Association of Obsessive-Compulsive Disorder With Objective Indicators of Educational Attainment: A Nationwide Register-Based Sibling Control Study

Key Points

Question  How is obsessive-compulsive disorder (OCD) associated with objective indicators of educational attainment?

Findings  This population-based cohort study included 2 115 554 individuals, of whom 15 120 were diagnosed with OCD, and found that people with OCD were significantly more likely to fail all courses in compulsory school and less likely to achieve each level of education from primary to postgraduate education. The association was greatest when OCD was first diagnosed before age 18 years.

Meaning  Obsessive-compulsive disorder, particularly when it has an early age at onset, has a pervasive and profound association with decreased achievement across all educational levels.

Importance  To our knowledge, the association of obsessive-compulsive disorder (OCD) and academic performance has not been objectively quantified.

Objective  To investigate the association of OCD with objectively measured educational outcomes in a nationwide cohort, adjusting for covariates and unmeasured factors shared between siblings.

Design, Setting, And Participants  This population-based birth cohort study included 2 115 554 individuals who were born in Sweden between January 1, 1976, and December 31, 1998, and followed up through December 31, 2013. Using the Swedish National Patient Register and previously validated International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes, we identified persons with OCD; within the cohort, we identified 726 198 families with 2 or more full siblings, and identified 11 482 families with full siblings discordant for OCD. Data analyses were conducted from October 1, 2016, to September 25, 2017.

Main Outcomes and Measures  The study evaluates the following educational milestones: eligibility to access upper secondary school after compulsory education, finishing upper secondary school, starting a university degree, finishing a university degree, and finishing postgraduate education.

Results  Of the 2 115 554 individuals in the cohort, 15 120 were diagnosed with OCD (59% females). Compared with unexposed individuals, those with OCD were significantly less likely to pass all core and additional courses at the end of compulsory school (adjusted odds ratio [aOR] range, 0.35-0.60) and to access a vocational or academic program in upper secondary education (aOR, 0.47; 95% CI, 0.45-0.50 and aOR, 0.61; 95% CI, 0.58-0.63, for vocational and academic programs, respectively). People with OCD were also less likely to finish upper secondary education (aOR, 0.43; 95% CI, 0.41-0.44), start a university degree (aOR, 0.72; 95% CI, 0.69-0.75), finish a university degree (aOR, 0.59; 95% CI, 0.56-0.62), and finish postgraduate education (aOR, 0.52; 95% CI, 0.36-0.77). The results were similar in the sibling comparison models. Individuals diagnosed with OCD before age 18 years showed worse educational attainment across all educational levels compared with those diagnosed at or after age 18 years. Exclusion of patients with comorbid neuropsychiatric disorders, psychotic, anxiety, mood, substance use, and other psychiatric disorders resulted in attenuated estimates, but patients with OCD were still impaired across all educational outcomes.

Conclusions and Relevance  Obsessive-compulsive disorder, particularly when it has an early onset, is associated with a pervasive and profound decrease in educational attainment, spanning from compulsory school to postgraduate education.

Educational achievement is not only important from an individual perspective, given its critical link to social and economic success,^1-3 but it is also relevant from a public health and societal perspective. Education is one of the strongest predictors of health⁴ and is associated with a country’s ability to increase its standard of living, compete in global markets, and promote participation in civic affairs.^5-7

The age at onset of a disorder is an important factor in predicting the course of illness and psychosocial outcomes, including education;^8,9 early-onset psychiatric disorders can lead to truncated educational attainment.^10-12 There is some evidence from cross-sectional studies that children and adolescents with psychiatric disorders are at increased risk for underachieving in school or dropping out prematurely compared with the general population.^10,13 However, little is known about the association of obsessive-compulsive disorder (OCD) with educational attainment.

Obsessive-compulsive disorder is a relatively common psychiatric disorder. It has a lifetime prevalence of 2.3%,¹⁴ which is similar in both sexes;¹⁵ the disorder tends to follow a chronic waxing and waning course.¹⁶ One-third of patients with OCD develop the disorder before age 15 years, and about 50% report onset in childhood and adolescence.¹⁷ Clinical experience suggests that OCD might be negatively associated with the person’s education, not only in childhood and adolescence, but also in early adulthood. However, this topic has rarely been empirically investigated.

A handful of studies conducted in specialist pediatric OCD clinics have found impairments in self-reported or parent-reported educational outcomes.^18-21 In addition, 2 cross-sectional studies of 3570 individuals by the National Survey of American Life found an association between OCD (n = 57) and fewer self-reported years of education.^20,21 A study using data from the 2007 Australian National Survey of Mental Health and Wellbeing retrospectively examined the association of early-onset psychiatric disorders with the early termination of secondary education in 2055 people aged 20 to 34 years.²² The results showed that boys (but not girls) with early-onset OCD (before age 16 years) had higher school dropout rates compared with their unaffected counterparts (odds ratio [OR], 4.3).

However, these studies were limited by methodological issues, including small sample sizes, recruitment mainly from specialist clinics, cross-sectional design, and retrospectively collected self-reported or parent-reported data. Furthermore, none of the previous studies could strictly adjust for important confounders, such as psychiatric comorbidity or familial factors (eg, genetic or shared environmental factors), that might influence both OCD and educational attainment. Finally, previous studies have been limited to the educational achievements in childhood and adolescence; thus, how OCD is associated with later stages of education (eg, university studies) is unknown. It is plausible that some individuals with OCD might be able to compensate for their initial difficulties, find alternative routes to achieve higher education, or return to the education system at a later stage, but this has not been explored.

There is therefore a clear gap in the literature for studies based on larger, prospectively collected samples and objectively measured educational outcomes. This population-based study aims to investigate the association of OCD with educational attainment across the person’s lifespan—from compulsory school through postgraduate education—taking into account a number of measured covariates, such as parity and parental age at the time that an individual was born. In addition, we used a sibling control design to control for unmeasured familial confounders shared by full siblings (such as genetic factors, parental psychopathology, and socioeconomic status). To ensure that the observed associations are not entirely explained by comorbid conditions, we performed sensitivity analyses in subgroups in which all individuals with comorbid conditions were excluded. We hypothesized that OCD would be associated with academic underachievement across all educational levels, particularly in individuals who experienced pediatric or adolescent onset of the disorder.

The regional ethical review board of Stockholm approved the study. The requirement for informed consent was waived because the study was based on existing registers, and the data on the included individuals were deidentified.

The data were obtained by linking the following Swedish national population-based registers through the individuals’ unique personal identification numbers (after recoding these for anonymity):²³ (1) the National School Register, which holds information on individual school performance from all municipal and independent schools from December, 31, 1988;²⁴ (2) the Longitudinal Integration Database for Health Insurance and Labor Studies, acronymized LISA under its Swedish name,²⁵ which integrates annual data on the labor market, education sector, and social sectors for all individuals living in Sweden; (3) the Swedish National Patient Register (NPR),²⁶ which covers inpatient hospital admissions since 1969 and outpatient care since 2001, with International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes in use from 1997 to the present; (4) the Multi-Generation Register,²⁷ which connects every person who was born in Sweden from 1932 to the present and who was registered at least once as living in Sweden between 1961 and the present to their parents, enabling database users to obtain a family genealogy for each participant; (5) the Migration Register,²⁸ which provides information about migration into and out of Sweden; and (6) the Cause of Death Register,²⁹ which records information on dates and causes of all deaths since 1961.

The initial study cohort consisted of 2 328 201 individuals who were born in Sweden between January 1, 1976, and December 31, 1998, and followed up until December 31, 2013. Of these, 1 195 489, or 51.3%, were male. Individuals diagnosed with organic brain disorders (ICD-10 codes F00-F09) and/or mental retardation (ICD-10 codes F70-F79) were excluded from the cohort (n = 20 221). Additionally, individuals with 2 parents born outside of Sweden and those who were missing data on the origin of their parents were excluded (n = 192 426).³⁰ The final study cohort therefore consisted of 2 115 554 individuals (of whom 1 085 445, or 51.0%, were male). For the sibling-comparison analyses, we identified a subsample of 726 198 families with at least 2 singleton full siblings (ie, siblings of either sex who shared the same biological mother and father) during the same period.

The exposure was defined as receiving a diagnosis of OCD according to ICD-10 criteria (code F42), as recorded in the NPR.³¹ Individuals with a lifetime diagnosis of OCD were considered exposed and those without the disorder were considered nonexposed.

The National School Register contains information on each student’s eligibility to access upper secondary education after successful completion of compulsory education at age 15 or 16 years in Sweden. Because of different admission requirements before 1998 and after 2010, this study retrieved information from the National School Register for a subcohort of individuals with data between these years (n = 1 234 042). Eligibility to access upper secondary education is based on the school grades in the final year of compulsory school. The admission requirements vary depending on the student’s choice for a vocational program or an academic program. Students aiming to enter a vocational program were required to pass the 3 core courses (Swedish, English, and mathematics), and students aiming to enter an academic program (which is preparation for higher education, such as a university degree) were required to pass the 3 core courses and at least 9 additional courses. The National School Register also includes information on the individual school courses, which were coded as binary variables (passed vs failed). For the purposes of this study, students are dichotomized as eligible or ineligible for access to upper secondary education.

Data on the following binary educational outcomes for the full cohort (n = 2 115 554) were retrieved from the LISA database:²⁵ finishing upper secondary school, starting a university degree, finishing a university degree, and finishing postgraduate education (ie, a master’s or doctoral degree). Each outcome was dichotomized as achieved vs not achieved.

Logistic regression models were fitted to compare exposed and nonexposed individuals on all binary outcomes (passing specific courses, eligibility to progress to secondary education, and achievement of subsequent educational levels). First, crude associations with OCD were modeled separately for each outcome. Subsequently, models were adjusted for sex, year of birth, maternal age and paternal age at the birth of the individual participant, and parity. Results were expressed as OR with 95% confidence intervals (CIs).

A fixed-effects model was implemented in the subsample of clusters of all full siblings of individuals with OCD. By design, these stratified logistic regression models control for shared familial confounders³² and, in particular, for genetic factors and unmeasured shared confounders, such as socioeconomic status or unchanging parental traits. Furthermore, we adjusted for all measured confounders listed above, which typically vary between siblings.

To examine whether individuals with a pediatric or adolescent onset of the disorder had more profound educational impairment than those with adult onset, we repeated our main analyses, stratifying exposed persons into 2 groups: individuals first diagnosed with OCD before age 18 years (n = 4296) and individuals first diagnosed at 18 years or older (n = 10 824).

Sensitivity analyses were performed in subgroups from which all individuals with comorbid conditions were excluded. These conditions were organized in 7 groups: (1) attention-deficit/hyperactivity disorder (ADHD) (ICD-10 code F90); (2) other neuropsychiatric disorders (pervasive developmental disorders, Tourette syndrome and chronic tic disorders, and learning disabilities; ICD-10 codes F84, F95, and F81); (3) schizophrenia, schizotypal, and delusional disorders (ICD-10 codes F20-F29); (4) phobic, anxiety, and reaction to severe stress and adjustment disorders (ICD-10 codes F40-F41 and F43); (5) affective disorders (ICD-10 codes F30-F39); (6) substance use disorders (ICD-10 codes F10-F19); and (7) other disorders (dissociative disorders, somatoform disorders, other neurotic disorders, and eating disorders; ICD-10 codes F44, F45, F48, F50-F59). All disorders were defined as at least 1 registered diagnosis in the NPR. These models also adjusted for sex, year of birth, maternal and paternal age at the birth of the individual participant, and parity.

All analyses were stratified by sex and were conducted using SAS, version 9.4 (SAS Institute Inc). Data analyses were conducted from October 1, 2016, to September 25, 2017.

Descriptive characteristics of the study cohort are presented in Table 1. Of the 2 115 554 individuals included, 15 120 received a lifetime diagnosis of OCD. The proportion of females in the OCD cohort (8996 of 15 120, or 59.5%) was significantly larger than in the unexposed population (1 021 113 of 2 100 434, or 48.6%). As expected, those with a diagnosis of OCD also had more psychiatric comorbidity than those without OCD (12 234 of 15 120, or 80.9%, vs 284 905 of 2 100 434, or 13.6%) (P < .001). Of the 726 198 families with at least 2 singleton children, 11 482 (1.6%) included full siblings who were discordant for OCD.

Individuals with OCD were less likely to be eligible to access a vocational or an academic program in upper secondary school compared with the general population (adjusted OR [aOR], 0.47; 95% CI, 0.45–0.50 and aOR, 0.61; 95% CI, 0.58-0.63, for vocational and academic programs, respectively). The results remained largely unchanged in the sibling comparison models (aOR, 0.53; 95% CI, 0.47-0.60 and aOR, 0.65; 95% CI, 0.58-0.71, for vocational and academic programs, respectively). This pattern was similar across both sexes (Table 2).

Analyses of the specific school courses revealed that individuals with a lifetime diagnosis of OCD were significantly less likely to pass all courses in the last year of compulsory education (aOR range, 0.35-0.60). For example, students with OCD were 43%, 40%, and 53% less likely to pass each of the core courses (Swedish language, English language, and mathematics, respectively) (Table 3). A very similar pattern of results emerged in the sibling comparison models (aOR range, 0.42-0.60) (Table 3).

Overall, female students with OCD tended to experience lower odds of success in various courses compared with males with OCD, but these differences were not statistically significant (in that their confidence intervals overlapped) in the full sibling comparisons (eTable 1 in the Supplement).

Compared with population controls, individuals with OCD were significantly less likely to achieve each of the assessed educational levels during the 22-year study period. In the adjusted models, the individuals with OCD were 57% less likely to complete upper secondary school, 28% less likely to start a university degree, 41% less likely to finish a university degree, and 48% less likely to complete postgraduate education compared with those without OCD (Table 2). The results remained similar in the sibling comparison model, although the estimates for postgraduate studies had broader confidence intervals and so were less precise. Female persons with OCD were less likely to start a university degree and also less likely to finish a university degree compared with males with OCD, as indicated by the OR’s non-overlapping CIs. However, these differences were no longer significant in the full sibling comparison.

Overall, individuals first diagnosed with OCD before age 18 years (n = 4296) had worse outcomes across all educational levels, compared with individuals first diagnosed at age 18 years or older (n = 10 824), both in the adjusted model and the sibling comparison model (Table 4).

When individuals with different groups of comorbidities were excluded from the analyses, the results still showed significantly worse educational outcomes for the OCD group compared with the general population; however, in general, the magnitude of these differences was smaller. For example, after the exclusion of individuals with ADHD, the group with OCD was still 32% to 44% less likely to be eligible for upper secondary school entry and 51% less likely to finish upper secondary school than the group without OCD (Table 5). Similar results were obtained when these analyses were stratified by sex (Table 5) or age of first OCD diagnosis (eTable 2 in the Supplement).

To our knowledge, this is the first study examining the association of OCD with prospectively and objectively measured educational outcomes at a nationwide level. The main finding was that OCD is associated with pervasive academic underachievement. The association was global rather than being limited to a particular course; patients were more likely to fail each of the core and additional courses in compulsory school and less likely to achieve each level of education, from primary school to postgraduate education. As expected, the association was greatest when OCD was diagnosed at an early age, in childhood or adolescence. For example, patients first diagnosed with OCD before age 18 years were 55% to 62% less likely to progress beyond compulsory education compared with population controls. The corresponding figures for patients diagnosed at 18 years or older were 32% to 48%. The results were not simply explained by the high rates of psychiatric comorbidity; when individuals with relevant comorbidities were excluded, the magnitude of the results was attenuated, but persons with OCD were still significantly impaired across all educational outcomes.

An interesting finding was that the impairment appeared greatest up to the end of upper secondary school (since patients with OCD were 57% less likely to complete this educational level) and was somewhat improved during university education (where persons with OCD were 28% less likely to start and 41% less likely to finish a degree). We speculate that some individuals with OCD might cope better with their symptoms as they age, perhaps because of the natural history of the disorder or their receiving evidence-based treatment that allows them to resume their education. These individuals might be able to find alternative routes to access university, such as the locally funded school system for adults who have failed to complete primary or secondary school (known as komvux in Sweden). Thus, early educational failure does not necessarily condemn individuals to lifelong educational ostracism.

The examination of the educational attainment of this uniquely large cohort of patients with OCD greatly increases our understanding of the burden and societal impact of this disorder and its relationship to productivity impairment. According to the Organization for Economic Cooperation and Development (OECD), which is composed of 35 countries including Sweden, national unemployment rates among 25- to 64-year-olds are directly proportional to educational levels; in 2015, unemployment rates were 4.8% for those with tertiary education (ie, university studies), 7.3% for those with upper secondary education, and 12.5% for those with only primary education.³³ Additionally, individuals with tertiary education earned on average 55% more than those with upper secondary education,³⁴ and the difference in salaries continues to increase with further education beyond the tertiary level.³⁵ Hence, it is likely that the profound educational underachievement observed in this study has a direct effect on the socioeconomic status of the affected individuals and, indirectly, on the wealth of society at large.

Intuitively, our results highlight the need to detect and treat OCD at an early stage to increase the chances that affected individuals fulfill their educational potential. Whether access to and receipt of evidence-based treatment for OCD is associated with a higher likelihood of progressing to higher educational levels, as has been recently suggested in ADHD,³⁶ is an important question for future research. School-based strategies aimed at reducing scholastic underachievement in this group seem to be warranted; these might include educating school staff to identify early signs of OCD or creating guidelines for schools and higher education institutes. This approach has been used in other childhood-onset disorders, such as ADHD.³⁷ These strategies may require unprecedented collaborative efforts between mental health services, education authorities, and policy makers.

Strengths of the study include the use of a large, nationally representative cohort with data collected from records kept by government agencies or other organizations, which ensured minimal risk of selection, recall, and report biases. Second, the study benefited from a long-term follow-up period, which allowed the examination of educational levels from childhood (compulsory school) to adulthood (up to postgraduate education). Third, the use of sibling comparison models allowed strict control of unmeasured confounders, such as genetic and environmental factors, that are shared by siblings. Finally, this study used ICD-10 codes for OCD, which have been shown to have excellent interrater reliability and validity in the Swedish NPR, with positive predictive values ranging from 0.91 to 0.96.³⁸

Several limitations of the study need to be considered. First, patients included in the NPR might not be representative of all people with OCD in the population, as the register was only amended to include outpatient care in 2001 (and therefore only includes inpatient admissions between 1997 and 2000). In addition, it only includes specialist care visits and not patients seen by general practitioners. Similarly, the NPR does not include patients diagnosed by professionals other than specialist physicians, which means that only a fraction of all Swedish patients with OCD could be included in our analyses.³⁹ These limitations also apply to the identification of the comorbid psychiatric conditions.

Second, we used the date of first OCD diagnosis as a proxy for the age at onset of the disorder to identify individuals with a pediatric or adolescent onset. However, the date of first diagnosis does not necessarily correspond to the age at disorder onset, because many individuals with OCD do not seek help when they first experience symptoms.^40-42 Thus, many individuals in the adult-onset group probably had undiagnosed OCD before age 18 years.⁴³ An additional limitation is that the NPR does not include measures of symptom severity, which may have a clear association with the eventual educational attainment of an individual.

Finally, we were not able to investigate other interesting outcomes, such as absenteeism from school. This would have added valuable information on the association of OCD with education.

Obsessive-compulsive disorder, particularly when it has an early age at onset, is associated with pervasive and profound decreases in educational attainment across all levels, spanning from compulsory school to postgraduate education.

Corresponding Author: Ana Pérez-Vigil, MD, Karolinska Institutet, Department of Clinical Neuroscience, Child and Adolescent Psychiatry Research Center, Gävlegatan 22 (Entré B), Floor 8, SE-11330 Stockholm, Sweden (ana.perez.vigil@ki.se).

Accepted for Publication: September 29, 2017.

Published Online: November 15, 2017. doi:10.1001/jamapsychiatry.2017.3523

Author Contributions: Dr Pérez-Vigil had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analyses.

Study concept and design: Pérez-Vigil, Fernández de la Cruz, Brander, Isomura, Jangmo, Larsson, Mataix-Cols.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Pérez-Vigil, Fernández de la Cruz, Mataix-Cols.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pérez-Vigil.

Obtained funding: Pérez-Vigil, Fernández de la Cruz, Mataix-Cols.

Administrative, technical, or material support: Larsson, Mataix-Cols.

Study supervision: Fernández de la Cruz, Mataix-Cols.

Conflict of Interest Disclosures: Dr Mataix-Cols and Dr Fernández de la Cruz receive royalties for contributing articles to UpToDate, a part of Wolters Kluwer Health. Dr Larsson has served as a speaker for Eli-Lilly and Shire and has received a research grant from Shire Pharmaceuticals (all outside the submitted work). No other disclosures are reported.

Funding/Support: This work was supported by a research grant from the International OCD Foundation awarded to Dr Fernández de la Cruz. Ms Pérez-Vigil is supported by a grant from the Alicia Koplowitz Foundation. Dr Fernández de la Cruz is supported by a Junior Researcher grant from the Swedish Research Council for Health, Working Life, and Welfare (FORTE grant 2015-00569). Mr Brander is supported by a Karolinska Institutet PhD stipend (KID-funding). Dr Rück is supported by grant K2013-61P-22168 from the Swedish Research Council.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.