Sick Leave and Work Participation Among Parents of Children with Autism Spectrum Disorder in the Stockholm Youth Cohort: A Register Linkage Study in Stockholm, Sweden

The study population consisted of the biological families (n = 326,183) of children in the Stockholm Youth Cohort (SYC) previously described by Idring et al., a cohort including all children ages 0–17 who resided in Stockholm County at some time in the years 2001–2007 (Idring et al. 2012). The following exclusion criteria were applied to the study population: Families without any children aged 4–17 in 2006 (n = 116,986) were excluded because children under four years old are unlikely to have been diagnosed with ASD. Families where the 4–17 year old/s had not lived in Stockholm from 2001–2006 (2002–2006 for four year olds) (n = 27,705) were excluded to ensure they had adequate time be diagnosed/registered in the Stockholm health care system. Families in which the mother gave birth in 2006 or 2007 (n = 9,774) were excluded since approximately 21 % of women who were pregnant in 2006 and 2007 received pregnancy benefits, which are presented together with sickness benefits data in the sources from the Swedish Social Insurance. Families with more than six children (n = 766) were excluded because having a large family may in itself impact parental sick leave and work participation. Parents who had children with more than one partner potentially could have been represented multiple times and in order to prevent this, only the family from the first partner was kept and subsequent families were excluded (n = 11,586). Families who had a child with ASD, but that child was not in the 4–17 age range were excluded (n = 413). Families with missing data were excluded, both those who were absent from the longitudinal integration database for health insurance and labor market studies (LISA), (Statistics Sweden 2011a) (n = 4,350) and those who were present in LISA but were missing data on sick leave and/or work participation (n = 5,036). After all exclusions, the final analytical sample consisted of 149,567 families.

Biological families (mother, father and siblings) were identified via the Multigenerational Register (Statistics Sweden 2011b) using record linkage. Children with ASD were identified by Idring et al. with a multiple (four) register approach in order to maximize case identification, and covering ASD case status as of December 31, 2007. One register used was the Clinical Database for Child and Adolescent Psychiatry in Stockholm which covers in and outpatient care within Stockholm since 2001 as well as diagnoses based on DSM IV (American Psychiatric Association 1994). A second source was the outpatient care from the VAL databases, which covers public health care services since 1997 and includes diagnoses from ICD (World Health Organization (WHO) 1992) with good coverage of diagnosis from 2006. A third source was the Habilitation Register, which records the use of habilitation services in Stockholm since 1998. The Habilitation services center helps individuals with disabilities, including but not limited to ASD, and having a diagnosis is a prerequisite for receiving services. Finally, the inpatient care from the National patient register, with as good as complete coverage for psychiatric clinics from 1973 onwards and which includes diagnoses from the ICD systems, was used. The first three of these four registers are maintained by the Stockholm County Council (SLL), while the fourth register is maintained by the National Board of Health and Welfare (Ludvigsson et al. 2011). Additionally, Idring et al. identified ASD with and without ID using recorded diagnoses meeting criteria for intellectual disability in ICD 9 (317–319), ICD 10 (F70–79) and DSM IV (317–319) and using records from the Habilitation register which categorize recipients into having autism with or without ID. These ASD cases were validated by Idring et al. (2012) in a study where they examined 177 randomly selected cases. Of these 96 % were accurate with an ASD diagnosis according to the journals. Idring et al. also cross validated SYC cases with a twin study including questionnaire data (Lichtenstein et al. 2010).

Multiple logistic regressions were performed to obtain odds ratios (OR) with confidence intervals (CI) of 95 % to examine the association between sick leave, long term sick leave, not being in the labor force, low income and having a child with or without ASD. In the first model we adjusted for parental age. In the second model we introduced parental contact with psychiatric care prior to birth of first child. In the third model we added the potentially confounding covariates parental birth country and number of children aged 4–17 in the family. In the final model we adjusted for socioeconomic covariates; parental education, two parent/single parent household, and receiving social assistance. Being a parent to a child with ASD was also analyzed stratified by ASD with or without ID. After the final model, we also checked the effect of mother’s/father’s situation by adjusting for the other parent’s covariates and each outcome. SAS 9.2 (SAS Institute Inc., Cary, NC, USA) was used to perform statistical analysis.

Ethical permission for this study was obtained from the Regional Ethical Review Board in Stockholm (2007/545-31).

Descriptive results are shown in Tables 1 and 2. The proportion of parents with a child with ASD taking sick leave and not being in the labor force or earning less was higher than for parents without a child with ASD. A higher percentage of parents who took sick leave, were not in the labor force or had a low income were born outside of Sweden. For example 10.5 % of fathers born outside of Sweden were on long term sick leave versus 2.7 % of fathers born in Sweden, and 29.6 % of mothers born outside of Sweden were not in the labor force versus 10.0 % of mothers born in Sweden. Likewise, parents with less than 10 years of education, single household parents, or parents receiving social assistance were more likely to have all four outcomes. For example, 36 % of mothers with less than 10 years of education were not in the labor force versus 9.7 % of mothers who have more than 12 years education, and 44.4 % of lone fathers earned a low income versus 27.5 % of fathers living in a two parent household. Parental contact with psychiatric care prior to the birth the first child impacted the outcomes with a higher proportion of these parents being on sick leave, not being in the labor force, and earning a low income. For example 17.2 % of fathers and 14.2 % of mothers among this group were on long term sick leave.

Parents of children who have ASD were more likely to take sick leave (measured by sick leave and long term sick leave) and to have lower work participation (measured by not being in the labor force and low income) than other parents (Tables 3, 4). A significant association was seen with all outcomes for mothers and fathers of a child with ASD, with the exception of fathers taking sick leave 15–365 days in model 4. The association was stronger among mothers. For mothers, the OR for sick leave/long term sick leave if they had a child with ASD was OR 1.3 (95 % CI 1.2–1.4)/OR 2.0 (95 % CI 1.8–2.3) in the crude model, while for fathers they were OR 1.2 (95 % CI 1.1–1.4)/OR 1.5 (95 % CI 1.2–1.7). For not being in the labor force and low income the results were OR 1.6 (95 % CI 1.5–1.8) and OR 1.5 (95 % CI 1.4–1.6) for mothers, and OR 1.3 (95 % CI 1.2–1.4) and OR 1.2 (95 % CI 1.1–1.3) for fathers. When stratifying by ASD with or without ID, the association with sick leave 15–365 days was strengthened among mothers and fathers who had a child with ASD without ID and was not significant among mothers and fathers who had a child with ASD with ID. The association with long term sick leave was strengthened for mothers of children with ASD without ID and weakened, but still significant, for mothers of children with ASD with ID. Among fathers of children with ASD with ID the association with long term sick leave was strengthened and the association among fathers of children with ASD without ID was weakened. Having a child with ASD with ID strengthened the association to not being in the labor force for both mothers and fathers. Having a child with ASD without ID weakened the association with not being in the labor force, and among fathers the association was not significant in all models. Finally, when considering low income, the associations were strengthened for mothers and fathers who had a child with ASD with ID and weakened for a child with ASD without ID, but the association remained. The OR was weakened slightly when adjusting for parental contact with psychiatric care. When adjusting for country of birth and the number of children the results were slightly strengthened. In the final model, the associations with outcomes of parents with a child with ASD/ASD without ID/ASD with ID and sick leave tended to be slightly weakened. In the final model for work participation the results were similar for ASD, slightly weakened for ASD with ID and strengthened for ASD without ID. In addition to the four models mentioned above, the data was re-analyzed with adjustments for the other parent’s covariates and outcomes. These extra analyses did not change the results.

The main strength of the study is that it was based on a cohort containing the total population of children 0–17 years living in Stockholm sometime between 2001 and 2007 and that Sweden has high quality registers (Allebeck 2009) enabling us to identify parents for the study using record linkage. Additional strengths are that Sweden has universal health care which increases the likelihood that ASD cases are detected (Idring et al. 2012), and that the ASD cases in the study are validated (see “Methods”). Finally, because the study looks at sick leave that exceeds two weeks it is unlikely that sick leave goes unreported. People with flexible jobs might not report when they are sick if they can make up missed work on weekends or other times, but this would be hard to maintain for sickness lasting two weeks. Also, in order to be on sick leave a doctor’s certificate is required and a person on sick leave must continually return to the doctor to get their sick leave extended. Once deemed fit to work the person is required to return to or look for work, which means that parents in the study are not “choosing” to be on sick leave. Additionally, people on social benefits are required to do activities to help them reenter the work force which makes it less likely that parents “choose” to be receiving social benefits.

Despite the strengths, there are several limitations with our study which should be noted. Although we controlled for parental psychiatric care before birth of first child we believe residual confounding might occur if parents have psychiatric problems and have not sought psychiatric care. Psychiatric illness proceeding parenthood results in the problem of reverse causality preventing us from distinguishing between the parent’s condition and the effect of parenting a child with ASD. Another limitation is that despite universal health care cases of ASD in the Stockholm population may go undetected, especially among children with ASD without ID, which may lead to misclassification: parents with a child with an undetected ASD classified as not having a child with ASD will result in weakened associations. Additionally, selection bias might occur since we exclude 11,586 mother/fathers because one or both of them have children with another partner, keeping only the “first” family, in order to prevent parents from being counted more than once. Since the likelihood of having a child with ASD increases with age (Croen et al. 2007), and since these parents are likely to be older, we may have selected a larger number of parents who did not have a child with ASD than if we had instead chosen to keep the most recent family. However, there is no reason to believe that this group differs regarding the outcomes.

To make the comparison group representative of the Stockholm population we included parents with children with other disabilities. Some parents in the comparison group may have had children with other conditions which could have affected the four outcomes, possibly attenuating the results. It would have been interesting to compare parents who have children with other (non ASD) disabilities as well as to study comorbidities among the children with ASD (for example OCD, anxiety, and depression.). However, data for these other diagnoses were not readily available, but could perhaps be studied in the future.

Another limitation arises from the complicated nature of the outcomes. We do not know why people are sick or why they are not in the labor force. Not being employed will affect well-being in different ways depending on whether or not the person is willingly or unwillingly outside of the labor force. Also, while having an income below the 20th percentile very likely indicates part time work, we do not know how many hours parents work or if they work more hours but at a very low wage. Additionally, parents in well paid jobs may work part time but exceed the 20 %, thus for these reasons low income is a crude measure for work participation.

We controlled for a variety of possible confounders such as being born outside of Sweden, socioeconomic factors (level of education, single parent household, receiving social assistance), and parental psychiatric contact. However, there are many other factors which can contribute and are not covered in the scope of this study that might be better suited for a qualitative study.

In conclusion, this study found that being a parent of a child with ASD is associated with higher maternal and paternal sick leave and lower work participation. The study found that ASD without ID but not ASD with ID was associated with sick leave lasting 15–365 days, but that both ASD with and without ID were associated with long term sick leave, not being in the labor force and low income. The findings are of particular interest because they appear in a society that has developed many policies to support parents with children with disabilities. In such a society it might be expected that these parents have comparable levels of sick leave, being in the work force, and income. However our findings suggest that despite Swedish policies aimed at helping families of children with ASD, both with well-being and with ability to work, that these parents remain a vulnerable group for which additional support might be warranted. It can also be noted that being on sick leave, outside of the work force or earning a low income will have long reaching impact on these parents because of Sweden’s pension system which is based on an individual’s life time earnings. It is recommended that further studies be done to see what support mothers and fathers would find most beneficial and what support they are lacking.