Human social isolation and stress: a systematic review of different contexts and recommendations for future studies

Abstract Objectives The emergence of the coronavirus disease 2019 (COVID-19) pandemic and subsequent lockdowns and social distancing measures adopted worldwide raised questions about the possible health effects of human social isolation. Methods We conducted a systematic review on PubMed, Scopus, and Embase electronic databases using terms related to human social isolation – defined as the isolation of an individual from regular routines and usual social contact – and psychological stress, searching for simulated or naturalistic isolation environments. We present the main results, as well as the validity and limitations of each model. PROSPERO registry number: CRD42021241880. Results Despite the diversity of contexts reviewed, some outcomes almost ubiquitously relate to psychological stress, i.e., longer periods, expectation of a longer period, confinement, lack of social interaction, and support. Based on the results, and considering that most studies were not designed for the purpose of understanding isolation itself, we propose a group of recommendations for future experimental or naturalistic research on the topic. Conclusion Evidence on the impact of different situations in which individuals are subjected to social isolation can assist in development of directed preventive strategies to support people under similar circumstances. Such strategies might increase the general public’s compliance with social distancing as a non-pharmacological intervention for emerging infectious diseases.


Supplemental Material S2
Main results

Experimental and quasi-experimental models
The samples of the five experimental and quasi-experimental studies were composed only by males.In three of them subjects were young (under 30 years of age) and in the other two, the age was not described.Duration of isolation ranged from two 8-hour sessions to 21 days.
Stress was assessed in four of the five experimental and quasi-experimental studies included in this review using either the Standard Stress Scale or the Multidimensional Personality Questionnaire 21,24- 27 .In a study evaluating the effects of compatibility, crowding, group size and leadership seniority on several outcomes, groups composed of three men usually reported less stress than two-men groups.When in a crowded condition, however, three-men groups indicated feeling the most stressed 24 .Another study investigated the effects of privacy, expectations about length of social isolation and degree of contact with the outside world in socially isolated groups and found that the expectation of a longer isolation period generated more stress and led to higher quitting rates.Groups expecting longer isolation with more privacy and no outside contact reported the greater stress, while groups expecting shorter missions with outside stimulation and privacy were the least stressed.In this same study, more than half of the participants did not complete the protocol, 67% of which expected a longer isolation period 26 .Another study with a 1-week duration showed no statistical difference regarding stress levels when comparing groups that were confined but allowed to have social contact, confined and socially isolated and controls.Quitting rates, nonetheless, were higher in the socially isolated group 25 .
Anxiety was also investigated in four of the five articles included.When comparing the effects of crowding, group size and leadership seniority on this outcome, two-men-group subjects registered significantly higher state anxiety throughout confinement than those in three-men groups; moreover, triads led by senior men registered the lowest state anxiety 24 .In another study, confined groups reported greater incidence of worry than the control group 25 .When subjects had different expectations about the length of isolation, groups expecting shorter periods maintained a fairly stable and significantly lower level of anxiety state throughout confinement; the expectation of longer isolation under conditions of privacy produced the greatest levels of anxiety, with gradual increases over days; groups expecting longer isolation periods with privacy and without outside contact showed a significantly sharp increase of anxiety after five days of protocol, when all individuals in this condition quitted 26 .In a study in which subjects spent two sessions of eight hours confined and in social isolation, either with or without sensory deprivation (visual and auditory restriction achieved through the use of blacked-out goggles and acoustical earmuffs), anxiety was greater in the sensory deprivation condition 27 .
Several objective outcomes were assessed.A 10-day confinement period produced changes in the distribution of leukocyte subpopulations, natural killer (NK) cells and CD69-expressing lymphocytes in confined subjects 23 .Alterations in occipital alpha frequency measured by electroencephalography were observed after one week of confinement 25 .When subjects were confined and socially isolated for eight hours, breathing rate was lower in the group under sensory deprivation, but no differences were found in other physiological measurements, such as heart rate, basal skin conductance, fluctuations in skin conductance and breathing amplitude.Endocrine responses were also measured, and some hormones (17-KGS, LH, 17-KS and TSH) were altered by the sensory deprivation condition 27 , although these where punctual measurements and we are unable to infer the rhythmicity of synthesis or secretion of these hormones.
Two of the articles assessed cognitive outcomes and showed that the social isolation group had the worst performance in an intellectual battery (consisting of tests measuring numerical reasoning, number facility, abstract reasoning, verbal reasoning, verbal fluency, space relations, digit span, recall, recognition, cancellation of numbers, and dexterity) and loss of contact with reality.Subjective phenomena (i.e.experiences of a hallucinatory-like nature, inefficient thought processes, reminiscence and vivid memories and changes in body image and self-appraisal) resulted from confinement, despite http://doi.org/10.47626/2237-6089-2021-0452Human social isolation and stress: a systematic review of different contexts and recommendations for future studies -Tonon et al.
http://doi.org/10.47626/2237-6089-2021-0452social contact or lack thereof 25 .In another study, participants placed in the sensory deprivation condition during an eight-hour confinement and social isolation scored significantly higher on the questionnaire measuring loss of reality contact, hallucinations, dreams, and body-image changes 27 .
Two studies assessed affective outcomes.Men confined for 10 days in groups of five had their mood assessed through the face scale test, and the results showed that the subjects' moods were low at the beginning and then improved toward the end of the experiment 23 .When depression symptoms were evaluated, they were found to be higher in participants subjected to sensory deprivation 27 .A higher incidence of subjective restlessness and restless acts was observed in groups confined for 7 days than in non-confined ambulatory controls, regardless of social isolation 25 .
The effects of compatibility, crowding, group size and leadership seniority on hostility showed that groups led by junior leaders tended to display more hostility.Also, incompatibility between members of the group led to significantly higher hostility and greater hostility toward partners was revealed by subjects in less crowded groups than by crowded subjects 24 .

Space mission and analogues
All data collected from five space analogues and an official space flight reflect homogenous samples of mostly males under 40 years of age.These studies follow a longitudinal design, with the duration of isolation ranging from a few months to 520 days.Overall, no significant difference was reported in scores of perceived stress, anxiety, or mood disturbances.However, one controlled study observed that crewmembers presented less dysphoria, expressiveness, independence, anger and aggression compared to mission control personnel 15 .
One study reported increased subjective feelings of sickness in the assessments of days 360 and 510 compared to baseline 11 .This study also reported significantly higher salivary cortisol levels from the 360th day onwards.Another study observed significant differences on the second half of the mission (4th, 5th and 6th months) compared to the first 13 .Thus, from the 4th month of isolation on, subjects exercised less, increased the masseter tone by 6-14% and showed an increase in collateral acts (a roll of psychosomatic events that might indicate stress and anxiety behavior) assessed by Ethological Monitoring.Moreover, cortisol levels increased in the first month, but were normalized in the following months.Similarly, Basner and colleagues 14 described greater feelings of sickness, tiredness, fatigue and loss of pleasure in the second half of the mission, while vigor and activity were lowest in the second mission quarter, although no changes in sleep quality were reported.Nonetheless, Luger and colleagues 12 reported increased sleep duration, bed rest time and frequency of wake-ups from 4 weeks of isolation onwards.Moreover, compared to mission control personnel, Luger found poorer social competence and teamwork in isolated individuals in the first two weeks 12 .However, Basner reported five times less conflicts in crew members 14 .Finally, Yuan described that conflicts and pressure tended to decrease throughout confinement, while behavioral and social activities were lower in the 2nd month.

Submarine mission
Only one study conducted in a submarine setting was retrieved 17 , in which one sample was recruited for a 40-day mission and another, for a 10-day mission, while also recruiting office workers and a control group of military recruits.One of the purposes of this study was to observe distinct personality traits and coping strategies in this homogenous sample of men (only one woman was recruited).Submariners reported low frequency of health complaints in all measurements and these scores were even lower than the control group at baseline.In addition, no significant differences between scores in the first and last weeks of the mission for any of the crews were found.However, the authors recorded an increase in cortisone values of the submariners, which also reported increased homesickness and stress related to social factors and feeling isolated after the missions compared with baseline.Cortisone levels and health complaints were higher in crew members on the 10-day than in the 40-day mission.http://doi.org/10.47626/2237-6089-2021-0452Human social isolation and stress: a systematic review of different contexts and recommendations for future studies -Tonon et al. http://doi.org/10.47626/2237-6089-2021-0452

Antarctic
The five Antarctic studies are composed of similar samples of mostly males under 60 years of age.These studies follow a cohort design, with the duration of isolation ranging from two months to one year.
Stress-related outcomes were evaluated in four out of the five Antarctic studies included in this review.A longitudinal assessment of psychological adaptation during a winter-over in Antarctica showed that stress reactions decreased around the third-quarter of the stay, contradicting the "third-quarter phenomenon" -the idea that this period is when most discomfort is reported 18 .A study conducted during the summer months investigated the coping strategies used by participants to deal with the Antarctic environment and maintain emotional stability.At the end of the stay there was a significant decrease in coping strategies based on seeking social support and problem analysis and solution; coping by acceptance and attempting to get something positive out of the situation increased, though not significantly 19 .Another investigation on stress coping strategies, performed during the winter months, showed that all observed coping strategies (active problem solving, palliative reactions, avoidance and passive expectancies, and comforting cognitions) decreased during midwinter, the darkest and coldest period, which also corresponded to the third-quarter of the expedition 20 .A study about sex differences in stress responses during confinement in Antarctica showed that stress levels remained low during the whole year, not differing between sexes or according to time of data collection.When completing the Post-Traumatic Stress Scale (applied 2 months before, in the middle and 3-4 months after the expedition), participants from both sexes reported no increased feelings of anxiety when talking about the prior month, but exhibited a small, but not significant, increase in negative feelings during the past few days at all time points 21 .
Affective outcomes were also assessed in four of the five Antarctic studies.Participants spending 50 weeks in Antarctica showed stable reports of pessimistic feelings, sadness, and lack of motivation before the middle of the isolation period, which then significantly increased until the end of the stay 18 .During a summer campaign of 2-5 months subjects presented no significant differences in mood states between baseline and end of the isolation period.However, a small increase of confusion and fatigue, and a small reduction of vigor were observed 19 .Another study of winter-over stay in Antarctica showed that negative affective states (e.g.feeling afraid, upset, ashamed) were continuously low during the entire isolation period, while positive affective states (e.g.feeling excited, inspired, alert) were least reported during the mid-winter months 20 .In a study evaluating the association between attachment style and resilience in extreme environments, participants with insecure attachment style reported stronger negative mood in comparison to participants with secure attachment style throughout the entire mission.Positive mood was stable among individuals with both secure and insecure attachment styles throughout the isolation period 22 .
In regard to physical symptoms and sleep, one winter-over study showed significant decrease of somatic reactions (e.g.insomnia, appetite alterations, headaches) until the middle of the isolation period, after which the number of somatic reactions stabilized 18 .In another study, participants completed a sleep diary during the first week of every month for ten months and the results indicated a negative trend over time in subjective sleep quality, with a slight increase at the end of the stay 20 .
Anxiety was evaluated in two winter-over Antarctic studies.Strewe and colleagues reported that state and trait anxiety scores were low throughout the entire period and no differences between sexes were observed 21 .Caputo and colleagues found that prolonged isolation and confinement did not influence individual anxiety, whereas state anxiety was constant throughout the mission.Similarly, attachment style did not influence absolute anxiety nor its progression during the isolation period 22 .
The study aiming to identify sex differences in stress and immune responses during confinement also included objective measurements.Salivary cortisol concentrations in the morning were significantly higher in females than males in April, May and July, and in both sexes the cortisol ratio between morning and evening values was higher throughout the expedition than at baseline.No sex differences were observed in relation to cytokine profiles after in vitro stimulation, but significant time-dependent changes were noticed: the concentration of IFN-γ rose continuously and peaked around the middle of the expedition (but showed significant changes to baseline data collection only for males), then declined gradually to reach baseline levels at post data collection (PDC); both sexes showed significant increases http://doi.org/10.47626/2237-6089-2021-0452Human social isolation and stress: a systematic review of different contexts and recommendations for future studies -Tonon et al. http://doi.org/10.47626/2237-6089-2021-0452 in IL-10 concentrations at PDC; IL-2 concentrations peaked in June (but did not reach statistical significance in comparison to baseline values) and TNF concentrations declined during the expedition in comparison to baseline.Blood cell counts indicated that hemoglobin increased significantly during the whole isolation period compared to baseline and hematocrit values were significantly higher than baseline in the months of March, July, September and October in females.Significant differences in thrombocyte concentration in comparison to baseline were found in females in June and July and for males in March and July.Leukocytes and granulocyte levels were higher during confinement with a dip in winter whereas lymphocytes were significantly elevated throughout the entire isolation period 21 .The study evaluating the association between attachment style and genomic and physiological resilience in extreme environments showed that participants with a secure attachment style exhibited significantly lower salivary cortisol concentrations compared with participants with insecure attachment style one month after arriving at the station.Prolonged exposure to confinement and isolation induced significantly different gene expression profiles in all subjects, deregulating genes related to mitochondrial function, protein synthesis, immune response and circadian rhythms.Individuals with insecure attachment style, however, showed a higher proportion of differentially expressed genes than participants with secure attachment style 22 .
Décamps and Rosnet assessed occupational and social reactions in participants during a winterover in Antarctica and found that, while no significant variation in the number of occupational reactions (e.g.overinvestment in work, overestimation of workload, inability to complete any task) were observed throughout the stay, social reactions (e.g.withdrawal within oneself, criticism, distrust) generally increased significantly during the stay, with a significant decrease following the middle of the isolation period 18 .

Pandemics
The pandemic requires us to look for effective preventive strategies, like social distancing, which necessarily brings individuals into isolation.In the reviewed studies, the mental health variables most commonly assessed are stress, depression and anxiety, although a few articles found significant results in other psychological aspects, like boredom, subjective isolation 2 , insomnia 6 , and general worries 7 .In the pandemic's model, the duration of isolation seems to be an important predictor of psychological impact 1,2,7 , although this factor was not assessed in all studies.
The association between demographic variables and the presence of psychological symptoms varied among studies.One study 6 showed that being under 40 years old, having a history of psychiatric disorders, being male and being unmarried were associated with symptoms of depression, anxiety, insomnia, and acute stress.In another article 7 , women and people who were in a relationship but not cohabitating showed higher psychological impairment, and students and unemployed respondents showed higher levels for depression.The same article found that participants aged 18-24, followed by 25-34 showed the highest psychological impact.It also states that participants with the highest levels of psychological impact, stress, and anxiety were students, and the least affected were retired people.Another study 9 found that the depressed group was younger, less educated and more likely to be unmarried.Between the student population, one article found that women were more likely to become depressed, anxious, stressed and lonely 10 , while another found that male and younger students were more prone to depression 8 .
Being a frontline worker is associated with elevated risk for insomnia and acute stress symptoms 6 .Being a healthcare worker is also associated with higher stress level 2 .Not surprisingly, being part of the high-risk population in case of COVID-19 infection is associated with higher levels of psychological impact, including anxiety, stress and depressive symptoms 7 and is negatively associated with well-being 5 .Participants with confirmed or suspected disease, or in proximity with people infected with SARS-CoV-2 or who tested positive for COVID-19 had a tendency towards greater psychological impact 1,6,7 .Moreover, one study found that in-person contact face-to-face, in general, had a positive connection with stress 7 .