We conducted a cross-sectional study in Japan and Sweden from March 18 to June 15, 2021 and collected the data through an online survey. The only eligibility criterion was that the age of the participants should be at least 18 years old. Data collection was outsourced to Asmark companies in Japan and Prolific in Sweden, and data collection was carried out through each company’s online research platform. Each company asked pooled participants to participate in the study ― that is, to respond online. Participants answered the questions by accessing the websites. They were informed in the first half of the questionnaire that the survey content included information about the COVID-19 hospitalization experience and post-COVID conditions, so participants could withdraw their participation, if they wished, depending on the degree of psychological distress. A small monetary compensation was paid as a reward to the participants through the research company.
The study was planned and designed by researchers in Japan and Sweden according to the STROBE statement . The protocol for the current observational study was reviewed and approved by the Chiba University Graduate School of Medicine Ethics Review Committee (approval number 4129). The online survey was written in the native languages of Japan and Sweden, and it took about 20 min to complete.
We recruited 763 participants from Japan and Sweden. Of the participants, 135 had been infected with COVID-19 and 628 had never been infected with COVID-19. Data were collected in Japan from March 18 to 22, 2021 and in Sweden from April 5 to June 15, 2021 (e.g., a slight difference in time).
We collected the following background information about the of the participants: age, gender (woman, male and prefer not to say), race, occupation (regular, non-regular, unemployed, college student), family structure (living with family, living alone, sharing a house with someone other than the family), academic background, financial situation (household annual income of less than 4270,000 JPY (about 38,366 dollars) or more/ 380,000 SEK (about 44,408 dollars or more), having enough savings to live for about half a year if you lose your current job (yes, no), living area (up to prefecture), history of mental illness (depression, bipolar disorder, schizophrenia, anxiety, PTSD, obsessive–compulsive disorder, panic disorder, eating disorder, substance use disorder, etc.; free description), presence or absence of history of physical illness (high blood pressure, asthma, diabetes, etc.; free description).
Data on COVID-19
We collected information about the participants’ experience with COVID-19 (presence of infection, time of infection, time required for recovery, acute symptoms, physical sequelae, and vaccination). Participants were asked about symptoms related to COVID-19 and responded “yes” or “no” to the following items: heat, cough, fatigue/tiredness, dyspnea, olfaction disorder, dysgeusia, increased sputum, chest pain, joint pain, muscle pain, headache, hair loss, insomnia, anxiety, depression, and sore throat. If they experienced any symptoms related to COVID-19 other than the items mentioned above, the participants were free to mention them.
Criteria for mental disorders
Mental health was evaluated by four psychological measure scales to assess fear of COVID-19, depression, general anxiety, post-traumatic stress. The Fear of COVID-19 Scale (FCV-19S) is a self-rating scale with seven items that can quantify the fear of COVID-19. The total scores range from 7 to 35 points, and the higher the score, the stronger the fear of COVID-19 [28, 29]. Patient Health Questionnaire-9 (PHQ-9), with 9 items, is a self-rating scale to assess severity of depression. The total score of PHQ-9 shows 1 to 4 points are mild, 5 to 9 points are mild, 10 to 14 points are moderate, 15 to 19 points are moderate to severe, and 20 to 27 points are severe [30, 31]. General Anxiety Disorder-7 -item (GAD-7), with 7 items, is a self-rating scale to assess severity of general anxiety. The total score of GAD-7 shows o to 4 points are minimal anxiety, 5 to 9 points are mild, 10 to 14 points are moderate, 15 to 21 points are severe [32, 33]. Impact of Event Scale-Revised (IES-R) is a self-rating scale for measuring traumatic stress symptoms. This scale consists of 8 items of invasion symptoms, 8 items of avoidance symptoms, and 6 items of hypervigilance symptoms, for a total of 22 items [34, 35]. We used cutoffs as the criterion for the incidence of each clinical measure. The cutoff for FCV-19S was 18 points or more , 10 points or more for PHQ-9 , 10 points or more for GAD-7 , and 25 points or more for the IES-R .
Dealing with bias
People who have been recovering for some time may not be able to remember the early aftereffects of recovery following a COVID-19 infection. To address this potential recall bias, we asked about sequelae and mental health at the time of the survey. Because the data were collected via the internet, there may be a selection bias as the survey only reach those who have access to the internet and are interested in health. To address this selection bias, we conducted the survey in two countries (Japan and Sweden) in which internet usage is extremely high. Internet usage rates have been over 90% in both countries for a long time . Although there are some regulations in both countries, strict lockdown has never been implemented since the onset of the COVID-19 pandemic. Therefore, we assumed that ordinary people in both countries 2021 would be interested in a survey on COVID-19.
The assumed effect size calculated by G*Power was 0.30, two-tail. The power was set at 0.80 and the significance level at 0.05. The sample size needed to obtain sufficient power by the F test was estimated to be 37 people in each group (non-infected group, infected without post-COVID conditions, and infected with one). To collect at least 74 infected people, the final total sample size was set to 800 after taking into account the proportion of infected people and missing data.
We used SPSS Version 26 (IBM Corporation, Armonk, NY, USA) for four statistical analyses. A two-sided P value of 0.05 was considered statistically significant, and p-values for the Mann–Whitney U test were deemed statistically significant at the Bonferroni corrected p 0.05. Firstly, we performed the Kruskal–Wallis tests among the uninfected people, the infected people with post-COVID conditions, and those without sequelae. Secondly, we conducted the Bonferroni-corrected Mann–Whitney U test as Post hoc comparisons to verify which group had a significant difference. Thirdly, we conducted a Chi-squared test to compare the three groups mentioned above and verify if there was a significant difference in the proportion of people at high risk of clinically significant mental illness. Finally, we conducted logistic regression analyses to evaluate the impact of post-COVID conditions on mental health. In a logistic regression model, we assessed whether the nine variables (age, country, gender, mental illness, physical illness, days to recovery, hospitalization, post-COVID conditions, and ventilator) were associated with determinants of clinical symptom incidence. In the logistic regression model, we also calculated the adjusted odds ratio (OR) with a 95% confidence interval (CI) for the risk of clinically significant symptoms of depression, anxiety, and post-traumatic stress for the participants with post-sequelae after COVID conditions-19.