Persistent COVID-19 symptoms in community-living older adults from the Canadian Longitudinal Study on Aging (CLSA) | Communications Medicine – Nature.com

This is the one of the first studies reporting COVID-19 symptom persistence launched in an existing population-based cohort of older adults, allowing us to estimate not only the background symptom cumulative incidence in the cohort, but also to explore a broader range of pre-pandemic risk factors for post-acute COVID-19 symptom persistence. Whereas other studies have not differentiated between mild and moderate/severe symptoms, we consistently found higher symptom cumulative incidence and severity in the COVID-19 group beyond the background level observed in community-living middle-aged and older adults during the same time frame. In our study of primarily non-hospitalized patients with COVID-19, over two-thirds reported at least one symptom persisting >1 month and over half reported at least one symptom persisting >3 months. This extends previous work and further substantiates long-COVID as an important syndrome in both hospitalized and non-hospitalized individuals.

We demonstrated a higher cumulative incidence and severity of symptoms in participants with COVID-19 compared to those without. Lund et al.²⁰ was the only other population-based study examining symptom as the time of COVID-19 to include a comparison group. They used information from Danish prescription, patient, and health insurance registries and propensity-score weighting to compare PCR-test positive and negative individuals with respect to hospital-based diagnoses and prescriptions, as well as general practitioner and outpatient clinical visits in 2 weeks to 6 months after diagnosis. While they did not find differences in overall hospital-based diagnoses or prescriptions between those who tested positive and negative, they did find increased prior event adjusted rate ratios for general practitioner and outpatient clinic visits, suggesting that persistent symptoms may lead to increased healthcare utilization but not to the initiation of hospital-based acute treatment.

Most symptom persistence studies to date have been conducted in hospitalized patients. While hospital-based cohorts include the most severe cases of COVID-19, our study provides important evidence for similar levels of persistent symptoms in a population-based cohort where the vast majority of participants with COVID-19, like the general population, did not require hospitalization. In recent systematic reviews, the median prevalence of ≥1 post-acute symptom has been reported as 70.0% (IQR, 46.3–78.9%) by Nalbandian et al.⁶ and 72.5% (IQR, 55.0–80.0%) by Nasserie et al.⁵. This is in line with our estimate of 67.8% having at least one symptom persisting >1 month and higher than 53.4% persisting >3 months in community-living individuals with COVID-19. The largest multicenter hospital-based study, PHOSP-COVID, found that only 28.8% of patient were fully recovered at a median of 5.9 months post-discharge²⁸. In our study, only 5.8% of individuals with COVID-19 reported any hospitalization. There are fewer studies of non-hospitalized individuals with COVID-19. Hernandez-Romieu et al.¹⁹ reported 68% of participants had a new diagnosis (i.e., a new ICD-10 code) within 1–6 months, Jacobson et al.¹³ reported 66.9% of participants had at least one symptom at a median of 4 months, and Gaber reported 45% still had symptoms 3-4 months post COVID-19 diagnosis. While the estimated cumulative incidence of persistent symptoms is slightly lower in primarily non-hospitalized groups, it is nonetheless substantial. The largest community-based study examining 29 persistent symptoms including participants from the REACT-2 study²⁹ found 37.7% of participants reported symptoms at 12 weeks. Similar to studies in hospitalized and non-hospitalized individuals with COVID-19^{5,6,12,13,14,19,21,28,29}, we found the most commonly reported persistent symptoms were fatigue, shortness of breath, dry cough, and muscle and joint pain; however, Matta²¹ and Whitaker²⁹ also reported sleep problems, which we did not capture. We also found that over a third of individuals with COVID-19 who reported moderate/severe shortness of breath, decreased sense of smell, trouble with balance, and wet cough indicated that these symptoms persisted for >3 months, consistent with NICE guidance on long COVID (symptoms lasting >12 weeks)³⁰. Importantly, many of these symptoms can be improved with rehabilitation and these data suggest there may be unmet needs for rehabilitative care among community-dwelling patients with COVID-19 who did not require hospital treatment^31,32.

Finally, we found the rate of persistent symptoms was higher in females, those with pre-pandemic multimorbidity and lower levels of subjective social status but did not differ significantly by age. There have been mixed results on the association between symptom persistence and age with some studies reporting an association with symptom persistence¹² or post-COVID-19 healthcare utilization¹⁹ and others not^13,15. Although we found differences in the pattern of mild and moderate/severe symptom cumulative incidence by age group qualitatively, we did not find an association between age group and the rate of persistent symptoms in our regression analysis. While the associations with sex^{13,14,15,19,28,29} and multimorbidity^{13,15,19,28,29,33} have been previously reported, no studies to date have examined pre-pandemic subjective social status. As many studies have used administrative data or patient records, this type of measure may not be readily available. Until recently no other studies have found an association between an indicator of socioeconomic status and symptom persistence^15,28,29,34 Interestingly, the two largest studies including community samples in the UK had conflicting results. Thompson et al.³⁴ conducted a meta-analysis of 10 longitudinal studies and electronic health records. The authors found no significant relationship between a postal-code based index of multiple deprivation in the longitudinal studies but found those in the least deprived areas had increased odds of long-COVID compared to those in the most deprived areas. In contrast, Whitaker et al.²⁹ found an increased risk of persistent symptoms (>12 weeks) for those living in areas of higher deprivation compared to those living in areas of lower deprivation. Our results using individual-level subjective social status as a predictor of symptom persistence at 3 months support the results of Whitaker et al. Subjective social status has been shown to explain variance beyond more objective measures of SES such as income and education³⁵. Studies have also reported an association between race and symptom persistence^13,36 and in post-COVID-19 healthcare utilization¹⁹. The National Institutes of Health has identified the short- and long-term effects of COVID-19 on health and how to reduce differential outcomes among racial and ethnic groups as a research priority³⁷. Our finding that lower subjective social status (reflecting the relative perception that individuals have of their place in the social hierarchy) is associated with COVID-19 symptom persistence may indicate that other social factors should be considered in addition to race in future studies.

This study has many strengths including its design nested within an existing population-based nationally generalizable cohort²³ with pre-morbid data on many aspects of health. This study also has some limitations within which to interpret the results. While the CLSA cohort has been shown to be generalizable to the target population in Canada on many factors²³, the response rate of the COVID-19 study was 67.2% which may lead to participation bias. This is not unusual compared to other studies in non-hospitalized patients that were not based on administrative data, where most response rates were less than 50%⁵. One advantage of using the CLSA data is that we can understand how our COVID-19 questionnaire study population compares to the full CLSA and the target population in Canada, which is not possible for most of the non-hospital-based and hospital-based cohorts published to date. Nonetheless, while our sample size is large, the actual number of participants with COVID-19 was relatively small. While the number of COVID-19 positive participants aligns with Canadian prevalence statistics in community-living older adults for the period in which the Exit questionnaire was administered³⁸, the small sample size is reflected in the width of our 95% CIs. Furthermore, by excluding participants reporting they “very likely” had COVID-19 but had no positive COVID-19 test or physician diagnosis, we may have left out some participants who in fact had COVID-19. However, our results were robust to a number of sensitivity analyses. We also do not have the exact date that participants experienced COVID-19. It is possible that some participants had recovered from COVID-19 less than 1 month prior to the questionnaire administration, and thus we may be underestimating the cumulative incidence of symptom persistence. Finally, we cannot comment on symptom persistence longer than 3 months.

One additional potential limitation is that, despite that our COVID-19 prevalence estimates are in-line with other PCR-test-based Canadian data sources at the time, our definition of COVID-19 is based on self-report. Self-reported data are common in population-based community cohorts and while they may be more susceptible to recall bias, a positive COVID-19 test or physician diagnosis was such a salient event in the first year of the pandemic that recall bias is likely be minimal. Furthermore, measuring COVID-19 antibody status retrospectively also has limitations. Matta et al. examined the association between antibody-based results and people’s self-reported belief that they had COVID-19 with persistent symptoms in the CONSTNANCES cohort. Because that author found significantly larger associations with persistent symptoms with belief compared to antibody results they concluded that “physical symptoms persisting 10-12 months after the COVID-19 pandemic first wave may be associated more with the belief in having experienced COVID-19 infection than with actually being infected with the SARS-CoV-2 virus”. This may be problematic however because their conclusions were based on measuring antibody status using dry blood spots collected up to 8 months post COVID-19 infection. The authors did not consider the issue of false positives or the issue of decaying antibodies with time since infection which could decrease the presumed specificity of their assay³⁹. Even with a reported specificity of 97.5% the number of false positives can be substantial with a low cumulative incidence condition in a large population-based study. While the authors note in the discussion that the negative predictive value of the test in quite high (99.4%) they do not take into account that the positive predictive value of the assay is only 59.2%. This is not to say that population-based seroprevalence studies are not important or reliable, but rather that the totality of evidence needs to be considered to move our understanding forward. With the new omicron variant, many countries have substantially reduced PCR testing. Self-reported COVID-19 status and symptoms may become increasingly salient. Evidence from large well-established cohorts will allow us to examine factors related to health and aging pre- during- and eventually post-pandemic.

Clinical and public health implications

All follow-up studies of hospitalized COVID-19 patients post acute-care that incorporated assessments of health-related quality of life and functional capacity measures have universally reported substantial deficits in these domains⁶. However, because most COVID-19 infected individuals are managed in the community⁴⁰, it is of major public health importance to better understand the longer-term consequences of COVID-19 in the general population. With millions of individuals experiencing COVID-19 illness, persistent symptoms are a burden on individuals and their families as well as on outpatient care and public health. We found that a non-trivial proportion of people living in the community who may not have been hospitalized for COVID-19 still experience symptoms 1 month and even 3 months post infection. Many of the symptoms, for example, shortness of breath, fatigue, and pain, are amenable to rehabilitation. These data support the WHO recommendation that rehabilitation has an important role to play in promoting recovery after COVID-19³¹.