OBJECTIVES: Parental death during childhood has been linked to increased mortality and mental health problems in adulthood. School failure may be an important mediator in this trajectory. We investigated the association between parental death before age 15 years and school performance at age 15 to 16 years, taking into account potentially contributing factors such as family socioeconomic position (SEP) and parental substance abuse, mental health problems, and criminality.
METHODS: This was a register-based national cohort study of 772 117 subjects born in Sweden between 1973 and 1981. Linear and logistic regression models were used to analyze school performance as mean grades (scale: 1–5; SD: 0.70) and school failure (finished school with incomplete grades). Results are presented as β-coefficients and odds ratios (ORs) with 95% confidence intervals (CIs).
RESULTS: Parental death was associated with lower grades (ORs: –0.21 [95% CI: –0.23 to –0.20] and –0.17 [95% CI: –0.19 to –0.15]) for paternal and maternal deaths, respectively. Adjustment for SEP and parental psychosocial factors weakened the associations, but the results remained statistically significant. Unadjusted ORs of school failure were 2.04 (95% CI: 1.92 to 2.17) and 1.51 (95% CI: 1.35 to 1.69) for paternal and maternal deaths. In fully adjusted models, ORs were 1.40 (95% CI: 1.31 to 1.49) and 1.18 (95% CI: 1.05 to 1.32). The higher crude impact of death due to external causes (ie, accident, violence, suicide) (OR: –0.27 [90% CI: –0.28 to –0.26]), compared with natural deaths (OR: –0.16 [95% CI: –0.17 to –0.15]), was not seen after adjustment for SEP and psychosocial situation of the family.
CONCLUSIONS: Parental death during childhood was associated with lower grades and school failure. Much of the effect, especially for deaths by external causes, was associated with socially adverse childhood exposures.
- childhood socioeconomic position
- parental death
- parental mental health
- parental substance abuse
- school performance
- CI —
- confidence interval
- ICD —
- International Classification of Diseases
- OR —
- odds ratio
- SEP —
- socioeconomic position
What’s Known on This Subject:
Many children experience the death of a parent during childhood. The long-term consequences of this life event, including school performance, and the importance of the psychosocial circumstances of the home have not been well elucidated in previous studies.
What This Study Adds:
Both maternal and paternal deaths during childhood were associated with lower grades and school failure. Many of the associations (and especially for death due to external causes) were associated with socioeconomic disadvantage and psychosocial problems in the family.
The death of a parent constitutes a potentially traumatic life event that may affect the life course of a child in many different ways. Parental loss during childhood has been adversely associated with mental health problems, health risk behaviors, and mortality,1–3 and it may also affect children’s school performance. Poor school performance has been associated with adverse health behaviors and outcomes in adolescence and adulthood, including alcohol and drug abuse4,5 and mental health problems.6,7 School performance may predict educational achievement and occupational position later in life and thus the individual’s own adult socioeconomic position (SEP), an important determinant of health and longevity.8,9
Families in which a parent dies often differ from other families with regard to having a more disadvantaged background (eg, lower household income and educational attainment).10 A recent report from the Swedish National Agency for Education stressed the importance of socioeconomic background for children’s school performance,11underlining the significance of considering socioeconomic factors in studies of parental death and school performance. Other factors associated with deaths in middle age, such as substance abuse and/or mental health problems, may also be important contributing factors.12,13 Parental death can lead to placement in foster care, which may have a negative impact on school performance and educational achievement.14 The age of the child at the time of the parent’s death, and a differential impact of deaths due to external versus natural causes, is also plausible. Deaths due to external causes (accident, violence, and suicide) are often more unexpected and traumatic, and they may have a stronger effect on children’s outcomes.3
To the best of our knowledge, few studies have been conducted on parental death and school performance. In a single previous register-based study from Norway,15 family disruption, caused by divorce or paternal death, was associated with lower transition rates from lower to upper secondary school. These results were restricted to education after compulsory school. Similar results were seen in a British cohort study,16 in which paternal death in early adolescence was associated with reduced educational attainment as an adult. Recently, bereaved children have been shown to have lower educational aspirations, more difficulties at work, and fewer plans for career development.17 The majority of previous studies have focused on fathers’ death, and few studies have analyzed a potentially different impact of paternal versus maternal loss.15,16
The aim of the present study was to investigate the association between parental death during childhood (ie, before age 15 years) and school performance in the final compulsory school year at age 15 to 16 years, and the potential contribution of childhood socioeconomic and psychosocial factors. We used data from national registers on the total Swedish population to separate the effects of mothers’ and fathers’ deaths and natural and external causes of death, as well as to investigate the impact of the child’s age at the time of the parent’s death.
There is a long tradition of national registers with high-quality data on health and socioeconomic indicators in Sweden. Linkage of data from these registers is made possible by use of the unique personal identity number assigned to all Swedish residents at birth (or time of immigration), which constitutes an exceptional possibility for research based on national cohorts. For research purposes, the personal identity numbers are replaced by random reference numbers, and all data are analyzed anonymously. This particular study was approved by the ethics committee in the Stockholm region.
The study population consisted of all subjects born in Sweden between 1973 and 1981, who were alive and resident in the country during the year of their 16th birthday and for whom information on school performance was available (N = 772 117).
Subjects for whom information on mean grades was missing and who received disability benefits or retirement benefits at an early age were excluded (n = 17 479 [2.3%]) because their missing values were assumed to be caused by severe health problems and not school failure. Other subjects without information on mean grades were included in the category of school failure (n = 3028 [0.4%]).
Information on time and cause of death was collected from the Cause of Death Register. Causes of death were classified according to the International Classification of Diseases (ICD [revisions 8–10]) as natural deaths (ie, deaths caused by diseases [ICD-8 code: 0000-7969; ICD-9 code: 000-796; and ICD-10 code: A00-R99]) or deaths caused by external causes (ie, deaths caused by accident, violence, and suicide [ICD-8 code: 8000-9999; ICD-9 code: 800-999; and ICD-10 code: V01-Y98]).
Information on school performance was obtained by linkage to the National School Register and analyzed as mean grade points and school failure. The grading system in Sweden during 1988 to 1996 consisted of a scale of 1 to 5, where 5 is the highest grade. School failure was defined as having finished school with incomplete grades (ie, including individuals without grades in at least 1 subject and individuals without grades). In analyses of school failure, all individuals were included, and for analyses of mean grades, individuals with mean grades from 1 to 5 were included (N = 769 089).
Socioeconomic and Psychosocial Factors
Information on the highest socioeconomic level of any adult in the household (childhood SEP) was obtained from the National Population and Housing Censuses in 1985 and 1990. This construct consists of a combination of educational level, income, and occupational position18 and was analyzed in 6 categories; unskilled and skilled worker; lower, middle, and higher levels nonmanual; and unclassified (including unemployed, farmers, and self-employed). Parents’ highest educational level was retrieved from the educational register (measured when the child was 16 years old) and analyzed in 5 categories; compulsory school ≤9 years, upper secondary school 2 or 3 years, and university <3 years and ≥3 years. An indicator of whether the child had ever been in societal care, based on records in the Register of Children and Young Persons Subjected to Child Welfare Measures from 1973 to 2008, was also included.
Indicators of parental psychosocial problems were defined as at least 1 recorded case before the child’s 15th birthday (during 1973–2008). Substance abuse was defined by using hospital admissions with an ICD diagnosis indicating alcohol or illicit drug use, and mental health problems were defined by using hospitalizations with a diagnosis indicating psychiatric disorder and/or self-inflicted injuries, according to the Patient Discharge Register. Parental criminality was defined as having been convicted of a crime that resulted in a sentence of probation, prison, or psychiatric care, according to the Register of Court Convictions.
Linear regression models were used to analyze mean grade points as outcome, and logistic regression was used to analyze school failure, with calculation of 95% confidence intervals (CIs) for both outcomes. The results are presented as β-coefficients and odds ratios (ORs). The effect of losing 1 (either) parent or both parents was analyzed, as well as the impact of fathers’ and mothers’ deaths (analyzed separately). The effect of natural deaths and deaths caused by external causes was analyzed separately, as was the variation of the effect according to the age of the child at the time of the parent’s death. All analyses were performed separately for boys and girls.
Analyses were adjusted for year of birth (included because of the slight increase in mean grades over the years), gender, SEP of the childhood household, and indicators of psychosocial problems in the family. Additional adjustments for parental country of birth, geographic residency, and social welfare (a dichotomous indicator of whether the household had ever received social welfare) did not affect the results and were not included in the final models (results available on request). We also calculated Cohen’s d (equal to the mean difference divided by the SD) to estimate the strength of the effect (0.2 is considered small, 0.5 medium, and 0.8 large).19
All analyses were conducted by using SAS version 9.3 (SAS Institute, Inc, Cary, North Carolina).
Characteristics of the Study Population
A total of 22 420 subjects (2.9%) experienced the loss of a parent during childhood. For paternal death, 43.1% were deaths caused by external factors, and the corresponding number for maternal deaths was 27.1%. Mean age of the child at the time of death was 9.1 years for fathers’ death and 9.6 years for mothers’ death.
The SEP of the childhood household was lower among children who lost a parent compared with the group of children with no parental loss (Table 1). Among children who lost either parent, only 26.2% belonged to the 2 highest SEP categories (middle and higher level nonmanual) compared with 36.3% in the reference group. There were also large differences with regard to indicators of parental psychosocial problems.
Parental Death and School Performance
The mean grade point for boys was 3.08, and for girls, it was 3.36 (SD: 0.70 in the entire study population). However, only minimal differences were found between girls and boys in impact of parental death on school performance. Lower mean grades and higher proportions of school failure were seen among children who lost a parent, compared with children who did not lose a parent, during childhood (Table 2). The proportion with school failure among children who had experienced the death of either parent was 6.5% versus 3.5% in the group of children with no parental loss.
Parental death was associated with lower mean grades (ORs: –0.21 [95% confidence interval (CI): –0.23 to –0.20] for paternal death and –0.17 [95% CI: –0.18 to –0.15] for maternal death). The unadjusted effect size of parental death (mother or father) on mean school grades expressed as Cohen’s d was 0.30. The associations became weaker after inclusion in the models of household SEP and indicators of parental psychosocial problems (Table 3). For children who lost both parents, there was no association with school performance after adjustment for parental alcohol and drug abuse, mental health problems, and criminality. Adjustments where indicators of mothers’ and fathers’ psychosocial problems were included separately in the models did not change the results (results available on request).
Household SEP was defined as the highest socioeconomic level of any adult in the household, which might imply a slight overadjustment in analyses of the effect of parental death. Thus, in additional analyses, we adjusted for the surviving parent’s highest educational level instead of household SEP. In analyses of paternal death, the result from the linear regression adjusted for mothers’ highest education level was –0.11 (95% CI: –0.12 to –0.10), and the result from analyses of maternal death adjusted for fathers’ highest education level was –0.10 (95% CI: –0.11 to –0.08).
Because there was nesting within families and confounding by birth order, we performed additional analyses including only the first-born singleton child in each family (n = 561 216). This restriction did not change the results; the results from fully adjusted linear regression models in this restricted population were –0.09 (95% CI: –0.11 to –0.08) for paternal deaths and –0.10 (95% CI: –0.12 to –0.08) for maternal deaths.
Parental death was associated with higher ORs of finishing primary school with incomplete grades (ie, school failure), as shown in Table 4. In fully adjusted models, the OR for school failure was 1.34 (95% CI: 1.26 to 1.42) among children who lost either 1 of their parents. Also in these analyses, the impact of losing both parents was greatly attenuated when the analysis was adjusted for childhood SEP and parental psychosocial problems.
Cause of Death and Age of the Child
Mean grades were lower (2.95 [95% CI: 2.94 to 2.97] vs 3.06 [95% CI: 3.05 to 3.07]) and the proportion of school failure was higher (7.3% vs 5.9%) among children who lost a parent due to external causes compared with children who lost a parent due to natural causes (Table 2). The unadjusted effect of deaths from external causes was clearly higher than that from natural causes. However, this difference was not seen in the fully adjusted models (Table 5). There were no differences in school performance in children who lost a parent depending on age of the child at the time of the parent’s death.
Children in Societal Care
Among children who experienced parental death (either father or mother), 10.6% (compared with 1.6% among children without parental loss) had been placed in societal care at least once before the year of their 15th birthday. Analyses in which children who had been in societal care were excluded (n = 14 640 [1.9%]) produced results similar to the main analyses. After adjustment for age, gender, family SEP, and parental psychosocial variables, the results from linear regression models of mean grade points, after exclusion of these subjects, were –0.09 (95% CI: –0.10 to –0.08) for paternal deaths and –0.09 (95% CI: –0.10 to –0.07) for maternal deaths.
In this national cohort study, we used register data to analyze the association between parental deaths during childhood and school performance at age 15 to 16 years. Both maternal and paternal deaths were associated with lower mean grades and school failure. The estimated effect size for losing either parent was, however, relatively small according to Cohen’s d.19 Much of this effect was associated with lower SEP of the childhood household and psychosocial problems in the family (ie, parental substance abuse, psychiatric disorders, severe criminality).
Similar to previous studies associating school performances with socioeconomic factors,20,21 we found that some of the association between parental death and school performance could be explained by childhood SEP. In addition, the slightly higher impact of fathers’ deaths compared with mothers’ deaths seen in unadjusted models was not present after adjustment for household SEP.
We also adjusted the analyses for parental substance abuse, mental health problems, and criminality as indicators of adverse childhood psychosocial exposure, and compared the effects of parental death from external causes, which are often associated with these risk factors, with death from other causes. Our findings suggest that much of the lower school performance in bereaved children is related to family background characteristics, rather than the loss itself. This finding is congruent with the previously suggested hypothesis that much of the association between growing up in bereaved families and negative long-term outcomes are caused by family characteristics that existed before the death.1,22,23 These factors are also important predictors of parental death, but there seems, however, to be a small true effect of parental death on school performance, which is independent of the cause of death when these risk factors have been accounted for.
We also found that children who lost a parent were more likely to have been placed in societal care. A previous Swedish register-based study reported poorer school performance for children in foster care, compared with the majority population.14 Nonetheless, in the present study, we found similar results in relation to school performance after excluding children who had ever been placed in societal care.
There are few previous studies in which parental death was investigated in relation to school performance in a large national cohort. The strengths of the present study include the large sample size (covering the entire Swedish population born in 1973–1981), the longitudinal design, and high-quality register data on exposure, outcome, and multiple potentially confounding factors in the family environment.24
In Swedish schools during this time period, grades were not given before ages 15 to 16 years. A limitation of the study therefore is that we have no information on school performance before the death of the parent. Subjects with missing information on mean grades were more likely to having lost a parent (4.6% vs 2.9%). However, this group constituted a small proportion (2.2%) of the study population and would probably have had a marginal effect on the results. Furthermore, the same results were seen in additional analyses in which subjects with missing information on school failure (0.4%) were excluded.
Our indicators of parental psychosocial problems (defined as at least 1 recorded case before the child’s 15th birthday) were crude. However, hospital admissions with these problems usually indicate more chronic problems that have influenced the family environment for an extended time. Psychosocial problems of this kind may also be a consequence of the event of death in the family (with an increased risk of psychopathology in the surviving parent) and may thus be considered as mediators, rather than confounders, in the association between parental death and subsequent school performance. The situation in which the psychosocial and socioeconomic factors reflect the situation after the death of the parent may result in an underestimation of the results. However, no differences were seen in additional analyses in which we adjusted for the highest educational level of the surviving parent instead of household SEP, indicating that a potential underestimation of the associations is limited.
Our findings show lower mean grades and a higher proportion of school failure at age 15 to 16 years in children who had experienced maternal or paternal death during childhood compared with children who did not. Parental death was associated with psychosocial problems in the family and socioeconomic adversity, circumstances that seemed to explain much of the negative influence of parental death on school performance. Our findings indicate that special attention should be given in schools to bereaved children to prevent decline in school performance, and that interventions from health services to support bereaved children should be designed so that not only psychological needs, but also social adversity and psychosocial problems in the family environment, are adequately addressed.
- Accepted January 9, 2014.
- Address correspondence to Anders Hjern, MD, PhD, Centre for Health Equity Studies, Stockholm University, 106 91 Stockholm, Sweden. E-mail:
Dr Berg conceptualized and designed the study, conducted the analyses, interpreted the results, drafted the initial manuscript, and reviewed and revised the manuscript; Drs Rostila and Saarela interpreted the results and critically reviewed the manuscript; Dr Hjern conceptualized and designed the study, coordinated register linkage and created the data set for the study, interpreted the results, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: This study was supported by the Swedish Research Council for Health, Working Life and Welfare (grant 2009-1669).
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Jablonska B,
- Lindberg L,
- Lindblad F,
- Rasmussen F,
- Ostberg V,
- Hjern A
- Björkenstam C,
- Weitoft GR,
- Hjern A,
- Nordström P,
- Hallqvist J,
- Ljung R
- Erikson R
- Marmot M
- Fauth B,
- Thompson M,
- Penny A
- Swedish National Agency for Education
- Wahlbeck K,
- Westman J,
- Nordentoft M,
- Gissler M,
- Laursen TM
- ↵Statistics Sweden. Socioekonomisk indelning (SEI) [Socioeconomic Index (SEI)]. Stockholm, Sweden: Statistiska Centralbyrån (Meddelanden i samordningsfrågor, MIS); 1982:4
- Cohen J
- Jackson M,
- Erikson R,
- Goldthorpe J,
- Yaish M
- Parsons S
- Copyright © 2014 by the American Academy of Pediatrics