SARS-CoV-2, the virus that causes COVID-19, first emerged in the UK in January 2020.Reference Ball, Wace, Smyth and Brown1 The UK government closed schools on 20 March 2020, and imposed a nationwide lockdown on 23 March 2020.Reference Lyons2 This lockdown immediately generated COVID-19-related stressful events, including stressors that were social (restricted face-to-face interactions, including with intimate and familial relations), occupational (unemployment, furlough, reduced hours, working from home) and educational (cancelled courses, changed schedules, remote studying). In addition, infection outbreaks led to a number of individuals being placed in quarantine. A recent Chinese study showed that quarantine restricted every aspect of life and increased mental burden.Reference Ben-Ezra, Sun, Hou and Goodwin3 Although isolation restricts freedom of movement, other restrictions in COVID-19-related stressful events may stem from different sources. According to the conservation of resources theoryReference Hobfoll4 and related frameworks,Reference Hou, Lai, Ben-Ezra and Goodwin5 these major disruptions in daily lives can provoke psychological distress. During the COVID-19 pandemic, individuals may experience loss of those resources normally used to effectively cope with, and manage stressful situations. We suggest that this threat is cumulative across COVID-19-related stressful events, with the overwhelming effect a psychological version of a ‘cytokine storm’, an immune system failure with serious psychological implications.
We examine associations among COVID-19-related stressful events, underlying health conditions associated with increased mortality from COVID-19, being in isolation and serious mental illness. In addition, we hypothesise that a cumulative loss of resources will be associated with elevated risk of serious mental illness, with the larger the number of events generated by COVID-19, the greater the mental toll. We collect national cross-sectional data at two time points: the first survey during the peak of the pandemic (average new daily cases 3206, with 2078 new deaths), and the second data wave as the pandemic plateaued (average new daily cases 1042, with 137 new deaths; Fig. 1). Understanding the differential effects of disruption caused by COVID-19 during two very different periods should provide clinicians with a better understanding of its negative effect on mental health outcomes over the life course of the pandemic.
Method
All procedures performed in studies involving human participants were in accordance with the ethical standards of the Ethics Committee of Ariel University (approval number AU-SOC-MBE-20200225), and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Each participant signed an electronic informed consent form.
Study 1
Sampling
We conducted a sample of the UK population by using an internet panel (N = 1293), with a random and stratified sampling. Participants were recruited between 30 March and 2 April 2020 (i.e. 1 week after the UK national lockdown). The mean age of the participants was 51.51 years (s.d. 14.75, range 18–75), 53.3% were female (n = 689), 69.0% reported being in a committed relationship (n = 892) and 27.2% (n = 352) reported having a background medical condition (hypertension, diabetes, cardiovascular disease, chronic respiratory disease, chronic obstructive pulmonary disease and cancer). The response rate was 37%. See Table 1 for more information.
a Original algorithm as proposed by Shevlin et al.Reference Shevlin, Hyland, Ben-Ezra, Karatzias, Cloitre and Vallières7
Measurements
Being in isolation was measured by the item: ‘Are you currently in quarantine because of the coronavirus?’ (0 for no, 1 for yes).
COVID-19-related stressful events were measured by three questions: ‘In the past month, has the COVID-19 pandemic affected your relationships or social life?’, ‘In the past month, has the COVID-19 pandemic affected your work or ability to work?’ and ‘In the past month, has the COVID-19 pandemic affected your school or college work?’. Each question was rated on a scale with response categories of 0 for not at all, 1 for slightly, 2 for moderately, 3 for very much and 4 for extremely. Each question was aggregated based on the two highest categories versus the rest (low disruption of COVID-19-related stressful events for categories 0–2 versus high disruption in COVID-19-related stressful events for categories 3–4). In addition, we created a severity index of 0–3 to indicate the number of COVID-19-related stressful events.
Outcome variable
Screening for serious mental illness was measured using the six-item K6 scale,Reference Kessler, Barker, Colpe, Epstein, Gfroerer and Hiripi6 with direct reference to COVID-19 pandemic in the instructions. Items included: ‘During the past 30 days, how often did you feel: 1. nervous; 2. hopeless; 3. restless or fidgety; 4. so depressed that nothing could cheer you up; 5. everything was an effort; 6. worthless?’. Each category was rated as one of following: 0, none of the time; 1, a little of the time; 2, some of the time; 3, most of the time; or 4, all of the time. Scores ranged from 0 to 24, with ≥13 indicating elevated risk of serious mental illness.Reference Kessler, Barker, Colpe, Epstein, Gfroerer and Hiripi6 Cronbach α was satisfactory (0.90).
Statistical analyses
A multivariate logistic regression with elevated risk of serious mental illness (K6 score ≥ 13) as the outcome measure entered the following variables: demographics (age, gender, marital status, background illness, isolation) and three COVID-19-related stressful events. A second logistic regression also included an index for cumulative COVID-19-related stressful events (ranging from zero to three) instead of each single event, as in the former logistic regression. Each category in the index was compared with the reference group (no COVID-19-related stressful events). For each variable, we calculated the odds ratio and 95% confidence interval, using SPSS version 25 for Windows (IBM).
Study 2
The above study included only a limited number of COVID-19-related stressful events, and so did not cover a number of major aspects of life. In our second study, we aimed to replicate the findings of study 1, using a different national sample and an expanded list of stressful events. We collected data as the pandemic's first wave in the UK plateaued and the four constituent countries moved out of lockdown.
In this second study, we considered an additional aspect of the mental health toll of the COVID-19 pandemic: ICD-11 adjustment disorder. Contrary to serious mental illness, a broad concept based on multiple constructs mainly drawing on depression and anxiety,Reference Kessler, Barker, Colpe, Epstein, Gfroerer and Hiripi6 the ICD-11 adjustment disorder is specifically associated with stress, providing a distinct mental disorder (adjustment disorder: code 6B43 in ICD-11) alongside a specific narrative description to provide and guide specific diagnosis.Reference Shevlin, Hyland, Ben-Ezra, Karatzias, Cloitre and Vallières7
Based on study 1, we hypothesised that COVID-19-related stressful events will be associated with elevated risk of serious mental illness and probable ICD-11 adjustment disorder. Based on a dose–response model,Reference Shevlin, Hyland, Ben-Ezra, Karatzias, Cloitre and Vallières7 we predicted that the more stressful events a person experiences, the higher the likelihood that they will exhibit elevated risk of serious mental illness and ICD-11 adjustment disorder.
Sampling
We conducted a sample of the UK population, using an internet panel (N = 1073), with random stratified sampling. Participants were recruited between 17 and 23 June 2020. The mean age of the participants was 54.36 years (s.d. 12.14, range 20–75), 55.1% were female (n = 591), 70.5% reported being in a committed relationship (n = 757) and 28.6% (n = 307) reported having a background medical condition (hypertension, diabetes, cardiovascular disease, chronic respiratory disease, chronic obstructive pulmonary disease and cancer). The response rate was 42%. See Table 1 for more information.
Measurements
We used similar measures as in study 1, with the following exceptions. COVID-19-related stressful events were measured by the International Adjustment Disorder Questionnaire (IADQ).Reference Shevlin, Hyland, Ben-Ezra, Karatzias, Cloitre and Vallières7 The IADQ comprises two parts. First is a COVID-19-modified checklist of stressful events (stressors list) with eight events: financial problems (e.g. difficulty paying bills, being in debt); work problems (e.g. unemployment, redundancy, retirement, problems/conflicts with colleagues, change of job role); educational problems (e.g. difficulty with course work, deadline pressure); housing problems (e.g. stressful home move, difficulty finding a secure residence, lack of secure residence); relationship problems (e.g. break-up, separation or divorce, conflict with family or friends, intimacy problems); personal health problems (e.g. illness onset or deterioration, medication issues, injury or disability); a loved one's health problems (e.g. illness onset or deterioration, medication issues, injury or disability) and caregiving problems (e.g. emotional stress, time demands). Participants checked ‘yes’ for each event applicable to them, creating a cumulative score categorised as follows: 0, zero events; 1, one event; 2, two events; 3, three events and 4, four or more events (the number of participants reporting more than four events was 67 (6.1%)). The second IADQ component assesses adjustment disorder core symptoms, duration and functional impairment. The first six questions measure two symptoms clusters (‘preoccupation’ and ‘failure to adapt’). Three subsequent questions examine functional impairment in life domains (social, occupational (educational) and other domains of life). These nine questions were rated as follows: 0, not at all; 1, a little bit; 2, moderately; 3, quite a bit and 4, extremely. The tenth question assesses duration of symptoms (coded as 0 for no and 1 for yes).
The algorithm for a probable diagnosis of ICD-11 adjustment disorder requires the presence of a psychosocial stressor (score ≥1 on the IADQ stressor list), at least one preoccupation symptom rated ≥2, at least one failure-to-adapt symptom rated ≥2, and evidence of functional impairment rated ≥2.Reference Shevlin, Hyland, Ben-Ezra, Karatzias, Cloitre and Vallières7 Cronbach α was satisfactory for IADQ (0.97).
Outcome variables
To limit the tautological risk of using the stressor list as both an independent and dependent variable, we computed a second, modified algorithm, which excludes the IADQ stressor list.
To use the second part of the IADQ (adjustment disorder scale) as an outcome measure, we viewed the COVID-19 pandemic as a global stressor, thus satisfying the criteria for exposure to a stressful event. We used a slightly modified algorithm to prevent a tautology in this study, utilising the stressor list as part as the dependent and independent variables (measuring exposure to COVID-19-related stressful events as the independent variable and as part of the dependent variable). The modified algorithm used in the study is identical to algorithm presented above excluding the IADQ stressor list.
The results section includes the prevalence of probable ICD-11 adjustment disorder among the UK population according to the IADQ original algorithm, and the prevalence of ICD-11 adjustment disorder when using the modified algorithm.
Serious mental illness
Screening for serious mental illnessReference Kessler, Barker, Colpe, Epstein, Gfroerer and Hiripi6 was measured in the same way as in study 1. Cronbach α was satisfactory for K6 (0.92).
Statistical analysis
A multivariate logistic regression used elevated risk of serious mental illness (K6 score ≥ 13) as the outcome measure, with the following variables entering the equation: demographics (age, gender, marital status, background illness, isolation) and exposure to COVID-19-related stressful event (with all events entered simultaneously). The second logistic regression was the same, except that we used an index for cumulative COVID-19-related stressful events (ranging from zero to eight) instead of the eight single events. Each category in the index was compared with the reference group (zero COVID-19-related stressful events). The same method was used, with probable ICD-11 adjustment disorder as the outcome variable (based on the modified algorithm). For each variable, we calculated the odds ratio and 95% confidence interval, and conducted a comparison of proportions, comparing the percentage of those with elevated risk of serious mental illness in each sample (independent sample comparison). Finally, we compared the models associated with elevated psychological distress with the modified algorithm for ICD-11 probable adjustment disorder, using a chi-square test.
To test the dose–response model for stress, we compared the association of each stressful event and the total events score with elevated risk of serious mental illness and modified algorithm for ICD-11 probable adjustment disorder, respectively, using Steiger's z-test.Reference Steiger8 Data were analysed with SPSS version 25 for Windows (IBM).
Results
For study 1, elevated risk of serious mental illness was found in 16.6% of the sample (n = 215). Elevated risk of serious mental illness was associated with disruption in social (odds ratio 2.59; 95% CI 1.81–3.71; P < 0.001) and occupational domains (odds ratio 1.53; 95% CI 1.06–2.21; P = 0.022). (Table 1). Elevated risk of serious mental illness was associated with cumulative COVID-19-related stressful events: odds ratio of 1.65 for one COVID-19-related stressful event (95% CI 1.03–2.64; P = 0.037), 2.43 for two COVID-19-related stressful events (95% CI 1.53–3.87; P < 0.001) and 5.56 for three COVID-19-related stressful events (95% CI 3.60–8.60; P < 0.001) (Table 2).
The results of study 2 showed that elevated risk of serious mental illness was found in 14.3% of the sample (n = 153). The proportion comparison between the two samples indicated that elevated risk of serious mental illness was not significant (difference of 2.3%; 95% CI –0.64% to 5.20%; χ 2 = 2.360; d.f. 1; P = 0.125).
Probable ICD-11 adjustment disorder according to the IADQ algorithm was found among 15.9% of the sample (n = 171), and probable ICD-11 adjustment disorder according to the IADQ modified algorithm was found among 17.8% of the sample (n = 191).
Elevated risk of serious mental illness was significantly associated with housing problems (odds ratio 2.71; 95% CI 1.39–5.30; P = 0.003), the participant's own health problems (odds ratio 3.43; 95% CI 2.20–5.30; P < 0.001) and caregiving problems (odds ratio 1.83; 95% CI 1.14–2.94; P = 0.013). Elevated risk of serious mental illness was significantly associated with cumulative COVID-19-related stressful events compared with those participants reporting no COVID-19-related stressful events. These results were significant for those who had experienced at least one COVID-19-related stressful event (odds ratio 2.19; 95% CI 1.15–4.15; P = 0.017), with an increased likelihood of elevated risk of serious mental illness for those reporting experiencing four or more COVID-19-related stressful events (odds ratio 12.41; 95% CI 7.05–21.84; P < 0.001). In addition, being younger was associated with elevated risk of serious mental illness for both measures. For more information, see Table 3. These results echoed the results of study 1.
A similar pattern emerged with regard to probable ICD-11 adjustment disorder according to the modified algorithm. Probable ICD-11 adjustment disorder was significantly associated with the participant's own health problems (odds ratio 2.07; 95% CI 1.38–3.10; P < 0.001), the participant's loved one's health problems (odds ratio 1.85; 95% CI 1.22–2.80; P = 0.004) and caregiving problems (odds ratio 2.59; 95% CI 1.68–3.99; P < 0.001). Using the modified algorithm, probable ICD-11 adjustment disorder was significantly associated with cumulative COVID-19-related stressful events compared with those participants reporting no COVID-19-related stressful events. These results were significant for those who had experienced at least one COVID-19-related stressful event (odds ratio 2.45; 95% CI 1.27–4.72; P = 0.007), with an increased likelihood of probable ICD-11 adjustment disorder for those reporting experiencing four or more COVID-19-related stressful events (odds ratio 17.81; 95% CI 9.92–31.97; P < 0.001). In addition, being younger and currently being in isolation were associated with elevated serious mental illness for both measures. For more information, see Table 4.
In addition, we tested the models for association with the aforementioned (independent) variables examining the elevated risk of serious mental illness and the modified algorithm for probable ICD-11 adjustment disorder. The –2log likelihood in the first model (Table 3) associated with elevated risk of serious mental illness was 723.272. This compares with 828.809 for the –2log likelihood for the model for probable ICD-11 adjustment disorder (Table 4). The results of the chi-square test were significant at P < 0.005. Hence, there was better model fit for the elevated risk of serious mental illness.
Finally, we tested the dose–response model for elevated risk for serious mental illness. A Steiger's z-score of 1.609 indicates that the dose–response model was not significant for elevated risk of serious mental illness. However, the modified algorithm for probable ICD-11 adjustment disorder showed a significant Steiger's z-score of 2.375, indicating a dose–response association.
Discussion
The results of study 1 showed that elevated risk of serious mental illness was associated with disruption in social and occupational COVID-19-related stressful events, but not with those more educational aspects. As more COVID-19-related stressful events were generated, the risk of serious mental illness increased, suggesting the risk of overwhelming an individual's psychological capacities during times of enhanced stress.Reference Holmes, O'Connor, Perry, Tracey, Wessely and Arseneault9 These COVID-19-related stressful events have important implications for daily functioning and preserving psychological well-being. In terms of social relationships, an inability to seek help outside the home means that COVID-19 has been associated with increased domestic problems.Reference Townsend10 In the context of the UK, it is estimated that the current gross domestic product will shrink by 9% and around 7.6 million jobs will be lost. Roughly 24% of the UK labour force are at risk because of COVID-19-related lockdowns.Reference Gant11 Those with the lowest income are the most vulnerable.Reference Allas, Canal and Hunt12 This may suggest that the most vulnerable groups are prone to exhibit elevated serious mental illness when compared with less vulnerable groups.
The results pattern of study 2 were similar to those in study 1, albeit with different COVID-19-related stressful events. Elevated serious mental illness and probable ICD-11 adjustment disorder were associated with participants’ own health and caregiving problems and, to a lesser extent, housing problems and the health problems of a loved one. In addition, the model associated with elevated serious mental illness was found to be more robust than the model associated with the modified ICD-11 adjustment disorder. This may be because the K6 scale is capturing more aspects of anxiety and depression.
Cumulative stressful events take their medical toll. This has been found in the association between cumulative stress and disrupted hypothalamic-pituitary-adrenocortical axis functioning,Reference Suglia, Staudenmayer, Cohen, Enlow, Rich-Edwards and Wright13 lower oxytocin level in pregnant womenReference Samuel, Hayton, Gold, Feeley, Carter and Zelkowitz14 and Takotsubo cardiomyopathy.Reference Rosman, Dunsiger and Salmoirago-Blotcher15 This adds to the known detrimental effect of psychological stress on health (e.g. immunoregulatory balance) and the association with inflammatory disease,Reference Marshall16 cardio vascular disease,Reference Brotman, Golden and Wittstein17 immune systemReference Gu, Tang and Yang18 and immune inflammatory reactions.Reference Iwata, Ota and Duman19 Psychological stress also takes an economic toll directly and indirectly, and increases the burden on the health system.Reference Kalia20
The results of study 2 support a dose–response model of stress and post-traumatic stress disorder;Reference Dohrenwend and Dohrenwend21 the more COVID-19-related stressful events were experienced by the participants, the more the mental toll, evidenced in an increased likelihood of probable ICD-11 adjustment disorder. The dose–response association of cumulative disruptions in COVID-19-related stressful events or cumulative COVID-19-related stressful events on probable ICD-11 adjustment disorder echoes previous findings.Reference Kaysen, Rosen, Bowman and Resick22,Reference Hamer, Kivimaki, Stamatakis and Batty23
From a clinical point of view, these disruptions and COVID-19-related stressful events are important, as they can readily lead to an excessive burden on the general population. Clinicians may expect to see more referrals based on resource loss and any ineffective coping associated with this. Another consequence of the COVID-19 pandemic is that populations generally at lower risk of mental disorder before COVID-19 may become at risk because of the disruption to their everyday lives.Reference Marsh24–Reference Hou, Liu, Liang, Ho, Kim and Seong26 Our findings suggest that clinicians should be aware of disruption in COVID-19-related stressful events as a source of serious mental illness, especially when there is risk of a ‘storm’ that overwhelms an individual's capacity to cope. Although the association between a medical cytokine storm and health outcomes from COVID-19 diseaseReference Jose and Manuel27,Reference Zhou, Yu, Du, Fan, Liu and Liu28 has recently been questioned,Reference Sinha, Matthay and Calfee29 COVID-19 is a global stressful event, with salient psychosocial consequence.
General limitations and conclusions
We recognise several study limitations. Our study was cross-sectional and, although we recruited respondents from a wide range of ages, our responses were self-reported. In addition, we used two unrelated samples recruited at different time points. This is a weakness when compared with a longitudinal study. We also report relatively low response rates (37% and 42%, respectively), although these response rates are similar to other large internet surveys.Reference Ben-Ezra, Hyland, Karatzias, Maercker, Hamama-Raz and Lavenda30 We had no information on past psychological conditions. We did not consider additional psychological consequences of anxiety, such as stereotyping and prejudice, reported during the SARSReference Washer31 and COVID-19 pandemics.32,Reference Brooks, Webster, Smith, Woodland, Wessely and Greenberg33 Another potential bias is the possibility that people who are distressed about COVID-19 may be more likely to fill out a survey, thus inflating the strength of the relationship.
The cost of psychological stress is known to be associated with medical conditions and illnesses,Reference Cohen, Janicki-Deverts and Miller34 and increased risk for mortality.Reference Dohrenwend and Dohrenwend21 Psychiatrists should be aware that certain COVID-19-related stressful events can lead to serious psychological problems.Reference Marsh24 They need to pay particular attention to patients who report cumulative COVID-19-related stressful events, and consider them for mental health assessment and treatment.
In sum, to our knowledge, this is first study to empirically examine the association between COVID-19-related stressful events and cumulative COVID-19-related stressful events, with an elevated risk of serious mental illness at two different time points. Moreover, we report a novel association between single COVID-19-related stressful events and cumulative COVID-19-related stressful events with probable ICD-11 adjustment disorder.
Future studies should take a longitudinal approach, and use biological markers of stress to gain a better insight into the health costs of experiencing specific and cumulative COVID-19-related stressful events.
Data availability
The data that support the findings of this study are available from the corresponding author, M.B.-E., upon reasonable request..
Author contributions
M.B.-E. collected data, performed the statistical analysis and drafted the initial manuscript. M.B.-E., W.K.H. and R.G. provided the concept and design of study and were involved in the analysis. M.B.-E., W.K.H. and R.G. were involved in interpretation of data. W.K.H. and R.G. revised the manuscript for important intellectual content. All authors read and approved the final manuscript.
Funding
M.B.-E. was supported by Ariel's University Research Authority internal funding (grant RA2000000106).
Declaration of interest
None.
eLetters
No eLetters have been published for this article.