Article Text
PDF
Abstract
Introduction The COVID-19 pandemic had large impacts on mental health; however, most existing evidence is focused on the initial lockdown period and high-income contexts. By assessing trajectories of mental health symptoms in India over 2 years, we aim to understand the effect of later time periods and pandemic characteristics on mental health in a lower-middle income context.
Methods We used data from the Real-Time Insights of COVID-19 in India cohort study (N=3709). We used covariate-adjusted linear regression models with generalised estimating equations to assess associations between mental health (Patient Health Questionnaire (PHQ-4) score; range 0–12) and pandemic periods as well as pandemic characteristics (COVID-19 cases and deaths, government stringency, self-reported financial impact, COVID-19 infection in the household) and explored effect modification by age, gender and rural/urban residence.
Results Mental health symptoms dropped immediately following the lockdown period but rose again during the delta and omicron waves. Associations between mental health and later pandemic stages were stronger for adults 45 years of age and older (p<0.001). PHQ-4 scores were significantly associated with all pandemic characteristics considered, including estimated COVID-19 deaths (PHQ-4 difference of 0.10 units; 95% CI 0.06 to 0.13), government stringency index (0.14 units; 95% CI 0.11 to 0.18), self-reported major financial impacts (1.20 units; 95% CI 1.09 to 1.32) and COVID-19 infection in the household (0.36 units; 95% CI 0.23 to 0.50).
Conclusion While the lockdown period and associated financial stress had the largest mental health impacts on Indian adults, the effects of the pandemic on mental health persisted over time, especially among middle-aged and older adults. Results highlight the importance of investments in mental health supports and services to address the consequences of cyclical waves of infections and disease burden due to COVID-19 or other emerging pandemics.
- COVID-19
- Mental Health & Psychiatry
- Epidemiology
- Public Health
- Cohort study
Data availability statement
Data are available in a public, open access repository. Data are available to download via the Gateway to Global Aging website: covid.g2aging.org. Users must sign a data use agreement before being granted access to the data.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
The COVID-19 pandemic had wide-reaching impacts, and prior evidence suggests that the initial lockdown phase of the pandemic was associated with negative mental health outcomes.
The conditions surrounding the COVID-19 pandemic have evolved over time, and later pandemic stages were characterised by a lifting of lockdown measures and higher disease burden.
WHAT THIS STUDY ADDS
This study shows that the mental health impacts of the COVID-19 pandemic persisted over time, especially for middle-aged and older adults.
Later waves of the COVID-19 pandemic were associated with increases in mental health symptoms for middle-aged and older adults, but not younger adults.
Pandemic characteristics, including estimated cases and deaths, as well as COVID-19 infection in the household, were also associated with a higher burden of mental health symptoms.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Mental health outcomes should be considered in quantifying the overall burden and impact of COVID-19.
Mental health supports and services are needed during cyclical waves of COVID-19 or other emerging infections.
Introduction
COVID-19 has had wide-reaching impacts since the WHO first declared COVID-19 as a global pandemic on 11 March 2020.1 In reaction to the initial spread of COVID-19, many countries implemented strict lockdown periods, closing businesses, schools and requiring people to stay home.2 3 Concerns over short-term and long-term mental health consequences of the pandemic and lockdown period quickly emerged and were later backed up by quantitative evidence.4–8 Although the majority of the empirical evidence comes from high-income settings, data from India and other low-income and middle-income countries also highlight the impact of the initial lockdown period on mental health outcomes in these contexts.9 10
Since the initial stages of the pandemic, the direct impacts of COVID-19 have continued to grow, with over 7.2 million reported and 17.7 million estimated deaths globally through 1 December 2022.11 The state of the pandemic has also evolved over time, with rapid shifts due to new variants, waves and the development of vaccines. Despite potential impacts of the constantly changing landscape on mental health, longitudinal studies of mental health that assess multiple phases of the pandemic, including the lockdown period as well as the delta and omicron waves, have been relatively rare.
Existing studies of mental health trajectories during the COVID-19 pandemic have predominantly focused on the initial lockdown period in 2020 and have been concentrated in high-income settings.12–16 Evidence from these studies covering the early stages of the pandemic largely suggests that mental health symptoms peaked in the initial lockdown phase, and recovered as lockdown measures were lifted.12–16 However, these studies are unable to make conclusions about the impacts of later waves and changes during the pandemic. Although fewer studies assessing longer time periods exist, some evidence from Argentina, Australia, Switzerland and the USA suggests that trends may be more complex, with both increases and decreases in mental health symptom burden during later periods of the pandemic.17–21
India is the world’s most populous country, with a population of over 1.4 billion people, and has had high levels of COVID-19 infections and deaths during the COVID-19 pandemic.22–24 Critical challenges to the healthcare infrastructure also exist, including problems with equitability and accessibility,25 that likely served to further exacerbate pandemic outcomes and differences in physical and mental health across segments of the population. However, to our knowledge, only one prior study from India assessed mental health in the general population over more than two time periods.26 Data showed that the initial lockdown period and the second wave (delta) led to similar levels of depression in rural but not urban settings.26 However, the study was limited to two distinct regions in India and did not include follow-up beyond June 2021. As the impacts of COVID-19 continue to persist and evolve, it is important to understand the effects of later pandemic periods on mental health. The Indian context presents an important opportunity to study mental health outcomes throughout all stages of the pandemic due to the presence of distinct pandemic phases coinciding with strict lockdown measures (initial phase), large numbers of pandemic-related deaths (delta wave) and high numbers of infections (omicron wave). In other countries, such as China or the USA, the different phases were less distinct—China largely avoided an extreme delta wave due to the successful continuation of lockdown protocols, and the USA had a less extreme lockdown period and delta wave. The separation in time and stark differences between these phases in India can enhance our ability to disentangle the effects of the specific factors impacting mental health.
We aim to assess mental health over the course of the COVID-19 pandemic in India using data from the Real-Time Insights of COVID-19 in India (RTI COVID-India) survey, a nationwide survey with nine waves of data collection spanning from May 2020 to May 2022. By leveraging data over the course of the pandemic, we describe the impacts of various phases of the pandemic on mental health and identify characteristics of the pandemic most strongly associated with poor mental health. We also assess whether the impact of the pandemic and pandemic characteristics on mental health varies by age, gender and urban/rural residence.
Methods
Sample
The RTI COVID-India survey included respondents aged 18 and older from 1766 households across India.27 28 The study sampled households from the Harmonized Diagnostic Assessment of Dementia for the Longitudinal Aging Study in India (LASI-DAD), a nationally representative sample of adults 60 years and older from 18 states and union territories. For all LASI-DAD households with a valid phone number, two randomly selected household members over the age of 18 (one man and one woman if possible) were invited to participate in round 1. In subsequent rounds, these same participants were reinterviewed, but if unavailable additional randomly selected household members were invited. In wave three, all primary LASI-DAD respondents (60 years and older) were additionally targeted for enrolment if not previously enrolled. Nine rounds of telephone-based surveys were administered from May 2020 to May 2022 (online supplemental appendix figure 1). Respondents contributed an average of 5.6 (SD=2.5) rounds of data collection, with data collected from a total of 21 108 phone surveys and 3797 individuals. Additional details on the sample selection and characteristics of the RTI COVID-India survey are available elsewhere.27 We excluded individuals without valid data on items from the Patient Health Questionnaire (PHQ-4) (682 surveys; 233 respondents), without valid data on pandemic characteristics of interest (1558 surveys; 52 respondents) or without valid data on covariates (24 surveys; 4 respondents).
Supplemental material
Mental health measure
The PHQ-4 was used as a screening tool for anxiety and depression and was administered across all rounds.29 The PHQ-4 asks respondents how often they have been bothered by: (1) feeling nervous, anxious or on edge, (2) not being able to stop or control worrying, (3) feeling down, depressed or hopeless and (4) having little interest or pleasure in doing things in the last 2 weeks. Included response options are: not at all (0), several days (1), more than half the days (2) and nearly every day (3). Therefore, the full range of the scale is 0–12 and higher scores indicate higher mental health symptom burden. Due to the brief nature of the tool, we consider the four-item scale as a single marker of poor mental health. In primary models, we treat the 0–12 PHQ-4 score as a continuous measure of mental health. In a sensitivity analysis, we also compare primary results with those from models substituting the 0–12 PHQ-4 score with estimates of poor mental health derived from a measurement model (item response theory graded response model). These models relax the standard assumptions that all items contribute equally to total score and that differences between response options are equal. Variance in the factor represents the shared variance among the indicators and does not include random error specific to each item. Scores were scaled to have a mean of 0 and SD of 1. In online supplemental analyses we conducted two sensitivity analyses. First, we compared primary results with results from models for the presence of abnormal PHQ-4 scores (PHQ-4>2) as a binary outcome to assess the clinical relevance of findings.29 For binary models, we used Poisson regression with robust variance to estimate prevalence ratios. Second, we also re-estimated models using estimated factor scores for poor mental health from item response theory models.
Pandemic measures
We assessed both reported and estimated state-level numbers of cases and deaths (per 100 000 population (from the 2011 Indian census)) due to COVID-19 in India. Reported administrative data from the Indian government was accessed at https://data.covid19bharat.org. Because official reported numbers are known to undercount numbers of cases and deaths, we also used estimated infections and excess mortality from the Institute of Health Metrics and Evaluation (IHME).11 22 23 To capture the impact of lockdown measures, we used a state-specific index measure of government stringency, which captures the intensity of closings of public spaces and restrictions on gatherings and movement (details in online supplemental appendix 3).30 To facilitate comparisons between indicators, we converted them into z-scores. Because respondents’ mental health is likely affected by the conditions during the time period leading up to the survey, we created summary indicators based on the mean level of cases, deaths and government stringency over the time period leading to the interview date. We used a period of 3 months for the health-related indicators and a period of 1 month for the stringency index. Policies are often enacted and enforced quickly, and therefore, are likely to have more immediate impacts on mental health. In contrast, the effects of population health burden may last longer and be influenced by ongoing, longer-term trends. The longer, 3-month period for health-related indicators was selected to ensure that the characteristics of the environment and any ongoing waves of infections would be appropriately represented in calculated cumulative means, given that the largest waves of infections were 4 months (delta) and 2 months (omicron) in length. Based on the real-time and lagged metrics of cases, deaths and government response, the Indian Council of Medical Research COVID-19 timeline and limited data on variant seroprevalence trends, we divided the COVID-19 pandemic into six time periods: lockdown, first wave, low cases (winter 2020/2021), delta wave, low cases (fall 2021) and omicron wave (figure 1).31 32
Trajectories in pandemic markers over the course of the COVID-19 pandemic during the time frame of data collection for the Real-Time Insights of COVID-19 in India study. All pandemic markers are standardised to z-scores to facilitate comparisons in trajectories over time. All metrics are reported or estimated at the state level; overall levels shown here are weighted based on the distribution of participants across the Indian states and territories. (A) Current values of pandemic markers. (B) Mean pandemic markers over the time periods used for analysis (3 months of case and death indicators, 1 month for government stringency index).
We used prior household infection as an indicator of the direct impact and experience of COVID-19 infection. In constructing this indicator, we considered information from reports of diagnosed infections from any household member in waves 2–8, reports of confirmed or suspected infections from any household member in waves 7–8, and dates of reported or suspected household infections from a single female household member in wave 9. Finally, we used data on self-reported financial impact of the pandemic (no impact/minor impact/moderate impact/major impact/too soon to tell) reported by a male household member to assess the effect of perceived financial distress.
Covariates and effect modifiers
We considered self-reported age, gender and educational attainment (none/less than secondary/secondary/some graduate) from survey data as well as rural/urban residence as covariates. We also calculated prepandemic per-capita household consumption based on reports of food, non-food and healthcare expenditures in the past year. We used quintiles of prepandemic per-capita consumption as a covariate in models.
Statistical analysis
We used descriptive statistics (proportions and numbers for binary variables, means and IQRs for continuous variables) to characterise included covariates and pandemic markers over the six pandemic periods and nine waves considered in analyses. We also visualised trajectories of PHQ-4 scores over time. To assess the association between each pandemic indicator and mental health, we used linear regression models estimated with generalised estimated equations (GEEs) with an exchangeable correlation structure to account for correlation attributable to the inclusion of multiple surveys for a given respondent. For each indicator, we estimated unadjusted models and models adjusted for age (estimated with a natural cubic spline with 2 internal knots (33rd and 66th percentiles) and 2 external knots (5th and 95th percentile)), gender, educational attainment, per capita household consumption quintile and rural/urban residence. To understand individual contributions of each pandemic indicator while adjusting for other indicators, we estimated a single GEE model including covariates as well as all pandemic characteristics except time period and administrative cases and deaths. We chose to use estimated rather than reported cases and deaths in this single model to avoid bias due to under-reporting, particularly in later stages of the pandemic.
Finally, we assessed the effect modification of the associations between each pandemic characteristic and mental health by age category (<30/30–44/45–59/60–74/75+), gender and rural/urban residence. Stratified results to visualise effect modification were estimated using linear combinations of coefficients from GEE models including an interaction between the pandemic characteristic and effect modifiers of interest. Models controlled for the same covariates used in previous models but excluded any covariate considered as an effect modifier. χ2 tests comparing models with an interaction to models without an interaction were used to test the overall statistical significance of effect modification. In sensitivity analyses, we re-estimated all models described above using a binary indicator for abnormal PHQ score (>2). For binary models, we used Poisson regression with robust variance to estimate prevalence ratios. We also re-estimated all models using estimated factor scores for poor mental health from item response theory models.
All descriptive statistics used survey weights to correct for sampling processes and selection bias; we present unweighted regression models in main analyses and weighted regression models in sensitivity analyses. Item response theory models were estimated in Mplus V.8; all other models were estimated in R V.4.2.2.
Patient and public involvement
Respondents were not actively involved in the design of this study or the analysis of data. However, during the consent process, respondents were informed of how their participation would enable scientific research across a wide range of topics. The authorship team is composed primarily of academic researchers at the University of Southern California and researchers from partner institutions in India, who have collaborated closely in the design and implementation of the RTI-COVID India survey.
Results
We included data from 3709 respondents collected across 18 844 phone surveys. Demographics and characteristics, including age, gender, educational attainment, per capita consumption quintile and rural/urban residence were stable across wave and pandemic period (table 1, online supplemental appendix tables 1,2). Although reported cases were similar between the delta and omicron waves, estimated cases were higher during the omicron wave. Reported and estimated deaths were higher during the delta wave. Government stringency index was highest during the lockdown phase, while self-reported financial impact did not vary substantially over time. PHQ-4 score varied over time and was highest during the lockdown phase and at the end of the delta wave (table 1, figure 2). During the pandemic, a sizeable portion of participants had at least mild mental health symptoms (varying from 22.2% to 39.4%) or moderate mental health symptoms (varying from 5.1% to 14.0%). A smaller (varying from 1.0% to 4.0%) but important group had severe symptoms. The proportions of individuals in each of these categories also varied across pandemic periods, and changes over time are reflected in shifts in the continuous PHQ-4 score. All age groups had similarly high PHQ-4 scores in the initial phases of the pandemic but in later stages of the pandemic, PHQ-4 sores were higher among the older age groups (figure 2).
Characteristics of the pandemic and the Real-Time Insights of COVID-19 in India sample (N=3709) across the 6 pandemic time periods
Estimated smoothed trajectories of PHQ4 scores for participants in the Real-Time Insights of COVID-19 in India sample (N=3709) over the course of the pandemic, overall (A) and by age category (B). Trajectories were fit with generalised additive models with flexible penalised cubic splines. Grey shaded regions reflect time periods with ongoing data collection. PHQ-4, Patient Health Questionnaire-4.
Crude and adjusted associations between pandemic periods from regression models were similar and suggested that mean PHQ-4 scores in all subsequent pandemic periods were lower than mean PHQ-4 score during the lockdown phase (figure 3). However, there was variation over time, and in adjusted models, mean PHQ-4 score during the delta wave was only slightly lower than during the lockdown period (difference of −0.18, 95% CI −0.31 to −0.06 units). PHQ-4 scores were also higher during the omicron wave (difference of −0.16, 95% CI −0.30 to −0.02 units compared with lockdown) as compared with the period immediately following the initial lockdown period (first wave (−0.65, 95% CI −0.75 to −0.54) and winter 2021 (−0.64, 95% CI −0.74 to −0.55)).
Associations between COVID-19 pandemic periods and COVID-19 pandemic markers and PHQ-4 scores in the Real-Time Insights of COVID-19 in India sample (N=3709). Associations with all continuous markers (cases, deaths, government stringency index) are shown as associations for 1 SD unit difference. Adjusted models are adjusted for age (spline), gender, educational attainment, per capita consumption quintile and rural/urban residence. PHQ-4, Patient Health Questionnaire-4; IHME, Institute for Health Metrics and Evaluation.
Estimated case counts were positively associated with PHQ-4 score, such that a 1 SD higher estimated case count was associated with 0.07 (95% CI 0.04 to 0.11) unit higher PHQ-4 score (figure 3). The estimated association between reported case counts and PHQ-4 score was larger (0.15 (95% CI 0.10 to 0.19)), likely because cases were more likely to be under-reported during the later stages of the pandemic when the disease was less severe. Reported and estimated deaths as well as the government stringency index were positively associated with PHQ-4 score; mean effect sizes ranged from 0.10 to 0.15 units. The estimated association between mental health and prior household infection (0.36; 95% CI 0.23 to 0.50) was larger than the estimated association between mental health and contextual pandemic markers. The largest estimated association was observed for self-reported financial impact; compared with those reporting no impact, those reporting a major financial impact had 1.20 (95% CI 1.09 to 1.32) unit higher PHQ-4 scores. Individual associations for pandemic markers remained stable when estimated in a single model (online supplemental appendix table 3), indicating that associations between each pandemic marker and PHQ-4 score were largely independent.
Age was an important effect modifier of the association between pandemic period and pandemic markers and PHQ-4 scores (figure 4); adding interaction terms significantly improved model fit for all pandemic markers (p<0.001) (online supplemental appendix table 4). PHQ-4 scores were more evenly elevated over the course of the entire pandemic in respondents 45 years of age and older, whereas in younger respondents (under 44 years of age), PHQ-4 scores dropped and stayed lower after the initial lockdown period (figure 4, online supplemental appendix table 4). Associations between PHQ-4 scores and pandemic markers including self-reported financial impact, cases, deaths and prior household infection were stronger among older respondents (for all: p<0.001). However, the opposite was true for government stringency index. Gender differences were largest for pandemic period (p=0.004) and financial impact (p<0.001); there were stronger associations between the lockdown period and financial effects of the pandemic for men compared with women. PHQ-4 score was also more strongly associated with later pandemic periods (omicron and delta waves) (p=0.055), and the association with estimated deaths was stronger (p=0.002) in urban compared with rural areas.
Effect modification of the association between COVID-19 pandemic periods and COVID-19 pandemic markers and PHQ-4 scores by age group in the Real-Time Insights of COVID-19 in India sample (N=3709). All models are adjusted for gender, educational attainment, per capita consumption quintile and rural/urban residence. PHQ-4, Patient Health Questionnaire-4; IHME, Institute for Health Metrics and Evaluation.
Results from models using the binary outcome of abnormal PHQ-4 score (>2) showed similar findings, though the effects of later pandemic waves were stronger than in primary models; the probability of abnormal PHQ-4 scores during the delta and omicron waves was not significantly different from lockdown (online supplemental appendix figures 2 and 3), (online supplemental appendix table 5). Results from models using estimated factor scores for mental health were consistent with findings from main analytic models (online supplemental appendix figures 4 and 5), (online supplemental appendix table 6). Results from weighted regression models were closer to results reported for younger participants, as the RTI COVID-India study oversampled older adults compared with the general population of India (online supplemental appendix figures 6 and 7), (online supplemental appendix table 7).
Discussion
In a nationwide sample from India, mental health symptoms fluctuated over the course of the pandemic. High mental health symptom burden was especially apparent during the lockdown period, but symptom burden was also high at later points in the pandemic in response to COVID-19 waves (omicron, delta), particularly among older respondents (aged 45 and older). Estimated and reported cases, estimated and reported deaths, government stringency, self-reported financial burden, and reported infection in the household were all associated with poorer mental health.
Our results on mental health during the lockdown period align with a large number of studies that have reported high mental health symptom burden in the initial lockdown stage of the pandemic in India.33 34 The literature on mental health outcomes in later stages of the pandemic in India is more scarce and somewhat mixed; studies have reported both increasing and decreasing symptom burden over time.26 35 36 Our findings suggest that such inconsistent evidence could be due to differences in the timing of survey waves and longitudinal follow-ups; we found both increasing and decreasing symptom burden during different time periods. The number of assessments and length of follow-up time allowed us to link subsequent waves of COVID-19 and the impacts of COVID-19 (eg, deaths) to mental health outcomes. In other contexts, including Australia, Poland, the USA and the UK, prior studies have also found evidence of poorer mental health outcomes in response to later COVID-19 waves, in line with our findings.20 37–39 While we focused on trends in mean mental health outcomes at various stages of the pandemic, previous studies have also found that some people may suffer from severe and persistent poor mental health outcomes as a consequence of exposure to acute emergency environments such as the lockdown phase or delta and omicron waves.40 41 Identification, characterisation and analysis focusing on this subgroup are beyond the scope of this present paper but should be the topic of future research.
We were also able to assess correlates of poor mental health and compare the relative magnitude of associations across different pandemic attributes. Significant associations across all attributes considered suggests that such characteristics could help policy-makers target time periods or regions with greater need for mental health supports and services. Although the magnitude of some estimated coefficients was small in SD units of reported or estimated cases and deaths (eg, difference of 0.07 95% CI 0.04 to 0.11 in PHQ-4 score per SD increase in estimated cases), these health burden indices abruptly changed by 2–5 SD during the omicron and delta waves; therefore, larger changes in mental health would be expected with the sudden onset of new waves of infections. The summation of these smaller effects may account for the larger effects observed for specific time periods; for example, comparing PHQ-4 scores during the delta wave to those observed immediately following the lockdown period (difference of 0.46, 95% CI 0.34 to 0.58 units). Furthermore, the clinical importance of findings is underscored by the consistency of results for the binary outcome of abnormal PHQ-4 score (>2).
Similarly to other studies across contexts, we found that financial stress was strongly associated with poor mental health.42–44 In our study, perceived pandemic-induced financial burden had the strongest associations with poor mental health among all the pandemic characteristics considered, highlighting the importance of economic conditions. This finding aligns with prior work suggesting that perceived financial burden is an important mediator of the effects of job and income loss on mental health.45 Although policies affecting financial stress were largely concentrated in the early lockdown period, our data indicated that perceived financial stress was high over the course of the COVID-19 pandemic in India. Financial stress may have larger consequences in settings such as India due to higher levels of poverty and food insecurity compared with other settings.46 Future work should consider objective measures of economic shocks in addition to the self-report measure considered here.
Although the financial effects of government lockdown policies have been previously emphasised,47 we found that the association between government stringency index and poor mental health outcomes was independent of perceived financial stress. The independence of the effects of government stringency and financial distress suggests that social and psychosocial factors impacted by lockdown, including loneliness and fear, may also play an important role.9 48 Prior cross-national research has also found an association between policy stringency and poor mental health.49 However, other work suggests that prompt implementation of restrictions with clear messaging may actually be beneficial.50 Considered together, this evidence underscores the importance of effective and quick communication and action. In contrast, delayed or draw-out lockdown periods with poor communication, as was experienced in India,51 are more likely to have negative impacts on mental health. Additionally, in our study, associations between policy stringency and mental health were strongest for younger participants and were attenuated in older adults, who, given their higher degree of vulnerability, may feel more protected in more controlled environments.
We also found that increases in reported and estimated cases and deaths, and COVID-19 infection in the household were associated with poorer mental health. Prior research in the USA has also documented associations between reported cases and worse mental health outcomes, however, this work did not compare associations for reported versus estimated infections.52 In our study, the association between COVID-19 deaths and mental health was stronger in urban areas, where mass open-air funeral pyres and higher population density may have magnified fear and awareness of potential negative consequences of COVID-19.53 Overall, results indicate that the presence of severe disease likely has negative effects on mental health, which should be considered when quantifying and summarising the overall impact of COVID-19.
Differences in trajectories of mental health and associations between pandemic characteristics and mental health across age groups point to differences in the experience of the pandemic. While mental health symptoms were elevated across all age groups during the initial lockdown phase, during later stages of the pandemic, elevated symptoms were observed primarily among adults 45 years and older. In comparison, among younger adults, mental health symptoms improved after the lockdown phase, with minor fluctuations later in the pandemic. Furthermore, associations between measures of pandemic burden (cases, deaths) were also stronger among those 45 years and older, while they were largely null among younger adults. Prior evidence comparing the prepandemic and lockdown periods indicated that the negative effects of the pandemic on mental health were larger for younger adults.14 16 However, based on this study, conclusions that older adults were able to cope with the COVID-19 pandemic better than younger adults may have been premature.54 Instead, current evidence from India suggests that the mental health impacts of pandemic burden (household infections, deaths) after the initial lockdown phase may be larger among older adults. Older adults may have been more affected by these later waves of the pandemic given their increased vulnerability to the health impacts of COVID-19.55
Taken together, our results have important policy implications. Although the lockdown period had negative consequences for mental health,4–8 in this study, we show that subsequent waves of uncontrolled disease spread also had negative impacts on mental health for middle-aged and older adults. These negative consequences should be considered when evaluating the implementation of policy-based approaches to managing the COVID-19 pandemic or other future pandemics. Furthermore, the observed mental health consequences of the COVID-19 pandemic on middle-aged and older adults highlights the importance of issues such as the destigmatisation of mental health issues and the promotion of policies to ensure the availability of mental health supports and services during future public emergencies. Additionally, both an understanding of the associations between different contextual factors (eg, cases, self-reported financial impact) and mental health outcomes and an understanding of effect modifiers such as age, gender and urban/rural status can help policy-makers with resource allocation and targeted interventions. Ultimately, these lessons likely extend beyond the COVID-19 pandemic to future pandemics or other acute stressors or shocks that contribute to poor mental health.
This study is the first large-scale nationwide effort to assess mental health throughout the COVID-19 pandemic in India. Because of the design, scale and setting of the study, we were able to disentangle associations between a variety of different pandemic characteristics and poor mental health. In the context of study strengths, limitations should be considered. First, the implementation of telephone assessments may have introduced selection bias. Although India has high rates of mobile phone ownership (93% of households), there are gaps by gender and socioeconomic status.56 To account for this potential bias, we used survey weights adjusted for non-response in sensitivity analyses; weight construction included a poststratification raking algorithm to ensure the weighted sample matched national population demographics. Second, due to concerns about participant burden, we were unable to implement detailed measures of mental health across all time points. Instead, we used the PHQ-4, which is commonly used as a screening tool for anxiety and depression. The use of this brief measure limits the precision and interpretation of findings. The scale only includes four items and is not diagnostic, generally serving as an indicator for further inquiry in clinical settings. However, in sensitivity analyses, we used item-response theory methods to estimate latent mental health with maximal precision, and patterns of results were consistent in models using these estimated scores, as well as in models assessing trajectories and correlates of reporting abnormal PHQ-4 scores (>2). Despite its brevity, the PHQ-4 correlates strongly with more comprehensive measures of mental health and shows strong criterion and construct validity.29 Third, attrition across study waves may bias estimates. However, analysis of reasons for drop-out indicates that only 27% of observed household-level drop-out was for reasons identified as having the highest likelihood of being associated with mental health; 24% and 25% of observed drop-out were for reasons that had either moderate or less potential to be related to mental health, respectively (additional details in online supplemental appendix). Furthermore, we hypothesise that those who attrit would be more likely to have worse mental health outcomes in response to pandemic stressors, which would make our results conservative. Because we found significant effects across a range of pandemic-related characteristics, we believe that any bias would be unlikely to impact the interpretation of results or the overall conclusions. Fourth, we did not consider the impact of the General Elections, which occurred in spring 2021 in four states and one union territory. However, the elections may be an effect modifier of observed results if they altered the context and the environment during the pandemic. This should be a topic of future research. Finally, we were unable to examine the impacts of the vaccination campaign on mental health because of the strong correlation in time between the delta wave and vaccination; other contexts may be better suited to tease out these associations.
Overall, this study highlighted the complexity of mental health trajectories over the course of the COVID-19 pandemic in India. The initial lockdown period was associated with a high burden of mental health symptoms, but subsequent pandemic waves and associated mortality or household infections were also associated with worsening mental health, particularly among middle-aged and older adults. Results suggest that the quantification of the consequences of COVID-19 throughout the pandemic should include the mental health impacts of COVID-19 in both the early and later stages of the pandemic. Our findings further highlight the importance of continued attention to mental health in the context of COVID-19 and future emerging pandemics and support the need for ongoing services and interventions aimed at bolstering mental health in the general population.
Data availability statement
Data are available in a public, open access repository. Data are available to download via the Gateway to Global Aging website: covid.g2aging.org. Users must sign a data use agreement before being granted access to the data.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by Institutional Review Boards at the University of Southern California (study number UP-20-00277) and the All India Institute of Medical Sciences (study number RP-29/2020). Participants gave informed consent to participate in the study before taking part.
Acknowledgments
We acknowledge all members of the study team who made data collection possible. We also acknowledge the contribution of all participants who consented to take part in this study during the COVID-19 pandemic.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Seye Abimbola
Twitter @enichols94
Contributors EN wrote the statistical analysis plan, analysed the data, and drafted and revised the paper. She is guarantor. SP managed data collection and data monitoring, and revised the paper. PK, JB and SD designed data collection tools, implemented data collection, monitored data collection and revised the paper. MA and SS designed data collection tools, monitored the data, contributed to analysis and revised the paper. DEB, ABD and JL secured study funding, designed data collection tools, supervised all data collection and revised the paper.
Funding National Institute on Aging, National Institutes of Health (U01AG065958, R01AG051125, R01 AG030153).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.