Table of Contents
The study CORONA-MONITORING lokal (CoMoLo) and its follow-up
The CoMoLo baseline study was a population-based study to investigate the seroprevalence of SARS-CoV-2 antibodies in four selected municipalities in Germany that were particularly affected by the COVID-19 pandemic in the year 2020 (i.e. with a reported cumulative SARS-CoV-2 incidence of more than 500 cases per 100,000 inhabitants before the start of the study). At baseline, random samples of adults (aged ≥ 18 years, n = 8999) from the local population registration offices were asked to visit a temporary study center to collect a blood sample and an oropharyngeal swab and to complete a short written questionnaire. About one to two weeks later, a detailed web-based or telephone-based questionnaire was completed [14, 15]. Study periods were May/June 2020 for Kupferzell (response proportion 63%), June/July 2020 for Bad Feilnbach (response proportion 59%), September 2020 for Straubing (response proportion 30%) and November/December 2020 for Berlin-Mitte (response proportion 29%) [16, 17]. Further details of the study have been described previously [14,15,16,17].
In 2021/2022, a follow-up study was conducted with participants who had given consent for re-contact (n = 8372) to investigate, among other aspects, (a) long-term health-consequences in association with infection status for all participants and (b) seroprevalence in comparison to baseline for participants in one municipality (Straubing). To achieve these aims, participants from all municipalities were asked between October 2021 and January 2022 to complete a detailed web-based or telephone-based questionnaire on health-related topics (response proportion among those eligible for follow-up-invitation 65%). All participants of Straubing were also invited for on-site follow-up blood sampling.
Study population
For the present data analysis on long-term health consequences of SARS-CoV-2 infections, the study population consisted of baseline study participants from the four municipalities who completed the detailed follow-up questionnaire on health-related topics in 2021/2022 (n = 5472). The flow chart (Fig. 1) reflects the numbers of participants who were stepwise excluded according to defined criteria (i.e. missing information to define infection status at baseline and follow-up (n = 391) as well as first-time SARS-CoV-2 infection during follow-up (n = 264)) and the final number of participants included in the present analysis (n = 4817). In addition, the numbers of the study population differentiated by infection status are given.
Flow chart with stepwise exclusion criteria applied to define the study population
For one municipality (Straubing), blood samples were taken from all participants at follow-up. Thus, in addition to the information from the follow-up questionnaire, blood sample results could be considered for an expanded definition of infection status at follow-up for a sensitivity analysis. The final number of participants included in the sensitivity analysis (n = 1298) is also shown in the flow chart.
Exposure assessment of SARS-CoV-2 infection status
For all four municipalities, the same criteria were applied to define infection status. The applied criteria were dependent by the availability of PCR tests and vaccines. At baseline in the year 2020, PCR tests were not widely feasible and vaccines against SARS-CoV-2 were not yet available. At the time of follow-up in the 2021/2022 both were widely available.
Participants with SARS-CoV-2 infection at baseline were defined based on (a) detectable IgG antibodies against the virus’ S1 antigen from venous blood specimens (Anti-SARS-CoV-2-QuantiVac-ELISA (IgG), Euroimmun, Lübeck, Germany) (n = 289) or a positive PCR test result based on an oropharyngeal swab (SARS-CoV-2 RT-PCR testing targeting the E gene and the orf1ab region of SARS-CoV-2) (n = 6) taken at the study center [14, 15] or (b) a self-reported history of laboratory-based physician diagnosis of COVID-19 (n = 29) in the short written questionnaire completed at the study center or (c) a self-reported positive laboratory test (n = 15) or a physician diagnosis of COVID-19 (n = 11) in the detailed questionnaire one to two weeks later.
Participants without a SARS-CoV-2 infection were defined as participants (a) without an infection at baseline (as described above) and (b) without a first-time infection during follow-up. First-time infection during follow-up was defined based on a self-reported history of a positive PCR test in the detailed follow-up questionnaire.
For one municipality (Straubing), venous blood samples were taken from all participants at follow-up, so that in a sensitivity analysis of the present study, detectable IgG antibodies against the virus’ N-antigen (Elecsys Anti-SARS-CoV-2, Roche, Basel, Switzerland) were additionally considered for defining a first-time SARS-CoV-2 infection during follow-up.
Outcome assessment of health consequences at follow-up
All health consequences at follow-up were assessed in the detailed follow-up questionnaire. Recurrent or persistent health complaints since baseline participation at the study center were assessed based on a list of 18 complaints (in the order as listed in Table 2). The total number of health complaints was categorized into 0, 1–2, 3–4 and ≥ 5. In case of indicating a complaint, participants were further asked whether the complaint still exists until today. Incident physician-diagnosed diseases since baseline participation at the study center were assessed based on a list of eight disease groups, i.e. comprising cardiovascular disease (e.g. myocarditis, heart failure/cardiac insufficiency, heart attack, stroke, venous thrombosis), lung disease (e.g. pneumonia, pulmonary embolism, pulmonary fibrosis), gastrointestinal disease, liver disease, kidney disease, neurological disease (e.g. Parkinson's disease, neuropathies), mental illness (e.g. depression, anxiety disorder), and metabolic disease (e.g. diabetes mellitus). Multiple answers regarding recurrent or persistent health complaints and incident diseases were possible.
Current subjective health was measured by the question “What is your state of health in general?” (response options: very good, good, fair, bad, very bad). Subjective memory impairment was defined as self-reported memory worsening with associated worries [18]. Different aspects of current health-related quality of life (i. e. physical, mental, and social aspects) were assessed by the German-language version of the Patient-Reported Outcomes Measurement Information System (PROMIS)-29 Profile v2.1 [19]. The PROMIS-29 Profile v2.1 measures seven domains: physical function, fatigue, sleep disturbance, pain interference (as domains of physical well-being), anxiety, depression (as domains of mental health), and ability to participate in social roles and activities (as domain of social well-being) with each including four items (each scored on a 5-point scale) and, additionally, a single item on pain intensity (scored on a 10-point scale). Higher scores reflect more of the domain’s concept being measured, i. e. better physical function and ability to participate in social roles and activities, but worse anxiety, depression, fatigue, sleep disturbance, pain interference and pain intensity [20]. Raw domain scores were obtained and converted into standardized t-scores with response pattern scoring following the PROMIS scoring manual [21].
Assessment of baseline characteristics (covariables)
Age, sex, school qualification and smoking status were assessed in the short-written baseline questionnaire. All other information was self-reported in the detailed baseline questionnaire. Educational status (lower, medium, high) was defined by information on school qualification (4 categories) and on vocational qualification (14 categories) using the 2011 version of the International Standard Classification of Education (ISCED) [22]. Smoking status (current, ex-, never smoking) was assessed by the information of daily, occasionally, no more, or never smoking. Body mass index (BMI) was calculated from self-reported body height and body weight (without clothes and shoes) and a BMI ≥ 30 kg/m2 was considered as obesity (yes/ no). Presence of chronic illness, i.e. of any chronic disease or long-term health problem lasting for at least six months (yes/no), and presence of any severely or moderately health-related limitation in usual everyday activities lasting for at least six months (yes/no) were recorded, which both are established central indicators of the general health status [23]. Presence of depression or anxiety disorders in the past two weeks was defined as an indicator of mental health by the Patients Health Questionnaire-4 (PHQ-4), which consists of a 2-item depression scale (PHQ-2) and a 2-item anxiety scale (GAD-2), and was assumed in case of a sum score of at least six points (possible score range: 0–12 points) [24].
Statistical analysis
All statistical analyses were performed using Stata (version 17.0, StataCorp, College Station, TX, USA). Proportions and their 95% confidence intervals (95% CI) for categorical variables as well as means and their 95% CI for continuous variables, along with numbers of observations, were reported for description of possible health consequences of SARS-CoV-2 infection at follow-up. For logistic and linear regression analyses, suspected potential confounding variables for the association between SARS-CoV-2 infection and long-term health consequences were included as covariables. The basic models were adjusted for the baseline information on sociodemographic characteristics (age, sex, municipality, and educational level) and the fully-adjusted models were additionally adjusted for baseline information on health-related characteristics (obesity, smoking status, chronic disease/health problem, health-related activity limitations, depressive/anxiety disorders) and follow-up time (days). As results between basic and fully-adjusted models differed only marginally, we present only results from the fully-adjusted models. Since the impact of a risk factor depends on both its frequency in the population and the strength of its association with the health outcome of interest, we also examined population attributable risks (PAR). PAR were derived by an equation postulated by Miettinen, taking the relative frequency of infections among participants with the respective health consequence and the respective odds ratio from the fully-adjusted model into consideration [25].
Proportions of missing information in baseline variables ranged from 0.04% for educational level, 0.4% for chronic disease/health problem and health-related activity limitation, 1.4% for depressive/anxiety disorders to 1.9% for smoking status. Persons with missing information were excluded from the descriptive analysis of baseline characteristics (as described in footnote of Table 1). To handle missing information for regression analyses, multiple imputation of missing baseline information using multiple regression analyses with the fully conditional specification (chained equations) approach was performed, assuming missing at random [26]. A total of 15 imputed datasets were created. Imputed data sets were analyzed combined to obtain odds ratios (OR) and 95% CI from logistic regression analysis or β coefficients and 95% CI from linear regression analysis, respectively, taking the within-variation and between-variation of imputed data sets into account.
