Photo: iStock

JAMANetwork.com

Key Points

Question: Is depression during youth associated with higher risks of subsequent morbidity and mortality?

Findings:  In this Swedish population-based cohort study of nearly 1.5 million individuals, 2.5% were diagnosed with depression between ages 5 and 19 years, and there were increased relative and absolute risks of being diagnosed with a wide range of medical conditions and of early death among individuals with prior youth depression compared with the general population. When adjusted for psychiatric comorbidity, particularly for substance use disorders and anxiety disorders, relative risks decreased but persisted.

Meaning: This study presents evidence suggesting an association between youth depression, medical conditions across numerous diagnostic domains, and mortality, providing a foundation for future research.

Importance: Early-onset depression has been linked to poor health outcomes. However, it is unclear the extent to which this disorder is associated with specific diseases and premature death and whether these associations remain after controlling for psychiatric comorbidity.

Objective: To quantify the association of youth depression with subsequent diagnoses of numerous somatic diseases and mortality.

Design, Setting, and Participants: A population-based cohort study was conducted using Swedish national registers containing data on all individuals born in Sweden between 1982 and 1996. A total of 1 487 964 participants were followed up from age 5 years through 2013 if no censoring occurred. Data analysis was performed from January 15, 2019, to August 10, 2020.

Exposures: Youth depression was defined as having received at least 1 diagnosis of depression from inpatient or outpatient care between ages 5 and 19 years.

Main Outcomes and Measures: This study examined 69 somatic conditions diagnosed after youth depression, as well as all-cause and cause-specific mortalities. Overall and sex-specific hazard ratios (HRs), together with 95% CIs, were estimated using Cox proportional hazards regression with attained age as underlying timescale and time-varying exposure, and adjusted for birth year and sex. All analyses were repeated controlling for psychiatric comorbidities. Absolute risk differences were calculated using standardization with Cox proportional hazards regression.

Results: Of 1 487 964 individuals included in the analysis, 51.2% were male. A total of 37 185 patients (2.5%; 67.4% female) had an inpatient or outpatient contact for depression between ages 5 and 19 years (mean [SD] age at first recorded diagnosis of depression, 16.7 [2.1] years for males and 16.7 [1.8] years for females). Age at the end of follow-up ranged between 17 and 31 years. Individuals with youth depression had higher relative risks for 66 of the 69 somatic diagnoses. Strong associations were observed for certain injuries, especially self-harm in females (HR, 14.4; 95% CI, 13.8-15.1), sleep disorders (HR, 8.1; 95% CI, 7.6-8.7), viral hepatitis (HR, 6.1; 95% CI, 5.4-6.8), all-cause mortality (HR, 5.9; 95% CI, 5.3-6.6), and cause-specific mortalities, especially death by intentional self-harm (HR, 14.6; 95% CI, 12.6-16.9). Most associations were attenuated but persisted after adjusting for psychiatric comorbidity. The absolute risk difference of a specific disease within 12 years from the first diagnosis of depression during youth ranged from −0.2% (95% CI, −1.0% to 0.6%) for arthropathies among males to 23.9% (95% CI, 22.7%-25.0%) for the broader category of injuries among females.

Conclusions and Relevance: In this Swedish population cohort study, patients with depression diagnosed during their youth appeared to have increased risks for many somatic diseases as well as for mortality, even after controlling for other psychiatric disorders. These findings suggest that several medical conditions should be considered when investigating youth depression.

Depressive disorders are common and often recurrent psychiatric illnesses, with a lifetime prevalence of approximately 16% and a wide range of ages of onset.¹ In youths, the prevalence is less than 1% for those aged 8 years and younger, 2.8% for children younger than 13 years, and 5.6% for adolescents aged 14 to 18 years.² Youth depression has been associated with increased risks of adverse health outcomes, such as substance misuse,³ atherosclerosis, and early cardiovascular disease,⁴ and premature death.⁵ These findings suggest the need for additional investigation regarding other somatic conditions following diagnosis of depression at an early age, as well as mortality risks among young patients.

To date, most large observational studies have focused on adult populations.⁶ The extent to which depression diagnosed at a young age differs from adult-onset depression in terms of somatic morbidities, sex-specific patterns, and mortality has rarely been explored.⁷ Furthermore, previous research has investigated the morbidity between depression and medical conditions by examining depressive disorders that develop after diagnosis of another disease.^8-10 In many cases the development of depressive symptoms would arguably result from the patient’s impaired quality of life due to another medical condition.¹¹ Therefore, exploring the occurrence of medical conditions following depression could provide insights into morbidity patterns and possible prevention opportunities for some medical outcomes through early interventions. Another limitation of previous research is the lack of investigation into psychiatric comorbidities, such as substance misuse and anxiety disorders. These psychiatric conditions are associated with depression, as well as with several medical conditions and higher mortality risks.¹² Investigating the role of these psychiatric comorbidities in the association between depression and medical outcomes could facilitate understanding of their specificity and guide more personalized risk assessments. This study aimed to describe the association of youth depression with a wide range of subsequent diagnoses of somatic diseases as well as premature death and explore the potential role of psychiatric comorbidities.

We conducted a cohort study within the total Swedish population using several national registers. The cohort selection process is shown in Figure 1. From the Medical Birth Register,¹³ we identified all individuals born in Sweden between 1982 and 1996 (N = 1 593 009) with follow-up through 2013. We excluded individuals with severe congenital malformations, stillbirths, and children who died neonatally (n = 74 105). Data on clinical diagnoses were collected from the National Patient Register,¹⁴ which captures all hospitalizations since 1987 and 80% of outpatient diagnoses since 2001, coded according to the International Classification of Diseases, Eighth Revision (ICD-8) (1973-1986), ICD-9 (1987-1996), and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) (1997 to present) editions. Because pediatric depression is rarely diagnosed during the first years of life,¹⁵ we excluded 7 individuals diagnosed between ages 0 and 4 years. We obtained information on migrations and deaths from the Total Population Register¹⁶ and the Cause of Death Register¹⁷ and excluded 30 933 children who emigrated or died before the observation period began (aged 5 years). The final cohort consisted of 1 487 964 individuals who were followed up until their first diagnosis of the outcomes of interest, death, emigration, or December 31, 2013, whichever occurred first. This study was approved, necessitating no informed consent, by the Regional Ethical Review Board in Stockholm, Sweden because data are deidentified. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Depression diagnosed during childhood or adolescence, referred to hereafter as youth depression, was defined as the patient having received at least 1 diagnosis of depression between ages 5 and 19 years within inpatient or outpatient care (ICD-9 codes 296B and 311x and ICD-10 codes F32 and F33). We identified 69 somatic diseases for which there were a sufficient number of affected individuals (ie, >80 patients). These diseases were defined by grouping diagnostic codes describing similar medical conditions according to previous literature^18,19 and ICD-10 categorization. Next, we further clustered these somatic diagnoses into 14 broader groups of medical disorders. A complete list of all specific diseases, medical disorder groups, and corresponding ICD-8, ICD-9, and ICD-10 codes is presented in eTable 1 in the Supplement.

To investigate the risk of premature mortality, we collected information on date and primary cause of death, excluded all individuals with missing data (n = 20), and explored all-cause and cause-specific mortality. We categorized causes of mortality into 3 groups: (1) death by intentional self-harm, (2) death by other external causes, and (3) death by natural causes (disease states) (eTable 2 in the Supplement).

We identified various psychiatric comorbidities diagnosed during follow-up, based on ICD-8, ICD-9, and ICD-10 codes, and clustered these in 4 groups (eTable 3 in the Supplement): (1) substance use disorders (SUDs); (2) anxiety disorders; (3) earlier-onset disorders, comprising disorders likely to develop before depression and consisting of attention-deficit/hyperactivity disorder, autism spectrum disorder, and intellectual disability²⁰; and (4) later-onset disorders, defined as illnesses that tend to be diagnosed after youth depression, consisting of mania, bipolar disorder, schizoaffective disorders, and schizophrenia.

Descriptive comparisons were performed using Pearson χ² tests. We used Cox proportional hazards regression models with attained age as the underlying timescale to estimate hazard ratios (HRs) and 95% CIs for every outcome. Youth depression was modeled as a time-varying exposure, resulting in 2 separate observational records for exposed individuals. That is, individuals were assumed to be unexposed before diagnosis of depression and exposed from the first depression diagnosis. If the outcome of interest occurred before exposure to depression, that date indicated the end of the follow-up time and the observation contributed only to the unexposed time. Sample sizes and number of exposed individuals for each outcome are specified in eTable 4 in the Supplement. When estimating risks for a single specific disease or disorder group, information on all other diseases was not taken into consideration.

We obtained information on familial relationships from the Multi-Generation Register²¹ and used a robust sandwich estimator, clustered on mothers, to account for the dependence between individuals in the same family, thereby adjusting the estimates’ precision. In addition, we adjusted for sex and birth year and conducted sex-specific analyses. We first performed all analyses without considering psychiatric comorbidity. Next, we adjusted for the 4 groups of psychiatric disorders separately and then combined. P < .05 was considered statistically significant; we also used a Bonferroni-corrected significance level, determined with 2-sided, unpaired testing, of P < .0006 to account for 81 analyses.

To compute absolute risks, we matched 10 unexposed comparison individuals by sex and birth year for each child or adolescent exposed to depression diagnosis. We followed up the participants from the date of the first depression diagnosis of the exposed until the outcome, censoring, or 12 years after the depression diagnosis, whichever occurred first, and calculated the cumulative incidence as the probability of receiving the outcome diagnosis before consecutive time points, using the package stdReg.²² Absolute risk differences were estimated as the difference between exposed and unexposed groups throughout follow-up.

In addition, to investigate earlier-onset depression more specifically, we conducted a sensitivity analysis limiting the exposure to children and young adolescents (ie, aged 5-16 years) and excluding 22 308 individuals who were diagnosed with depression between ages 17 and 19 years (eTable 5 in the Supplement). Analyses were performed using Stata, version 15.1 (StataCorp LLC) and R, version 3.6.1 (R Foundation for Statistical Computing).

Descriptive statistics for the study population are given in the Table. The cohort consisted of 1 487 964 individuals (51.2% male, 48.8% female), alive and living in Sweden at their fifth birthday. If no censoring occurred, age at the end of follow-up was between 17 and 31 years. A total of 37 185 patients (2.5%; 67.4% female) had an inpatient or outpatient contact for depression between ages 5 and 19 years (mean [SD] age at first recorded diagnosis of depression, 16.7 [2.1] years for males and 16.7 [1.8] years for females). A total of 22 308 of 37 185 patients (60.0%) had their first depression onset between age 17 and 19 years. The prevalence of depression increased across advancing birth years, reflecting the later initiation of the outpatient register in the National Patient Register (from 2001), and possibly, changes in the psychiatric health care system and diagnostic practices. The prevalence of comorbid psychiatric disorders was higher among youth-depression patients compared with the rest of the population. Particularly, in the youth depression group, 7530 individuals (20.3% [56.8% females]) were diagnosed with attention-deficit/hyperactivity disorder vs 37 669 individuals (2.6%) in the nondepressed group; 7911 (21.3% [65.3% females]) were diagnosed with SUDs vs 63 687 (4.4%) in the nondepressed group; and 17 047 (45.8% [73.2% females]) were diagnosed with an anxiety disorder vs 66 240 (4.6%) in the nondepressed group.

In total, 360 patients (1.0%) with youth depression died during follow-up compared with 6254 individuals (0.4%) who died in the nondepressed group (eTable 6 in the Supplement). Intentional self-harm was the leading cause of death among the individuals with youth depression (224 [62.2%]), followed by other external causes of death (93 [25.8%]). Individuals without depression diagnoses primarily died due to other external causes (2645 [42.3%]) and natural causes (2407 [38.5%]).

Compared with the general population, patients with youth depression had significantly higher relative risks of being diagnosed with any of 66 of the 69 explored somatic diseases at any time subsequent to their first inpatient- or outpatient-recorded depression episode (Figure 2; sex-specific HRs and associated P values are available in eTable 7 in the Supplement).

The highest HRs were found for injuries, particularly in females, with an HR for intentional self-harm of 14.4 (95% CI, 13.8-15.1). Within the endocrine and metabolic disorder group, males presented higher relative risks for most diseases compared with females, such as for obesity (4.3; 95% CI, 3.7-5.0); both sexes had an increased risk for type 2 diabetes (3.8; 95% CI, 3.1-4.8). Among nervous system diseases, HRs ranged from 8.1 (95% CI, 7.6-8.7) for sleep disorders to 1.8 (95% CI, 1.7-2.0) for migraine. Compared with the general population, the youth depression group had considerable increases in the hazards of viral hepatitis (HR, 6.1; 95% CI, 5.4-6.8), hypothyroidism (HR, 4.8; 95% CI, 3.6-6.4 in males and 2.7; 95% CI, 2.4-3.0 in females), liver diseases (HR, 3.4; 95% CI, 2.7-4.4), and kidney diseases (HR, 3.2; 95% CI, 2.7-3.8). In addition, among patients with youth depression, other significant sex differences were observed: compared with males, females had elevated relative risks of some infections (ie, gastrointestinal, genitourinary, and respiratory), cough, and some skin disorders; males had higher HRs for thyroid gland disorders, other endocrine gland disorders, celiac disease, dermatitis and eczema, and connective tissue disorders.

In all analyses, all-cause mortality and cause-specific mortality relative risks were elevated in the exposed individuals, and sex-specific HRs showed no statistically significant difference (eTable 8 in the Supplement). Both males and females with youth depression had nearly 6-fold higher HRs for all-cause mortality compared with those without a clinical diagnosis (HR, 5.9; 95% CI, 5.3-6.6). Examination of cause-specific mortality indicated that youth with depression were at increased risk for death by intentional self-harm (14.6; 95% CI, 12.6-16.9), death from other external causes (3.7; 95% CI, 3.0-4.6), and death by natural causes (2.1; 95% CI, 1.6-2.9).

When adjusted for comorbid psychiatric disorders, patterns by sex for every outcome were consistent, although all estimates decreased to different extents (eFigure 1 in the Supplement). Controlling for earlier-onset disorders partially attenuated all associations, whereas later-onset disorders had almost no effect on the associations. When adjusted for SUDs, and especially for anxiety disorders, all estimates were considerably attenuated. Next, we controlled for all co-occurring psychiatric disorders in a fully adjusted model (eTable 9 in the Supplement). As shown in Figure 2, all diseases presented generally similar patterns to those noted in the psychiatric-unadjusted analysis, although all associations decreased. For example, the HR for self-harm decreased to 3.3 (95% CI, 3.1-3.4). Higher relative risks persisted for all nervous system diseases, most endocrine and metabolic disorders, hypothyroidism, liver diseases, and kidney diseases. The HR for viral hepatitis was considerably attenuated (1.2; 95% CI, 1.0-1.3). In addition, relative risks decreased but persisted for all-cause mortality as well as for cause-specific mortalities (eFigure 2 and eTable 10 in the Supplement).

Sex- and time-specific risk differences of receiving a diagnosis of each medical condition following a diagnosis of youth depression are shown in Figure 3. Cumulative incidences and risk differences for each outcome are presented in eFigures 3, 4, 5, 6, and 7 in the Supplement. The lowest absolute risk difference of being diagnosed with a somatic disease within 12 years from the first diagnosis of depression during youth was for arthropathies among males (HR, −0.2%; 95% CI, −1.0% to 0.6%), and the highest was for the broader category of injuries among females (HR, 23.9%; 95% CI, 22.7%-25.0%). Among patients with youth depression, sex differences were observed: females showed increased absolute risks for several specific diseases, especially genitourinary infections (risk difference 12 years after a depression diagnosis: HR, 14.5%; 95% CI, 13.6%-15.5%); males presented increased risk differences for all-cause mortality as well as for every cause-specific mortality.

When repeating all analyses limiting the exposure to earlier ages at onset of depression (ie, up to age 16 rather than 19 years), we found the same patterns of association for all outcomes (eTable 11 and eTable 12 in the Supplement).

To our knowledge, this was the largest and most comprehensive population-based study to explore the association between youth depression and a broad range of somatic diseases as well as premature mortality. In addition, we investigated the contribution of psychiatric comorbidities. We found that individuals receiving a diagnosis of depression during childhood or adolescence appeared to have increased relative and absolute risks of extensive later morbidity and mortality, particularly death by intentional self-harm. Sex differences were observed for many of the examined medical outcomes. In particular, compared with males, females had substantially higher relative and absolute risks for injuries, an increased risk difference for genitourinary infections, and moderately higher relative risks of gastrointestinal, genitourinary, and respiratory infections; cough; and some skin disorders. Conversely, males had comparatively elevated relative risks for obesity, thyroid gland and other endocrine gland disorders, celiac disease, connective tissue disorders, and dermatitis and eczema. Both sexes with youth depression presented increased relative risks for nervous system disorders, type 2 diabetes, viral hepatitis, kidney disease, and liver disease, among others.

These results are in line with several findings linking depression and adverse outcomes. First, we replicated associations between poor mental health and nervous system disorders.^9,23,24 In particular, we identified substantially higher risks of sleep disorders in patients with youth depression, in accordance with previous hypotheses that sleep disturbances may be both depression symptoms and contributors to depression.²⁵ Furthermore, our findings support earlier evidence for the morbidity of depression with suicidal behavior,²⁶ hypothyroidism,²⁷ type 2 diabetes,²⁸ and other endocrine disorders.²⁹ We additionally noted that youths with depression have elevated risks of premature mortality primarily attributable to suicide and other external causes of death.³⁰ The same patterns were observed when restricting the exposure to the youngest patients, providing evidence that the results were not mainly associated with patients whose depression was assessed during late adolescence (ie, between ages 17 and 19 years).

When adjusted for psychiatric comorbidity, all associations were attenuated but persisted, and consistent patterns by sex were observed. Associations between youth depression and adverse outcomes may be partially explained by earlier-onset disorders. Later-onset disorders appeared to contribute little to the associations, which may be owing to the relatively young age of the study population. Substance use disorders and anxiety disorders explained a substantial part of the observed associations of youth depression with somatic diseases and mortality. These results were in accordance with those of previous studies, supporting associations of anxiety disorders and SUDs³¹ with mortality and wide-ranging morbidity. In the fully adjusted model, the risk of viral hepatitis substantially decreased, suggesting that the observed association with youth depression in the unadjusted model may be explained by these comorbid psychiatric disorders. This result aligns with previous findings linking youth depression and viral hepatitis through risk-taking behaviors.³² Nevertheless, even after controlling for psychiatric disorders, we found that youths with depression still appeared to have elevated risks of several adverse outcomes, suggesting that these associations may only partially be explained by co-occurring psychiatric disorders.

Although previous population studies on depression mainly focused on adult populations, to our knowledge, this was the largest study exploring depression assessed at a young age and its adverse outcomes. Further investigation should examine the extent to which youth-onset depression differs from adult-onset depression in terms of comorbidities, sex-specific patterns, and mortality.

This study supports and extends the results of previous research on youth depression and adverse health outcomes, possibly providing new insights into the contribution of psychiatric comorbidities. These findings highlight the hypothesis that the morbidity of youth depression extends beyond psychiatric and neurologic disorders, increasing the burden of disease and posing quality-of-life and public health challenges. Consequently, clinical efforts to comprehensively manage both psychiatric and somatic diagnoses are needed. Our results may serve as a useful foundation for further research on the complex etiologic factors and sequelae of this mental disorder, as well as on the causal mechanisms underlying the associations found in this study. It is important for future studies to establish whether the observed increased health problems follow causally from depression, stem from shared genetic^33-35 or environmental risk factors,^36,37 or whether the associations are due to reverse causality. Furthermore, these results could guide clinical intervention, risk assessment, and prediction tools tailored to young patients with depression.

The study had several limitations. The registers used in these analyses mainly capture the more severe cases, limiting the generalizability of the findings. Furthermore, the outpatient register was introduced in 2001, leading to a possible loss of early diagnoses for the youngest individuals in our cohort. A number of factors related to clinics, physicians, and patients could have affected the time at which diagnoses were recorded. Therefore, although the documented timing of diagnoses likely approximates their ordering, the diagnostic records do not provide definitive information about when health problems occurred. Moreover, it is plausible that the increased risks among individuals with depression compared with the general population may be partly explained by detection bias; because these patients are likely to be in contact with clinicians, they may also be more likely to receive other diagnoses. In addition, because the follow-up did not extend beyond early adulthood, it was not possible to investigate later-onset psychiatric comorbidities and medical conditions.

The findings of this Swedish population-based cohort study suggest that somatic and psychiatric disorders should be considered when investigating youth depression. More research is needed to identify whether depression at a young age leads to adverse health outcomes or common causes underlie both. Discovery of disease mechanisms that may serve as intervention targets in early life should be prioritized in light of the substantial disease burden associated with youth depression diagnoses.