Epidemiology and Psychiatric Sciences Mental impact of Covid-19 among Spanish cambridge.org/eps healthcare workers. A large longitudinal survey Original Article *Jordi Alonso and Gemma Vilagut should be considered joint first authors and Víctor Pérez-Sola and Philippe J. Alonso1,2,3, * , G. Vilagut1,2, *, I. Alayo1,2, M. Ferrer1,2,3, F. Amigo1,2, Mortier should be considered joint senior authors. A. Aragón-Peña4,5, E. Aragonès6,7, M. Campos8, I. del Cura-González5,9,10, † The MINDCOVID Working Group is formed by: I. Urreta2,11, M. Espuga12, A. González Pinto13,14, J. M. Haro14,15,16, Jordi Alonso, Itxaso Alayo, Manuel Alonso, Mar Álvarez, N. López Fresneña17, A. Martínez de Salázar18, J. D. Molina14,19,20,21, Benedikt Amann, Franco F. Amigo, Gerard Anmella, Andres Aragón, Nuria Aragonés, Enric Aragonès, Ana Isabel Arizón, R. M. Ortí Lucas14,22, M. Parellada14,17, J. M. Pelayo-Terán14,23,24, Angel Asunsolo, Alfons Ayora, Laura Ballester, Puri Barbas, Josep Basora, Elena Bereciartua, Inés Bravo Ignasi Bolibar, A. Pérez Zapata25, J. I. Pijoan2,26, N. Plana2,25, M. T. Puig16,27,28,29, C. Rius2,30, Xavier Bonfill, Alberto Cotillas, Andres Cuartero, Concha de Paz, Isabel del Cura, Maria Jesus del Yerro, Domingo Diaz, C. Rodriguez-Blazquez31,32, F. Sanz3,33,34, C. Serra2,35,36, R. C. Kessler37, Jose Luis Domingo, Jose I. Emparanza, Mireia Espallargues, Meritxell Espuga, Patricia Estevan, M. Isabel R. Bruffaerts38, E. Vieta14,39, V. Pérez-Solá14,16,35, *, P. Mortier1,2, * Fernandez, Tania Fernandez, Montse Ferrer, Yolanda and MINDCOVID Working group1, † Ferreres, Giovanna Fico, M. Joao Forjaz, Rosa Garcia Barranco, J. Manuel Garcia TorrecillasC. Garcia-Ribera, 1 Araceli Garrido, Elisa Gil, Marta Gomez, Javier Gomez, Ana Health Services Research Unit, IMIM-Institut Hospital del Mar d’Investigacions Mèdiques, Barcelona, Spain; 2CIBER Gonzalez Pinto, Josep Maria Haro, Margarita Hernando, Epidemiología y Salud Pública (CIBERESP), Madrid, Spain; 3Department of Medicine and Life Sciences, Universitat Maria Giola Insigna, Milagros Iriberri, Nuria Jimenez, Xavi Pompeu Fabra, Barcelona, Spain; 4Epidemiology Unit, Regional Ministry of Health, Community of Madrid, Madrid, Jimenez, Amparo Larrauri, Fernando Leon, Nieves Lopez- Spain; 5Fundación Investigación e Innovación Biosanitaria de AP, Comunidad de Madrid, Madrid, Spain; 6Institut Fresneña, Carmen Lopez, Mayte Lopez-Atanes Juan d’Investigació en Atenció Primària IDIAP Jordi Gol, Barcelona, Spain; 7Atenció Primària Camp de Tarragona, Antonio Lopez-Rodriguez, German Lopez-Cortacans, Alba Institut Català de la Salut, Tortosa, Spain; 8Service of Prevention of Labor Risks, Medical Emergencies System, Marcos, Jesus Martin, Vicente Martin, Mercedes Martinez- Generalitat de Catalunya, Barcelona, Spain; 9Research Unit, Primary Care Management, Madrid Health Service, Cortés, Raquel Martinez-Martinez, Alma D. Martinez de Madrid, Spain; 10Department of Medical Specialities and Public Health, King Juan Carlos University, Madrid, Spain; Salazar, Isabel Martinez, Marco Marzola, Nelva Mata, Josep 11 Hospital Universitario Donostia, San Sebastián, Spain; 12Occupational Health Service, Hospital Universitari Vall Maria Molina, Juan de Dios Molina, Emilia Molinero, Philippe Mortier, Carmen Muñoz, Andrea Murru, Jorge d’Hebron, Barcelona, Spain; 13Hospital Universitario Araba-Santiago, Vitoria-Gasteiz, Spain; 14CIBER Salud Mental Olmedo, Rafael M Ortí, Rafael Padrós, Meritxell Pallejà, (CIBERSAM), Madrid, Spain; 15Parc Sanitari Sant Joan de Déu, Barcelona, Spain; 16Universitat Autònoma de Raul Parra, Julio Pascual, Jose Maria Pelayo, Rosa Pla, Barcelona (UAB), Barcelona, Spain; 17Hospital General Universitario Gregorio Marañón, Madrid, Spain; 18UGC Nieves Plana, Coro Perez Aznar, Beatriz Perez Gomez, Salud Mental, Hospital Universitario Torrecárdenas, Almería, Spain; 19Villaverde Mental Health Center, Clinical Aurora Perez Zapata, Jose Ignacio Pijoan, Elena Management Area of Psychiatry and Mental Health, Psychiatric Service, Hospital Universitario 12 de Octubre, Polentinos, Beatriz Puertolas, Maria Teresa Puig, Alex Madrid, Spain; 20Research Institute Hospital 12 de Octubre (i+12), Madrid, Spain; 21Faculty of Health Sciences, Quílez, M. Jesus Quintana, Antonio Quiroga, David Rentero, Francisco de Vitoria University, Madrid, Spain; 22Hospital Clínic Universitari, Valencia, Spain; 23Servicio de Cristina Rey, Cristina Rius, Carmen Rodriguez-Blazquez, Psiquiatría y Salud Mental, Hospital el Bierzo, Gerencia de Asistencia Sanitaria del Bierzo (GASBI), Gerencia M. Jose Rojas, Yamina Romero, Gabriel Rubio, Mercedes Regional de Salud de Castilla y Leon (SACYL), Ponferrada, León, Spain; 24Area de Medicina Preventiva y Salud Rumayor, Pedro Ruiz, Margarita Saenz, Jesus Sanchez, Ignacio Sanchez-Arcilla, Ferran Sanz, Consol Serra, Victoria Pública, Universidad de León, León, Spain; 25Príncipe de Asturias University Hospital, Alcalá de Henares, Madrid, Serra-Sutton, Manuela Serrano, Silvia Sola, Sara Solera, Spain; 26Hospital Universitario Cruces/OSI EEC, Bilbao, Spain; 27Department of Epidemiology and Public Health, Miguel Soto, Alejandra Tarrago, Natividad Tolosa, Mireia Hospital de la Santa Creu i Sant Pau, Barcelona, Spain; 28Biomedical Research Institute Sant Pau (IIB Sant Pau), Vazquez, Margarita Viciola, Eduard Vieta, Gemma Vilagut, Barcelona, Spain; 29CIBER Enfermedades Cardiovasculares (CIBERCV), Madrid, Spain; 30Agència de Salut Pública Sara Yago, Jesus Yañez, Yolanda Zapico, Luis Maria Zorita, de Barcelona, Barcelona, Spain; 31National Center of Epidemiology, Instituto de Salud Carlos III (ISCIII), Madrid, Iñaki Zorrilla, Saioa L. Zurbano, and Victor Perez-Solá. Spain; 32CIBER Enfermedades Neurodegenerativas (CIBERNED), Madrid, Spain; 33Research Progamme on Biomedical Informatics (GRIB), Hospital del Mar Medical Research Institute (IMIM), Barcelona, Spain; 34Instituto Cite this article: Alonso J et al (2022). Mental impact of Covid-19 among Spanish healthcare Nacional de Bioinformatica – ELIXIR-ES, Barcelona, Spain; 35Parc de Salut Mar PSMAR, Barcelona, Spain; 36CiSAL- workers. A large longitudinal survey. Centro de Investigación en Salud Laboral, IMIM/UPF, Barcelona, Spain; 37Department of Health Care Policy, Epidemiology and Psychiatric Sciences 31, e28, Harvard Medical School, Boston, MA, USA; 38Center for Public Health Psychiatry, Universitair Psychiatrisch 1–16. https://doi.org/10.1017/ Centrum, KU Leuven, Leuven, Belgium and 39Fundació Clínic per a la Recerca Biomèdica, Barcelona, Spain S2045796022000130 Received: 25 October 2021 Abstract Revised: 17 March 2022 Accepted: 19 March 2022 Aims. Longitudinal data on the mental health impact of the coronavirus disease 2019 (Covid-19) pandemic in healthcare workers is limited. We estimated prevalence, incidence and persistence of Key words: Depression; epidemiology; mental health; post- probable mental disorders in a cohort of Spanish healthcare workers (Covid-19 waves 1 and 2) traumatic stress disorder; stressful life events -and identified associated risk factors. Methods. 8996 healthcare workers evaluated on 5 May–7 September 2020 (baseline) were Author for correspondence: Jordi Alonso, E-mail: jalonso@imim.es invited to a second web-based survey (October–December 2020). Major depressive disorder (PHQ-8 ≥ 10), generalised anxiety disorder (GAD-7 ≥ 10), panic attacks, post-traumatic stress disorder (PCL-5 ≥ 7), and alcohol use disorder (CAGE-AID ≥ 2) were assessed. Distal (pre- © The Author(s), 2022. Published by Cambridge University Press. This is an Open Access article, pandemic) and proximal (pandemic) risk factors were included. We estimated the incidence distributed under the terms of the Creative of probable mental disorders (among those without disorders at baseline) and persistence Commons Attribution licence (http://creative (among those with disorders at baseline). Logistic regression of individual-level [odds ratios commons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, (OR)] and population-level (population attributable risk proportions) associations were esti- provided the original article is properly cited. mated, adjusting by all distal risk factors, health care centre and time of baseline interview. Results. 4809 healthcare workers participated at four months follow-up (cooperation rate = 65.7%; mean = 120 days S.D. = 22 days from baseline assessment). Follow-up prevalence of any disorder was 41.5%, (v. 45.4% at baseline, p < 0.001); incidence, 19.7% (S.E. = 1.6) and per- sistence, 67.7% (S.E. = 2.3). Proximal factors showing significant bivariate-adjusted associations https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press 2 J. Alonso et al. with incidence included: work-related factors [prioritising Covid-19 patients (OR = 1.62)], stress factors [personal health-related stress (OR = 1.61)], interpersonal stress (OR = 1.53) and financial factors [significant income loss (OR = 1.37)]. Risk factors associated with per- sistence were largely similar. Conclusions. Our study indicates that the prevalence of probable mental disorders among Spanish healthcare workers during the second wave of the Covid-19 pandemic was similarly high to that after the first wave. This was in good part due to the persistence of mental dis- orders detected at the baseline, but with a relevant incidence of about 1 in 5 of HCWs without mental disorders during the first wave of the Covid-19 pandemic. Health-related factors, work-related factors and interpersonal stress are important risks of persistence of mental dis- orders and of incidence of mental disorders. Adequately addressing these factors might have prevented a considerable amount of mental health impact of the pandemic among this vul- nerable population. Addressing health-related stress, work-related factors and interpersonal stress might reduce the prevalence of these disorders substantially. Study registration number: NCT04556565 Introduction environments, inadequate/insufficient material and human resources, increased workload (Galanis et al., 2021). Given the Many studies and several systematic reviews (Boden et al., 2021a; scarcity of longitudinal studies among healthcare workers, knowl- Chigwedere et al., 2021; de Kock et al., 2021; Phiri et al., 2021; edge about risk factors associated with poor mental health trajec- Santabarbara et al., 2021; Sun et al., 2021) suggest that prevalence tories during Covid-19 is limited. One study among Portuguese of mental health problems was extremely high among healthcare nurses showed that fear of personal infection and of infecting workers during the first wave of the coronavirus disease 2019 others were both associated with higher depression, anxiety and (Covid-19) pandemic, including depression, anxiety, post- stress scores (Sampaio et al., 2021). Another study of 361 health- traumatic stress, insomnia and other symptoms. Prevalence of care workers in New York reported increased levels of stress asso- anxiety was estimated to be 37% and of depression 36% pooled ciated with a higher number of Covid-19 cases in the community, across 44 of these studies (Sun et al., 2021). However, substantial while stress was reduced among workers with high emotional variability of results exists across studies. Mental disorders and support and high resilience at baseline (Hirten et al., 2020). The suicidal ideation and behaviours among healthcare workers dur- influence of distal and proximal risk factors on the evolution of ing the initial phases of the pandemic were higher than those the mental health impact of Covid-19 among healthcare workers among the general population (Mortier et al., 2021). requires further research. In the general population, several longitudinal studies suggest This paper aims to assess the evolution of mental disorders that initial high levels of anxiety, depression and other (prevalence, incidence and persistence) after the Covid-19 pan- Covid-related psychopathology tend to remit with time (Hirten demic in a large cohort of healthcare workers after 4 months et al., 2020; McFadden et al., 2021; Miguel-Puga et al., 2021; follow-up in Spain. The paper also examines the association of Sampaio et al., 2021; Sasaki et al., 2021; Van Steenkiste et al., proximal and distal risk factors with follow-up mental disorders. 2021). But evidence about this trajectory among healthcare work- The period studied corresponds with the final part of wave 1 and ers, who have continued to be exposed to a very high workload the initial part of wave 2 of the pandemic in Spain. related to the pandemic, is more limited and results are mixed. Some studies suggest that distress and fear/worry of Covid-19 remained high or even increased in follow-up surveys (Jordan et al., 2021; Lopez Steinmetz et al., 2022; McFadden et al., Methods 2021; Miguel-Puga et al., 2021; Sasaki et al., 2021) while others Study design, population sampling and follow-up indicate that depression and anxiety tended to decrease at follow-up (Hirten et al., 2020; Roberts et al., 2021; Rodriguez A multicentre, observational cohort study of healthcare workers et al., 2021; Sampaio et al., 2021; Van Steenkiste et al., 2021). was carried out in a convenience sample of 18 healthcare institu- Some of these studies are based on a relatively small number of tions from 6 Autonomous Communities in Spain including hos- healthcare workers or on short follow-up periods (Hirten et al., pitals, primary care and public healthcare centres. Institutional 2020; Sampaio et al., 2021; Van Steenkiste et al., 2021) or do representatives invited all employed workers to participate using not focus on specific mental disorders (Hirten et al., 2020; the institution’s administrative email distribution lists (i.e., census Roberts et al., 2021; Sampaio et al., 2021; Sasaki et al., 2021). sampling) to a web-based survey platform (qualtrics.com). Therefore, knowledge about the evolution of mental disorders Informed consent was obtained from all participants. Up to two among healthcare workers over the course of the pandemic is reminder emails were sent within a 2–4-week period after the ini- limited. tial invitation. Workers completing the baseline interview (May Many factors have been associated with the mental health 5th through September 7th, just after the peak of the first wave impact of early Covid-19 pandemic phases among healthcare of the Covid-19 pandemic in Spain) and providing an e-mail workers. Distal risk factors include existing mental disorders, address were re-contacted for the follow up interview approxi- female sex, nurse profession and lower income (Kunzler et al., mately 4 months after, between October 9th and December 2021). Proximal risk factors include frequency/intensity of expos- 11th 2020, at the ascending curve of the pandemic’s second ure to the Covid-19 pandemic, longer working hours in high-risk wave (mean follow-up time of 120 days; median = 118; IQ range https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press Epidemiology and Psychiatric Sciences 3 = 105–136 days). More information can be found elsewhere average weekly hours worked; changes in assigned functions, (Alonso et al., 2021; Mortier et al., 2022). team, or working location; changes of a team or assigned func- tions, and no changes; perceived frequency of lack of protective equipment into a 5-level Likert scale from 0 (‘none of the Measures time’) to 4(‘all of the time’), having to make decisions regarding We used a conceptual framework to study the mental health prioritising care among Covid19 patients, having patients in impact of Covid 19 considering: (i) current mental disorders care that died from Covid-19 infection. We also assessed the fre- (negative mental health status), (ii) proximal (pandemic) risk fac- quency of direct exposure to Covid-19 infected patients during tors (infection, work-related factors and health-related stress), professional activity, 5-level Likert type item, ranging from 0 assessed in the baseline survey and distal (pre-pandemic) risk fac- (‘none of the time’) to 4(‘all of the time’). tors (social and economic characteristics and clinical vulnerabil- Health-related stress: Personal health-related stress, as the ities) (Boden et al., 2021b). mean of two items on a 5-level Likert type scale ranging from 0 to 4 with higher values indicating higher levels of stress (Cronbach α = 0.80); and health-related stress of loved ones Probable current mental disorders assessed with two items on a 5-level Likert type scale (concern Major depressive disorder (MDD): We used the Spanish version about loved ones being infected with Covid-19, and concerns of the Patient Health Questionnaire (PHQ-8), with the cut-off about the health of loved ones). The mean value of the corre- point = 10+ of the sum score to indicate current MDD. The sponding items is obtained with a range from 0 (lower levels of PHQ-8 shows high reliability (>0.8) and good diagnostic accuracy stress) to 4 (higher levels of stress) (Cronbach α = 0.85). for MDD (AUC > 0.90) (Wu et al., 2020). Nevertheless, the Financial factors: Having suffered a significant loss in personal instrument can overestimate the prevalence of depressive disor- or familial income due to the Covid-19 pandemic; financial stress, ders (Levis et al., 2020). calculated using two 5-level Likert-type items (Dohrenwend et al., Generalised anxiety disorder (GAD): We used the seven-item 1978), and stress regarding job loss or loss of income because of GAD scale (GAD-7), which has a good performance to detect Covid-19, obtaining a 2-item combined score (Cronbach α = 0.82). anxiety (AUC > 0.8) (Kroenke et al., 2010; Kessler et al., 2013). Interpersonal stress: Mean score of four 5-level Likert type We used the Spanish version of the GAD-7 (Garcia-Campayo items (i.e., love life, relationships with family, problems getting et al., 2010) and considered the cut-off point of 10+ to indicate along with people at work or school, other problems experienced a current GAD. by loved ones) ranging from 0 to 4, with higher values indicating Panic attacks: the number of panic attacks in the 30 days prior higher levels of stress (Cronbach α = 0.79). to the interview, assessed with an item from the World Mental Family functioning: The Brief Assessment of Family Health-International College Student-WMH-ICS (Kessler et al., Functioning Scale (BAFFS) (Mansfield et al., 2019), a three-item 2013; Blasco et al., 2016). A dichotomous variable indicated the version of the general functioning scale from the Family presence/absence of panic attacks. Assessment Device (FAD) (Epstein et al., 1983) that was devel- Posttraumatic Stress Disorder (PTSD): assessed using the 4-item oped to assess perceived satisfaction or distress with general fam- version of the PTSD checklist for DSM-5 (PCL-5) (Blasco et al., ily functioning. The Spanish version of the scale is available with 2016; Zuromski et al., 2019) which generates diagnoses that closely good psychometric results (Barroilhet et al., 2009). Cronbach’s α parallel those of the full PCL-5 (AUC > 0.9), making it well-suited for internal consistency in our study was α = 0.50. for screening (Weathers et al., 2013). We used the Spanish version Parental stress: 4-item version derived from the Parental Stress of the questionnaire (Resick et al., 2020), and considered a cut-off Scale (Berry and Jones, 1995). Items were summed and re-scaled point of 7 to indicate current PTSD. ranging from 0 (lower parental stress) to 4 (highest parental Substance Use Disorder (SUD): We used the CAGE adapted to stress). The Spanish version has been developed showing adequate include drugs questionnaire (CAGE-AID), that consists of 4 items psychometric properties (Oronoz et al., 2007). Internal consist- focusing on Cutting down, Annoyance by criticism, Guilty feel- ency obtained in our study was α = 0.74. ing, and Eye-openers, proven useful in helping to make a diagno- sis of Alcohol Use Disorder (Dohrenwend et al., 1978; Hinkin Distal risk factors et al., 2001; Zuromski et al., 2019) and Substance Use Disorder (Hinkin et al., 2001). The questionnaire has been adapted into Demographics and professional characteristics: Age category; gen- Spanish (Diez Martinez et al., 1991). Cut-off point of 2+ was con- der; country of birth; marital status; having children in care; pro- sidered to indicate current SUD (Mdege and Lang, 2011). fession; and workplace. Prior lifetime mental disorders: Lifetime mental disorders prior Proximal risk factors to the onset of the Covid-19 outbreak assessed using a checklist Covid-19 exposure and infection status: Whether the respondent from the Composite International Diagnostic Interview (CIDI) had been hospitalised for Covid-19 infection and/or had a posi- (Navarro-Mateu et al., 2013) screening lifetime mood, anxiety, sub- tive Covid-19 test or medical diagnosis not requiring hospitalisa- stance use problems and ‘other’ mental disorders. The number of tion, whether the respondent had been isolated or quarantined chronic physical health conditions was also assessed for a list of because of exposure to Covid-19 infected person(s), and whether conditions considered as vulnerable to Covid-19. they had loved ones infected with Covid-19. Work-related factors: Perceived lack of care centre prepared- Ethical considerations ness (i.e., lack of coordination, communication, personnel, super- vision at work, training for assigned tasks) using four 5-level The study complies with the Declaration of Helsinki and the Likert-type items ranging from ‘none of the time’ to ‘all of the Code of Ethics and was approved by the IRB Parc de Salut Mar time’ (summed score rescaled to a 0.0–4.0; Cronbach α = 0.86); (2020/9203/I) and by the corresponding IRBs of all the https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press 4 J. Alonso et al. participating centres. Registered at ClinicalTrials.gov (https://clin- individuals who agreed to participate (n = 10 360) divided by icaltrials.gov/ct2/show/NCT04556565). unique first survey page visitors (n = 11 507), was 90.0%, and of them, 8328 professionals finalised the baseline interview, repre- senting a survey completion rate of 80.4% (Eysenbach, 2004). Statistical analysis Of the total 8996 unique healthcare workers that provided suffi- Missing item-level data among respondents were imputed using cient information of the baseline evaluation, n = 7318 provided multiple imputations (MI) by chained equations (Van Bureen, their email to be re-contacted and n = 4809 (cooperation rate = 2012) (12 imputed datasets, 10 iterations per imputation). 65.7%) answered the follow-up interview between 9th of Sample characteristics are reported as weighted percentages and October and 11th of December 2021, with a mean number of S.E. Differences of baseline characteristics between those who did 120.1 days (S.D. = 22.2) between baseline and follow up assess- and those who did not participate in the follow-up survey (lost ments. Those participating in the follow-up survey differed only to follow up) were evaluated using the modified Rao-Scott χ 2 marginally in some baseline study characteristics (follow-up par- test. To correct the bias caused by lost to follow up missing values, ticipants were older, with higher income, with a higher propor- inverse-probability weighting (IPW) (Seaman et al., 2012) was tion being medical doctors, and having suffered of Covid-19 on applied, as the inverse of the probability of completing the their own or their loved ones and had suffered less health-related follow-up survey on observed related baseline covariates, estimated and financials stresses). These differences were very small, using a logistic regression model. Additionally, post-stratification although statistically significant due to the high numbers involved weights through raking were used to adjust for potential deviations (online Supplementary Table 1). Inverse probability weights were from target population distributions in terms of age, gender, profes- applied to restore these differences as explained above. sion and healthcare centre. Pooled MI-based parameter estimates and standard errors (S.E.) and statistical inferences were obtained Any mental disorder prevalence, incidence and persistence from the weighted analysis of these MI datasets. Prevalence of any probable mental disorder at baseline and at At 4-month follow-up, the prevalence of any probable mental dis- follow up was estimated overall and stratified by distal and prox- order was 41.5%, somewhat lower than at baseline (45.4%, Mc imal risk factors. Prevalence of any probable mental disorder at Nemar test p-value < 0.001) (Table 1 and Fig. 1a). Prevalence at follow-up stratified among individuals with a negative baseline follow-up was higher for younger age groups, female gender, screen (i.e., incidence) and among individuals with a positive base- country of birth other than Spain, lower income, auxiliary nurse line screen (i.e., persistence) were also estimated. Logistic regression profession, and several pre-pandemic mental disorders and higher estimated the association of each proximal risk factor with probable number of chronic physical health conditions. Incidence of prob- mental disorders adjusting each time for all distal risk factors, able mental disorder (among those without baseline disorder) was health care centre and time of baseline survey. Regression coeffi- 19.7% (S.E. = 1.6), being significantly higher among those with cients were exponentiated to obtain odds ratios (OR) and 95% con- lower income and those with some pre-pandemic mental disor- fidence intervals (CIs). All Likert-type variables were analysed as ders (Table 1, middle, and Fig. 1b). More than two thirds continuous. Potential deviations from a continuous linear effect (67.7%; S.E. = 2.3) with a baseline probable mental disorder, per- in the logit were assessed graphically. Population-attributable risk sisted with a disorder at follow-up. Persistence was significantly proportions (PARP) (Krysinska and Martin, 2009) and associated associated to the same distal factors than incidence, except for 95% confidence intervals obtained with bootstrap resampling (200 lower-income levels, for which higher persistence frequency did replications), were simulated based on individuals’ predicted prob- not reach statistical significance, and except for a number of abilities estimated by the multivariable logistic regression equations chronic physical health conditions, significantly associated to per- (Nock et al., 2014). PARP provide estimates of the proportions of sistence, but not to incidence (Table 1, three last columns, and mental disorders that could potentially be attributed to specific Fig. 1b). predictor variables assuming a causal pathway between these pre- There was a trend towards a lower severity of both PHQ-8 and dictor variables and the outcome. GAD-7 scores between baseline and 4-months follow-up assess- All analyses were adjusted by the week of the interview ments (see online Supplementary Tables 2 and 3). Online (entered as a continuous variable) to account for possible differ- Supplementary Tables 4 and 5 present unadjusted and adjusted, ences of baseline predictor variables by the time of assessment respectively, associations between distal risk factors and probable at baseline. Since the correlation between the response time at mental disorders prevalence, incidence and persistence. baseline and the time of follow up was r = −0.93, we only adjusted by one of these variables. Association between proximal risk factors and any mental Variance estimates were obtained using the Taylor series lin- disorder earisation method considering weighting and within-health care centre clustering of data. MI were carried out using package Table 2 presents associations of prevalence, incidence and per- mice from R (Van Buuren and Groothuis-Oudshoorn, 2011) sistence of any probable mental disorder with each proximal Analyses were performed using R v4.1.032 (Team, 2021) and and work-related risk factor in our study, adjusting by all distal SAS v9.4 (INC SI, 2014). factors (i.e., those presented in Table 1). Each row presents a dif- ferent model. An overall multivariate model was not primarily attempted, in order to minimise over adjustment bias Results (Schistermann et al., 2009). (Online Supplementary Table 6 shows the bivariate associations, and online Supplementary Participation Table 7, the fully adjusted model.) All risk factors evaluated A total of 8996 healthcare workers participated in the baseline were significantly associated with a higher follow-up prevalence evaluation. The survey participation rate, calculated as unique of probable mental disorders, except for having been hospitalised https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press Epidemiology and Psychiatric Sciences Table 1. Prevalence, incidence and persistence of any of mental disorders, total and by distal risk factors. Spanish healthcare workers, MINDCOVID study (absolute numbers and weighted proportions) Among those with NO baseline disorders Full follow-up sample (n = 4809) (n = 2681) Among those with any baseline disorder (n = 2128) Any baseline Any follow-up Incidence of any disorder Persistence of any disorder disorder (n = 2128) disorder (n = 1870) (n = 480) (n = 1390) na %b (S.E.) %b (S.E.) %b (S.E.) na % b (S.E.) %b (S.E.) na %b (S.E.) %b (S.E.) Total 45.4 (1.8) 41.5 (1.9) 19.7 (1.6) 67.7 (2.3) Age 50 years or more 2031 43.5 (2.5) 39.5 (2.1) 37.8 (2.0) 1237 48.2 (2.6) 17.6 (1.4) 795 37.8 (2.7) 68.6 (2.5) 30–49 years 2275 45.5 (1.6) 49.1 (2.1) 44.0 (2.3) 1223 42.4 (2.1) 21.1 (2.0) 1052 49.2 (2.1) 67.7 (2.5) 18–29 years 503 11.0 (1.9) 53.7 (3.5) 45.8 (2.7) 222 9.3 (1.4) 23.7 (4.3) 281 13.0 (2.7) 64.9 (4.4) Modified Rao-Scott 9.65 (225 106) [<0.001]* 9.33 (239 061) [<0.001]* 2.34 (220 640) [0.097] 0.43 (213 426) [0.652] (R-S) χ 2 test Gender Female 3899 77.6 (1.3) 47.2 (2.0) 42.8 (2.1) 2096 75.1 (1.6) 20.6 (2.1) 1803 80.7 (1.3) 67.5 (2.4) Male 910 22.4 (1.3) 39.1 (2.0) 37.0 (1.7) 585 24.9 (1.6) 16.8 (2.0) 325 19.3 (1.3) 68.5 (3.6) Modified Rao-Scott 10.72 (11 875) [0.001]* 8.63 (110 196) [0.003]* 1.40 (113 171) [0.236] 0.09 (188 852) [0.766] (R-S) χ 2 test Country of birth Spain 4582 95.4 (0.5) 45.1 (2.0) 41.1 (1.9) 2570 95.9 (0.5) 19.4 (1.5) 2012 94.8 (0.9) 67.6 (2.3) Other 227 4.6 (0.5) 51.8 (3.7) 49.0 (3.3) 111 4.1 (0.5) 26.3 (6.3) 116 5.2 (0.9) 70.2 (5.2) Modified Rao-Scott 1.66 (15 765) [0.197] 4.09 (156 397) [0.043]* 1.23 (1 137 771) [0.266] 0.30 (15 353) [0.586] (R-S) χ 2 test Marital status Single, divorced, legally 2272 47.3 (2.2) 49.0 (1.8) 43.4 (2.7) 1199 44.1 (2.0) 18.9 (2.9) 1073 51.0 (2.8) 68.9 (3.0) separated, or widowed Married 2537 52.7 (2.2) 42.2 (2.2) 39.7 (1.7) 1482 55.9 (2.0) 20.2 (1.3) 1054 49.0 (2.8) 66.4 (2.6) Modified Rao-Scott 11.97 (123 177) [<0.001]* 2.52 (194 691) [0.112] 0.19 (127 056) [0.666] 0.65 (159 316) [0.421] (R-S) χ 2 test Income More than 4500€ 1659 28.8 (1.5) 38.0 (1.4) 33.5 (2.1) 1050 32.8 (1.8) 15.4 (1.6) 609 24.1 (1.6) 63.0 (3.5) Between 2200€–4500€ 1730 36.1 (1.2) 42.6 (2.5) 41.0 (2.0) 982 37.9 (1.4) 20.0 (1.6) 748 33.9 (1.5) 69.2 (2.6) Less than 2200€ 1420 35.1 (2.2) 54.4 (2.3) 48.5 (2.5) 649 29.3 (2.1) 23.9 (3.4) 771 42.0 (2.6) 69.2 (2.7) Modified Rao-Scott 22.14 (2309) [<0.001]* 24.04 (2285) [<0.001]* 3.86 (2875) [0.021]* 2.34 (2767) [0.097] (R-S) χ 2 test (Continued ) 5 https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press Table 1. (Continued.) 6 Full follow-up sample (n = 4809) Among those with NO baseline disorders Among those with any baseline disorder (n = 2128) (n = 2681) Any baseline Any follow-up Incidence of any disorder Persistence of any disorder disorder (n = 2128) disorder (n = 1870) (n = 480) (n = 1390) na %b (S.E.) %b (S.E.) %b (S.E.) na % b (S.E.) %b (S.E.) na %b (S.E.) %b (S.E.) Children in care Yes 1995 41.2 (1.4) 43.7 (2.6) 40.1 (2.0) 1139 42.5 (1.1) 21.2 (2.1) 856 39.7 (2.7) 64.4 (2.9) No 2814 58.8 (1.4) 46.6 (2.1) 42.5 (2.3) 1542 57.5 (1.1) 18.6 (1.7) 1272 60.3 (2.7) 69.9 (2.9) Modified Rao-Scott 0.92 (134 650) [0.337] 1.07 (133 524) [0.301] 1.47 (113 670) [0.225] 2.47 (118 185) [0.116] (R-S) χ 2 test (DF) [p-value] Profession Medical doctor 1650 26.3 (2.8) 36.5 (1.2) 34.5 (2.7) 1041 30.6 (3.0) 17.4 (2.1) 608 21.2 (2.8) 64.2 (3.8) Nurse 1406 31.0 (1.2) 49.4 (2.4) 42.4 (2.4) 697 28.7 (1.2) 19.6 (1.7) 709 33.7 (1.6) 65.8 (3.3) Auxiliary nurse 387 13.7 (3.2) 56.8 (4.7) 51.7 (4.2) 157 10.8 (2.5) 24.1 (2.5) 230 17.1 (4.2) 72.7 (4.1) Other profession 555 9.0 (0.8) 41.0 (3.5) 39.1 (3.7) 324 9.7 (0.9) 17.8 (2.5) 231 8.1 (1.1) 69.8 (3.0) involved in patient care Other profession not 812 20.0 (2.4) 45.2 (3.7) 43.3 (4.4) 462 20.1 (1.9) 21.8 (4.3) 350 19.9 (3.7) 69.4 (3.0) involved in patient care Modified Rao-Scott 5.85 (413 247) [<0.001]* 3.37 (425 209) [0.009]* 1.08 (424 948) [0.363] 1.22 (412 096) [0.299] (R-S) χ 2 test (DF) [p-value] Workplace Hospital 2818 57.5 (0.9) 47.1 (2.6) 40.9 (2.5) 1525 55.7 (1.2) 18.4 (1.8) 1293 59.7 (1.4) 66.2 (2.6) Primary care 1581 36.2 (0.9) 44.2 (1.2) 43.6 (1.9) 899 37.0 (1.2) 22.2 (1.7) 682 35.2 (1.3) 70.6 (3.5) Others 410 6.3 (0.4) 36.6 (4.0) 34.5 (3.0) 258 7.3 (0.7) 16.6 (2.2) 153 5.1 (0.6) 65.4 (5.8) Modified Rao-Scott 2.61 (258 618) [0.074] 1.58 (229 265) [0.206] 1.84 (23 356) [0.159] 0.77 (239 577) [0.463] (R-S) χ 2 test (DF) [p-value] Pre-pandemic mood disorder Yes 528 11.4 (0.6) 70.3 (2.7) 64.3 (4.0) 164 6.2 (0.5) 32.7 (7.2) 364 17.6 (1.1) 77.7 (3.5) No 4281 88.6 (0.6) 42.2 (1.7) 38.5 (1.7) 2517 93.8 (0.5) 18.8 (1.4) 1764 82.4 (1.1) 65.5 (2.3) Modified Rao-Scott 77.96 (11 962) [<0.001]* 38.00 (113 592) [<0.001]* 4.42 (161 749) [0.035]* 16.30 (17 138) [<0.001]* (R-S) χ 2 test (DF) J. Alonso et al. [p-value] Pre-pandemic anxiety disorder Yes 1710 36.8 (0.9) 61.8 (2.1) 54.6 (2.6) 670 25.8 (0.7) 26.9 (3.8) 1040 50.1 (1.4) 71.7 (2.5) No 3099 63.2 (0.9) 35.8 (1.8) 33.8 (1.8) 2012 74.2 (0.7) 17.2 (1.2) 1088 49.9 (1.4) 63.7 (2.9) https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press Modified Rao-Scott 105.07 (130) [<0.001]* 78.80 (11 171) [<0.001]* 8.22 (17 328) [0.004]* 7.45 (12 399) [0.006]* Epidemiology and Psychiatric Sciences (R-S) χ 2 test (DF) [p-value] Pre-pandemic substance use disorder Yes 60 1.5 (0.1) 87.9 (5.3) 68.0 (6.7) 12 0.3 (0.1) 20.4 (10.2) 49 2.9 (0.4) 74.6 (7.2) No 4749 98.5 (0.1) 44.8 (1.8) 41.1 (1.9) 2670 99.7 (0.1) 19.7 (1.6) 2079 97.1 (0.4) 67.5 (2.3) Modified Rao-Scott 30.40 (12 790) [<0.001]* 10.72 (12 301) [0.001]* 0.01 (125 024) [0.916] 0.99 (1877) [0.319] (R-S) χ 2 test (DF) [p-value] Other pre-pandemic mental disorder Yes 38 0.8 (0.1) 58.1 (8.7) 60.3 (11.7) 16 0.6 (0.2) 41.3 (16.9) 21 1.0 (0.2) 74.1 (9.2) No 4772 99.2 (0.1) 45.3 (1.8) 41.3 (1.8) 2665 99.4 (0.2) 19.5 (1.5) 2107 99.0 (0.2) 67.6 (2.3) Modified Rao-Scott 2.39 (184 740) [0.122] 3.04 (1 350 096) [0.081] 2.43 (11 914 069) [0.119] 0.50 (1 831 474) [0.478] (R-S) χ 2 test (DF) [p-value] Number of pre-pandemic lifetime mental disorders Two or more 377 8.4 (0.7) 77.1 (3.0) 65.0 (4.6) 84 3.5 (0.4) 27.3 (8.7) 293 14.3 (1.3) 76.2 (4.7) Exactly one 1550 32.8 (0.9) 57.3 (2.1) 52.7 (2.7) 689 25.6 (0.8) 28.5 (3.6) 861 41.3 (2.0) 70.7 (2.2) None 2882 58.8 (0.9) 34.3 (1.9) 31.9 (1.8) 1909 70.8 (0.7) 16.1 (1.1) 973 44.4 (1.6) 62.1 (3.1) Modified Rao-Scott 75.44 (21 250) [<0.001]* 35.58 (24 262) [<0.001]* 7.58 (24 445) [<0.001]* 6.25 (22 828) [0.002]* (R-S) χ 2 test (DF) [p-value] Number of physical health conditions Two or more 157 3.9 (0.5) 63.6 (5.1) 60.8 (4.1) 57 2.6 (0.5) 21.3 (4.8) 100 5.5 (0.7) 83.5 (4.2) Exactly one 1013 22.0 (0.8) 49.4 (2.6) 45.5 (2.2) 502 20.4 (1.1) 22.7 (2.1) 511 23.9 (1.3) 68.9 (2.9) None 3639 74.1 (1.0) 43.3 (2.0) 39.2 (2.0) 2122 77.0 (1.2) 18.8 (1.7) 1516 70.6 (1.6) 66.1 (2.3) Modified Rao-Scott 7.94 (21 032) [<0.001]* 13.12 (21 656) [<0.001]* 1.83 (2269) [0.163] 7.48 (25 242) [0.001]* (R-S) χ 2 test (DF) [p-value] R-S, Rao-Scott; DF, Degrees of freedom. Note. Covid-19, coronavirus disease 2019; S.E., standard error. a Unweighted numbers. b Weighted percentage [using post-stratification weights and inverse-probability weighting (IPW)]. *Statistically significant (α = 0.05). 7 8 J. Alonso et al. Fig. 1. (a) Prevalence of probable current mental disor- ders among Spanish Healthcare Workers [n = 4809, baseline (blue bars); 4-month follow-up (red bars)]. (b) Probable current mental disorders among Spanish healthcare workers at 4-month follow-up sur- vey: Incident/New Disorder Onset (green bars) and Persistent/Recurrent Disorder (red bars) (n = 4809). MDD, major depressive disorder; GAD, generalised anx- iety disorder; PTSD, posttraumatic stress disorder; SUD, substance use disorder. **p-value <0.001 for the multiple imputation chi-square pooling of the McNemar’s test comparing paired prevalence at base- line and follow up for the mental disorders evaluated. *p-value <0.05 for the multiple imputations χ 2 pooling of the McNemar’s test comparing paired prevalence at baseline and follow up for the mental disorders evaluated. for Covid-19 (probably due to low numbers in the sample), and persistence. PARPs were generally higher for incidence of prob- for having other family, friends or others infected with able mental disorders than for persistence of baseline probable Covid-19. The same risk factors showed a positive association disorders. with the prevalence of any probable mental disorder after the 1st wave of the pandemic (baseline assessment) as with prevalence during the 2nd wave (follow-up assessment) (see columns 3 and 4 Discussion of Table 2). Nevertheless, associations tended to be marginally This study is, to our knowledge, the first one to estimate the higher for baseline than for follow up data, for work-related fac- prevalence of probable mental disorders among healthcare work- tors (number of hours worked), for personal health-related stress, ers during the second wave of the Covid-19 pandemic focusing on for health-related stress of loved ones and for interpersonal stress. the most relevant mental disorders and including a large sample. Covid-19 own infection was a risk factor for new onset (OR = 1.54 Two major results arise from this longitudinal study. First, the 95% IC = 1.24, 1.91) but not for persistence (1.09; 95% IC = 0.92, prevalence of probable mental disorders during the second wave 1.30), while a larger number of hours worked was a risk factor for of the Covid-19 pandemic was high and very similar to that dur- persistence but not for new onset (Table 2, columns 7 and 10). ing the first wave. This was mainly due to the high persistence of baseline disorders (more than two-thirds of those with disorders at baseline continued to have these disorders at follow-up); never- Population attributable risk proportions (PARPs) theless, the incidence of new probable mental disorders was also Table 3 shows attributable risk proportions (PARPs) of proximal important (one in 5 of healthcare workers free of mental disorders risk factors of any probable mental disorder, adjusting by distal during the first wave developed one in the second wave). And factors. Health-related, work-related, and interpersonal stress pre- second, health-related and work-related factors as well as inter- sented the highest association with both prevalence, incidence and personal stress were important risk factors for both onset and https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press Epidemiology and Psychiatric Sciences Table 2. Adjusted associations between proximal risk factors and any probable mental disorders diagnosis (bivariate adjusted for all distal factors). Spanish healthcare workers, MINDCOVID study (absolute numbers and weighted proportions) Among those with NO Among those with any Full follow-up sample (n = 4809) baseline disorders (n = 2681) baseline disorder (n = 2128) Any baseline Any follow-up Incidence of any b disorder disorder b disorder Persistence of any % (S.E.) or (n = 2128) (n = 1870) % (S.E.) or (n = 480) %b (S.E.) or disorder (n = 1390) Med (S.E.) Med (S.E.) Med (S.E.) na (IQR) OR (95% CI) OR (95% CI) na (IQR) OR (95% CI) na (IQR) OR (95% CI) Covid-19 infection Having been hospitalised for 67 1.4 (0.2) 1.48 (0.87–2.51) 1.57 (0.93–2.64) 32 1.2 (0.3) 1.46 (0.61–3.51) 35 1.6 (0.2) 1.48 (0.56–3.89) Covid-19 Positive Covid-19 test or medical 869 16.1 (2.0) 1.08 (0.94–1.23) 1.25 (1.11–1.42)* 461 15.2 (2.0) 1.54 (1.24–1.91)* 408 17.2 (2.1) 1.09 (0.92–1.30) Covid-19 diagnosisc None of the above 3873 82.5 (2.1) (Ref) (Ref) 2189 83.6 (2.0) (Ref) 1685 81.1 (2.3) (Ref) F value (ndf, ddf) [ p value]f 1.42 (216 394) [0.243] 9.59(21 492) [<0.001]* 8.01(23 901) [<0.001]* 1.02(23 745) [0.360] Type of loved ones infected with Covid-19 Partner, children, or parents 781 13.7 (2.0) 1.79 (1.27–2.52)* 1.41 (1.21–1.64)* 406 11.9 (2.2) 1.06 (0.71–1.59) 375 15.8 (1.8) 1.22 (0.96–1.55) Other family, friends or othersd 2964 58.6 (0.9) 1.28 (0.99–1.66) 1.11 (0.93–1.33) 1637 58.1 (1.3) 0.91 (0.71–1.15) 1327 59.2 (1.6) 1.10 (0.78–1.55) None of the above 1065 27.7 (2.4) (Ref) (Ref) 639 30.0 (3.0) (Ref) 426 25.0 (2.1) (Ref) f F value (ndf, ddf) [ p value] 6.03 (212 311) [0.002]* 12.05 (23 480) [<0.001]* 1.73 (21 531) [0.177] 1.43 (21 122) [0.240] Covid-19 isolation Having been isolated or 1337 25.6 (1.7) 1.34 (1.14–1.58)* 1.48 (1.19–1.83)* 671 23.0 (1.4) 1.51 (1.08–2.11)* 667 28.8 (2.2) 1.29 (1.00–1.66)* quarantined because of Covid-19 Not having been isolated or 3472 74.4 (1.7) (Ref) (Ref) 2011 77.0 (1.4) (Ref) 1461 71.2 (2.2) (Ref) quarantined because of Covid-19 F value (ndf, ddf) [ p value]f 12.76 (119 879) [<0.001]* 12.70 (123 489) [<0.001]* 5.81 (11 566) [0.016]* 3.88 (13 853) [0.049]* Frequency of direct exposure to 2.8 (0.2) 1.35 (1.24–1.47)* 1.34 (1.20–1.50)* 2.6 (0.2) 1.28 (1.13–1.45)* 3.0 (0.2) 1.21 (1.04–1.42)* Covid-19 patients (scale 0–4) (2.0–3.9) (1.6–3.6) (2.2–4.1) F value (ndf, ddf) [ p value]f 41.27 (1482) [<0.001]* 27.30 (16 456) [<0.001]* 15.55 (112 051) [<0.001]* 5.84 (16 834) [0.016]* Perceived lack of preparedness 1.6 (0.1) 1.69 (1.53–1.86)* 1.53 (1.39–1.68)* 1.3 (0.1) 1.36 (1.24–1.49)* 1.9 (0.1) 1.25 (1.05–1.49)* (rescaled 0–4) (0.7–2.4) (0.5–2.0) (1.2–2.7) F value (ndf, ddf) [ p value]f 90.43 (1304) [<0.001]* 72.50 (1916) [<0.001]* 19.15 (1.27) [<0.001]* 6.10(16 782) [0.014]* Average weekly hours worked 51 h or more 765 14.0 (0.7) 1.59 (1.30–1.94)* 1.23 (1.02–1.50)* 397 13.0 (0.9) 0.68 (0.40–1.14) 368 15.3 (1.0) 1.36 (1.08–1.72)* 41–50 h 1140 22.7 (2.4) 1.39 (1.10–1.75)* 1.32 (1.06–1.64)* 581 20.7 (2.3) 1.08 (0.76–1.55) 559 25.1 (2.9) 1.31 (0.91–1.88) 40 h or less 2905 63.2 (2.5) (Ref) (Ref) 1704 66.3 (2.5) (Ref) 1201 59.5 (2.7) (Ref) (Continued ) 9 https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press 10 Table 2. (Continued.) Full follow-up sample (n = 4809) Among those with NO Among those with any baseline disorders (n = 2681) baseline disorder (n = 2128) Any baseline Any follow-up Incidence of any Persistence of any b disorder disorder b disorder b disorder (n = 1390) % (S.E.) or (n = 2128) (n = 1870) % (S.E.) or (n = 480) % (S.E.) or Med (S.E.) Med (S.E.) Med (S.E.) na (IQR) OR (95% CI) OR (95% CI) na (IQR) OR (95% CI) na (IQR) OR (95% CI) F value (ndf, ddf) [ p value]f 13.51 (25 901) [<0.001]* 6.18 (25 061) [0.002]* 1.50 (221 121) [0.223] 4.25 (2158) [0.016]* Work related changes Changed to specific Covid-19 1080 21.1 (3.7) 1.48 (1.17–1.88)* 1.45 (1.10–1.89)* 540 18.5 (3.6) 1.40 (0.99–1.98) 540 24.2 (4.0) 1.19 (0.82–1.72) related work location Changed of team or assigned 1642 33.7 (3.2) 1.38 (1.14–1.66)* 1.30 (1.17–1.44)* 871 31.7 (3.3) 1.18 (0.91–1.53) 770 36.0 (3.6) 1.13 (0.98–1.29) functionse No changes 2088 45.3 (1.6) (Ref) (Ref) 1270 49.9 (1.7) (Ref) 818 39.8 (2.0) (Ref) f F value (ndf, ddf) [ p value] 7.75 (21 936) [<0.001]* 16.71 (2195) [<0.001]* 2.45 (21 102) [0.086] 1.60 (21 025) [0.203] Work – perceived frequency of lack of 2.7 (0.1) 1.59 (1.42–1.78)* 1.44 (1.29–1.60)* 2.5 (0.1) 1.29 (1.12–1.47)* 3.0 (0.1) 1.21 (1.05–1.40)* protective equipment (scale 0–4) (2.0–3.5) (1.6–3.2) (2.3–3.8) F value (ndf, ddf) [ p value]f 57.47(11 131) [<0.001]* 43.51 (14 586) [<0.001]* 11.91 (1654) [0.001]* 7.02(114 892) [0.008]* Decisions regarding patients Having to make decisions 890 15.8 (1.8) 1.51 (1.23–1.86)* 1.83 (1.64–2.03)* 460 14.6 (2.1) 1.62 (1.27–2.06)* 430 17.3 (1.5) 1.74 (1.40–2.18)* regarding prioritising care among Covid-19 patients Not having to make decisions 3919 84.2 (1.8) (Ref) (Ref) 2222 85.4 (2.1) (Ref) 1698 82.7 (1.5) (Ref) regarding prioritising care among Covid-19 patients F value (ndf, ddf) [ p value]f 15.15(119 328) [<0.001]* 95.58 (1169) [<0.001]* 15.22 (16 058) [<0.001]* 23.36 (12 258) [<0.001]* Having patient(s) in care that died 1927 37.2 (3.2) 1.32 (1.19–1.46)* 1.36 (1.11–1.68)* 997 34.6 (3.2) 1.34 (1.05–1.71)* 931 40.3 (3.7) 1.20 (0.87–1.65) from Covid-19 infection Not having patient(s) in care that 2882 62.8 (3.2) (Ref) (Ref) 1685 65.4 (3.2) (Ref) 1197 59.7 (3.7) (Ref) died from Covid-19 infection F value (ndf, ddf) [ p value]f 24.29 (1601) [<0.001]* 8.44 (12 041) [0.004]* 5.62 (11 577) [0.018]* 1.22 (127 142) [0.269] Personal health-related stress 1.6 (0.1) 2.20 (1.99–2.43)* 1.94 (1.77–2.12)* 1.2 (0.0) 1.61 (1.41–1.83)* 2.0 (0.1) 1.50 (1.36–1.66)* (rescaled to 0–4) (0.8–2.4) (0.5–1.9) (1.3–2.8) F value (ndf, ddf) [ p value]f 184.58 (1181) [<0.001]* 141.37 (1.86) [<0.001]* 39.67 (1184) [<0.001]* 50.55 (1187) [<0.001]* Health-related stress loved ones 2.4 (0.1) 2.23 (1.99–2.50)* 1.91 (1.71–2.14)* 2.0 (0.1) 1.49 (1.27–1.75)* 3.0 (0.1) 1.52 (1.29–1.79)* J. Alonso et al. (rescaled to 0–4) (1.5–3.2) (1.2–2.8) (2.1–3.5) F value (ndf, ddf) [ p value]f 139.02(1118) [<0.001]* 105.93 (1251) [<0.001]* 23.37 (13 557) [<0.001]* 25.17 (12 990) [<0.001]* https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press Epidemiology and Psychiatric Sciences Significant loss of personal or familial income due to Covid-19 Yes 885 20.2 (1.2) 1.55 (1.31–1.83)* 1.42 (1.16–1.74)* 417 16.7 (1.0) 1.37 (0.80–2.34) 468 24.4 (1.8) 1.12 (0.85–1.47) No 3924 79.8 (1.2) (Ref) (Ref) 2265 83.3 (1.0) (Ref) 1659 75.6 (1.8) (Ref) f F value (ndf, ddf) [ p value] 25.36 (11 428) [<0.001]* 11.63( 111 044) [0.001]* 1.35 (128 579) [0.246] 0.66 (18 307) [0.418] Financial stress (rescaled to 0–4) 0.6 (0.0) 1.42 (1.32–1.53)* 1.36 (1.28–1.45)* 0.4 (0.0) 1.22 (1.07–1.39)* 0.9 (0.0) 1.21 (1.11–1.32)* (0.0–1.6) (0.0–1.2) (0.1–1.9) F value (ndf, ddf) [ p value]f 76.79(11 075) [<0.001]* 63.00(1.90) [<0.001]* 8.11(11 662)[0.004]* 18.05(1762) [<0.001]* Interpersonal stress (rescaled to 0–4) 1.3 (0.1) 2.41 (2.12–2.74)* 1.96 (1.72–2.22)* 0.8 (0.0) 1.53 (1.35–1.73)* 1.8 (0.0) 1.50 (1.24–1.82)* (0.5–2.1) (0.2–1.6) (1.1–2.5) F value (ndf, ddf) [ p value]f 139.75(1171) [<0.001]* 96.91 (11 064) [<0.001]* 37.17(1291) [<0.001]* 17.82 (17 883) [<0.001]* Family functioning scale score 0.3 (0.0) 1.59 (1.46–1.72)* 1.41 (1.30–1.54)* 0.2 (0.0) 1.18 (0.99–1.41) 0.5 (0.0) 1.21 (1.07–1.37)* (rescaled to 0–4) (0.0–0.9) (0.0–0.8) (0.0–1.2) F value (ndf, ddf) [ p value]f 53.05 (1.24) [<0.001]* 32.65 (1.35) [<0.001]* 3.02 (1113) [0.085] 9.72 (1359) [0.002]* Parental stress scale score 0.0 (0.1) 1.26 (1.05–1.50)* 1.17 (0.99–1.39) 0.0 (0.0) 1.08 (0.90–1.28) 0.0 (0.1) 1.10 (0.87–1.38) (rescaled to 0–4) (0.0–0.8) (0.0–0.7) (0.0–0.9) F value (ndf, ddf) [ p value]f 6.31 (1991) [0.012]* 3.43 (12 287) [0.064] 0.69 (1357) [0.407] 0.64 (12 445) [0.423] CI, confidence interval; IQR, interquartile range; Med, median; S.E., standard error; ndf, numerator degrees of freedom; ddf, denominator degrees of freedom. Note. OR, odds ratio; 95% CI, 95% confidence interval; Covid-19, coronavirus disease 2019; S.E., standard error; IQR, interquartile range. All analyses adjust for time of survey (weeks), health centre membership, and all distal factors. a Unweighted numbers. b Weighted percentage (using post-stratification weights and inverse-probability weighting (IPW)). c The category ‘positive Covid-19 test or medical Covid-19 diagnosis’ excludes those having been hospitalised for Covid-19. d The category ‘other family, friends, or others’ excludes having a partner, children, or parents infected with Covid-19. e The category ‘changed of team or assigned functions’ excludes those that changed to a specific Covid-19-related work location. f F-test to evaluate joint significance of categorical predictor levels base on multiple imputation. *Statistically significant (α = 0.05). 11 12 J. Alonso et al. Table 3. Adjusted population attributable risk proportions (PARP) for the association between proximal risk factor domains and any probable mental disorders diagnosis. Spanish healthcare workers, MINDCOVID study Full follow-up sample (n = 4809) Among those with NO disorders Among those with any disorder baseline (n = 2681) baseline (n = 2128) Any baseline disorder Any follow-up Incidence of any disorder Persistence of any disorder (n = 2128) disorder (n = 1870) (n = 480) (n = 1390) PARP % (S.E.) PARP % (S.E.) PARP % (S.E.) PARP % (S.E.) Infection-related 13.9 (3.7)* 11.0 (4.3)* 4.5 (8.9) 4.8 (4.5) factors WOrk-related factors 69.3 (3.1)* 65.1 (3.8)* 57.4 (7.8)* 35.6 (6.9)* Health-related stress 80.6 (2.1)* 74.9 (3.1)* 59.2 (7.1)* 48.5 (7.0)* Financial stress 20.3 (2.1)* 20.0 (2.4)* 15.5 (4.8)* 9.0 (2.7)* Interpersonal stress 71.0 (2.7)* 62.1 (3.2)* 43.8 (7.3)* 35.5 (5.8)* Total for all proximal 93.8 (1.1)* 88.4 (2.2)* 74.1 (7.2)* 69.4 (7.1)* factors PARP, population-attributable risk proportion; S.E., standard error. *Statistically significant (α = 0.05). persistence of mental disorders. Some of these factors are poten- outcomes using longer follow-ups and large enough samples, to tially preventable (Kisely et al., 2020). Thus, adequately addressing assess the extent to which the likely mental health impact due these factors might be effective in reducing the high prevalence of to the Covid-19 pandemic may become sustained in time mental disorders in this critical segment of the population. among healthcare workers. Risk factors Our results in context with previous knowledge Most proximal risk factors analysed in our study were associated Frequency of mental health impacts with the prevalence of probable mental disorders after the 1st As noted in the introduction, longitudinal studies assessing the wave (baseline assessment) and during the 2nd wave of the pan- impact of the Covid-19 pandemic among healthcare workers are demic. Covid-19 infection, which was frequent among healthcare scarce and results are mixed. Some studies indicate the persistence workers, increased the risk of new onset of probable mental dis- of psychological stress (Sasaki et al., 2021) while others suggest a orders. These associations suggest that it is necessary to imple- lowering of depression and anxiety, in particular after 5 or more ment interventions to mitigate the mental health impact of the months from the beginning of the pandemic (Paul et al., 2021). In pandemic among healthcare workers. Large population attribut- general, evidence is limited due to a small number of studies, very able risk proportions found indicate a potentially high benefit, short follow-up periods and with a small number of cases (Hirten if a causal interpretation can be assumed. Although many inter- et al., 2020; Sampaio et al., 2021; Van Steenkiste et al., 2021). Our ventions have been suggested to mitigate the effects of infectious study captures a longer follow-up than most of the previous stud- disease epidemics on population mental health (Leon et al., 1997; ies among healthcare workers with a considerably larger sample, North and Pfefferbaum, 2013; Kisely et al., 2020), evidence of and it clearly suggests that mental health impact is maintained their effectiveness among healthcare workers is very limited and during the second wave of the pandemic (4 months after the base- of insufficient quality (Pollock et al., 2020). Specific e-health line assessment). This finding contrasts with studies in the general interventions for healthcare workers are of increasing interest population, which show that the high impact at the beginning of (Drissi et al., 2021). Our results support previous recommenda- the first wave of the pandemic tends to decline after 3–5 weeks of tions that when selecting interventions aimed at supporting front- the first wave (Gonzalez-Sanguino et al., 2020; Daly and line workers’ mental health, organisational, social, personal and Robinson, 2021; Fancourt et al., 2021; Robinson and Daly, psychological factors may all be important (Pollock et al., 2020). 2021). Nevertheless, some general population studies show a per- sistently high level of mental health impact through the initial Persistent and new-onset probable cases phases of the pandemic (Kikuchi et al., 2020; McGinty et al., We stratified the analyses according to the mental health status 2020) while still another study shows that a substantial proportion after the first wave of the pandemic (baseline assessment). This of the general population maintained high levels of symptoms has allowed us to estimate the ‘persistence’ of probable mental dis- (McPherson et al., 2021). A possible explanation of the different orders and the rate of appearance of ‘incident’ disorders. This trajectory among healthcare workers is the maintained levels of approach is quite novel in the literature of the impact of proximal stressors they faced, like other essential workers (Paul Covid-19, and our results suggest: (a) that most of the prevalent et al., 2021). Also, we found that Panic Attacks and GAD tended disorders at follow up are at the expense of workers that had to be less persistent than MDD and PTSD. This might indicate already a disorder at baseline; and (b) that, nevertheless, new they will become the more lasting mental health impacts of probable cases appear during the second wave and they are asso- Covid-19 among healthcare workers. But the 4-month follow-up ciated with almost exactly the same proximal risk factors that are period in our study might be short to assess stable trends. More associated with persistence. The first finding suggests that persist- research is needed, assessing proximal stressors and mental health ent mental disorders may be an emergent issue among healthcare https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press Epidemiology and Psychiatric Sciences 13 workers after the Covid-19 pandemic. And the second, strongly the baseline assessment because of a slow roll-in period of health suggests the need to implement more effective attenuation of institutions. But we did adjust by a centre and by assessment date work-related and health-related factors during the evolution of to minimise any possible bias. the pandemic. Similar to a recently published study with Spanish healthcare workers (Mediavilla et al., 2021), our results Conclusions suggest a high persistent negative mental health impact of Covid-19. Mental health monitoring and facilitating access to ser- Notwithstanding the above-mentioned limitations, our study vices are needed to decrease the likelihood of chronicity among indicates that the prevalence of probable mental disorders healthcare workers. among Spanish healthcare workers during the second wave of the Covid-19 pandemic was similarly high to that after the first Strengths and limitations wave. This was in good part due to the persistence of mental dis- Our study has several strengths. First, it is based on a large sample orders detected at the baseline, but with a relevant incidence of of healthcare workers of a wide distribution of public health ser- about 1 in 5 of HCWs without mental disorders during the vices in Spain, including hospitals, primary care and emergency first wave of the Covid-19 pandemic. Health-related factors, work- and public health departments. All workers in each institution related factors and interpersonal stress are important risks of per- were invited to the study, based on a clear sampling frame and sistence of mental disorders and of incidence of mental disorders. providing a wide picture of exposure to acute stressors. Our Adequately addressing these factors might have prevented a con- study is unique in following the same healthcare workers and dif- siderable amount of mental health impact of the pandemic among ferentiating new onset of probable disorders and persistence of this vulnerable population. disorders present after wave one. Supplementary material. The supplementary material for this article can The study has also some limitations that should be considered. be found at https://doi.org/10.1017/S2045796022000130 First, the follow-up cooperation rate was incomplete (65.7%). In addition, the cooperation at baseline was even lower (Alonso Data. The data will be available in the future in a repository that the funding et al., 2021), although the real figure is difficult to assess, since institution (Instituto de Salud Carlos III) is currently developing. For any add- data from one large hospital in our study showed that the propor- itional information contact with the corresponding author. tion of emails ‘seen’ by target workers was just over 26% at base- Acknowledgements. The authors would like to sincerely thank all healthcare line. Although differences between workers participating in the workers participating in the study. They also thank Puri Barbas for the manage- follow-up survey and those not participating were very small, ment of the project and Carme Gasull for helping with the manuscript preparation. inverse probability weighting has been applied to correct for differential probabilities of participation at follow up surveys Author contributors. J. A., G. V. and P. M. reviewed the literature. J. A., based on baseline characteristics. Additionally, we used post- G. V., P. M., M. F., E. A., V. P. S., J. M. H., R. C. K. and R. B. conceived stratification so that each survey matches the distribution of age, and designed the study. E. A., J. D. M., N. L., T. P., J. M. P. T., J. I. P., J. I. E., M. E., N. P., A. G. P., C. R., E. A., I. C. G., A. A. P., M. C., A. P. Z., gender and profession in each of the health institutions included E. V., C. S. and V. P. S. acquired the data. G. V., I. A., F. A. and PM cleaned in the study. Moreover, response rates across the 18 participating and analysed the data. J. A., G. V. and P. M. drafted the initial version of the institutions were not correlated with the prevalence of any mental manuscript. All authors reviewed the initial draft and made a critical contribu- disorder (Spearman R = −0.12, p = 0.65). Also, data come from tion to the interpretation of the data and approved the manuscript. The cor- one country only which limits the external validity. Second, it is responding author attests that all listed authors meet authorship criteria and important to note that we used screening measures of mental dis- that no others meeting the criteria have been omitted. orders. While they have shown to have acceptable validity for Financial support. Instituto de Salud Carlos III/ Ministerio de Ciencia identifying individuals with a high risk of a particular mental dis- e Innovación/ FEDER (J. A., grant number COV20/00711); Project “PI17/ order, results cannot be interpreted as clinical diagnoses. 00521”, funded by Instituto de Salud Carlos III (ISCIII) and co-funded by Moreover, the PHQ might overestimate the prevalence of depres- the European Union, PERIS, Health Dpt, Generaliat de Catalunya (I. A., grant sion (Levis et al., 2020). Thus, we consider them only ‘probable’ number SLT017/20/000009); ISCIII-FSE+, Miguel Servet (P. M., grant number mental disorders. But it is important to clarify that a positive CP21/00078); ISCIII-FSE, Sara Borrell (P. M., grant number CD18/00049), result in each of these measures indicates relevant psychopath- Generalitat de Catalunya (2017SGR452). Additional partial funding was ology. One advantage of using these particular measures is that received from the Gerencia Regional de Salud de Castilla y León (SACYL) results can be compared with many existing data with other popu- (J. M. P. T., grant number GRS COVID 32/A/20). lations and time periods. Nevertheless, two-phase studies would Conflict of interest. E. A. reports personal fees from Lündbeck, Esteve, provide a more valid estimation (Dunn et al., 1999). Third, Schwabe and Roche. E. V. reports personal fees from Abbott, Allergan, both distal and proximal factors used in the analyses were gath- Angelini, Lundbeck, Sage and Sanofi, grants from Novartis and Ferrer and ered at baseline. While there is the possibility that proximal fac- grants and personal fees from Janssen, outside the submitted tors varied during the follow-up, we assumed this was minimal work. J. D. M. reports personal fees from Janssen and Angelini, personal given the follow-up period was 4 months on average. Moreover, fees and non-financial support from Otsuka, Lundbeck and Accord, outside in the analysis, we wanted to ensure that all proximal/distal factors the submitted work. J. M. P. T. reports personal fees from Angelini, Janssen occurred before the outcome (incidence/persistence). Finally, 4 and Lunbeck, and grants from Janssen, outside the submitted months of follow-up is a short time period for a complete under- work. R. C. K. was a consultant for Datastat, Inc., Holmusk, RallyPoint standing of the mental impact dynamics of the Covid-19 pan- Networks, Inc., Sage Therapeutics and he has stock options in Mirah, demic. Longer follow-up periods are necessary. A paper P. Y. M., and Roga Sciences. All other authors reported no conflict of interest. indicates that 56% of Spanish healthcare workers remained symp- Ethical standards. The authors assert that all procedures contributing to tomatic or worsened over time in terms of psychological distress this study comply with the ethical standards of the relevant national and insti- (Mediavilla et al., 2021). In addition, length of follow-up differed tutional committees on human experimentation and with the Helsinki by healthcare workers. This was due to an extended duration of Declaration of 1975, as revised in 2000. https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press 14 J. Alonso et al. References systematic review and meta-analysis. Journal of Advanced Nursing 77, 3286–3302. Alonso J, Vilagut G, Mortier P, Ferrer M, Alayo I, Aragon-Peña A, Garcia-Campayo J, Zamorano E, Ruiz MA, Pardo A, Perez-Paramo M, Aragones E, Campos M, Cura-Gonzalez ID, Emparanza JI, Espuga M, Lopez-Gomez V, Freire O and Rejas J (2010) Cultural adaptation into Forjaz MJ, Gonzalez-Pinto A, Haro JM, Lopez-Fresneña N, Salazar Spanish of the generalized anxiety disorder-7 (GAD-7) scale as a screening ADM, Molina JD, Orti-Lucas RM, Parellada M, Pelayo-Teran JM, tool. Health and Quality of Life Outcomes 8, 8. Perez-Zapata A, Pijoan JI, Plana N, Puig MT, Rius C, Gonzalez-Sanguino C, Ausin B, Castellanos MA, Saiz J, Lopez-Gomez A, Rodriguez-Blazquez C, Sanz F, Serra C, Kessler RC, Bruffaerts R, Ugidos C and Muñoz M (2020) Mental health consequences during the Vieta E and Perez-Sola V and MINDCOVID Working Group (2021) initial stage of the 2020 coronavirus pandemic (COVID-19) in Spain. Mental health impact of the first wave of COVID-19 pandemic on Brain, Behavior, and Immunity 87, 172–176. Spanish healthcare workers: a large cross-sectional survey. Revista de Hinkin CH, Castellon SA, Dickson-Fuhrman E, Daum G, Jaffe J and Jarvik Psiquiatria y Salud Mental (Engl Ed) 14, 90–105. L (2001) Screening for drug and alcohol abuse among older adults using Barroilhet S, Cano-Prous A, Cervera-Enguix S, Forjaz MJ and a modified version of the CAGE. American Journal on Addictions 10, Guillen-Grima F (2009) A Spanish version of the family assessment device. 319–326. Social Psychiatry and Psychiatric Epidemiology 44, 1051–1065. Hirten RP, Danieletto M, Tomalin L, Choi KH, Zweig M, Golden E, Kaur S, Berry JO and Jones WH (1995) The parental stress scale: initial psychometric Helmus D, Biello A, Pyzik R, Calcogna C, Freeman R, Sands BE, evidence. Journal of Social and Personal Relationship 12, 463–472. Charney D, Bottinger EP, Keefer L, Suarez-Farinas M, Nadkarni GN Blasco MJ, Castellvi P, Almenara J, Lagares C, Roca M, Sese A, Piqueras JA, and Fayad ZA (2020) Factors associated with longitudinal psychological Soto-Sanz V, Rodriguez-Marin J, Echeburua E, Gabilondo A, Cebria AI, and physiological stress in health care workers during the COVID-19 pan- Miranda-Mendizabal A, Vilagut G, Bruffaerts R, Auerbach RP, Kessler demic. medRxiv preprint. https://doi.org/10.1101/2020.12.21.20248593. RC and Alonso J and Universal Study Group (2016) Predictive models INC SI (2014) SAS Software 9.4. Cary, NC: SAS Inst. Inc. Mark Co. for suicidal thoughts and behaviors among Spanish University students: Jordan JA, Shannon C, Browne D, Carroll E, Maguire J, Kerrigan K, rationale and methods of the UNIVERSAL (University & mental health) Hannan S, McCarthy T, Tully MA, Mulholland C and Dyer KFW project. BMC Psychiatry 16, 122. (2021) COVID-19 staff wellbeing survey: longitudinal survey of psycho- Boden M, Cohen N, Froelich JM, Hoggatt KJ, Abdel Magid HS and logical well-being among health and social care staff in Northern Ireland Mushiana SS (2021a) Mental disorder prevalence among populations during the COVID-19 pandemic. British Journal of Psychiatry Open 7, e159. impacted by coronavirus pandemics: a multilevel meta-analytic study of Kessler RC, Santiago PN, Colpe LJ, Dempsey CL, First MB, Heeringa SG, COVID-19, MERS & SARS. General Hospital Psychiatry 70, 124–133. Stein MB, Fullerton CS, Gruber MJ, Naifeh JA, Nock MK, Sampson Boden M, Zimmerman L, Azevedo KJ, Ruzek JI, Gala S, Abdel Magid HS, NA, Schoenbaum M, Zaslavsky AM and Ursano RJ (2013) Clinical Cohen N, Walser R, Mahtani ND, Hoggatt KJ and McLean CP (2021b) reappraisal of the composite international diagnostic interview screening Addressing the mental health impact of COVID-19 through population scales (CIDI-SC) in the army study to assess risk and resilience in service- health. Clinical Psychology Review 85, 102006. members (Army STARRS). International Journal of Methods in Psychiatric Chigwedere OC, Sadath A, Kabir Z and Arensman E (2021) The impact of Research 22, 303–321. epidemics and pandemics on the mental health of healthcare workers: a sys- Kikuchi H, Machida M, Nakamura I, Saito R, Odagiri Y, Kojima T, tematic review. International Journal of Environmental Research and Public Watanabe H, Fukui K and Inoue S (2020) Changes in psychological dis- Health 18, 6695. tress during the COVID-19 pandemic in Japan: a longitudinal study. Daly M and Robinson E (2021) Acute and longer-term psychological distress Journal of Epidemiology 30, 522–528. associated with testing positive for COVID-19: longitudinal evidence from a Kisely S, Warren N, McMahon L, Dalais C, Henry I and Siskind D (2020) population-based study of US adults. Psychological Medicine, 1–8, [Epub Occurrence, prevention, and management of the psychological effects of ahead of print]. doi: 10.1017/S003329172100324X. emerging virus outbreaks on healthcare workers: rapid review and de Kock JH, Latham HA, Leslie SJ, Grindle M, Muñoz SA, Ellis L, Polson R meta-analysis. BMJ 369, m1642. and O’Malley CM (2021) A rapid review of the impact of COVID-19 on the Kroene K, Spitzer RL, Williams JB and Lowe B (2010) The patient health mental health of healthcare workers: implications for supporting psycho- questionnaire somatic, anxiety, and depressive symptom scales: a systematic logical well-being. BMC Public Health 21, 104. review. General Hospital Psychiatry 32, 345–359. Diez Martinez S, Martin Moros JM, Altisent Trota R, Aznar Tejero P, Krysinska K and Martin G (2009) The struggle to prevent and evaluate: appli- Cebrian Martin C, Imaz Perez FJ and del Castillo Pardo M (1991) cation of population attributable risk and preventive fraction to suicide pre- Brief questionnaires for the early detection of alcoholism in primary health vention research. Suicide & Life-Threatening Behavior 39, 548–557. care. Atencion Primaria 8, 367–370. Kunzler AM, Rothke N, Gunthner L, Stoffers-Winterling J, Tuscher O, Dohrenwend BS, Krasnoff L, Askenasy AR and Dohrenwend BP (1978) Coenen M, Rehfuess E, Schwarzer G, Binder H, Schmucker C, Exemplification of a method for scaling life events: the peri life events Meerpohl JJ and Lieb K (2021) Mental burden and its risk and protective scale. Journal of Health and Social Behavior 19, 205–229. factors during the early phase of the SARS-CoV-2 pandemic: systematic Drissi N, Ouhbi S, Marques G, de la Torre Diez I, Ghogho M and Janati review and meta-analyses. Globalization and Health 17, 34. Idrissi MA (2021) A systematic literature review on e-mental health solu- Leon AC, Olfson M, Portera L, Farber L and Sheehan DV (1997) Assessing tions to assist health care workers during COVID-19. Telemedicine psychiatric impairment in primary care with the Sheehan disability scale. Journal and e-Health 27, 594–602. International Journal of Psychiatry in Medicine 27, 93–105. Dunn G, Pickles A, Tansella M and Vazquez Barquero JL (1999) Two-phase Levis B, Benedetti A, Ioannidis J, Sun Y, Negeri Z, He C, Wu Y, Krishnan epidemiological surveys in psychiatric research. British Journal of Psychiatry A, Bhandari P, Neupane D, Imran M, Rice D, Riehm K, Saadat N, Azar 174, 95–100. M, Boruff J, Cuijpers P, Gilbody S, Kloda L, McMillan D, Patten S, Epstein NB, Baldwin LM and Bishop DS (1983) The McMaster family assess- Shrier I, Ziegelstein R, Alamri S, Amtmann D, Ayalon L, Baradaran ment device. Journal of Marital and Family Therapy 9, 171–180. H, Beraldi A, Bernstein C, Bhana A, Bombardier C, Carter G, Chagas Eysenbach G (2004) Improving the quality of web surveys: the checklist for M, Chibanda D, Clover K, Conwell Y, Diez-Quevedo C, Fann J, reporting results of internet E-surveys (CHERRIES). Journal of Medical Fischer F, Gholizadeh L, Gibson L, Green E, Greeno C, Hall B, Haroz Internet Research 6, e34. E, Ismail K, Jetté N, Khamseh M, Kwan Y, Lara M, Liu S, Loureiro S, Fancourt D, Steptoe A and Bu F (2021) Trajectories of anxiety and depressive Löwe B, Marrie R, Marsh L, McGuire A, Muramatsu K, Navarrete L, symptoms during enforced isolation due to COVID-19 in England: a longi- Osório F, Petersen I, Picardi A, Pugh S, Quinn T, Rooney A, Shinn E, tudinal observational study. The Lancet Psychiatry 8, 141–149. Sidebottom A, Spangenberg L, Tan P, Taylor-Rowan M, Turner A, van Galanis P, Vraka I, Fragkou D, Bilali A and Kaitelidou D (2021) Nurses’ Weert H, Vöhringer P, Wagner L, White J, Winkley K and Thombs B burnout and associated risk factors during the COVID-19 pandemic: a (2020) Patient Health Questionnaire-9 scores do not accurately estimate https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press Epidemiology and Psychiatric Sciences 15 depression prevalence: individual participant data meta-analysis. Journal of 2-month follow-up observational study of 21,874 adults in England during Clinical Epidemiology 122, 115–128. the COVID-19 pandemic. medRxiv preprint. https://doi.org/10.1101/2021. Lopez Steinmetz LC, Herrera CR, Fong SB and Godoy JC (2022) A longitu- 04.20.21255817. dinal study on the changes in mental health of healthcare workers during Phiri P, Ramakrishnan R, Rathod S, Elliot K, Thayanandan T, Sandle N, the COVID-19 pandemic. Psychiatry 85, 56–71. Haque N, Chau SW, Wong OW, Chan SS, Wong EK, Raymont V, Mansfiedl AK, Keitner GI and Sheeran T (2019) The brief assessment of Au-Yeung SK, Kingdon D and Delanerolle G (2021) An evaluation of family functioning scale (BAFFS): a three-item version of the general func- the mental health impact of SARS-CoV-2 on patients, general public and tioning scale of the family assessment device. Psychotherapy Research 29, healthcare professionals: a systematic review and meta-analysis. 824–831. EClinicalMedicine 34, 100806. McFadden P, Neil RD, Moriarty J, Gillen P, Mallett J, Manthorpe J, Currie D, Pollock A, Campbell P, Cheyne J, Cowie J, Davis B, McCallum J, McGill K, Schroder H, Ravalier J, Nicholl P, McFadden D and Ross J (2021) A cross- Elders A, Hagen S, McClurg D, Torrens C and Maxwell M (2020) sectional examination of the mental wellbeing, coping and quality of working Interventions to support the resilience and mental health of frontline health life in health and social care workers in the UK at two time points of the and social care professionals during and after a disease outbreak, epidemic COVID-19 pandemic. Epidemiologia 2, 227–242. or pandemic: a mixed-methods systematic review. The Cochrane Database McGinty EE, Presskreischer R, Anderson KE, Han H and Barry CL (2020) of Systematic Reviews 11, CD013779. Psychological distress and COVID-19-related stressors reported in a longi- Resick PA, Monson CM and Chard KM (2020) Cognitive Processing Therapy tudinal cohort of US adults in April and July 2020. JAMA 324, 2555–2557. for Posttraumatic Stress Disorder (PTSD). Washington, DC: Department of McPherson KE, McAloney-Kocaman K, McGlinchey E, Faeth P and Veteran Affairs. Armour C (2021) Longitudinal analysis of the UK COVID-19 psycho- Roberts T, Daniels J, Hulme W, Hirst R, Horner D, Lyttle MD, Samuel K, logical wellbeing study: trajectories of anxiety, depression and Graham B, Reyanrd C, Barrett M, Foley J, Cronin J, Umana E, Vinagre J COVID-19-related stress symptomology. Psychiatry Research 304, 114138. and Carlton E (2021) Psychological distress and trauma in doctors provid- Mdege ND and Lang J (2011) Screening instruments for detecting illicit drug ing frontline care during the COVID-19 pandemic in the United Kingdom use/abuse that could be useful in general hospital wards: a systematic review. and Ireland: a prospective longitudinal survey cohort study. BMJ Open 11, Addictive Behaviors 36, 1111–1119. e049680. Mediavilla R, Fernández-Jiménez E, Martinez-Morata I, Jaramillo F, Robinson E and Daly M (2021) Explaining the rise and fall of psychological Andreo-Jover J, Morán-Sánchez I, Mascayano F, Moreno-Küstner B, distress during the COVID-19 crisis in the United States: longitudinal evi- Minué S, Ayuso-Mateos J, Bryant R, Bravo-Ortiz M and Martínez-Alés G dence from the understanding America study. British Journal of Health (2021) Sustained negative mental health outcomes among healthcare workers Psychology 26, 570–587. over the first year of the COVID-19 pandemic: a prospective cohort study. Rodriguez RM, Montoy JC, Hoth KF, Talan DA, Harland KK, Eyck PT, medRxiv preprint. https://doi.org/10.1101/2021.11.21.21266594. MOwer W, Krishnadasan A, Santibañez S and Mohr N and Project Miguel-Puga JA, Cooper-Bribiesca D, Avelar-Garnica FJ, Sanchez-Hurtado COVERED Emergency Department Network (2021) Symptoms of anx- LA, Colin-Martinez T, Espinosa-Poblano E, Anda-Garay JC, Gonzalez- iety, burnout, and PTSD and the mitigation effect of serologic testing in Diaz JI, Segura-Santos OB, Vital-Arriaga LC and Jauregui-Renaud K emergency department personnel during the COVID-19 pandemic. (2021) Burnout, depersonalization, and anxiety contribute to post-traumatic Annals of Emergency Medicine 78, 35–43 e2. stress in frontline health workers at COVID-19 patient care, a follow-up Sampaio F, Sequeira C and Teixeira L (2021) Impact of COVID-19 outbreak study. Brain and Behavior 11, e02007. on nurses’ mental health: a prospective cohort study. Environmental Mortier P, Vilagut G, Ferrer M, Alayo I, Bruffaerts R, Cristobal-Narvaez P, Research 194, 110620. del Cura-Gonzalez I, Domenech-Abella J, Felez-Nobrega M, Olaya B, Santabarbara J, Bueno-Notivol J, Lipnicki DM, Olaya B, Perez-Moreno M, Pijoan JI, Vieta E, Perez-Sola V, Kessler RC, Haro JM and Alonso J Gracia-Garcia P, Idoiaga-Mondragon N and Ozamiz-Etxebarria N (2021) (2021) Thirty-day suicidal thoughts and behaviors in the Spanish adult gen- Prevalence of anxiety in health care professionals during the COVID-19 eral population during the first wave of the Spain COVID-19 pandemic. pandemic: a rapid systematic review (on published articles in Medline) Epidemiology and Psychiatric Sciences 30, e-19, [Epub ahead of print]. with meta-analysis. Progress in Neuro-Psychopharmacology & Biological Mortier P, Vilagut G, Alayo I, Ferrer M, Amigo F, Aragonès E, Psychiatry 107, 110244. Aragón-Peña A, Asúnsolo del Barco A, Campos M, Espuga M, Sasaki N, Asaoka H, Kuroda R, Tsuno K, Imamura K and Kawakami N González-Pinto A, Haro J, López Fresneña N, Martínez de Salázar A, (2021) Sustained poor mental health among healthcare workers in Molina J, Ortí-Lucas R, Parellada M, Pelayo-Terán J, Pérez-Gómez B, COVID-19 pandemic: a longitudinal analysis of the four-wave panel survey Pérez-Zapata A, Pijoan J, Plana N, Polentinos-Castro E, Portillo-Van over 8 months in Japan. Journal of Occupational Health 63, e12227. Diest A, Puig M, Rius C, Sanz F, Serra C, Urreta-Barallobre I, Kessler Schisterman E, Cole S and Platt R (2009) Overadjustment bias and unneces- R, Bruffaerts R, Vieta E, Pérez-Solá V and Alonso J and sary adjustment in epidemiologic studies. Epidemiology (Cambridge, Mass.) MINDCOVID Working Group (2022) Four-month incidence of suicidal 20, 488–495. thoughts and behaviors among healthcare workers after the first wave of Seaman SR, White IR, Copas AJ and Li L (2012) Combining multiple imput- the Spain COVID-19 pandemic. Journal of Psychiatric Research, 149, 10–17. ation and inverse-probability weighting. Biometrics 68, 129–137. Navarro-Mateu F, Moran-Sanchez I, Alonso J, Tormo MJ, Pujalte ML, Sun P, Wang M, Song T, Wu Y, Luo J, Chen L and Yan L (2021) The psy- Garriga A, Aguilar-Gaxiola S and Navarro C (2013) Cultural adaptation chological impact of COVID-19 pandemic on health care workers: a sys- of the Latin American version of the World Health Organization tematic review and meta-analysis. Frontiers in Psychology 12, 626547. Composite International Diagnostic Interview (WHO-CIDI) (v 3.0) for Team RC (2021) R: A Language and Environment for Statistical Computing. use in Spain. Gaceta Sanitaria 27, 325–331. Vienna, Austria: R Foundation for Statistical Computing. Nock MK, Stein MB, Heeringa SG, Ursano RJ, Colpe LJ, Fullerton CS, Van Bureen S (2012) Flexible Imputation of Missing Data, Chapter 2. Multiple Hwang I, Naifeh JA, Sampson NA, Schoenbaum M, Zaslavsky AM, Imputation. Boca Raton: Chapman and Hall/CRC Press. Kessler RC and Army SC (2014) Prevalence and correlates of suicidal behav- Van Buuren S and Groothuis-Oudshoorn K (2011) mice: multivariate imput- ior among soldiers: results from the army study to assess risk and resilience in ation by Chained equations in R. Journal of Statistical Software 45, 1–67. servicemembers (Army STARRS). JAMA Psychiatry 71, 514–522. Van Steenkiste E, Schoofs J, Gilis S and Messiaen P (2021) Mental health North CS and Pfefferbaum B (2013) Mental health response to community impact of COVID-19 in frontline healthcare workers in a Belgian Tertiary disasters: a systematic review. JAMA 310, 507–518. care hospital: a prospective longitudinal study. Acta Clinica Belgica 1–8 Oronoz B, Alonso-Arbiol I and Balluerka N (2007) A Spanish adaptation of [Epub ahead of print]. doi: 10.1080/17843286.2021.1903660. the parental stress scale. Psicothema 19, 687–692. Weathers FW, Litz BT, Keane TM, Palmieri PA, Marx BP and Schnurr PP Paul E, Wan Mak H, Fancourt D and Bu F (2021) Comparing mental health (2013) The PTSD Checklist for DSM-5 (PCl-5). Washington, USA: trajectories of four different types of key workers with non-key workers: a Department of Veterans Affairs. https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press 16 J. Alonso et al. Wu Y, Levis B, Riehm KE, Saadat N, Levis AW, Azar M, Rice DB, Boruff J, P, Picardi A, Reuter K, Rooney AG, Santos IS, Shaaban J, Sidebottom Cuijpers P, Gilbody S, Ioannidis JP, Kloda LA, McMillan D, Patten SB, A, Simning A, Stafford L, Sung S, Tan PL, Turner A, van Weert HC, Shrier I, Ziegelstein RC, Akena DH, Arroll B, Ayalon L, Baradaran HR, White J, Whooley MA, Winkley K, Yamada M, Benedetti A and Baron M, Bombardier CH, Butterworth P, Carter G, Chagas MH, Chan Thombs BD (2020) Equivalency of the diagnostic accuracy of the PHQ-8 JC, Cholera R, Conwell Y, Ginkel dM-v, Fann JR, Fischer FH, Fung D, and PHQ-9: a systematic review and individual participant data Gelaye B, Goodyear-Smith F, Greeno CG, Hall BJ, Harrison PA, meta-analysis. Psychological Medicine 50, 1368–1380. Harter M, Hegerl U, Hides L, Hobfoll SE, Hudson M, Hyphantis T, Zuromski KL, Ustun B, Hwang I, Keane TM, Marx BP, Stein MB, Inagaki M, Jette N, Khamseh ME, Kiely KM, Kwan Y, Lamers F, Liu Ursano RJ and Kessler RC (2019) Developing an optimal short-form SI, Lotrakul M, Loureiro SR, Lowe B, McGuire A, Mohd-Sidik S, of the PTSD checklist for DSM-5 (PCL-5). Depression and Anxiety 36, Munhoz TN, Muramatsu K, Osorio FL, Patel V, Pence BW, Persoons 790–800. https://doi.org/10.1017/S2045796022000130 Published online by Cambridge University Press