We conducted a cohort study on a group of 958 outpatients who had mostly experienced mild symptoms of COVID-19. The study used longitudinal data and followed a case-control design. The participants had received care at the post-COVID outpatient clinic of the University Hospital of Cologne (UHC) between April 6th, 2020 and August 11th 2021 and presented for convalescent plasma donation six weeks after the onset of symptoms. Subsequently, each individual was invited for longitudinal follow-up visits to monitor clinical courses and humoral immune responses following each individuals SARS-CoV-2 infection. As a result, ongoing PCS at month 4 and 7 post-infection was described at the very beginning of the pandemic after mostly mild COVID-19. A more comprehensive overview of the cohort is given in the reference (1).

Out of 958 initial patients, we included 222 (23.2%) outpatients for which follow-up data of 15 months after infection was available. Patients were seen by a clinician over a period of up to 15 months including four scheduled visits, which are defined as T1 (baseline), T2 (6 months), T3 (10 months), and T4 (15 months) in the following. In case of clinical deterioration, patients additionally presented to the UHC as part of unscheduled study visits. At each visit (total visits n=908: scheduled visits n=888, unscheduled visits n=20, Figure 1 ) patients completed a self-reported questionnaire which was evaluated by a trained physician after anamnesis, physical examination and - if needed - symptom-oriented diagnostics such as electrocardiography, pulmonary spirometry, or imaging techniques. In addition, blood samples were collected at all visits.

Cases were defined as patients who presented with long-term PCS at T3 (10 months post-infection) and T4 (15 months post-infection), whereas patients who did not exhibit PCS were considered as controls ( Figure 2 ). We estimated the i) probability of PCS in patients who received a COVID-19 vaccination, ii) monitored individual clinical courses and distinct symptom clusters, iii) compared sociodemographic factors, such as age, working status, and iv) compared SARS-CoV-2-specific antibody response in cases and controls. Analyses were stratified according to specific age groups, the vaccination and working status.

The following inclusion criteria were specified: (I) age of 18 years and older at the time of diagnosis, (II) diagnosed with SARS-CoV-2 infection using polymerase chain reaction (PCR), and (III) received care at the post-COVID outpatient clinic of the UHC. PCS was defined according to the World Health Organization (WHO) (15) by one or a number of newly emerged symptoms three months after initial SARS-CoV-2 infection such as fatigue, dyspnea, anosmia or ageusia that lasted for at least 2 months and could not be explained by an alternative diagnosis. For detailed analysis of clinical courses, individual symptoms were merged into distinct symptom clusters (SC) based on organ systems, prevalence, and concurrent time of occurrence (SC1 anosmia and ageusia, SC2 fatigue and dyspnea, SC3 sore throat and cough, SC4 headache, SC5 diarrhoea). In other words, most frequent symptoms at baseline (anosmia and ageusia; sore throat and cough) and T4 (fatigue and dyspnea) were merged into distinct clusters based on our previous work (1). In particular, anosmia and ageusia, fatigue and dyspnea and sore throat and cough were stated at the same time in the majority of patients and were thus merged to SC1, SC2 and SC3, respectively. Clusters were classified only if both symptoms were present at the respective time point.

Depending on guidelines of national authorities and the local availability of vaccines either a vector-based, or a messenger ribonucleic acid (mRNA)-based, or a heterologous regimen was administered ( Figure 2 ). Importantly vaccine regimens were not administered as part of this study but as part of each individual decision following the government vaccination program. Patients were evaluated both before and after receiving vaccination at the UHC outpatient clinic. The query did not specify whether patients were vaccinated as a prophylactic measure against re-infections or for therapeutic treatment of PCS symptoms

SARS-CoV-2-specific immunoglobulin G (IgG) titers against the receptor binding domain (RBD) of the SARS-CoV-2 spike protein were measured by the SARS-CoV-2 IgG II Quant assay provided by Abbott on the Alinity i (Abbott, Abbott Park, IL, United States). IgG titers were reported in Binding Antibody Units per milliliter (BAU/ml). According to manufacturer instructions, sample values ≥ 7.1 BAU/ml were interpreted as reactive. Samples with anti-RBD IgG titers above the upper limit of quantification of 11360 BAU/ml were not further diluted.

We performed binomial regression models on the long-term outcome PCS ( Table 1 ). The unadjusted odds ratio (OR) with 95% confidence interval (CI) for the probability to develop PCS at T3 (10 months) and T4 (15 months) were reported for defined variables as follows: (i) distinct symptom clusters (SC1-SC5) at baseline, (ii) PCS incidence, (iii) vaccination status, and (iv) SARS-CoV-2 antibody titers. Chi-square tests were performed to identify associations between the different variables and PCS. In addition, metric SARS-CoV-2-specific humoral antibody response (SARS-CoV-2 IgG in BAU/ml) was tested for statistical significance between visits. Normality was assessed by Kolmogorov–Smirnov or Shapiro–Wilk test, respectively. To test for differences in only non-parametrically distributed antibody titers, Kruskal-Wallis tests were performed as applicable. Descriptive analyses for sociodemographic and clinical data are presented as absolute numbers with percentages and median with interquartile range (IQR) as appropriate in Tables 2 and S1 . P-values of 0.05 and lower were considered as statistically significant. STATA version 17.0 and GraphPad Prism version 9.4.0 (GraphPad Software, La Jolla, CA) were used to compile the analyses and graph the data.

The study received ethical approvals from the ethics committee of the UHC reference number: 16_054 and 20-1187_1.

A total of 222 patients with PCR-confirmed SARS-CoV-2 infection who received follow-up care at the outpatient clinic of the UHC between April 6th 2020 and August 11th 2021 were included in the analyses. The majority of patients included in our study were women [61.3% (136/222)]. The median age of all patients was 52 years (IQR 40-59) ( Table 2 ). Most patients were healthy and did not have any pre-existing comorbidities (153/222; 68.9%). Of those that did, the main pre-existing conditions were hypertension (23/222; 10.4%), malignancies (12/222; 5.4%) and chronic lung disease (8/222; 3.6%). More detailed sociodemographic and clinical characteristics are listed in Table 2 . Since our study cohort exhibited disease symptoms between February and May 2020, it is most probable that the SARS-CoV-2 infections of our cohort were caused by the wild-type variant of Wuhan. No patient had received any specific therapy during his initial COVID-19. All patients had a mild course of COVID-19 in the outpatient setting (WHO disease grade I-III (16); 100.0%, 222/222, Table 2 ). Patients were followed up for a median time of 15 months (interquartile range (IQR) 15-16, ( Figure 1A ). Among all patients, 77.0% (171/222) have received at least one vaccination during the observation period between 10 and 15 months after symptom onset. 59.6% (102/171) and 40.4% (69/171) of vaccinated patients received single and double vaccination, respectively. Vaccinations were administered at a median of 60 days (IQR 31-84) prior to 15 months visit. Messenger ribonucleic acid (mRNA)-based vaccination regimens (116/171; 67.8%) were the most used ( Figure 2 ), followed by vector-based (46/171; 26.9%) and heterologous (4/171; 2.3%) regimens. In 2.9% (5/171) of patients, the type of vaccination was unknown. While baseline symptoms of vaccinated and unvaccinated patients did not differ, vaccinated patients stated to have more pre-existing conditions (34.5% versus 19.6%, Table 2 ). At 6, 10 and 15 months after symptom onset 14.9% (33/222), 15.3% (34/222) and 14.9% (33/222) of patients were on sick leave, respectively ( Tables 2 , S3 ).

The most common reported COVID-19 related symptoms at baseline ( Table 2 ) were cough (72.5; 161/222), change in taste or smell (58.6%; 130/222 and 54.5%; 121/222 respectively), headache (55.0%; 122/222) as well as muscle and/or body aches (51.8%; 115/222). A total of 31.1% (69/222) patients reported symptoms consistent with PCS after a median time of 15 (IQR 15-16) months after infection ( Figures 1C , 2 , Table 2 ). Similar rates of PCS (31.5%; 70/222 and 36.5%; 81/222) were found after a median time of 6 (IQR 6-8) or 10 months (IQR 15-16), respectively ( Figure 1C , Table 2 ). Among patients with a PCS at T4 ( Table S1 ), the most frequent symptoms ( Figure 1C ) after 15 months were fatigue (65.2%; 45/69), dyspnea (37.7%; 26/69), ageusia (34.8%; 24/69) and diarrhoea (23.2%; 16/69). Over time, the frequencies and characteristics of the symptoms changed. Based on the clinical presentation of patients, symptoms were assigned to five distinct symptom clusters (SC1-SC5) ( Figure 3 , Table 2 ). Comparison of SC1-SC5 at each visit revealed distinct clinical manifestations in early and long-term disease courses, respectively. While clusters such as anosmia and ageusia (SC1), sore throat and cough (SC3) and headache (SC4) were most common at 6 weeks after symptom onset (SC1: n=104, SC3: n=60 and SC4: n=122), they were least present at month 15 (SC1: n=1, SC3: n=0 and SC4: n=4). In contrast, the prevalence of fatigue and dyspnea (SC2) was low in the beginning (SC2: n=1) and increased over time (SC2: n=17). Only the frequency of diarrhoea (SC5) showed two peaks: at first and at last visit (T1: n=37, T4: n=16, Figure 3A ).

While SARS-CoV-2 RBD immunoglobulin G (IgG) titers gradually declined from T1 to T3, IgG levels at T4 increased in all patients after vaccination (median BAU/ml (IQR); T1: 111 (43-284), T2: 51 (21-114), T3: 40 (15-94), T4: 2308 (320-4334), Figure 1B ). Figure S1 shows SARS-CoV-2 IgG titers (BAU/ml) stratified by (A) presence/absence of (PCS⁺ versus PCS⁻), (B) sex (female versus male) and (C) age (18-59 versus 60-79), respectively. Pre-vaccination SARS-CoV-2 IgG titers were comparably distributed across visits when comparing patients with and without PCS, female, and male patients or younger (18-59 years) and older (60-79 years) patients ( Figure S1 , Table S2 ). After vaccination between T3 and T4, significantly increased titers (p <.0001 for all comparisons) were observed in all three stratified groups. (PCS⁺ p <.0001; PCS⁻ p <.0001); female p <.0001; male p <.0001; age 18-59 p <.0001; age 60-79 p <.0001, Figures S1A–C , Table S2 ). Interestingly, antibody titers (BAU/ml) were distributed differently in symptom clusters SC1-SC5, with SC2 and SC4 having the highest and SC1 and SC5 having the lowest levels of SARS-CoV-2 IgG ( Figure 3B ). However, no significant differences were observed.

Patients were stratified based on their vaccination status at last visit [vac⁺: 77.0% (171/222), vac⁻: 23,0% (51/222)], and the presence of PCS was assessed over a total period of 15 months ( Figure 4A ). The percentage of patients with PCS was overall comparable at the different time points and did not appear to be affected by vaccination (vac⁺ versus vac⁻: T4 30.4% (52/171) versus 33.3% (17/51), p= .6921 Figure 4A ). To further evaluate if vaccination can influence PCS recovery, we assessed the clinical courses and intraindividual outcomes over time of the 70 patients that were identified with PCS at T2 (6 months, Figure 4B ). Regardless of whether vaccinated or not, 21.4% (15/70) of PCS patients recovered prior to vaccination between T2 and T3 and 20.0% (11/55) of PCS patients recovered after vaccination between T3 and T4 ( Figure 4B ). In addition, intraindividual outcomes of the 81 patients with PCS at T3 were also assessed ( Figure 4C ). Here, 25.8% (16/62) of vaccinated and 26.3% (5/19) of unvaccinated PCS patients recovered until T4 (p =.9646). In addition, according to the unadjusted binomial regression model ( Table 1 ) there was no significant difference in the probability of experiencing a PCS after 15 months between those who were vaccinated and those who were not (OR 0.87; 95% CI: 0.45 1.75; p =0.6921).

Unadjusted binomial logistic regression models revealed several symptoms during acute COVID-19 that were associated with an increased risk of PCS after 10 and 15 months, respectively ( Table 1 ): Diarrhoea (10 months: OR 3.33; 95%CI 1.62 6.9, p =.0007; 15 months: OR 3.27; 95%CI 1.54 6.64, p =.0009) and headache (10 months: OR 2.55; 95%CI 1.43 4.62, p =.0013; 15 months: OR 1.85; 95%CI 1.04 3.26, p =.0390) were risk factors for PCS at T3 and T4, respectively. Moreover, as expected, presence of PCS strongly increased the risk of PCS at the following visit (10 months: OR 17.77; 95%CI 8.67 36.79, p <.0001; 15 months: OR 41.91; 95%CI 17.74 93.22, p <.0001, Table 1 ).