A patient-centered view of symptoms, functional impact, and priorities in post-COVID-19 syndrome: Cross-sectional results from the Québec Action Post-COVID cohort

Procedures

The project (2022-8066) was approved by the Research Ethics Board of the McGill University Health Centre. People interested in participating were directed to the QAPC website to register. The study coordinator recorded their contact information, generated a study identification number and invited them into the study. Upon invitation, they were directed to an online web-portal “Research Electronic Data Capture” (REDCap) to enter their unique identification number, allowing them entry into the data capture platform. Following an e-consent process, they recorded their health outcomes information. All were asked to consent to open data sharing for secondary analyses and to be re-contacted for additional studies.

Recruited participants also were provided a password to access online self-management resources addressing breathing difficulties, cognitive symptoms, fatigue, and increasing physical activity. To manage mental health symptoms, participants were directed to BounceBack®, a free program offered by the Canadian Mental Health Association to help manage low mood, mild to moderate depression, anxiety, stress or worry(21).

Measurement Framework

The Wilson-Cleary model guided the health outcomes assessment(22). A lightweight yet comprehensive approach used patient-centered assessment tools to characterize symptoms and their functional and quality of life impact at recruitment and over time.

As little was known at this study’s outset about the health effects beyond a symptom inventory(23), we adopted an individualized approach to outcome measurement. The Patient Generated Index (PGI)(20) asks people to nominate areas of their life affected by a health situation, here the sequelae of COVID. Each area is then rated on severity, from 0, “not at all”, to 10, “worst imaginable severity”. The person is then asked to consider areas where they most desire improvement and distribute a theoretical 10 tokens across the nominated areas. A total score is generated by multiplying the severity rating by the number of priority tokens allocated and summed. This process avoids the over-reporting that can occur when people are asked to choose from a list of symptoms, rather than spontaneously declaring them(24), and provides information about priorities in the person’s own words.

The only measure that has been recommended to date for assessing the health impact of post-COVID is the SF-36(25). We used it in its publicly available form (RAND-36)(26). We also used an internationally recognized health utility measure, the EQ-5D-5L(27, 28), which queries 5 functions: walking, usual activities, self-care, pain/discomfort, and anxiety/depression. These were supplemented with a series of visual analogue scales to cover the additional areas of fatigue, sleep, distress, shortness of breath (SOB), health rating, and overall quality of life (29, 30). Two additional single questions asked about fatigue impact (31) and minutes of exercise over a week. Questionnaires on post-exertional malaise (PEM) (32), post-traumatic stress disorder (PTSD)(33) and cognitive concerns(34) were also administered. Demographic information, co-morbidity, and COVID experience and vaccine history were also collected. Assessment tools had strong evidence for validity with robust measurement properties, were available in English and French, and in most cases had existing Canadian or Quebec norms(35-37) or comparative values(34)aiding interpretation. Each participant’s information was summarized, interpreted, and shared with them in the form of a personal dashboard, similar to one we had previously designed and tested for people living with HIV(38).

Data Analysis

Descriptive statistics were calculated for each variable after recruitment of 100, 200, 300 and 414 participants and values on study measures did not vary across these waves suggesting that early and late responders did not differ on severity. Two people had missing data on language and on hospitalization. Where relevant, the effect of sex and age were estimated using linear, quantile (median), ordinal, or logistic regression depending on the distribution of the variable under study. When there was a sex effect, data were presented separately for men and women. For a sex effect to be considered important, the confidence interval around the estimated effect had to exclude the null value and the difference in prevalence between men and women had to be greater than 10%(39). Text threads were analysed using natural language processing (i.e., stemming and lemmatization).

Cluster analysis (k-means) was used to identify the extent to which the functional consequences of fatigue, cognition, sleep, distress, physical function, and shortness of breath clustered in individuals. For this analysis, all variables were converted to a score out of 100, with 100 being the worst level. The effects of the symptom burden of PCS were estimated using linear regression by regressing health and quality of life outcomes on cluster membership, adjusted for age and sex. Quantitative analyses were conducted using SAS V9.4.

Interpretation

This study of adults self-identifying with PCS in Quebec, Canada found long-lasting, widespread, and severe effects across the full range of health outcomes (symptoms, function, health perception, and QOL). The symptom of most concern to this sample was fatigue, with over 80% reporting PEM and over 90% needing to rest during the day. Cognitive symptoms were also very prevalent with people scoring 40 (out of 100) points lower than expected on a measure of self-reported cognitive concerns.

Concentration difficulties were the most common cognitive issue. Respiratory symptoms were also frequent. Symptoms had substantial functional impact, with close to half the sample on sick leave. Health-related quality of life as assessed with the EQ-5D index values for the QAPC sample was 0.63, comparable to values reported for Canadians with stroke, multiple sclerosis, or advanced cancer(17, 18) and far below the normative value for Quebec (0.80).(37)

A similar profile of PCS burden has been reported by others(15, 23, 40-45) although it is difficult to compare studies owing to the differences in how samples were assembled. PCS is by definition a self-reported syndrome and hence how people are recruited into studies will affect estimates of symptom burden. The QAPC sample was predominately women (76%), as in other studies with self-identified PCS samples(46). In contrast, in the UK census study that is the best available population-level data on self-reported PCS to date (6), the women to men ratio was ≈3:2. However, we found that women and men were similar on the outcomes we assessed. We observed effects of age on some variables.

Paradoxically, while older age was associated with higher likelihood of hospitalization during acute infection, it was associated with lower PCS symptom burden and better QOL.

We assessed symptom clustering, seeking evidence for sub-syndromes that might have different etiologies or contributors. The observed pattern is consistent with a range of severity rather than sub-syndromes, with fatigue an early feature in even the mildest stages, and additional symptoms accruing with increasing severity. A recent systematic review of 151 studies from 32 countries with a total of 1,285,407 participants who would meet criteria for PCS(12) reported that the proportion of people with at least one symptom decreased over time from 56.0% (at 1–3 months) to 37.8% (at 6–12 months). This indicates that over a third of people reporting PCS can be symptomatic for a year or more. While the cross-sectional observations here do not address the evolution of PCS, participants had symptoms for nearly a year at study entry, on average. On-going longitudinal follow-up will shed light on the evolution and day-to-day variability in this seemingly chronic phase of COVID-19.

This study is affected by the same limitations as much of the existing literature on PCS, as there is no base population to sample from and people entering the study were likely those with the most severe and persistent symptoms. Further, while over 90% of the QAPC sample reported a positive COVID test, infection status was not independently verified. There was widespread availability of PCR testing in the early waves of the pandemic in Quebec, and home tests were provided free of charge in pharmacies thereafter. Study participants likely had SARS-CoV-2 infection with persistent disabling chronic symptoms and most likely to seek help.

While the findings from this self-identified sample do not provide information on prevalence, they can inform healthcare service planning: People identified relief from fatigue, pain, and from cognitive, respiratory, and cardiac symptoms as their top priorities for treatment. This information could help prioritize current management recommendations which are based on reported frequency and expert opinion.(47) Rehabilitation approaches may be particularly helpful, given the nature, functional impact, and chronicity of the most bothersome symptoms. Most people had multiple symptoms, indicating that a multidisciplinary approach to assessment and treatment is warranted. Management of the most common chronic symptoms falls within the scope of practice of physical therapy, occupational therapy, psychology, and nursing. Social services addressing the financial and functional impact of PCS will also be important.

These initial results from the QAPC study provide good information as to the range of problems experienced, their severity, and patient-centered priorities for intervention in the Canadian context. There is a need for people with PCS to have access to clinical care for management of complications, rehabilitation, as well as support for self-management of symptoms if appropriate. Further work is needed to identify the most effective interventions and to identify risk and resilience factors in the evolution of symptoms over time.