Epidemiology and Psychiatric COVID-19 after two years: trajectories of Sciences different components of mental health in the cambridge.org/eps Spanish population I. Bayes-Marin1 , M. Cabello-Toscano1,2 , G. Cattaneo3,4 , Original Article J. Solana-Sánchez3,4 , D. Fernández5,6,7 , C. Portellano-Ortiz1,3 , J. M. Tormos3,4 , A. Pascual-Leone3,8,9 and D. Bartrés-Faz1,2,3 Cite this article: Bayes-Marin I, Cabello- Toscano M, Cattaneo G, Solana-Sánchez J, 1 Fernández D, Portellano-Ortiz C, Tormos JM, Departament de Medicina, Facultat de Medicina i Ciències de la Salut – Campus Clínic, Universitat de Barcelona, Pascual-Leone A, Bartrés-Faz D (2023). COVID- Barcelona, Spain; 2Institut de Recerca Biomèdica August Pi i Sunyer (IDIBAPS), Barcelona, Spain; 3Institut 19 after two years: trajectories of different Guttmann, Institut Universitari de Neurorehabilitació adscrit a la Universitat Autónoma de Barcelona, Barcelona, components of mental health in the Spanish Spain; 4Fundació Institut d’Investigació en Ciències de la Salut Germans Trias i Pujol, Badalona, Spain; 5Instituto population. Epidemiology and Psychiatric de Salud Carlos III, Centro de Investigación Biomédica en Red de Salud Mental, CIBERSAM, Monforte de Lemos 3-5, Sciences 32, e19, 1–13. https://doi.org/10.1017/ Pabellón 11, 28029, Madrid, Spain; 6Serra-Húnter fellow. Department of Statistics and Operations Research (DEIO), S2045796023000136 Universitat Politècnica de Catalunya ⋅ BarcelonaTech (UPC), 08028 Barcelona, Spain; 7Institute of Mathematics of UPC – BarcelonaTech (IMTech), 08028 Barcelona, Spain; 8Hinda and Arthur Marcus Institute for Aging Research Received: 9 September 2022 Revised: 24 February 2023 and Deanna and Sidney Wolk Center for Memory Health, Hebrew SeniorLife, Harvard Medical School, Boston, MA, Accepted: 12 March 2023 USA and 9Department of Neurology, Harvard Medical School, Boston, MA, USA Keywords: Abstract COVID-19; growth mixture models; mental health; trajectories Aims. Our study aimed to (1) identify trajectories on different mental health components during a two-year follow-up of the COVID-19 pandemic and contextualise them according Corresponding author: to pandemic periods; (2) investigate the associations between mental health trajectories and I. Bayes-Marin, E-mail: ivet.bayes@ub.edu several exposures, and determine whether there were differences among the different mental health outcomes regarding these associations. Methods. We included 5535 healthy individuals, aged 40–65 years old, from the Barcelona Brain Health Initiative (BBHI). Growth mixture models (GMM) were fitted to classify indivi- duals into different trajectories for three mental health-related outcomes (psychological dis- tress, personal growth and loneliness). Moreover, we fitted a multinomial regression model for each outcome considering class membership as the independent variable to assess the association with the predictors. Results. For the outcomes studied we identified three latent trajectories, differentiating two major trends, a large proportion of participants was classified into ‘resilient’ trajectories, and a smaller proportion into ‘chronic-worsening’ trajectories. For the former, we observed a lower susceptibility to the changes, whereas, for the latter, we noticed greater heterogeneity and susceptibility to different periods of the pandemic. From the multinomial regression models, we found global and cognitive health, and coping strategies as common protective factors among the studied mental health components. Nevertheless, some differences were found regarding the risk factors. Living alone was only significant for those classified into ‘chronic’ trajectories of loneliness, but not for the other outcomes. Similarly, secondary or higher education was only a risk factor for the ‘worsening’ trajectory of personal growth. Finally, smoking and sleeping problems were risk factors which were associated with the ‘chronic’ trajectory of psychological distress. Conclusions. Our results support heterogeneity in reactions to the pandemic and the need to study different mental health-related components over a longer follow-up period, as each one evolves differently depending on the pandemic period. In addition, the understanding of modifiable protective and risk factors associated with these trajectories would allow the char- acterisation of these segments of the population to create targeted interventions. © The Author(s), 2023. Published by Cambridge University Press. This is an Open Introduction Access article, distributed under the terms of The COVID-19 pandemic posed an extraordinary health, social and economic challenge to the the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), world. Due to the rapid spread of the virus, governments had to implement restrictive policies which permits unrestricted re-use, distribution such as lockdowns or stay-at-home orders (COVID-19 Mental Disorders Collaborators, 2021). and reproduction, provided the original article Although these restrictive policies varied between countries, they affected people’s daily lives is properly cited. globally, in terms of their work, livelihood, leisure activities and social interactions (Prati and Mancini, 2021). In the case of Spain, in the two years following the start of the pandemic, dif- ferent containment measures were put into place, interleaving periods of strict lock-down con- finement (e.g., home confinement, closure of schools and businesses, use of facemasks outdoors/indoors) with those of more relaxed measures (progressive return to work, the 2 I. Bayes‐Marin et al. opening of restaurants and shops, use of facemasks only in some subjective well-being variables (Pellerin et al., 2021) as predictors enclosed spaces, etc.) (Red Nacional de Vigilancia Epidemiológica. of these trajectories. According to Keyes et al. (2020), mental Instituto de Salud Carlos III, 2022). health is a conjunction of emotional (positive and negative A large body of knowledge has been generated regarding the affect and psychological distress), psychological (positive func- impact of the pandemic and confinement in relation to mental tioning variables, as meaning in life, personal growth, autonomy health (Salari et al., 2020; Prati and Mancini, 2021; Wu et al., and environmental mastery) and social wellbeing (social integra- 2021). Whether through cross-sectional or longitudinal studies, tion, social contribution and social acceptance), being more than it has been reported prevalence rates or mean scores of depressive just the absence of psychopathology (Keyes et al., 2020). or anxiety symptoms, assuming that the response to the pandemic Accordingly, it might be hypothesised that we could find changes is homogeneous, i.e., the same among individuals (Shevlin et al., in these other components of mental health. For example, Baños 2023). In contrast, a systematic review based on longitudinal stud- et al. (2022) found in a sample of Spanish residents that the scores ies declared that the effect of lockdowns on depression and anx- on positive functioning variables (meaning in life, gratitude, resili- iety was small and significant, but also highly heterogenous (Prati ence, compassion and life satisfaction) worsened from the begin- and Mancini, 2021). Similarly, a meta-review of mental health ning of the lockdown, whereas emotional distress improved by the during the COVID-19 pandemic, found an increase of mental end of the first Spanish state of alarm (June 21st, 2020) (Baños health problems from 20 to 36%, but also a high heterogeneity et al., 2022). Thus, an in-depth study of the impact of the among studies (de Sousa et al., 2021). It is worth mentioning COVID-19 pandemic on mental health should not be limited to that this evidence come from studies carried out at most up to its effect on psychological distress, but on the different compo- one year after the pandemic, with a lack of studies that have ana- nents of wellbeing affecting mental health. lysed longer-term consequences on mental health. According to Likewise, people classified into different trajectories differed in Taylor (2019), pandemics are dynamic events and as such their terms of several predictors at baseline. As reported in previous reactions were likely to vary over time (Taylor, 2019). For this rea- research, being younger, female, reporting lower income, less edu- son, the results should be contextualised at different times of the cation and having a previous mental health diagnosis, were factors pandemic and the events occurring in each period. In addition, in consistently associated with ‘chronic’ and ‘worsening’ trajectories order to evaluate change from pre-pandemic status, baseline (Pierce et al., 2021; Saunders et al., 2021; Shilton et al., 2021). information is needed, and this condition has been less available Fewer research studies examined modifiable determinants asso- in the performed research (Ahrens et al., 2021; Ellwardt and Präg, ciated with these mental health patterns such as emotion regula- 2021; Pierce et al., 2021). tion, coping strategies and locus of control (Ahrens et al., 2021; In agreement with the assumption that psychological adjust- Joshi et al., 2021; Shilton et al., 2021). ment in front of an adverse event is heterogeneous and may Altogether, the study of the impact of the pandemic on mental vary over time, different studies have been carried out on mental health should take into account the heterogeneity of responses to health trajectories (Ahrens et al., 2021; Batterham et al., 2021; a crisis situation. Prevalence or incidence rates would not be suf- Ellwardt and Präg, 2021; Joshi et al., 2021; Pellerin et al., 2021; ficient to estimate its impact. In this sense, the study of mental Pierce et al., 2021; Saunders et al., 2021; Shilton et al., 2021). health trajectories over a long follow-up would make it possible Most of these studies identified trajectories based on depression to identify subgroups of the population in a situation of greater and anxiety symptoms measures, using individual-centred statis- vulnerability, as well as to visualise the most critical moments tical techniques, as growth mixture models (GMM) or latent of the pandemic. Furthermore, the understanding of modifiable class growth analysis. These techniques rely on the assumption protective and risk factors associated with these trajectories that individuals can be assigned to homogeneous subgroups would allow the characterisation of these segments of the popula- (i.e., distinct trajectories) based on similarities on given outcomes tion to create targeted interventions. The resulting body of knowl- (Nguena Nguefack et al., 2020). The abovementioned investiga- edge would have considerable practical implications for pressing tions identified from two (Joshi et al., 2021) to five trajectories public health efforts. of depression or/and anxiety symptoms (Ahrens et al., 2021; Therefore, this study aimed to (1) identify trajectories based on Batterham et al., 2021; Ellwardt and Präg, 2021; Pellerin et al., different mental health components (emotional, psychological 2021; Pierce et al., 2021; Saunders et al., 2021; Shilton et al., and social wellbeing) during a two-year follow-up of the 2021). In general terms, the results showed that a large proportion COVID-19 pandemic, and contextualise them according to rele- of the sample was classified in a stable trajectory over time (called vant events in each pandemic period; (2) investigate the associa- ‘resilient trajectory’), while a smaller proportion showed worse tions between mental health trajectories and sociodemographic, scores or worsening over the follow-up period (‘chronic’ and personality, coping, subjective well-being and lifestyles variables, ‘deteriorating trajectories’) (Ahrens et al., 2021; Batterham et al., and to determine whether there were differences among the dif- 2021; Ellwardt and Präg, 2021; Joshi et al., 2021; Pellerin et al., ferent mental health outcomes regarding these associations. 2021; Pierce et al., 2021; Saunders et al., 2021; Shilton et al., 2021). These results support the model put forward by Bonanno (2004), which argued that resilience is extremely com- Method mon, finding higher proportions in the so-called ‘resilient’ trajec- Study design and participants tory, where hardly any changes were observed throughout the follow-up in the face of a stressor (Bonanno, 2004). Middle- aged volunteers (40 to 65 years) participating in the Nevertheless, these studies focused on psychological distress as Barcelona Brain Health Initiative (BBHI), an ongoing prospective outcome measure, using mostly sociodemographic variables, and longitudinal study that aims to understand and characterise the in some cases personality (Saunders et al., 2021), loneliness determinants of brain health maintenance, were invited to partici- (Ahrens et al., 2021; Shevlin et al., 2023), coping strategies pate in the current study. Briefly, BBHI study participants are (Joshi et al., 2021; Lin et al., 2021; Pellerin et al., 2021) and community-dwelling individuals, free from any self-reported Epidemiology and Psychiatric Sciences 3 neurological or psychiatric diagnosis at the time of the recruit- symptoms in the last 2 weeks, rated 0 ‘not at all’, 1 ‘several ment, who answer annual questionnaires regarding demographic, days’, 2 ‘more than half the days’, or 3 ‘nearly every day’. A socio-economic, self-perceived health and lifestyles (general score of six or higher represent the cut-off point for a potential health, physical activity, cognitive ability, socialisation, sleep, case of depression/anxiety (Kroenke et al., 2009). However, in nutrition and vital plan) information. The BBHI recruitment our analyses, we used the continuous form where higher scores took place in 2017 through an intensive dissemination campaign mean greater psychological distress. including conferences, radio and TV interviews and social media advertisements. For further details of the cohort and study proto- Psychological col see Cattaneo et al. (Cattaneo et al., 2018). This domain was constituted by ‘personal growth’, one of the The present work refers to a BBHI sub-study designed to positive functioning variables extracted from the Ryff investigate mental health during the COVID-19 pandemic Psychological wellbeing scale (SPWB) (Ryff, 1995; Ryff and (Bartrés-Faz et al., 2021; Pascual-Leone et al., 2021). BBHI parti- Keyes, 1995). SPWB measure consists of 39 items, constituted cipants who had completed the annual questionnaires before the by six sub-scales evaluating six aspects of positive functioning. COVID-19 widespread were invited to participate in subsequent Participants are asked to indicate how accurately each item brief evaluations (March, April, June and October 2020, March, describes themselves by rating on a 5-level Likert scale ranging July and October 2021 and February 2022) during the different from 1 ‘least like me’ to 5 ‘most like me’. Higher scores indicate periods of the COVID-19 pandemic (See Fig. 1). In this sub- better positive functioning. In particular, ‘personal growth’, is study, several measures regarding mental health, subjective well- constituted by seven items and refers to one’s openness to new being, quality of life and coping strategies, were included to experiences and growth. explore the effects of the pandemic on health and well-being. In the present study, we included both the annual general Social follow-up questionnaires and the COVID-19 assessments, consid- Keyes’ social wellbeing definition includes different factors of the ering the observations two years before the pandemic (2018 and subjective evaluation of personal life circumstances and function- 2019 annual questionnaires) as baseline data. We decided not to ing in society, such as social contribution, integration, actualisa- include the 2017 annual questionnaire as we considered the infor- tion, acceptance and coherence. In the present study, we used mation from two points before the pandemic as a good baseline the UCLA 3-Item Loneliness Scale (Rico-Uribe et al., 2016), as on the individual’s mental health status. a proxy measure of social well-being. The UCLA items are related Figure 1 summarises the periods covered by our study (from to social integration since refer to the feeling of being excluded or early 2018 to February 2022), highlighting the time points isolated from others. (Rico-Uribe et al., 2016).In this brief ques- when the questionnaires were launched (orange dots), the rele- tionnaire, respondents were asked how often they felt that they: vant highlights of the pandemic (blue dots) and their correspond- lacked companionship, were left out, and were isolated from ence with the epidemic periods established by the national others, on a 3-level Likert scale coded from 1 ‘hardly ever’, to 3 epidemiological surveillance network of the Carlos III National ‘often’. Higher scores indicate greater loneliness. Health Institute (Red Nacional de Vigilancia Epidemiológica. Instituto de Salud Carlos III, 2022). These periods were defined Exposures by this national epidemiological surveillance network by analysing the evolution of incidence rates in the Spanish population. We included other variables, such as sociodemographic, self- The study was approved by the Catalan Union of Hospitals perceived quality of life and health, lifestyles related to health, ethics committee [Unió Catalana d’Hospitals] (approval refer- among other psychological measures to characterise the mental ences: CEIC 17/06 and CEI 18/07). Moreover, written informed health trajectories. consent was obtained from all participants in accordance with The following sociodemographic variables were considered: the Code of Ethics of the World Medical Association sex (male/female), age (continuous), current marital status (single, (Declaration of Helsinki). married, divorced, widowed), living alone (yes/no), educational level (primary or less, secondary, higher education), occupation (employed, unemployed, retired), monthly family income Measures (<1000€, 1000–2000€, 2000–5000€, >5000€), and if the person Outcomes lives in a town or in a city (town/city). According to Keyes et al. (2020) definition of mental health, we Furthermore, to evaluate self-perceived general health and cog- selected different variables as proxies for the emotional, psycho- nitive function we used the Patient-Reported Outcomes logical and social components. This selection was made according Measurement Information System (PROMIS) of global health to the availability of longitudinal measures including baseline data (Ader, 2007) and the PROMIS Applied Cognition – General and similarity to the constructs assessed (Keyes et al., 2020). Concerns scale (Fieo et al., 2016), respectively. The PROMIS Global Health is composed by ten items representing five Emotional domains (physical function, pain, fatigue, emotional distress, To assess psychological distress, we used the Patient Health social health) that are used to assess global physical health. Questionnaire 4 items (PHQ-4) (Kroenke et al., 2009), a screening Concerning the cognitive function scale, is comprised by eight and accurate measurement of core symptoms or signs of depres- items assessing self-reported cognitive troubles or deficits. In sion (‘be bothered by little interest or pleasure in doing things’, ‘be both measures, higher scores mean better general health and bet- bothered by feeling down, depressed, or hopeless’) and anxiety ter cognitive functioning. (‘feeling nervous, anxious or on edge’, ‘be bothered by not In addition, we included some variables related to lifestyles, as being able to stop or control worrying’). Participants were asked sleeping problems and tobacco consumption. Sleeping problems to indicate how often they have been bothered by four possible (i.e., difficulty to fall asleep, wake up at night) were assessed 4 I. Bayes‐Marin et al. Figure 1. Timing of data acquisition and periods relative to the development of the COVID-19 pandemic in Spain. Note: Timeline showing the periods covered by the present study, according to the epidemic periods in Spain, as defined by the national epidemiological surveillance network of the Carlos III National Health Institute. Questionnaires launching is presented with orange dots, whereas blue dots represent relevant highlights of the pandemic. through the Jenkins Sleep Evaluation Questionnaire, a 4-item were evaluated with the Brief Resilience and Coping Scale questionnaire with scores ranging from 0 (no sleep problems) (BRCS) (Sinclair and Wallston, 2004), where higher scores to 20 (most sleep problems) (Jenkins et al., 1988). Moreover, mean better resilience and coping ability. tobacco coded as yes/no was included in our analyses. Related to this, we added the Engaged Living Scale to assess an We also considered the big five personality traits (extraversion, engaged response style (Trompetter et al., 2013), and three of the emotional stability, agreeableness, conscientiousness and open- six scales from the SPWB: autonomy (a sense of autonomy in ness to experience), assessed via the International Personality thought and action), environmental mastery (the ability to man- Item Pool (Goldberg, 1992). Resilience and coping strategies age complex environments to suit personal needs and values) and Epidemiology and Psychiatric Sciences 5 positive relations with others (the establishment of quality ties to To study the relationship between latent trajectory member- other) (Ryff, 1995). For each of these scales, higher scores are ship and the described exposures, we first fitted univariable mod- indicative of better functioning. els for each outcome variable (online Supplementary Table 4). We Furthermore, perceived stress (the Perceived Stress Scale aimed to explore interactions or possible confounding effects to (Cohen et al., 1983)) was included as a continuous measure. In avoid misinterpretations. Then, we conducted a multinomial this case, higher scores mean worse level of that construct. regression model for each outcome considering class membership as the independent variable to assess the association with several exposures. For each model, the most stable-resilient trajectory was considered the reference category. These multivariable models Statistical analysis were additionally adjusted for sex, age, living alone, monthly fam- We performed a descriptive analysis of the exposures at baseline. ily income and educational level. Due to potential multicollinear- Continuous variables were described by mean ± S.D. values, while ity between some of the exposures we checked the significance categorical variables were presented by the absolute number of and magnitude of correlations through a correlation matrix before individuals and its corresponding percentage (%) within the sam- running the model (online Supplementary Fig. 3). Regression ple. We considered the information extracted from the annual models were run in 20 imputed datasets and results combined questionnaires before the COVID-19 pandemic as the baseline. using Rubin’s rules (Little and Rubin, 2002). In the case of the variables ‘resilience and coping strategies’ and Additional tests were performed to ensure internal consistency ‘perceived stress’, no pre-pandemic data were available. These (Cronbach’s alpha) and intraclass reliability (intraclass correlation two variables were collected in different assessments and to coefficient, ICC) of all the scales in the study, since these were increase the sample size, we considered as baseline the first avail- administered in their translated version (Spanish and Catalan). able observation of each subject on each of these two variables. ICC was only calculated for longitudinal assessments (i.e., To identify mental health trajectories, we first fitted multiple PHQ-4, UCLA-3 and ‘personal growth’ from SPWB) and limited general mixed effects models for each outcome (psychological to pre-pandemic observations. distress, personal growth and feelings of loneliness), to explore All statistical analyses were performed in R version 3.6.2 (R the extent of between-individual heterogeneities (as also recom- Core Team, 2019), and run in RStudio, version 1.3.1093 mended in (Herle et al., 2020)). These models separately can (RStudio Team, 2020). allow the estimation of random intercepts, random slopes or both. In this line, these models were compared using a Results Chi-squared test to find the best design option and do model selection (i.e., the one with the lowest residual sum of squares) In Table 1 are presented the main characteristics of the total sam- (online Supplementary material, Table 2). Second and guided ple (N = 5536) at baseline. Our analytical sample was charac- by the results in the previous step, we fitted a GMM with random terised by higher number of females than males (67.39% vs. intercepts and slopes for each outcome to classify individuals into 32.60%) and by a high proportion of persons with high education latent trajectories based on their score on the outcome variables (70.82%). The mean age was 51.17 (S.D. = 6.93). From the total without covariates (Nagin and Tremblay, 2005; Berlin et al., sample, 14.43% were living alone, 8.83% were unemployed and 2014; Nagin, 2014). The number of trajectories was determined 4.11% had a monthly household income lower than 1000€, by analysing group models from 1 to 5 trajectories. According whereas in 15.93% it was more than 5000€. Moreover, most of to the Bayesian information criterion (BIC) and the Akaike infor- the sample (73.80%) was living in an urban area. All scales mation criterion (AIC), where the lowest value indicates the better showed high internal consistencies (Cronbach’s alpha ranging fit, the optimal model was selected (Schwarz, 1978; Akaike, 1998). from 0.75 to 0.95) and good intraclass reliability (UCLA-3: ICC Moreover, average posterior probabilities above 0.70 were consid- = 0.75, PHQ-4: ICC = 0.75, ‘personal growth’ from SPWD: ICC ered as indicators of optimal fit (Tein et al., 2013; Nylund-Gibson = 0.79). and Choi, 2018). Trajectories sample size was also considered since inadequate sample size (lower than 5% can lead to conver- Mental health trajectories gence problems, insufficient power to identify classes and chan- ging solutions) (Nylund-Gibson and Choi, 2018). The time The first step was to determine the optimal number of latent tra- variable within the GMM was ‘months of the study’, although jectories according to the fit indices (online Supplementary for a clearer presentation of the results, we used the pandemic material Tables from 5 to 7). Although in most outcomes the periods established by the national epidemiological surveillance information criteria (BIC and AIC) pointed to the five- and four- network of the Carlos III National Health Institute when plotting class solutions, the size of the latent classes (<5.00%) and the pos- these. terior probabilities (<0.70), lead these solutions to be discarded. Then, multiple imputation by chained equations was used to Consequently, the 3-class solution provided the best fit. In the deal with missing data in some of the exposures (online case of ‘personal growth’, one of the posterior probabilities was Supplementary Table 3), assuming missing-at-random (MAR), slightly lower than 0.70, but the three-class solution was selected which can handle variables of varying types (Lepkowski et al., to allow comparability with the other outcomes and to explore 2001; van Buuren, 2007). The imputation model included the out- this sub-sample characteristics. come (i.e., trajectories membership) and all the variables In the case of psychological distress (N = 5530, see Fig. 2a), we described in the exposures section, generating 20 imputed datasets identified a trajectory composed by individuals with PHQ-4 (He, 2010). To check imputation quality, we compared imputed scores above the clinical cut-off pre and during the pandemic. and observed data using density and stripplots of van Buuren This sub-group was termed ‘chronic’ trajectory (1: n = 518 and Greenacre (van Buren and Greenacre, 2018) (online (9.36%)) and showed some fluctuations across periods (e.g., Supplementary Figs 1 and 2, respectively). there was a significant increase of psychological distress when 6 I. Bayes‐Marin et al. Table 1. Main characteristics of the sample at baseline the de-escalation plan took place (period 2 > period 1: t = 2.383 p = 0.017) and with the notification of a new variant of Characteristics N = 5536 SARS-CoV-2 (VOC B.1.1.7 -Alpha)) (period 4 > period 3: t = Sex, n (%) 2.869 p = 0.004). Conversely, most individuals showed stable tra- jectories (2: n = 1940 (35.08%) and 3: n = 3072 (55.55%)) across Male 1805 (32.60) the follow-up period. These trajectories differed essentially in Female 3731 (67.39) the intercept, but we considered them as ‘resilient’ trajectories Age, mean (S.D.) 51.17 (6.93) according to Bonanno’s (2004) definition and were named as ‘resilient’ and ‘moderately resilient’, respectively. Marital status, n (%) From the three-trajectories of ‘personal growth’ (N = 5,535, see Married 3358 (60.65) Fig. 2b), one group (3: n = 1996 (36.06%)) was characterised by Single 1015 (18.33) higher levels of this construct (meaning better perception of per- sonal growth), that was sustained over time, so we termed the Divorced 1029 (18.58) ‘resilient’ trajectory. Conversely, we identified another group Widowed 134 (2.42) (‘worsening’ trajectory, 1: n = 423 (7.64%)) that had higher scores Living alone (yes), n (%) 799 (14.43) before the pandemic and that decreased significantly at the first period of the pandemic (i.e., when the Spanish Government Educational level, n (%) declared the state of emergency; period 1 > pre: t = 8.885 p < Primary education or less 248 (4.49) 0.001) and reported a steady and sustained decline over the Secondary education 1367 (24.69) follow-up. Finally, most of the sample (2: n = 3116 (56.29%)) was classified into a group (‘progressively ascending’ trajectory) Higher education 3921 (70.82) characterised by lower scores at baseline with a slight increase Occupation, n (%) during the studied period. However, this change was not signifi- Employed 4492 (81.14) cant and its name was merely descriptive. Finally, of the three trajectories of loneliness (N = 4,066, see Unemployed 489 (8.83) Fig. 2c), two of them (2: ‘chronic – high loneliness’, n = 468 Retired 555 (10.02) (11.51%), and 3: ‘chronic – medium loneliness’, n = 828 Household income, n (%) (20.36%)) showed a similar pattern, such that those with higher scores of perceived loneliness before the pandemic showed a <1000€ 228 (4.11) decrease at the beginning of the pandemic (i.e., when Spanish 1000–2000€ 1238 (22.36) Government declared the state of emergency and lockdown was 2000–5000€ 3188 (57.58) implemented; period 1 > pre: 2 t = 4.331 p < 0.001, 3 t = 10.329 p < 0.001), which increased again in period 2 (when the >5000€ 882 (15.93) de-escalation plan began; period 2 > period 1: 2 t = −4.699 p < Living in a city (yes), n (%) 4086 (73.80) 0.001, 3 t = −1.975 p = 0.048). From the third period on, there Smoking (yes), n (%) 753 (13.60) Cronbach’s was a decrease until the sixth period, where there was newly an α increase in perceived loneliness. Conversely, most of the sample (1: ‘resilient – no loneliness’, n = 2770 (68.12%)) had low and Global health, mean (S.D.) 37.96 (5.62) 0.84 stable scores during the study-period, meaning low perceived Cognitive function, mean (S.D.) 49.23 (8.97) 0.95 loneliness. Sleeping problems, mean (S.D.) 8.69 (4.01) 0.68 In addition, we calculated the proportions of participants clas- sified in the resilient trajectories of each mental health outcome Personality traits, mean (S.D.) and the overlap among them. We aimed to see whether those Extraversion 31.72 (7.08) 0.85 individuals who were resilient in one mental health component Emotional stability 33.64 (7.82) 0.88 were also resilient in the others. Of these results, it should be noted that 65.91% of the participants classified in the trajectory Agreeableness 41.08 (5.20) 0.78 ‘resilient - no loneliness’ were the same individuals as those clas- Conscientiousness 38.31 (6.01) 0.78 sified in the trajectories ‘resilient’ and ‘ moderately resilient’ of the Openness to experience 36.23 (6.03) 0.79 psychological distress variable. Engaged living scale, mean (S.D.) 60.74 (9.35) 0.93 Association between mental health trajectories and exposures Autonomy, mean (S.D.) 47.35 (1.79) 0.75 Environmental mastery, mean (S.D.) 36.17 (2.28) 0.75 To explore possible interactions or confounding effects among the exposure variables, we performed univariable regression models Positive relations with others, mean 27.83 (6.45) 0.84 for each mental health component (online Supplementary (S.D.) Table 4). From these results, highlight the significant associations Brief resilience and coping scale, 15.52 (2.38) 0.75 found in some socio-demographic variables, such as living alone, mean (S.D.) occupation, household income and educational level, smoking, sleep- Perceived stress, mean (S.D.) 17.63 (7.25) 0.88 ing problems and some personality traits. These associations largely Note. The analyses were performed after the multiple imputation, combining 20 imputed disappear in the multivariable models when we adjusted for sex, age, datasets using Rubin’s rules as described in the ‘Statistical Analysis’ section. living alone, monthly family income and educational level. Epidemiology and Psychiatric Sciences 7 Figure 2. Latent trajectories of different components of mental health. Note: The different trajectories were termed as follow: psychological distress (1: ‘chronic’ (n = 518), 2: ‘resilient’ (n = 1,940), and 3: ‘moderately resilient’ (n = 3,072)), personal growth (1: ‘worsening’ (n = 423), 2: ‘progressively ascending’ (n = 3,116), and 3: ‘resilient’ (n = 1,996)), and loneliness (1: ‘resilient – no loneliness’ (n = 2,770), 2: ‘chronic – high loneliness’ (n = 468), and 3: ‘chronic – medium loneliness’ (n = 828)). *Trajectories used as the reference category when multinomial regression models were performed. Blue dots indicate significant changes along the trajectories according to relevant highlights of the pandemic. In particular, we found significant changes in the following periods: period 1 (Spanish Government declared state of emergency), period 2 (beginning of the de-escalation plan), period 3 (Spanish Government declared a new state of emergency), period 4 (notification of a new variant of SARSCoV-2 (VOC B.1.1.7 – Alpha), and started COVID-19 vaccination in Spain), and period 5 (end of the second state of emergency). In Table 2 the significant results from the multinomial regres- were risk factors for being classified in these trajectories. sion models performed for each of the mental health outcomes Among the protective factors, we found better health status, are presented, expressed as relative risk ratios with 95% confi- higher scores on the ‘engagement with life’ and the SPWB ‘posi- dence intervals (CI). We excluded marital status from the analyses tive relations with others’ scales, and in the case of those classified due to a high collinearity (0.72) with the variable living alone into the ‘chronic – high loneliness’, higher scores on the resilience (online Supplementary Fig. 3). and coping strategies scale (BRCS), compared to those classified For psychological distress, females, former smokers, having in the ‘resilient – no loneliness’ class. sleeping problems and higher perceived stress, were risk factors to be classified into the ‘chronic’ trajectory but also for the ‘mod- Discussion erately resilient’ trajectory, compared to those in the ‘resilient’ one. Conversely, higher age, better global health and cognitive Mental health during the COVID-19 pandemic attracted much function, higher emotional stability (personality trait and coping attention, and numerous studies on this topic have been con- strategies (BRCS), were protective factors for the ‘chronic’ and ducted (Salari et al., 2020; Prati and Mancini, 2021; Wu et al., ‘moderately resilient’ trajectories, taking as a reference the ‘resili- 2021). However, the vast majority focused on psychological dis- ent’ class. tress as a measure of mental health (Ahrens et al., 2021; In the case of ‘personal growth’, in addition to some similar- Batterham et al., 2021; Ellwardt and Präg, 2021; Joshi et al., ities, we observed differences in the risk and protective factors 2021; Pellerin et al., 2021; Pierce et al., 2021; Saunders et al., of the ‘worsening’ and ‘progressively ascending’ trajectories, com- 2021; Shilton et al., 2021), which is a conjunction of emotional, pared to the ‘resilient’ class. Regarding similarities, we observed psychological and social components (Keyes et al., 2020). Our that older age was a risk factor, and that variables such as person- objective was to identify mental health trajectories considering ality trait ‘openness to experience’ and higher scores on the BRCS these components as indicators of mental health and to determine (i.e., better resilience and coping strategies) were protective fac- whether they were affected in the same way during the different tors. Concerning the differences, those with lower scores in ‘per- stages of the pandemic. Moreover, we aimed to investigate if the sonal growth’ and who experienced a small increase during associated variables differed or coincided among the different follow-up (‘progressively ascending’ trajectory), also have as pro- trajectories. tective factors a better health status, better cognitive function and For the three outcomes studied (psychological distress, per- higher scores in the SPWB scales of ‘positive relations with others’ sonal growth and feelings of loneliness), we identified three latent and ‘environmental mastery’. Conversely, higher and secondary trajectories. Of these, we differentiated two major trends, a large education were risk factors for those classified in the ‘worsening’ proportion of people who were in ‘resilient’ trajectories (i.e., better trajectory, compared to primary education or less. previous functioning with stable trajectories during the follow-up As for the loneliness results, we observed similarities between period), and a smaller proportion of participants who were part of the two trajectories with high scores (‘chronic – high loneliness’ ‘chronic-worsening’ trajectories (i.e., low functioning and/or with and ‘chronic – medium loneliness’). In both trajectories, variables changes during follow-up). For the ‘resilient’ trajectories, we also such as being a female, living alone, and higher perceived stress observed a lower susceptibility to the changes that occurred in 8 I. Bayes‐Marin et al. Table 2. Results from the multivariable models to explore the association between latent trajectory membership and exposures in the mental health constructs Psychological distress Personal growth Loneliness ‘Moderately ‘Progressively ‘Chronic – high ‘Chronic – medium Variables ‘Chronic’ resilient’ ‘Worsening’ ascending’ loneliness’ loneliness’ Sex Male (ref.) – – – – – – Female 2.59 (1.91–3.51) 1.71 (1.48–1.99) 1.07 (0.14–2.45) 0.82 (0.71–0.95) 1.36 (1.02–1.81) 1.29 (1.06–1.56) Age 0.95 (0.93–0.97) 0.97 (0.96–0.98) 1.02 (0.83–1.37) 1.02 (1.01–1.04) 1.00 (0.98–1.02) 0.99 (0.98–1.00) Living alone No (ref.) – – – – – – Yes 1.32 (0.91–1.91) 0.95 (0.77–1.17) 1.06 (0.77–1.46) 0.91 (0.74–1.11) 3.15 (2.33–4.28) 2.06 (1.63–2.61) Occupation Employed (ref.) – – – – – – Unemployed 1.06 (0.68–1.65) 0.96 (0.74–1.25) 1.40 (0.95–2.07) 1.07 (0.83–1.38) 1.09 (0.71–1.67) 1.12 (0.82–1.52) Retired 0.88 (0.52–1.50) 1.00 (0.77–1.29) 0.95 (0.62–1.45) 0.98 (0.76–1.11) 0.94 (0.59–1.48) 1.27 (0.95–1.72) Household income <1000€ (ref.) – – – – – – 1000–2000€ 0.65 (0.35–1.21) 1.15 (0.77–1.71) 1.40 (0.77–2.57) 1.42 (0.99–2.04) 0.91 (0.53–1.57) 1.02 (0.66–1.58) 2000–5000€ 0.62 (0.34–1.15) 0.92 (0.62–1.36) 1.35 (0.74–2.46) 1.52 (1.06–2.17) 0.56 (0.31–0.98) 0.85 (0.54–1.32) >5000€ 0.73 (0.36–1.46) 0.85 (0.56–1.30) 1.03 (0.53–1.99) 1.40 (0.95–2.07) 0.63 (0.33–1.20) 0.75 (0.46–1.22) Living in a city No (ref.) – – – – – – Yes 1.17 (0.87–1.57) 1.07 (0.92–1.25) 0.88 (0.69–1.12) 1.05 (0.91–1.22) 1.08 (0.82–1.41) 2.06 (0.84–1.22) Educational level Primary education or – – – – – – less (ref.) Secondary education 0.76 (0.42–1.40) 0.94 (0.65–1.37) 2.79 (1.11–7.02) 0.62 (0.43–0.90) 0.93 (0.52–1.65) 0.99 (0.61–1.62) Higher education 0.81 (0.45–1.45) 1.14 (0.79–1.64) 2.55 (1.02–6.36) 0.45 (0.32–0.65) 1.16 (0.66–2.06) 1.09 (0.69–1.74) Global health 0.88 (0.85–0.90) 0.94 (0.92–0.96) 1.00 (0.98–1.03) 0.97 (0.95–0.98) 0.95 (0.92–0.97) 0.96 (0.94–0.98) Cognitive function 0.90 (0.89–0.92) 0.94 (0.93–0.95) 1.00 (0.98–1.02) 0.98 (0.97–0.99) 0.97 (0.96–0.99) 0.99 (0.98–1.00) Smoking No (ref.) – – – – – – Yes 1.96 (1.38–2.79) 1.27 (1.03–1.56) 1.07 (0.78–1.47) 1.06 (0.87–1.29) 1.35 (0.99–1.84) 1.02 (0.80–1.30) Sleeping problems 1.15 (1.11–1.19) 1.07 (1.05–1.10) 0.97 (0.94–1.00) 0.98 (0.96–1.00) 1.01 (0.98–1.04) 1.01 (0.98–1.03) Personality traits Extraversion 1.00 (0.97–1.02) 0.99 (0.98–1.01) 0.99 (0.97–1.01) 0.98 (0.97–0.99) 0.99 (0.97–1.01) 0.99 (0.98–1.01) Emotional stability 0.92 (0.89–0.94) 0.95 (0.93–0.96) 0.98 (0.96–1.00) 0.99 (0.98–1.01) 0.95 (0.93–0.97) 0.97 (0.95–0.98) Agreeableness 1.02 (0.99–1.06) 1.01 (1.00–1.03) 1.00 (0.97–1.03) 0.98 (0.96–1.00) 1.00 (0.97–1.03) 1.00 (0.98–1.02) Conscientiousness 1.00 (0.97–1.03) 0.99 (0.98–1.01) 0.99 (0.97–1.01) 0.99 (0.98–1.01) 0.99 (0.97–1.02) 0.99 (0.97–1.00) Openness to 1.03 (1.00–1.06) 1.00 (0.99–1.02) 0.96 (0.93–0.98) 0.95 (0.94–0.97) 1.04 (1.01–1.06) 1.02 (1.00–1.04) experience Engaged living scale 0.96 (0.94–0.98) 0.98 (0.97–0.99) 0.99 (0.97–1.00) 0.96 (0.95–0.97) 0.97 (0.96–0.99) 0.98 (0.96–0.99) Autonomy 0.98 (0.92–1.04) 0.98 (0.95–1.02) 1.03 (0.98–1.08) 1.00 (0.96–1.04) 1.03 (0.97–1.11) 0.99 (0.94–1.04) Environmental mastery 0.98 (0.92–1.05) 1.00 (0.96–1.04) 0.96 (0.90–1.02) 0.88 (0.84–0.92) 1.00 (0.94–1.07) 1.03 (0.98–1.08) Positive relationships 0.94 (0.92–0.97) 0.98 (0.97–1.00) 0.99 (0.97–1.01) 0.98 (0.96–0.99) 0.86 (0.84–0.88) 0.93 (0.91–0.94) with others (Continued ) Epidemiology and Psychiatric Sciences 9 Table 2. (Continued.) Psychological distress Personal growth Loneliness ‘Moderately ‘Progressively ‘Chronic – high ‘Chronic – medium Variables ‘Chronic’ resilient’ ‘Worsening’ ascending’ loneliness’ loneliness’ Brief resilience and 0.87 (0.81–0.93) 0.96 (0.92–1.00) 0.88 (0.83–0.95) 0.89 (0.85–0.93) 0.91 (0.86–0.97) 0.98 (0.94–1.03) coping scale Perceived stress 1.12 (1.09–1.15) 1.06 (1.04–1.07) 1.01 (0.99–1.03) 1.00 (0.99–1.01) 1.07 (1.05–1.10) 1.04 (1.02–1.05) Note. Relative risk ratios (95% CI) from multinomial logistic regression models. Models were run in 20 imputed datasets and results combined using Rubin’s rules. Models were adjusted for sex, age, living alone, monthly family income, and educational level. Boldface indicates statistically significant results. each period of the pandemic, reaffirming Bonanno’s (2004) model of togetherness that was generated to deal with stay-at-home and the results of research conducted on mental health trajector- orders, such as video calls to family and friends or the ‘20:00 h ies during the COVID-19 pandemic (Ahrens et al., 2021; applause’, where thousands of people applauded frontline health Batterham et al., 2021; Ellwardt and Präg, 2021; Joshi et al., professionals from windows or balconies acknowledging them 2021; Pellerin et al., 2021; Pierce et al., 2021; Saunders et al., their work and commitment. All these aspects may have helped 2021; Shilton et al., 2021). In the case of the so-called ‘chronic- to intensify social bounds, cooperation and a sense of belonging worsening’ trajectories, we observed greater heterogeneity and in the initial stages of the pandemic outbreak. However, in the susceptibility to different periods of the pandemic. For example, present study including a much-extended follow-up assessment, regarding psychological distress (emotional component), those indicated that this initial effect declined after the end of home participants classified in the ‘chronic’ trajectory had higher scores confinement until the initial levels of loneliness were reached at baseline than when the state of alarm was declared (period 1), (Bartrés-Faz et al., 2021). When we characterised these groups and these scores increased at later points in the pandemic (e.g., of individuals (i.e., ‘chronic – high loneliness’ and ‘chronic – period 2, when the de-escalation plan was initiated (‘new normal- medium loneliness’), we observed that they were mostly females, ity’); or period 4, when the Alpha variant was reported). However, people who were living alone and individuals with high perceived in the social component (loneliness variable), those people who stress. Unlike for psychological distress, fewer studies have been felt lonelier before the pandemic (chronic - high loneliness), carried out on loneliness. In much of the research, it has been reduced their scores when the state of alarm was decreed (period used as a predictor of mental health and rarely as an outcome 1) and home confinement was imposed, returning to their previ- (Ahrens et al., 2021; Shevlin et al., 2023). Studies performed in ous scores when the de-escalation and the period of new normal- different countries, that have focused on loneliness during the ity began (period 2). pandemic, have found somewhat controversial results. Some With respect to psychological distress, one possible explan- research found an increase of loneliness during the acute phase ation for the results obtained is that people classified within this of the outbreak (Bu et al., 2020; Luchetti et al., 2020), whereas trajectory already had levels of anxious-depressive symptoms other reported a reduction in perceived loneliness in this phase above the cut-off point before the pandemic, predisposing them (Bartrés-Faz et al., 2021). These findings suggest that the results to higher vulnerability. This explanation is further supported by need to be contextualised, as the effect of the pandemic on lone- the results of the multinomial regression models, where we liness may depend on contextual aspects, such as the restrictions observed higher perceived stress as a risk factor and a negative applied in each country. association with higher scores in resilience and coping strategies, In the case of the psychological dimension of mental health and with the personality trait ‘emotional stability’. Our results (‘personal growth’), we identified fewer changes during follow-up, were in line with previous research. For instance, higher perceived yet some aspects deserve to be mentioned. According to our stress during COVID-19 lockdown was found to be a predictor results, we found that more than half of the sample (those classi- for worse mental health (based on GHQ-28 scores) in a longitu- fied in the ‘progressively ascending’ class) had low scores in ‘per- dinal study conducted in Germany (Ahrens et al., 2021). In sonal growth’, being people with a feeling of personal stagnation the same way, previous mental health diagnosis has been consist- or lack of a sense of improvement or expansion in life. These par- ently associated to ‘chronic’ or ‘worsening’ trajectories (Pierce ticipants experienced an improvement at follow-up, although not et al., 2021; Saunders et al., 2021), which could be extrapolated statistically significant. In contrast, a small proportion of the sam- to the scores above the PHQ-4 cut-off at baseline in our ple (‘worsening’ trajectory) presented a large decrease in scores study. Furthermore, in the investigation conducted by Saunders from the onset of the pandemic (period 1) compared to their et al. (2021), personality traits such as ‘emotional stability’ was baseline scores. Faced with both scenarios, we wondered what also associated with trajectories with worse anxiety scores variables would be associated with these trajectories to character- (based on the GAD-7), in particular trajectories called ‘moder- ise them. In both cases, older age was a risk factor compared to ate/moderately-severe symptoms that become severe over time’ the ‘resilient’ class. This differed from what was found in the lit- and ‘severe initial anxiety that decreases to normal range, pre- erature in studies on emotional distress variables during the pan- dominantly during lockdown’ (Saunders et al., 2021). Taken demic, where younger subjects were more vulnerable (Ellwardt together, all these factors may be acting synergistically posing and Präg, 2021; Pierce et al., 2021; Saunders et al., 2021; these individuals in a more vulnerable situation. Shilton et al., 2021). Nevertheless, a review concerning the impact Regarding loneliness, the decrease in scores in the initial per- of age on mental health changes during the pandemic found het- iod of the pandemic (period 1), was also observed in a previous erogeneous findings in the literature, suggesting that the effect of report by our group, attributing this initial change to the spirit age may depend on contextual variables but also on the mental 10 I. Bayes‐Marin et al. health outcome studied (Lebrasseur et al., 2021). Our study allows From the perspective of practice and policy, our study provides to contextualise these findings in terms of a particular age group useful information for risk identification. Our research allows to (40 to 65 years) and one of the domains of mental health. In add- identify and characterise groups of more resilient people and ition, both trajectories (‘worsening’ and ‘progressively ascending’), others who are in a situation of chronicity or vulnerability. had in common higher resilience and coping strategies, and the Furthermore, the fact that we have separated different aspects of personality trait ‘openness to experience’ as positive factors asso- mental health (psychological distress, personal growth and feel- ciated to these trajectories. This could be translated into a lower ings of loneliness) and contextualised the fluctuations by consid- adaptive capacity as well as a tendency towards conservativeness ering the relevant events of the pandemic, makes our study of and less openness to experience. However, they differed in a potential great value. In this sense, it allows for the detection of lower risk of being classified in the ‘progressively ascending’ key temporal moments in which to target interventions to stra- class in the case of better self-reported health, better cognition tegically prevent to promote a better emotional, psychological and higher scores in ‘positive relations with others’ and ‘environ- and social status. This knowledge could be extrapolated to the mental mastery’, i.e., quality ties to others and the ability to man- current situation, where other social and economic threats have age complex situations, respectively. increased, such as the rising price of basic needs (electricity, gas Given these results, and with the calculation of the overlapping and food), inflation and eventual recession. Exposure to these fac- of individuals classified in trajectories considered ‘resilient’, we tors could affect people’s health, and the results of these studies reaffirm our initial hypothesis that the different components of could be used to guide preventive strategies. mental health should be analysed separately. We found that within so-called ‘resilience’ there was also heterogeneity, as the Strengths and limitations proportion of overlapping in the ‘resilient’ individuals among out- comes was only above 50% for psychological distress and loneli- The strengths of the present work include a two-year follow-up ness, while for ‘personal growth’ and loneliness it was 26.29%. from the start of the COVID-19 pandemic and the inclusion of The greatest overlap, that was found between loneliness and emo- baseline information. To the best of our knowledge, no previous tional distress, was consistent with that reported in the literature, study has carried out such a long follow-up (Ahrens et al., where both variables have been consistently related (Bu et al., 2021; Batterham et al., 2021; Ellwardt and Präg, 2021; Joshi 2020; Ahrens et al., 2021). Moreover, each outcome was suscep- et al., 2021; Pellerin et al., 2021; Pierce et al., 2021; Saunders tible to different stages of the pandemic and the variables asso- et al., 2021; Shilton et al., 2021). In our study, we analysed data ciated with the trajectories presented some differences. These considering the previous two years as the baseline, until variations included that living alone was only a significant risk February 2022, when the large expansion of the Omicron variant factor for loneliness (‘chronic-high/medium loneliness’ trajector- occurred. This is particularly important because, according to ies), but not for the other outcomes. Likewise, monthly household Taylor, pandemics are dynamic events and therefore changes in income was only related to one of the trajectories of ‘personal mental health outcomes are expected to occur over time, includ- growth’ in the adjusted models. Furthermore, lifestyles such as ing a return to baseline levels (Taylor, 2019). This could be smoking behaviour and sleeping problems were associated with observed with a long follow-up and not just at the beginning of the ‘chronic’ class of the psychological distress measure, which the pandemic when lockdown and other covid measures were could be related to a maladaptive strategy and a consequence of implemented. Furthermore, we interpreted the fluctuations in experienced distress, respectively. For the same class, predictors the trajectories in terms of the periods of greatest interest for as ‘emotional stability’ and perceived stress, well-known distress- the pandemic, contextualising the changes in the analysed mental related variables, were found to be risk factors also for the health outcomes, suggesting that certain changes might be related ‘chronic’ trajectories of loneliness, but not for ‘personal growth’. to the events taking place in each covid period. This made our In addition, from the analysis of the variables associated with study a richer investigation as it was not limited to two major per- the different trajectories, we also observed some similarities. iods (e.g., pre-covid/covid or lockdown/new normality), but Predictors such as better overall health and better cognitive allowed us to observe the evolution of psychological, emotional function were protective factors in all of the studied variables. and social outcomes at different points and to identify the most The relationship between physical and mental health status has critical moments of the pandemic. Moreover, as we mentioned been commonly reported in the literature, suggesting a bidirec- earlier, we identified trajectories based on proxy measures of tional relationship (Druss and Walker, 2011). Likewise, anxious- different components of mental health, not just psychological dis- depressive symptomatology has been widely recognised as a risk tress, since mental health is more than the absence of anxious- factor for cognitive impairment (Chodosh et al., 2010; depressive symptoms. Therefore, the approach of our study was Zaninotto et al., 2018). Similarly, the personality trait ‘openness under Keyes et al.’s (2020) definition of mental health and consid- to experience’, and some SPWB scales (‘engaged living scale’ ered emotional, psychological and social elements as indicators of and ‘positive relations with others’) were positively associated mental health (Keyes et al., 2020). The fact that we found differ- with better mental health outcomes (i.e., ‘resilient’ trajectories). ences in trajectories and associated variables among mental health Finally, emphasise the role of coping strategies, as it was positively outcomes reinforces our hypothesis and the need for more holistic associated with those trajectories with better functioning in all the studies on mental health. Finally, the inclusion of several predic- analysed outcomes. Previous research found frequent use of dys- tors, such as socio-demographic variables, personality traits, some functional coping strategies and less frequent use of emotion- lifestyles and variables regarding subjective well-being and coping focused coping strategies in those participants classified into the strategies, provided a good overview of the risk and protective trajectory ‘high-increasing depressive symptoms’ (Joshi et al., factors that characterise each of the trajectories. 2021). The role of coping strategies is of particular interest as it However, some limitations deserve to be mentioned. First, we is a modifiable factor, which can be trained and serve as a pre- did not use a random sample and it could have introduced some ventive strategy for future crises. bias limiting the sample representativeness and result Epidemiology and Psychiatric Sciences 11 generalisability. For example, there was an oversampling of [PID2019-104830RB-I00/ DOI (AEI): 10.13039/501100011033].. J.M.T. was females and participants with higher education. Ideally, we should partly supported by AGAUR (2018 PROD 00172), Fundació Joan Ribas have fitted the models in a randomised design, but such design is Araquistain and ‘La Marató de TV3’ Fundation (201735.10). This research was furthermore supported by the Government of Catalonia (2017SGR748). not possible to pursue in the current context. Future research We also acknowledge support from the Spanish Ministry of Science and could use post-randomisation techniques based on matching or Innovation and State Research Agency through the ‘Centro de Excelencia weighting-based random sampling methods that specifically Severo Ochoa 2019-2023’ Program (CEX2018-000806-S), and support from target potentially varying background characteristics. Secondly, the Generalitat de Catalunya through the CERCA Program. there were differences in the number of observations among periods and variables collected. This fact, although inherent to a Conflict of interest. None. longitudinal study, entailed a large number of missingness in Ethical standards. The authors assert that all procedures contributing to most of the predictors, so multiple imputation procedures were this work comply with the ethical standards of the relevant national and insti- performed. In our case the complete case analysis could not be tutional committees on human experimentation and with the Helsinki considered due to a drastic reduction of the sample size. Declaration of 1975, as revised in 200. Nevertheless, the use of multiple imputation procedures is widely advocated when missing data occur in one or more covariates in a Availability of data and materials. The authors encourage interested regression model and under an MAR assumption, and in order to investigators to reach out and we will honour all reasonable and scientifically ensure the quality of the imputed data, all necessary diagnostics motivated requests for data access and make the raw data available when required. were performed (Sterne et al., 2009; White and Carlin, 2010). Thirdly, despite having longitudinal information on some of the exposure variables, multinomial regression models included only baseline scores. Some of these variables, such as occupation, References sleep problems, resilience and coping strategies and perceived Ader DN (2007) Developing the Patient-Reported Outcomes Measurement stress, might have changed during follow-up. Due to differences Information System (PROMIS). Medical Care 45, S1–S2. in the number of observations and the period of collection of Ahrens KF, Neumann RJ, Kollmann B, Brokelmann J, von Werthern NM, each variable, longitudinal analysis was discarded. However, Malyshau A, Weichert D, Lutz B, Fiebach CJ, Wessa M, Kalisch R, future studies should consider analysing the exposure variables Plichta MM, Lieb K, Tüscher O and Reif A (2021) Impact of COVID-19 lockdown on mental health in Germany: longitudinal observa- longitudinally, as their possible changes could explain part of tion of different mental health trajectories and protective factors. the results found. Finally, the identification of trajectories in Translational Psychiatry 11, 1–10. two of the mental health components was based on screening Akaike H (1998) Information theory and an extension of the maximum like- measures, such as the PHQ-4 and the UCLA-3. While much of lihood principle. In Parzen E, Tanabe K and Kitagawa G (eds), Selected the research in this field has used these or similar measures (Bu Papers of Hirotugu Akaike. New York: Springer, pp. 199–213. et al., 2020; Fancourt et al., 2021; Pierce et al., 2021; Saunders Baños RM, Garcés JJ, Miragall M, Herrero R, Vara MD and Soria-Olivas E et al., 2021; Shevlin et al., 2023), researchers and policymakers (2022) Exploring the heterogeneity and trajectories of positive functioning should be aware of the accuracy limitations with such tools, and variables, emotional distress, and post-traumatic growth during strict con- interpret the results with caution. finement Due to COVID-19. Journal of Happiness Studies 23, 1683–1708. Bartrés-Faz D, Macià D, Cattaneo G, Borràs R, Tarrero C, Solana J, Tormos Supplementary material. The supplementary material for this article can JM and Pascual-Leone A (2021) The paradoxical effect of COVID-19 out- be found at https://doi.org/10.1017/S2045796023000136 break on loneliness. BJPsych Open 7, e30. Batterham PJ, Calear AL, McCallum SM, Morse AR, Banfield M, Farrer Acknowledgments. Authors thank the Barcelona Brain Health Initiative LM, Gulliver A, Cherbuin N, Rodney Harris RM, Shou Y and Dawel A project participants, whose invaluable contribution made this research (2021) Trajectories of depression and anxiety symptoms during the possible. COVID-19 pandemic in a representative Australian adult cohort. The Medical journal of Australia 214, 462–468. Authors contributions. G.C., J.S-S., J.M.T., A.P-L., D. B-F. conceived and Berlin KS, Parra GR and Williams NA (2014) An introduction to latent vari- designed the study. I.B-M., M.C-T., G.C. and J.S-S. had full access to all of able mixture modeling (Part 2): longitudinal latent class growth analysis the study data, verified the data, and take responsibility for the integrity of and growth mixture models. Journal of Pediatric Psychology 39, 188–203. the data and the accuracy of the data analysis. I.B-M. and M.C-T. performed Bonanno GA (2004) Loss, trauma, and human resilience: have we underesti- the analyses. I.B-M. and M.C-T. wrote the first draft of the paper and incor- mated the human capacity to thrive after extremely aversive events? The porated revisions. I.B-M., M.C-T. and D. B-F. contributed to the interpretation American Psychologist 59, 20–28. of results. I.B-M., M.C-T., G.C., J.S-S., D.F., C.P-O., J.M.T., A.P-L., D. B-F. crit- Bu F, Steptoe A and Fancourt D (2020) Loneliness during a strict lockdown: ically revised and approved the final manuscript. trajectories and predictors during the COVID-19 pandemic in 38,217 United Kingdom adults. Social Science & Medicine 265, 113521. Financial support. This work was supported by a grant from the Agència de Cattaneo G, Bartrés-Faz D, Morris TP, Solana Sánchez J, Macià D, Tarrero Gestió d’Ajuts Universitaris i de Recerca (AGAUR) ‘PANDÈMIES 2020’ (ref. C, Tormos JM and Pascual-Leone A (2018) The Barcelona brain health 2020PANDE00043) and a grant from ‘La Marató de TV3’ MARATÓ 2020 initiative: a cohort study to define and promote determinants of brain COVID-19 (ref. 202129–31). Supported in part by the Spanish Ministry of health. Frontiers in Aging Neuroscience 10, 321. Science, Innovation and Universities (MICIU/FEDER; grant number Chodosh J, Miller-Martinez D, Aneshensel CS, Wight RG and Karlamangla RTI2018-095181-B-C21) and an ICREA Academia 2019 grant award to AS (2010) Depressive symptoms, chronic diseases, and physical disabilities D. B-F. Partially, this research has received funding from ‘La Caixa’ as predictors of cognitive functioning trajectories in older Americans. Foundation (grant number LCF/PR/PR16/11110004), and from Institut Journal of the American Geriatrics Society 58, 2350–2357. Guttmann and Fundació Abertis. I.B-M. was supported by a postdoctoral fel- Cohen S, Kamarck T and Mermelstein R (1983) A global measure of per- lowship related to ‘PANDÈMIES 2020’ (AGAUR; 2020PANDE00043). D.F. ceived stress. Journal of Health and Social Behavior 24, 385–396. has been supported by grant 2021 SGR 01421 (GRBIO) administrated by COVID-19 Mental Disorders Collaborators (2021) Global prevalence and the Departament de Recerca I Universitats de la Generalitat de Catalunya burden of depressive and anxiety disorders in 204 countries and territories (Spain) and by the Ministerio de Ciencia e Innovación (Spain) in 2020 due to the COVID-19 pandemic. The Lancet 398, 1700–1712. 12 I. Bayes‐Marin et al. de Sousa Jr GM, Tavares VDdO, de Meiroz Grilo MLP, Galvão Coelho ML, Pascual-Leone A, Cattaneo G, Macià D, Solana J, Tormos JM and de Lima-Araújo GL, Barreto Schuch F and Galvão-Coelho NL (2021) Bartrés-Faz D (2021) Beware of optimism bias in the context of the Mental health in COVID-19 pandemic: a meta-review of prevalence COVID-19 pandemic. Annals of Neurology 89, 423–425. meta-analyses. Frontiers in Psychology 12, 703838. Pellerin N, Raufaste E, Corman M, Teissedre F and Dambrun M (2021) Druss BG and Walker ER (2011) Mental disorders and medical comorbidity. Mental health trajectories during the COVID-19 lockdown: evidence for The Synthesis Project Research Synthesis Report 21, 1–26. resilience and the role of psychological resources and flexibility. Scientific Ellwardt L and Präg P (2021) Heterogeneous mental health development dur- Reports 12, 10674. ing the COVID-19 pandemic in the United Kingdom. Scientific Reports 11, Pierce M, McManus S, Hope H, Hotopf M, Ford T, Hatch SL, John A, 15958. Kontopantelis E, Webb RT, Wessely S and Abel KM (2021) Fancourt D, Steptoe A and Bu F (2021) Trajectories of anxiety and depressive Mental health responses to the COVID-19 pandemic: a latent class tra- symptoms during enforced isolation due to COVID-19 in England: a longi- jectory analysis using longitudinal UK data. The Lancet Psychiatry 8, tudinal observational study. The Lancet Psychiatry 8, 141–149. 610–619. Fieo R, Ocepek-Welikson K, Kleinman M, Eimicke JP, Crane PK, Cella D Prati G and Mancini AD (2021) The psychological impact of COVID-19 pan- and Teresi JA (2016) Measurement equivalence of the Patient Reported demic lockdowns: a review and meta-analysis of longitudinal studies and Outcomes Measurement Information System® (PROMIS®) applied cogni- natural experiments. Psychological Medicine 51, 201–211. tion – general concerns, short forms in ethnically diverse groups. R Core Team (2019) R: A Language and Environment for Statistical Psychological test and assessment modeling 58, 255–307. Computing. Vienna, Austria: R Foundation for Statistical Computing. Goldberg LR (1992) The development of markers for the big-five factor struc- Red Nacional de Vigilancia Epidemiológica. Instituto de Salud Carlos III ture. Psychological Assessment 4, 26–42. (2022) Informes COVID-19. Available at https://www.isciii.es/QueHacemos/ He Y (2010) Missing data analysis using multiple imputation: getting to the Servicios/VigilanciaSaludPublicaRENAVE/EnfermedadesTransmisibles/ heart of the matter. Circulation Cardiovascular Quality and Outcomes 3, Paginas/InformesCOVID-19.aspx (Accessed 13 May 2022). 98–105. Rico-Uribe LA, Caballero FF, Olaya B, Tobiasz-Adamczyk, Koskinen S, Herle M, Micali N, Abdulkadir M, Loos R, Bryant-Waugh R, Hübel C, Leonardi M, Haro JM, Chetterji S, Ayuso-Mateos JL and Miret M Bulik CM and De Stavola BL (2020) Identifying typical trajectories in lon- (2016) Loneliness, social networks, and health: a cross-sectional study in gitudinal data: modelling strategies and interpretations. European Journal of three countries. PLOS ONE 11, e0145264. Epidemiology 35, 205–222. RStudio Team (2020) RStudio: Integrated Development Environment for R. Jenkins CD, Stanton BA, Niemcryk SJ and Rose RM (1988) A scale for the Boston, MA: RStudio, Inc. estimation of sleep problems in clinical research. Journal of Clinical Ryff CD (1995) Psychological well-being in adult life. Current Directions in Epidemiology 4, 313–321. Psychological Science 4, 99–104. Joshi D, Gonzalez A, Griffith L, Duncan L, MacMillan H, Kimber M, Ryff CD and Keyes CLM (1995) The structure of psychological well-being Vrkljan B, Mackillop J, Beauchamp M, Kates N and Raina P (2021) revisited. Journal of Personality and Social Psychology 69, 719–727. The trajectories of depressive symptoms among working adults during Salari N, Hosseinian-Far A, Jalali R, Vaisi-Raygani A, Rasoulpoor S, the COVID-19 pandemic: a longitudinal analysis of the InHamilton Mohammadi M, Rasoulpoor S and Khaledi-Paveh B (2020) Prevalence COVID-19 study. BMC Public Health 21, 1–10. of stress, anxiety, depression among the general population during the Keyes CLM, Yao J, Hybels CF, Milstein G and Proeschold-Bell RJ (2020) Are COVID-19 pandemic: a systematic review and meta-analysis. changes in positive mental health associated with increased likelihood of Globalization and Health 16, 57. depression over a two year period? A test of the mental health promotion Saunders R, Buckman JEJ, Fonagy P and Fancourt D (2021) Understanding and protection hypotheses. Journal of Affective Disorders 270, 136–142. different trajectories of mental health across the general population during Kroenke K, Spitzer RL, Williams JBW and Löwe B (2009) An ultra-brief the COVID-19 pandemic. Psychological Medicine 52, 1–9. screening scale for anxiety and depression: the PHQ-4. Psychosomatics Schwarz G (1978) Estimating the dimension of a model. The Annals of 50, 613–621. Statistics 6, 461–464. Lebrasseur A, Fortin-Bédard N, Lettre J, Raymond E, Bussières EL, Lapierre Shevlin M, Butter S, McBride O, Murphy J, Gibson-Miller J, Hartman TK, N, Faieta J, Vincent C, Duchesne L, Ouellet MC, Gagnon E, Tourigny A, Levita L, Maon L, Martinez AP, Mckay R, Stocks TVA, Bennett K, Lamontagne ME and Routier F (2021) Impact of the COVID-19 pandemic Hyland P and Bentall RP (2023) Refuting the myth of a ‘tsunami’ of men- on older adults: rapid review. JMIR Aging 4, e26474. tal ill-health in populations affected by COVID-19: evidence that response Lepkowski J, Raghunathan TE, Lepkowski JM and Solenberger PW (2001) to the pandemic is heterogeneous, not homogeneous. Psychological A multivariate technique for multiply imputing missing values using a Medicine 53, 429–437. sequence of regression models. Survey Methodology 27, 85–95. Shilton T, Mancini AD, Perlstein S, Didomenico GE, Visoki E, Greenberg Lin T, Yi Z, Zhang S and Veldhuis CB (2021) Predictors of psychological dis- DM, Brown LA, Gur RE, Waller R and Barzilay R (2021) Prospective pre- tress and resilience in the post-COVID-19 Era. International Journal of dictors of risk and resilience trajectories during the early stages of the Behavioral Medicine 29, 506–516. COVID-19 pandemic: a longitudinal study. [Preprint] Available at https:// Little RJA and Rubin DB (2002) Statistical Analysis with Missing Data. www.medrxiv.org/content/10.1101/2021.10.08.21264752v1 (Accessed on 1 Hoboken: John Wiley & Sons. June 2022). Luchetti M, Lee JH, Aschwanden D, Sesker A, Strickhouser JE, Terracciano Sinclair VG and Wallston KA (2004) The development and psychometric A and Sutin AR (2020) The trajectory of loneliness in response to evaluation of the brief resilient coping scale. Assessment 11, 94–101. COVID-19. The American Psychologist 75, 897–908. Sterne JAC, White IR, Carlin JB, Royston P and Carpenter JR (2009) Nagin DS (2014) Group-based trajectory modeling: an overview. Annals of Multiple imputation for missing data in epidemiological and clinical Nutrition and Metabolism 65, 205–210. research: potential and pitfalls. BMJ 338, b2393. Nagin DS and Tremblay RE (2005) Developmental trajectory groups: fact or Taylor S (2019) The Psychology of Pandemics: Preparing for the Next Global a useful statistical fiction? Criminology: An Interdisciplinary Journal 43, Outbreak of Infectious Disease. Newcastle: Cambridge Scholars Publishing. 873–904. Tein JY, Coxe S and Cham H (2013) Statistical power to detect the correct Nguena Nguefack HL, Pagé MG, Katz J, Choinière M, Vanasse A, Dorais M, number of classes in latent profile analysis. Structural Equation Modeling Samb OM and Lacasse A (2020) Trajectory modelling techniques useful to 20, 640–657. epidemiological research: a comparative narrative review of approaches. Trompetter HR, ten Klooster PM, Schreurs KMG, Fledderus M, Westerho GJ Clinical Epidemiology 12, 1205–1222. and Bohlmeijer ET (2013) Measuring values and committed action with the Nylund-Gibson K and Choi AY (2018) Ten frequently asked questions about Engaged Living Scale (ELS): psychometric evaluation in a nonclinical sample latent class analysis. Translational Issues in Psychological Science 4, 440–461. and a chronic pain sample. Psychological Assessment 25, 1235–1246. Epidemiology and Psychiatric Sciences 13 van Buren S and Greenacre M (2018) Flexible Imputation of Missing Data. Wu T, Jia X, Shi H, Niu J, Yin X, Xie J and Wang X (2021) Prevalence London: Chapman & Hall/CRC. of mental health problems during the COVID-19 pandemic: a system- van Buuren S (2007) Multiple imputation of discrete and continuous data by atic review and meta-analysis. Journal of Affective Disorders. 281, fully conditional specification. Statistical Methods in Medical Research 16, 91–98. 219–242. Zaninotto P, Batty GD, Allerhand M and Deary IJ (2018) Cognitive function White IR and Carlin JB (2010) Bias and efficiency of multiple imputation trajectories and their determinants in older people: 8 years of follow-up in compared with complete-case analysis for missing covariate values. the English longitudinal study of ageing. Journal of Epidemiology and Statistics in Medicine 29, 2920–2931. Community Health 72, 685–694.