Background Exposure to childhood household dysfunction increases the risk of psychiatric morbidity. Although school performance also has been linked with psychiatric morbidity, limited research has considered school performance as a mediating factor. To address this gap in the literature, the current register study examined whether school performance mediates the association between childhood household dysfunction (experienced between birth and age 14 years) and psychiatric care utilisation in young adulthood.
Methods We used a Swedish cohort of 96 399 individuals born during 1987–1991. Indicators of childhood household dysfunction were familial death, parental substance abuse and psychiatric morbidity, parental somatic disease, parental criminality, parental separation/single-parent household, public assistance recipiency and residential instability. Final school grades from the 9th year of compulsory school were used to create five categories. Estimates of risk of psychiatric care utilisation (measured as inpatient, outpatient and primary care) after the age of 18 years were calculated as HRs with 95% CIs. Mediation was tested with the bootstrap approach.
Results Cumulative exposure to childhood household dysfunction was positively associated with psychiatric care utilisation. Specifically, individuals exposed to three or more indicators with incomplete school grades had the highest risk (HR=3.7 (95% CI 3.3 to 4.1) after adjusting for demographics), compared to individuals exposed to no indicators with highest grades. School performance was found to mediate the relationship.
Conclusions Our findings suggest that future efforts to prevent or mitigate the negative effects of childhood household dysfunction on psychiatric morbidity may benefit from integration of strategies that improve school performance among vulnerable youth.
- Lifecourse / Childhood Circumstances
Statistics from Altmetric.com
Research strongly suggests that childhood household dysfunction increases the risk of psychiatric disorders.1–7 Childhood household dysfunction is characterised by parental separation or divorce, substance abuse in the home, a mentally ill family member, an incarcerated parent, etc.1 Indicators of childhood household dysfunction tend to occur in clusters rather than as single events or experiences.1–6 ,8–10 Moreover, clustered risk factors in childhood have a strong graded relationship to psychiatric morbidity later in life, in terms of depression,9 substance abuse disorder4 and psychotropic drug consumption.8 Accordingly, the greater the number of exposures, the greater the risk of developing a mental illness.
Conceptually, exposure to childhood household dysfunction may shape adult health through different processes.11 They may influence health outcomes through biological imprint processes; for example, early life stress may contribute to cause enduring brain dysfunction that adversely affects health and quality of life throughout the lifespan.1 ,12 ,13 Pathways from a psychological and psychosocial perspective may include the damaging impact of childhood household dysfunction on the child's emotion regulation, self-worth and self-esteem,3 ,6 which in turn may influence long-term adverse health outcomes.
Among well-established and important mechanisms through which exposure to household dysfunction in childhood may adversely impact later health, educational attainment and school performance have been pointed out.11 ,14 ,15 Exposure to household dysfunction may lead to worse school performance,16–18 and poor school performance in turn is a powerful predictor of psychiatric morbidity.14 ,19–21 One study demonstrated a mediating effect whereby children who experienced childhood adversity tended to have lower levels of education or lower grade point average (GPA), which led to poor adult health.11
Despite earlier findings, little emphasis has been placed on school performance as a possible link between cumulative exposure to childhood household dysfunction and the increased risk of later psychiatric morbidity. To address this gap in the literature, the current register-based study used a large sample of approximately 96 000 adolescents born in the Stockholm County between 1987 and 1991 to test the hypotheses that (1) cumulative exposure to childhood household dysfunction increases the risk of psychiatric care utilisation and (2) the association of childhood household dysfunction with psychiatric care utilisation is mediated by school performance.
The study population was defined as all individuals born in Stockholm County, Sweden between 1987 and 1991 (n=116 088), recorded in the Medical Birth Register. This register includes data on all deliveries in Sweden since 1973. We applied certain exclusion criteria: owing to a high proportion of missing data on important variables, those who were adopted were excluded, as well as those not registered in the National School Register. To minimise problems with reverse causation, we also excluded those with a history of childhood psychopathology. Our target population comprised 96 399 individuals (figure 1).
The unique Swedish personal identity number was used to link information from several population-based registers as follows: the Causes of Death Register comprises information on all deaths of Swedish residents. The National Patient Register includes all individuals admitted to psychiatric or general hospitals. In addition, the Swedish VAL database, containing data on public healthcare services in Stockholm County, was used. The Total Enumeration Income Survey contains data on the income, and governmental benefits. The Total Population Register includes information on age, sex, place of residence and other demographic characteristics. The Longitudinal Integration Database for Health Insurance and Labor Market Studies integrates existing data from the labour market, educational and social sectors. The Register of Court Convictions contains all court convictions in Sweden for persons 15 years of age or older. The National School Register holds information on individual school performance from all public and non-public schools. Families were linked together through the Multi-Generation register, which contains all known relationships between children and parents (born in 1932 or later) since 1961.
Indicators of childhood household dysfunction
Indicators were selected on the basis of prior research demonstrating them to have significant adverse health implications.1 ,8 ,22–29 The eight indicators: familial death, parental substance abuse and psychiatric morbidity, parental somatic disease, parental criminality, parental separation/single-parent household, public assistance recipiency and residential instability, occurring between birth and age 14 years, are defined in table 1. The selection of conditions constituting the indicator parental severe somatic disease is based on earlier research.25 ,26 To assess cumulative exposure to the study indicators, the total number was summed. Each indicator was weighted equivalently in the analyses.
School performance: The GPA was based on the student’s 16 best subjects in the final (9th) year of compulsory school (age 15–16). The child earned 10–20 points per subject, yielding a total maximum grade point of 320. The maximum number of points earned was used to create five categories: one category included youth with incomplete grades (ie, missing values on more than five subjects) and four quartile categories based on the mean maximum grade point. The mean values were 173.8 (100–190), 212.2 (195–230), 249.7 (235–265) and 292.8 (270–320), respectively.
Psychiatric care utilisation
The individuals were followed from age 18 years (2005–2011) until at most 31 December 2011 to track psychiatric care utilisation, in terms of a psychiatric diagnosis (International Classification of Diseases (ICD) 10: F00-F99), during inpatient/outpatient and/or primary care. The outcomes were dichotomised; hence, we only considered the first psychiatric contact for each type of care.
All analyses were adjusted for sex and age. Owing to known associations between immigrant status (including second generation) and mental health in the Swedish population,30 we adjusted for parental country of birth, for which the following four categories were created: both parents born in Sweden, one parent born in Sweden and the other foreign-born, one/both parents born in another Nordic country, and one/both parents foreign-born elsewhere).
Multivariate analyses were performed using Cox hazards models of time to first psychiatric contact during which a psychiatric diagnosis was assigned. Entry date was defined as the date of the 18th birthday, and exit date as the date of first psychiatric contact, date of death, date of emigration or the end of follow-up (ie, 31 December 2011). Hence, the oldest (ie, born 1987) were followed for a maximum of 7 years. The youngest (ie, born in 1991) were followed for 3 years at most. Cluster robust SEs were used in order to take potential family level clustering into account.
Among the different procedures to examine mediation, we employed the bootstrap method. This method, used to estimate the indirect effects in simple mediation models,31 is considered a powerful approach for estimating mediation and indirect effects. It has been specifically designed to draw valid and reliable conclusions in instances when the assumption of multivariate normality cannot be met.31 ,32 With the SAS PROCESS macro, provided by Preacher and Hayes,33 we used a non-parametric bootstrapping method with 5000 resamples to derive the 95% CI to test for the mediation effect.32 In addition, the Sobel test was used to measure the statistical reliability of the degree of mediation.34 We first regressed the hypothesised mediator (school performance) on the independent variable (IV), that is, cumulative number of indicators using ordinary least squares regression. Second, the dependent variable (DV), that is, psychiatric care utilisation, was regressed on the IV using logistic regression. Third, the DV was regressed on both the IV and the hypothesised mediator using logistic regression. Separate coefficients for each equation were estimated and mediation was considered to occur if the relationship between the IV and DV could be partially or totally accounted for by the hypothesised mediator and if significant indirect effects were demonstrated.35
In an attempt to take genetic factors into account, we excluded parental psychiatric inpatient care from the indicators and considered it a potential confounder.
All statistical analyses were conducted with SAS V.9.4.
Our cohort of 96 399 children was evenly distributed between the sexes (49% women and 51% men). Half of the cohort was exposed to at least one indicator of childhood household dysfunction, and 23% experienced two or more indicators (table 2).
Nearly 10% had incomplete grades (8803 of 96 399), and individuals in this group were exposed to household dysfunction to a larger extent than all other grade groups. Parental separation/single-parent household was the most common indicator; 43% of the cohort was exposed to this indicator, including 65% of the ‘incomplete grades’ group compared to only 30% of the ‘highest grade points’ (quartile 4) group. Accumulation of indicators was negatively associated with school grades, with higher rates in lower school grade quartiles.
The studied indicators of childhood household dysfunction were highly intercorrelated (see online supplementary table 1). Specifically, the following pairs of variables were highly correlated: familial death with parental somatic disease, parental substance abuse with parental criminality, and single-parent household with household public assistance.
As shown in table 3, in the total cohort, nearly 14% had at least one psychiatric contact during the follow-up period, and 10% (vs 3%) used specialised psychiatric outpatient (vs inpatient) care. Psychiatric care utilisation rates revealed that individuals experiencing childhood adversity had higher rates of all measures of psychiatric care. Moreover, children in lower grade groups utilised both inpatient and outpatient care and primary care to a larger extent than those in higher grade groups.
Table 4 presents the crude and multiadjusted HRs with 95% CIs for risk of psychiatric care utilisation by average school grade (in quartiles), and exposure to childhood household dysfunction. The crude HRs displayed a gradual increase in the risk of psychiatric care utilisation with decreasing levels of grades (ie, decreasing school performance). The highest risks were found mainly among individuals with incomplete grades, who experienced multiple indicators of childhood household dysfunction. This group had an approximate sixfold risk of inpatient care (HR 6.3, 95% CI 5.2 to 7.7), an approximate fourfold risk of specialised outpatient care (HR 3.8, 95% CI 3.3 to 4.3), and a 2.5-fold risk of primary care (HR 2.4, 95% CI 2.1 to 2.7). Adjustments for demographic variables including sex, birth year and parental country of birth had little effect on the estimates (table 4, model II). The dose–response relationship between childhood household dysfunction, school performance and psychiatric care utilisation was not as evident for primary care as for the other outcome measures. In each grade group separately, the risk of psychiatric care increased as the number of indicators increased, for example, those in the highest grade group with 3+ indicators had a twofold risk (95% CI 1.6 to 2.1) of any psychiatric care.
In the mediation analyses (see online supplementary figure 1), cumulative childhood household dysfunction (IV) predicted school performance (mediator) (β=−0.6, p<0.0001). In the second step, cumulative exposure to household dysfunction (IV) predicted psychiatric care utilisation (DV) (β=0.21, p<0.0001). In the third step, which included all three variables (IV, DV and mediator), the relationship of cumulative exposure to childhood household dysfunction (IV) with psychiatric care utilisation (DV) decreased, although it remained significant. The mediation analysis based on the bootstrapping method estimated the indirect effect of childhood household dysfunction on psychiatric care through the mediator (ie, school performance) to be 0.047 (95% CI 0.043 to 0.051), indicating the presence of a small but significant indirect effect of school performance. Testing the SE of the indirect effect (ie, the Sobel test) indicated a significant mediation effect of school performance (z=23.5, p<0.0001).
Studied separately, every indicator of childhood household dysfunction was associated with psychiatric care utilisation (see online supplementary table 2), regardless of school performance. Results revealed that, for instance, for those in the highest grade group, exposure to childhood household dysfunction such as parental substance abuse and household public assistance was associated with a higher risk of psychiatric care utilisation (HR 1.5, 95% CI 1.3 to 1.8 and 1.7, 95% CI 1.5 to 1.9, respectively).
We also performed separate analyses where we excluded all individuals with indications of parental psychiatric care. This did not have any substantial effects on the pattern of the results (data not shown).
As hypothesised, and in line with prior research,1–5 ,8 ,9 ,36 our cohort study of almost 100 000 young individuals in Stockholm, Sweden, showed that childhood household dysfunction was associated (in a dose response manner) with a markedly increased risk of psychiatric care utilisation in young adulthood. Moreover, the association of childhood household dysfunction with psychiatric care utilisation was partly mediated by school performance.
Further, the impact of childhood household dysfunction appeared to be cumulative, with the risk of psychiatric care utilisation increasing with the number of indicators incurred. Similar findings, based on different contexts, have been reported in the USA1 ,3 ,4 ,9 and in other parts of the Western world.2 ,5 ,8 ,10 Earlier studies typically used cross-sectional survey data.2 ,5 ,10 Our study expands this literature by providing longitudinal evidence that the impact of childhood household dysfunction is cumulative, as has been shown in other settings.1 ,5 ,10
Although there is extensive research investigating childhood adversity and later morbidity,1 ,3–5 ,9 research specifically examining the mediating role of school performance on the association between childhood household dysfunction and psychiatric morbidity is scarce. We found a significant negative relationship between childhood household dysfunction and school performance, that is, the higher the number of indicators, the lower the GPA. These results support earlier findings suggesting that childhood household dysfunction in combination with other adversities in the family,16 ,37 rather than single indicators in isolation, most often lead to poor school performance. One recent study from the USA examined if childhood adversity, in terms of socioeconomic disadvantage, in combination with educational attainment predicted future health.11 Findings revealed that because education had a larger impact on health than did childhood socioeconomic context, adults from disadvantaged childhoods who achieved high education levels often had better life expectancies than adults from advantaged childhoods who achieved low education levels.11 The researchers found that educational attainment mediated the association between childhood SEP and health.11 These results are in line with findings from a Swedish study in which school performance strongly mediated the association between childhood SES and suicidal behaviour.15 Even though these studies examine parental SES rather than childhood household dysfunction, it is known that low parental SES is likely to be accompanied by psychosocial adversity related to alcohol misuse and family disruption.38 Our current findings demonstrate that school performance is an important mediator through which childhood household dysfunction translates into risk for psychiatric care utilisation. Youth accumulating adversities in childhood have a higher risk of psychiatric care utilisation, and perform worse in school than those without childhood household dysfunction. Thus, the combination of cumulative household dysfunction and lower school grades seems to be particularly detrimental to later health. However, our findings also reveal that childhood household dysfunction increased the risk of psychiatric care, regardless of school performance, evidenced by the fact that, when comparing those in the highest grade group with and without household dysfunction, those exposed had a higher risk.
There are different possible explanations for the current findings. Researchers in the field of neuroscience have suggested biological explanations for the association between childhood household dysfunction and negative health outcomes.12 Cumulative exposure to childhood stress may lead to allostatic overload, causing enduring brain dysfunction, which in turn may affect health and quality of life throughout the lifespan.13 Psychological explanations suggest that childhood household dysfunction and, in particular, cumulative exposure may lead to lower self-esteem and self-worth, which in turn may increase the risk of psychiatric morbidity.3 ,6 This may also hold true for poor school performance. It has been suggested that education promotes good health in part by moderating the effects of social stressors and strengthening individuals’ social networks.39 Education may improve health because it enhances a sense of personal control encouraging and enabling a healthy lifestyle.40 Hence, improving vulnerable children's school performance may be of importance.41 Researchers have pointed out that school performance of children experiencing household dysfunction can be improved with appropriate interventions.42 Ongoing psychosocial problems could have a negative impact on both school performance and mental health. Some of the problems may be explained by individual factors, such as lower cognitive ability and the presence of behavioural problems.14 ,27 IQ has been shown to be a good predictor of school results,43 and a high IQ is protective for psychiatric morbidity.44 In the current study, the finding that school performance mediated the association of childhood household dysfunction with psychiatric care utilisation contributes to growing evidence that school performance may play an important role in the prediction of psychiatric morbidity.14 ,19 ,20 Given that household dysfunction is associated with increased levels of stress, an alternative explanation for the mediating effect of school performance may also be that household dysfunction predisposes to school dropout or lower GPA due to increased stress. This, in turn, may result in additional adversities (eg, not finding a job due to poor education) leading to depressive and anxious symptoms and possible subsequent pursuit of psychiatric care. Another possible explanation may be that childhood household dysfunction, as we know, leads to early mental health problems, which in turn may lead to worse school performance. Hence, improving vulnerable children's school performance needs to be accompanied by prevention programmes aiming to improve mental health among young adults. Although school performance may play an essential role in the relationship between childhood household dysfunction and later psychiatric morbidity, still the childhood environment in itself is of great importance for mental health later in life. Our findings suggest that even those in higher school grade groups that grow up in a dysfunctional household are at increased risk of subsequent psychiatric morbidity. Thus, from a practice perspective, detecting childhood household dysfunction early is crucial in order to determine whether children exposed to stressful environments are on a trajectory that could lead to increased risk for later psychiatric disorder.
The studied indicators examined in this study reflect social disadvantage in childhood, though they may also capture a degree of genetic susceptibility to morbidity. It is difficult to disentangle hereditary risk for mental disorder from social disadvantage in childhood. It is also well known that the presence of a family history of psychiatric morbidity is a predictor of offspring psychiatric morbidity.24 In an attempt to take genetic factors into account, we excluded parental psychiatric inpatient care from the list of indicators and considered it as a potential confounder in separate analyses, which yielded little effect on risk estimates.
Strengths and limitations
This study has several methodological strengths, including the longitudinal population-based design, use of nationwide registers with high completeness and no loss to follow-up, and large cohort size (which maximised statistical power). Other studies with similar research questions have often been retrospective and based on self-reported information, entailing risk for recall bias (eg, under-reporting of household dysfunction). 45
This study also had methodological weaknesses. We evaluated a set of indicators that did not consider timing, sequencing, persistence or severity, which have otherwise been suggested as important considerations. Important indicators of childhood adversity, including childhood abuse and neglect, peer and school-related events, were not possible to capture with our register data. Future studies may benefit from examining whether earlier timing of exposure to childhood household dysfunction is linked with prolonged exposure to adverse effects associated with household dysfunction. Earlier research has stressed the importance of taking age at exposure into account when studying the association between childhood adversities and adult health, as exposure at a specific period in the life course may be particularly damaging to the individual.46 ,47 The use of only inpatient and outpatient care events to measure psychiatric morbidity limits the sample to only those actively seeking care, though we were able to able to cover a broad spectrum of psychiatric morbidity. Finally, this is an observational study with obvious limitations regarding causality issues. We cannot provide evidence that poor school performance is a casual risk factor for mental ill health, or that improved school performance and educational attainment would reduce the prevalence of long-term mental health problems for persons growing up in adverse family environments. However, from our data, and a long row of other studies, it seems a plausible hypothesis.
The results of this study show a strong association between accumulation of childhood household dysfunction and later psychiatric care utilisation, with mediating effects of school performance. The potential role of school performance as a pathway between childhood household dysfunction and other health outcomes needs to be further investigated in future studies. In the light of prior evidence that children from adverse family backgrounds tend to demonstrate school performance below their potential, based on their cognitive capacity,27 programmes aimed at boosting school performance may prove a promising pathway for mental health prevention.
What is already known on this subject
Former studies have demonstrated an association between exposure to childhood household dysfunction and psychiatric care utilisation. Despite earlier findings, little emphasis has been placed on school performance as serving a possible link between cumulative exposure to childhood household dysfunction and the increased risk of later psychiatric morbidity.
What this study adds
Our study expands this literature in an important way by providing longitudinal evidence that the impact of childhood household dysfunction is cumulative. In addition, findings revealed that school performance mediated the relationship between childhood household dysfunction and psychiatric care utilisation.
The authors thank Lena Jörgensen for valuable help with data extraction and data management.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Contributors EB, CD and BB originated the idea. EB analysed the data and wrote the manuscript draft. All authors contributed to the interpretation of the results and to the writing of the final article.
Funding This study was supported by a grant from the Swedish Council for Working Life and Social Research (grant number 2013-2729). BV's work was supported by a grant from the Bank of Sweden Tercentenary Foundation.
Competing interests None declared.
Ethics approval Approved by the ethical committee in Stockholm, Sweden (dnrs: 2010/1185-31/1 and 2013/1118-32).
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The data in this study cannot be made available in the manuscript, the supplemental files, or a public repository. According to the Swedish Ethical Review Act, the Personal Data Act and the Administrative Procedure Act, data can only be made available after a legal review for researchers who meet the criteria for access to this type of sensitive and confidential data. For questions about this, please contact Christina Dalman responsible for the data set.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.