Prevention, the American Psychiatric Institute for Research and Education, the George West Mental Health Foundation, NovaDel Pharma Inc, and the National Foundation for Mental Health; holds patents on a method and devices for transdermal delivery of lithium (US 6,375,990 B1) and on a method to estimate serotonin and norepinephrine transporter occupancy after drug treatment using patient or animal serum (provisional filing April 2001); and holds equity in Reevax, BMC-JR LLC, and CeNeRx. Dr Kalali is on the advisory board or speakers bureau of AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb, GlaxoSmithKline, Janssen Pharmaceutica, Pfizer Inc, and Shire. Dr Keller has been a consultant to or has received honoraria from Collegium, Cypress Bioscience, Cyberonics, Eli Lilly and Co, Forest Laboratories, Inc, Janssen Pharmaceutica, Organon, Otsuka America Pharmaceutical, Inc, Pfizer Inc, Pharmastar, Sepracor, Vela Pharmaceuticals Inc, and Wyeth Pharmaceuticals; has received grants from or performed research for Eli Lilly and Co, Forest Laboratories, Inc, Pfizer Inc, and Wyeth Pharmaceuticals; and has been on the advisory board of Abbott Laboratories, Bristol-Myers Squibb, Cyberonics, Cypress Bioscience, Eli Lilly and Co, Forest Laboratories, Inc, GlaxoSmithKline, Janssen Pharmaceutica, Novartis, Organon, Pfizer Inc, Sepracor, and Wyeth Pharmaceuticals. Dr Charney has consulting agreements with Abbott Laboratories, AstraZeneca, Bristol-Myers Squibb, Cyberonics, Gene Logic Inc, the Institute of Medicine, Neurogen Corp, the Neuroscience Education Institute, Novartis Pharmaceuticals Corp, OREXIGEN Therapeutics, Inc, Organon International, Otsuka America Pharmaceutical, Inc, Quintiles Transnational, and Sepracore Inc; and has a confidentiality agreement with Forest Laboratories, Inc, and Novartis Pharmaceuticals Corp. Dr Schatzberg is a consultant to Eli Lilly and Co, Wyeth Pharmaceuticals, Corcept Therapeutics, Bristol-Myers Squibb, Novartis, Abbott Laboratories, Forest Laboratories Inc, Quintiles Transnational, and Lundbeck; is a cofounder of Corcept Therapeutics and has equity in Forest Laboratories, Pfizer Inc, and Merck and Co; and has received research funding from GlaxoSmithKline and Wyeth Pharmaceuticals. References 1. Center for Drug Evaluation and Research, US Food and Drug Administration, Background information on the Suicidality Classification Project. http://www.fda.gov/cder/drug/antidepressants/classificationProject.htmMarch 14, 2006 2. GlaxoSmithKline, US medical information letter: use of PAXIL or PAXIL CR in pediatric patients. http://www.gsk.com/media/paroxetine/letter.pdfMarch 13, 2006 3. Harris G Study finds less youth antidepressant use. http://www.nytimes.com/2004/09/21/business/21drug.html?ex=1169614800&en=83960e4e3bdeff4a&ei=5070#.March 15, 2006 4. Rosack J New data show declines in antidepressant prescribing. Psychiatric News September2 2005;40 (17) 1- 39Google Scholar 5. National Cancer Institute, US National Institute of Health, Joinpoint regression program. http://srab.cancer.gov/joinpointMarch 22, 2006 6. Martin RMMay MGunnell D Did intense adverse media publicity impact on prescribing of paroxetine and the notification of suspected adverse drug reactions? analysis of routine databases, 2001-2004. Br J Clin Pharmacol 2006;61224- 228Google ScholarCrossref 7. Vector One: National (VONA): TRx MQT: August 2000 to March 2005. Verispan Web site.http://www.verispan.com/productsJanuary 26, 2007 8. Vector One: National (VONA): TRx MQT: June 2002 to March 2005. Verispan Web site.http://www.verispan.com/productsJanuary 26, 2007 9. Physician Drug & Diagnosis Audit: TRx QTR: February 2004, QTR February 2005. Verispan Web site.http://www.verispan.com/products January 26, 2007 10. American College of Neuropsychopharmacology, Executive summary: preliminary report of the Task Force on SSRIs and Suicidal Behavior in Youth: January 21, 2004. http://www.acnp.org/Docs/ACNP%20Task%20Force%20Report%20on%20SSRIs%20and%20Suicide%20in%20Youth.pdf June 15, 2004 11. Medco Health Solutions, Inc, News release page: FDA warning on pediatric antidepressants results in significant reduction in use. http://phx.corporate-ir.net/phoenix.zhtml?c=131268&p=irol-newsArticle&ID=616944&highlight= June 21, 2004 12. Harris G Doctors say they will cut antidepressant use. http://query.nytimes.com/gst/fullpage.html?res=980CE6DD1F30F935A2575AC0A9629C8B63&sec=&spon=&pagewanted=all March 15, 2006 13. Stong C FDA black box warning on antidepressants creates concerns for clinicians. NeuroPsychiatry Rev. December2004;5- 9Google Scholar 14. ParentsMedGuide.org Web site. http://www.parentsmedguide.org June 14, 2006 15. Mann JJEmslie GBaldessarini RJBeardslee WFawcett JAGoodwin FKLeon ACMeltzer HYRyan NDShaffer DWagner KD ACNP Task Force report on SSRIs and suicidal behavior in youth. Neuropsychopharmacology 2006;31473- 492Google ScholarCrossref 16. Japsen B Do these drugs need a warning: the FDA says yes, but doctors who disagree are taking their case to the AMA. Chicago Tribune June9 2005;Google Scholar 17. Minderd J FDA advisers want antidepressant suicidality balck-box warning for young adults. www.medpagetoday.com/tbpring.cfm?tbid=4702&topicid=183January 19, 2007

Neighbors, Harold W.;Caldwell, Cleopatra;Williams, David R.;Nesse, Randolph;Taylor, Robert Joseph;Bullard, Kai McKeever;Torres, Myriam;Jackson, James S.

doi: 10.1001/archpsyc.64.4.485pmid: 17404125

Abstract Context Little is known about differences in the unmet need for mental health service use between African Americans and Caribbean blacks. Objective To extend the National Survey of Black Americans by examining 12-month mental health service use for African Americans and Caribbean blacks from the recently completed National Survey of American Life. Design and Setting National household probability samples of noninstitutionalized African Americans and Caribbean blacks (blacks from Caribbean area countries now living in the United States) conducted between February 2001 and June 2003, using a slightly modified World Mental Health version of the World Health Organization's Composite International Diagnostic Interview. Participants A total of 3570 African Americans and 1621 Caribbean blacks 18 years and older (N = 5191). Main Outcome Measures Proportion of respondents with 12-month DSM-IV disorders who sought help in the specialty mental health, general medical, human service, and complementary-alternative medicine treatment sectors. The percentage receiving minimally adequate treatment was also assessed. Results Overall, 10.1% of respondents used some form of mental heath care services in the past year. Use of services was much higher among those who met criteria for a 12-month DSM-IV disorder (31.9%) than among those who did not (5.4%). Forty-nine percent of respondents with serious mental illness used services, whereas 39.3% had contact with mental health care specialists. The youngest and oldest age groups were least likely to obtain any services. Among African Americans, women were more likely than men to use general medical care and services from any sector. Respondents with the most years of education showed the highest use of services. Conclusions The underuse of mental health services among black Americans remains a serious concern. Educational interventions that focus on both consumers and mental health care professionals are needed. Twenty-five years ago, the National Survey of Black Americans (NSBA) produced the first national data on how symptoms of distress are defined and responded to by black Americans.1,2 The NSBA found that most black Americans did not seek mental health services in response to emotional distress. Predating the DSM-III, the mental health need-assessment approach taken by the NSBA grew out of an epidemiologic tradition that emphasized how variation in personal problem definitions is related to patterns of help-seeking behavior.3 Interestingly, because personal distress was defined from a lay community perspective and not within a medical diagnostic taxonomy, it was difficult to draw firm conclusions about the extent of unmet need for mental health treatment on the basis of the NSBA.4,5 In this article, we use data from the recently conducted National Survey of American Life (NSAL) to examine help seeking for mental disorders in an ethnically diverse sample of black Americans. The NSAL extends the NSBA in 2 important ways. First, the NSAL uses the Composite International Diagnostic Interview to estimate service use among persons with DSM-IV criteria for selected mental disorders. Second, it addresses the issue of black ethnic variation by including samples of both African Americans and Caribbean blacks. It is estimated that Caribbean-descended and immigrant groups constitute 10% to 15% of the United States' black population. Studies6 of multiple racial and ethnic groups reveal that groups of color are as likely to differ from each other as they are to differ from white Americans. Unfortunately, no studies have addressed black ethnic variation in help seeking for mental disorders within the United States. The few studies7-9 that have examined help seeking among Caribbean blacks have been conducted in the United Kingdom. As a result, many questions remain unanswered regarding whether Caribbean blacks and African Americans actually differ in mental health service use. Studying blacks of different ethnic origins is important for public mental health service professionals because of questions about the contribution of culture to population group differences in behavior.10-13 Hypotheses related to assumed differences in such social processes as group identity, acculturation, nativity, and immigration suggest that sociodemographic factors have differential effects on treatment seeking across different ethnic groups.14-17 Although such an initial demographic analysis cannot speak directly to culture, it begins to identify directions for future research on differences in psychosocial processes related to culture and mental health.18 Given the virtual absence of findings in this area, we take an exploratory, descriptive approach to this first article on mental health services. Nevertheless, on the basis of findings from the general services literature and our previous work with African Americans, we have some expectations. We predict that Caribbean blacks will be less likely than African Americans to use medical and mental health services. We predict significant differences in the use of services for other demographic variables, although we are unsure about how uniform these relationships will be across the 2 ethnic groups and across the multiple service domains explored. Specifically, we predict that both income and education will show a positive relationship with use of services, that women will be more likely than men to use services, that insured people will be more likely than uninsured people to seek professional help, and that the oldest respondents (≥65 years) will be least likely to use services. In summary, the NSAL is an excellent resource to explore the extent to which both groups receive mental health services and the nature of ethnic differences in the unmet need for mental health care. No national studies have measured the prevalence of mental disorders in conjunction with help seeking in representative national samples of both African American and Caribbean blacks. This article describes the use of general medical, specialty mental health care, human services, and complementary-alternative medical resources for mental health problems and selected, discrete mental disorders. Methods Sample The NSAL was part of a National Institute of Mental Health Collaborative Psychiatric Epidemiology Surveys initiative that also included the National Comorbidity Survey Replication (NCS-R) and the National Latino and Asian American Study.19 The NSAL was an integrated national household probability sample of 3570 African Americans and 1621 blacks of Caribbean descent 18 years and older. The African American sample was selected exclusively from geographic segments in proportion to the African American population; the Caribbean black sample was selected from the African American segments and additional metropolitan segments in which blacks of Caribbean descent made up more than 10% of the population.20 In both the African American and Caribbean black samples, it was necessary for respondents to self-identify their race as black. Those self-identifying as black were included in the Caribbean black sample if they answered affirmatively to any of these inclusion criteria: (1) West Indian or Caribbean descent, (2) from a Caribbean area country, and/or (3) parents or grandparents were born in a Caribbean area country. Most interviews (88%) were conducted face to face and 12% by telephone, using a computer-assisted instrument and lasting an average of 2 hours 20 minutes. Data collection was completed between February 2, 2001, and June 30, 2003. The overall response rate was 72.3%: 70.7% for African Americans and 77.7% for Caribbean blacks. Measures Diagnostic Assessment We measured DSM-IV disorders, both lifetime and 12 month, with the World Mental Health Composite International Diagnostic Interview, a structured diagnostic interview; mental disorders sections were modified versions of those developed for the World Mental Health project.21 The 18 twelve-month mental disorders assessed were as follows: anxiety disorders (panic disorder, agoraphobia, social phobia, generalized anxiety disorder, posttraumatic stress disorder, and obsessive-compulsive disorder, which was assessed using the Composite International Diagnostic Interview Short Form),22 mood disorders (major depressive disorder, dysthymia, and bipolar I and II disorders), substance disorders (alcohol abuse, alcohol dependence, drug abuse, and drug dependence), childhood disorders (oppositional defiant disorder, conduct disorder, and attention-deficit/hyperactivity disorder, asked only of respondents in the 18- to 44-year age range), and eating disorders. Severity of Mental Disorder Respondents who reported 12-month suicidal ideation or attempts, who had at least 1 nonaffective psychotic symptom plus ever being treated for psychosis, or who met 12-month criteria for at least 1 disorder were divided into 1 of 3 severity gradients: serious, moderate, or mild. Severity was primarily assessed using measures of role impairment derived from the Sheehan Disability Scale.23 The significant positive relationship between the severity measure and 30-day disability, ranging from a low of 0.96 disability day for respondents with mildly severe mental disorders to more than 5 disability days for those with serious mental disorders, speaks to the validity of the disability measure. Service Use Respondents were asked if they had made contact with anyone from a list of health care professionals for problems with their emotions, nerves, mental health, or use of alcohol or drugs in the past 12 months. Health care professionals were categorized into a mental health sector (psychiatrists, psychologists, counselors and social workers seen in mental health settings, other mental health care professionals, and mental health hotlines) and a general medical sector (general physicians, family physicians, physician specialists, nurses, occupational therapists, and other health care professionals). The term nonpsychiatrist refers to psychologists, counselors, and social workers seen in a mental health care setting. The non–health care sector included human services (religious and spiritual advisers and counselors and social workers seen in non–mental health settings) and complementary-alternative medicine (herbalists, chiropractors, spiritualists, self-help groups, and Internet support groups). Twelve-month service use was defined as making at least 1 visit to a service provider within the 12 months before the interview. Minimally Adequate Treatment Minimally adequate treatment was defined separately for each 12-month disorder in a manner consistent with that used in the NCS-R,24 as reporting either (1) at least 4 visits with any physician and receiving appropriate pharmacotherapy for at least 60 days during the past year or (2) at least 8 psychotherapy visits, each averaging 30 minutes or more, with any other health care professional within the health care or human services treatment sectors. Complementary-alternative medicine was considered adequate only for substance disorders and only if respondents attended at least 8 self-help sessions of any duration during the past year. Appropriate pharmacotherapy for disorders included antidepressants for depression and dysthymia, mood stabilizers or antipsychotics for bipolar disorders, antidepressants or benzodiazepines for anxiety disorders, and disulfiram for substance disorders. Sociodemographic Correlates Sociodemographic correlates include race/ethnicity (African American or Caribbean black), age (18-29, 30-44, 45-59, or ≥60 years), sex, highest level of education attained (0-11, 12, 13-15, or ≥16 years), marital status (married or cohabiting, previously married, or never married), household income (<$18 000, $18 000-$31 999, $32 000-54 999, or ≥$55 000), employment status (working vs not working), and whether the respondent had health insurance. Analysis strategy Cross-tabulations are presented to illustrate ethnic differences in 12-month service use. The Rao-Scott χ2 represents a complex design-corrected measure of association. Logistic regression was used to examine the main effect of ethnicity on service use, adjusted for demographic variables and having any 12-month DSM-IV disorder. To account for multiple comparisons, χ2 values were estimated for the overall type III effects of each categorical predictor variable within the contexts of the multivariate models. Standard errors and 95% confidence intervals reported in this article reflect adjustment for the sampling design. Unless otherwise stated, P <.05 on a 2-sided design–based test of significance represented the cutoff for assessing statistical significance. All analyses were conducted using SAS statistical software, version 9.13, which uses the Taylor expansion approximation technique for calculating the complex design–based estimates of variance.25 Since the NSAL used a multistage sample design, involving both clustering and stratification, specialized statistical techniques to account for the complexity of the design and associated standard errors were used. Standard errors calculated on the basis of a simple random sample would not reflect the true variation of estimates in the NSAL, resulting in an increased likelihood of type I errors (declaring a result to be significant when it is not). Because standard errors adjusted for complex design are usually larger than nonadjusted standard errors, differences may appear to be large yet not statistically significant. Furthermore, the Caribbean black sample is significantly more clustered than the African American sample, so the standard errors for the Caribbean black sample are usually higher than those for the African American sample when correctly estimated. Results Table 1 focuses on the demographic correlates of 12-month service use in response to problems with emotions, nerves, mental health, or use of alcohol or other drugs in the past 12 months. Overall, 442 (10.1%) of the NSAL respondents used some form of services for mental heath care in the past year. African Americans and Caribbean blacks differ in the use of nonhealth services (132 [4.1%] and 40 [1.8%], respectively), with African Americans more likely to use help. The youngest and oldest age groups are least likely to obtain any services in response to mental health problems. Women are more likely than men, those not married are more likely than married individuals, and those working are more likely than those not working to use any services. Table 1 also gives the demographic correlates of use for African Americans and Caribbean blacks. Among African Americans, age is related to the use of all service sectors but only to the use of psychiatrists for Caribbean blacks. Among African Americans but not Caribbean blacks, women are more likely than men to contact general medical care, nonhealth sectors, or any services. Both African Americans and Caribbean black respondents with 16 or more years of education have the highest use of nonpsychiatric mental health professionals. Among African Americans, previously married respondents report more use of services than those who are currently married or living with their partner. Employed African Americans are more likely than those not working to use all service sectors except for nonpsychiatric and nonhealth sectors. Insured Caribbean blacks are more likely than uninsured individuals to use psychiatrists or all services combined. Insurance coverage has no influence on the use of services by African Americans. Table 2 indicates use of services by sex and ethnicity by level of mental disorder severity. Although only 179 respondents (4.8%) without mild, moderate, or serious mental disorder use any services, 84 respondents (48.8%) with serious disorder use any services. A similar relationship is seen for each service sector. Examining those with serious disorders, similar percentages of both African Americans (50 [39.1%]) and Caribbean blacks (17 [41.5%]) obtain help from any mental health service. Within the mental health services sector, however, a much higher percentage of African Americans compared with Caribbean blacks (42 [34.4%] and 14 [18.6%], respectively) seek help from psychiatrists. The reverse occurs for the use of nonpsychiatrist mental health professionals; 10 Caribbean blacks (37.5%) and 27 African Americans (19.4%) obtained help from these types of professionals for serious disorders. Roughly comparable percentages of both ethnic groups with serious disorders seek help from the general medical care sector. Among African Americans with serious disorders, a higher percentage of men than women use both psychiatrists (19 [43.7%] vs 23 [27.9%]) and nonpsychiatrist mental health therapists (10 [24.8%] vs 17 [15.7%]). On the other hand, a higher percentage of women than men with serious mental illness seek the help of general medical care professionals. For Caribbean blacks, the opposite occurs; women use more mental health services than men, but men use more medical services than women. Table 3 presents service sector use for each disorder separately for African Americans and Caribbean blacks. Use of services was much higher among those who met criteria for a 12-month DSM-IV disorder than among those who did not; 238 (31.9%) of those with a disorder obtained some type of help, whereas only 204 (5.4%) of those without a disorder did so. The use of any services for any mood disorder is higher for African Americans: 92 (43.5%) compared with 30 Caribbean blacks (22.9%). The same is true for the use of psychiatrists (37 [17.8%] and 11 [4.0%], respectively) and general medical care (45 [21.0%] vs 6 [12.5%], respectively). A sizeable percentage of Caribbean blacks, however, use psychiatrists for bipolar disorder (6 [16.1%]), which is much more than for dysthymia (1 [1.1%]) and major depression (9 [4.1%]). Comparatively large percentages of African Americans use psychiatrists for major depression (32 [18.6%]), dysthymia (11 [20.9%]), and bipolar disorder (10 [20.7%]). Similar differences are found between African Americans and Caribbean blacks in the use of psychiatrists for any anxiety disorder (44 [14.4%] vs 13 [4.0%]), but again, not for nonpsychiatric mental health professionals, from whom 41 African Americans (14.3%) and 12 Caribbean blacks (16.4%) received care. African Americans and Caribbean blacks are more similar in the use of any health services for any anxiety disorder (87 [28.6%] and 28 [29.4%], respectively) but not for any mood disorder (73 [34.6%] and 25 [20.4%], respectively). In general, Caribbean blacks are more likely to obtain mental health care from nonpsychiatrist mental health professionals than from psychiatrists for each disorder type. These differences are not present for African Americans. Table 4 gives the results of multivariate logistic regression analyses that estimated the effect of ethnicity and other demographic measures to each service use sector, adjusting for any 12-month mental disorder. Ethnicity is not related to specialty mental health service use. African Americans, however, are 2.7 times more likely than Caribbean blacks to use non–health care services. Table 4 also indicates that the use of any services is associated with being 30 to 44 and 45 to 59 years old, female, and insured and having 16 or more years of education. Age is similarly related to the use of any mental health services. Those 18 to 29 years old are significantly less likely than the older age groups to use general medical care for treatment of mental problems. Women are more likely than men to use general medical care and any non–health care services. Those with insurance are more likely than the uninsured to use a psychiatrist or any health services. Previously married respondents are more likely than the married and never married to use non–health care services. Those with the highest level of education were more likely to use all health-related services sectors than those with lower educational levels. Having a disorder increases significantly the use of all service sectors. Table 5 indicates the proportion of African Americans and Caribbean blacks who are receiving minimally adequate treatment by service sectors. Overall, 63 (26.2%) received minimally adequate treatment, but the percentages varied noticeably in the service sector, ranging from 18 (10.5%) in the general medical sector to 56 (30.0%) for any mental health care services (41 [29.3%] to psychiatrists and 28 [28.5%] to nonpsychiatric mental health care professionals). The percentages of patients who are receiving minimally adequate treatment are higher for Caribbean blacks than for African Americans, but large standard errors make conclusions about Caribbean blacks problematic. Comment The NSAL has several strengths. First, the NSAL assesses the presence of mental disorders, thereby addressing a major limitation of data gathered in previous mental health surveys that focused on black Americans. Second, the study includes a large representative sample that permits the identification of mental health differences among groups often lumped together within the black American population. These types of analyses are critical because of changing immigration patterns and diverging socioeconomic conditions that have occurred within the black population in the last 25 years. Third, our study used novel geographical screening procedures that ensured that every African American household in the continental United States had a known probability of selection.1,26,27 In addition, new methods were developed to ascertain the influences of structurally missing members of black households (eg, young men in prisons) on sampling and disorder estimates.1 Fourth, all respondents were selected from the targeted geographic segments in proportion to the African American and Caribbean black population, making this the first national sample of people of different racial and ethnic groups who live in the same contexts and geographical areas (high- and low-density, urban and rural areas). In addition to these strengths, a few limitations should be noted. First, the World Mental Health Composite International Diagnostic Interview does not include the DSM-IV diagnoses of schizophrenia or other nonaffective psychoses. In addition, the NSAL did not collect data on specific phobias or intermittent explosive disorder. Second, because homeless and institutionalized individuals were not included, prevalence estimates are likely lower than reported. Those who did not speak English were not included, which underrepresents French-, Spanish-, and Creole-speaking Caribbean blacks. These groups are relatively rare in the United States, so their exclusion should have minimal effects. Another limitation is the global nature of our designation of Caribbean ancestry, which is characterized by heterogeneity that we were not able to fully explore owing to sample size limitations. Because of the high clustering of Caribbean blacks and a relatively smaller sample size, their adjusted standard errors are sometimes large. A final limitation is systematic nonresponse to various questions. Some nonrespondents may have met the criteria for a DSM-IV disorder. Despite the popular view that African Americans and Caribbean blacks represent different cultural heritages, they did not differ much in the use of services. Differences observed between these 2 ethnic groups were largely due to different relationships among demographic groups to mental health service use. This finding suggests the presence of interactions among ethnicity, use of services, and a third demographic variable. We tested the effect of sex and ethnicity on any services use and found that the relationship of sex to use depended on ethnicity. African American women were significantly more likely to use services than African American men. No sex effect was found among Caribbean blacks (data available from the author). The absence of sex effects on use among Caribbean blacks is surprising, since black women are typically more likely to use services than black men.28 The idea that requesting help is antithetical to male socialization may not be uniform across all black men. Future work will focus more explicitly on ethnic differences in the social construction of masculine identities. The findings showed that age differences in mental health care use deserve attention in future analyses of the NSAL. The youngest and the oldest groups, especially among African Americans, used services the least. These age differences are consistent with other research in this field.18,29 Age at onset of mood disorders, which tends to occur at approximately 30 years, may account at least partially for the lower use among young people.30 Older respondents are known to underuse mental health services because of greater perceived stigma. There was little evidence that respondents with higher incomes are more likely to use services. Education, on the other hand, showed a positive relationship with service use. These findings are consistent with the notion that, although related, income and education capture distinctive aspects of socioeconomic position.31 Education is likely a proxy for knowledge, greater attentiveness to mental health information, and awareness of the availability and acceptability of seeking help for mental health problems.32 Differential access to services based on income may be less striking in this sample because of working people having health insurance and poor people having Medicaid.17 Placing income and education in the same model may account for some of the same variance, and the effects of income may be mediated through education. The lack of an income effect might also be attributed to the sizeable proportions of both African Americans (65.9%) and Caribbean blacks (62.9%) who had mental health care insurance. This level of insurance coverage is comparable to that in whites, and as a result, statistical power was not a problem on the basis of a restricted range of insurance coverage. There may be more of an insurance effect than we were able to capture given our additive modeling approach. Clearly, we need to know more about how both socioeconomic status/position and insurance in combination affect use across all service sectors. African Americans and Caribbean blacks who sought professional help for mental health problems used general medical care almost as much as specialty mental health care. The relative accessibility of primary care physicians and the limitations that most health insurance plans put in place to control the use of specialty mental health care make this the most likely pattern of use.33-35 The large percentages of African Americans and Caribbean blacks who go to their primary care physicians for help with mental health problems might be receiving inappropriate levels of care. We believe that professionals trained especially to deal with mental health problems (ie, psychiatrists, psychologists, and social workers) are best suited to handle the treatment of these mental disorders. Although not the focus of the present study, the proportion of NSAL respondents who obtained 12-month service use (10.1%) is noticeably lower than the percentage reported by the NCS-R (17.9%).24 This finding is compelling evidence that the black-white difference in the use of mental health services remains an issue worthy of more in-depth investigation. Differences were also found in the sociodemographic correlates of 12-month service use between the NSAL and the NCS-R; specifically, not having a low family income, previously being married, and not living in a rural area. Income and marital status were not significant predictors of use in the NSAL, whereas education was. These patterns suggest interesting interactions among race, sociodemographic predictors, and service use that can be explored once the NSAL and NCS-R data are merged. Some findings in the literature suggest that although the black-white gap in use may be narrowing,36 racial disparities may occur in the quality of mental health treatment.37 Rates of minimally adequate treatment are lower in the NSAL (26.2%) compared with the NCS-R (32.7%). Although the level of minimally adequate treatment provided by the general medical sector is comparable across the 2 studies (10.5% and 12.7%, respectively), minimally adequate treatment received from psychiatrists is noticeably lower in the NSAL (29.3% and 44.5%, respectively).24 Such differences in treatment adequacy are worthy of attention in future work. Many black Americans who do not use services rely on help from informal support networks and alternative helpers, such as ministers.38,39 We were not able, in this first article, to address specifically the role of faith-based organizations and particularly the helping role of clergy, which our previous work has shown to be important.40 We have begun to explore the use of clergy, and preliminary results indicate a much higher clergy use for mood and anxiety disorders among African Americans than Caribbean blacks. The more pressing policy question, however, is whether the seriousness of the emotional challenges confronting all black Americans is appropriately matched with the help sources to which these groups turn. Many mental disorders require the attention of trained mental health care professionals. Despite the positive aspects of informal help, social support is as much a barrier to mental health care as an acceptable treatment alternative.40-43 The mental health need-assessment tradition from which the NSAL flows relied more on lay conceptualizations of distress than on professional judgments of need. Although good clinical, scientific, and policy reasons exist for the development of highly structured survey instruments that can classify respondents by DSM-IV criteria, this should not be the only approach to assessing need for mental health services; not everyone in need of mental health treatment meets the criteria for a disorder,37 and meeting these criteria may not be serious enough to warrant treatment.30,44 People decide to seek professional help not because they know that they have a particular disorder but because the level of distress experienced has exhausted the personal and social resources used to cope with the emotional pain.44 The NSAL embraced each of these epidemiologic traditions, and future work will explore both the lay taxonomy that motivates the search for help and how well the conceptualization of distress represented by the DSM-IV predicts the need for services.45-48 Our findings demonstrate that underuse of mental health services for both African Americans and Caribbean blacks remains a serious concern. As a result, educational interventions that focus on both black consumers and mental health care professionals are needed. Primary care physicians need to be educated on how best to identify black individuals with serious mental health problems and disorders. Mental health care professionals must incorporate knowledge about ethnic differences in idioms of distress and how to overcome feelings of mistrust into their therapeutic approach. Mental health educational programs must facilitate, among black consumers, the recognition and definition of symptom clusters that need to be treated by mental health care professionals. Clearly, ways must be found to increase the use of mental health care and to increase the quality of that care among all black groups, irrespective of their ethnic heritage. The consequences in terms of needless pain and suffering and unnecessary losses in productivity are too great to ignore. Correspondence: Harold W. Neighbors, PhD, Program for Research on Black Americans, Research Center for Group Dynamics, The Institute of Social Research, Room 5067, University of Michigan, PO Box 1248, 426 Thompson St, Ann Arbor, MI 48106-1248 ([email protected]). Submitted for Publication: November 3, 2005; final revision received March 21, 2006; accepted May 2, 2006. Financial Disclosure: None reported. Funding/Support: The NSAL is supported by the National Institute of Mental Health (grant U01-MH57716) with supplemental support from the Office of Behavioral and Social Science Research at the National Institutes of Health and the University of Michigan. Dr Neighbors is also supported by a Robert Wood Johnson Foundation Investigator Award in Health Policy Research (grant 050594). Acknowledgment: We appreciate the assistance provided in all aspects of the NSAL study by the Program for Research on Black Americans faculty and research staff, including Jamie Abelson, MSW, Raymond Baser, MS, Deborah Coral, BA, Carl Hill, PhD, Lisa Martin, MPH, Carmel Salhi, BA, Phyllis Stillman, BA, and Julie Sweetman, MS. We thank the staff at the Survey Research Center's Survey Research Operations department for their assistance with instrumentation and fieldwork for the NSAL study. References 1. Jackson JS Life in Black America. Newbury Park, Calif: Sage Publications; 1991; 2. Neighbors HWJackson JS Mental health in Black America: psychosocial problems and help seeking behavior. Neighbors HWJackson JSeds. Mental Health in Black America. Thousand Oaks, Calif: Sage;1996;1- 13Google Scholar 3. Neighbors HWJackson JS The use of informal and formal help: four patterns of illness behavior in the black community. Am J Community Psychol 1984;12629- 644PubMedGoogle ScholarCrossref 4. Gurin GVeroff JFeld S Americans View Their Mental Health. New York, NY Basic Books 1960; 5. Neighbors HW Seeking professional help for personal problems: black Americans' use of health and mental health services. Community Ment Health J 1985;21156- 166PubMedGoogle ScholarCrossref 6. Borrell LNLynch JNeighbors HWGillespie B Is there homogeneity in periodontal health between African Americans and Mexican Americans? Ethn Dis 2002;1297- 110PubMedGoogle Scholar 7. Bhui KStandfield SHull SPriebe SMole FFeder G Ethnic variation in pathways to and use of specialist mental health services in the UK. Br J Psychiatry 2003;182105- 116PubMedGoogle ScholarCrossref 8. Mclean CCampbell CCornish F African-Caribbean interactions with mental health services in the UK: experiences and expectations of exclusion as reproductive of health inequalities. Soc Sci Med 2003;56657- 669PubMedGoogle ScholarCrossref 9. Morgan CMallett RHutchnson GLeff J Negative pathways to psychiatric care and ethnicity: the bridge between social science and psychiatry. Soc Sci Med 2004;58739- 752PubMedGoogle ScholarCrossref 10. Airhihenbuwa CO Health and Culture: Beyond the Western Paradigm. Thousand Oaks, Calif Sage 1995; 11. Sussman LKRobins LNEarls F Treatment-seeking for depression by Black and White Americans. Soc Sci Med 1987;24187- 196PubMedGoogle ScholarCrossref 12. US Department of Health and Human Services, Mental Health: Culture, Race, and Ethnicity: A Supplement to Mental Health: A Report to the Surgeon General. Rockville, Md: US Dept of Health and Human Services; 2001; 13. New Freedom Commission on Mental Health, Transforming Mental Health Care in America: Final Report. Rockville, Md: US Dept of Health and Human Services; 2003.;SMA-03-3832 14. Atdjian SVega WA Disparities in mental health treatment in U.S. racial and ethnic minority groups: implications for psychiatrists. Psychiatr Serv 2005;561600- 1602PubMedGoogle ScholarCrossref 15. Halpern D Minorities and mental health. Soc Sci Med 1993;36597- 607PubMedGoogle ScholarCrossref 16. Jackson JSTorres MCaldwell CHNeighbors HWNesse RMTaylor RJTrierweiler SJWilliams DR The National Survey of American Life: a study of racial, ethnic, and cultural influences on mental disorder and mental health. Int J Methods Psychiatr Res 2004;13196- 207PubMedGoogle ScholarCrossref 17. Snowden LR Barriers to effective mental health services for African Americans. Ment Health Serv Res 2001;3181- 187PubMedGoogle ScholarCrossref 18. Snow LF Folk medical beliefs and their implications for care of patients: a review base on studies among Black Americans. Ann Intern Med 1974;8182- 96PubMedGoogle ScholarCrossref 19. Colpe LMerikangas KCuthbert BBourdon K National Institute of Mental Health [guest editorial]. Int J Psychiatr Res 2004;13193- 195Google ScholarCrossref 20. Jackson JSNeighbors HWNesse RTrierweiler STorres M Methodological innovations in the National Survey of American Life. Int J Methods Psychiatr Res 2004;13289- 298PubMedGoogle ScholarCrossref 21. Kessler RCMerikangas KR The national co-morbidity survey replication (NCS-R): background and aims. Int J Methods Psychiatr Res 2004;1360- 68PubMedGoogle ScholarCrossref 22. Kessler RCAndrews GMroczek DUstun BWittchen H The World Health Organization Composite International Diagnostic Interview Short Form (CIDI-SF). Int J Methods Psychiatr Res 1998;7171- 185Google ScholarCrossref 23. Leon ACOlfson MPortera LFarber LSheehan D Assessing psychiatric impairment in primary care with the Sheehan Disability Scale. Int J Psychiatry Med 1997;2793- 105PubMedGoogle ScholarCrossref 24. Wang PSLane MOlfson MPincus HAWells KBKessler RC Twelve-month use of mental health services in the United States. Arch Gen Psychiatry 2005;62629- 640PubMedGoogle ScholarCrossref 25. SAS Institute Inc, SAS/STAT User's Guide, Version 9.1. Cary, NC: SAS Institute;2005; 26. Bowman JT Conceptual and methodological problems in survey research on Black Americans. Liu WT ed Methodological Problems in Minority Research Chicago, Ill: Pacific/Asian American Mental Health Research Center;1982;Google Scholar 27. Hess I Sampling for Social Surveys: 1947-1980. Ann Arbor: University of Michigan Institute for Social Research;1985; 28. Neighbors HWHoward CS Sex differences in professional help use among adult blacks. Am J Community Psychol 1987;15403- 417PubMedGoogle ScholarCrossref 29. Snowden LRPingitore D Frequency and scope of mental health service delivery to African Americans in primary care. Ment Health Serv Res 2002;4123- 130PubMedGoogle ScholarCrossref 30. Kessler RCBerglund PDemler OJin RWalters E Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the national comorbidity survey replication. Arch Gen Psychiatry 2005;62593- 602[published correction appears in Arch Gen Psychiatry 2005;62768Merikangas, Kathleen R, added]PubMedGoogle ScholarCrossref 31. Krieger NWilliams DRMoss N Measuring social class in U.S. public health research: concepts, methodologies, and guidelines. Annu Rev Public Health 1997;18341- 378PubMedGoogle ScholarCrossref 32. Williams DR Socioeconomic differentials in health: a review and redirection. Soc Psychol Q 1990;5381- 99Google ScholarCrossref 33. Daley MC Race, managed care, and the quality of substance abuse treatment. Adm Policy Ment Health 2005;32457- 476PubMedGoogle ScholarCrossref 34. Greenberg GARosenheck RA Change in mental health services delivery among blacks, whites, and Hispanics in the department of Veterans Affairs. Adm Policy Ment Health 2003;3131- 43PubMedGoogle ScholarCrossref 35. Grembowski DEMartin DPatrick DLDiehr PKaton WWilliams BEngelberg RNovak LDickstein DDeyo RGoldberg HI Managed care, access to mental health specialists, and outcomes among primary care patients with depressive symptoms. J Gen Intern Med 2002;17258- 269PubMedGoogle ScholarCrossref 36. Cooper-Patrick LGallo JJPowe NRSteinwachs DMEaton WWFord DE Mental health utilization by African Americans and whites: the Baltimore Epidemiologic Catchment Area Follow-up. Med Care 1999;371034- 1045PubMedGoogle ScholarCrossref 37. Institute of Medicine, Improving the Quality of Health Care for Mental and Substance-Use Conditions: Committee on Crossing the Quality Chasm: Adaptation to Mental Health and Addictive Disorders: Board on Health Care Services, Institute of Medicine. Washington, DC: National Academies of Press; 2006; 38. Taylor RJChatters LM Extended family networks of older black adults. J Gerontol 1991;46S210- S217PubMedGoogle ScholarCrossref 39. Taylor RJHardison CBChatters L Kin and non-kin as sources of informal assistance. Neighbors HWJackson JSeds. Mental Health in Black America. Newbury Park, Calif: Sage Publishers;1996;130- 145Google Scholar 40. Neighbors HWMusick MWilliams DR The African American minister as a source of help for serious personal crises: bridge or barrier to mental health care? Health Educ Behav 1998;25759- 777PubMedGoogle ScholarCrossref 41. Friedson E Client control and medical practice. Am J Surg 1960;65374- 382Google Scholar 42. Pescosolido BA Illness careers and network ties: a conceptual model of utilization and compliance. Adv Med Sociol 1991;2161- 184Google Scholar 43. Pescosolido BABoyer CHorwitz AVScheid TL How do people come to use mental health services? Horwitz AVScheid TLeds. Handbook for the Study of Mental Health. Cambridge, England: Cambridge University Press;1999;392- 411Google Scholar 44. Mechanic D Is the prevalence of mental disorders a good measure of the need for services? Health Aff (Millwood) 2003;228- 20PubMedGoogle ScholarCrossref 45. Flewelling RLEnnett STRachal JVTheisen AC National Household Survey on Drug Abuse: Race/Ethnicity, Socioeconomic Status and Drug Abuse 1991. Rockville, Md: Substance Abuse and Mental Health Services Administration, Office of Applied Studies;1993.;DHHS publication SMA 93-2062 46. Regier DANarrow WE Defining clinically significant psychopathology with epidemiologic data. Helzer JEHudiak JJeds. Defining Psychopathology in the 21st Century: DSM-V and Beyond. Washington, DC: American Psychopathological Association;2002;19- 30Google Scholar 47. Wakefield JSpitzer R Why requiring clinical significance does not solve epidemiology's and DSM's validity problem: response to Regier and Narrow. Helzer JEHudiak JJeds. Defining Psychopathology in the 21st Century: DSM-V and Beyond. Washington, DC: American Psychopathological Association2002;Google Scholar 48. Woodward A Access to substance abuse treatment and mental health services: A literature review. Council CL ed Health Services Utilization by Individuals with Substance Abuse and Mental Disorders. Rockville, Md: Substance Abuse and Mental Health Services Administration;2004.;DHHS publication SMA 04-3949Google Scholar

Wakefield, Jerome C.;Schmitz, Mark F.;First, Michael B.;Horwitz, Allan V.

doi: 10.1001/archpsyc.64.4.433pmid: 17404120

Abstract Context Symptoms of intense bereavement-related sadness may resemble those of major depressive disorder (MDD) but may not indicate a mental disorder. To avert false-positive diagnoses, DSM criteria for MDD exclude uncomplicated bereavement of brief duration and modest severity. However, the DSM does not similarly exempt depressive reactions to other losses, even when they are uncomplicated in duration and severity. Objective To test the validity of the DSM exclusion of uncomplicated depressive symptoms only in response to bereavement but not in response to other losses. Design Community-based epidemiological study. Participants From the National Comorbidity Survey (NCS) of 8098 persons aged 15 to 54 years representative of the US population, we identified individuals who met MDD symptom criteria and whose MDD episodes were triggered by either bereavement (n = 157) or other loss (n = 710). Intervention We divided the bereavement and other loss trigger groups into uncomplicated and complicated cases by applying the NCS algorithm for uncomplicated bereavement to the reactions to other losses. We then compared uncomplicated bereavement and uncomplicated reactions to other losses on a variety of disorder indicators and symptoms. Main Outcome Measures Nine disorder indicators, as follows: number of symptoms, melancholic depression, suicide attempt, duration of symptoms, interference with life, recurrence, and 3 service use variables. Results Episodes of uncomplicated depression triggered by bereavement and by other loss have similar symptom profiles and are not significantly different for 8 of 9 disorder indicators. Moreover, uncomplicated reactions, whether triggered by bereavement or other loss, are significantly lower than complicated reactions on almost all disorder indicators. Conclusion The NCS data do not support the validity of uniquely excluding uncomplicated bereavement but not uncomplicated reactions to other losses from MDD diagnosis. A common criticism of the DSM symptom-based diagnostic criteria is that their failure to consider stressful contexts results in false-positive diagnoses, that is, the classification of psychiatrically normal persons as mentally disordered.1,2 False-positive diagnoses can potentially lead to stigmatization, inappropriate care, and inflated epidemiological prevalence rates that undermine the credibility of the diagnostic system.2 Conversely, restricting diagnostic criteria to exclude persons without mental disorder but with symptoms can inadvertently lead to false-negative diagnoses, that is, classification of genuinely disordered individuals as nondisordered, potentially leading to failure to obtain needed treatment. This false-negative/false-positive tension is nowhere more apparent than in the DSM-IV bereavement exclusion for major depressive disorder (MDD). A diagnosis of MDD requires at least 1 major depressive episode (MDE) not caused by bipolar or nonaffective psychotic disorders. Criteria for an MDE require at least 5 of 9 symptoms including sadness or lack of interest or pleasure, at least 2 weeks' duration, clinically significant impairment or distress, and exclusion of substance-induced and general medical etiologies. However, some individuals who meet these symptom and impairment criteria are not experiencing a mood disorder but intense normal sadness in response to bereavement.3 The DSM criteria for MDE use an exclusion criterion in an attempt to prevent false-positive diagnoses, requiring that the symptoms cannot be better explained by bereavement. Bereavement can sometimes trigger genuine mood disorders. Thus, excluding all cases of bereavement from MDEs would yield false-negative diagnoses. The DSM addresses this problem by using unusual duration, impairment, or symptoms to identify exceptions to the bereavement exclusion that likely represent true disorder. The DSM-IV, for example, classifies bereavement responses as MDEs if the symptoms last more than 2 months or if there is marked functional impairment, morbid preoccupation with worthlessness, suicidal ideation, psychotic symptoms, or psychomotor retardation. The DSM uses the term “complicated bereavement” for bereavement that triggers MDD. However, this term has recently become widely used to also denote a nondepressive mourning-related pathologic condition including such symptoms as unremitting yearning and sense of loss.4-6 To avoid confusion, we refer to bereavement-triggered depressive disorder as complicated bereavement-triggered depression. The DSM criteria for MDD ignore the many other kinds of serious losses that can cause intense symptoms of normal sadness. (We use the term “sadness” as a generic label for normal and abnormal depressive responses to various losses.) This asymmetry raises the question of whether the DSM is justified in singling out bereavement as the only type of loss that produces normal intense sadness symptomatically similar to MDE. Historical precedent, common sense, and research on loss responses all suggest that many types of losses can trigger intense normal sadness. From early Greek and Roman physicians through Kraepelin and Freud to pre-DSM-III diagnostic manuals, psychiatric thought generally differentiated depressive disorder from symptomatically similar sadness resulting from various losses including not only bereavement but also romantic betrayal and rejection, economic misfortune, severe physical illness, loss of cherished possessions, and failure to attain important goals, among others.7-9 As is the case in bereavement,10,11 ample research suggests that many other types of loss, such as marital dissolution, unexpected job loss, and natural disasters, can trigger intense sadness that soon after the loss may satisfy MDD symptom criteria yet often naturally desists with time or when circumstances improve.12-21 Such intense sadness responses to major nonbereavement losses are generally considered normal.7,8,22 Moreover, the evidence suggests that intense sadness is a biologically designed response to a broad range of circumstances, including separation from a love object and loss of social status.23,24 Evolutionary approaches to the distinction between normal and disordered functioning,25-28 therefore, imply that depressive symptoms should not automatically be classified as a disorder, even at levels that satisfy DSM symptom criteria, without considering the nature of their trigger. From this perspective, bereavement may be considered a model for other types of loss responses, which might similarly be grounds for exclusion from a diagnosis of MDD. Like bereavement, other stressors may also trigger complicated, truly disordered reactions. Consequently, if the bereavement exclusion were extended to other stressors to avert false-positive diagnoses, the distinction between complicated vs uncomplicated bereavement would also have to be applied to other stressors to avoid false-negative diagnoses. Our overarching view is that the DSM uncomplicated vs complicated bereavement distinction reflects likely nondisorder vs disorder but that grief is not unique in this regard and the same approach has comparable validity for reactions to other types of loss. We tested 2 specific hypotheses predicted by this view. First, uncomplicated bereavement and uncomplicated reactions to other losses are no different across a range of variables generally considered indicative of disorder (eg, duration, recurrence, and service use). Second, uncomplicated responses to bereavement and to other losses are less severe on such disorder indicators than are complicated reactions to either type of trigger. In effect, the second hypothesis is a validity check on the uncomplicated-complicated distinction. Methods Sample It is difficult to compare complicated and uncomplicated episodes in clinical samples, which generally exclude individuals falling under the bereavement exclusion and contain few nondisordered individuals. Thus, we used publicly available data from the first-wave National Comorbidity Survey (NCS), a community-based epidemiological survey administered between September 14, 1990, and February 6, 1992, to 8098 persons aged 15 to 54 years who are representative of the US population.29 The NCS uses DSM-III-R–derived algorithms for diagnosing disorders, including MDD. Note that DSM-III-R criteria for MDD differ from DSM-IV criteria because they do not require clinically significant distress or impairment in addition to symptoms. The NCS operationalization of DSM-III-R MDD criteria require that a respondent satisfy 4 criteria. First, during a 2-week index episode (for individuals with multiple episodes, the index episode is the episode with the most symptoms), the respondent must report at least 1 symptom from each of 5 symptom groups or more (constructed to reflect DSM MDD symptom groups), none of which are from organic causes. For example, 1 symptom group includes lost appetite, lost weight, increased appetite, or increased weight; another includes trouble in concentrating, slow thinking, or inability to make decisions. As in the DSM, 1 of the 5 endorsed symptom groups must be either the sad-blue-gloomy group or loss of interest. Second, the condition is not covered by the bereavement exclusion (see “Uncomplicated vs Complicated Responses”). Third, the respondent must never have had mania or hypomania. Fourth, the respondent does not have delusions or hallucinations indicative of nonaffective psychotic disorder, as indicated by either psychotic diagnoses or the occurrence of psychotic ideation for 2 weeks outside of affective episodes. Analytic sample To test our hypotheses, we formed an overall analytic sample consisting of 4 subsamples: uncomplicated bereavement triggered (n = 56; 6.5% of the analytic sample); complicated bereavement triggered (n = 101; 11.6%); uncomplicated other loss triggered (n = 174; 20.1%); and complicated other loss triggered (n = 536; 61.8%). These groups are derived from the NCS sample in several steps; they exclude a considerable number of cases in which the NCS data set does not contain the necessary information. The Figure shows the sequence of steps by which the analytic sample and its subgroups were identified. (The precise algorithms are available from the authors on request.) To persons meeting lifetime NCS criteria for MDD, we added those who meet other MDD criteria but are eliminated from MDD by the bereavement exclusion (n = 65), yielding the MDD pool (n = 1308). All subsequent analyses use this expanded MDD pool to enable comparisons of uncomplicated and complicated bereavement. We next identified 2 MDD pool subsamples who reported that either grief or other triggering events caused their episodes (typical NCS questions were as follows: “Did that period of feeling sad/blue occur just after someone close to you died?” “Was there anything else going on in your life at that time which caused you to feel sad/blue?”). For analytic purposes, it is essential to identify the trigger of the index episode because that is the only episode for which the NCS reports detailed symptom information. (The index episode for single-episode cases is the individual's only episode; for multiple-episode cases, it is the individual's worst episode in terms of number of symptoms, or, if no episode is worse than others, the most recent episode.) To identify bereavement-triggered cases (n = 157), we used the NCS algorithm for bereavement to identify individuals whose index episode of MDD was bereavement-triggered. The NCS instrument is such that only by identifying those multiple-episode cases with all grief episodes could we identify cases in which the index episode was bereavement triggered; thus, even for multiple-episode cases, this is a pure bereavement sample. Analogously, other loss–triggered cases (n = 710) included those in which the index episode was triggered by other loss, and multiple-episode cases included those in which no other episodes were bereavement triggered. This algorithm allows nonindex other loss–triggered episodes to be untriggered, but the NCS data do not permit us to determine exactly how often this occurred. However, the overall rate of untriggered episodes in the sample was so low (5% of single-episode cases and 4% of index episodes in multiple-episode cases) that this approximation to a purely other loss–triggered group seems warranted. Of single-episode MDD pool cases (n = 410), 93% were either bereavement triggered (n = 108) or other loss triggered (n = 276) and are included in the analysis. We excluded 5% (n = 20) because they were untriggered and 1% (n = 6) because of missing data. Among multiple-episode MDD pool cases (n = 898), 54% (n = 483) had index episode triggers identifiable as grief or other loss and were included in the analysis. We excluded cases with missing data (1%; n = 6), those with untriggered index episodes (4%; n = 34), and those in which the type of index episode trigger (bereavement vs other type of loss) could not be inferred from the data (42%; n = 375). The type of index episode trigger sometimes cannot be inferred because if an individual reports experiencing both types of triggers at various times, there is no way within the NCS data of establishing which type specifically applies to the index episode. Our resulting analytic sample (n = 867; 44% single-episode cases and 56% multiple-episode cases) consists of the 66% of the MDD pool that can be established to be either bereavement-triggered (n = 157) or other loss–triggered (n = 710) cases. Mean (SD) demographic data for the analytic sample were as follows: age, 33.8 (9.8) years; sex, 60.9% (2.7) female; educational achievement, 13.0 (0.1) years; and race/ethnicity, 80% (3.2) white. These data did not differ significantly from those for the excluded 34% of the MDD pool. Table 1 indicates that these demographic characteristics are not significantly different between the bereavement- and other loss–triggered categories. Uncomplicated vs complicated responses To divide trigger groups into uncomplicated and complicated categories, we strictly extended the NCS algorithm for differentiating uncomplicated vs complicated bereavement-triggered episodes to nonbereavement, other loss–triggered episodes. The NCS algorithm for complicated bereavement is derived from DSM-III-R criteria. The DSM-III-R exclusion clause, which differs slightly from the DSM-IV clause, states that “Morbid preoccupation with worthlessness, suicidal ideation, marked functional impairment or psychomotor retardation, or prolonged duration suggest bereavement complicated by major depression.”30(p223) The NCS interprets the term “suggest” to mean that no one symptom is sufficient by itself to imply MDD. Therefore, the NCS requires 2 symptoms or more for complicated bereavement, unlike the later 1-symptom DSM-IV approach. The NCS adds a sixth symptom, suicide attempt, to the DSM-III-R list; it operationalizes prolonged duration as longer than 12 weeks, and operationalizes marked impairment by agreement with the item, “Kept you from working or from seeing friends or relatives.” Bereavement and other loss–triggered episodes are considered uncomplicated unless they are classifiable as complicated based on having 2 or more of the 6 NCS duration, impairment, and other symptom features. We could not evaluate each episode in multiple-episode cases for complicatedness because the data only contain index episode symptoms. We followed the NCS algorithm in classifying multiple-episode bereavement cases as uncomplicated or complicated based on which category applied to the index episode. The rationale was that the index episode was the individual's worst episode and complicated symptoms indicate severity; thus, an uncomplicated index episode likely implies all uncomplicated episodes. We applied this same procedure in other loss–triggered cases to identify uncomplicated vs complicated multiple-episode cases. There were no significant demographic differences among the 4 uncomplicated and complicated bereavement- and other loss–triggered groups for age, sex, educational achievement, or race/ethnicity (Table 2). Disorder indicators To evaluate whether the 4 groups (uncomplicated vs complicated bereavement-triggered and uncomplicated vs complicated other loss–triggered episodes) differ in disorder vs nondisorder status, we compared the groups on descriptive variables that have face validity and are commonly used to indicate disorder. Three indicators concern index episode features: severity (mean number of the 9 MDD symptoms); melancholic depression (percentage of patients satisfying DSM-IV–type criteria for melancholic depression); and suicide attempt. Although the suicide attempt indicator is contaminated because it is also an NCS complicatedness symptom, we included it because we think it is pragmatically important to establish whether cases involving suicide attempts would be classified as disorders. Six other disorder indicators concern lifetime history; thus, for multiple-episode cases, these indicators do not necessarily apply specifically to the index episode but set an upper boundary for that episode. These 6 indicators are as follows: duration (mean duration of the longest episode); interference with life (whether episode or episodes interfered with life or activities “a lot”); ever saw a mental health professional because of depression; ever took medication for depression; was ever hospitalized because of depression; and recurrence (mean number of episodes). Note that duration is contaminated to some extent because duration longer than 12 weeks is a symptom of complicatedness. To reduce the effect of outliers on recurrence and duration means, we coded respondents reporting more than 20 depressive episodes as having 20 episodes, and respondents reporting their longest episode as lasting more than 104 weeks as having a longest episode of 104 weeks. Interference with life and service use indicators follow the clinical significance disorder indicators of Narrow et al.31 Service use indicators are based on disorder-specific questions (eg, seeing a professional because of depression) to minimize confounding by service use for comorbid conditions.32 A long tradition considers melancholic depression as indicating disorder.7 We constructed criteria that approximate DSM-IV melancholic depression criteria, requiring inability to enjoy usual activities plus 3 or more of the following: retardation observed by others, or agitation; feel bad in the morning; early awakening (at least 2 hours early); lost weight; and excessive feelings of guilt. Statistical analyses Tests of Significance All data used in the analyses were weighted and corrected for sampling design. Statistical analyses were performed using the survey estimation procedures in STATA software, version 9 (StataCorp, College Station, Tex), which calculate weighted coefficients and use Taylor series linearization to calculate SE. Analysis of variance with planned comparisons were used to determine the mean differences between groups. When doing the planned comparisons, the nature of our 2 hypotheses dictated different statistical procedures. We used 2-tailed tests to evaluate the hypothesis that uncomplicated bereavement-triggered vs other loss–triggered episodes are not different on disorder indicators. Because we predicted that uncomplicated categories would be lower than complicated categories on disorder indicators, we used 1-tailed tests of our second hypothesis. Statistical Power Because our hypotheses propose no significant differences in mean values of indicator variables between 2 groups (uncomplicated bereavement-triggered vs uncomplicated other loss–triggered episodes), statistical power is an important concern. According to Cohen,33 small, medium, and large effect sizes would correspond to mean differences of 0.2, 0.5, and 0.8 SD, respectively, which would require roughly 400, 60, and 25 cases, respectively, to achieve good statistical power (power = .80) for the 2-tailed tests and roughly 300, 50, and 20 cases, respectively, for the 1-tailed tests. An example of a small mean difference would be 10 weeks' duration, 0.3 in interference score, and 6% for proportion of cases seeing a mental health provider. Medium effect size for the same analyses would be 26 weeks' duration, 0.75 in interference score, and 15% for proportion of cases seeing a mental health provider. Most of the analyses in this study contained sufficient numbers of cases to find small or medium and larger effect sizes in the mean differences. Results Data in Table 3 confirm 8 of 9 predictions stemming from our first hypothesis that uncomplicated bereavement-triggered and other loss–triggered cases do not differ on indicators of disorder. The sole exception is that more individuals in the other loss–triggered group than the bereavement-triggered group (12.4% vs 4.6%) reported that their condition interferes with life “a lot.” We also compared uncomplicated bereavement-triggered cases with uncomplicated other loss–triggered cases on MDE symptom groups because substantial differences might suggest a difference in disorder status. (Symptom differences between uncomplicated and complicated cases are assured by definition.) No significant differences were found for 8 of 9 symptom groups (see Table 4). The only significant difference is that the uncomplicated bereavement-triggered group endorsed the suicide and death thought symptom group more frequently (80.6%) than the uncomplicated other loss–triggered group (42.9%). This difference is entirely attributable to a significant difference between the 2 groups in “thought about death” (78.6% vs 33.3%; t = 6.3), which is understandable because the bereaved have recently been exposed to the death of a loved one. Table 3 also gives the results of tests of our second hypothesis, that uncomplicated bereavement- and other loss–triggered cases are lower on disorder indicators than complicated bereavement- and other loss–triggered cases. Seven indicators fully confirm this hypothesis: severity, melancholic depression, suicide attempt, duration, interference, saw a professional, and took medication. The remaining 2 indicators partially confirm the hypothesis. Results for hospitalization support 3 of 4 subhypotheses, but, contrary to our prediction, individuals in the uncomplicated bereavement group are not less likely to be hospitalized than are those in the complicated other loss group. Results for recurrence confirm 2 of 4 subhypotheses but do not indicate that either uncomplicated bereavement- or uncomplicated other loss–triggered cases are significantly less likely than complicated bereavement cases to recur. Examination of significant differences between complicated and uncomplicated cases indicate that the effect sizes are medium to large. Comment The DSM bereavement exclusion acknowledges that some intense episodes of sadness that satisfy symptomatic criteria for MDD are not disorders; however, the exclusion includes only bereavement and not responses to other losses. Based on traditional distinctions between depressive disorder and normal intense sadness and on an evolutionary view of sadness and disorder, we hypothesized that the bereavement exclusion represents a valid attempt to avert false-positive diagnoses that applies equally to other losses. Using NCS data, we tested our hypothesis that bereavement and other losses are symmetric against the contention, implied by the DSM MDD criteria, that bereavement is categorically different from other stressors on the dimensions assessed in this study. The results overwhelmingly support our hypotheses. They confirm 8 of 9 predicted relationships of no difference between uncomplicated categories, regardless of the type of trigger, and 33 of 36 predicted directional differences between uncomplicated and complicated categories. The results have substantial implications for MDD diagnosis, especially inasmuch as bereavement or some other loss reportedly precedes more than 90% of index episodes in MDD cases in the NCS. The results do not support the current categorical distinction between uncomplicated bereavement-triggered and uncomplicated other loss–triggered episodes. Rather, they imply that if the current criteria correctly label uncomplicated depressive episodes after death of a loved one as nondisorders, then uncomplicated episodes that occur after other losses are also plausibly nondisordered. Moreover, the differences between uncomplicated and complicated cases suggest that the bereavement exclusion reflects a valid distinction. Overall, our results suggest a potentially serious problem with MDD in the diagnosis of responses to major nonbereavement losses. These results also have implications for MDD prevalence. The NCS reports a lifetime prevalence of 14.9% for MDD, a figure that has caused much skepticism.2 Extending the bereavement exclusion to our other loss–triggered group, of which 24.5% of cases (2.2% of the total NCS sample) were uncomplicated, reduces NCS MDD prevalence to 12.7%. However, our analysis excludes many triggered cases because the trigger was unidentifiable. Excluding from diagnosis all NCS MDD cases with a trigger that were uncomplicated decreases MDD prevalence to 11.3%, an overall reduction of almost one fourth (24.2%). The 1-year NCS MDD rate of 8.6% would be similarly reduced about one fourth, to 6.5%. Despite overall strong support for our hypotheses, there are several exceptions. Contrary to our first hypothesis, significantly fewer uncomplicated bereavement- than uncomplicated other loss–triggered cases report that depressive episodes interfered with life a lot (4.6% vs 12.4%; t = 2.62). This finding is not repeated with other clinical-significance measures, and both uncomplicated categories are substantially lower on interference than either complicated bereavement-triggered or complicated other loss–triggered categories (grief, 4.6% vs 46.2%; other loss, 12.4% vs 47.3%). One possible explanation for this finding is that bereavement is a socially acknowledged and frequently ritualized experience that often legitimates excuses from normal responsibilities. In contrast, social norms are less likely to allow withdrawal from normal role functioning in other loss–triggered cases; thus, greater interference with expectable levels of social engagement and responsibility might be experienced. Contrary to our second hypothesis, the 5% of hospitalized patients with uncomplicated bereavement-triggered sadness is not significantly lower than the 8.9% of hospitalized patients with complicated other loss–triggered sadness (t = 0.87). We speculate that if bereavement is more likely than other triggers to be considered a health risk, providers may occasionally hospitalize patients with uncomplicated sadness as a preventive measure. This is an area worthy of further investigation. Also contrary to our second hypothesis, neither uncomplicated bereavement-triggered episodes nor uncomplicated other loss–triggered episodes are significantly lower than complicated bereavement in mean recurrence. However, these findings have a simple explanation; bereavement episodes, by definition, occur only after the death of a loved one and, thus, are severely constrained insofar as recurrence. Thus, the failure of this prediction seems less a disconfirmation than an anomalous situation. This study has several limitations. The age range of the NCS sample (15-54 years) means that the sample does not include the elderly, a major group affected by bereavement and other losses, and a target of much recent depression research; it remains for future research to demonstrate whether the present results can be generalized to this important group. Another limitation is that we eliminated from analysis a substantial number of NCS multiple-episode MDD cases in which it was reported that the index episode was triggered, because the nature of the NCS data does not allow us to identify the type of trigger (bereavement vs other loss). In addition, we conservatively accepted the NCS algorithm for identifying uncomplicated vs complicated episodes and did not empirically examine alternative formulations. Furthermore, the accuracy of the NCS respondents' self-reports of triggering events is unknown; respondents may misremember whether there was an event or whether the timing of an event was before an episode or may misattribute the cause of an episode to an event when they were coincidental. However, depressive symptoms can be misattributed to death of a loved one and to other losses, and the crucial complicated vs uncomplicated distinction we used is based on symptom reports and not on causal self-attributions alone. Another limitation is that we were unable to examine the qualitative nature of nonbereavement triggers; such qualitative data were collected but are not publicly available because of institutional review board restrictions. Consequently, we were unable to examine distinctions among nongrief stressors or to make finer discriminations of proportionality between stressor and symptoms, for example, to judge when a reported trigger is not major and perhaps is so mild as to potentially suggest disorder even in uncomplicated episodes. Future studies should be designed to identify and evaluate the stressors directly. Uncomplicated bereavement-triggered and uncomplicated other loss–triggered groups might differ in ways we did not detect; we can only say that at the level of detection these data make possible, there is no reason to consider other loss–triggered conditions as substantially different from bereavement conditions on disorder indicators and, thus, no support for the current DSM asymmetry in applying the complicated vs uncomplicated distinction to bereavement but not to other stressors. Conclusions These results, even if replicated, cannot by themselves fully resolve the issue of when to consider uncomplicated depressive episodes as nondisordered. While there are clear cases of complicated conditions that are disorders and uncomplicated conditions that are nondisorders, there is also growing awareness that depressive symptoms occur on a continuum; thus, many clinical patients might not fall clearly to one side or the other of the disorder-nondisorder divide. Moreover, it might be argued that our results for the disorder indicators are consistent with the position that uncomplicated episodes in response to both bereavement and other stressors are merely mild disorders. For example, Zisook et al,34 also noting the asymmetry between bereavement and other losses in DSM MDD criteria, claim that the bereavement exclusion should be eliminated: “No other life event (or precipitant) negates the diagnosis of depression when the full syndrome occurs. It is not clear why death of a loved one should cancel out the diagnosis of major depressive disorder, either.”34(p229) Our findings, therefore, directly challenge only the current asymmetry between bereavement and other loss situations in the MDD bereavement exclusion. Given the absence of an objective gold standard for differentiating clinical depression from normal sadness, further empirical research should explore the prognostic, treatment, and policy implications of different classifications of uncomplicated nonbereavement episodes, as well as the conceptual underpinnings of MDD. Traditionally, diagnosis of depressive disorder reflected the notion that sadness in response to loss is natural and normal and that the indication of disorder lies in the sadness being without sufficient cause in given environmental contexts or being disproportional to actual loss.7 The DSM MDD complicated bereavement exclusion can be interpreted to mean that clinical depression should be diagnosed if the response is symptomatically out of proportion even to loss of a loved one. In attempting to extend the bereavement exclusion to reactions other than grief, further research is required to determine whether DSM-type symptom severity criteria should also be adopted for other stressors or, as much research suggests,12,35,36 some more complex measure of proportionality that considers the typical or individual meaning of the stressor is necessary and clinically feasible. In addition, some other losses (eg, financial reversal or marital alienation) may be less clear-cut than death of a loved one and it may be harder to judge the severity of these losses, leading to challenges to reliable measurement. The importance of reliability, however, does not obviate the need to address substantial failures of validity. Although death is clear-cut, the bereavement exclusion itself depends on the complicated-uncomplicated distinction, which is not clear-cut. In extending our results to clinical practice, we do not intend to suggest that treatment after major stressors is appropriate only for individuals experiencing complicated episodes. So-called uncomplicated episodes can involve substantial suffering and, in vulnerable individuals, may evolve into complicated episodes. Treatment, including psychotherapy or medication,33 may sometimes be appropriate for intense normal sadness. Nevertheless, the current DSM distinction between complicated vs uncomplicated bereavement-triggered depression has implications for treatment planning, prognosis, and stigmatization. Our results suggest that the same reasons dictate recognition of intense normal sadness responses to stressors other than bereavement. If further research confirms these findings, the DSM MDD bereavement exclusion likely should be reconsidered in DSM-V, with equal attention to bereavement and nonbereavement triggers of intense sadness. Correspondence: Jerome C. Wakefield, PhD, DSW, School of Social Work, New York University, 1 Washington Sq N, New York, NY 10003 ([email protected]). Submitted for Publication: May 23, 2006; final revision received August 30, 2006; accepted August 30, 2006. Financial Disclosure: None reported. References 1. Spitzer RLWakefield JC DSM-IV diagnostic criterion for clinical significance: does it help solve the false positives problem? Am J Psychiatry 1999;1561856- 1864PubMedGoogle Scholar 2. Regier DAKaelbar CTRae DSFarmer MEKnauper BKessler RCNorquist GS Limitations of diagnostic criteria and assessment instruments for mental disorders: implications for research and policy. Arch Gen Psychiatry 1998;55109- 115PubMedGoogle ScholarCrossref 3. Clayton PJHerjanic MMurphy GEWoodruff R Jr Mourning and depression: their similarities and differences. Can Psychiatr Assoc J 1974;19309- 312PubMedGoogle Scholar 4. Horowitz MJSiegel BHolen ABonanno GAMilbrath CStinson CH Diagnostic criteria for complicated grief disorder. Am J Psychiatry 1997;154904- 910PubMedGoogle Scholar 5. Prigerson HGFrank EKasl SVReynolds CFAnderson BZubenko GSHouck PRGeorge CJKupfer DJ Complicated grief and bereavement-related depression as distinct disorders: preliminary empirical validation in elderly bereaved spouses. Am J Psychiatry 1995;15222- 30PubMedGoogle Scholar 6. Lichtenthal WGCruess DGPrigerson HG A case for establishing complicated grief as a distinct mental disorder in DSM-V. Clin Psychol Rev 2004;24637- 662PubMedGoogle ScholarCrossref 7. Jackson SW Melancholia and Depression: From Hippocratic Times to Modern Times. New Haven, Conn: Yale University Press;1986; 8. Radden Jed The Nature of Melancholy: From Aristotle to Kristeva. New York, NY: Oxford University Press; 2000; 9. MacDonald M Mystical Bedlam: Madness, Anxiety, and Healing in Seventeenth-Century England. New York, NY: Cambridge University Press;1981; 10. Clayton PJ The model of stress: the bereavement reaction. Dohrenwend BPed. Adversity, Stress, and Psychopathology. New York, NY: Oxford University Press; 1998;96- 110Google Scholar 11. Clayton PJedDarvish HSedRose RMedKlerman GLed Course of depressive symptoms following the stress of bereavement. Barrett JE Stress and Mental Disorder. New York, NY: Raven Press;1979;Google Scholar 12. Kessler RC The effects of stressful life events on depression. Annu Rev Psychol 1997;48191- 214PubMedGoogle ScholarCrossref 13. Booth AAmato P Divorce and psychological stress. J Health Soc Behav 1991;32396- 407PubMedGoogle ScholarCrossref 14. Bruce ML Divorce and psychopathology. Dohrenwend BP ed. Adversity, Stress, and Psychopathology. New York, NY: Oxford University Press;1998;219- 232Google Scholar 15. Bruce MLKim KLeaf PJJacobs S Depressive episodes and dysphoria resulting from conjugal bereavement in a prospective community sample. Am J Psychiatry 1990;147608- 611PubMedGoogle Scholar 16. Kessler RCTurner JBHouse JS Unemployment, reemployment, and emotional functioning in a community sample. Am Sociol Rev 1989;54648- 657Google ScholarCrossref 17. Price RHChoi JNVinokur AD Links in the chain of adversity following job loss. J Occup Health Psychol 2002;7302- 312PubMedGoogle ScholarCrossref 18. Dooley DPrause JHam-Rowbottom KA Underemployment and depression: longitudinal relationships. J Health Soc Behav 2000;41421- 437PubMedGoogle ScholarCrossref 19. Chen HChung HChen TFang LChen JP The emotional distress in a community after the terrorist attack on the World Trade Center. Community Ment Health J 2003;39157- 165PubMedGoogle ScholarCrossref 20. Galea SAhern JResnick HKilpatrick DBucuvalas MGold JVlahov D Psychological sequelae of the September 11 terrorist attacks in New York City. N Engl J Med 2002;346982- 987PubMedGoogle ScholarCrossref 21. Knudsen HKRoman PMJohnson JADucharme LJ A changed America? the effects of September 11 on depressive symptoms and alcohol consumption. J Health Soc Behav 2005;46260- 273PubMedGoogle ScholarCrossref 22. Kleinman AGood B Introduction: culture and depression. Kleinman AGood BedsCulture and Depression. Berkeley: University of California Press;1985;1- 33 23. Archer J The Nature of Grief: The Evolution and Psychology of Reactions to Loss. New York, NY: Routledge & Kegan Paul Plc; 1999; 24. Keller MCNesse RM Is low mood an adaptation? evidence for subtypes with symptoms that match precipitants. J Affect Disord 2005;8627- 35PubMedGoogle ScholarCrossref 25. Nesse RM Is depression an adaptation? Arch Gen Psychiatry 2000;5714- 20PubMedGoogle ScholarCrossref 26. Wakefield JC Evolutionary versus prototype analyses of the concept of disorder. J Abnorm Psychol 1999;108374- 399PubMedGoogle ScholarCrossref 27. Wakefield JC The concept of mental disorder: on the boundary between biological facts and social values. Am Psychol 1992;47373- 388PubMedGoogle ScholarCrossref 28. Wakefield JC Disorder as harmful dysfunction: a conceptual critique of DSM-III-R's definition of mental disorder. Psychol Rev 1992;99232- 247PubMedGoogle ScholarCrossref 29. Kessler RCMcGonagle KAZhao SNelson CBHughes MEshleman SWittchen HUKendler KS 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 30. American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition . Washington, DC: American Psychiatric Association;1987; 31. Narrow WERae DSRobins LNRegier DA Revised prevalence estimates of mental disorders in the United States: using a clinical significance criterion to reconcile 2 surveys' estimates. Arch Gen Psychiatry 2002;59115- 123PubMedGoogle ScholarCrossref 32. Wakefield JCSpitzer RL Lowered estimates—but of what? Arch Gen Psychiatry 2002;59129- 130[published correction appears in Arch Gen Psychiatry. 2002;59:416]PubMedGoogle ScholarCrossref 33. Cohen J A power primer. Psychol Bull 1992;112155- 159Google ScholarCrossref 34. Zisook SShuchter SRPedrelli PSable JDeaciuc SC Bupropion sustained release for bereavement: results of an open trial. J Clin Psychiatry 2001;62227- 230PubMedGoogle ScholarCrossref 35. Brown GW Social roles, context and evolution in the origins of depression. J Health Soc Behav 2002;43255- 276PubMedGoogle ScholarCrossref 36. Dohrenwend BP The role of adversity and stress in psychopathology: some evidence and its implications for theory and research. J Health Soc Behav 2000;411- 19PubMedGoogle ScholarCrossref

Odgers, Candice L.;Caspi, Avshalom;Broadbent, Jonathan M.;Dickson, Nigel;Hancox, Robert J.;Harrington, HonaLee;Poulton, Richie;Sears, Malcolm R.;Thomson, W. Murray;Moffitt, Terrie E.

doi: 10.1001/archpsyc.64.4.476pmid:

doi: 10.1001/archpsyc.64.4.397pmid: N/A

Ettinger et alArticle carried out a magnetic resonance imaging investigation of the thalamus in monozygotic twins with or without schizophrenia. Twin pairs concordant for schizophrenia but not twin pairs discordant for schizophrenia showed significant volumetric reductions of the thalamus, suggesting that this abnormality may reflect the increased genetic liability seen in the concordant group. Frodl et alArticle examined the impact of the BDNF Val66Met polymorphism on hippocampal and amygdala volumes. Sixty patients with major depression had smaller hippocampal volumes as compared with 60 healthy controls. Moreover, patients, as well as controls, carrying the Met-BDNF allele showed smaller hippocampal volumes than subjects homozygous for the Val-BDNF allele. Therefore, the BDNF polymorphism may be involved in hippocampal development and a smaller hippocampal volume may be a risk factor for major depression. Miklowitz et alArticle evaluated the effectiveness of intensive psychotherapy (in combination with pharmacotherapy) in 293 outpatients with bipolar I or II depression. Patients randomly assigned to intensive psychotherapy (30 sessions of family-focused therapy, interpersonal and social rhythm therapy, or cognitive-behavioral therapy) had more rapid recoveries and were more likely to remain well during the 1-year study than patients assigned to 3 sessions of psychoeducation. Wakefield et alArticle, using National Comorbidity Survey data, compared uncomplicated bereavement with uncomplicated reactions to other stressors on multiple variables indicating pathologic conditions. Results indicate that uncomplicated bereavement and uncomplicated other-loss reactions are similar on almost all indicators and are similarly different from complicated episodes. The authors suggest extending the bereavement exclusion to other stressors in DSM-V. Scherk et alArticle conducted a meta-analysis of randomized controlled trials on the efficacy of second-generation antipsychotics (SGAs) in the treatment of acute mania. Twenty-four studies with 6187 patients were included. The SGAs were more efficacious than placebo and of similar efficacy as mood stabilizers. Data suggest that combining SGAs and mood stabilizers is the most efficacious treatment of acute mania. Analyzing a nationally distributed sample, Feinberg et alArticle found that genotype interacted with both parental negativity and low warmth, predicting overall antisocial behavior but not depression. Genetic influence was greater for adolescent antisocial behavior when parenting was more negative or less warm. This study demonstrates that a continuous measure of parenting in the normative range moderates the influence of genotype on antisocial behavior. To determine the effects of the publicity concerning the potential for increased suicidality with selective serotonin reuptake inhibitor use in children and adolescents, Nemeroff et alArticle analyzed the Verispan retail pharmacy prescription database, which captures nearly one half of all prescriptions in the United States. The number of children and teenagers who were prescribed antidepressants decreased after the Food and Drug Administration Public Health Advisory meetings. Moreover, psychiatrists are now prescribing a higher percentage and generalists a lower percentage of antidepressants to individuals 18 years and younger. Using a 30-year prospective study, Odgers et alArticle assessed whether youth with childhood-onset life-course-persistent antisocial behavior become adult men with poor physical health. Health outcomes included injury, sexually transmitted diseases, cardiovascular risk, immune function, and dental disease, suggesting the health costs of childhood-onset conduct disorder are considerable. Neighbors et alArticle examined the 12-month mental health service use in a national probability sample of African American and Caribbean black adults 18 years and older. Respondents who met criteria for a 12-month DSM-IV disorder reported greater use of services compared with those without a disorder. However, mental health service sectors were generally underused by both African American and Caribbean black individuals. In a large population-based study of hospitalized persons aged 18 to 44 years, Westover et alArticle found an association between amphetamine abuse and hemorrhagic stroke, as well as between cocaine abuse and both ischemic and hemorrhagic stroke, in a multivariate logistic regression model. Amphetamine abuse has increased more than other drugs of abuse from 2000 to 2003 and was associated with a higher risk of in-hospital death from hemorrhagic stroke.

Ettinger, Ulrich;Picchioni, Marco;Landau, Sabine;Matsumoto, Kazunori;van Haren, Neeltje E.;Marshall, Nicolette;Hall, Mei-Hua;Schulze, Katja;Toulopoulou, Timothea;Davies, Nadia;Ribchester, Tracy;McGuire, Philip K.;Murray, Robin M.