The longitudinal patterns of depression subtypes and stressors in depression severity in the Canadian longitudinal study on aging (CLSA)

Abstract

Aim

The current study aims to characterize the longitudinal patterns of depression subtypes and investigate the associations among the stability of depression subtypes, COVID‐19‐related stressors, and depression severity.

Methods

The study utilized data from the Canadian Longitudinal Study on Aging, which is a national, long‐term study of Canadian adults aged 45 and older (n = 12,957). Latent profile analysis was used to identify latent depression subtypes. Latent transition analysis was then applied to assess the stability of these subtypes over time. Hierarchical multivariate linear regression was used to explore the relationships among these identified depression subtypes, COVID‐19‐related stressors, and depression severity among males and females, respectively.

Results

Distinct depression subtypes were identified. Except for atypical depression, other depression subtypes showed greater stability over time. We also found that melancholic depression (B = 9.432) and typical depression (B = 6.677) were strongly associated with depression severity during the pandemic. Health‐related stressors (B = 0.840), conflict (B = 3.639), difficulties accessing resources (B = 0.927), separation from family (B = 0.840), and caregiving experience (B = 0.764), were significantly associated with increased depression severity. Sex‐specific analyses also revealed differences in the associations between stressors and depression severity between males and females.

Conclusions

This study contributes valuable insights into the latent clustering of depression subtypes and their stability. Stressors were associated with increased depression severity, with distinct associations observed among males and females. These findings have implications for targeted early interventions and integrated clinical management strategies by providing the evidence base for tailored mental health care during and after the pandemic.

Depression and depression subtypes

Depression in late life is a growing public health challenge as the population aged 65 and older rapidly increases domestically and worldwide. ¹ Compared to younger patients, older adults with depression more commonly have a poorer disease course, a high disease burden, ² , ³ reduced life satisfaction, ⁴ higher healthcare service use, and poor functional status. ⁵ , ⁶ However, research on late‐life depression has been hindered by the heterogeneity of depression. ⁷ , ⁸ The diagnosis of depression cannot reflect the wide array of possible depressive symptom combinations. A diagnosis of major depression in the Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria 5th edition is made if five out of nine symptoms are met, and several of these symptoms are opposites (e.g. weight gain or loss). ⁹ Two people with the same diagnosis of depression may have few symptoms in common. Although DSM‐5 contains several specifiers such as depression with atypical (characterized by increased sleep and appetite) or melancholic features (characterized by decreased sleep and appetite, and the presence of psychomotor symptoms), these have not proved sufficient to predict prognosis and treatment response. ¹⁰ , ¹¹ Furthermore, previous longitudinal studies demonstrated low stability of depression categories described in the DSM‐5. ¹²

A different approach to examining the heterogeneity of depression is through person‐based methods, such as latent class analysis (LCA) to identify symptoms that tend to co‐occur in patient subgroups. ¹³ These techniques cluster patients based on the congregation of different depressive symptoms without a pre‐conceived hypothesis. Studies have identified an ‘atypical’ subtype and a ‘typical’ or ‘melancholic’ subtype using LCA approaches. ¹⁴ , ¹⁵ , ¹⁶ These subtypes and their most distinguishing symptoms such as appetite and weight are also linked with distinct biological and genetic correlates and different neural activities. ¹⁷ , ¹⁸ Importantly, some pharmacological interventions are sensitive to certain depression subtypes. For instance, monoamine oxidase inhibitors are effective for atypical depression. ¹⁹ Therefore, subtyping depression has clinical significance as it ties to potential clinical utility. ²⁰ In addition, cohort studies of older adults have shown that depressive symptoms predict mortality, ²¹ , ²² and poor health outcomes as a gradient across the spectrum of symptoms which may not be limited to those who only have depressive symptoms above a certain threshold. ²³

Depression subtyping requires further research. Earlier data‐driven studies have found that depression subtypes in older adults were mainly based on severity, probably due to the inclusion of persons without a formal diagnosis of depression. ²⁴ , ²⁵ , ²⁶ Furthermore, previous work has yielded inconclusive findings between increase and decrease of sleep, appetite, weight, and psychomotor symptoms in various population groups, while these distinctions may be crucial in identifying depression subtypes. ¹⁴ , ¹⁵ , ²⁷ Likewise, studies pointed out that the DSM melancholic or atypical specifiers cannot help to create more homogeneous depression subgroups. ²⁸ , ²⁹ In addition, clinical symptoms may change over time, with some findings suggesting lower stability on melancholic and atypical features/subtypes. ³⁰ , ³¹ To further examine the validity and clinical usefulness of depression subtypes, an in‐depth exploration of symptoms subtypes and their stability over time is needed to identify their associations with related factors and the transition of subjects across subtypes over time.

COVID‐19‐related stressors and depression subtypes

The COVID‐19 pandemic has placed significant psychological and physical health burdens on individuals and societies. ³² The literature has consistently shown a substantial increase in the prevalence of depression worldwide since the outbreak of COVID‐19. ³³ , ³⁴ The overall pooled global prevalence of major depression in the general population from January to March 2020 was 31.4%. ³⁵ Females and unemployed individuals were more likely to report mental disorders during the pandemic. ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ The COVID‐19‐related stressors, such as contracting SARS‐CoV‐2 virus, financial strain, domestic conflicts, and family re‐arrangements are associated with depression and other mental disorders during the pandemic. ⁴² , ⁴³ , ⁴⁴ , ⁴⁵

In line with the diathesis‐stress model, exposure to COVID‐19‐related stressors makes people more prone to depression and other mental disorders. ⁴⁶ Although the associations between depression and COVID‐19‐related stressors have been suggested, several critical issues remain to be addressed. Depressive symptoms vary from time to time. It is reasonable to postulate that more variations in depressive symptoms during the pandemic given exposure to COVID‐19‐related stressors. Most studies on mental disorders during the pandemic were conducted by web‐based surveys using convenience samples, ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ which are subject to selection bias and inaccurate results. Additionally, no study has comprehensively investigated the associations among depression subtypes, COVID‐19‐related stressors, and depression severity during the pandemic. Studying a representative sample of the general population is pertinent to fully understanding the pandemic's mental health impact. ⁵¹ Importantly, sex differences have been widely recognized in many aspects of depression in terms of rates, symptoms, risk factors, and disease course. ⁵² , ⁵³ Incorporating sex analyses in the study is required to improve population mental health care and promote sex equity goals. ⁵⁴

Given the importance and inconclusive findings on depression subtyping and the limited knowledge of stability across different subtypes, it is necessary to deeply dive into depression subtypes and their stability over time. From a clinical perspective, it is also meaningful to examine to what extent different subtypes are attributable to depression severity while considering stressors. To address these knowledge gaps, the current study aimed to (1) identify the unique clustering of depression subtypes and test the stability of depression subtypes before the outbreak of the pandemic in a large representative population cohort; (2) examine whether the stability of depression subtypes before the pandemic and COVID‐19‐related stressors were linked to the severity of depressive symptoms during the pandemic; and (3) identify whether there were sex differences in these relationships during the pandemic.

Methods

Study cohort

Data used were from the Canadian Longitudinal Study on Aging (CLSA) baseline, Follow‐up 1 and 2 (FUP 1 and FUP 2), and COVID‐19 baseline and exit surveys (See Fig. 1 for the flow chart of sample selection). The CLSA is a large, national, longitudinal study of more than 50,000 Canadians who are being followed for at least 20 years. ⁵⁵ The CLSA consists of a Comprehensive cohort and a Tracking cohort. The present study selected those participants of the Comprehensive cohort who also participated in both CLSA COVID‐19 baseline and exit surveys. The Comprehensive Cohort, with 30,097 participants, underwent in‐home interviews and visited a CLSA data collection site for physical and clinical assessment. Recruitment involved random selection of the 11 data collection sites in seven Canadian provinces. Follow‐ups were conducted every 3 years, with baseline data collected between 2011 and 2015. ⁵⁶ CLSA has received ethical approval from the Hamilton Integrated Research Ethics Board. All CLSA participants provided written informed consent and their anonymity was preserved using methods approved by the Ethics Committee. This study received Ethics Approval from the Institutional Research Ethics Board at the Douglas Research Centre and conforms to the provisions of the Declaration of Helsinki.

Fig. 1.

Flow chart of the process of selection of study subjects.

Participants

The present study cohort includes (1) adults aged 45 years and older; and (2) those CLSA participants of the comprehensive cohort who also participated in both CLSA Covid‐19 baseline and exit surveys. Finally, 12,957 participants eligible across the five timepoints were included in the current study.

Measures

Depressive symptoms were measured using the Center for Epidemiologic Studies Short Depression Scale (CES‐D10). ⁵⁷ It includes three items on depressed affect, five items on somatic symptoms, and two items on positive affect. The CES‐D10 is reliable and valid in assessing depressive symptoms in adults, with the internal consistency of 0.86, the test–retest reliability of 0.85, the convergent validity of 0.91 and the divergence of 0.89. ⁵⁸ , ⁵⁹ Depression subtypes were derived by person‐centered approaches which were measured at the COVID‐19 baseline survey (April–May 2020). Whereas depression severity was determined by the total score of CES‐D10 items which was measured at the COVID‐19 exit survey (September–December 2020). A score of 10 was used to define the cut‐off point for non‐severe and severe depression, respectively. ⁶⁰ Additionally, for individuals who suffered depression at least once across the years of 2011 to 2020, we classified them based on the consistency of their depression subtypes across these time points. Specifically, those who maintained the same depression subtype for more than two consecutive time points were categorized as a chronic specific depression subtype, except for those who exhibited recovery at FUP2, were designated as “recovered.” In contrast, individuals who had a new depression diagnosis at FUP2 were identified as “new cases.” Melancholic depression subgroup was marked by infrequent positive affects, frequent depressive affects, and somatic complaints. Typical depression subgroup exhibited pronounced depressive affects, appetite loss and weight loss, whereas the atypical depression subgroup exhibited appetite gain and weight gain. Different subtypes of depression were characterized by three distinct depression subtypes across time. Our findings revealed a total of seven identifiable longitudinal trends: no depression; recovered; new cases; different subtypes of depression; typical depression; melancholic depression; and atypical depression.

Five COVID‐19‐related stressors were measured using a self‐reported questionnaire at the COVID baseline survey: (1) Health‐related stressors were identified by asking participants to indicate whether they were ill or if someone close to them was ill or had died due to COVID‐19 or non‐COVID‐19‐related reasons; (2) Difficulties with accessing resources was identified by asking participants to indicate whether they had experienced loss of income, and difficulties in accessing necessary supplies, food and usual healthcare including prescription medications and treatments; (3) Conflict was identified by asking participants to report whether they had experienced increased verbal or physical conflict; (4) Separation from family was identified by asking participants to report whether they were separated from family during the pandemic; (5) Caregiving experience assessed whether participants had spent increased time in caregiving or whether they were unable to care for people who required assistance due to a health condition or limitation.

Statistical analysis

Descriptive statistics, including measures such as frequencies and percentages, were employed to summarize key demographic variables. To identify the unique clustering of depression subtypes, we first used latent profile analysis (LPA) to discover unique classes of community‐dwelling middle‐aged and older adults based on their responses to the CES‐D items at the survey time points of Baseline, FUP1, and FUP2. The optimal number of latent classes was determined by the model fit among the four consecutive models with two to five classes each, which was assessed with criteria favoring lower values of Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), sample size‐adjusted BIC, higher entropy value, and a statistically significant outcome in the bootstrap likelihood ratio test (BLRT). ⁶¹ We then performed latent transition analysis (LTA) to identify the stability of the latent memberships of depression subtypes. To determine an optimal number of clusters, several LTA models were estimated, adding another cluster (k) to each consecutive model and comparing entropy and fit indices to the previous model (k‐1). Hierarchical multivariate linear regression models were then performed to examine how trajectories of depression subtypes and the COVID‐19‐related stressors exposures are associated with depression severity. Age and sex were included as covariates because they are critical demographic factors that are associated with both stressors and depression. Therefore, the current study includes Model 1: depression subtypes across the year of 2011 to 2020; Model 2: Model 1+ COVID‐19‐related stressors at COVID‐19 baseline survey; Model 3: Model 2 + age, sex. The model fit assessment includes R‐square and the change in R‐square. These measures assess the proportion of variance explained by the model and the improvement in explanatory power when additional factors are added. Descriptive analysis and regression models were performed using Stata, version 15, and LPA and LPA were performed by Mplus, version 8.

Results

The study cohort predominantly comprises participants aged 55–64, with the smallest representation from those 75 and older. Most participants are married or in common‐law partnerships. The majority hold post‐secondary degrees, with males generally earning higher incomes than females. Most participants fall in the middle‐income range, and the cohort is overwhelmingly White. Additionally, over 80% of participants are non‐immigrants. The overall depression score for the participants ranged from 4.35 to 6.02 across the 10 years.

Identified latent depression subtypes among the study cohort

The characteristics of the baseline CLSA cohort are presented in Table S1. Person‐centered methods were used to identify the latent clustering of depression subtypes based on depressive symptoms collected on the Baseline, FUP1 and FUP2 surveys, separately. A total of 2901 participants with at least one diagnosis of depression across the three timepoints were included in this analysis. In the present study, four profile models with the best model fit were selected for all three timepoints (Tables S2–S4). Profile 1 was labeled as “No/Mild depression” for being characterized by having the lowest probabilities of depressed mood, feeling lonely, etc., which was the most common depression subtype. Profile 3 was labeled as “Melancholic depression”, which exhibited high probabilities of symptoms like depressed mood, loneliness, insomnia, and trouble concentrating. Profile 2 was named “Typical depression” for displaying severe typical symptoms, including decreased appetite, weight loss, and insomnia. In contrast, Profile 4, named “Atypical depression,” had featured symptoms like increased appetite and weight gain. More details are in Figs S1–S3.

Latent transition in depression subtypes from baseline to FUP2

Based on the four‐profile LPA solution, LTA was conducted to identify the longitudinal transition patterns of depression subtypes from baseline to FUP2. LTA was adjusted for age, sex, marital status, education attainment, ethnicity, the presence of depression, age onset of depression and considered population weights. The findings on the transition probabilities from baseline to FUP1 and from FUP1 to FUP2 are shown in Tables S5 and S6, respectively. Table S5 presents transition probabilities from baseline to FUP1 for participants categorized into diverse depression subtypes. Participants with no symptoms/mild symptoms were more likely to remain at FUP1 (54.7%) and less likely to transit to “Melancholic depression” (8.0%). Conversely, participants initially categorized as “Melancholic depression” had the highest probability of sustaining their initial classification at follow‐up (66.4%) and were less likely to transit to other subtypes. In a similar pattern, participants with typical depression had a higher probability of remaining in the same class (36.2%). Lastly, individuals with atypical depression exhibited a higher likelihood of transiting to no/mild depression (31.7%) and a lower likelihood of transiting to melancholic depression (19.5%). Table S6 shows transition probabilities from FUP1 to FUP2 among four depression subtypes. Individuals initially having no symptoms/mild symptoms had a higher probability (50.8%) of maintaining their classification and a reduced probability of transiting to other depression subtypes. Likely, participants initially diagnosed with melancholic depression exhibited a pronounced probability (53.6%) to remain. Those originally characterized as having typical depression showed a higher probability (41.3%) of remaining rather than transiting. Conversely, individuals with atypical depression had a relatively high probability (36.1%) of transition to no/mild depression. Transition probabilities of identified depression subtypes over time are presented in Fig. 2.

Fig. 2.

Transit probabilities of depression subtypes across time.

How COVID‐19 stressors and depression subtypes before the COVID‐19 pandemic were related to depression severity during the COVID‐19 pandemic

We extended our analysis to explore how COVID‐19 stressors and depression subtypes relate to depression severity during the pandemic (Table 1). Only depression subtypes were included in the initial model – Model 1. We found that recovered, new cases, different types of depression, typical depression, melancholic depression, and atypical depression were associated with significantly higher depression severity scores in comparison to individuals without depression (P < 0.01). Notably, melancholic depression (B = 10.063) had the strongest association with depression severity, followed by typical depression (B = 7.395), different types of depression (B = 7.066) and typical depression (B = 6.965). COVID‐related stressors were then included in Model 2. Health‐related stressors (B = 0.841), conflict (B = 3.668), difficulties in accessing resources (B = 0.944), separation from family (B = 0.827), and stressors related to caregiving experience (B = 0.799) were consistently linked with higher depression severity scores (P < 0.01). Model 3 further included sex and age and indicated that being a female (B = 0.729) and individuals aged between 45 to 54 years old (B = 0.377) were associated with higher depression severity scores. Model 3 had the best fit, with these factors collectively accounting for a significant portion of the variability in depression severity during the COVID‐19 pandemic (R² = 0.303).

Table 1.

Hierarchical regression analysis of depression subtypes and Covid‐related stressors in depression severity during the Covid‐19 pandemic (n = 12,957)

	Model 1			Model 2			Model 3
Predictor	B	SE B	β	B	SE B	β	B	SE B	β
Block1
Depression subtypes
Recovered	4.361**	0.185	0.195	3.913**	0.177	0.175	3.891**	0.177	0.174
New cases	5.890**	0.233	0.209	5.372**	0.223	0.191	5.364**	0.223	0.190
Different types of depression	7.066**	0.250	0.233	6.426**	0.239	0.213	6.403**	0.239	0.212
Typical depression	7.395**	0.403	0.151	6.687**	0.388	0.136	6.677**	0.388	0.136
Melancholic depression	10.063**	0.318	0.262	9.433**	0.306	0.244	9.432**	0.306	0.244
Atypical depression	6.965**	0.340	0.169	6.026**	0.327	0.146	6.027**	0.327	0.146
Block 2
Health‐related stressors (yes)				0.841**	0.088	0.076	0.840**	0.088	0.076
Conflict (yes)				3.668**	0.174	0.169	3.639**	0.174	0.167
Difficulties with accessing resources (yes)				0.944**	0.087	0.087	0.927**	0.088	0.086
Separation from family (yes)				0.827**	0.084	0.080	0.840**	0.084	0.081
Caregiving experience (yes)				0.799**	0.109	0.059	0.764**	0.110	0.057
Block 3
Sex (female)							0.729**	0.087	0.071
Age
45–54y							0.377**	0.104	0.032
65–74y							−0.010	0.105	−0.001
75+ y							0.152	0.145	0.009
F		342.872			319.883			259.488
R2		0.232			0.297			0.303
ΔR2		0.232			0.065			0.006

^**Indicates P < 0.01; Model 1: depression subtypes across the years of 2011 to 2020; Model 2: Model 1+ COVID‐19‐related stressors at COVID‐19 baseline survey; Model 3: Model 2 + age, sex.

β, standardized coefficients; B, unstandardized coefficients; SE, Standard error.

Furthermore, we discovered that all the identified depression subtypes (including recovered, new cases, different types of depression, typical depression, melancholic depression, and atypical depression) were associated with depression severity among females. Among these subtypes, melancholic depression (B = 9.259) and different types of depression (B = 6.634) displayed the strongest associations. Our results also revealed that all stressors were significantly related to depression severity. Similarly, for males, all depression subtypes were significantly associated with higher depression severity, especially melancholic depression (B = 9.573) and typical depression (B = 7.693). Additionally, all stressors except for separation from the family were significantly associated with depression severity (P ≤ 0.01) with an additional 6.5% of the variance in depression severity compared to the initial model, indicating the substantial roles of these COVID‐19‐related stressors in explaining variations in depression severity during the pandemic among males (Tables 2 and 3).

Table 2.

Hierarchical regression analysis of depression subtypes and Covid‐related stressors in depression severity during the Covid‐19 pandemic among females (n = 6821)

	Model 1			Model 2
	B	SE B	Β	B	SE B	Β
Block 1
Depression subtypes
Recovered	4.158**	0.248	0.195	3.711**	0.239	0.174
New cases	5.317**	0.330	0.187	4.879**	0.317	0.171
Different types of depression	7.170**	0.337	0.246	6.634**	0.323	0.229
Typical depression	6.602**	0.566	0.135	5.943**	0.549	0.120
Melancholic depression	10.067**	0.433	0.269	9.259**	0.417	0.248
Atypical depression	6.879**	0.437	0.182	5.863**	0.422	0.155
Block 2
Health‐related stressors (yes)				0.995**	0.130	0.086
Conflict (yes)				3.791**	0.254	0.167
Difficulties with accessing resources (yes)				1.059**	0.130	0.092
Separation from family (yes)				0.742**	0.129	0.065
Caregiving experience (yes)				0.739**	0.154	0.055
F		165.731			155.266
R2		0.219			0.284
ΔR2		0.219			0.065

^**Indicates P < 0.01; Model 1: depression subtypes across the years of 2011 to 2020; Model 2: Model 1+ COVID‐19‐related stressors at COVID‐19 baseline survey.

β, standardized coefficients; B, unstandardized coefficients; SE, Standard error.

Table 3.

Hierarchical regression analysis of depression subtypes and Covid‐related stressors in depression severity during the Covid‐19 pandemic among males (n = 6136)

Predictor	Model 1			Model 2
	B	SE B	β	B	SE B	β
Block 1
Depression subtypes
Recovered	4.445**	0.280	0.189	4.083**	0.268	0.174
New cases	6.533**	0.321	0.242	5.953**	0.309	0.220
Different types of depression	6.553**	0.374	0.209	5.840**	0.359	0.187
Typical depression	8.404**	0.562	0.177	7.693**	0.539	0.163
Melancholic depression	9.917**	0.465	0.254	9.573**	0.451	0.242
Atypical depression	6.612**	0.562	0.140	5.986**	0.538	0.127
Block 2
Health‐related stressors (yes)				0.612**	0.117	0.061
Conflict (yes)				3.556**	0.231	0.177
Difficulties with accessing resources (yes)				0.847**	0.114	0.087
Separation from family (yes)				0.113	0.127	0.011
Caregiving experience (yes)				0.733**	0.153	0.056
F		164.789			152.294
R2		0.230			0.295
ΔR2		0.230			0.065

^**Indicates P < 0.01; Model 1: depression subtypes across the years of 2011 to 2020; Model 2: Model 1+ COVID‐19‐related stressors at COVID‐19 baseline survey.

β, standardized coefficients; B, unstandardized coefficients; SE, Standard error.

Discussion

This study provides one of the first evidence of the stability of latent depression subtypes stratified based on their depressive symptoms before the COVID‐19 pandemic in a longitudinal national cohort of community‐based middle‐aged and older adults. We also discovered significant associations between these identified depression subtypes and COVID‐19‐related stressors, and depression severity during the COVID‐19 pandemic. Importantly, sex‐specific analyses revealed distinct associations between depression subtypes, COVID‐19‐related stressors, and depression severity among males and females. Separation from family during the COVID‐19 pandemic was associated with increased severity of depression only for females.

In line with a recent review, depression subtypes were characterized by variations in distinctive combinations of depressive symptoms probabilities. ⁶² The current study discovered that atypical depression exhibited atypical or reversed neurovegetative symptoms, primarily characterized by increased appetite and weight gain. In contrast, melancholic depression and typical depression, displayed typical neurovegetative symptoms, such as trouble concentrating, insomnia, and depressed mood, with typical depression characterized by distinct decreased appetite, and weight loss. Notably, we observed that all depression subtypes were likely to remain in the same depression subtype group over time, except for atypical depression. The research on the stability of depression subtypes has been focused on a specific depression subtype. For melancholic depression, previous findings reported a low recovery rate among those with the first depressive episode. ⁶³ We also found a high stability of melancholic depression across time. For typical depression, studies have shown that the symptoms of typical depression are more likely to have a persistent disease course and co‐morbid with other psychiatric disorders. ⁶⁴ In contrast, atypical depression was less stable over time. ⁶⁵ This phenomenon is particularly pronounced among the elderly since advanced age and the aging process are frequently linked with declining physical health, such as weight loss and muscle weakness, as well as somatic comorbidities, particularly in elderly individuals experiencing depression. These physiological processes may impact the manifestation of late‐life depression, diminishing the discriminative power of symptoms like changes in appetite and weight over time. ⁶⁶

Notably, melancholic depression and typical depression before the COVID‐19 pandemic were the two depression subtypes with the strongest associations with depression severity during the pandemic. These two depression subtypes had distinct symptoms that were associated with poor disease course. For instance, melancholic depression was indicated as a more severe variant of mood disorder with an especially virulent course that tends to exhibit a strong correlation with the overall severity of depression. ⁶⁷ , ⁶⁸ Similarly, typical depression is featured by the state of hyperarousal and fear, often originating from a sense of personal unworthiness and pessimism about the future, and seems to consistently predict the trajectory of the subsequent depressive conditions. ⁶⁹ , ⁷⁰ This study also unveiled the significant associations between proximal stressors, specifically COVID‐19‐related stressors, and worsening depression. COVID‐19‐related stressors including health‐related stressors, conflict, difficulties accessing resources, separation from family, and caregiving experience stressors also related to the severity of depression. Our study is consistent with previous evidence indicating difficulty in accessing resources including food, healthcare, and prescription medications/treatments during the pandemic was associated with severe depression. ⁷¹ The COVID‐19 pandemic has introduced additional challenges that can have a significant impact on middle‐aged and older adults. ⁷² They often face unique health and socioeconomic challenges that could influence the severity of depression. Compared to younger adults, they tend to face more financial pressure and family commitments, which make them more susceptible to and experience more negative impacts from COVID‐19‐related stressors, e.g., unemployment, family conflict, etc ⁷³ Furthermore, health‐related vulnerabilities, such as chronic illnesses and age‐related immunity declines, make this demographic more susceptible to COVID‐19‐related stressors. This distress is reflected in increased rates of various mental problems such as anxiety, depression, and trauma‐related stress disorder, as also observed in surveyed American older adults. ⁷⁴ The pandemic also made it physically, emotionally and economically more difficult for caregivers to provide care. Access to healthcare and resources may also be more difficult for them due to their chronic health issues and/or technology barriers they face. The demands of caregiving could add additional burdens. The inability to access healthcare services, particularly for individuals with pre‐existing mental health conditions, frailty, or multiple health issues, for example, the elderly, can result in new or worsening depressive conditions over time. ⁷² Beyond these practical difficulties, separation from family members was also associated with depression severity for females. During the pandemic, certain older individuals had conveyed a feeling of amplifying negative perceptions, experiencing a sense of being unnecessary and forgotten, particularly due to quarantine measures and restrictions. ⁷⁴ The limitations imposed during this time have contributed to a heightened sense of isolation and marginalization among some seniors, intensifying their sentiments of being overlooked and unimportant. These findings were comparable to similar studies conducted in various countries. ⁷⁵ , ⁷⁶ Notably, the current results showed that verbal or physical conflict experiences were closely linked to increased depressive symptoms during the COVID‐19 pandemic. Prior studies have identified interpersonal conflict as a predisposing factor for depression among the elderly. ⁷⁷ Recent investigations suggest a significant decline in family functioning during the pandemic since prolonged periods spent together at home, coupled with disruptions to coping mechanisms that typically mitigate familial conflict, such as social support, logically implies an escalation in interpersonal conflicts. ⁷⁸ The current study also confirmed that females exhibited a higher likelihood of experiencing more severe depression during the pandemic. Females might be undergoing more pronounced elevations in anxiety and depressive symptoms amid the pandemic, given the inclination of the feminine mental health profile towards internalizing disorders. ⁷⁹ In our sex‐dimorphism analysis, we did observe that separation from family during the pandemic was also associated with depression severity among females. The effects of separation from family on depression severity connect with various factors, such as cultural, social, psychological, and individual differences. Traditionally, females tend to place a higher value on interpersonal relationships and family connections. As a result, the absence or separation from family may have a more pronounced impact on females who may have received significant support from familial relationships. ⁸⁰ For some cultures, females/women are more likely to expect to have more familial roles, i.e., caregivers, and maintainers of family bonds. Therefore, separation from family may be perceived as more challenging for females/women based on societal expectations. Additionally, females may experience different psychosocial stressors related to separation from family, such as concerns about family responsibilities, and caregiving roles. ⁸¹ These factors can also contribute to the increased stress sensitivity to separation from family. The current study sheds light on how societal norms and expectations can exacerbate stress and depression in females during periods of isolation or separation, emphasizing the need for targeted mental health interventions that take into account the female‐specific emotional and social support systems. Recognizing sexual disparities can guide the development of more effective mental health support strategies that are tailored to the unique challenges faced by different sex groups, potentially leading to more equitable mental health outcomes. This finding could serve as an evidence base to direct interventions that specifically address the psychosocial stressors linked to family dynamics and caregiving roles that disproportionately affect females.

Strengths and limitations

This study offers a comprehensive examination of how latent depression subtypes evolved over time and to what extent COVID‐related stressors and the stability of depression subtypes are related to depression severity. The analyses were conducted with robust depression measures, rigorous statistical analyses, and a large population‐based sample. Sex‐specific analyses demonstrated differential patterns of depression subtypes as well as various COVID‐19 stressors attributable to depression severity. The current study not only sheds light on precision medicine in terms of finer‐mapping depression diagnosis and specific clinical management but also informs dedicated intervention strategies and more targeted and effective mental health care in the preparation of public emergencies. Depression subtyping and its stability over time plays a crucial role in clinical practice by offering tailored treatment approaches, providing insights into the disease course and anticipated potential complications, leading to more effective interventions and clinical management, allowing the potential for research advancements in the underlying mechanisms of depression, and facilitating more appropriate allocating resources to meet the demands.

There are several limitations to be noted. First, the study primarily focuses on a Canadian national cohort of middle‐aged and older adults. The generalizability of research findings to other populations with significantly different sociodemographic characteristics may be limited. Second, depressive symptoms were measured by a self‐reported questionnaire (CES‐D). Even though the psychometric of CES‐D in population studies has been proved, ⁸² the presence of depression in the study is not equivalent to clinical diagnosis. Additionally, participants might underreport or overreport their experiences, which is susceptible to response bias. Third, these identified depression subtypes warrant further replication studies to explore underlying neurobiological mechanisms. Last, the current study only controlled for age and sex, which may not sufficiently address potential confounding (i.e., unknown confounding) that can lead to spurious results. Therefore, future research is warranted to incorporate a wider range of confounders to validate the current findings.

Conclusion

The present study discovers the latent depression subtypes and their stability prior to the COVID‐19 pandemic in a national cohort of middle‐aged and older Canadians. The significant associations between pre‐pandemic depression subtypes and COVID‐19‐related stressors, and heightened depression severity during the pandemic were noted. Sexual dimorphism in the severity of depressive symptoms was also discovered. The findings of this study underscore the need for tailored mental health interventions based on distinct depression subtypes and their stability. Sex‐specific approaches are crucial given the differential impacts of COVID‐19‐related stressors on males and females. It is essential to integrate mental health support into public health responses to pandemics and ensure that resources are allocated effectively to address mental health needs during crises, facilitating early interventions and continuous monitoring to manage depression effectively.

Funding Information

This study was funded by a Catalyst Grant: Analysis of CLSA Data, from the Canadian Institutes of Health Research, grant number 23ca014.

Disclosure statement

The authors declare no competing interests.

Author contributions

XFM and YYS had full access to all of the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: XFM. Acquisition, analysis, and interpretation of data: All authors. Drafting of the manuscript: XFM, YYS. Critical revision of the manuscript for important intellectual content: All authors. Statistical analysis: XFM, YYS. Administrative, technical, or material support: MZL, NS. Study supervision: XFM.

Supporting information

Data S1 Supporting Information