One versus two biological parents with mental disorders: Relationship to educational attainment in the next generation

Background Both maternal and, separately, paternal mental illness are associated with diminished academic attainment among children. However, the differential impacts of diagnostic type and degree of parental burden (e.g. one v. both parents affected) on these functional outcomes are unknown. Methods Using the Swedish national patient (NPR) and multi-generation (MGR) registers, 2 226 451 children (1 290 157 parental pairs), born 1 January 1973–31 December 1997, were followed through 31 December 2013. Diagnostic status of all cohort members was defined for eleven psychiatric disorders, and families classed by exposure: (1) parents affected with any disorder, (2) parents affected with a disorder group (e.g. neuropsychiatric disorders), and (3) parents affected with a specific disorder (e.g. ADHD). Pairs were further defined as ‘unaffected,’ ‘single-affected,’, or ‘dual-affected.’ Among offspring, the study evaluated fulfillment of four academic milestones, from compulsory (primary) school through University (college). Sensitivity analyses considered the impact of child's own mental health, as well as parental education, on main effects. Results Marked reductions in the odds of achievement were observed, emerging at the earliest levels of schooling for both single-affected [adjusted odds ratio (aOR), 0.50; 95% CI 0.49–0.51] and dual-affected (aOR 0.29, 95% CI 0.28–0.30) pairs and persisting thereafter [aOR range (single), 0.52–0.65; aOR range (dual), 0.30–0.40]. This pattern was repeated for analyses within diagnosis/diagnostic group. Main results were robust to adjustment for offspring mental health and parent education level. Conclusions Parental mental illness is associated with profound reductions in educational attainment in the subsequent generation, with children from dual-affected families at uniquely high risk.

energy program; vehicle and transport program; business and administration program, etc.) and the other 6 programs refer to academic program (e.g., business management and economics; arts; humanities; natural science; social science; technology) (The Swedish National Agency for Education, 2013).
Each program consists of several courses which provide a certain number of points. All programs include 8 core subjects, such as Swedish/Swedish as a second language, English language, mathematics, civics, religion, science studies, physical education and health, and artistic activities (and history since 2011), as well as program-specific subjects. Students are graded at the end of each course, in accordance with the national knowledge requirements laid down for each course. In addition, students are graded after completion of a diploma project. A final grade, which is awarded upon completion of the whole program, is registered in a schoolleaving certificate. The certificate denotes the status of having completed upper secondary education, which is recorded in the Educational Register and the Longitudinal Integration Database for Health Insurance and Labour Market Studies (LISA; Swedish acronym for Longitudinell Integrationsdatabas för Sjukförsäkrings-och Arbetsmarknadsstudier; held by Statistics Sweden; https://www.scb.se/) (Ludvigsson, Svedberg, Olen, Bruze, & Neovius, 2019). Since 1990, LISA annually collects educational data on each individual aged 16 years and above who is alive and registered in Sweden as of December 31st each year along with information on completeness/incompleteness and length of education (in years) at each educational level (Ludvigsson, Svedberg, Olen, Bruze, & Neovius, 2019). The prospective and uniform collection of data on education in LISA allows to follow each individual throughout his/her lifetime educational development.
In the current study, LISA was used as a source of data on post-compulsory education. The outcome called 'finishing upper secondary school' was defined by individual-level data from LISA on completeness/incompleteness of either vocational or academic program and the length of upper secondary education. For this outcome, all analyses were based on a sub-cohort of individuals born in 1973-1994 (i.e., who were aged 19 years and above by the end of follow-up on December 31 st , 2013) and who did not die or emigrated from Sweden before age 19.
Higher education Higher education is a voluntary form of post-secondary education, which is offered by universities and university colleges (also called tertiary education). Higher education in Sweden is provided at the undergraduate level, including Bachelor's degree and Master's degree education as well as at the postgraduate level, i.e. Doctoral degree education (Deen, 2007). Postgraduate level education is beyond the focus of the current study. For citizens of Sweden (or citizens of European Union or European Economic Area country, or Switzerland), a higher education in Sweden is tuition-free (Willemse &De Beer, 2012). Through the Swedish Board of Student Finance (https://www.csn.se/), students with Swedish citizenship are eligible for the state financial support for their university studies in a form of student loans or study grants that minimizes the influence of financial aspect on the decision-making process of attaining higher education. Each educational level within higher education is built on achieving a prior level. Thus, to study at bachelor level, students must have successfully completed upper secondary education or its equivalent. To enter a program at the master's level, students have a Bachelor's degree from Sweden or a foreign country, or have a corresponding qualification. Higher education institutions can vary in educational process, but it is overall organized through offering courses and programs; with specific entry requirements for each course and program. The duration and extent of programs and courses is measured by the higher education credits and one full-time academic year is equal to 60 credits. Bachelor's degree requires of at least 180 higher education credits (3 years in length) and Master's degree requires additional 60 or 120 credits (additional 1-year education results in a Magister degree [in Swedish: Magisterexamen] and 2-year education leads to a Master degree [in Swedish: Masterexamen]; both degrees are translated as a 'Degree of Master' in English). At each level, the students can earn a general qualification, an artistic qualification, or a professional qualification. The latter qualification leads to a specific profession and implies that students must have completed an occupation-specific program in order to be licensed to work in such profession (for example: a Degree of Bachelor of Science in Social Work, a Degree of Master of Science in Medicine, etc.). As indicated above, annually updated information on educational status of each Swedish resident (aged 16+) is registered in LISA (Ludvigsson, Svedberg, Olen, Bruze, & Neovius, 2019). For education at post-secondary level, LISA records the current length of studies, the number of obtained higher education credits, presence or absence of a degree, type of the program taught, and whether education is undertaken at universities/university colleges or not.
In the current study, the outcome called 'starting university' was defined by information from LISA on having undertaken post-secondary education for up to 3 years in length without being awarded a degree. The outcome called 'finishing university' was denoted by LISA records of having studied at universities/university colleges for 3 years or more, i.e., having fulfilled the requirements for a degree. For the former outcome, the analyses were based on a sub-cohort of individuals born in 1973-1992 (i.e., who were aged 21 years and above by the end of follow-up on December 31 st , 2013) and who did not die or emigrated from Sweden before age 21. For the latter outcome, the sub-cohort of individuals born in 1973-1988 was used, i.e., the analyses were conducted among those who were aged 25 years and above by the end of follow-up and who did not die or emigrated from Sweden before age 25.

Summary of Education Outcome Definitions
For each educational outcome in our analyses, separate sub-cohorts of index offspring were created, capturing individuals with 1) the minimum person-time necessary to achieve the corresponding outcome (Fig. S4) and 2) similar definitions of the target outcomes. Thus, eligibility to access USS was assessed in offspring who graduated from compulsory school in 1998-2013 (n=1,416,867), with the graduation range defined due to the equivalence of eligibility criteria used in this period. For finishing USS, the sub-cohort included individuals born in 1973-1994 (i.e., aged 19+ by the end of follow-up) and who were alive and living in Sweden at age 19 (n=1,985,519). For starting university education, the sub-cohort consisted of those born 1973-1992, who were alive/in Sweden at age 21 (n=1,781,145). Finally, the sub-cohort for finishing university included those born 1973-1988, who were alive/in Sweden at age 25 years (n=1,344,192). All age cut-offs were drawn from categories defined in reporting on the Swedish educational system.

Note 3. Description of the 'uncleaned population' used for comparison
For the main analyses, the comparison pairs were selected from the general population using an 'uncleaned' technique (Gottesman et al., 2010). This applies to all analyses for exposure type 1 (see below Supplemental Fig S2A for details), exposure type 2 (see Supplemental Fig 2B), and exposure type 3 (see Supplemental Fig S3). This approach imposes no restriction on parental diagnoses and represents, for each of the exposure types, the remainder of the full study population. This method retains and reflects the high rates of comorbidity inherent to psychiatric populations and ensures real-world, conservative measures of associations. Below we present the examples for different exposure types. The parental psychiatric disorders that compose the exposure variables include 11 disorders as: attention-deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD), schizophrenia, bipolar disorder, major depressive disorder (MDD), generalized anxiety disorder (GAD), agoraphobia, social phobia, obsessive-compulsive disorder (OCD), substance use disorders (SUD), and Tourette's disorder and chronic tic disorders (TD/CTD) (see below Supplemental Table S1A for the ICD-codes).
The exposure type 1 is defined as: parent(s) affected with any of 11 disorders of interest (i.e., pairs in which ≥ 1 member has a disorders of interest). These pairings included: (i) dualaffected pairs (i.e., both parents affected; pairs permitted to share or differ in their respective diagnoses), and (ii) single-affected pairs (i.e., only one parent affected) (Supplemental Fig S2A). The uncleaned comparison population for offspring exposed to having dual-affected parents includes the rest of the offspring from the sub-cohort (please note that for each specific educational outcome a separate sub-cohort is used as reported in Supplemental Fig 1A and 1B), i.e., offspring of pairs where either none of parents are affected by any of 11 disorders or only one parent is affected. The uncleaned comparison population for offspring exposed to having single-affected parent consists of offspring of parents not affected by any of 11 disorders.
The exposure type 2 is defined as: the disorders of interest are clustered into five operational groups: 1) neuropsychiatric disorders (ADHD, ASD, and TD/CTD), 2) SUD, 3) GAD and MDD, 4) agoraphobia, social phobia, and OCD, and 5) schizophrenia and bipolar disorder, and within each disorder group, exposure variables are generated-same as in exposure type 1 -to distinguish dual-affected and single-affected pairs (e.g., 'dual-affected' with neuropsychiatric disorders would indicate both members have ADHD, ASD, or TD/CTD [in any combination], while in single-affected pair only one parent has neuropsychiatric disorders [in any combination] but not the other; same for all other disorder groups) (Supplemental Fig S2B). Using the group of neuropsychiatric disorders as an example, the uncleaned comparison population for offspring exposed to having dual-affected parents includes offspring of pairs where either none of parents are affected by neuropsychiatric disorders (but parents might be dual-or single-affected by any of the remaining disorder groups), or only one parent is affected by neuropsychiatric disorders. The uncleaned comparison population for offspring exposed to having single-affected parent includes offspring from families where none of parents are affected by neuropsychiatric disorders (but parents might be dual-or single-affected by any of the remaining disorder groups). Same principals apply for each other disorder groups.
The exposure type 3 is defined as: parent(s) affected with a specific disorder of interest: exposure groups are also generated for each of 11 disorders of interest. For each specific disorder, we identified all (i) within-disorder dual-affected pairs (i.e., mother and father with the same specific disorder, e.g., with OCD), (ii) cross-disorder dual-affected pairs (mother and father with different specific disorders, e.g., mother with OCD, father with MDD), and (iii) single-affected pairs (only one parent has a specific disorder of interest, e.g., OCD, but the other parent id free from this particular disorder) (Supplemental Fig S3). Using pairs dual-affected with OCD as an example, the uncleaned comparison population for offspring exposed to having within-disorder dual-affected parents consists of offspring of pairs where either none of parents are affected by OCD (but they might be dual-or single-affected by any other disorders), or only one parent is affected by OCD. Using pairs where mother is affected with OCD and father is affected with MDD as an example, the uncleaned comparison population for offspring exposed to having cross-disorder dual-affected parents consists of offspring from any other families where such combination of maternal and paternal disorders was not identified (but mother and/or father might be affected by other disorders). For pairs where only one, but not the other parent is affected with a specific disorder (e.g., OCD), uncleaned comparison population consists of offspring where none of parents have OCD, but they might be affected (one or both) by any other disorders.   Note: White box with dotted margins denotes the sub-cohort for the analysis of compulsory educational outcome, the other white boxes represent the sub-cohorts for the analyses of post compulsory outcomes.  Note: All models are adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970) and clustered by family identification number with robust standard error estimation (sandwich estimator). 'Within-disorder affected parental pairs' imply that both parents have records of the same disorder. 'Cross-disorder affected parental pairs' imply that parents have records of different disorders out of the eleven psychiatric disorders in question. For each educational outcome, separate sub-cohorts of index offspring were created, comprising the individuals who had the time necessary to achieve the corresponding outcome. mXXX denotes maternal diagnoses and fXXX denotes paternal diagnoses.
Empty cells indicate the results of analyses with no exposed individuals or if ≤5 exposed cases or exposed non-cases were available (as a result, Tourette's disorder and chronic tic disorders are not at all reported as maternal or paternal diagnoses due to small numbers of exposed cases or non-cases). Figures in bold denote the statistically significant associations and figures in Italics denote non-significant associations.
a 'Uncleaned population' imposes no restriction on parental diagnoses in the comparison group and represents, in each relevant analysis, the rest of study population. This approach reflects the high rates of psychiatric comorbidity and ensures obtaining real-world, conservative measures of associations.
Abbreviation: ADHD, attention-deficit hyperactivity disorder; AGO, agoraphobia; ASD, autism spectrum disorder; BIP, bipolar disorder; GAD, generalized anxiety disorder; MDD, major depressive disorder; OCD, obsessivecompulsive disorder; OR, odds ratio; SCZ, schizophrenia; SOC, social phobia; SUD, substance use disorders. Note: All models are adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970) and clustered by family identification number with robust standard error estimation (sandwich estimator). mXXX denotes maternal diagnoses and fXXX denotes paternal diagnoses. 'Within-disorder affected parental pairs' imply that both parents have records of the same disorder. 'Cross-disorder affected parental pairs' imply that parents have records of different disorders out of 11 psychiatric disorders in question. For each educational outcome, separate sub-cohorts of index offspring were created, comprising the individuals who had the time necessary to achieve the corresponding outcome. Empty Note: All models are adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970) and clustered by family identification number with robust standard error estimation (sandwich estimator). mXXX denotes maternal diagnoses and fXXX denotes paternal diagnoses. 'Within-disorder affected parental pairs' imply that both parents have records of the same disorder. 'Cross-disorder affected parental pairs' imply that parents have records of different disorders out of 11 psychiatric disorders in question. For each educational outcome, separate sub-cohorts of index offspring were created, comprising the individuals who had the time necessary to achieve the corresponding outcome. Empty cells indicate the results of analyses with no exposed individuals or if ≤5 exposed cases or exposed non-cases were available (as a result, Tourette's disorder and chronic tic disorders are not at all reported as maternal or paternal diagnoses due to small numbers of exposed cases or non-cases). Figures in bold denote the statistically significant associations and figures in Italics denote non-significant associations.
a 'Uncleaned population' imposes no restriction on parental diagnoses in the comparison group and represents, in each relevant analysis, the rest of study population. This approach reflects the high rates of psychiatric comorbidity and ensures obtaining real-world, conservative measures of associations. Note: All models are adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970) and clustered by family identification number with robust standard error estimation (sandwich estimator). mXXX denotes maternal diagnoses and fXXX denotes paternal diagnoses. 'Within-disorder affected parental pairs' imply that both parents have records of the same disorder. 'Cross-disorder affected parental pairs' imply that parents have records of different disorders out of 11 psychiatric disorders in question. For each educational outcome, separate sub-cohorts of index offspring were created, comprising the individuals who had the time necessary to achieve the corresponding outcome. Empty cells indicate the results of analyses with no exposed individuals or if ≤5 exposed cases or exposed non-cases were available (as a result, Tourette's disorder and chronic tic disorders are not at all reported as maternal or paternal diagnoses due to small numbers of exposed cases or non-cases). Figures in bold denote the statistically significant associations and figures in Italics denote non-significant associations.
a 'Uncleaned population' imposes no restriction on parental diagnoses in the comparison group and represents, in each relevant analysis, the rest of study population. This approach reflects the high rates of psychiatric comorbidity and ensures obtaining real-world, conservative measures of associations.
Abbreviation: ADHD, attention-deficit hyperactivity disorder; AGO, agoraphobia; ASD, autism spectrum disorder; BIP, bipolar disorder; GAD, generalized anxiety disorder; MDD, major depressive disorder; OCD, obsessivecompulsive disorder; SCZ, schizophrenia; SOC, social phobia; SUD, substance use disorder subsection .5) ≥10 years a In the National Patient Register, the diagnoses are coded according to the Swedish version of ICD-codes in its 8 th , 9 th , or 10 th revision. b In the Prescribed Drug Register, drug classes are defined following the Anatomical Therapeutic Chemical (ATC) classification system codes. c Diagnosis defined through both ICD-code and prescription of drugs for attention-deficit/hyperactivity disorder as recorded in the Prescribed Drug Register by the corresponding ATC-codes. The validity of this definition has been previously described (Larsson et al., 2013). d The Swedish codes for Tourette's disorder and chronic tic disorders have been previously validated (Ruck et al., 2015) and a previously described algorithm has been used to minimize the inclusion of individuals with only transient tics (Brander et al., 2018;Fernandez de la Cruz et al., 2017;Mataix-Cols et al., 2015;Perez-Vigil et al., 2018). e For sensitivity analysis #2, stricter definition of parental schizophrenia and bipolar disorder have been used that required at least two discharge diagnoses of a corresponding disorder to be recorded (offspring diagnoses were left unchanged). Table S1B. Study Design: The RECORD Statement -Checklist of Items to be reported in observational studies using routinely collected health data (Benchimol et al., 2015) Item No. Page 10, Additional Information Table S2. Educational outcomes in offspring with parents single-affected by each specific disorder (one disorder at the time).

STROBE items
Adjusted OR (95% CI) for achieving each educational outcome in all index offspring and offspring free from any 11 psychiatric disorders with parents single-affected by each specific disorder (one disorder at the time), compared to individuals with parents from the general population (uncleaned population a ). Note: All models are clustered by family identification number with robust standard error estimation (sandwich estimator). For each disorder, 'single-affected parental pairs' imply that one parent (either mother or father) has a disorder in question and the other parent is free from such disorder. For each educational outcome, separate sub-cohorts of index offspring were created, comprising the individuals who had the time necessary to achieve the corresponding outcome.
a 'Uncleaned population' imposes no restriction on parental diagnoses in the comparison group and represents, in each relevant analysis, the rest of study population. This approach reflects the high rates of psychiatric comorbidity and ensures obtaining real-world, conservative measures of associations. b Adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970)  Abbreviation: ADHD, attention-deficit hyperactivity disorder; AGO, agoraphobia; ASD, autism spectrum disorder; BIP, bipolar disorder; CI, confidence interval; GAD, generalized anxiety disorder; MDD, major depressive disorder; OCD, obsessive-compulsive disorder; OR, odds ratio; SCZ, schizophrenia; SOC, social phobia; SUD, substance use disorders; TD/CTD, Tourette's disorder and chronic tic disorders; USS, upper secondary school.  Abbreviations: ADHD, attention-deficit hyperactivity disorder; AGO, agoraphobia; ASD, autism spectrum disorder; BIP, bipolar disorder; CI, confidence intervals; GAD, generalized anxiety disorder; MDD, major depressive disorder; OCD, obsessive-compulsive disorder; OR, odds ratio; SCZ, schizophrenia; SOC, social phobia; SUD, substance use disorders; TD/CTD, Tourette's disorder and chronic tic disorders; USS, upper secondary school.  Note: Columns highlighted in orange indicate main effect moved from significant to non-significant on adjustment for parental education level Abbreviations: ADHD, attention-deficit hyperactivity disorder; AGO, agoraphobia; ASD, autism spectrum disorder; BIP, bipolar disorder; CI, confidence intervals; GAD, generalized anxiety disorder; MDD, major depressive disorder; OCD, obsessive-compulsive disorder; OR, odds ratio; SCZ, schizophrenia; SOC, social phobia; SUD, substance use disorders; TD/CTD, Tourette's disorder and chronic tic disorders; USS, upper secondary school. Table S7. Sensitivity analysis #1 (comparison with cleaned population): educational outcomes in offspring of parents dual-and singleaffected by five groups of psychiatric disorders (one group at the time). OR (95% CI) for achieving compulsory and post compulsory educational outcomes, compared to individuals with parents from the general population (cleaned population a ).   Note: All models are clustered by family identification number with robust standard error estimation (sandwich estimator). Within each group, 'dual-affected parental pairs' imply that both parents have records of the disorders from the group in question (both parents diagnosed with either the same or different disorders within the group; for example, attentiondeficit hyperactivity disorder, or autism spectrum disorder, or Tourette's disorder and chronic tic disorders in the analysis of neuropsychiatric disorders). Within each group, 'singleaffected parental pairs' imply that one parent has at least one disorder from the group in question and the other parent is free from any disorders from the corresponding group. For each educational outcome, separate sub-cohorts of index offspring were created, comprising the individuals who had the time necessary to achieve the corresponding outcome.

Dual-affected parental pairs
a 'Cleaned population' implies that the reference group is composed of the offspring with both parents being free from any 11 psychiatric disorders. b Adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970) c Offspring who graduated from compulsory school in 1998-2013. The graduation years were chosen due to the different eligibility criteria used prior to 1998. d Neuropsychiatric disorder group includes attention-deficit hyperactivity disorder, autism spectrum disorder, and Tourette's disorder and chronic tic disorders. e Offspring born in 1973-1994 (i.e., aged ≥19 by the end of follow-up) and who was alive and living in Sweden at age 19 years. f Offspring born in 1973-1992 (i.e., aged ≥21 by the end of follow-up) and who was alive and living in Sweden at age 21 years. g Offspring born in 1973-1988 (i.e., aged ≥25 by the end of follow-up) and who was alive and living in Sweden at age 25 years.
# p-value <0.05, § p-value <0.01, * p-value <0.001 Abbreviation: ADHD, attention-deficit hyperactivity disorder; AGO, agoraphobia; ASD, autism spectrum disorder; BIP, bipolar disorder; CI, confidence intervals; GAD, generalized anxiety disorder; MDD, major depressive disorder; OCD, obsessive-compulsive disorder; OR, odds ratio; SCZ, schizophrenia; SOC, social phobia; SUD, substance use disorders; TD/CTD, Tourette's disorder and chronic tic disorders; USS, upper secondary school. ote: All models are clustered by family identification number with robust standard error estimation (sandwich estimator). 'Dual-affected pairs' imply that both parents have a record of at east one of 11 disorders (being diagnosed with either the same disorder or with different ones), and 'single-affected pairs' imply that one parent has a record of any such disorders, the other s free from any 11 disorders. For each educational outcome, separate sub-cohorts of index offspring were created, comprising the individuals who had the time necessary to achieve the orresponding outcome.
'Cleaned population' implies that the reference group is composed of the offspring with both parents being free from any 11 psychiatric disorders. Adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970) Offspring who graduated from compulsory school in 1998-2013. The graduation years were chosen due to the different eligibility criteria used prior to 1998. Offspring born in 1973Offspring born in -1994., aged ≥19 by the end of follow-up) and who was alive and living in Sweden at age 19 years.
Offspring born in 1973-1992 (i.e., aged ≥21 by the end of follow-up) and who was alive and living in Sweden at age 21 years. Offspring born in 1973Offspring born in -1988., aged ≥25 by the end of follow-up) and who was alive and living in Sweden at age 25 years.
p-value <0.001 bbreviation: CI, confidence intervals; OR, odds ratio; USS, upper secondary school. Note: All models are clustered by family identification number with robust standard error estimation (sandwich estimator). For each disorder, 'single-affected parental pairs' imply that one parent (either mother or father) has a disorder in question and the other parent is free from such disorder. For each educational outcome, separate subcohorts of index offspring were created, comprising the individuals who had the time necessary to achieve the corresponding outcome. Within each sub-cohort, a total number of offspring of affected pairs varies between the disorders, while total number of offspring of unaffected pairs is the same, that is: for eligibility to access USS -1,169,427 individuals, for finishing USS -1,627,976 individuals, for starting university -1,458,407 individuals, for finishing university -1,099,309 individuals. Each disorder is analyzed as a separate entity, therefore, offspring of parents affected by disorder(s) other than the disorder in question are excluded from the corresponding analysis.
a 'Cleaned population' implies that the reference group is composed of the offspring with both parents being free from any 11 psychiatric disorders. b Adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970) c Offspring who graduated from compulsory school in 1998-2013. The graduation years were chosen due to the different eligibility criteria used prior to 1998. d Offspring born in 1973-1994 (i.e., aged ≥19 by the end of follow-up) and who was alive and living in Sweden at age 19. e Offspring born in 1973-1992 (i.e., aged ≥21 by the end of follow-up) and who was alive and living in Sweden at age 21. f Offspring born in 1973-1988 (i.e., aged ≥25 by the end of follow-up) and who was alive and living in Sweden at age 25.
* p-value <0.001 Abbreviation: ADHD, attention-deficit hyperactivity disorder; AGO, agoraphobia; ASD, autism spectrum disorder; BIP, bipolar disorder; CI, confidence intervals; GAD, generalized anxiety disorder; MDD, major depressive disorder; OCD, obsessive-compulsive disorder; OR, odds ratio; SCZ, schizophrenia; SOC, social phobia; SUD, substance use disorders; TD/CTD, Tourette's disorder and chronic tic disorders; USS, upper secondary school.  ote: All models are clustered by family identification number with robust standard error estimation (sandwich estimator). 'Dual-affected parental pairs' imply that both parents have repeated ecords of either schizophrenia (SCZ-R) or bipolar disorder (BIP-R) (e.g., SCZ-R/SCZ-r or BIP-R/BIP-R or SCZ-R/BIP-R). 'Single-affected parental pairs' imply that one parent has either CZ-R or BIP-R and the other parent is free from any of such disorders. For each educational outcome, separate sub-cohorts of index offspring were created, comprising the individuals who ad the time necessary to achieve the corresponding outcome.
'Uncleaned population' imposes no restriction on parental diagnoses in the comparison group and represents, in each relevant analysis, the rest of study population. This approach reflects he high rates of psychiatric comorbidity and ensures obtaining real-world, conservative measures of associations. Adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970) Offspring who graduated from compulsory school in 1998-2013 (n=1,416,867  Note: All models are clustered by family identification number with robust standard error estimation (sandwich estimator). For each disorder, 'single-affected parental pairs' imply that one parent (either mother or father) has a disorder in question and the other parent is free from such disorder. For each educational outcome, separate sub-cohorts of index offspring were created, comprising the individuals who had the time necessary to achieve the corresponding outcome.
a 'Uncleaned population' imposes no restriction on parental diagnoses in the comparison group and represents, in each relevant analysis, the rest of study population. This approach reflects the high rates of psychiatric comorbidity and ensures obtaining real-world, conservative measures of associations. b Adjusted for offspring year of birth (5-year categories starting from 1973), offspring sex, and maternal and paternal year of birth in decades (<1940, 1950s, 1960s, 1970s, ≥1970) c Individuals who graduated from compulsory school in 1998-2013 (n=1,416,867). The graduation years were chosen due to the different eligibility criteria used prior to 1998. d Offspring born in 1973-1994 (i.e., aged ≥19 by the end of follow-up) and who was alive and living in Sweden at age 19 years (n=1,985,519). e Offspring born in 1973-1992 (i.e., aged ≥21 by the end of follow-up) and who was alive and living in Sweden at age 21 years (n=1,781,145). f Offspring born in 1973-1988 (i.e., aged ≥25 by the end of follow-up) and who was alive and living in Sweden at age 25 years (n=1,344,192).
* p-value <0.001 Abbreviation: BIP_r, repeated diagnoses of bipolar disorder; CI, confidence intervals; OR, odds ration; SCZ_r, repeated diagnoses of schizophrenia; USS, upper secondary school. Table S13. Sensitivity Analysis #3 (if both parents neither died before 1997 nor emigrated prior to 1997and never return to Sweden): educational outcomes in offspring with parents dual-and single-affected by any of 11 psychiatric disorders. OR (95% CI) for achieving compulsory and post compulsory educational outcomes, compared to individuals with parents from the general population (uncleaned population a ).  Table S14. Sensitivity Analysis #3 (if both parents neither died before 1997 nor emigrated prior to 1997and never return to Sweden): educational outcomes in offspring with parents dual-or single affected by five groups of psychiatric disorders (one group at the time). OR (95% CI) for achieving compulsory and post compulsory educational outcomes, compared to individuals with parents from the general population (uncleaned population a ).     Table S15. Sensitivity Analysis #3 (if both parents neither died before 1997 nor emigrated prior to 1997and never return to Sweden): educational outcomes in offspring with parents single-affected by each specific disorder (one disorder at the time). OR (95% CI) for achieving compulsory and post compulsory educational outcomes, compared to individuals with parents from the general population (uncleaned population a ).

Offspring of affected pairs
Offspring of unaffected pairs