Skip to main content
NCBI home page
UKPMC Funders Author Manuscripts logoLink to UKPMC Funders Author Manuscripts
. Author manuscript; available in PMC: 2026 Jun 16.
Published in final edited form as: Soc Sci Med. 2025 Nov 21;389:118830. doi: 10.1016/j.socscimed.2025.118830

COVID-19, economic downturn, and long-term trajectories of population mental health: Evidence from two nationally representative British birth cohorts at the intersection of gender and socioeconomic position

Darío Moreno-Agostino a,b,*, George B Ploubidis a,b,#, Jayati Das-Munshi b,c,#
PMCID: PMC7619173  EMSID: EMS214121  PMID: 41313899

Abstract

Background

We examined long-term trajectories of mental (ill-)health in two British generations (‘Baby Boomers’ and ‘Generation X’) across the life-course, including the COVID-19 lockdowns and the following cost-of-living increases. We analysed inequalities by generation, gender, socioeconomic position (SEP), and their intersections, and explored the relationship between inflation and mental (ill-)health post-lockdown.

Methods and findings

We used data from the National Child Development Study (NCDS/1958, n = 8215) and the 1970 British Cohort Study (BCS/1970, n = 7789), with repeated measures of psychological distress (Malaise Inventory) between ages 23–64.5 (NCDS/58) and 26–52.5 (BCS/70). We used multilevel growth curve models to study long-term trajectories, and negative binomial regression models to analyse associations with inflation/cost-of-living in the 2021–2023 period. Distress increased during the pandemic but declined post-lockdown (second quadratic spline: BNCDS/58 = −0.12 [-0.17, −0.08], p < 0.001; BBCS/70 = −0.16 [-0.21, −0.11], p < 0.001). Women and individuals from disadvantaged childhood SEPs started their trajectories with significantly (p < 0.001) higher distress levels in both cohorts (women: BNCDS/58 = 0.72 [0.62, 0.82], BBCS/70 = 0.73 [0.62, 0.83]; manual-class background: BNCDS/58 = 0.24 [0.14, 0.35], BBCS/70 = 0.23 [0.12, 0.35]; rented housing: BNCDS/58 = 0.34 [0.22, 0.46], BBCS/70 = 0.30 [0.15, 0.45]). Inequalities were larger for women from disadvantaged SEPs born in 1958, indicating intersectional effects. None of these inequalities significantly reduced in the long term. Inflation/cost-of-living was significantly associated with distress, but effects did not vary by gender, concurrent SEP, or their intersection.

Conclusions

Despite post-pandemic improvements, persistent inequalities by gender and childhood SEP remain. Considering the high levels of socioeconomic adversity in the UK, action is needed to reduce these inequalities and prevent their transmission across generations.

1. Introduction

The global COVID-19 pandemic came with a deterioration in population mental health, with disproportionate impacts among disadvantaged groups (Gibson et al., 2021; Sun et al., 2023). In the UK context, this global shock disrupted the long-term trajectories of mental health, widening pre-existing inequalities such as those by gender (Moreno-Agostino et al., 2023, 2024; Patel et al., 2022). In many countries, this period was followed by a rapid increase in the cost of living, with global consumer prices reaching levels unseen since the 2008 financial crisis (World Bank, 2025). In the UK, similar inflation levels had not been seen since 1990, with the Consumer Prices Index including owner occupiers’ housing costs (CPIH) reaching a peak of 9.6 % in October 2022 (Office for National Statistics, 2025). However, there is no population-based, longitudinal evidence on how long-term trajectories of mental (ill-)health have developed after the COVID-19 lockdowns, a period largely coinciding with large cost-of-living increases, and whether some of the inequalities that widened during the lockdown period have changed.

Economic shocks can impact population mental health, particularly among those socioeconomically disadvantaged (Broadbent et al., 2023; England et al., 2024; Talamonti et al., 2023). From a long-term perspective, socioeconomic adversity plays a fundamental and complex role on mental (ill-)health across the life course and even across life courses (e.g., through intergenerational transmission) (Kirkbride et al., 2024; Link and Phelan, 1995). Even early life socioeconomic disadvantage, over which people have little to no agency, can impact mental health later in life and lead to further socioeconomic adversity (Ridley et al., 2020). Despite being one of the wealthiest economies globally, levels of poverty have remained consistently high in the UK (Francis-Devine, 2024), inextricably related to decades of austerity measures (Ruckert and Labonte, 2017; Stuckler et al., 2017). This has led some scholars to propose that ‘poverty pandemic’ may be a more accurate term than the widely used ‘cost-of-living crisis’ (Morris et al., 2023), as the situation seems “endemic, not a short-term crisis to weather” (Haggar et al., 2023). Studying the trajectories of mental (ill-) health in the population from a life-course perspective can help to understand the potential role of the cost-of-living increases both in the context of other recent (e.g., COVID-19 pandemic) and longer-term (e.g., early life socioeconomic disadvantage) phenomena.

When studying how population mental health levels may have changed after the lockdown period, there are additional sources of complexity to consider. First, understanding whether and how the large (and, during the pandemic, widening (Moreno-Agostino et al., 2023; Patel et al., 2022)) gender inequalities in mental (ill-)health have changed is key to grasp the extent of the action needed to close them. Importantly, the systems of oppression that underlie the gender and socioeconomic inequalities in mental health (including sexism, the material impacts of socioeconomic deprivation, and classism) are complex and interlocked, rather than independent from each other. This idea is central to intersectionality theories (Bauer, 2014; Bowleg, 2008). Importantly, inequalities at specific intersections will remain undetected unless explicitly acknowledged when quantitatively analysing social inequalities (Bowleg and Bauer, 2016). Second, there is compelling evidence of a deterioration in multiple health outcomes across generations in the UK, or a ‘generational health drift’ (Gimeno et al., 2024, 2025). This includes mental (ill-)health outcomes, with younger generations (particularly those born in 1970s and onwards) experiencing worse mental health levels than older generations at similar or same ages (Armitage et al., 2023; McElroy et al., 2023; Moreno-Agostino et al., 2023; Patalay and Gage, 2019). People from different generations have lived through different socio-historical contexts in which power and privilege are differently distributed (Collins and Bilge, 2020). For example, the percentage of women aged 25–54 in paid employment (including self-employment) went from 57 % in 1975 to 78 % in 2017 (Roantree and Vira, 2018). And, although it is far from being resolved, the percentual gender pay gap among all employees has slowly reduced over time, from 27.5 % in 1997 to 14.4 % in 2022 (Office for National Statistics, 2024). The Equal Pay Act, a key milestone for the legal basis of gender pay equality, was introduced in 1970, with amendments and further legislation being introduced in the 80s, 90s, and most recently consolidated in the Equality Act 2010 (Francis-Devine and Ferguson (2020). By taking place at different times in their lifespan, societal conditions and events like these can have different implications for, in this example, women and men of different generations.

Considering all the above, studying the long-term trajectories of mental (ill-)health from a life-course, cross-generational, and intersectional perspective has multiple advantages. First, it can inform whether the observed increasing trends in psychological distress observed during the COVID-19 pandemic have continued, stopped, or reversed as populations exited the lockdown periods. Second, it can inform how social inequalities observed during the COVID-19 pandemic (e.g., widening gender inequalities) have evolved in the post-lockdown era. Third, it can provide further insights into how the generational decline (or ‘generational health drift’ (Gimeno et al., 2024)) in mental health is evolving. And fourth, when coupling this cross-generational approach with a focus on early life determinants of long-term trajectories, it can serve as a measure of societal progress, by informing whether certain social inequalities have meaningfully reduced across generations and in the long term.

Therefore, the two primary aims of this study were #1) to understand how long-term trajectories of mental (ill-)health developed in post-lockdown Britain, a period largely coinciding with cost-of-living increases; and #2) to examine differences in those long-term trajectories by generation, gender, childhood socioeconomic position, and their intersections. As a secondary aim, we also aimed #3) to examine the relationship between inflation and population mental (ill-)health, and whether this relationship varied by gender, socioeconomic position, and their intersections.

2. Methods

2.1. Sample

We used data from two British birth cohorts: the 1958 National Child Development Study (NCDS/58) (Power and Elliott, 2006) and the 1970 British Cohort Study (BCS/70) (Sullivan et al., 2022). These are two nationally representative birth cohort studies following up the lives of people born in Britain in a single week in 1958 and 1970, respectively. Data on the cohort members have been collected since their birth and throughout their life courses. The most recent main survey data collections (sweeps) took place between February 2019–April 2024 in NCDS/58 and between January 2020–January 2024. As such, they started before the COVID-19 pandemic, were interrupted during the lockdown period, and resumed after the lockdowns. However, data collection continued during the pandemic as part of the COVID-19 Surveys (Brown et al., 2021), which collected data from members of these and other cohorts at three time-points during the COVID-19 pandemic: May 2020 (during first national lockdown), September–October 2020 (before second national lockdown), and February–March 2021 (during third national lockdown) (Institute for Government, 2021). A total of 8215 (NCDS/58) and 7789 (BCS/70) people took part in the pilots, dress rehearsals, main stage, or mop-up web surveys in the most recent data collection, corresponding to response rates of 73.3 % and 65.3 %, respectively. The target population for the present study were adults born in Britain in 1958 or 1970, still alive and residing in the UK during the most recent main survey sweep. All procedures involving human subjects/patients were approved by the National Health Service (NHS) Research ethics Committee. All participants provided oral informed consent.

2.2. Measures

2.2.1. Main outcome: psychological distress

Psychological distress (a measure of mental ill-health including depressive and anxiety symptomatology) was the main outcome in both the within-person long-term trajectory and between-person post-lockdown analyses. It was measured with the nine-item version of the Malaise Inventory (Rodgers et al., 1999). This questionnaire explores whether the respondent ‘often’ experiences a series of general mental ill-health experiences, with “yes/no” response options. Hence, the sum-score ranges between 0 (lowest psychological distress) and 9 (highest psychological distress) at any time-point. The questionnaire was administered at ages 23, 33, 42, 50, 62, 62.5, 63, and 64.5 in NCDS/58 and at ages 26, 29, 34, 42, 46, 50, 50.5, 51, and 52.5 in BCS/70. Previous studies have provided evidence on the appropriateness of a sum-score approach and on the invariance of the resulting measure across time-points, genders, and cohorts (Gondek et al., 2022; Moreno-Agostino et al., 2023; Ploubidis et al., 2019). Due to the aims of this study and the use of additional newly collected data, we extended this previous evidence by including the most recent data collection time-points and by analysing the measurement invariance at the intersection of gender and socioeconomic position, both within and across cohorts (an often overlooked aspects in quantitative research on intersectional inequalities (Else-Quest and Hyde, 2016)). Additional details on this approach are available in eAppendix 1 (Supplementary Material).

2.2.2. Social identities and positions

We used an inter-categorical approach to intersectional complexity (McCall, 2005), provisionally adopting categorical variables as proxies of the systems of oppression that may lead to inequities in the outcomes under study.

We used sex assigned at birth as a proxy for the gender system. Our position is that differences between women and men in the outcome under study will be more likely due to differences in power and privilege–in other words, to sexism–than due to inherent biological characteristics. Therefore, we refer to gender rather than sex inequalities throughout.

We used two alternative childhood socioeconomic position indicators as proxies for social class and classism, harmonised across NCDS/58 and BCS/70: parental social class and childhood housing tenure. A six-category harmonised variable representing the cohort member’s father’s social class at ages 11 (NCDS/58) or 10 (BCS/70) was obtained from the UK Data Service (UKDS) based on the work by Dodgeon et al. (2019). This was further dichotomised into manual (including the skilled manual, partly skilled, and unskilled categories) and non-manual (including the skilled non-manual, managerial and technical, and professional categories) for the purposes of this study. A three-category harmonised variable capturing whether the cohort member lived in an accommodation that was rented, owned at one time-point, or owned at both time-points at ages 7&11 (NCDS/58) or 5&10 (BCS/70) was obtained from the UKDS based on the work by Wood et al. (2019).

2.2.3. Confounders

As ‘time’ was the main exposure in the long-term trajectory analyses, we considered it to be unconfounded in itself as nothing would cause both time and the outcome under study. Although multiple processes unfold with time, which may explain changes in the outcome, the aims of the long-term trajectory analyses were to describe the trajectories and differences in these across groups, rather than to identify the underlying mechanisms that may explain them.

2.3. Statistical analysis

2.3.1. Within-person long-term trajectory analyses

Multilevel growth curve models (Bryk and Raudenbush, 1987; Curran et al., 2010) were used to understand the change in psychological distress over time, including the period with large cost-of-living increases (aim #1), and differences in these changes by generation, gender, childhood socioeconomic groups, and their intersections (aim #2).

Cohort members whose participation in the most recent main survey sweep took place before the COVID-19 pandemic (1662 in NCDS/58 and 116 in BCS/70) or within the period spanning the COVID-19 Surveys data collection (44 participants from BCS/70 whose interviews took place between September 11, 2020 and October 7, 2020) were not included in these analyses. Age at the most recent sweep was set at the median weighted age among included cohort members, which was 64.5 in NCDS/58 and 52.5 in BCS/70 (roughly corresponding to the second half of 2022). Therefore, the resulting repeated measures of psychological distress spanned ages 23, 33, 42, 50, 62, 62.5, 63, and 64.5 in NCDS/58 and ages 26, 29, 34, 42, 46, 50, 50.5, 51, and 52.5 in BCS/70.

Models with different growth parameters and random effects (to accommodate individual variation in the included growth parameters) were tested and selected upon the best model fit according to Akaike’s Information Criteria (AIC) and Bayesian Information Criteria (BIC), with lower AIC and BIC indicating better fit. Candidate models were selected based upon the descriptive analysis and visualisation of imputed and weighted life-course data. The selected models were then estimated in each cohort, first overall, then by gender and by each of the childhood socioeconomic position indicators, and finally by the intersection of gender and each of the childhood socioeconomic position indicators, by including the appropriate interaction terms between the growth parameters and the group variables. To aid with the interpretation of the results, marginal predicted levels at each of the time-points (and for each of the groups in the analyses by gender, childhood socioeconomic position, and their intersection) were obtained and visualised.

As a check of the sensitivity of the results to model specification and sample selection, a set of additional overall models were estimated where continuous age was included instead of median age as the main time variable at the latest sweep. These analyses included all participants regardless of whether their most recent main survey sweep interview had taken place before/during the COVID-19 pandemic.

To more specifically characterise any potential change in the gender, socioeconomic, and intersectional inequalities over time and across cohorts, an additional set of analyses were conducted comparing the psychological distress levels in the most recent sweep (roughly, as abovementioned, the second half of 2022) with three relevant previous time-points: 1) the earliest time-point in the long-term trajectory (age 23 in NCDS/58 or 26 in BCS/70; 1981 and 1996, respectively), 2) the most recent pre-pandemic assessment (age 50 in NCDS/58 and 46 in BCS/70; 2008 and 2016, respectively), and 3) the point of highest psychological distress during the COVID-19 pandemic (age 62.5 in NCDS/58 or 50.5 in BCS/70; September-October 2020). These analyses were conducted separately for each cohort and, then, pooling together the data from the two cohorts, adding cohort as a main and interaction variable with the existing terms to explore any cohort differences in the change across time-points, social inequalities, and their potential change over time.

2.3.2. Between-person post-lockdown analyses

To provide further insights on the potential shorter-term relationship between inflation and mental (ill-)health both in general and by pre-existing sources of disadvantage (aim #3), a separate group of analyses were conducted focusing on the post-lockdown period and leveraging the time variability in the data collection. Additional details on the rationale for these analyses, as well as the measurement and analytical approach used, are available in eAppendix 2 (Supplementary Material).

2.3.3. Missing data

Inverse probability weighting (IPW) and multiple imputation by chained equations (MICE) were used to deal with missing data, both under the assumption that data were missing at random (MAR) after conditioning on observed variables (Enders, 2023).

Non-response weights were derived and used to help restore sample representativeness to the target population (i.e., people born in Britain in 1958 or 1970, still alive and residing in the UK at the time of the most recent sweep), using relevant predictors of non-response from across the individuals’ life-courses (Silverwood et al., 2024). Full details on the derivation of the weights and their effectiveness to restore representativeness to the target population are available in the cohort studies’ user guides (Brown et al., 2025; Sedovic et al., 2024).

Multiple Imputation by Chained Equations (MICE) was used to flexibly deal with item-missingness (White et al., 2011). To increase the plausibility of the MAR assumption, we used cohort members’ prior life-course data which has been found to be related to attrition and/or to the missing values themselves as auxiliary variables, in line with previous research (Ploubidis et al., 2021) and with the Centre for Longitudinal Studies (CLS) missing data strategy (Silverwood et al., 2024). The complete list of auxiliary variables included in the imputation models is available in eAppendix 3 (Supplementary Material). Fifty imputed datasets were created, discarding the first 10 iterations of each chain. The analytical models were then conducted over the imputed datasets using Rubin’s rules to pool the estimates and standard errors (Enders, 2023).

All data used in this study are freely available to bona fide researchers at the UK Data Service (https://doi.org/10.5255/UKDA-Series-200001 and https://doi.org/10.5255/UKDA-Series-2000032). All code used for the data management, imputation, and analysis is available at the Open Science Framework (https://doi.org/10.17605/OSF.IO/MSPRQ).

We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist (eAppendix 5, Supplementary Material).

All analyses were conducted in Stata MP 19.5 (StataCorp, 2025), except measurement invariance testing, which was conducted in Mplus 8.9 (Muthén & Muthén, 2017).

3. Results

The overall sample for this study included 6553 participants from the NCDS/58 cohort (n = 3,277, 50.0 % women) and 7629 participants from the BCS/70 cohort (n = 4,015, 52.6 % women), after excluding cohort members whose latest main survey sweep took place before the COVID-19 pandemic onset (n = 1662 in NCDS/58 and n = 116 in BCS/70) or within the period spanning the COVID-19 Surveys data collection (n = 44 in BCS/70, interviewed between September–October 2020). Included participants contributed a median and interquartile range of 7 (5, 8) repeated observations in both cohorts (MNCDS/58 = 6.34, MBCS/70 = 6.56). The percentage of participants in a disadvantaged childhood socioeconomic position was, overall, smaller in BCS/70 (21.9 % rented at both time points, 47.4 % manual parental social class) compared to NCDS/58 (36.9 % and 51.8 %, respectively). Similar percentages of missing data were found in the socioeconomic position indicators across both cohorts, although this was slightly higher for the outcome variables in BCS/70 than in NCDS/58. Further descriptive details of the analytical samples, along with the percentage of missing data for the key variables of interest, are available in Table 1, and an extended table including a comparison with the overall samples of each cohort study is available in eAppendix 6 (Supplementary Material).

Table 1. Descriptive information from the analytical samples.

NCDS/58,
N = 6553		 BCS/70,
N = 7629
Gender, N (%)
    Women 3277 50.0 % 4015 52.6 %
    Men 3276 50.0 % 3614 47.4 %
Childhood SEP, N (%)
Parental social class
    Manual 3394 51.8 % 3618 47.4 %
    Non-manual 2116 32.3 % 2864 37.5 %
    Missing 1043 15.9 % 1147 15.0 %
Childhood housing tenure
    Owned at both time points 2250 34.3 % 3418 44.8 %
    Owned at one time point 414 6.3 % 565 7.4 %
    Rented at both time points 2419 36.9 % 1667 21.9 %
    Missing 1470 22.4 % 1979 25.9 %
Gender and childhood SEP, N (%)
Gender * parental social class
    Women * Manual 1699 25.9 % 1898 24.9 %
    Women * Non-manual 1048 16.0 % 1529 20.0 %
    Women * missing 530 8.1 % 588 7.7 %
    Men * Manual 1695 25.9 % 1720 22.5 %
    Men * Non-manual 1068 16.3 % 1335 17.5 %
    Men * missing 513 7.8 % 559 7.3 %
Gender * childhood housing tenure
    Women * owned at both time points 1093 16.7 % 1819 23.8 %
    Women * owned at one time points 218 3.3 % 312 4.1 %
    Women * rented at both time points 1239 18.9 % 883 11.6 %
    Women * missing 727 11.1 % 1001 13.1 %
    Men * owned at both time points 1157 17.7 % 1599 21.0 %
    Men * owned at one time point 196 3.0 % 253 3.3 %
    Men * rented at both time points 1180 18.0 % 784 10.3 %
    Men * missing 743 11.3 % 978 12.8 %
Malaise inventory, M (SD)
    Age 23 1.11 1.45 Age 26 1.69 1.71
    Age 33 0.90 1.42 Age 29 1.45 1.66
    Age 42 1.42 1.68 Age 34 1.59 1.84
    Age 50 1.40 1.87 Age 42 1.77 1.93
    Age 62 1.22 1.69 Age 46 1.69 2.06
    Age 62.5 1.50 1.88 Age 50 1.63 1.93
    Age 63 1.43 1.85 Age 50.5 1.97 2.10
    Age 64. 1.38 1.78 Age 51 1.86 2.06
Age 52.5 1.66 2.00
Malaise inventory missingness, N (%)
    Age 23 624 10.2 % Age 26 2512 32.0 %
    Age 33 478 7.8 % Age 29 1252 15.9 %
    Age 42 142 2.3 % Age 34 1632 20.8 %
    Age 50 272 4.4 % Age 42 1696 21.6 %
    Age 62 2783 45.3 % Age 46 1596 20.3 %
    Age 62. 1797 29.3 % Age 50 4511 57.4 %
    Age 63 1463 23.8 % Age 50.5 3507 44.6 %
    Age 64.5 22 0.4 % Age 51 3143 40.0 %
Age 52.5 819 10.4 %
Open in a new tab

Note. BCS/70: 1970 British Cohort Study; M: mean; N: frequency; NCDS/58: 1958 National Child Development Study; SD: standard deviation; SEP: socioeconomic position. Results based on unweighted data. Analytical samples exclude participants who took part in the latest main survey sweep prior to the COVID-19 pandemic onset (n = 1662 in NCDS/58 and n = 116 in BCS/70) or within the period spanning the COVID-19 Surveys data collection (n = 44 in BCS/70, interviewed between September–October 2020).

3.1. Long-term trajectory analyses

Evidence supporting the measurement invariance of the nine-item Malaise Inventory was found at the level needed to ensure valid comparisons of the psychological distress levels across time-points, genders, childhood socioeconomic position groups, and their intersections, as well as across birth cohorts. Further details on the approach and its results are available in eAppendix 1 (Supplementary Material).

The model comparison strategy supported the use of a piecewise model with a cubic spline and a quadratic spline, with a knot at the latest pre-pandemic assessment, as well as random intercepts and slopes for the linear terms within each of the splines. Further details on the model selection approach and its results are available in eAppendix 7 (Supplementary Material).

The overall long-term trajectory models in both birth cohorts showed that, after an initial increase in distress during the pandemic (mid-2020 to mid-2021) (Bspline2_linear_NCDS/58 = 0.33 [0.21, 0.44], p < 0.001; Bspline2_linear_BCS/70 = 0.37 [0.24, 0.50], p < 0.001), levels decreased towards the post-lockdown period (towards the second half of 2022) (Bspline2_quadratic_NCDS/58 = −0.12 [−0.17, −0.08], p < 0.001; Bspline2_quadratic_BCS/70 = −0.16 [−0.21, −0.11], p < 0.001), suggesting a bounce back to pre-pandemic levels (Fig. 1). Similar results were found in the models with interaction terms. We found large inequalities at the first time-point in the long-term trajectory (age 23 in NCDS/58, age 26 in BCS/70), with women having significantly higher levels of distress than men (Bwomen_NCDS/58 = 0.72 [0.62, 0.82], p < 0.001; Bwomen_BCS/70 = 0.73 [0.62, 0.83], p < 0.001) (Fig. 2), and those with parents from a non-manual social class (Bnon-manual_NCDS/58 = −0.24 [−0.35, −0.14], p < 0.001; Bnon-manual_BCS/70 = −0.23 [−0.35, −0.12], p < 0.001) (Fig. 3) or living in an owned property during childhood (Bowned7&11_NCDS/58 = −0.34 [−0.46, −0.22], p < 0.001; Bowned5&10_BCS/70 = −0.30 [−0.45, −0.15], p < 0.001) (Fig. 4) having significantly lower levels of distress than those with parents from a manual social class or living in a rented property at ages 7&11 (NCDS/58) or 5&10 (BCS/70).

Fig. 1. Overall long-term trajectories of psychological distress across NCDS/58 (n = 6553) and BCS/70 (n = 7629).

Fig. 1

Open in a new tab

Note. 95 % confidence intervals for the marginal predicted mean psychological distress levels from the multilevel growth curve models. Results based on weighted and imputed data.

Fig. 2. Long-term trajectories of psychological distress by gender across NCDS/58 (n = 6553) and BCS/70 (n = 7629).

Fig. 2

Open in a new tab

Note. 95 % confidence intervals for the marginal predicted mean psychological distress levels from the multilevel growth curve models. Results based on weighted and imputed data.

Fig. 3. Long-term trajectories of psychological distress by parental social class during childhood (age 10/11) across NCDS/58 (n = 6553) and BCS/70 (n = 7629).

Fig. 3

Open in a new tab

Note. 95 % confidence intervals for the marginal predicted mean psychological distress levels from the multilevel growth curve models. Results based on weighted and imputed data.

Fig. 4. Long-term trajectories of psychological distress by childhood housing tenure (ages 5&10/7&11) across NCDS/58 (n = 6553) and BCS/70 (n = 7629).

Fig. 4

Open in a new tab

Note. 95 % confidence intervals for the marginal predicted mean psychological distress levels from the multilevel growth curve models. Results based on weighted and imputed data.

In NCDS/58, inequalities in the starting points were also found at the intersection of gender and childhood housing tenure (Bwomen*owned7&11_NCDS/58 = −0.27 [−0.49, −0.05], p = 0.014) (Fig. 5) and, to a lesser extent, at the intersection of gender and childhood parental social class (Bwomen*non-manual_NCDS/58 = −0.21 [−0.42, 0.00], p = 0.048) (Fig. 6), although in this latter case they were almost non-statistically significant at the 95 %CI, showing larger childhood socioeconomic position inequalities in women than in men. Evidence suggestive of inequalities at those intersections in the starting points was not found in BCS/70.

Fig. 5. Long-term trajectories of psychological distress by gender and childhood housing tenure (ages 5&10/7&11) across NCDS/58 (n = 6553) and BCS/70 (n = 7629).

Fig. 5

Open in a new tab

Note. M: men; W: women. 95 % confidence intervals for the marginal predicted mean psychological distress levels from the multilevel growth curve models. Results based on weighted and imputed data.

Fig. 6. Long-term trajectories of psychological distress by gender and parental social class during childhood (age 10/11) across NCDS/58 (n = 6553) and BCS/70 (n = 7629).

Fig. 6

Open in a new tab

Note. M*: men; *M: manual parental social class; *NM: non-manual parental social class; W*: women. 95 % confidence intervals for the marginal predicted mean psychological distress levels from the multilevel growth curve models. Results based on weighted and imputed data.

In NCDS/58, a larger linear decrease among women (Bspline1_linear*women_NCDS/58 = −0.04 [−0.07, −0.01], p = 0.020) in the first segment of the long-term trajectory (up to the COVID-19 pandemic onset) was reflected in the gender inequality decreasing towards the age of 30 (Fig. 2). We did not find evidence of significant differences in the change over time by childhood socioeconomic position or their intersection with gender in any of the two cohorts.

The coefficients from the long-term trajectory models (multilevel growth curve models), are available in eAppendix 8 (Supplementary Material).

Models estimated with continuous age and with the extended sample provided very similar results. Estimates and plots of the marginal mean predicted levels from these models are available in eAppendix 9 and eAppendix 10 (Supplementary Material), respectively.

3.2. Differences across key time-points

Table 2 shows the estimates and 95 % CIs from the analyses comparing the psychological distress levels and inequalities between the most recent survey sweeps in both birth cohorts and three key time-points.

Table 2. Estimates and 95 % confidence intervals (CIs) from analyses comparing psychological distress levels between most recent main sweeps and earliest time-point, most recent pre-pandemic assessment, or point of highest psychological distress during COVID-19 pandemic.

NCDS/58 (n = 6553)
First (age 23) vs last (age 64.5) Pre-pandemic (age 50) vs last (age 64.5) Pandemic (age 62.5) vs last (age 64.5)
B (95 % CI) p B (95 % CI) p B (95 % CI) p
Overall
Time 0.35 (0.26, 0.45) <0.001 0.01 (–0.07, 0.09) 0.791 –0.16 (–0.23, –0.08) <0.001
Gender (ref.: Men)
Time 0.41 (0.28, 0.54) <0.001 0.00 (–0.13, 0.12) 0.974 –0.06 (–0.17, 0.04) 0.248
Women 0.72 (0.61, 0.83) <0.001 0.58 (0.45, 0.70) <0.001 0.79 (0.67, 0.91) <0.001
Time * Women –0.12 (–0.30, 0.06) 0.208 0.03 (–0.15, 0.21) 0.765 –0.19 (–0.35, –0.04) 0.016
Parental social class (ref.: Manual)
Time 0.38 (0.25, 0.50) <0.001 0.08 (–0.04, 0.20) 0.179 –0.13 (–0.23, –0.02) 0.019
Non-manual –0.24 (–0.35, –0.12) <0.001 –0.10 (–0.23, 0.04) 0.162 –0.21 (–0.34, –0.08) 0.002
Time * Non-manual –0.07 (–0.26, 0.13) 0.505 –0.20 (–0.39, –0.01) 0.039 –0.08 (–0.24, 0.07) 0.287
Gender (ref.: Men) and parental social class (ref.: Manual)
Time 0.45 (0.28, 0.63) <0.001 0.07 (–0.11, 0.24) 0.452 –0.01 (–0.15, 0.13) 0.882
Women 0.81 (0.66, 0.95) <0.001 0.62 (0.45, 0.79) <0.001 0.90 (0.73, 1.06) <0.001
Non-manual –0.11 (–0.27, 0.05) 0.175 –0.03 (–0.22, 0.16) 0.734 –0.06 (–0.23, 0.12) 0.532
Time * Women –0.16 (–0.41, 0.09) 0.210 0.03 (–0.22, 0.27) 0.823 –0.24 (–0.45, –0.03) 0.026
Time * Non-manual –0.13 (–0.41, 0.16) 0.377 –0.20 (–0.47, 0.08) 0.162 –0.15 (–0.37, 0.07) 0.176
Women * Non-manual –0.25 (–0.50, –0.01) 0.044 –0.12 (–0.39, 0.15) 0.397 –0.29 (–0.56, –0.02) 0.034
Time * Women * Non-manual 0.13 (–0.31, 0.56) 0.563 0.00 (–0.38, 0.38) 0.983 0.14 (–0.20, 0.47) 0.417
Childhood housing tenure (ref.: Rented at both time-points)
Time 0.35 (0.22, 0.49) <0.001 0.05 (–0.07, 0.18) 0.401 –0.17 (–0.28, –0.05) 0.007
Owned at both time-points –0.35 (–0.47, –0.23) <0.001 –0.24 (–0.38, –0.10) 0.001 –0.35 (–0.50, –0.21) <0.001
Time * Owned at both time-points 0.00 (–0.19, 0.20) 0.977 –0.10 (–0.30, 0.09) 0.294 0.00 (–0.17, 0.16) 0.961
Gender (ref.: Men) and childhood housing tenure (ref.: Rented at both time-points)
Time 0.36 (0.18, 0.53) <0.001 0.02 (–0.18, 0.22) 0.846 –0.07 (–0.23, 0.08) 0.363
Women 0.83 (0.66, 0.99) <0.001 0.76 (0.55, 0.96) <0.001 0.99 (0.79, 1.20) <0.001
Owned at both time-points –0.19 (–0.34, –0.05) 0.007 –0.06 (–0.24, 0.13) 0.545 –0.16 (–0.34, 0.03) 0.102
Time * Women 0.00 (–0.27, 0.26) 0.980 0.07 (–0.21, 0.34) 0.627 –0.18 (–0.43, 0.06) 0.144
Time * Owned at both time-points 0.03 (–0.21, 0.28) 0.781 –0.10 (–0.36, 0.16) 0.463 –0.02 (–0.24, 0.20) 0.882
Women * Owned at both time-points –0.27 (–0.50, –0.04) 0.021 –0.33 (–0.60, –0.06) 0.017 –0.35 (–0.64, –0.07) 0.016
Time * Women * Owned at both time-point –0.07 (–0.44, 0.31) 0.733 –0.01 (–0.38, 0.36) 0.968 0.01 (–0.32, 0.35) 0.932
BCS/70 (n=7629)
First (age 26) vs last (age 52.5) Pre-pandemic (age 46) vs last (age 52.5) Pandemic (age 50.5) vs last (age 52.5)
B (95 % CI) p B (95 % CI) p B (95 % CI) p
Overall
Time 0.02 (–0.07, 0.11) 0.644 –0.08 (–0.15, –0.01) 0.022 –0.33 (–0.41, –0.26) <0.001
Gender (ref.: Men)
Time 0.00 (–0.12, 0.11) 0.975 –0.22 (–0.32, –0.13) <0.001 –0.33 (–0.44, –0.22) <0.001
Women 0.73 (0.63, 0.84) <0.001 0.48 (0.36, 0.59) <0.001 0.75 (0.63, 0.87) <0.001
Time * Women 0.04 (–0.12, 0.21) 0.600 0.28 (0.13, 0.42) <0.001 0.00 (–0.15, 0.15) 0.984
Parental social class (ref.: Manual)
Time 0.02 (–0.11, 0.14) 0.783 –0.08 (–0.17, 0.02) 0.105 –0.34 (–0.46, –0.22) <0.001
Non-manual –0.27 (–0.40, –0.15) <0.001 –0.24 (–0.37, –0.11) <0.001 –0.26 (–0.41, –0.12) <0.001
Time * Non-manual 0.01 (–0.19, 0.20) 0.939 –0.01 (–0.16, 0.14) 0.891 0.02 (–0.16, 0.19) 0.862
Gender (ref.: Men) and parental social class (ref.: Manual)
Time –0.05 (–0.22, 0.12) 0.566 –0.22 (–0.36, –0.08) 0.002 –0.34 (–0.51, –0.17) <0.001
Women 0.73 (0.56, 0.89) <0.001 0.55 (0.38, 0.72) <0.001 0.84 (0.65, 1.02) <0.001
Non-manual –0.30 (–0.48, –0.11) 0.002 –0.15 (–0.33, 0.02) 0.089 –0.17 (–0.38, 0.04) 0.107
Time * Women 0.14 (–0.11, 0.39) 0.280 0.29 (0.09, 0.49) 0.005 0.00 (–0.22, 0.22) 0.993
Time * Non-manual 0.12 (–0.16, 0.40) 0.382 0.00 (–0.22, 0.21) 0.981 0.02 (–0.24, 0.27) 0.901
Women * Non-manual 0.03 (–0.24, 0.31) 0.814 –0.17 (–0.41, 0.08) 0.187 –0.19 (–0.48, 0.09) 0.180
Time * Women * Non-manual –0.23 (–0.65, 0.19) 0.279 –0.03 (–0.34, 0.28) 0.860 0.00 (–0.37, 0.36) 0.991
Childhood housing tenure (ref.: Rented at both time-points)
Time 0.08 (–0.11, 0.27) 0.399 0.00 (–0.16, 0.16) 0.960 –0.32 (–0.48, –0.16) <0.001
Owned at both time-points –0.31 (–0.47, –0.15) <0.001 –0.28 (–0.45, –0.12) 0.001 –0.37 (–0.54, –0.19) <0.001
Time * Owned at both time-points –0.15 (–0.36, 0.07) 0.177 –0.13 (–0.32, 0.05) 0.164 –0.04 (–0.23, 0.15) 0.680
Gender (ref.: Men) and childhood housing tenure (ref.: Rented at both time-points)
Time 0.07 (–0.18, 0.32) 0.592 –0.18 (–0.38, 0.03) 0.100 –0.41 (–0.63, –0.19) <0.001
Women 0.86 (0.61, 1.12) <0.001 0.54 (0.28, 0.80) <0.001 0.70 (0.43, 0.98) <0.001
Owned at both time-points –0.25 (–0.45, –0.04) 0.020 –0.24 (–0.46, –0.02) 0.031 –0.39 (–0.64, –0.15) 0.001
Time * Women 0.02 (–0.34, 0.39) 0.892 0.34 (0.04, 0.64) 0.024 0.17 (–0.14, 0.48) 0.273
Time * Owned at both time-points –0.12 (–0.41, 0.16) 0.397 –0.08 (–0.32, 0.17) 0.533 0.08 (–0.19, 0.34) 0.568
Women * Owned at both time-points –0.12 (–0.42, 0.18) 0.419 –0.09 (–0.39, 0.22) 0.588 0.05 (–0.28, 0.39) 0.751
Time * Women * Owned at both time-point –0.05 (–0.46, 0.37) 0.831 –0.10 (–0.45, 0.25) 0.571 –0.22 (–0.59, 0.15) 0.235
Open in a new tab

Note. BCS/70: 1970 British Cohort Study; B: coefficient; CI: confidence interval; NCDS/58: 1958 National Child Development Study; p: significance level. Results based on weighted and imputed data. Analyses including childhood housing tenure exclude cohort members who lived in a rented or owned house at only one of the time-points; sample size in these analyses is nNCDS/58 = 4703 and nBCS/70 = 5106.

Differences between the most recent pre-pandemic assessments and the latest sweeps (ages 50/64.5 in NCDS/58 and 46/52.5 in BCS/70) were not significant in NCDS/58 (Bprepandemic_time_NCDS/58 = 0.01 [−0.07, 0.09], p = 0.791) and marginally significant and negative in BCS/70 (Bprepandemic_time_BCS/70 = −0.08 [−0.15, −0.01], p = 0.022), in line with the notion of a “bounce back” to (or even slight improvement compared to) pre-pandemic levels.

In addition to the above-reported inequalities at earliest time-points (akin to the starting points in the long-term trajectory analyses, age 23 in NCDS/58 and 26 in BCS/70) by gender, parental social class during childhood, and childhood housing tenure (and their intersections in NCDS/58), we found similar inequalities at the height of the COVID-19 pandemic (age 62.5 in NCDS/58 and 50.5 in BCS/70). Similar results were found in the comparisons with the pre-pandemic time-point, although in this case the difference by parental social class during childhood was not significant in NCDS/58 (Bnon-manual_NCDS/58 = −0.10 [−0.23, −0.04], p = 0.162), widening again towards the post-pandemic period (Bprepandemic_time*non-manual_NCDS/58 = −0.20 [−0.39, −0.01], p = 0.039). In NCDS/58 there was a larger decrease in psychological distress between the pandemic and latest time-point among women than men (Bpandemic_time*women_NCDS/58 = −0.19 [−0.35, −0.04], p = 0.016). In BCS/70, however, we found evidence suggestive of a widening gender inequality when comparing the pre-pandemic and most recent assessments (Bprepandemic_time*women_BCS/70 = 0.28 [0.13, 0.42], p < 0.001).

The pooled analyses with the additional cohort terms (main effects and interactions) showed that, in line with Fig. 1, overall levels of psychological distress were significantly higher in BCS/70 than NCDS/58 in the three time comparisons. The cohort difference was significantly smaller in the most recent sweep when compared to the earliest time-point and to the difference during the pandemic (Bfirst_time*BCS/70 = −0.33 [−0.45, −0.21], p < 0.001; Bpandemic_time*BCS/70 = −0.18 [−0.29, −0.07], p = 0.002), but not significantly different when using the most recent pre-pandemic assessment as the comparison time-point (Bpre-pandemic_time*BCS/70 = −0.09 [−0.20, −0.01], p = 0.087). The full results from the pooled analyses are available in eAppendix 11 (Supplementary Material).

3.3. Between-person post-lockdown analyses

We found a significant relationship between inflation and psychological distress in the post-lockdown period, and we did not find evidence that this relationship significantly varied by gender, (concurrent) socioeconomic position, or their intersection. Detailed results from these analyses are available in eAppendix 12 and eAppendix 13 (Supplementary Material).

4. Discussion

In this study, we used data from two long-standing birth cohorts, representative of the British population born in 1958 and 1970, with three aims: first, to understand how long-term trajectories of mental (ill-)health evolved during the post-lockdown period, which overlapped with large cost-of-living increases; second, to examine differences in the trajectories by generation, gender, childhood socioeconomic position, and their intersections; and, third, to explore differences in mental (ill-) health by cost-of-living levels. Regarding the first and second aims, we found that, after a period of increased distress coinciding with the COVID-19 pandemic, levels reduced towards the post-lockdown period. These post-lockdown reductions were larger among the younger cohort (born in 1970 vs 1958) and among women from the 1958 cohort than men. Across most of the adult lifespan of both cohorts (ages 23–64.5 in NCDS/58 and 26–52.5 in BCS/70, spanning 41.5 and 26.5 years, respectively), women and those from a disadvantaged childhood socioeconomic position had, on average, higher levels of distress than men and those from an advantaged childhood socioeconomic position. We also found evidence suggesting that the inequalities by childhood socioeconomic position were larger in women from the older cohort (NCDS/58). In most cases, we did not find evidence of different changes over time by gender, childhood socioeconomic position, or their intersection. The only evidence we found suggesting a reduction of these inequalities was for women from the older cohort (NCDS/58) between the height of the pandemic and the latest data collection around 2022. In turn, partial evidence suggested a widening in the gender inequalities in the younger cohort (BCS/70) between the pre- and post-pandemic periods, as well as a temporary reduction in the inequalities by parental social class in childhood at age 50 in the older cohort (NCDS/58), just to widen again towards the most recent (post-pandemic) sweep. Our time-point-by-time-point comparisons also indicated a slight widening of childhood socioeconomic inequalities in the older cohort (NCDS/58) between the pre-pandemic and post-pandemic periods, suggesting that these inequalities may have increased. Finally, regarding the third aim, we found that inflation was associated with the incidence of psychological distress symptoms, but we did not find evidence suggesting that this relationship substantially varied by gender, concurrent socioeconomic position, or their intersections.

Taken altogether, these results offer a mixed view of the population mental health in post-lockdown Britain. The two generations of adults under study seem to be ‘bouncing back’ from the high levels of psychological distress experienced during the pandemic (Moreno-Agostino et al., 2023). However, our results also suggest that the improvement is smaller in the older cohort (NCDS/58). This is consistent with emerging evidence on depressive symptomatology in the pandemic aftermath among English adults aged 50 and older, with larger representation of adults aged 60–74 (Zaninotto et al., 2025). However, it would not be expected based on previous cross-generational evidence including older (1946) cohorts at similar ages (Gondek et al., 2022; Moreno-Agostino et al., 2023) or from evidence on mental health around retirement age in Britain (Fleischmann et al., 2020), both of which would have suggested an improvement in mental health around this age. Indeed, although the younger cohort (BCS/70, ‘Generation X’) had higher levels of psychological distress than the older cohort (NCDS/58, ‘Baby Boomers’) throughout the lifespan, in line with previous cross-generational evidence (Gondek et al., 2022; Moreno-Agostino et al., 2023), our study suggests that those inequalities may be decreasing towards older ages. It is possible that this is a result of experiencing multiple population-wide shocks (e.g., COVID-19 pandemic and cost-of-living increases) at a particularly sensitive period (retirement age). However, this is uncharted territory in a scholarship otherwise suggesting a decline in multiple health (including mental health) outcomes in younger generations (Gimeno et al., 2024, 2025), so further continued monitoring is needed.

We also found evidence suggesting a narrowing gender inequality in psychological distress in the NCDS/58 cohort between the height of the pandemic and the post-lockdown period. On the one hand, it is possible that psychological distress levels for men in their 60s are declining or not improving as fast as for women in recent years. Considering the challenging economic situation and the crucial transition into retirement, this could be consistent with previous evidence suggesting that, although women’s mental health tends to be poorer at all time-points, men’s mental health can deteriorate more during economic crises (Glonti et al., 2015) or with multiple inflation hardships (Louie et al., 2023), in line with sexist conceptions of the man as the ‘breadwinner’. This could also be consistent with the finding that these and other (but not all) sexist conceptions and attitudes have declined in Britain over the last few decades (Allen and Stevenson, 2023), which could reflect in them having a larger influence over older cohorts. On the other hand, however, the gender inequalities in the most recent NCDS/58 data collection were not different when compared to the ones earlier in life (ages 23 and 50). This suggests that these ‘improvements’ may be due to the fact that, since the pandemic came with a widening of the pre-existing gender inequalities (Moreno-Agostino et al., 2023), women had a longer way to ‘bounce back’, without really impacting the gender inequalities in the long run. Unfortunately, this is consistent with the findings in BCS/70, where the gender inequalities in the post-lockdown period were not significantly different than during the height of the pandemic, earlier in life (age 26) or, more dishearteningly, compared to prior to the pandemic, where part of our evidence actually suggests a widening in the gender inequalities. A reason for this lack of improvement in the long run may be found in the same report on the change in (some) sexist attitudes, which suggests that behavioural changes have not followed attitudinal ones, with the division of domestic labour, for instance, still resting mostly on women’s shoulders (Allen and Stevenson, 2023).

Like gender inequalities, childhood socioeconomic inequalities persisted throughout the entire long-term trajectories, without consistent evidence of these narrowing. On the one hand, we found that the proportion of people in the relatively disadvantaged childhood socioeconomic position was smaller in the younger generation, meaning that the proportion of individuals in the socioeconomically disadvantaged long-term trajectory has decreased across generations. On the other hand, we provide one of the longest-term empirical studies to date, which suggests early life socioeconomic disadvantage continues to influence mental health inequalities even after more than half a century. As a fundamental cause of health (including mental health) inequalities (Link and Phelan, 1995), early life socioeconomic inequalities can persist across the life course (and transmit across life courses) due to complex chains of privilege and disadvantage which start even before birth (Houweling and Grunberger, 2024). Our study adds to a growing field supporting the notion that early life prevention of socioeconomic disadvantage will be crucial, in order to reduce life-course inequalities and prevent further intergenerational transmission of health inequalities (Kirkbride et al., 2024). In light of the growing number of children in poverty in the UK (Social Metrics Commission, 2024), immediate measures to reduce socioeconomic adversity and counter decades of austerity measures (Ruckert and Labonte, 2017; Stuckler et al., 2017), including the reversal of the two-child benefit cap (Taylor-Robinson et al., 2024), may be needed to prevent these disadvantage cycles to further reproduce into generations currently in their infancy and youth (University of York Cost of Living Research Group, 2023). This is also consistent with the results of our analysis focused on the post-lockdown period, which suggests that inflation had a negative relationship with mental health during this period and that most of the inequalities by gender and concurrent socioeconomic position were independent of inflation levels or predating the period with highest inflation levels.

At the intersection of both gender and childhood socioeconomic inequalities, we found that mental health inequalities by childhood socioeconomic position were larger in women than in men, but only in the older (NCDS/58) cohort. We did not find evidence of these intersectional inequalities decreasing over time within that cohort, and we did not find evidence of such inequalities in the younger (BCS/70) cohort. These cohort differences may be, in part, due to increases in educational attainment and, relatedly, labour market participation among women across generations (Allen and Stevenson, 2023), suggesting that these inequalities can be prevented.

4.1. Strengths and limitations

We used data from two long-standing birth cohort studies, representing two generations (‘Baby Boomers’ and ‘Generation X’) of the British population. To the best of our knowledge, this is the longest population-based study of long-term trajectories of mental (ill-)health within the same individuals, with trajectories spanning up to more than four decades. We used a robust approach to the measurement of mental (ill-)health across the life course, ensuring that the measures were equivalent not only over time within the same individuals but also by gender, childhood socioeconomic position, cohort, and their intersections, an aspect that is often overlooked in quantitative research on intersectional inequalities (Else-Quest and Hyde, 2016). We also used the rich information collected from individuals across their life courses to inform our approach to dealing with missing data (Silverwood et al., 2024). By using these approaches, we maximised the plausibility that our results are generalisable to the British population at these ages.

However, the results from our study must be interpreted considering some limitations. First, despite our efforts to deal with missing data, including the use of non-response weights to restore representativeness to the target populations under study (people born in Britain in 1958 and 1970, still alive and residing in the UK at the most recent sweep), our findings may not be generalisable to other sectors within the British adult population, including migrant and displaced communities or ethnically minoritised groups. Our results may also not be generalisable to the populations of other countries, particularly those with substantially different geopolitical contexts and economic conditions and constraints, and which make up most of the global population. Related to this, the overlap between the UK’s exit from the European Union (Brexit) and the COVID-19 pandemic not only limits the generalisability of these results to other contexts but also makes it difficult to disaggregate the relative and combined contributions of both events. Other geopolitical events such as the Russian invasion of Ukraine may have partly confounded the relationship between inflation and psychological distress, and we were unable to adjust for this due to the overlap between these two events. Second, it is possible that parental mental health during childhood may have partly confounded the pattern of inequalities in the long-term distress trajectories by childhood SEP. We were unable to appropriately account for this due to the lack of good-quality measures of parental mental health that unequivocally preceded the measures of childhood SEP (and the risk of taking away part of the effect of parental SEP if adjusting for those). Third, we used a relatively simple approach to categorising individuals into gender and socioeconomic groups. While we are aware that there is a larger complexity in how these social identities and positions can be approached and categorised, we aimed to maximise comparability across the cohorts, minimise the presence of missing data, ensure temporal ordering, and limit model complexity. However, we acknowledge that these conform to an inherently limited set of intersecting identities and positions that are provisionally adopted as proxies of the systems of oppression underlying any of the inequalities we found (McCall, 2005), and that both gender (Bauer, 2023) and SEP (Galobardes et al., 2007) are complex and multidimensional concepts. Although we used multiple indicators of childhood and adulthood SEP in our analyses, with consistent results, it is possible that research using different SEP indicators (e.g., employment or income) may lead to different results. Fourth, despite the robust measurement approach we used, it is possible that the set of questions used to measure mental (ill-)health was already impacted by gender bias upon the very development of the questionnaire (Hill and Needham, 2013), which goes beyond the reach of the measurement invariance testing. Finally, we may have been limited in terms of statistical power to detect some ‘effects’ (in statistical terms), particularly those involving multiple interactions, which include most intersectional terms. We coupled the analytical approaches with data visualisations to avoid exclusively relying on statistical tests. The visualisations were generally consistent with the observation that social inequalities have not substantially reduced across the life course. It is important to note that the absence of an ‘intersectional effect’ does not imply the absence of intersectional experiences (Evans and Erickson, 2019). Furthermore, future studies may use other methods (e.g., three-level intersectional MAIHDA models (Bell et al., 2024)) to either replicate our findings or extend them to additional intersectional strata.

5. Conclusions

This study shows that, after a period of increased levels of psychological distress during the early phases of the COVID-19 pandemic, levels of psychological distress in British adults born in 1958 and 1970 have reduced close to pre-pandemic levels. Our study also confirms long-range impacts of childhood socioeconomic disadvantage on adult mental health even after more than half a century, mental health inequalities by gender at all time points and, in the 1958 cohort, inequalities at the intersection of childhood socioeconomic position and gender, that persisted for most of the life course. Although inflation was associated with psychological distress, inequalities by gender and adult socioeconomic position were independent of (or predated high levels of) inflation. This suggests that efforts must be made to reduce gender and socioeconomic inequalities from early in life, coupled with interventions (and further research into optimal interventions) to reverse unjust inequalities in population mental health.

Supplementary Material

Appendix A.Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.socscimed.2025.118830.

Supplementary Material

Acknowledgements

We would like to thank the members of the NCDS/58 and BCS/70 cohorts for generously giving up their time across their life courses, as well as the Centre for Longitudinal Studies team members for collecting, managing, and making these data useable and accessible.

The 1958 National Child Development Study (NCDS/58) and the 1970 British Cohort Study (BCS/70) are supported by the Centre for Longitudinal Studies, Resource Centre 2015–2020 [ES/M001660/1] and 2022 [ES/W013142/1] grants, along a host of other co-funders. The COVID-19 data collections were funded by the UKRI grant Understanding the economic, social and health impacts of COVID-19 using lifetime data: evidence from 5 nationally representative UK cohorts [ES/V012789/1].

The views expressed are those of the authors and not necessarily those of the Wellcome Trust, ESRC, King’s College London, or other funders.

Funding

DMA, GBP, and JDM are part supported by the Economic and Social Research Council (ESRC) Centre for Society and Mental Health at King’s College London [grant number ES/S012567/1]. DMA is supported by the Wellcome Trust [grant number 304283/Z/23/Z]. GBP is supported by the UK Research and Innovation (UKRI) Centre for Longitudinal Studies resource centre [grant numbers ES/M001660/1, ES/W013142/1]. JDM is in receipt of funding from UKRI funding for the Population Mental Health Consortium [grant number MR/Y030788/1] which is part of Population Health Improvement UK (PHIUK), a national research network which works to transform health and reduce inequalities through change at the population level. JDM has also received funding from the Health Foundation working together with the Academy of Medical Sciences, for a Clinician Scientist Fellowship. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Footnotes

CRediT authorship contribution statement

Darío Moreno-Agostino: Writing – review & editing, Writing – original draft, Visualization, Methodology, Formal analysis, Data curation, Conceptualization. George B. Ploubidis: Writing – review & editing, Supervision, Funding acquisition, Conceptualization. Jayati Das-Munshi: Writing – review & editing, Supervision, Funding acquisition, Conceptualization.

EA statement

The most recent sweeps of the 1970 British Cohort Study (BCS70) and the 1958 National Child Development Study (NCDS) have been granted ethical approval by the National Health Service (NHS) Research Ethics Committee and all participants have given informed consent. No additional ethical approval was necessary for this secondary data analysis.

Declaration of competing interest

No competing interests to declare.

Data availability

All used data are freely available to bona fide researchers at the UK Data Service (https://ukdataservice.ac.uk/). Code is publicly available at the Open Science Framework (https://doi.org/10.17605/OSF.IO/MSPRQ).

References

  1. Allen J, Stevenson I. Gender Roles (British Social Attitudes: the 40th Report. National Centre for Social Research; 2023. Issue https://natcen.ac.uk/publications/bsa-40-gender-roles. [Google Scholar]
  2. Armitage JM, Kwong ASF, Tseliou F, Sellers R, Blakey R, Anthony R, Rice F, Thapar A, Collishaw S. Cross-cohort change in parent-reported emotional problem trajectories across childhood and adolescence in the UK. Lancet Psychiatry. 2023;10(7):509–517. doi: 10.1016/S2215-0366(23)00175-X. [DOI] [PubMed] [Google Scholar]
  3. Bauer GR. Incorporating intersectionality theory into population health research methodology: challenges and the potential to advance health equity. Soc Sci Med. 2014;110:10–17. doi: 10.1016/j.socscimed.2014.03.022. [DOI] [PubMed] [Google Scholar]
  4. Bauer GR. Sex and gender multidimensionality in epidemiologic research. Am J Epidemiol. 2023;192(1):122–132. doi: 10.1093/aje/kwac173. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bell A, Evans C, Holman D, Leckie G. Extending intersectional multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) to study individual longitudinal trajectories, with application to mental health in the UK. Soc Sci Med. 2024;351:116955. doi: 10.1016/j.socscimed.2024.116955. [DOI] [PubMed] [Google Scholar]
  6. Bowleg L. When black + Lesbian + woman ≠ Black Lesbian woman: the methodological challenges of qualitative and quantitative intersectionality research. Sex Roles. 2008;59(5):312–325. doi: 10.1007/s11199-008-9400-z. [DOI] [Google Scholar]
  7. Bowleg L, Bauer G. Invited reflection:quantifying intersectionality. Psychol Women Q. 2016;40(3):337–341. doi: 10.1177/0361684316654282. [DOI] [Google Scholar]
  8. Broadbent P, Thomson R, Kopasker D, McCartney G, Meier P, Richiardi M, McKee M, Katikireddi SV. The public health implications of the cost-of-living crisis: outlining mechanisms and modelling consequences. Lancet Reg Health Eur. 2023;27:100585. doi: 10.1016/j.lanepe.2023.100585. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Brown M, Goodman A, Peters A, Ploubidis GB, Sanchez A, Silverwood R, Smith K. COVID-19 Survey in Five National Longitudinal Studies: Waves 1, 2 and 3 User Guide (Version 3) UCL Centre for Longitudinal Studies and MRC Unit for Lifelong Health and Ageing. 2021 [Google Scholar]
  10. Brown M, Mendonca S, Mohamad-Zaki N, Moulton V, Peters A, Ploubidis G, Sanchez-Galvez A, Sanchez C, Sedovic M, Agalioti-Sgompou V, Silverwood R, et al. National Child Development Survey: Age 62 Survey (Version 1) UCL Centre for Longitudinal Studies. 2025 [Google Scholar]
  11. Bryk AS, Raudenbush SW. Application of hierarchical linear models to assessing change. Psychol Bull. 1987;101(1):147–158. doi: 10.1037/0033-2909.101.1.147. [DOI] [Google Scholar]
  12. Collins PH, Bilge S. Intersectionality. John Wiley & Sons; 2020. Intersectionality revisited; pp. 219–241. [Google Scholar]
  13. Curran PJ, Obeidat K, Losardo D. Twelve frequently asked questions about growth curve modeling. J Cognit Dev. 2010;11(2):121–136. doi: 10.1080/15248371003699969. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Dodgeon B, Morris T, Crawford C, Parsons S, Vignoles A, Oldfield Z, O’Neill D. CLOSER work package 2: harmonised socio-economic measures user guide (revised) CLOSER; 2019. [Google Scholar]
  15. Else-Quest NM, Hyde JS. Intersectionality in quantitative psychological Research:II Methods and techniques. Psychol Women Q. 2016;40(3):319–336. doi: 10.1177/0361684316647953. [DOI] [Google Scholar]
  16. Enders CK. Missing data: an update on the state of the art. Psychol Methods. 2023 doi: 10.1037/met0000563. [DOI] [PubMed] [Google Scholar]
  17. England C, Jarrom D, Washington J, Hasler E, Batten L, Edwards A, Lewis R. Methodological approaches to measuring mental health in a cost-of-living crisis: a rapid review. Health Policy. 2024;144:105062. doi: 10.1016/j.healthpol.2024.105062. [DOI] [PubMed] [Google Scholar]
  18. Evans CR, Erickson N. Intersectionality and depression in adolescence and early adulthood: a MAIHDA analysis of the national longitudinal study of adolescent to adult health, 1995-2008. Soc Sci Med. 2019;220:1–11. doi: 10.1016/j.socscimed.2018.10.019. [DOI] [PubMed] [Google Scholar]
  19. Fleischmann M, Xue B, Head J. Mental health before and after retirement-assessing the relevance of psychosocial working conditions: the whitehall II prospective study of British civil servants. J Gerontol B Psychol Sci Soc Sci. 2020;75(2):403–413. doi: 10.1093/geronb/gbz042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Francis-Devine B. Commons Library Research Briefing Number 7096. published by the House of Commons Library; 2024. Apr 8, Poverty in the UK: Statistics. 2024. [Google Scholar]
  21. Francis-Devine B, Ferguson D. 50 Years of the Equal Pay Act. House of Commons Library; 2020. Retrieved 28/02/2025 from. https://commonslibrary.parliament.uk/50-years-of-the-equal-pay-act/ [Google Scholar]
  22. Galobardes B, Lynch J, Smith GD. Measuring socioeconomic position in health research. Br Med Bull. 2007;81–82:21–37. doi: 10.1093/bmb/ldm001. [DOI] [PubMed] [Google Scholar]
  23. Gibson B, Schneider J, Talamonti D, Forshaw M. The impact of inequality on mental health outcomes during the COVID-19 pandemic: a systematic review. Canadian Psychology-Psychologie Canadienne. 2021;62(1):101–126. doi: 10.1037/cap0000272. [DOI] [Google Scholar]
  24. Gimeno L, Goisis A, Dowd JB, Ploubidis GB. Cohort differences in physical health and disability in the United States and Europe. J Gerontol B Psychol Sci Soc Sci. 2024;79(8) doi: 10.1093/geronb/gbae113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Gimeno L, Moreno-Agostino D, Danka M, Guo Y, Goisis A, Dowd JB, Ploubidis GB. The generational health drift: a systematic review of evidence from the British birth cohort studies. medRxiv. 2025:2025.2006.2012.25329414. doi: 10.1080/00324728.2026.2652038. [DOI] [PubMed] [Google Scholar]
  26. Glonti K, Gordeev VS, Goryakin Y, Reeves A, Stuckler D, McKee M, Roberts B. A systematic review on health resilience to economic crises. PLoS One. 2015;10(4):e0123117. doi: 10.1371/journal.pone.0123117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Gondek D, Bann D, Patalay P, Goodman A, McElroy E, Richards M, Ploubidis GB. Psychological distress from early adulthood to early old age: evidence from the 1946, 1958 and 1970 British birth cohorts. Psychol Med. 2022;52(8):1471–1480. doi: 10.1017/S003329172000327X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Haggar T, Hewlett K, Hall S, Piggott H, Regan Z, Hesketh R, Wojciechowska M, Dacombe R. Breaking Point: the cost-of-living Crisis in London, and what Can be Done About it. 2023 [Google Scholar]
  29. Hill TD, Needham BL. Rethinking gender and mental health: a critical analysis of three propositions. Soc Sci Med. 2013;92:83–91. doi: 10.1016/j.socscimed.2013.05.025. [DOI] [PubMed] [Google Scholar]
  30. Houweling TAJ, Grunberger I. Intergenerational transmission of health inequalities: towards a life course approach to socioeconomic inequalities in health - a review. J Epidemiol Community Health. 2024;78(10):641–649. doi: 10.1136/jech-2022-220162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Institute for Government. Timeline of UK government coronavirus lockdowns. 2021. https://www.instituteforgovernment.org.uk/charts/uk-government-coronavirus-lockdowns .
  32. Kirkbride JB, Anglin DM, Colman I, Dykxhoorn J, Jones PB, Patalay P, Pitman A, Soneson E, Steare T, Wright T, Griffiths SL. The social determinants of mental health and disorder: evidence, prevention and recommendations. World Psychiatry. 2024;23(1):58–90. doi: 10.1002/wps.21160. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Link BG, Phelan J. Social conditions as fundamental causes of disease. J Health Soc Behav. 1995:80–94. Spec No https://core.ac.uk/reader/77145022?utm_source=linkout. [PubMed] [Google Scholar]
  34. Louie P, Wu C, Shahidi FV, Siddiqi A. Inflation hardship, gender, and mental health. SSM Popul Health. 2023;23:101452. doi: 10.1016/j.ssmph.2023.101452. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. McCall L. The complexity of intersectionality. Signs: Journal of women in culture and society. 2005;30(3):1771–1800. [Google Scholar]
  36. McElroy E, Tibber M, Fearon P, Patalay P, Ploubidis GB. Socioeconomic and sex inequalities in parent-reported adolescent mental ill-health: time trends in four British birth cohorts. J Child Psychol Psychiatry Allied Discip. 2023;64(5):758–767. doi: 10.1111/jcpp.13730. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Moreno-Agostino D, Chanfreau J, Knowles G, Pelikh A, Das-Munshi J, Ploubidis GB. Gender inequalities in the disruption of long-term life satisfaction trajectories during the COVID-19 pandemic and the role of time use: evidence from a prospective cohort study. BJPsych Open. 2024;10(6):e217. doi: 10.1192/bjo.2024.817. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Moreno-Agostino D, Fisher HL, Goodman A, Hatch SL, Morgan C, Richards M, Das-Munshi J, Ploubidis GB. Long-term psychological distress trajectories and the COVID-19 pandemic in three British birth cohorts: a multi-cohort study. PLoS Med. 2023;20(4):e1004145. doi: 10.1371/journal.pmed.1004145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Morris S, Phoenix A, Stevenson O. The Cost-of-Living Crisis in the UK: All In It Together? UCL Public Policy, Policy Commission Report. London: UCL; 2023. [Google Scholar]
  40. Muthén LK, Muthén BO. In: Mplus User’s Guide. eighth Muthén M, editor. 2017. [Google Scholar]
  41. Office for National Statistics. Gender pay gap in the UK: 2024. 2024. https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/bulletins/genderpaygapintheuk/2024 .
  42. Office for National Statistics. Consumer price inflation time series. 2025. [Accessed 20 February 2025]. https://www.ons.gov.uk/economy/inflationandpriceindices/datasets/consumerpriceindices .
  43. Patalay P, Gage SH. Changes in millennial adolescent mental health and health-related behaviours over 10 years: a population cohort comparison study. Int J Epidemiol. 2019;48(5):1650–1664. doi: 10.1093/ije/dyz006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Patel K, Robertson E, Kwong ASF, Griffith GJ, Willan K, Green MJ, Di Gessa G, Huggins CF, McElroy E, Thompson EJ, Maddock J, et al. Psychological distress before and during the COVID-19 pandemic among adults in the United Kingdom based on coordinated analyses of 11 longitudinal studies. JAMA Netw Open. 2022;5(4):e227629. doi: 10.1001/jamanetworkopen.2022.7629. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Ploubidis GB, Batty GD, Patalay P, Bann D, Goodman A. Association of early-life mental health with biomarkers in midlife and premature mortality: evidence from the 1958 British birth cohort. JAMA Psychiatry. 2021;78(1):38–46. doi: 10.1001/jamapsychiatry.2020.2893. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Ploubidis GB, McElroy E, Moreira HC. A longitudinal examination of the measurement equivalence of mental health assessments in two British birth cohorts. Longitudinal and life course studies: Int J. 2019;10(4):471–489. doi: 10.1332/175795919X15683588979486. [DOI] [Google Scholar]
  47. Power C, Elliott J. Cohort profile: 1958 British birth cohort (National Child Development Study) Int J Epidemiol. 2006;35(1):34–41. doi: 10.1093/ije/dyi183. [DOI] [PubMed] [Google Scholar]
  48. Ridley M, Rao G, Schilbach F, Patel V. Poverty, depression, and anxiety: causal evidence and mechanisms. Science. 2020;370(6522) doi: 10.1126/science.aay0214. [DOI] [PubMed] [Google Scholar]
  49. Roantree B, Vira K. The rise and rise of women’s employment in the UK. 2018. https://www.ifs.org.uk/publications/5749 .
  50. Rodgers B, Pickles A, Power C, Collishaw S, Maughan B. Validity of the Malaise Inventory in general population samples. Soc Psychiatr Psychiatr Epidemiol. 1999;34(6):333–341. doi: 10.1007/s001270050153. [DOI] [PubMed] [Google Scholar]
  51. Ruckert A, Labonte R. Health inequities in the age of austerity: the need for social protection policies. Soc Sci Med. 2017;187:306–311. doi: 10.1016/j.socscimed.2017.03.029. [DOI] [PubMed] [Google Scholar]
  52. Sedovic M, Sanchez C, Peters A, Brown M, Ploubidis GB, Silverwood R, Wright L, Moulton V, Wielgoszewska B. 1970 British Cohort Study Age 51 Survey, User Guide (Version 1) UCL Centre for Longitudinal Studies. 2024 [Google Scholar]
  53. Silverwood R, Narayanan M, Dodgeon B, Katsoulis M, Ploubidis GB. Handling Missing Data in the CLS Cohort Studies: User Guide. 2024.
  54. Social Metrics Commission. Measuring poverty 2024. A report of the Social Metrics Commission. 2024. https://wpieconomics.com/wp-content/uploads/2024/11/SMC-2024-Report-Web-Hi-Res.pdf?_gl=1*6xy74n*_up*MQ..*_ga*MjA1MTIyMzIyMi4xNzQxNjk5MDk1*_ga_XGV3C75R5J*MTc0MTY5OTA5NC4xLjAuMTc0MTY5OTA5NC4wLjAuMA .
  55. StataCorp. Stata Statistical Software: Release 19. StataCorp LLC; 2025. [Google Scholar]
  56. Stuckler D, Reeves A, Loopstra R, Karanikolos M, McKee M. Austerity and health: the impact in the UK and Europe. Eur J Publ Health. 2017;27(Suppl. l_4):18–21. doi: 10.1093/eurpub/ckx167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Sullivan A, Brown M, Hamer M, Ploubidis GB. Cohort Profile Update: the 1970 British Cohort Study (BCS70) Int J Epidemiol. 2022 doi: 10.1093/ije/dyac148. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Sun Y, Wu Y, Fan S, Dal Santo T, Li L, Jiang X, Li K, Wang Y, Tasleem A, Krishnan A, He C, et al. Comparison of mental health symptoms before and during the covid-19 pandemic: evidence from a systematic review and meta-analysis of 134 cohorts. BMJ. 2023;380:e074224. doi: 10.1136/bmj-2022-074224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Talamonti D, Schneider J, Gibson B, Forshaw M. The impact of national and international financial crises on mental health and well-being: a systematic review. J Ment Health. 2023:1–38. doi: 10.1080/09638237.2023.2278104. [DOI] [PubMed] [Google Scholar]
  60. Taylor-Robinson D, Pickett KE, Bennett D. Reverse the benefit cap to tackle poor child health. BMJ. 2024;387:q2131. doi: 10.1136/bmj.q2131. [DOI] [PubMed] [Google Scholar]
  61. University of York Cost of Living Research Group. Sticking plasters and systemic solutions. Cost of living responses in the UK; 2023. https://www.york.ac.uk/media/policyengine/costoflivingresearchgroup/Cost%20of%20Living%20report%20(5)%20(1).pdf . [Google Scholar]
  62. White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30(4):377–399. doi: 10.1002/sim.4067. [DOI] [PubMed] [Google Scholar]
  63. Wood N, Stafford M, O’Neill D. CLOSER Work Package 9: Harmonised Childhood Environment and Adult Wellbeing Measures User Guide. CLOSER; 2019. [Google Scholar]
  64. World Bank. Inflation, Consumer Prices (Annual %) - United Kingdom. World; 2025. [Accessed 20 February 2025]. https://data.worldbank.org/indicator/FP.CPI.TOTL.ZG?end=2023&locations=GB-1W&start=1960&view=chart . [Google Scholar]
  65. Zaninotto P, Iob E, Di Gessa G, Steptoe A. Recovery of psychological wellbeing following the COVID-19 pandemic: a longitudinal analysis of the English longitudinal study of ageing. Aging Ment Health. 2025:1–9. doi: 10.1080/13607863.2025.2450260. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
EMS214121-supplement-Supplementary_Material.docx (3.5MB, docx)

Data Availability Statement

All used data are freely available to bona fide researchers at the UK Data Service (https://ukdataservice.ac.uk/). Code is publicly available at the Open Science Framework (https://doi.org/10.17605/OSF.IO/MSPRQ).

RESOURCES