key: cord-0325134-99go6vs3
authors: Joensen, A.; Danielsen, S.; Andersen, P. K.; Groot, J.; Strandberg-Larsen, K.
title: The impact of the initial and 2nd national COVID-19 lockdown on mental health in young people with and without pre-existing depressive symptoms
date: 2021-11-03
journal: nan
DOI: 10.1101/2021.11.03.21265861
sha: 66ce6b4c3d589e81f24b50dc8baa68ce82899919
doc_id: 325134
cord_uid: 99go6vs3

Background The evidence on mental well-being and loneliness among young people during the initial lockdown is mixed, and little is known about the long-lasting impact of the sequential lockdowns. We examine changes in young peoples mental health from before to during the initial and second more prolonged lockdown, and whether women and those with pre-existing depressive symptoms were disproportionally impacted. Methods Participants reported on mental health indicators in an ongoing 18-year data collection in the Danish National Birth Cohort and in a COVID-19 survey, including 8 data points: 7 in the initial lockdown, and 1 year post. Changes in quality of life (QoL), mental well-being, and loneliness were estimated with random effect linear regressions on longitudinal data (N=32,985), and linear regressions on repeated cross-sections (N=28,579). Findings Interim deterioration in mental well-being and loneliness was observed during the initial lockdown, and only in those without pre-existing depressive symptoms. During the second lockdown, a modest deterioration was again observed for mental well-being and loneliness. QoL likewise only declined among those without pre-existing symptoms, where women showed a greater decline than men. QoL did not normalise during the initial lockdown and remained at lower levels during the second lockdown. These findings were not replicated in the repeated cross-sections. Interpretation Except for an interim decrease in mental health during lockdown, and only in those without pre-existing depressive symptoms, this studys findings do not suggest a substantial detrimental impact of the lockdowns. Potential methodological differences in-between studies are a possible explanation for the mixed evidence. Funding The Velux Foundation

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint

Evidence before this study 92 We searched PubMed, PsycINFO, MedrXiv, and PsyArXiv with the terms ("Mental*" OR 93 "Psychological*" OR "Emotional*") AND ("Youth" OR "Young Adult*") AND ("COVID*" 94 OR "Coronavirus" OR "Lockdown*") for articles published in English between January 1 st 95 2020 and October 1 st 2021. Included studies varied in terms of quality of data used but 96 overall studies reported a detrimental impact of the lockdowns on young people's mental 97 health. However, the evidence on mental well-being and loneliness has shown to be 98 inconsistent and with signs of resilience. Young people, women, and those with a pre-existing 99 mental disorder have been identified as vulnerable subgroups, but only a few studies 100 investigating mental health in individuals with a pre-existing mental disorder included a pre-101 lockdown measurement. The included studies also demonstrated that there is a gap in the 102 evidence in understanding how mental health changed week by week, as well as the long-103 term impact over the course of the lockdowns. 104

105 Added value of this study 106

With longitudinal data, this study shows an interim impact of the initial and second lockdown 107 on mental health during the COVID-19 pandemic in young individuals without pre-existing 108 symptoms in Denmark. Since commencement of the initial lockdown, the levels of mental 109 health returned to before levels, but one year after the initial lockdown, the levels were still 110 lower than before lockdown in young people without pre-existing depressive symptoms. 111

Young individuals with pre-existing depressive symptoms did not experience more 112 detrimental impact of the lockdown, but rather indication of resilience or even improvements 113 in mental health were observed. A disproportional impact of the lockdown on women 114 compared to men was only observed for QoL, as women without pre-exiting depressive 115 symptoms experienced a greater decline in QoL than men without pre-existing depressive 116 symptoms. However, findings based on the repeated cross-sectional data did not show similar 117 interim impact -but instead no -or clinically irrelevant impact. Thus, taken together our 118 findings do not suggest a substantial lasting impact of the lockdowns on mental health among 119 young individuals. 120 121 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint Summary 131 Background: The evidence on mental well-being and loneliness among young people during 132 the initial lockdown is mixed, and little is known about the long-lasting impact of the 133 sequential lockdowns. We examine changes in young people's mental health from before to 134 during the initial and second more prolonged lockdown, and whether women and those with 135 pre-existing depressive symptoms were disproportionally impacted. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint following and through the initial lockdown, and during the second and more prolonged 191 lockdown. We further examined whether women and individuals with pre-existing depressive 192 symptoms were disproportionally impacted by the lockdowns, as we hypothesised these to be 193 most vulnerable. 194 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. The populations, in the present study, were restricted to participants with information on 212 household-socio-occupational status, maternal age at childbirth, parity, and maternal smoking 213 collected during pregnancy. In the analyses including the DNBC-18 and the COVID-19 214 survey, we further restricted our population to those eligible for the DNBC-18 before the 215 initial lockdown, Figure S1 . In total, 32,985 participants had complete data in the DNBC-18. 216

Of these, 7,431 and 8,808, respectively participated in wave 1 and wave 8 of the COVID-19 217 survey, Figure S1 . We also estimated the year-to-year and seasonal variation in mental health is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint

Participants reported their current educational enrolment and housing composition in the 263 DNBC-18. We also included information on gender, age, household-socio-occupational 264 status, maternal age at childbirth, parity, and maternal smoking collected during pregnancy. 265 These covariates were categorised as shown in Table S1 . 266 267

To account for differential attrition, we estimated inverse probability weights (IPW) by 269 logistic regressions with having data as outcome and the following predictors: gender, 270 household-socio-occupational status, maternal age at childbirth, parity, and maternal smoking 271 collected during pregnancy. Separate analyses were performed for each data point and on the 272 appropriate baseline population, Figure S1 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint across the waves of the COVID-19 survey, we exchanged the binary lockdown variable with 296 a variable indicating wave 1-8. Lastly, we restricted the before measure in the longitudinal 297 analyses to participants who completed DNBC-18 in year 2019 and period 2 or 3, to address 298 whether our results were biased by seasonal variation or the time gap between the pre-and 299 during lockdown measurement. 300 301 In the repeated cross-sectional setup, the during lockdown period was defined as the second 302 period in 2020 and onwards. We started out testing for interaction between lockdown and 303 period and omitted it if insignificant. We subsequently examined the disproportional impact 304 of lockdown on gender and depressive symptoms by including interaction terms as described 305 above. We performed these analyses unadjusted and adjusted for household-socio-306 occupational status, maternal age at childbirth, maternal smoking collected during pregnancy, 307 educational enrolment, and household composition. 308

The analyses were performed with SAS Software, version 9.4 (SAS Institute, North Carolina, 310 US) using the commands proc survey means, proc mixed/GLM and applying the weight 311 statement for IPW and random statement for random effect. 312 313

The funder of this study had no role in the study design, data collection, data analysis, data 315 interpretation, writing of the article, or in decision to publish. 316 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint

Slightly more than half of the participants were 18-20 years of age during the initial 318 lockdown. Moreover, more women than men participated in the DNBC-18, and were 319 undertaking education, living with parents, and from educated households. In the COVID-19 320 survey, seven out of ten participants were women in wave 1 and wave 8, while this 321 proportion slightly increased in wave 2-7, Table S1 . Moreover, participants in wave 2-7 were 322 more often without pre-existing depressive symptoms, under education, living with parents, 323 from educated households, nulliparous, non-smoking, and older mothers than participants 324 within the DNBC-18 and wave 1 and 8, Table S1 . 325 326 Before lockdown, women and young people with depressive symptoms reported lower 327 mental health than men and those without depressive symptoms, Table S2 . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint deteriorations in well-being and loneliness were greatest early in the initial lockdown. In 352 spring 2021, the overall changes in mental well-being and loneliness from before lockdown 353 were approximately the same as the change observed during the initial lockdown. When 354 investigating the degree of pre-existing symptoms, the deteriorations were greatest for the no 355 depressive symptom group and greatest improvements were seen for the severe group, Figure  356 S3. Restricting the before measure to data collected in spring and summer 2019 did not 357 change the overall conclusion, Figure S4 . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint

Within the longitudinal setup, this study demonstrates an interim deterioration in mental well-371 being and loneliness during the initial lockdown, and only in young people without pre-372 existing depressive symptoms. During the gradual re-opening of the second lockdown, the 373 mental well-being was equivalent to early in the initial lockdown, while the proportion of 374 loneliness was at levels during the reopening of the initial lockdown, thereby only slightly 375 increased. QoL likewise only declined following lockdown among young people without pre-376 existing symptoms, but women had a bigger decline in QoL than men. QoL did not normalise 377 among young people without pre-existing symptoms during the initial lockdown and 378 remained at lower levels in spring 2021. Overall, these deterioration in mental health seemed 379 rather modest, and were not replicated in repeated cross-sectional setup. Thus, taken together 380 our findings do not suggest a substantial lasting impact of the lockdowns on mental health 381 among young individuals. 382

The longitudinal data allow us to quantify the week-to-week variation in impact across the 384 entire span of the initial lockdown. In contrast, in the repeated cross-sections, the initial 385 lockdown is represented by one longer period. Mental well-being and loneliness seemed to 386 normalise during the gradual reopening of the initial lockdown, and this might explain why 387 we only observe deteriorations in QoL in young people without depressive symptoms in the 388 cross-sectional analyses. Feelings of loneliness and social isolation since the implementation 389 of the initial lockdown have been very dynamic and responsive to the government's 390 actions. 1,18 Additionally, in the COVID-19 survey it was explicitly stated that the aim was to 391 investigate how the COVID-19 pandemic impacted our living, which was not the case in the 392 ongoing DNBC-18, where COVID-19 was not mentioned and no adaptions were made. The 393 DNBC participants are regularly invited to complete age specific follow ups. Thus, it is 394 likely, however untestable, that the participants in the DNBC-18 deliberately compensated, so 395 their responses reflected their overall health and not solely their current lockdown situation. 396 397 Studies that have suggested a decline in young people's mental health are cross-sectional or 398 do not include a pre-lockdown measurement. 9,11-14,16 Most of the earlier findings do also 399 indicate that women were more detrimentally impacted by the lockdown than men, 2,3,7- is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint methodological differences. Regression to the mean is of concern, and longitudinal studies in 403 which adjustment for pre-lockdown mental health was performed, documented greater 404 deteriorations following the lockdown in individuals with pre-existing mental health 405 disorders. 7,8 Contrary, studies without adjustment for baseline values, as in our study, 406 document slightly improved or unchanged mental health following lockdown among those 407 with a pre-existing mental disorder. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint second and more prolonged lockdown. The repeated cross-sections allow us to quantify 437 seasonal variations and is only vulnerable to attrition if the participation in DNBC-18 438 systematically changed over year of birth or season, as opposed to the longitudinal setup 439 which is more vulnerable to attrition due to loss to follow-up. We attempted to reduce bias 440 from differential attrition by inverse probability weighting. The validity of this method relies 441 on a correctly specified model including all relevant predictors for loss to follow-up, which 442 cannot be assumed. 443

Interpretation of our findings likewise deserves consideration of a few limitations. In the 445 longitudinal setup, the baseline data was collected at age 18 years and three months for all 446 participants, whereas the participants' ages during lockdown was 18-24 years. Thus, the 447 timespan between the before and during lockdown measurement was greater for the older 448 participants. For older participants, changes in mental health may be underestimated, since 449 the pre-lockdown measurement represent a younger age than the follow-up measures, and on 450 average reporting on mental health instruments improves with age. 17 As an attempt to 451 preclude this, a sensitivity analyses was conducted to ensure that any observed deterioration 452 in mental health was likely attributable to the lockdown. Finally, the DNBC has previously 453 been shown to be healthier and more often from households with higher occupational levels 454 than the background population, and our population was mainly living with their parents and 455 studying. 30 Thus, the findings from this study cannot necessarily be generalised to all 18-24-456

year-olds or to other countries. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint *Repeated measures Random effect estimates and 95% CI presented (N=32,985) (Total number including repeated measures N=62,081) All models were weighted by IPW baseline population 1, Figure S2 (N=67,346) p-value for interaction between lockdown, gender, and pre-existing depressive symptoms (wave 1-8): QoL (p<0·001), mental well-being (p<0·001), and loneliness (p<0·001)

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint Figure 4 . Mean/ proportion [95% CI] of QoL, mental well-being, and loneliness stratified by gender and depressive symptoms, respectively (repeated cross-sectional setup) Figure 5 . Regression of changes in QoL, mental well-being, and loneliness from pre-to during lockdown (repeated cross-sectional setup) Crude and adjusted estimates and 95% CI are presented (N=28,579) *Adjusted for gender, period, household socio-occupational status, maternal age at childbirth, maternal smoking during pregnancy, educational enrolment, and housing composition. All models were weighted by IPW baseline population 2, Figure S2 (N=58,638) p-value for interaction between lockdown and depressive symptoms: QoL (p<0·001), mental well-being (p=0·3365), and loneliness (p<0·001)

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2021. ; https://doi.org/10.1101/2021.11.03.21265861 doi: medRxiv preprint

Standing together -at a distance': 471 Documenting changes in mental-health indicators in Denmark during the COVID-19 472 pandemic

Mental health and health behaviours 474 before and during the initial phase of the COVID-19 lockdown: Longitudinal analyses 475 of the UK Household Longitudinal Study

Increases in Loneliness Among Young Adults During 478 the COVID-19 Pandemic and Association With Increases in Mental Health Problems

López-Antón R, Santabárbara 481 J. Prevalence of depression during the COVID-19 outbreak: A meta-analysis of 482 community-based studies

COVID-19 pandemic and mental health consequences: 484 Systematic review of the current evidence

Mental Health During the First Year of the COVID-19 Pandemic: A Review and 489 Recommendations for Moving Forward

491 Longitudinal evidence for persistent anxiety in young adults through COVID-19 492 restrictions

Mental health before and during the 494 COVID-19 pandemic in two longitudinal UK population cohorts

Mental health during the 497 second wave of the covid-19 pandemic-polish studies

Decreasing mental well-being during the 500 COVID-19 pandemic: A longitudinal study among Danes before and during the 501 pandemic

Mental Health Burden of the COVID-19 503 Outbreak in Germany: Predictors of Mental Health Impairment

Psychological Distress and 506 Loneliness Reported by US Adults

Life satisfaction among adolescents 509 before and during the COVID-19 pandemic

Trajectories of anxiety and depressive symptoms during 512 enforced isolation due to COVID-19 in England: a longitudinal observational study

Longitudinal changes in mental health and the COVID-515 19 pandemic: Evidence from the UK Household Longitudinal Study

Depression reported by US

Mental health before and during the COVID-19 520 pandemic: a longitudinal probability sample survey of the UK population. The Lancet 521

Loneliness, worries, anxiety, and precautionary 523 behaviours in response to the COVID-19 pandemic: a longitudinal analysis of 200,000 524 Western and Northern Europeans

The mental health impact of the COVID-19 526 pandemic on people with and without depressive, anxiety, or obsessive-compulsive 527 disorders: a longitudinal study of three Dutch case-control cohorts

A Longitudinal 530 Investigation of the Effects of the COVID-19 Pandemic on the Mental Health of 531 Individuals with Pre-existing Severe Mental Illnesses

The Danish National Birth Cohort -its 534 background, structure and aim

Reliability and Validity of an Adapted Version of the Cantril 536 Ladder for Use with Adolescent Samples

Measuring mental well-being in 538 Denmark: Validation of the original and short version of the Warwick-Edinburgh 539 mental well-being scale (WEMWBS and SWEMWBS) and cross-cultural comparison 540 across four European settings

The Warwick-Edinburgh Mental Well-542 being Scale (WEMWBS): a valid and reliable tool for measuring mental well-being in 543 diverse populations and projects

Psychometric evaluation 545 of the Major Depression Inventory (MDI) as depression severity scale using the LEAD 546 (Longitudinal Expert Assessment of All Data) as index of validity

Regression to the mean: What it is and how 549 to deal with it

When is baseline 551 adjustment useful in analyses of change? An example with education and cognitive 552 change

ANCOVA versus change from baseline had more power in 554 randomized studies and more bias in nonrandomized studies

Simpson's Paradox, Lord's Paradox, and 557 Suppression Effects are the same phenomenon -The reversal paradox

Selection by socioeconomic factors into the 560 Danish National Birth Cohort