BJPsych Open (2021) 7, e30, 1–4. doi: 10.1192/bjo.2020.163 Short Report The paradoxical effect of COVID-19 outbreak on loneliness David Bartrés-Faz1, Dídac Macià1, Gabriele Cattaneo, Roger Borràs, Clara Tarrero, Javier Solana, José M. Tormos and Alvaro Pascual-Leone As in previous periods of quarantine, lockdown confinement appreciation by people for the labour and efforts of certain measures dictated to control SARS-CoV-2 would be expected to individuals, along with official campaigns designed to promote negatively affect mental health. We investigated the immediate feelings of inclusion and belonging, may have beneficial effects effects (over a 10 day period) of a strict nationwide stay-at-home on feelings of loneliness, a negative emotional state strongly order imposed in Spain, one of the countries most affected by regarded as a risk factor for impaired mental and general health the COVID-19 pandemic. Focusing our analysis on the feelings of status. Further assessments during the later stages of home loneliness, we obtained our measures within a social context confinement are now warranted. characterised by strong and continuous public and govern- mental support for increasing social bonds and cooperation in Keywords order to face the common public threat. Leveraging data from Psychological testing; rating scales; social functioning; commu- the Barcelona Brain Health Initiative, a prospective population- nity mental health teams; psychosocial interventions. based study cohort, the short UCLA Loneliness Scale was administered to 1604 participants 2 years and 1 year before the stay-at-home lockdown and repeated, on average, 10 days after Copyright and usage the official confinement order issued by the Spanish govern- © The Author(s), 2021. Published by Cambridge University Press ment. Ratings of loneliness remained stable during the 2 years on behalf of the Royal College of Psychiatrists. This is an Open before lockdown; however, they decreased significantly during Access article, distributed under the terms of the Creative the early stages of home confinement. This effect was particu- Commons Attribution licence (http://creativecommons.org/ larly significant for the item ‘feeling excluded from others’ and licenses/by/4.0/), which permits unrestricted re-use, distribu- was also observed among individuals who were confined alone. tion, and reproduction in any medium, provided the original work Overall, the results suggest that gestures and manifestations of is properly cited. In early April 2020, almost 1.5 million cases of COVID-19 had been Method confirmed in more than 200 countries, and over 80 000 deaths had already been recorded. Spain had the second highest number of con- Study participants were part of the Barcelona Brain Health Initiative firmed COVID-19 cases after the USA, and the highest rate of cases (BBHI, www.bbhi.cat/en), an ongoing longitudinal cohort study per inhabitant.1 As in many countries, the government in Spain investigating the determinants of brain and mental health in issued a nationwide stay-at-home order imposing strict lockdown middle and advanced age. The BBHI involves periodic medical, measures in mid-March.2 cognitive, brain imaging and biological assessments.9 The present The potentially negative psychological effects of quarantine study included a subsample of 1604 participants demographically have been described elsewhere,3 including observations of the asso- representative of the full BBHI cohort (n = 5592 as of April 2020). ciations between loneliness and mental health issues (e.g. anxiety, The mean age of the subsample was 55.7 years (s.d. 7.3); 65% depression, suicidal ideation) within the context of the COVID-19 were women, and 75% had completed higher education, 22% sec- pandemic,4 and of increases in feelings of loneliness following ondary and 3% only primary. The corresponding figures for the the stay-at-home restrictions in both general community samples5 full BBHI cohort were: mean age 54.4 years (s.d. 7.2), 67% and specific populations such as university students.6 However, women, 71% completed higher education, 25% secondary and 4% across the world people have been coming together during the only primary. pandemic in new ways to cope with the crisis and the challenge None of the respondents reported a diagnosis of a psychiatric or of confinement. For example, in Spain, every evening at 20:00 h, neurological disorder by a physician, and all completed the three from windows, front doors, balconies or terraces, thousands of waves of loneliness assessment using the short UCLA Loneliness people applauded frontline health professionals and acknowledged Scale,10 which consists of three items in which participants had them as ‘heroes of the nation’. Families and friends found to rate the frequency of several experiences (‘How often do you new ways to connect for online meals, drinks or games. As in feel isolated from others?’, ‘How often do you feel excluded?’ and other countries, the official motto of Spain’s governmental ‘How often do you feel that you lack company’) on a three-point COVID-19 campaign, ‘éste virus lo paramos unidos’ (united we Likert scale (options: ‘1: Hardly ever’, ‘2: Some of the time’, or ‘3: will stop this virus) reinforced the importance of social bonds and Often’). The scale was administered through the BBHI web platform cooperation. (https://bbhi.cat/en/), a personalised platform where participants In this study, we wanted to assess whether feelings of loneliness, are requested to fill in a series of questionnaires that assess their defined as the feeling that one’s desired quantity or quality of social self-perceived health and engagement in various lifestyles (cogni- connections is unfulfilled,7,8 in particular those referring to the tive, physical, sleep, social and nutrition habits). For the current subjective sense of social belonging, underwent changes during study, UCLA ratings were available for 2 years prior to the the early stages of the COVID-19 pandemic and, if so, whether COVID-19 lockdown and were obtained again some 10 days they increased or decreased. (range 9–13, s.d. 1.2) after official confinement. The authors certify that all procedures contributing to this study comply with 1 These authors contributed equally to this article. the ethical standards of the relevant national and institutional 1 Bartrés‐Faz et al Item 1 How often do you feel that you lack company? Responses Item 2 How often do you feel excluded? Often Item 3 How often do you feel isolated from others? Some of the time Hardly ever PRE-COVID-19 (2 y) Lack company Mean = 3.74 Exclusion s.d. = 1.18 Isolation 0 500 1000 1500 PRE-COVID-19 (1 y) Lack company Mean = 3.74 Exclusion s.d. = 1.17 Isolation 0 500 1000 1500 DURING-COVID-19 (1–2 weeks confinement) Lack Company Mean = 3.52 Exclusion s.d. = 0.9 Isolation 0 500 1000 1500 Frequency Fig. 1. Frequency of loneliness responses before and during COVID-19 confinement. committees on human experimentation and with the Helsinki Furthermore, this decrease did not occur homogeneously across Declaration of 1975, as revised in 2008. All procedures involving all items of the UCLA scale. Instead, one item (feelings of being human participants or patients were approved by the Unió excluded from others, item 2) drove most of the reduction. Thus, Catalana d’Hospitals ethics committee; approval numbers: CEIC the odds of ‘sometimes’ or ‘often’ feeling a lack of company (item 1), 17/06 (phase 1), CEI 18/07 (phase 2). Written informed consent excluded from others (item 2) or isolated (item 3) presented reduc- was obtained from all participants or patients. The data that tions of varying degree, the largest emerging for feelings of being support the findings of this study are available from the correspond- excluded (item 2). When the analysis was restricted to participants ing author, D.B.F., upon reasonable request. confined alone (N = 169), loneliness was overall higher than in those The longitudinal three-item UCLA score (the sum of the three not confined alone before and during COVID-19: (2 years before) = items of the scale) was modelled as a continuous variable using a 4.33, CI 4.15–4.53; (1 year before) = 4.28, CI 4.09–4.47 and (during linear mixed model allowing for a random intercept per partici- COVID-19) = 4.14, CI 3.95–4.30. Investigating the UCLA items pant. We also analysed each of the three individual items of the separately, we observed comparable patterns of significant reduc- scale separately, after dichotomisation of the responses. As there tions in loneliness during the confinement; once again, the largest were very few level 3 answers (‘often feeling lacking company, reductions were observed in feelings of exclusion (Table 1). excluded, or isolated’: <4%, see Fig. 1), we collapsed them with The differences in the decreases in item score probabilities level 2 (‘some of the times feeling …’) and opposed them to level 1 suggested a potential change in the internal coherence of the scale (‘hardly ever feeling …’). Thus, we obtained a separate binary longitudinal outcome for each of the three items of the scale, Table 1 Odds ratios of loneliness items before and during COVID-19 which we modelled with binomial linear mixed models including confinement a random intercept per participant. Fixed effect parameters could Odds ratio, hardly v. sometimes or often lonely be interpreted as odds ratios relative to the reference timepoint PRE (2 y)/PRE (1 y) DURING/PRE (1 y) (2 years before the COVID-19 outbreak). Confidence intervals All (N = 1604) were estimated from model parameter standard deviations. All ana- Lack of company 1.02, CI 0.83–1.26 0.44, CI 0.36–0.55a lyses were carried out with the functions lmer and glmer of the Exclusion 0.96, CI 0.78–1.21 0.23, CI 0.17–0.30a lme4 R package.11 Psychometric statistics were calculated following Isolation 0.98, CI 0.80–1.2 0.79, CI 0.64–0.98a the standard formulas implemented in the psych R package.12 Confined alone (N = 169) Lack of company 1.1, CI 0.60–2.03 0.47, CI 0.25, 0.86a Exclusion 1.05, CI 0.56–1.96 0.20, CI 0.09–0.44a Results Isolation 1.34, CI 0.79–2.27 1.39, CI 0.82–2.35 Odds ratios compare the odds of often or sometimes feeling lacking company, excluded or isolated compared with hardly ever in the two PRE-COVID-19 assessments (left col- Ratings from the three UCLA items remained stable for the two umn) and in the DURING-COVID-19 compared with the PRE-COVID-19 (1 y) assessment pre-COVID-19 years (expected fixed effect intercepts: (2 years (right column). Top: odds obtained from all participants having answered the three loneliness assessments. Bottom: odds obtained from participants confined alone after before) = 3.73, CI 3.70–3.75, (1 year before) = 3.73, CI 3.71– COVID-19 outbreak and who already lived alone before the outbreak. Odds ratios are derived from a binomial linear mixed model with a random intercept for each participant; 3.76). In clear contrast, scores fell significantly in the during 95% confidence intervals are obtained from the estimated standard deviation of the COVID-19 assessment (3.52, CI 3.50–3.54) (see also descriptive regression coefficients. a. 95% CI not including lack of effect (OR = 1). statistics in Fig. 1). 2 Effect of COVID‐19 outbreak on loneliness (cross-item correlation) after the pandemic outbreak. Indeed, or strengthening resources such as digital tools and the use of inter- we found that the DURING-COVID-19 pattern of scoring had net and social media to engage people creatively in new forms of altered the internal consistency of the UCLA scale. The two social support, improving access to information about local com- PRE-COVID-19 assessments exhibited good internal consistency munity groups and promoting social integration. This is aligned (Cronbach’s α = 0.76, CI 0.74–0.78, α = 0.75, CI 0.73–0.77) in agree- with initiatives that aim to enhance collective creativity and increase ment with the published literature,10 but in the DURING-COVID-19 the sense of ‘making things together’ which have been implemented evaluation, the consistency fell below psychometrically acceptable during the quarantine in charitable campaigns such as the Campaign levels (α = 0.63, CI 0.60–0.66). The main driver of this disruption to End Loneliness (see https://bemoreus.org.uk/the-quarantine-quilt- was again the change in scores for exclusion, its item-scale correl- connecting-through-creativity/). ation dropping from r = 0.63 and 0.62 in the two PRE-COVID-19 Pending future assessments to determine whether feelings of lone- assessments to r = 0.42 in the DURING-COVID-19 assessment. liness fluctuate further during more extended periods of confinement or in the face of changes in the social context, the present findings provide new evidence of the psychological responses to a state of con- Discussion finement, and may also offer valuable insights for future campaigns and interventions to reduce loneliness.14 Leveraging data from the BBHI study cohort, we observed that the influence of home confinement due to COVID-19 on the various David Bartrés-Faz , PhD, Department of Medicine, Faculty of Medicine and Health subcomponents of loneliness presented differences, to the point of Sciences, University of Barcelona, Spain; and Guttmann Institute, University Institute of altering the established internal consistency of the UCLA scale. In Neurorehabilitation affiliated to the Autonomous University of Barcelona (UAB), Spain; Dídac Macià, PhD, Department of Medicine, Faculty of Medicine and Health Sciences, other words, the subcomponents of the Loneliness scale were University of Barcelona, Spain; Gabriele Cattaneo, PhD, Guttmann Institute, University notably decoupled by the confinement. Feelings of exclusion from Institute of Neurorehabilitation affiliated to the Autonomous University of Barcelona (UAB), Spain; and Autonomous University of Barcelona, Spain; Roger Borràs, PhD, others (item 2) were the most reduced in all analyses, whereas Department of Child and Adolescent Psychiatry and Psychology, Institute of Cllinical scores on the Isolation subscale (item 3) decreased in the overall Neurosciences, Clinical and Provincial Hospital of Barcelona, Spain; Clara Tarrero, MS, sample but were unchanged in people living alone. These findings Guttmann Institute, University Institute of Neurorehabilitation affiliated to the Autonomous University of Barcelona (UAB), Spain; and Autonomous University of support the notion that loneliness is not a unitary, isolated construct Barcelona, Spain; Javier Solana, PhD, Guttmann Institute, University Institute of but rather represents a cluster of subjective and objective experi- Neurorehabilitation affiliated to the Autonomous University of Barcelona (UAB), Spain; and Autonomous University of Barcelona, Spain; JoséM. Tormos, MD, PhD, Guttmann ences of social integration and socioemotional states.10 Institute, University Institute of Neurorehabilitation affiliated to the Autonomous The observation that the UCLA scale item ‘feelings of exclusion’ University of Barcelona (UAB), Spain; and Autonomous University of Barcelona, Spain; Alvaro Pascual-Leone, MD, PhD, Guttmann Institute, University Institute of exhibited the greatest fall in score during a hard lockdown, which Neurorehabilitation affiliated to the Autonomous University of Barcelona (UAB), Spain; included an enforced reduction of physical social contacts, may be and Hinda and Arthur Marcus Institute for Aging Research, Hebrew SeniorLife, Harvard Medical School, Boston, Massachusetts, USA. related to the general contextual component, namely, that people knew that there was nothing to be done about the limitation on Correspondence: David Bartrés-Faz. Email: dbartres@ub.edu social contacts, and that everybody else was in the same situation. First received 3 Jun 2020, final revision 8 Dec 2020, accepted 9 Dec 2020 This interpretation may be aligned with findings of a recent longi- tudinal study during COVID-19 lockdown conducted among uni- versity students who, during the confinement, reported worrying less about the possibility that others were having more rewarding Funding experiences (i.e. fear of missing out) and perceived less competi- The research leading to these results received funding from ‘la Caixa’ Foundation (LCF/PR/ tion.6 In our study, during lockdown, contact with close family PR16/11110004), and also from the Guttmann Institute and Fundació Abertis. D.B.-F. was members and friends may have been more easily maintained funded by an ICREA Academia 2019 Award and by a Spanish Ministry of Science, Innovation and Universities (RTI2018-095181-B-C21) research grant. either within a shared household or using online methods, even though wider community life stopped, and this may have contribu- ted to the decrease in feelings of exclusion. Author contributions In addition to this effect, there are other plausible sociological G.C., J.S. and D.M. had full data access to all data in the study and take responsibility for the integrity of the data and the accuracy of the analyses. R.B. provided expert statistical analysis and psychological causes that might explain why feelings of support. Concept and design was undertaken by all authors. D.B.-F., A.P.-L. and D.M. drafted exclusion suddenly dropped during the first weeks of lockdown the manuscript. D.B.-F., A.P.-L. and J.M.T.carried out critical revision and supervision. in Spain. Some may appear obvious to many of us who lived through the experience, such as a sudden surge of patriotism, Data availability both spontaneous and fostered by the government’s communication The data that support the findings of this study are available from the corresponding author, campaign, or a shift from a focus on personal problems towards a D.B.-F., upon reasonable request. shared problem affecting the entire community. Our intriguing find- ings should invite replication with data from other countries, where Declaration of interest strict lockdowns were imposed, or even building on social identity None. theory propounding that times of uncertainty and crisis predispose individuals to powerful identification with groups,13 and should encourage further research into the psychological and political pre- dictors which, despite their observational nature, may shed light on References certain psychological and sociopolitical causal mechanisms. The main implications of our findings are that loneliness, and 1 Worldometer. COVID-19 Coronavirus Pandemic. Worldometer, 2020 (https:// in particular feelings of inclusion and social belonging, can be www.worldometers.info/coronavirus/?utm_campaign= homeAdvegas1?). improved without the need to enlarge the individual’s physical 2 Spanish Official Government Bulletin (Boletin Oficial del Estado). National Alarm State Order, March, 18, 2020. Official State Gazette Agency, 2020 (https://www. social network. Therefore, programmes designed to address loneli- boe.es/eli/es/rd/2020/03/14/463/con). ness should not solely aim to ‘physically’ reconnect people (e.g., 3 Brooks SK, Webster RK, Smith LE, Woodland L, Wessely S, Greenberg N, et al. Shared Lives UK; see https://sharedlivesplus.org.uk/news-campaigns- The psychological impact of quarantine and how to reduce it: rapid review of and-jobs/growing-shared-lives/) but should consider incorporating the evidence. Lancet 2020; 395(10227): 912–20. 3 Bartrés‐Faz et al 4 Tso IF, Park S. Alarming levels of psychiatric symptoms and the role of loneli- 10 Hughes ME, Waite LJ, Hawkley LC, Cacioppo JT. A short scale for measuring ness during the COVID-19 epidemic: a case study of Hong Kong. Psychiatry loneliness in large surveys: results from two population-based studies. Res Res 2020; 293: 113423. Aging 2004; 26(6): 655–72. 5 Killgore WDS, Cloonan SA, Taylor EC, Miller MA, Dailey NS. Three months 11 Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models of loneliness during the COVID-19 lockdown. Psychiatry Res 2020; 293: using lme4. J Stat Softw 2015; 67(1): 1–48. 113392. 12 Revelle W. psych: Procedures for Psychological, Psychometric, and Personality 6 Elmer T, Mepham K, Stadtfeld C. Students under lockdown: comparisons of Research. R package version 1.9.12. Northwestern University, 2019. Available students’ social networks and mental health before and during the COVID-19 from: https://CRAN.R-project.org/package=psych. crisis in Switzerland. PLoS One 2020; 15(7): e0236337. 13 Hogg MA. Subjective uncertainty reduction through self-categorization: a 7 Hawkley LC, Cacioppo JT. Loneliness matters: a theoretical and motivational theory of social identity processes. Eur Rev Soc Psychol 2000; empirical review of consequences and mechanisms. Ann Behav Med 2010; 11(1): 223–55. 40: 218–27. 14 Cacioppo S, Grippo AJ, London S, Goossens L, Cacioppo JT. Loneliness: clinical 8 Peplau LA, Perlman D. Loneliness: A Sourcebook of Current Theory, Research, import and interventions. Perspect Psychol Sci 2015; 10(2): 238–49. and Therapy. Wiley, 1982. 9 Cattaneo G, Bartrés-Faz D, Morris TP, Sánchez JS, Macià D, Tarrero C, et al. The Barcelona brain health initiative: a cohort study to define and promote determi- nants of brain health. Front Aging Neurosci 2018; 10: 321. 4