Recovery and Recurrence From Major Depression in Adolescence and Adulthood

ABSTRACT

Objective

The study aimed to estimate 5‐year recurrence rates of first‐episode major depressive disorder (MDD) and assess the impact of adolescence on recurrence likelihood after the first episode, compared to adults.

Methods

A pre‐registered retrospective cohort study that utilized epidemiological data from the Stockholm MDD Cohort (1997–2018), including all individuals registered with a depression diagnosis in Region Stockholm from 2010 to 2018. This dataset combines longitudinal information from primary and secondary care, socioeconomic data, drug dispensations, psychotherapy sessions, brain stimulation treatments, and inpatient treatment. The study included 9124 individuals (1727 adolescents aged 13–17 and 7397 adults aged 18–40) who experienced their first MDD episode between 2011 and 2012, with at least three months of remission. Propensity score weighting balanced cohorts for biological sex, socioeconomic status, depression severity, psychiatric comorbidities, and treatments.

Results

The 5‐year recurrence rates were 46.1% for adolescents and 49.0% for adults. The study had over 80% power to detect a minimum absolute difference in recurrence rates of approximately 5.5 percentage points. No significant difference in recurrence likelihood (p = 0.364) or time from remission to recurrence (median 379 days for adolescents, 326 days for adults, p = 0.836) was found between groups. Findings were consistent across bootstrap replicates and sensitivity analyses with extended remission periods.

Conclusions

Approximately half of individuals with a first MDD episode experience recurrence within five years. Recurrence rates were higher than expected for adults but consistent with expectations for adolescents. The study underscores the need for relapse prevention from adolescence through adulthood and indicates a similar clinical course of MDD across age groups.

Summary.

• Summations

  • ○
    Early recurrence and relapse prevention needs: Nearly half of individuals aged 13–40 experiencing a first episode of MDD have a recurrence within five years, exceeding previous estimates for adults but aligning with adolescent data.
  • ○
    Similar recurrence rates in adolescents and adults: There is no significant difference in the likelihood of 5‐year‐recurrence between diagnoses made during adolescence or adulthood.
  • ○
    Similar timing to recurrence in adolescents and adults: Time from remission to recurrence does not differ significantly between adolescents and adults, further supporting a consistent clinical course of MDD across these age groups.
• Limitations

  • ○
    Underrepresentation of comorbidities: The study's focus on first‐episode MDD and reliance on primary care data may underreport psychiatric comorbidities. While stringent inclusion criteria enhance internal validity, they limit generalizability to populations with higher rates of psychiatric comorbidities.
  • ○
    Exclusion of non‐psychiatric medical comorbidities: The study did not account for non‐psychiatric medical conditions that may influence MDD trajectories, limiting the understanding of broader health interactions and underscoring the need for future studies to incorporate such data.
  • ○
    Generalizability to other settings: While representative of a large urban healthcare system with universal access, findings may not generalize to populations with different healthcare structures, cultural contexts, or age ranges beyond 13–40 years.

1. Introduction

Depressive disorders during adolescence represent a critical public health issue, notably as a leading cause of long‐term disability and a substantial factor in suicide mortality rates [1, 2]. The initial manifestation of major depressive disorder (MDD) predominantly occurs from mid‐adolescence to the mid‐40s [2, 3]. However, approximately 40% of individuals encounter their inaugural episode before reaching 20 years of age, with the median onset age situated in the mid‐20s (approximately 25 years, range 18–43 years) [2, 3, 4, 5]. Episodes of MDD preceding puberty are notably rare [6]. Studies have indicated discernible distinctions in the natural course between adolescent and adult patients with depression. In the STAR*D study, which involved 4041 participants with nonpsychotic major depressive disorder (MDD) and utilized a retrospective recall design to ascertain the age of MDD onset, it was found that an onset of MDD before adulthood was associated with a more recurrent course and more severe index episodes, characterized by both atypical and possible melancholic features. Additionally, an earlier onset was linked to impaired social and occupational functioning and a lower quality of life [7]. This distinction in illness progression is acknowledged in contemporary diagnostic criteria, which incorporate an age of onset specifier for depressive disorders as detailed in the DSM‐5 [8]. Nevertheless, the debate on whether unipolar depression in adolescents and adults follow the same clinical trajectory persists, fueled by potential methodological biases in previous research due to variability in illness stage (e.g., episode count, illness duration) or prognostic factors. Future research has thus been encouraged to meticulously consider the stage of the disease in participants and to adopt longitudinal methods in their study designs [9, 10, 11]. To address potential methodological biases, our study employs propensity score weighting (PSW) to minimize confounding by balancing key covariates, such as treatment interventions and comorbidities, across age groups. This approach enables more robust causal inferences, addressing gaps in understanding age‐related differences in MDD trajectories.

Recovery and recurrence rates from adolescent MDD have been previously explored in both longitudinal samples of treatment studies and longitudinal epidemiologic studies of recruited community samples. Regarding the former, the translation of clinical trial findings to real‐world settings has been brought into question. For example, an analysis indicated that, from a nationally representative sample of adolescents from the National Comorbidity Survey: Adolescent Supplement, 61.9% and 42.2% would have been ineligible for inclusion in standard pharmacological and psychotherapy trials, respectively [12]. Studying the comparative analysis of overall MDD relapse rates between adolescents and adults requires a representative sample. Epidemiologic studies of community samples may thus offer more relevant evidence. A number of longitudinal epidemiologic studies have shown substantial continuity in depression over time and especially high recurrence risk among early onset cases and comorbid cases. These studies involved recruited community samples of 261–776 youths who were prospectively followed and re‐interviewed after a number of years [13, 14, 15, 16, 17]. Drawing on such studies, leading reviews and seminars on the topic report that approximately 50% of adolescents who remit will have another episode of depression within 5 years [1, 18].

In adults, there have been varying reports of MDD recurrence rates and there is a scarcity of reports from the general population sample without clear selection bias [19]. A Dutch psychiatric epidemiological cohort study in a representative adult population of 746 remitted MDD cases reported on a cumulative recurrence rate of 4.3% at 5 years [19]. A recent psychiatric epidemiological study involving 272,944 cases who received an MDD diagnosis in national longitudinal health registers from the Nordic region reported that—over the 38‐year‐study period—31% were reported to having received a diagnosis of recurrent MDD [20]. However, generalizing from these studies to the context of comparing MDD recurrence rates in adolescence and adulthood may be confounded from unaccounted for variations in illness stage and severity, potential underreporting of the diagnosis of recurrent MDD, and the wanting to more precisely confirm that the study truly concerns the first MDD episode (as depressive counts have been shown to predict recurrence likelihood). Despite the availability of extensive data, the precise understanding of MDD recurrence rates in adults remains an area of ongoing research.

The naturalistic trajectory and recurrence patterns of MDD have yet to be thoroughly explored across age groups in population‐based studies. Importantly, limitations of real‐world data necessitate a design structured to minimize selection bias to achieve reliable estimates. This gap hinders our ability to reliably determine both recurrence rates, time to recurrence and if these data differ significantly between adolescents and adults. Achieving a more unbiased comprehension of this information is crucial for clinical management as well as discerning any support for neurobiological (and thus also clinical management) distinctions, if any, between adolescent and adult MDD.

2. Aims

This study aims to investigate the clinical course of first episode MDD and how the age at onset (adolescence vs. adulthood) affects remittance timing and recurrence likelihood within 5 years. We categorize age into adolescence (13–17 years) and adulthood (18–40‐years old) at the first recorded MDD episode. By employing a propensity score weighted model to a the comprehensive Stockholm MDD Cohort and limiting our sample to first‐episode MDD cases, we aim to mitigate selection bias confounding.

3. Methods

The present study is a population‐based observational analysis using data from the Stockholm Major Depressive Disorder Cohort (SMC), which includes data starting from January 1, 1997, for all individuals who were diagnosed with MDD in Stockholm, Sweden, between January 1, 2010 and December 31, 2018 [21]. The data analysis was performed between September 5, 2023 and February 15, 2024. This cohort encompasses all individuals diagnosed with MDD in Stockholm from 2010 to 2018. For these individuals, we have extracted data spanning from 1997 to 2018. By the year 2018, Stockholm's population was approximately 2.4 million, representing nearly a quarter of Sweden's total population. Ethical clearance for this research was obtained from the Stockholm regional ethics committee (approval number: 2018/546–31). The requirement for individual participant consent was exempted due to the pseudonymization of all data. The SMC is registered in the EU PAS Register under the identification number 25664615 [22]. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

3.1. Study Participants

We identified patients initiating their first MDD episode (ICD‐10 code: F32) between 2011 and 2012, under 41 years of age (Supporting Information S2). For the prior years 1997–2010, we excluded individuals with any prior alcohol‐related (ICD‐10: F10), substance abuse (ICD‐10: F11‐19), bipolar disorder (ICD‐10 code: F31), MDD (ICD‐10 code: F32), recurrent MDD (ICD‐10 code: F33) and persistent mood [affective] disorders (ICD‐10 code: F34) related diagnosis recorded in either outpatient or inpatient care settings. For the remaining sample with first‐episode MDD between 2011 and 2012, baseline characteristics were retrieved including sex, age at first MDD diagnosis, socioeconomic grouping; and for the period starting from the first MDD episode and ending 365 days thereafter we retrieved the number of episodes spent in inpatient care with a primary psychiatric diagnosis (ICD‐10 code: F00‐99), the total number of days spent in inpatient care under the same conditions (length of each episode summarized across all episodes), comorbid psychiatric diagnoses (ICD‐10 codes: F00‐99), dispensed psychotropic medications, as well as psychotherapy, electroconvulsive therapy (ECT) and repetitive transcranial magnetic stimulation (rTMS) usage. A full list of variables, including ATC and ICD‐10 codes, can be found in the Supporting Information. The cohort was tracked from the first MDD episode until the study end in 2018. Detailed descriptions of the study participants are available in the Supporting Information.

3.1.1. Definitions and Choice of Primary Measure: Remission and Recurrence

In this study, the temporal progression from initial MDD diagnosis (ICD‐10 code: F32) to subsequent depressive events was systematically evaluated for each patient. Episodes were defined based on the interval between these events. An interval of 90 days or less indicated the continuation of the same depressive episode, while intervals exceeding 90 days marked the conclusion of an episode, determined by the date of the last recorded event related to depression. Sensitivity analyses were conducted by expanding the observational window to 180 and 365 days, respectively, aligning with methodologies previously employed by Lundberg et al.

Depressive events were identified through registration of either:

(1) Additional MDD diagnoses across all levels of healthcare (ICD‐10 code: F32, F33).

(2) Pharmacy dispensation prescription of antidepressants (AD; ATC code: N06A), and adjunctive treatments including lithium, risperidone, olanzapine, aripiprazole, and quetiapine (dosages above 100 mg), as well as non‐pharmacological interventions like electroconvulsive therapy (ECT), repetitive transcranial magnetic stimulation (rTMS), and psychotherapy.

A new recording of an MDD diagnosis, specifically under ICD‐10 codes F32 or F33, after the closure of a previous episode, signified the onset of a recurrent episode of MDD. Importantly, subsequent to an episode's closure, neither pharmacy dispensations, brain stimulation treatments, nor psychotherapy initiations were considered indicative of the onset of a recurrent episode. These interventions were used solely to determine that an MDD event was ongoing. Thus, only a newly registered MDD diagnosis contributed to the formal recognition of an MDD episode or its recurrence.

3.2. Statistical Analysis

3.2.1. Main Analysis

3.2.1.1. Propensity Score Weighing

Central to our analysis was the application of propensity score weighing (PSW), a technique utilized to equilibrate covariates and minimize the effects of confounding by indication [23]. This method was implemented to balance covariates between adolescent and adult groups, including sex, sociodemographic status, psychiatric comorbidities, inpatient psychiatric care, pharmacological treatments, and psychotherapy usage. This approach mitigates confound in comparisons, reducing confound by disparities in treatment or illness severity—and is consistent with the established methodologies in psychiatric registry‐based studies for addressing confounding variables [24, 25, 26]. It offers a structured approach that facilitates causal inference regarding the influence of age on MDD recurrence [23].

To achieve balanced samples between the adolescent and adult MDD groups, PSW and inverse probability of treatment weighting (IPTW) was implemented and accounting for variables such as sex, sociodemographic categories (ordered), inpatient total length of stay, substance use, and various psychiatric conditions, as well as treatments like electroconvulsive therapy (ECT), psychotherapy, and pharmacotherapy (antidepressants, mood stabilizers, anxiolytics, antipsychotics, etc.) (Supporting Information S4.). The subsequent investigation focused on estimating the Average Treatment effect on the Treated (ATT) using average marginal effects analysis, specifically examining the impact of being in adolescence at the timing of the first MDD episode on the likelihood of having a recurrence within five years, compared with those whose first episode occurs in adulthood. Conforming to the quasi‐binomial distribution, the model integrated propensity score weights and robust standard errors derived from the PSW dataset. This approach aligns with recommended practices for drawing causal inferences from PSW data [27]. To explore age as a continuous variable, a sensitivity analysis was conducted incorporating age as a continuous covariate in regression models and ATT approximations. Further sensitivity analyses were conducted by adjusting the length of the time‐window for remission classification from 90 to 180 and 365 days. For detailed information about the PSW process, power calculations, the estimation of the ATT and the internal validation of the treatment effect please see Supporting Information S6.

3.2.1.2. Estimating the Impact of Adolescence on Duration From Remission to MDD Recurrence

To investigate the impact of age group (adolescence versus adulthood) on the duration from remission to recurrence among individuals experiencing their first episode of MDD, we employed Cox proportional hazards models. These models were adjusted using PSW, robust standard errors and IPTW, facilitating a comparison between adolescent and adult cohorts. Notably, our primary analysis did not incorporate covariates, adhering to recommended practices for causal inference with PSW data [27]. We confirmed the proportional hazards assumption for our main analysis through a standard test for proportional hazards, enhancing the reliability of our findings. Internal validation was achieved through bootstrap replication.

In our sensitivity analyses, which extended the remission classification window to 180 and 365 days, we observed violations of the proportional hazards assumption (p < 0.05, test for proportional hazards). To accommodate these violations, we applied additive hazards models, also adjusted using PSW, robust standard errors, and IPTW, to maintain our causal inference approach. However, modeling constraints prevented the application of internal validation via bootstrap replication to these sensitivity analyses, as well as the graphical representation of these models.

4. Results

4.1. Baseline Characteristics and Adolescent Versus Adult Comparison

In a cohort study conducted in the Stockholm region between 2011 and 2012, we identified 9124 initial episodes of major depressive disorder (MDD) among individuals aged 13–40 years, who had no prior psychiatric diagnoses, including affective disorders, alcohol use disorders, or substance use disorders. The analysis included a propensity score weighted sample of 1727 adolescents and 7397 adults. Demographic and clinical variables were comparable between the age groups, although adults more often reported receiving benzodiazepine‐related drugs and psychotherapy (21.3%/31.9% for adults versus 4.0%/12.3% for adolescents). As of December 31, 2018, among individuals classified as in remission for at least 90 days, MDD recurrence rates were 46.1% for adolescents (n = 796) and 49.0% for adults (n = 3628). With longer remission classification periods, rates for adolescents adjusted to 42.6% (180 days) and 39.8% (365 days), and for adults to 46.1% (180 days) and 42.8% (365 days). The median time to recurrence from the start of the initial MDD episode was 1180.5 days (interquartile range [IQR]: 837–1646) for adolescents and 1099 days (IQR: 805–1503) for adults (Table 1).

TABLE 1.

Demographic and clinical characteristics of adolescents versus adults with first depressive episodes: Comorbidity, treatment and recurrence patterns.

	Adolescents	Adults
n	1727	7397
Age, mean (SD)	15.6 (1.3)	27.7 (6.5)
Sex (females, %)	68.2	65.1
Sociodemographic category, (%)
Low	45.1	39.0
Middle	16.0	19.9
High	38.8	41.1
Initial remission classification at 90‐day period
MDD recurrence, (%)	46.1	49.0
Days from initial MDD diagnosis to recurrence, median (IQR)	1180.5 (837–1646)	1099.0 (805–1503)
Days from initial MDD remission to first recurrence, median (IQR)	379.0 (56–949)	326.0 (78–811)
Initial remission classification at 180‐day period
MDD Recurrence, (%)	42.6	46.1
Days from initial MDD diagnosis to first recurrence, median (IQR)	1299.0 (947–1722)	1182.0 (896–1569)
Days from initial MDD remission to first recurrence, median (IQR)	408.0 (109–917)	318.0 (94–793)
Initial remission classification at 365‐day period
MDD Recurrence, (%)	39.8	42.8
Days from initial MDD diagnosis to first recurrence, median (IQR)	1441.5 (1126–1799)	1333.0 (1058–1699)
Days from initial MDD remission to first recurrence, median (IQR)	379.5 (89–823)	281.0 (73–718.5)
Diagnoses (%)
ADHD	0.6	0.8
Alcohol and substance abuse	1.7	1.4
Anxiety	3.3	3.3
Autism‐related	0.5	0.3
Eating disorder	1.5	0.2
Severe MDD diagnosis	1.5	1.8
Psychotic MDD diagnosis	0.2	0.4
Pharmacological treatment (%)
Antidepressant	31.9	36.7
Antipsychotic	4.3	8.0
Anxiolytic	2.0	3.6
Benzodiazepine‐related	4.0	21.3
Mood stabilizer	1.5	2.7
Other treatment (%)
ECT	0.1	1.1
Psychotherapy	12.3	31.9

Note: This table provides the demographic and clinical characteristics of adolescents and adults who experienced their first recorded episode of major depressive disorder (MDD) within the study timeframe. It includes detailed comorbidity patterns, treatment interventions, and rates of MDD recurrence. The definition of MDD recurrence is based on a new MDD diagnosis following a period meeting predefined remission criteria. The data represents an aggregation of diagnostic and treatment encounters from the commencement of the initial depressive episode across a full year. Participants were meticulously selected to have no prior recorded diagnosis in the decade leading to the start of the assessment period. This rigorous selection criterion aims to mitigate confounding by prior conditions and to precisely capture the progression from a first episode MDD, which may result in comorbidity patterns that diverge from those observed in general clinical practice. Disparities in demographic and clinical features between adolescent and adult groups have been adjusted through propensity score weighting. The methodology underlying these adjustments is further discussed in the associated comprehensive analysis. The exhaustive list of conditions and treatments included under the diagnostic and treatment categories is elaborated in Supporting Information S4.

Abbreviations: ADHD: Attention deficit hyperactivity disorder; ECT: Electroconvulsive therapy; IQR: Interquartile range; MDD: Major depressive disorder; SD: Standard deviation.

4.2. Five‐Year MDD Recurrence Likelihood

The propensity score weighted data demonstrated sufficient power (> 80%) to detect an absolute difference in recurrence rates between adolescents and adults of 5.53% (Supporting Information S9). The subsequent investigation focused on estimating the Average Treatment Effect on the Treated (ATT) using average marginal effects analysis, specifically examining the impact of being in adolescence at the timing of the first MDD episode on the likelihood of having a recurrence within five years, compared with those whose first episode occurs in adulthood. Analysis of average marginal effects—accounting for the whole range of co‐variates and their interactions with age groups—revealed no significant difference in five‐year recurrence likelihood between adolescents and adults (mean ratio: 0.96, p = 0.364, 95% confidence interval [CI] [0.878, 1.05]), a finding supported by 1000 bootstrap replicates (95% CI [0.940, 1.161]). These results remained consistent when age was included as a continuous covariate (p = 0.742).

In sensitivity analyses extending the remission classification from 90 to 180 and 365 days, these results were further validated in the adolescent versus adult comparison analyses (180 days: mean ratio: 0.931, p = 0.119, 95% confidence interval [CI] [0.85, 1.02], bootstrap replicated, 95% CI [0.905, 1.122]; 365 days: mean ratio: 0.966, p = 0.481, 95% CI [0.877, 1.06], bootstrap replicated, 95% CI [0.917, 1.148]).

4.3. Impact of Age on Duration From Remission to Recurrence

Cox proportional hazards models, employing propensity score weighting (PSW), robust standard errors and inverse probability of treatment weighting (IPTW) for causal inference approximation, indicated no significant effect of age group (adolescence versus adulthood) on the duration from MDD remission to recurrence (hazard ratio [HR] = 1.012, p = 0.836, 95% CI [0.907, 1.129]) (Figure 1). The results were consistent in 1000 bootstrap replicates (95% CI [0.901, 1.128]). Sensitivity analyses—extending the remission classification from 90 to 180 and 365 days—were conducted in using additive hazards models for both groups, due to violation of the proportional hazards assumption in the cox proportional hazards models. These models employed PSW, robust standard errors and IPTW for causal inference approximation and could not evince any age‐effect for adolescence versus adulthood on the duration from remission to recurrence (p = 0.475 and p = 0.254, respectively).

FIGURE 1.

Comparative Analysis of Recurrence Duration Post‐Remission in Adolescent and Adult Onset Major Depressive Disorder. Figure legend: This table presents findings from a comprehensive analysis examining the impact of being adolescent [13–17‐year‐olds] (compared to being in adulthood [18‐40‐year‐olds]) on the time from remission to recurrence in individuals with first episode Major Depressive Disorder (MDD). Employing Cox proportional hazards models with propensity score weighting (PSW), robust standard errors, and inverse probability of treatment weighting (IPTW), our analysis sought to approximate causal inference within two distinct age groups: Adolescents and adults. The hazard ratio (HR) for the effect of adolescence on the duration from MDD remission to recurrence was 1.012 (p = 0.836, 95% CI [0.907, 1.129]), indicating no significant age‐related differences. These findings were robust across 1000 bootstrap replicates (95% CI [0.901, 1.128]). To address violations of the proportional hazards assumption, sensitivity analyses employing additive hazards models and extending remission classification from 90 to 180 and 365 days were conducted. These models, also utilizing PSW, robust standard errors, and IPTW for causal inference, further supported the absence of a significant age effect on the duration from remission to recurrence, with p‐values of 0.475 and 0.254 for the 180 and 365 day window, respectively (not illustrated).

5. Discussion

This study identified a 5‐year relapse risk after a first episode of MDD of 46.1% in adolescents and 49.0% in adults, with rates adjusting to 42.6% and 39.8% for adolescents, and 46.1% and 42.8% for adults, at 180‐day and 365‐day remission classification periods, respectively. Our methodology addresses prior research limitations by comprehensively accounting for disease stage and severity, meticulously excluding prior relevant diagnoses that could have impacted upon recurrence likelihood to ensure a fair population‐wide sample of cases with first episode MDD without subsequent exclusions. This includes a sensitivity analysis of remission and recurrence, extending beyond conventional diagnostic registrations by incorporating pharmacological and psychotherapeutic interventions. Detailed adjustments for longer remission periods were also employed, alongside an analysis treating age as a continuous covariate, demonstrating that the results are robust to variations in remission classifications and not solely dependent on the categorical age cutoff. Notably, our findings suggest recurrence rates that exceed previously reported figures in adult general populations while corroborating adolescent‐focused research estimates. This underscores the need for a reevaluation and potential intensification of relapse‐prevention and early detection strategies in clinical practice. Early interventions that continue across developmental stages may be critical for reducing recurrence risk and improving long‐term outcomes.

Employing a propensity score‐weighted analysis to minimize selection bias and facilitate causal inference approximation, this study revealed no significant difference in the 5‐year recurrence likelihood of MDD between individuals diagnosed initially during adolescence and those in adulthood. This consistency held across all conducted internal validation and sensitivity analyses. As the inaugural investigation to directly compare MDD recurrence rates between those first diagnosed in adolescence and adulthood, our findings challenge existing assumptions regarding age‐specific clinical trajectories in first episode MDD. Earlier studies, such as the STAR*D trial, have reported a more severe and recurrent trajectory for early‐onset depression [7]. However, our findings suggest that under certain conditions, recurrence risk may be similarly substantial for later‐onset cases. This challenges previous assumptions and suggests that the recurrence risk across age groups may be influenced by common underlying mechanisms, such as genetic predispositions, neurobiological pathways, or environmental stressors, rather than being solely age‐dependent. Future research should aim to disentangle these shared factors to inform age‐agnostic strategies for preventing recurrence. Such insights could serve as an important reference for future revisions of clinical guidelines, particularly those concerning the management of MDD in adolescents. This research underscores the need for thoughtful consideration of how knowledge from adult populations might be integrated into adolescent treatment contexts, potentially enhancing therapeutic strategies and informing policy decisions.

5.1. Strengths and Limitations

Employing a Propensity Score Weighted model mitigated potential selection bias, enhancing causal inference and precision in estimating treatment effects. Excluding individuals with a history of alcohol or substance abuse, major affective disorders, or persistent mood disorders within ten years prior to their initial MDD diagnosis minimized confounding risks, ensuring outcomes were primarily attributable to MDD dynamics. This methodological decision enhanced the validity of conclusions regarding MDD recurrence rates. Additionally, the nuanced methodology for classifying ongoing MDD episodes provided a deeper understanding of MDD's clinical course. The regional focus of the cohort minimized the impact of diagnostic variability across regions.

This study has several limitations, including the absence of mortality data, which prevents differentiation between cases censored due to death versus other reasons. Additionally, the reliance on ICD‐10 codes for psychiatric comorbidities and MDD diagnoses may underestimate true rates of comorbid conditions, particularly anxiety disorders. These factors may influence recurrence patterns and warrant further investigation. Third, focusing on adolescent MDD excluded direct comparisons with childhood and adult recurrence rates. Reliance on clinical diagnoses to define MDD onset and recurrence may not consistently capture ‘disease‐free’ periods or unreported cases, introducing potential uncertainty. However, robust methodology, including propensity score weighting and comprehensive validation, mitigates these concerns. Fourth, while the study was powered to detect differences in recurrence rates around 5 percentage points, smaller disparities may exist, though they are unlikely to significantly impact policy decisions or understanding of neurobiological differences. Fifth, reliance on registered diagnoses could miss some episodes, particularly if treatment‐seeking behaviors varied by age. Additionally, the 5‐year follow‐up may not capture late recurrences, but consistent remission‐to‐recurrence times reduce this concern. Moreover, remission was defined as a period of at least 90 days without recorded MDD diagnoses, antidepressant dispensations, psychotherapy, or brain stimulation interventions. To address potential inaccuracies in this definition, sensitivity analyses with extended remission periods of 180 and 365 days were conducted, yielding consistent results. While this operationalization may imperfectly capture remission—potentially missing unreported depressive symptoms or treatment dropout—it is expected to affect adolescents and adults, similarly, thereby minimizing the likelihood of introducing between‐group bias. In addition, while the study focuses on first‐episode unipolar MDD, diagnostic transitions to other psychiatric disorders, such as bipolar disorder or schizophrenia, could occur during the 5‐year follow‐up period and may influence recurrence patterns. To mitigate this, individuals with prior diagnoses of bipolar disorder, persistent mood disorders, or psychotic disorders were excluded. However, subsequent diagnostic changes were not explicitly analyzed. Although the ICD coding system used in this study allows for tracking diagnostic transitions, our analysis was limited to the recurrence of MDD diagnoses. Future studies should consider the potential implications of diagnostic transitions on recurrence patterns. Finally, the age cap at 40 limits inferences for first MDD episodes in older individuals, due to challenges in identifying undiagnosed prior cases.

The relatively low rates of psychiatric comorbidities observed in our study may reflect its focus on first‐episode MDD cases and the reliance on primary care data, where comorbid conditions not directly addressed during consultations may be underreported. This limitation is further compounded by the stringent inclusion criteria, which excluded individuals with prior diagnoses of mood or other psychiatric disorders, potentially reducing the prevalence of comorbidities within the sample. While these exclusions strengthen the internal validity of the findings, they may also limit generalizability to populations with higher rates of psychiatric and non‐psychiatric comorbidities. Moreover, non‐psychiatric medical comorbidities, which may influence MDD trajectories, were not included in the dataset. Future studies should aim to address these gaps by incorporating data from specialized psychiatric care settings and capturing a broader range of medical conditions to provide a more comprehensive understanding of the interplay between comorbidities and MDD outcomes.

While stringent exclusion criteria were implemented to identify first‐episode cases by retrospectively examining medical records up to 10 years prior to the index MDD diagnosis, the possibility of undetected prior depressive episodes, particularly in adults, remains. This is due to potential gaps in historical psychiatric care documentation, which may influence recurrence rates and warrants careful interpretation. We also acknowledge potential variability in diagnostic documentation between adolescents and adults, which may impact observed recurrence rates. However, the use of propensity score weighting mitigates these differences, ensuring balanced comparisons between the groups.

6. Conclusions

Nearly half of all individuals experiencing a first episode of MDD undergo a recurrence within five years—a rate that exceeds previous estimates for adults but aligns with those reported for adolescents. Additionally, our analysis reveals no significant differences in the likelihood of 5‐year recurrence or in the interval from first remission to recurrence between those diagnosed with their first episode of MDD in adolescence versus adulthood.

The findings of this study have the potential to be broadly generalizable due to the comprehensive and representative nature of the cohort, which includes all individuals aged 13–40 years diagnosed with their first episode of MDD in the Stockholm region between 2011 and 2012. The dataset integrates extensive demographic, socioeconomic, and clinical data from a large catchment area of over 2 million people and includes both primary and secondary care. This enhances the external validity and applicability of the results to similar populations in other large, urban, Western healthcare settings with universal healthcare systems. The generalizability is further supported by a study design that minimizes selection bias and accounts for illness stage, severity, and other important prognostic factors. However, caution should be exercised when extrapolating these results to populations with different healthcare systems, cultural contexts, or those outside the specified age range.

Author Contributions

The authors were responsible for design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication.

Conflicts of Interest

Thomas Cars reported being an employee of and shareholder in Sence Research during the conduct of the study. The other authors declare no conflicts of interest.

Supporting information

Data S1.

Data Availability Statement

The data that support the findings of this study are available from the Swedish National Board of Health and Welfare, but restrictions apply to the availability of these data, which were used with ethical permission for the current study and therefore are not publicly available.