Two-year follow-up of patients with post-COVID-19 condition in Sweden: a prospective cohort study – The Lancet
Methods
Study design, setting and participants
,
,
,
Structured screening interviews, identical to those at a previous 4-month follow-up,
were performed. Reporting is in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cohort studies.
A total of 745 patients with a positive polymerase chain reaction (PCR) for SARS-CoV-2 were admitted to hospital for COVID-19 during the first pandemic wave between March 1st and May 31st, 2020, in Region Östergötland (population of approximately 450,000). This region is one of 21 Swedish healthcare regions and has three hospitals: one tertiary care university hospital with approximately 400 beds and two general hospitals with 241 and 76 beds, respectively. Thirty Intensive Care Unit (ICU) beds were available at the beginning of the pandemic, which increased to 52 during the first pandemic wave. After excluding non-COVID-19 related hospitalisations (i.e., coincidental COVID-19), in-hospital deaths, major comorbidities (e.g., dementia or terminal cancer) and dropouts, 433 individuals were screened for PCC at four months after infection. This was performed using a structured telephone interview, where 185/433 (42.7%) patients reported symptoms consistent with PCC severe enough to impair daily activities. These patients were considered eligible for the current study and thus recruitment was attempted among this sample. Interviews were performed via telephone by two of the authors (CW and GF) using a structured interview guide (described below). A third interviewer, with more experience in interpreter-mediated interviews, was used when interviews could not be performed in Swedish or English (23 cases (13.9%)).
Ethics
The Swedish Ethical Review Authority approved the study (Dnr 2020-03029, 2020-04443 and 2021-07038). In accordance with the ethics approval, the need for a written informed consent was waived given that the follow-up procedure also formed part of a clinical follow-up.
Variables and data sources
Interviews were standardized by use of this structured interview guide, instructing each of the three interviewers how to ask and respond. To further standardize how interviewers manage unusual or deviant responses all authors met for a weekly discussion. Interviewees were instructed to only consider symptoms related to COVID-19 and, when present, grade their respective impact on everyday life on a scale of 1–5 (1: no impact; 2: minor impact; 3: moderate impact; 4: high impact; 5: very high impact). General health was assessed by participants rating their current subjective health status on a five-point Likert scale, ranging from very good to very bad, similar to the first question regarding overall health of the WHO health survey questionnaire.
Dyspnea was evaluated using the modified Medical Research Council (mMRC) dyspnea scale, widely accepted for follow-up of shortness of breath post COVID-19.
,
Interview questions are available as supplementary information translated from Swedish to English. Data pertaining to comorbidities and health issues during the interim period up until the 24-month follow-up were retrieved from medical records.
The research team comprised medical specialists in infectious diseases, critical care, neurology, and rehabilitation medicine. The team met regularly and discussed individual needs of medical attention as disclosed by the interviews, and provided referrals to relevant caregivers when indicated.
These domains comprise symptoms related to vision (Domain I), sensorimotor dysfunction (Domain II), cognition (Domain III), affective symptoms (Domain IV), swallowing (Domain V), voice (Domain VI) and mental fatigue (Domain VII). Three symptoms (dizziness, hearing loss and altered smell/taste) did not fit the factor analysis, and another two (difficulty managing work/studies and experienced falls after discharge) were excluded as the first had a response rate below 50% and the second did not refer to the specific situation at the time of the interview.
for the entire cohort. According to the WHO CPS, patients with grade 4 or 5 were categorized as having moderate disease severity and cases with WHO CPS 6–9 as severe. Additionally, for ICU-treated patients, the Sequential Organ Failure Assessment (SOFA)
and Simplified Acute Physiology Score III (SAPS3)
scores were used to determine the severity of organ failures.
Statistics
Statistical analysis was performed using IBM SPSS vs. 27. Data are presented as means and standard deviations (SD) for normally distributed continuous variables; as medians and interquartile ranges (IQR) for non-normally distributed numeric variables; and as n (%) for categorical data. Comparisons over time for ordinal data were made using paired Wilcoxon signed-rank tests. Comparisons for normally distributed continuous variables were made using t-tests. Comparisons of occupational status (dichotomized) over time were made using McNemar’s tests. No imputation was performed. Normality was assessed using Shapiro–Wilks tests. A p-value <0.05 was used to denote statistical significance throughout the paper unless otherwise noted.
Role of the funding source
The study was funded by Region Östergötland. The funder had no role in data collection, analysis, interpretation, study design or writing of the report.
Discussion
In a Dutch study,
92% reported at least one residual symptom still present at 12 months after hospitalisation for COVID-19. Similar to our study, they reported significant improvements in muscle weakness and exertional dyspnea from three to 12 months, but contrastingly not in fatigue or cognitive symptoms.
Similar results to ours were also reported in a longitudinal Spanish study exploring symptom trajectories,
in which 81% of individuals had at least one symptom still present at eight months after hospital admission, with this number decreasing to 68% at 13 months. In a study from Verona, Italy,
where 51% of participants were hospitalised for COVID-19 and the rest were managed in outpatient care, 20% reported at least one symptom remaining at nine months. The lower percentage compared with the 30% observed in our study may be explained by specific differences between the cohorts. Firstly, only half of the Italian cohort was hospitalised (of which 10% were admitted to the ICU), indicating less severe disease than in our cohort. Secondly, 98% of the Italian cohort reported their pre-COVID-19 health status as very good or excellent, compared with 72% of our cohort (assessed on a similar 5-point Likert scale). Additionally, in line with the present study, a Chinese 2-year follow-up of initially hospitalised patients identified overall improvements regarding both physical and mental health but with a high symptom load remaining 2 years after infection.
,
This suggests that the self-reported prevalence of cognitive symptoms in our study might underestimate the true prevalence of such deficits. Additionally, earlier experiences from SARS and MERS infections portray long-term persistent neuropsychiatric deficits that are not necessarily correlated to initial clinical severity,
,
,
which is in many ways similar to PCC and the results of the present study.
but could also impair cognitive function.
,
Sweden had few restrictions and lock-down procedures as compared to other countries, described in detail by a commission designated by the Swedish government.
Large public gatherings of more than 50 individuals were at times prohibited. No strict lock-down was enforced. The Swedish government also issued economic support to many companies which enabled continued employment and continued work, albeit from a distance when possible. The Public Health Agency of Sweden also recommended that individuals aged 70 or older, as well as groups at high risk for severe COVID-19, take precautions. Kivi and colleagues presented that during the initial pandemic wave Swedish older adults rated their general well-being at a similar level to, or even higher than, before the pandemic.
In summary, our results may to some extent be explained as a possible consequence to pandemic-related restrictions and social distancing, as we have no control group to compare with, but we do not believe it to play a decisive part.
Furthermore, survivors from intensive care due to other diagnoses often experience long-lasting residual symptoms, such as neurocognitive, affective and pulmonary symptoms, as well as activity impairments.
Many of these symptoms are in line with those presented in our study. It is therefore challenging to determine which symptoms, if any, are specifically related to COVID-19 and which are more general post-ICU symptoms. However, as the prevalence of residual symptoms was similar in the non-ICU-treated subgroup, we suspect that the residual symptoms reported in this study cannot be solely explained as a post-ICU phenomenon.
,
and the most common causes for hospital admission in the interim period of the present study were indeed cardiovascular. This is in line with a recently published large cohort study by Wang and colleagues,
presenting a higher incidence of cardiovascular disease twelve months after initial SARS-CoV-2 infection. Hospital readmission rates after COVID-19 vary between countries. A meta-analysis by Ramzi revealed an all-cause one-year readmission rate of 10.7% in developed countries.
The presented 2-year readmission rate of 21.2% in our cohort may suggest that patients with PCC are at higher risk for hospital readmissions. However, since the readmission rate was not registered for the entire hospitalised cohort (including those without PCC) such a comparison is beyond the scope of this article.
,
,
Our results show that rates of sick leave in previously hospitalised patients with PCC may decrease but remain high two years later.
but whether vaccination post-infection ameliorates already established residual symptoms is still unclear, with some studies showing a reduction in symptoms and others no effect or even worsening of symptoms.
No effective vaccines were available before infection for the current cohort, but after infection a majority had received at least three doses at the 24-month follow-up. As such, individual vaccinations may have influenced our results in either way. However, our results suggest that vaccination after development of PCC leads to an improvement in self-rated health as compared to being unvaccinated.
several rehabilitation services for PCC are recommended, e.g., multidisciplinary rehabilitation teams, follow-up and referral systems, standardised symptom assessments, education and skills training regarding energy conservation routines. Many of these recommendations are in line with the clinical follow-up offered to our cohort at the 4-month assessment. Our results strengthen these recommendations, as there seems to be a need for medical attention in patients with PCC even two years after initial infection, with potential for significant improvement of many symptoms. However, it also indicates a need for further studies of specific interventions and their effectiveness in long-term rehabilitation of patients with PCC.
Strengths and limitations
Strengths of the present study include the long-term follow-up of a well-defined cohort as well as a high participation rate (91%), limiting information bias due to loss to follow-up. All interviews were conducted via telephone by medical professionals, which enhances data quality compared with a survey completed independently. Patients were interviewed both at four and at 24 months, thus making prognostic trajectories possible. Additionally, medical records were screened for health issues occurring in the interim period (including reinfections with SARS-CoV-2).
which is an indication of low risk of information bias. Furthermore, p-values should be interpreted with caution as potential confounders were not adjusted for, and no correction was made for multiple testing. Lastly, for some sub-analyses the sample size was low.
Conclusions
Our cohort of patients, who were hospitalised with COVID-19 during the first pandemic wave and showed symptoms indicating PCC at 4-months post-discharge, showed improved symptoms at two years post-admission, but also a high prevalence of persistent cognitive, sensorimotor and fatigue symptoms impacting on their everyday life. This implies a need to establish routines for long-term follow-up of patients previously hospitalised due to COVID-19 with PCC.
