Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Infant abuse diagnosis associated with abusive head trauma criteria: incidence increase due to overdiagnosis?

Infant abuse diagnosis associated with abusive head trauma criteria: incidence increase due to... Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 Infant abuse diagnosis based on criteria for abusive head trauma 641 39 Kepper M, Tseng T-S, Volaufova J, et al. Pre-school obesity is inversely 40 Ministry of Education, Culture, Sports, Science, and Technology. School nutrition associated with vegetable intake, grocery stores and outdoor play. Pediatr Obes 2016; report, 2015. Available at: http://www.mext.go.jp/b_menu/toukei/chousa05/eiyou/ 11:e6–8. gaiyou/1376285.htm (19 September 2016, date last accessed). ......................................................................................................... The European Journal of Public Health, Vol. 28, No. 4, 641–646 The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. doi:10.1093/eurpub/cky062 Advance Access published on 17 April 2018 ......................................................................................................... Infant abuse diagnosis associated with abusive head trauma criteria: incidence increase due to overdiagnosis? 1 2 3 4 1 5 Ulf Ho¨ gberg , Erik Lampa ,Go¨ ran Ho¨ gberg , Peter Aspelin , Fredrik Serenius , Ingemar Thiblin 1 Department of Women’s and Children’s Health, Uppsala University, Uppsala, Sweden 2 UCR-Uppsala Clinical Research Centre, Uppsala University, Uppsala, Sweden 3 Child and Adolescent Psychiatric Unit, Department of Women’s and Children’s Health, Karolinska Institute, Stockholm, Sweden 4 Department of Clinical Science, Intervention and Technology (CLINTEC), Karolinska Institute, Stockholm, Sweden 5 Forensic Medicine, Department of Surgical Sciences, Uppsala University, Uppsala, Sweden Correspondence: Ulf Ho¨ gberg, Uppsala University, Department of Women’s and Children’s Health, Akademiska Sjukhuset, SE-751 85 Uppsala, Sweden, Tel: +46 186115246, Fax: +46 186115583, e-mail: [email protected] Background: The hypothesis of this study is that the diagnosis of infant abuse is associated with criteria for shaken baby syndrome (SBS)/abusive head trauma (AHT), and that that changes in incidence of abuse diagnosis in infants may be due to increased awareness of SBS/AHT criteria. Methods: This was a population-based register study. Setting: Register study using the Swedish Patient Register, Medical Birth Register, and Cause of Death Register. The diagnosis of infant abuse was based on the International Classification of Diseases, 9th and 10th revision. Participants: All children born in Sweden during 1987–2014 with a follow-up until 1 year of age (N = 2 868 933). SBS/AHT criteria: subdural haemorrhage, cerebral contusion, skull fracture, convulsions, retinal haemorrhage, fractures rib and long bones. Outcomes: Incidence, rate ratios, aetiologic fractions and Probit regression analysis. Results: Diagnosis of infant abuse was strongly associated with SBS/AHT criteria, but not risk exposure as region, foreign-born mother, being born preterm, multiple birth and small for gestational age. The incidence of infant abuse has increased tenfold in Sweden since the 1990s and has doubled since 2008, from 12.0 per 100 000 infants during 1997–2007 to 26.5/100 000 during 2008–2014, with pronounced regional disparities. Conclusions: Diagnosis of infant abuse is related to SBS/ AHT criteria. The increase in incidence coincides with increased medical preparedness to make a diagnosis of SBS/ AHT. Hidden statistics and a real increase in abuse are less plausible. Whether the increase is due to overdiagnosis cannot be answered with certainty, but the possibility raises ethical and medico-legal concerns. ......................................................................................................... The incidence of maltreatment syndrome has shown a variation Introduction by country, from 11.5 per 100 000 infants in Sweden (1987–2009), n 1962 battered child syndrome was first described in a clinical to 34.6 in England (1997–2008), to 118.9 in Western Australia I study. In 1972 it was proposed that subdural haematomas could (1980–2005), with only England showing a declining trend. The be caused by whiplash shaking and in 1974 the term ‘shaken baby incidence of non-fatal AHT among infants, based on the case syndrome (SBS)’ was coined to describe a condition inflicted by definition of the US Center of Disease Control and Prevention violent shaking and identified by the triad retinal haemorrhage, (CDC), was 32.3 per 100 000, while a Canadian study on infant subdural haemorrhage and encephalopathy. The diagnosis of AHT found an incidence of 13.0–15.5 for 2002–2007. In Scotland abusive head trauma (AHT), departing from the three diagnostic in 1998–1999, the incidence of shaken impact syndrome among criteria for SBS, now has a broader categorization that may also infants was 24.6 per 100 000. include apnoea, seizures, fractures of the skull, metaphyses and shaft Shaken baby syndrome has, since 1997, been questioned as a of long bones and ribs, and inability of parents/carers to provide an diagnostic entity, and emerging imaging technology demands further 13–18 explanation for accidental trauma, the former findings and symptoms differential diagnostic considerations. Furthermore, the precision 4–7 being considered highly specific for non-accidental trauma. of the SBS/AHT diagnosis is lost when excluding the diagnostic criterion Shaken baby syndrome/AHT is not a diagnosis classified in the of incompatibility between the carer’s report of history and the International Classification of Diseases (ICD-9/ICD-10); however, investigating doctor’s interpretation of the findings. In 2016, a physical abuse and battered baby or child syndrome is defined as systematic literature review by the Swedish Agency for Health maltreatment syndrome. In an intercountry epidemiological study, Technology Assessment and Assessment of Social Service (SBU) Gilbert et al. described the maltreatment syndrome/assault and concluded that there is limited scientific evidence that the triad can be included in their classification the diagnoses of intracranial injury explained by isolated shaking, and that there is insufficient evidence to 8 20,21 and long bone fractures. assess the diagnostic accuracy of the triad to identify SBS/AHT. Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 642 European Journal of Public Health Table 1 Definitions of infant abuse, and other diagnoses, according to the Swedish Patient Registry, National Board of Health and Welfare (ICD-9: 1987–1996; ICD-10: 1997–2014) Diagnosis category ICD code Infant abuse diagnosis Observation for suspected Z03.8K (ICD-10); E967 (ICD-9) abuse Battered baby syndrome Y07.9 (ICD-10); 995F (ICD-9) Maltreatment syndrome T74.1 (ICD-10); 995F (ICD-9) Neglect and adandonment Y06 Other maltreatment Y07 Subdural haemorrhage I62.0, S06.5 (ICD-10) Skull fracture S02.0, S02.1, S02.8, S02.09 S02.00, S02.9 (ICD-10) Cerebral contusion S06.0, S06.1 (ICD-10) Convulsions R56, R56.8, G40–41, R56.0 (ICD-10) Retinal haemorrhage H35.6 (ICD-10) Rib fracture S22.3, S22.4 (ICD-10) Long bone fracture S42.2, S42.3, S42, 4, S42.7, S42.8, S52, S72, S82, T10, T12 (ICD-10) In Sweden, battered child syndrome was introduced as a 100 000 [95% confidence interval (CI)], and the two periods were diagnostic criterion during the 1960s, and the triad of SBS was compared in terms of rate ratios (I /I ) (95% CI) and aetiologic 1 0 described in the Swedish Medical Journal in 1994. In 2008, fractions of incidence (I -I /I ) (%). The following covariates were 1 0 1 Stockholm County Council published clinical guidelines for the in- selected for further analysis because of a proposed association with vestigation of SBS according to the AHT criteria and the document the diagnosis of infant maltreatment: maternal country of birth; has been adapted for use in other parts of Sweden. In 2009, the infant perinatal characteristics such as multiple birth, preterm Swedish Paediatric Society established a child abuse task force to birth (born at <37 weeks’ gestation), born Small-for-Gestational- foster awareness of and training in child abuse recognition and Age (SGA) [<2.5 standard deviation (SD)]; diagnosis by health reporting, and implementation of the guidelines for investigation care region. and diagnosis of AHT. Subsequently, regional hospital-based child Probit regression analysis was performed in a Structural Equation protection centres were established. Modelling (SEM) framework, in order to examine a conceptual This study examines the incidence of diagnosis of infant abuse model to explore the interplay of the variables of importance for over three decades in Sweden. The hypothesis of our study is that the outcome ‘diagnosis of infant abuse’. The conceptual model is diagnosis of infant abuse is based on SBS/AHT criteria, and that shown in figure 2. Implicit in the model is a latent diagnosis variable changes in incidence of diagnosis of infant abuse may be due to that is observed across the different diagnoses. The importance of increased awareness of SBS/AHT criteria. the different observed diagnoses for the latent diagnoses variable was assessed by the factor loadings, with higher loading values implying greater importance. Methods Probit regression, like logistic regression, models the probability This is a population-based register study of children born in Sweden of an outcome variable being equal to one of two values and can be between 1987 and 2014, with a follow-up until 1 year of age. written as: The number of children born during the period 1987–2014 PYðÞ ¼ 1jX¼ X b ; (N = 2 984 813) was retrieved from Statistics Sweden. The research database was linked to the Swedish Patient Register, the Swedish where X is the data matrix and  is the vector of regression coeffi- Medical Birth Register and the Swedish Cause of Death Register, cients. denotes the cumulative distribution function of the standard with diagnoses and conditions classified using ICD-9 (1987–1996) normal distribution. The coefficient of the latent diagnosis variable and ICD-10 (1997–2015) codes. Infant abuse was defined according in the SEM Probit model should be interpreted as a change in to the Swedish version of the ICD-9 and ICD-10 (table 1). For the z-score which allows calculation of the probability of an SD years 1997–2014, ICD-10 codes for other possible differential increase in the latent diagnosis variable. To investigate whether as- diagnoses (table 1), symptoms and conditions which might be sociations changed over time, the SEM model was divided into two 6,7,13 considered when investigating infant abuse, were searched for. groups defined by the time period and the coefficients compared. In all, 182 974 infants with diagnostic codes were found, for each a The statistical software package IBM SPSS 25.0 (SPSS Inc., control was selected, born in the same year and without any Chicago, IL, USA) and R version 3.3.1 were used for data diagnosis in the Patient Register (N = 731 901). In total, the analyses. The SEM models were fitted using the lavaan add-on sample consisted of 914 875 infants, 49% of all infants born package for R. during 1997–2014. For this study, we selected as criteria for SBS/ The study was approved by the Regional Ethical Committee in AHT to be associated with abuse diagnosis:subdural haemorrhage, Uppsala (2014-11-19 No. 383). cerebral contusion, skull fracture, convulsions, retinal haemorrhage, 4,6–8,20 fractures of rib and long bones (table 1). The baseline descriptive analysis was based on the incidence Results proportion of diagnosis of infant abuse per 100 000 by year (moving annual average). For thorough analysis of diagnosis, only Altogether 368 infants were diagnosed with maltreatment during the years 1987–2014; 12.3 per 100 000 infants. Seven of the children years with ICD-10 coding were selected, 1997–2014. We based our analyses on incidence of diagnosis in the population and, for cases diagnosed with abuse died, i.e. 0.23 deaths per 100 000 infants and diagnosed with abuse, for the whole period divided by the two a case fatality rate of 1.9%. A tenfold increase in diagnosis of infant periods 1997–2007 and 2008–2014, before and after the increased abuse was observed from the start to the end of the period. awareness of SBS/AHT criteria was established in the Swedish The increase started slowly from 1996, peaking during the years setting. Diagnoses were selected by the incidence proportion per 2008–2014 (figure 1). There was a doubling of the incidence of Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 Infant abuse diagnosis based on criteria for abusive head trauma 643 Infant abuse diagnosis in Sweden 1987-2014 Moving annual means for infants born 1987-2014 Figure 1 Infant abuse diagnosis in Sweden for children born in 1987–2014 per 100 000 infants (moving annual average) Figure 2 Conceptual model for diagnosis of infant abuse in Sweden in 1997–2014, performed in a Structural Equation Modelling (SEM) framework. Observed variables are given inside rectangles. The latent diagnosis variable is drawn as an ellipse. Arrows radiating out from the ellipse and pointing to the observed diagnoses represent factor loadings. Coefficients represented by arrows pointing to ‘Infant abuse diagnosis’ should be interpreted as the increase in z-score for the probability of observing the outcome. For the risk exposures, the coefficients represent the increase in z-score if the condition is present. For the latent diagnosis variable, the coefficient represents the increase in z-score for a standard deviation (SD) increase in the latent diagnosis variable. The coefficient for the Southern health region was set to zero as a reference for the other health regions. SGA = small for gestational age abuse diagnosis during the years 2008–2014 compared with 1997– and skull fracture, had substantial loadings on the latent diagnosis 2007, from 12.0 (n = 127)–26.5 (n = 210) per 100 000 infants [odds variable suggesting that these four diagnoses were the main drivers ratio (OR) 2.21 (95% CI 1.78–2.76)]. of the association with the probability for diagnosis of infant abuse. Of the 337 cases of infant abuse during the years 1997–2014, 137 All diagnoses taken together were strongly associated with this prob- had either concomitant diagnoses of subdural haemorrhage, skull ability [coefficient 0.752 (95% CI 0.709–0.795 (P < 0.01)]. If the fracture, cerebral contusion, traumatic brain injury, convulsions, baseline probability of diagnosis of infant abuse is around 0.01% retinal haemorrhage, rib fracture, or fractures of the long bones. with all other factors held at their reference values, an SD increase in Out of the residual, 200 cases, 11 had injury to the head and 14 to the latent diagnosis variable would push that probability to the body, nine had injury of the eye or orbital fracture, five had a (3.72 + 0.75) = 0.15%. Coefficients for diagnoses were increased fractured clavicle, three had burns, three had diagnosis of failure to for being born preterm and SGA. Coefficients for diagnosis of thrive; others had miscellaneous diagnoses such as infections, skin infant abuse were increased when being born preterm and having disease, nausea, obstipation and icterus. Altogether 125 had no other a foreign-born mother. diagnosis besides abuse (table 1), whereof 60 had the diagnosis ‘ob- With the exception of convulsions, no statistically significant servation for suspected abuse’ and 35 had a diagnosis of maltreat- changes in population incidence of subdural haemorrhage, skull ment syndrome. fracture, cerebral contusion, retinal haemorrhage, rib fracture or The results of the Probit regression analysis are presented in fractures of the long bones could be seen between the two periods figure 2 and Supplementary table S1. Four of the observed 1997–2007 and 2008–2014 (table 2). However, cases diagnosed with diagnoses: subdural haemorrhage, rib fracture, retinal haemorrhage infant abuse with those diagnoses had statistically significantly Per 100,000 infants 2014 Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 644 European Journal of Public Health Table 2 Diagnosis and background characteristics of infants or children with maltreatment syndrome in Sweden, 1997–2007 (n = 127) and 2008–2014 (n = 210), per 100 000 children below 1 year of age (1997–2007 n = 1, 062, 084; 2008–2014 n = 793, 183) Diagnosis and background Abuse diagnosis 1997–2007 2008–2014 2008–2014/1997–2007 Characteristics n Incidence (95% CI) n Incidence (95% CI) RR (95% CI) EFe (%) Subdural haemorrhage No 130 12.3 (10.4–14.5) 133 16.8 (14.2–19.7) Yes 20 1.89 (1.22–2.79) 23 2.90 (1.94–4.20) 1.36 (1.07–1.73) 26.4 Skull fracture No 876 82.5 (77.2–88.0) 612 77.1 (71.3–83.4) Yes 12 1.13 (0.65–1.85) 21 2.65 (1.06–2.8) 2.34 (1.15–4.76) 57.3 Cerebral contusion No 5357 504 (491–518) 3, 731 470 (455–486) Yes 4 0.37 (0.15–0.83) 5 0.63 (0.28–1.29) 1.67 (0.45–6.23) 40.3 Convulsions No 7225 680 (665–696) 5, 931 747 (729–767) Yes 7 0.66 (0.33–1.23) 11 1.39 (0.78–2.32) 2.10 (0.82–5.43) 52.5 Retinal haemorrhage No 70 6.59 (5.22–8.22) 49 6.18 (4.68–8.02) Yes 10 0.94 (0.52–1.61) 18 2.27 (1.44–3.43) 2.41 (1.11–5.22) 58.5 Long bone fractures No 1157 108 (103–115) 993 125 (117–133) Yes 22 2.07 (1.37–3.02) 35 4.41 (3.18–5.99) 2.13 (1.25–3.63) 53.1 Rib fracture No 26 2.44 (1.68–3.47) 20 2.52 (1.64–3.74) Yes 12 1.13 (0.65–1.85) 20 2.01 (1.25–3.12) 2.32 (1.09–4.57) 55.2 Preterm-born (<37 wks) Yes 18 27.3 (17.3–41.3) 27 57.2 (39.4–80.6) 2.09 (1.15–3.80) 52.2 Small for gestational age Yes 7 29.6 (14.6–55.3) 8 43.9 (22.6–79.2) 1.48 (0.54–4.09) 32.6 Multiple birth Yes 6 38.0 (17.9–74.0) 14 127 (76.2–202) 3.34 (1.28–8.68) 70.0 Mother Swedish-born Yes 96 11.1 (9.13–13.5) 124 12.3 (12.4–18.5) 1.87 (1.43–2.44) 46.4 Mother foreign-born Yes 31 15.6 (11.0–21.6 86 44.6 (35.9–54.2) 2.71 (1.80–4.09) 63.1 Southern Region Yes 27 13.2 (9.07–18.6) 53 33.4 (25.6–43.0) 2.54 (1.60–4.03) 60.6 South East Region Yes 22 20.4 (13.5–29.7) 19 24.2 (15.5–36.2) 1.59 (0.86–2.94) 37.1 Western Region Yes 13 7.2 (4.24–11.6) 32 20.8 (14.4–29.1) 2.89 (1.49–5.57) 65.3 Uppsala-Orebro Region Yes 36 17.3 (12.5–23.4) 32 24.4 (17.3–33.6) 1.41 (0.87–2.27) 29.2 Stockholm Region Yes 16 6.00 (3.72–9.28) 72 34.9 (27.7–43.8) 5.81 (3.38–9.81) 82.9 Northern Region Yes 12 12.8 (7.41–21.1) 5 10.8 (4.74–22.0) 0.21 (0.07–0.59) – Notes: Incidence proportion per 100 000 infants [95% confidence intervals (CIs)], rate ratio (RR) 2008–2014/1997–2007 and aetiologic fraction among the exposed (EFe), those born preterm or small for gestational age (SGA) and multiple births, for having an abuse diagnosis. Source: Swedish National Board of Health and Welfare; Source: Statistics Sweden and the National Board of Health and Welfare. increased rate ratios, during 2008–2014, for subdural haemorrhage, The increasing incidence in Sweden over time is intriguing. How skull fracture, retinal haemorrhage and long bone fractures; is it that Sweden, previously having one of the lowest rates of infant increased, though non-significant, rate ratios for cerebral maltreatment in Western societies, has manifested this increase? Is contusion and convulsions; and aetiologic fractions ranging from it because previous years, i.e. the 1980s and early 1990s, were 26.4% to 58.5% (table 2). Having an abuse diagnosis and being characterized by underdiagnosis and hidden statistics, or is the preterm born, Small-for-Gestational-Age or a twin was associated increase due to overdiagnosis? Alternatively, has there been a real with greatly increased incidences during the latter period, with stat- increase in infant abuse? istically significant rate ratio changes and aetiologic fractions ranging Whether there is an increase in true positive or in false positive from 32.6% to 70% (table 2). Both infants of Swedish mothers and cases, the Swedish doubling in cases coincides with the fact that there infants of foreign-born mothers had increased risk ratios in the latter has been an increased awareness of SBS/AHT among Swedish 24,28 period, especially infants of foreign-born mothers. The distribution paediatricians during the latter period of our study, and that of cases of infant maltreatment was uneven by region and period, doctors may therefore have become more likely to make an infant with threefold geographical differences. No significant changes in the abuse diagnosis. A similar findingwas reported from a New Zealand coefficients from the SEM were observed when comparing the hospital study where the incidence of diagnosis of AHT quadrupled models for the separate time periods. from 1991 to 2010 after the establishment of a specialist child protection team; however, the possibility of overdiagnosis was not discussed. Unlike our national data for the years 2008–2014, Discussion national statistics from New Zealand on infant maltreatment- related injury admissions did not show an increase during the This study shows that diagnosis of infant abuse was strongly 8 29 years 1995–2010 contrary to the hospital study. associated with SBS/AHT criteria. Over 27 years, the diagnosis of One argument proposed for the hidden statistics hypothesis is a infant abuse in Sweden increased tenfold from the late 1980s to previously low identification rate due to a low proportion of 2014, doubling from 12.0 per 100 000 infants during the period Swedish paediatricians receiving training on child abuse, and 1997–2007–26.5 during the period 2008–2014. Pronounced that there are hidden cases of child abuse among infant deaths regional differences were observed. The doubling incidence was classified as ‘unknown cause of death’. However, this hypothesis attributed to the SBS/AHT criteria, region, being preterm born or was disproved in a recent study where all records of infants deceased SGA and multiple births. in 1994–2013 were analyzed for diagnosis of AHT, and where, Our results support the interpretation that diagnosis of infant compared with international statistics on AHT death, a tenfold abuse is associated with SBS/AHT criteria. The high Swedish national incidence of infant abuse diagnosis is still lower, even lower incidence is reported. The low case fatality rate in our study, 1.9%, does not correspond to the AHT fatality rate of 30% during the latter period, than the infant maltreatment diagnosis 27 10 8 reported from hospital settings in the USA. Irrespective of the true reported from the USA, Canada and the UK, but is comparable with Scotland’s incidence of shaken impact syndrome. aetiology of the lethal intracranial lesions, this disparity does not Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 Infant abuse diagnosis based on criteria for abusive head trauma 645 support hypotheses of either hidden statistics or a real increase, but they had been interpreted in different ways. Neither we know supports the hypothesis of overdiagnosis. whether there is underreporting of ICD-codes on abuse of There are several arguments supporting the overdiagnosis children being subjected for intervention of Social Service. hypothesis. The regional variation suggests that the diagnosis of However, the direction of information and detection bias might infant abuse is due to differences in applying the SBS/AHT lead to underestimation of the findings of this study. One further criteria. In Stockholm, cases of infant abuse began to rise steeply, limitation is the change from ICD-9 to ICD-10, which may have concomitant with the establishment of a child protection team and affected incidence changes. However detailed analysis was solely the introduction of a ‘mental vaccination’ against shaking restricted to ICD-10 codes. Although the conceptual Probit programme in 2008–2010. Similar intervention programmes in regression model fit was reasonable, with a comparative fit index the USA have reported decreased incidence of AHT. A possible of 0.86 and a root mean square error of approximation of 0.004, the interpretation of a contrary effect of the Swedish campaigns is that model did not account for the information in the observed the increase in diagnosed cases of infant maltreatment did not covariance matrix (P < 0.001). indicate a real increase in incidence, but instead, a prevalence of overdiagnosis because of enhanced awareness and preparedness of Conclusions SBS/AHT among health care professionals. Further, population surveys showed that Swedish parents reported shaking their This study shows that the incidence of diagnosis of infant abuse is children in 18% in 2006, while after an information campaign strongly associated with the diagnostic criteria of SBS/AHT; further, against shaking, parental reporting of shaking was reduced to that the increase in incidence of diagnosis of abuse coincides with almost nil by 2011. increased preparedness of doctors to make an SBS/AHT diagnosis. The diagnostic process of SBS/AHT has been, and still is, Hidden statistics or a real increase are a less plausible explanation. mainstream thinking in health care and considered a scientific Whether the increase is due to overdiagnosis cannot be answered fact, but this position has been critically challenged for the past with certainty by this study, but the possibility that overdiagnosis of 13–16 two decades and the evidence has now been contested by the cases may be a reason for the increase does raise ethical concerns for first independent systematic literature review ever completed medico-legal assessment in the context of child protection. 20,21 regarding this diagnosis. Whether the increase in diagnosis of infant abuse is due to overdiagnosis, i.e. changes in diagnostic classification, this Supplementary data comprises ethical principles of beneficence, non-maleficence and Supplementary data are available at EURPUB online. justice. In the USA, the sum of infant homicide rates and accident mortality rates remained constant during the years 1980–2005, while the proportion of infant accident mortality rates to infant homicide Acknowledgements rates decreased compared with the period 1940–1979. This may be We wish to thank Henrik Passmark, at the Swedish National Board explained by the changed diagnostic classification, since 1980, with of Health and Welfare, for linkage of the registers, and Per Wikman more diagnosis of homicide cases and fewer cases of accidents being for database management. diagnosed. The possibility of overdiagnosis as a reason for the increase observed in the present study raises medico-ethical Conflicts of interest: None declared. concerns relating to harm in health care, as non-evidence-based knowledge being practised and may affect families seeking health care for their infants. Key points How to interpret the strong association between perinatal risk exposure and infant abuse diagnosis in this study, especially in This study shows that the diagnosis of infant abuse is relation to the steep increase in abuse incidence during the latter strongly associated with SBS/AHT criteria. period? One explanation could be the risk of metabolic bone disease The incidence of abuse has doubled since 2008 and coincides 36–38 in relation to being born preterm or SGA or being twin-born. with increased preparedness of doctors to make a diagnosis Whether preterm and multiple births are more prone to having a of SBS/AHT. birth-related subdural haemorrhage has not yet been reported. The increase may be due to hidden statistics, a real increase Foreign-born mothers had a higher incidence risk ratio, and a or overdiagnosis, the former two are less plausible. The pos- higher attributable risk, than Swedish-born mothers, and further sibility of overdiagnosis as a reason for the increase raises displayed a higher risk coefficient for abuse diagnosis but not for medico-ethical concerns of how to improve public health AHT diagnosis compared with Swedish-born mothers. The reasons policy on child protection. for this could not be ascertained in this study. Differential diagnostic to abuse diagnosis of the SBS/AHT criteria applied in this study was not part of the study aim but the extent is displayed in table 2, and will be further analyzed in forthcoming References studies. 1 Kempe CH, Silverman FN, Steele BF, et al. The battered-child syndrome. JAMA Strengths and limitations 1962; 181:17–24. 2 Guthkelch AN. Infantile subdural haematoma and its relationship to whiplash A strength of the present study is that the health registers covered the injuries. Br Med J 1971; 2:430–1. whole country during the study period. More than 1/3 of cases of 3 Caffey J. The whiplash shaken infant syndrome: manual shaking by the extremities infant abuse during the years 1997–2014 had no additional code with whiplash-induced intracranial and intraocular bleedings, linked with residual explaining the diagnosis of abuse. This is a major limitation in permanent brain damage and mental retardation. Pediatrics 1974; 54:396–403. that we had no access to the individuals’ details and no checks had been made of clinical records, laboratory tests, or radiological 4 Maguire SA, Kemp AM, Lumb RC, Farewell DM. Estimating the probability of abusive head trauma: a pooled analysis. Pediatrics 2011; 128:e550–64. or imaging reports. This hindered the possibility to establish to what extent diagnostic criteria, more non-symptomatic signs as retinal 5 Vinchon M, de Foort-Dhellemmes S, Desurmont M, Delestret I. Confessed abuse haemorrhage, thin subdural haemorrhage, classic metaphyseal versus witnessed accidents in infants: comparison of clinical, radiological, and lesions, or non-dislocated rib fractures, had been missed or how ophthalmological data in corroborated cases. Childs Nerv Syst 2010; 26:637–45. Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 646 European Journal of Public Health 6 Christian CW. The evaluation of suspected child physical abuse, Committee on 23 Jansson B. Shaken baby syndrome. Svara hja ¨rnskador till fo ¨ ljd av barnmisshandel Child Abuse and Neglect, American Academy of. Pediatrics. Pediatrics 2015; kan uppta ¨ckas i o ¨ gonbotten (Swedish). La ¨kartidningen 1994; 91:491–9. 135:e1337–54. 24 Frasier LD, Kelly P, Al-Eissa M, Otterman GJ. International issues in abusive head 7 Berkowitz CD. Physical abuse of children. N Engl J Med 2017; 376:1659–66. trauma. Pediatr Radiol 2014; 44 Suppl 4:S647–53. 8 Gilbert R, Fluke J, O’Donnell M, et al. Child maltreatment: variation in trends and 25 R Core Team (2016). V, Austria. R: A language and environment for statistical policies in six developed countries. Lancet 2012; 379:758–72. computing. R Foundation for Statistical Computing Vienna, Austria, 2016. Available at: https://www.R-project.org/. 9 Parks S, Sugerman D, Xu L, Coronado V. Characteristics of non-fatal abusive head trauma among children in the USA, 2003–2008: application of the CDC operational 26 Rosseel Y. An R package for structural equation modeling. J Stat Softw 2012; case definition to national hospital inpatient data. Inj Prev 2012; 18:392–8. 48:1–36. 10 Fujiwara T, Barr RG, Brant RF, et al. Using International Classification of Diseases, 27 Gumbs GR, Keenan HT, Sevick CJ, et al. Infant abusive head trauma in a military 10th edition, codes to estimate abusive head trauma in children. Am J Prev Med cohort. Pediatrics 2013; 132:668–76. 2012; 43:215–20. ˚ ˚ 28 Tegern G, Tingho ¨ g P, Flodmark O. Om Att Fo ¨rebygga Skakvald Mot Sma Barn 11 Barlow KM, Minns RA. Annual incidence of shaken impact syndrome in young (Swedish). Stockholm: Karolinska Institutet and Linko ¨ ping University, 2012. children. Lancet 2000; 356:1571–2. 29 Kelly P, John S, Vincent AL, Reed P. Abusive head trauma and accidental head 12 Wilkins B, Sunderland R. Head injury: abuse or accident. Arch Dis Child 1997; injury: a 20-year comparative study of referrals to a hospital child protection team. 76:393–7. Arch Dis Child 2015; 100:1123–30. 13 Barnes PD. Imaging of nonaccidental injury and the mimics: issues and 30 Lahne K, Otterman, G, Janson S. Alltfo ¨rfa ˚ oklara do ¨ dsfall bland barn utreds i controversies in the era of evidence-based medicine. Radiol Clin North Am 2011; Sverige (Swedish). English Summary La ¨kartidningen 2013; 110:CD3P. 49:205–29. 31 Andersson J, Thiblin I. Abusive Head Trauma mortality in Sweden is at least 13-fold 14 Guthkelch AN. Problems of infant retion-dural hemorrhage with minimal injury. lower than reported in other western countries, suggesting previously reported Houston J Health Law Policy 2012; 201–7. statistics may be exaggerated. Acta Paediatr 2017; 107:477–83. 15 Squier W, Mack J. The neuropathology of infant subdural haemorrhage. Forensic Sci 32 Keenan HT, Runyan DK, Marshall SW, et al. A population-based study of inflicted Int 2009; 187:6–13. traumatic brain injury in young children. JAMA 2003; 290:621–6. 16 Gabaeff SC. Challenging the pathophysiologic connection between subdural 33 Altman RL, Canter J, Patrick PA, et al. Parent education by maternity nurses and hematoma, retinal hemorrhage and shaken baby syndrome. West J Emerg Med 2011; prevention of abusive head trauma. Pediatrics 2011; 128:e1164–72. 12:144–58. 34 Janson S, Lindblad B, Svensson B. Violence against Children in Sweden. A National 17 Plunkett J. Recognizing abusive head trauma in children. JAMA 1999; 282:1421–2. Survey 2006-2007 & 2011 (in Swedish). Stockholm: Allmanna Barnhuset and 18 Donohoe M. Evidence-based medicine and shaken baby syndrome: part I: literature Karlstad University, 2007 and 2011. review, 1966-1998. Am J Forensic Med Pathol 2003; 24:239–42. 35 Riggs JE, Hobbs GR. Infant homicide and accidental death in the United States, ¨ ¨ 19 Hogberg G, Aspelin P, Hogberg U, Colville-Ebeling B. Circularity bias in Abusive 1940–2005: ethics and epidemiological classification. J Med Ethics 2011; 37:445–8. Head Trauma studies could be diminished with a new ranking scale. Egypt J Forensic 36 Miller M, Ward T, Stolfi A, Ayoub D. Overrepresentation of multiple birth Sci 2016; 6:6–10. pregnancies in young infants with four metabolic bone disorders: further evidence 20 SBU. Traumatic Shaking—The Role of the Triad in Medical Investigations of Suspected that fetal bone loading is a critical determinant of fetal and young infant bone Traumatic Shaking: A Systematic Review. Stockholm: Swedish Agency for Health strength. Osteoporos Int 2014; 25:1861–73. Technology Assessment and Assessment of Social Services (SBU), 2016. 37 Paterson CR, Monk EA. Clinical and laboratory features of temporary britle bone 21 Lynoe N, Elinder G, Hallberg B, et al. Insufficient evidence for ‘shaken baby disease. J Pediatric Endocrin Met 2014; 27:37–45. syndrome’ - a systematic review. Acta Paediatr 2017; 106:1021–7. 38 Carroll DM, Doria AS, Paul BS. Clinical-radiological features of fractures in 22 Barnmisshandel (Swedish). Socialstyrelsen Redovisar. Stockholm: Swedish Board of premature infants: a review. J Perinat Med 2007; 35:366–75. Health and Welfare, 1969. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Journal of Public Health Oxford University Press

Infant abuse diagnosis associated with abusive head trauma criteria: incidence increase due to overdiagnosis?

Loading next page...
 
/lp/ou_press/infant-abuse-diagnosis-associated-with-abusive-head-trauma-criteria-0Odn0ReKR7

References (53)

Publisher
Oxford University Press
Copyright
Copyright © 2022 European Public Health Association
ISSN
1101-1262
eISSN
1464-360X
DOI
10.1093/eurpub/cky062
Publisher site
See Article on Publisher Site

Abstract

Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 Infant abuse diagnosis based on criteria for abusive head trauma 641 39 Kepper M, Tseng T-S, Volaufova J, et al. Pre-school obesity is inversely 40 Ministry of Education, Culture, Sports, Science, and Technology. School nutrition associated with vegetable intake, grocery stores and outdoor play. Pediatr Obes 2016; report, 2015. Available at: http://www.mext.go.jp/b_menu/toukei/chousa05/eiyou/ 11:e6–8. gaiyou/1376285.htm (19 September 2016, date last accessed). ......................................................................................................... The European Journal of Public Health, Vol. 28, No. 4, 641–646 The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. doi:10.1093/eurpub/cky062 Advance Access published on 17 April 2018 ......................................................................................................... Infant abuse diagnosis associated with abusive head trauma criteria: incidence increase due to overdiagnosis? 1 2 3 4 1 5 Ulf Ho¨ gberg , Erik Lampa ,Go¨ ran Ho¨ gberg , Peter Aspelin , Fredrik Serenius , Ingemar Thiblin 1 Department of Women’s and Children’s Health, Uppsala University, Uppsala, Sweden 2 UCR-Uppsala Clinical Research Centre, Uppsala University, Uppsala, Sweden 3 Child and Adolescent Psychiatric Unit, Department of Women’s and Children’s Health, Karolinska Institute, Stockholm, Sweden 4 Department of Clinical Science, Intervention and Technology (CLINTEC), Karolinska Institute, Stockholm, Sweden 5 Forensic Medicine, Department of Surgical Sciences, Uppsala University, Uppsala, Sweden Correspondence: Ulf Ho¨ gberg, Uppsala University, Department of Women’s and Children’s Health, Akademiska Sjukhuset, SE-751 85 Uppsala, Sweden, Tel: +46 186115246, Fax: +46 186115583, e-mail: [email protected] Background: The hypothesis of this study is that the diagnosis of infant abuse is associated with criteria for shaken baby syndrome (SBS)/abusive head trauma (AHT), and that that changes in incidence of abuse diagnosis in infants may be due to increased awareness of SBS/AHT criteria. Methods: This was a population-based register study. Setting: Register study using the Swedish Patient Register, Medical Birth Register, and Cause of Death Register. The diagnosis of infant abuse was based on the International Classification of Diseases, 9th and 10th revision. Participants: All children born in Sweden during 1987–2014 with a follow-up until 1 year of age (N = 2 868 933). SBS/AHT criteria: subdural haemorrhage, cerebral contusion, skull fracture, convulsions, retinal haemorrhage, fractures rib and long bones. Outcomes: Incidence, rate ratios, aetiologic fractions and Probit regression analysis. Results: Diagnosis of infant abuse was strongly associated with SBS/AHT criteria, but not risk exposure as region, foreign-born mother, being born preterm, multiple birth and small for gestational age. The incidence of infant abuse has increased tenfold in Sweden since the 1990s and has doubled since 2008, from 12.0 per 100 000 infants during 1997–2007 to 26.5/100 000 during 2008–2014, with pronounced regional disparities. Conclusions: Diagnosis of infant abuse is related to SBS/ AHT criteria. The increase in incidence coincides with increased medical preparedness to make a diagnosis of SBS/ AHT. Hidden statistics and a real increase in abuse are less plausible. Whether the increase is due to overdiagnosis cannot be answered with certainty, but the possibility raises ethical and medico-legal concerns. ......................................................................................................... The incidence of maltreatment syndrome has shown a variation Introduction by country, from 11.5 per 100 000 infants in Sweden (1987–2009), n 1962 battered child syndrome was first described in a clinical to 34.6 in England (1997–2008), to 118.9 in Western Australia I study. In 1972 it was proposed that subdural haematomas could (1980–2005), with only England showing a declining trend. The be caused by whiplash shaking and in 1974 the term ‘shaken baby incidence of non-fatal AHT among infants, based on the case syndrome (SBS)’ was coined to describe a condition inflicted by definition of the US Center of Disease Control and Prevention violent shaking and identified by the triad retinal haemorrhage, (CDC), was 32.3 per 100 000, while a Canadian study on infant subdural haemorrhage and encephalopathy. The diagnosis of AHT found an incidence of 13.0–15.5 for 2002–2007. In Scotland abusive head trauma (AHT), departing from the three diagnostic in 1998–1999, the incidence of shaken impact syndrome among criteria for SBS, now has a broader categorization that may also infants was 24.6 per 100 000. include apnoea, seizures, fractures of the skull, metaphyses and shaft Shaken baby syndrome has, since 1997, been questioned as a of long bones and ribs, and inability of parents/carers to provide an diagnostic entity, and emerging imaging technology demands further 13–18 explanation for accidental trauma, the former findings and symptoms differential diagnostic considerations. Furthermore, the precision 4–7 being considered highly specific for non-accidental trauma. of the SBS/AHT diagnosis is lost when excluding the diagnostic criterion Shaken baby syndrome/AHT is not a diagnosis classified in the of incompatibility between the carer’s report of history and the International Classification of Diseases (ICD-9/ICD-10); however, investigating doctor’s interpretation of the findings. In 2016, a physical abuse and battered baby or child syndrome is defined as systematic literature review by the Swedish Agency for Health maltreatment syndrome. In an intercountry epidemiological study, Technology Assessment and Assessment of Social Service (SBU) Gilbert et al. described the maltreatment syndrome/assault and concluded that there is limited scientific evidence that the triad can be included in their classification the diagnoses of intracranial injury explained by isolated shaking, and that there is insufficient evidence to 8 20,21 and long bone fractures. assess the diagnostic accuracy of the triad to identify SBS/AHT. Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 642 European Journal of Public Health Table 1 Definitions of infant abuse, and other diagnoses, according to the Swedish Patient Registry, National Board of Health and Welfare (ICD-9: 1987–1996; ICD-10: 1997–2014) Diagnosis category ICD code Infant abuse diagnosis Observation for suspected Z03.8K (ICD-10); E967 (ICD-9) abuse Battered baby syndrome Y07.9 (ICD-10); 995F (ICD-9) Maltreatment syndrome T74.1 (ICD-10); 995F (ICD-9) Neglect and adandonment Y06 Other maltreatment Y07 Subdural haemorrhage I62.0, S06.5 (ICD-10) Skull fracture S02.0, S02.1, S02.8, S02.09 S02.00, S02.9 (ICD-10) Cerebral contusion S06.0, S06.1 (ICD-10) Convulsions R56, R56.8, G40–41, R56.0 (ICD-10) Retinal haemorrhage H35.6 (ICD-10) Rib fracture S22.3, S22.4 (ICD-10) Long bone fracture S42.2, S42.3, S42, 4, S42.7, S42.8, S52, S72, S82, T10, T12 (ICD-10) In Sweden, battered child syndrome was introduced as a 100 000 [95% confidence interval (CI)], and the two periods were diagnostic criterion during the 1960s, and the triad of SBS was compared in terms of rate ratios (I /I ) (95% CI) and aetiologic 1 0 described in the Swedish Medical Journal in 1994. In 2008, fractions of incidence (I -I /I ) (%). The following covariates were 1 0 1 Stockholm County Council published clinical guidelines for the in- selected for further analysis because of a proposed association with vestigation of SBS according to the AHT criteria and the document the diagnosis of infant maltreatment: maternal country of birth; has been adapted for use in other parts of Sweden. In 2009, the infant perinatal characteristics such as multiple birth, preterm Swedish Paediatric Society established a child abuse task force to birth (born at <37 weeks’ gestation), born Small-for-Gestational- foster awareness of and training in child abuse recognition and Age (SGA) [<2.5 standard deviation (SD)]; diagnosis by health reporting, and implementation of the guidelines for investigation care region. and diagnosis of AHT. Subsequently, regional hospital-based child Probit regression analysis was performed in a Structural Equation protection centres were established. Modelling (SEM) framework, in order to examine a conceptual This study examines the incidence of diagnosis of infant abuse model to explore the interplay of the variables of importance for over three decades in Sweden. The hypothesis of our study is that the outcome ‘diagnosis of infant abuse’. The conceptual model is diagnosis of infant abuse is based on SBS/AHT criteria, and that shown in figure 2. Implicit in the model is a latent diagnosis variable changes in incidence of diagnosis of infant abuse may be due to that is observed across the different diagnoses. The importance of increased awareness of SBS/AHT criteria. the different observed diagnoses for the latent diagnoses variable was assessed by the factor loadings, with higher loading values implying greater importance. Methods Probit regression, like logistic regression, models the probability This is a population-based register study of children born in Sweden of an outcome variable being equal to one of two values and can be between 1987 and 2014, with a follow-up until 1 year of age. written as: The number of children born during the period 1987–2014 PYðÞ ¼ 1jX¼ X b ; (N = 2 984 813) was retrieved from Statistics Sweden. The research database was linked to the Swedish Patient Register, the Swedish where X is the data matrix and  is the vector of regression coeffi- Medical Birth Register and the Swedish Cause of Death Register, cients. denotes the cumulative distribution function of the standard with diagnoses and conditions classified using ICD-9 (1987–1996) normal distribution. The coefficient of the latent diagnosis variable and ICD-10 (1997–2015) codes. Infant abuse was defined according in the SEM Probit model should be interpreted as a change in to the Swedish version of the ICD-9 and ICD-10 (table 1). For the z-score which allows calculation of the probability of an SD years 1997–2014, ICD-10 codes for other possible differential increase in the latent diagnosis variable. To investigate whether as- diagnoses (table 1), symptoms and conditions which might be sociations changed over time, the SEM model was divided into two 6,7,13 considered when investigating infant abuse, were searched for. groups defined by the time period and the coefficients compared. In all, 182 974 infants with diagnostic codes were found, for each a The statistical software package IBM SPSS 25.0 (SPSS Inc., control was selected, born in the same year and without any Chicago, IL, USA) and R version 3.3.1 were used for data diagnosis in the Patient Register (N = 731 901). In total, the analyses. The SEM models were fitted using the lavaan add-on sample consisted of 914 875 infants, 49% of all infants born package for R. during 1997–2014. For this study, we selected as criteria for SBS/ The study was approved by the Regional Ethical Committee in AHT to be associated with abuse diagnosis:subdural haemorrhage, Uppsala (2014-11-19 No. 383). cerebral contusion, skull fracture, convulsions, retinal haemorrhage, 4,6–8,20 fractures of rib and long bones (table 1). The baseline descriptive analysis was based on the incidence Results proportion of diagnosis of infant abuse per 100 000 by year (moving annual average). For thorough analysis of diagnosis, only Altogether 368 infants were diagnosed with maltreatment during the years 1987–2014; 12.3 per 100 000 infants. Seven of the children years with ICD-10 coding were selected, 1997–2014. We based our analyses on incidence of diagnosis in the population and, for cases diagnosed with abuse died, i.e. 0.23 deaths per 100 000 infants and diagnosed with abuse, for the whole period divided by the two a case fatality rate of 1.9%. A tenfold increase in diagnosis of infant periods 1997–2007 and 2008–2014, before and after the increased abuse was observed from the start to the end of the period. awareness of SBS/AHT criteria was established in the Swedish The increase started slowly from 1996, peaking during the years setting. Diagnoses were selected by the incidence proportion per 2008–2014 (figure 1). There was a doubling of the incidence of Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 Infant abuse diagnosis based on criteria for abusive head trauma 643 Infant abuse diagnosis in Sweden 1987-2014 Moving annual means for infants born 1987-2014 Figure 1 Infant abuse diagnosis in Sweden for children born in 1987–2014 per 100 000 infants (moving annual average) Figure 2 Conceptual model for diagnosis of infant abuse in Sweden in 1997–2014, performed in a Structural Equation Modelling (SEM) framework. Observed variables are given inside rectangles. The latent diagnosis variable is drawn as an ellipse. Arrows radiating out from the ellipse and pointing to the observed diagnoses represent factor loadings. Coefficients represented by arrows pointing to ‘Infant abuse diagnosis’ should be interpreted as the increase in z-score for the probability of observing the outcome. For the risk exposures, the coefficients represent the increase in z-score if the condition is present. For the latent diagnosis variable, the coefficient represents the increase in z-score for a standard deviation (SD) increase in the latent diagnosis variable. The coefficient for the Southern health region was set to zero as a reference for the other health regions. SGA = small for gestational age abuse diagnosis during the years 2008–2014 compared with 1997– and skull fracture, had substantial loadings on the latent diagnosis 2007, from 12.0 (n = 127)–26.5 (n = 210) per 100 000 infants [odds variable suggesting that these four diagnoses were the main drivers ratio (OR) 2.21 (95% CI 1.78–2.76)]. of the association with the probability for diagnosis of infant abuse. Of the 337 cases of infant abuse during the years 1997–2014, 137 All diagnoses taken together were strongly associated with this prob- had either concomitant diagnoses of subdural haemorrhage, skull ability [coefficient 0.752 (95% CI 0.709–0.795 (P < 0.01)]. If the fracture, cerebral contusion, traumatic brain injury, convulsions, baseline probability of diagnosis of infant abuse is around 0.01% retinal haemorrhage, rib fracture, or fractures of the long bones. with all other factors held at their reference values, an SD increase in Out of the residual, 200 cases, 11 had injury to the head and 14 to the latent diagnosis variable would push that probability to the body, nine had injury of the eye or orbital fracture, five had a (3.72 + 0.75) = 0.15%. Coefficients for diagnoses were increased fractured clavicle, three had burns, three had diagnosis of failure to for being born preterm and SGA. Coefficients for diagnosis of thrive; others had miscellaneous diagnoses such as infections, skin infant abuse were increased when being born preterm and having disease, nausea, obstipation and icterus. Altogether 125 had no other a foreign-born mother. diagnosis besides abuse (table 1), whereof 60 had the diagnosis ‘ob- With the exception of convulsions, no statistically significant servation for suspected abuse’ and 35 had a diagnosis of maltreat- changes in population incidence of subdural haemorrhage, skull ment syndrome. fracture, cerebral contusion, retinal haemorrhage, rib fracture or The results of the Probit regression analysis are presented in fractures of the long bones could be seen between the two periods figure 2 and Supplementary table S1. Four of the observed 1997–2007 and 2008–2014 (table 2). However, cases diagnosed with diagnoses: subdural haemorrhage, rib fracture, retinal haemorrhage infant abuse with those diagnoses had statistically significantly Per 100,000 infants 2014 Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 644 European Journal of Public Health Table 2 Diagnosis and background characteristics of infants or children with maltreatment syndrome in Sweden, 1997–2007 (n = 127) and 2008–2014 (n = 210), per 100 000 children below 1 year of age (1997–2007 n = 1, 062, 084; 2008–2014 n = 793, 183) Diagnosis and background Abuse diagnosis 1997–2007 2008–2014 2008–2014/1997–2007 Characteristics n Incidence (95% CI) n Incidence (95% CI) RR (95% CI) EFe (%) Subdural haemorrhage No 130 12.3 (10.4–14.5) 133 16.8 (14.2–19.7) Yes 20 1.89 (1.22–2.79) 23 2.90 (1.94–4.20) 1.36 (1.07–1.73) 26.4 Skull fracture No 876 82.5 (77.2–88.0) 612 77.1 (71.3–83.4) Yes 12 1.13 (0.65–1.85) 21 2.65 (1.06–2.8) 2.34 (1.15–4.76) 57.3 Cerebral contusion No 5357 504 (491–518) 3, 731 470 (455–486) Yes 4 0.37 (0.15–0.83) 5 0.63 (0.28–1.29) 1.67 (0.45–6.23) 40.3 Convulsions No 7225 680 (665–696) 5, 931 747 (729–767) Yes 7 0.66 (0.33–1.23) 11 1.39 (0.78–2.32) 2.10 (0.82–5.43) 52.5 Retinal haemorrhage No 70 6.59 (5.22–8.22) 49 6.18 (4.68–8.02) Yes 10 0.94 (0.52–1.61) 18 2.27 (1.44–3.43) 2.41 (1.11–5.22) 58.5 Long bone fractures No 1157 108 (103–115) 993 125 (117–133) Yes 22 2.07 (1.37–3.02) 35 4.41 (3.18–5.99) 2.13 (1.25–3.63) 53.1 Rib fracture No 26 2.44 (1.68–3.47) 20 2.52 (1.64–3.74) Yes 12 1.13 (0.65–1.85) 20 2.01 (1.25–3.12) 2.32 (1.09–4.57) 55.2 Preterm-born (<37 wks) Yes 18 27.3 (17.3–41.3) 27 57.2 (39.4–80.6) 2.09 (1.15–3.80) 52.2 Small for gestational age Yes 7 29.6 (14.6–55.3) 8 43.9 (22.6–79.2) 1.48 (0.54–4.09) 32.6 Multiple birth Yes 6 38.0 (17.9–74.0) 14 127 (76.2–202) 3.34 (1.28–8.68) 70.0 Mother Swedish-born Yes 96 11.1 (9.13–13.5) 124 12.3 (12.4–18.5) 1.87 (1.43–2.44) 46.4 Mother foreign-born Yes 31 15.6 (11.0–21.6 86 44.6 (35.9–54.2) 2.71 (1.80–4.09) 63.1 Southern Region Yes 27 13.2 (9.07–18.6) 53 33.4 (25.6–43.0) 2.54 (1.60–4.03) 60.6 South East Region Yes 22 20.4 (13.5–29.7) 19 24.2 (15.5–36.2) 1.59 (0.86–2.94) 37.1 Western Region Yes 13 7.2 (4.24–11.6) 32 20.8 (14.4–29.1) 2.89 (1.49–5.57) 65.3 Uppsala-Orebro Region Yes 36 17.3 (12.5–23.4) 32 24.4 (17.3–33.6) 1.41 (0.87–2.27) 29.2 Stockholm Region Yes 16 6.00 (3.72–9.28) 72 34.9 (27.7–43.8) 5.81 (3.38–9.81) 82.9 Northern Region Yes 12 12.8 (7.41–21.1) 5 10.8 (4.74–22.0) 0.21 (0.07–0.59) – Notes: Incidence proportion per 100 000 infants [95% confidence intervals (CIs)], rate ratio (RR) 2008–2014/1997–2007 and aetiologic fraction among the exposed (EFe), those born preterm or small for gestational age (SGA) and multiple births, for having an abuse diagnosis. Source: Swedish National Board of Health and Welfare; Source: Statistics Sweden and the National Board of Health and Welfare. increased rate ratios, during 2008–2014, for subdural haemorrhage, The increasing incidence in Sweden over time is intriguing. How skull fracture, retinal haemorrhage and long bone fractures; is it that Sweden, previously having one of the lowest rates of infant increased, though non-significant, rate ratios for cerebral maltreatment in Western societies, has manifested this increase? Is contusion and convulsions; and aetiologic fractions ranging from it because previous years, i.e. the 1980s and early 1990s, were 26.4% to 58.5% (table 2). Having an abuse diagnosis and being characterized by underdiagnosis and hidden statistics, or is the preterm born, Small-for-Gestational-Age or a twin was associated increase due to overdiagnosis? Alternatively, has there been a real with greatly increased incidences during the latter period, with stat- increase in infant abuse? istically significant rate ratio changes and aetiologic fractions ranging Whether there is an increase in true positive or in false positive from 32.6% to 70% (table 2). Both infants of Swedish mothers and cases, the Swedish doubling in cases coincides with the fact that there infants of foreign-born mothers had increased risk ratios in the latter has been an increased awareness of SBS/AHT among Swedish 24,28 period, especially infants of foreign-born mothers. The distribution paediatricians during the latter period of our study, and that of cases of infant maltreatment was uneven by region and period, doctors may therefore have become more likely to make an infant with threefold geographical differences. No significant changes in the abuse diagnosis. A similar findingwas reported from a New Zealand coefficients from the SEM were observed when comparing the hospital study where the incidence of diagnosis of AHT quadrupled models for the separate time periods. from 1991 to 2010 after the establishment of a specialist child protection team; however, the possibility of overdiagnosis was not discussed. Unlike our national data for the years 2008–2014, Discussion national statistics from New Zealand on infant maltreatment- related injury admissions did not show an increase during the This study shows that diagnosis of infant abuse was strongly 8 29 years 1995–2010 contrary to the hospital study. associated with SBS/AHT criteria. Over 27 years, the diagnosis of One argument proposed for the hidden statistics hypothesis is a infant abuse in Sweden increased tenfold from the late 1980s to previously low identification rate due to a low proportion of 2014, doubling from 12.0 per 100 000 infants during the period Swedish paediatricians receiving training on child abuse, and 1997–2007–26.5 during the period 2008–2014. Pronounced that there are hidden cases of child abuse among infant deaths regional differences were observed. The doubling incidence was classified as ‘unknown cause of death’. However, this hypothesis attributed to the SBS/AHT criteria, region, being preterm born or was disproved in a recent study where all records of infants deceased SGA and multiple births. in 1994–2013 were analyzed for diagnosis of AHT, and where, Our results support the interpretation that diagnosis of infant compared with international statistics on AHT death, a tenfold abuse is associated with SBS/AHT criteria. The high Swedish national incidence of infant abuse diagnosis is still lower, even lower incidence is reported. The low case fatality rate in our study, 1.9%, does not correspond to the AHT fatality rate of 30% during the latter period, than the infant maltreatment diagnosis 27 10 8 reported from hospital settings in the USA. Irrespective of the true reported from the USA, Canada and the UK, but is comparable with Scotland’s incidence of shaken impact syndrome. aetiology of the lethal intracranial lesions, this disparity does not Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 Infant abuse diagnosis based on criteria for abusive head trauma 645 support hypotheses of either hidden statistics or a real increase, but they had been interpreted in different ways. Neither we know supports the hypothesis of overdiagnosis. whether there is underreporting of ICD-codes on abuse of There are several arguments supporting the overdiagnosis children being subjected for intervention of Social Service. hypothesis. The regional variation suggests that the diagnosis of However, the direction of information and detection bias might infant abuse is due to differences in applying the SBS/AHT lead to underestimation of the findings of this study. One further criteria. In Stockholm, cases of infant abuse began to rise steeply, limitation is the change from ICD-9 to ICD-10, which may have concomitant with the establishment of a child protection team and affected incidence changes. However detailed analysis was solely the introduction of a ‘mental vaccination’ against shaking restricted to ICD-10 codes. Although the conceptual Probit programme in 2008–2010. Similar intervention programmes in regression model fit was reasonable, with a comparative fit index the USA have reported decreased incidence of AHT. A possible of 0.86 and a root mean square error of approximation of 0.004, the interpretation of a contrary effect of the Swedish campaigns is that model did not account for the information in the observed the increase in diagnosed cases of infant maltreatment did not covariance matrix (P < 0.001). indicate a real increase in incidence, but instead, a prevalence of overdiagnosis because of enhanced awareness and preparedness of Conclusions SBS/AHT among health care professionals. Further, population surveys showed that Swedish parents reported shaking their This study shows that the incidence of diagnosis of infant abuse is children in 18% in 2006, while after an information campaign strongly associated with the diagnostic criteria of SBS/AHT; further, against shaking, parental reporting of shaking was reduced to that the increase in incidence of diagnosis of abuse coincides with almost nil by 2011. increased preparedness of doctors to make an SBS/AHT diagnosis. The diagnostic process of SBS/AHT has been, and still is, Hidden statistics or a real increase are a less plausible explanation. mainstream thinking in health care and considered a scientific Whether the increase is due to overdiagnosis cannot be answered fact, but this position has been critically challenged for the past with certainty by this study, but the possibility that overdiagnosis of 13–16 two decades and the evidence has now been contested by the cases may be a reason for the increase does raise ethical concerns for first independent systematic literature review ever completed medico-legal assessment in the context of child protection. 20,21 regarding this diagnosis. Whether the increase in diagnosis of infant abuse is due to overdiagnosis, i.e. changes in diagnostic classification, this Supplementary data comprises ethical principles of beneficence, non-maleficence and Supplementary data are available at EURPUB online. justice. In the USA, the sum of infant homicide rates and accident mortality rates remained constant during the years 1980–2005, while the proportion of infant accident mortality rates to infant homicide Acknowledgements rates decreased compared with the period 1940–1979. This may be We wish to thank Henrik Passmark, at the Swedish National Board explained by the changed diagnostic classification, since 1980, with of Health and Welfare, for linkage of the registers, and Per Wikman more diagnosis of homicide cases and fewer cases of accidents being for database management. diagnosed. The possibility of overdiagnosis as a reason for the increase observed in the present study raises medico-ethical Conflicts of interest: None declared. concerns relating to harm in health care, as non-evidence-based knowledge being practised and may affect families seeking health care for their infants. Key points How to interpret the strong association between perinatal risk exposure and infant abuse diagnosis in this study, especially in This study shows that the diagnosis of infant abuse is relation to the steep increase in abuse incidence during the latter strongly associated with SBS/AHT criteria. period? One explanation could be the risk of metabolic bone disease The incidence of abuse has doubled since 2008 and coincides 36–38 in relation to being born preterm or SGA or being twin-born. with increased preparedness of doctors to make a diagnosis Whether preterm and multiple births are more prone to having a of SBS/AHT. birth-related subdural haemorrhage has not yet been reported. The increase may be due to hidden statistics, a real increase Foreign-born mothers had a higher incidence risk ratio, and a or overdiagnosis, the former two are less plausible. The pos- higher attributable risk, than Swedish-born mothers, and further sibility of overdiagnosis as a reason for the increase raises displayed a higher risk coefficient for abuse diagnosis but not for medico-ethical concerns of how to improve public health AHT diagnosis compared with Swedish-born mothers. The reasons policy on child protection. for this could not be ascertained in this study. Differential diagnostic to abuse diagnosis of the SBS/AHT criteria applied in this study was not part of the study aim but the extent is displayed in table 2, and will be further analyzed in forthcoming References studies. 1 Kempe CH, Silverman FN, Steele BF, et al. The battered-child syndrome. JAMA Strengths and limitations 1962; 181:17–24. 2 Guthkelch AN. Infantile subdural haematoma and its relationship to whiplash A strength of the present study is that the health registers covered the injuries. Br Med J 1971; 2:430–1. whole country during the study period. More than 1/3 of cases of 3 Caffey J. The whiplash shaken infant syndrome: manual shaking by the extremities infant abuse during the years 1997–2014 had no additional code with whiplash-induced intracranial and intraocular bleedings, linked with residual explaining the diagnosis of abuse. This is a major limitation in permanent brain damage and mental retardation. Pediatrics 1974; 54:396–403. that we had no access to the individuals’ details and no checks had been made of clinical records, laboratory tests, or radiological 4 Maguire SA, Kemp AM, Lumb RC, Farewell DM. Estimating the probability of abusive head trauma: a pooled analysis. Pediatrics 2011; 128:e550–64. or imaging reports. This hindered the possibility to establish to what extent diagnostic criteria, more non-symptomatic signs as retinal 5 Vinchon M, de Foort-Dhellemmes S, Desurmont M, Delestret I. Confessed abuse haemorrhage, thin subdural haemorrhage, classic metaphyseal versus witnessed accidents in infants: comparison of clinical, radiological, and lesions, or non-dislocated rib fractures, had been missed or how ophthalmological data in corroborated cases. Childs Nerv Syst 2010; 26:637–45. Downloaded from https://academic.oup.com/eurpub/article/28/4/641/4974823 by DeepDyve user on 19 July 2022 646 European Journal of Public Health 6 Christian CW. The evaluation of suspected child physical abuse, Committee on 23 Jansson B. Shaken baby syndrome. Svara hja ¨rnskador till fo ¨ ljd av barnmisshandel Child Abuse and Neglect, American Academy of. Pediatrics. Pediatrics 2015; kan uppta ¨ckas i o ¨ gonbotten (Swedish). La ¨kartidningen 1994; 91:491–9. 135:e1337–54. 24 Frasier LD, Kelly P, Al-Eissa M, Otterman GJ. International issues in abusive head 7 Berkowitz CD. Physical abuse of children. N Engl J Med 2017; 376:1659–66. trauma. Pediatr Radiol 2014; 44 Suppl 4:S647–53. 8 Gilbert R, Fluke J, O’Donnell M, et al. Child maltreatment: variation in trends and 25 R Core Team (2016). V, Austria. R: A language and environment for statistical policies in six developed countries. Lancet 2012; 379:758–72. computing. R Foundation for Statistical Computing Vienna, Austria, 2016. Available at: https://www.R-project.org/. 9 Parks S, Sugerman D, Xu L, Coronado V. Characteristics of non-fatal abusive head trauma among children in the USA, 2003–2008: application of the CDC operational 26 Rosseel Y. An R package for structural equation modeling. J Stat Softw 2012; case definition to national hospital inpatient data. Inj Prev 2012; 18:392–8. 48:1–36. 10 Fujiwara T, Barr RG, Brant RF, et al. Using International Classification of Diseases, 27 Gumbs GR, Keenan HT, Sevick CJ, et al. Infant abusive head trauma in a military 10th edition, codes to estimate abusive head trauma in children. Am J Prev Med cohort. Pediatrics 2013; 132:668–76. 2012; 43:215–20. ˚ ˚ 28 Tegern G, Tingho ¨ g P, Flodmark O. Om Att Fo ¨rebygga Skakvald Mot Sma Barn 11 Barlow KM, Minns RA. Annual incidence of shaken impact syndrome in young (Swedish). Stockholm: Karolinska Institutet and Linko ¨ ping University, 2012. children. Lancet 2000; 356:1571–2. 29 Kelly P, John S, Vincent AL, Reed P. Abusive head trauma and accidental head 12 Wilkins B, Sunderland R. Head injury: abuse or accident. Arch Dis Child 1997; injury: a 20-year comparative study of referrals to a hospital child protection team. 76:393–7. Arch Dis Child 2015; 100:1123–30. 13 Barnes PD. Imaging of nonaccidental injury and the mimics: issues and 30 Lahne K, Otterman, G, Janson S. Alltfo ¨rfa ˚ oklara do ¨ dsfall bland barn utreds i controversies in the era of evidence-based medicine. Radiol Clin North Am 2011; Sverige (Swedish). English Summary La ¨kartidningen 2013; 110:CD3P. 49:205–29. 31 Andersson J, Thiblin I. Abusive Head Trauma mortality in Sweden is at least 13-fold 14 Guthkelch AN. Problems of infant retion-dural hemorrhage with minimal injury. lower than reported in other western countries, suggesting previously reported Houston J Health Law Policy 2012; 201–7. statistics may be exaggerated. Acta Paediatr 2017; 107:477–83. 15 Squier W, Mack J. The neuropathology of infant subdural haemorrhage. Forensic Sci 32 Keenan HT, Runyan DK, Marshall SW, et al. A population-based study of inflicted Int 2009; 187:6–13. traumatic brain injury in young children. JAMA 2003; 290:621–6. 16 Gabaeff SC. Challenging the pathophysiologic connection between subdural 33 Altman RL, Canter J, Patrick PA, et al. Parent education by maternity nurses and hematoma, retinal hemorrhage and shaken baby syndrome. West J Emerg Med 2011; prevention of abusive head trauma. Pediatrics 2011; 128:e1164–72. 12:144–58. 34 Janson S, Lindblad B, Svensson B. Violence against Children in Sweden. A National 17 Plunkett J. Recognizing abusive head trauma in children. JAMA 1999; 282:1421–2. Survey 2006-2007 & 2011 (in Swedish). Stockholm: Allmanna Barnhuset and 18 Donohoe M. Evidence-based medicine and shaken baby syndrome: part I: literature Karlstad University, 2007 and 2011. review, 1966-1998. Am J Forensic Med Pathol 2003; 24:239–42. 35 Riggs JE, Hobbs GR. Infant homicide and accidental death in the United States, ¨ ¨ 19 Hogberg G, Aspelin P, Hogberg U, Colville-Ebeling B. Circularity bias in Abusive 1940–2005: ethics and epidemiological classification. J Med Ethics 2011; 37:445–8. Head Trauma studies could be diminished with a new ranking scale. Egypt J Forensic 36 Miller M, Ward T, Stolfi A, Ayoub D. Overrepresentation of multiple birth Sci 2016; 6:6–10. pregnancies in young infants with four metabolic bone disorders: further evidence 20 SBU. Traumatic Shaking—The Role of the Triad in Medical Investigations of Suspected that fetal bone loading is a critical determinant of fetal and young infant bone Traumatic Shaking: A Systematic Review. Stockholm: Swedish Agency for Health strength. Osteoporos Int 2014; 25:1861–73. Technology Assessment and Assessment of Social Services (SBU), 2016. 37 Paterson CR, Monk EA. Clinical and laboratory features of temporary britle bone 21 Lynoe N, Elinder G, Hallberg B, et al. Insufficient evidence for ‘shaken baby disease. J Pediatric Endocrin Met 2014; 27:37–45. syndrome’ - a systematic review. Acta Paediatr 2017; 106:1021–7. 38 Carroll DM, Doria AS, Paul BS. Clinical-radiological features of fractures in 22 Barnmisshandel (Swedish). Socialstyrelsen Redovisar. Stockholm: Swedish Board of premature infants: a review. J Perinat Med 2007; 35:366–75. Health and Welfare, 1969.

Journal

European Journal of Public HealthOxford University Press

Published: Aug 1, 2018

Keywords: infant; craniocerebral trauma; diagnosis; overdiagnosis; child; hematoma, subdural; infant, small for gestational age; retinal hemorrhage; skull fractures

There are no references for this article.