Introduction
Long COVID has been characterized as the continuation or development of new symptoms 3 months after the initial SARS-CoV-2 infection, with no other explanation [1]. Approximately 10–30% of persons infected with SARS-CoV-2 develop long COVID. The wide range in estimates of the prevalence of long COVID can be attributed to a number of factors, including a lack of standardized biomarkers and differences in study design, recruitment strategies, symptom assessment tools employed, and health care environments [Reference Greenhalgh2, Reference Alwan and Johnson3]. Although the exact numbers of those living with the condition are uncertain, at least 65 million people are estimated to be experiencing long COVID globally [Reference Lancet4, Reference Davis5].
Long COVID symptoms may be reported months or even years after infection [Reference Davis6–Reference Oelsner8]. The persistence of long COVID symptoms also leads to ongoing demands on healthcare services and can have substantial negative impacts on the ability to work, resulting in significant economic consequences for individuals and society [Reference Tene9–Reference Woldegiorgis12].
Because a specific diagnostic test is lacking, there is a reliance on self-reported symptoms to diagnose long COVID [Reference Wu13–Reference Krysa15]. The diverse symptomatology associated with this condition, however, makes developing a precise case definition difficult [Reference Iwasaki and Putrino16, Reference Ely, Brown and Fineberg17]. Even more challenging has been identification of key factors which can explain the varying clinical course of COVID over time [Reference Buonsenso and Tantisira18–Reference Ballouz20]. Early investigations revealed that symptoms associated with long COVID could increase, decrease, or plateau during the first year following SARS-CoV-2 infection [Reference Davis6, Reference Tran21]. Several clinical trajectories for individuals with long COVID have recently been identified including cohorts with rapidly decreasing versus highly persistent symptomatology, with the latter group likely responsible for the significant ongoing burden of long COVID from a healthcare and economic perspective [Reference Servier19, Reference Fischer22].
The main aim of this study was to investigate factors associated with persistence of symptoms among individuals who had long COVID 3 months after a first SARS-CoV-2 infection with the Omicron variant in Western Australia (WA).
Methods
This longitudinal study prospectively followed persons who had long COVID 3 months after SARS-CoV-2 infection out to 6 months. Persons 18 years of age or over with laboratory-confirmed infection between 16 July and 03 August 2022, and long COVID diagnosed in a previous study, were eligible for inclusion [Reference Woldegiorgis12].
In the earlier study, long COVID was defined as ‘the continuation or development of new symptoms 3 months after the initial SARS-CoV-2 infection with no other explanation’. In this follow-up study, ‘persistent long COVID’ is defined as the presence of one or more COVID-19 illness-related symptom/s or health issue/s 6 months post-SARS-CoV-2 infection among persons who had long COVID 3 months post-infection; ‘recovered long COVID’ is defined as persons who had long COVID at 3 months post-infection who no longer report related symptoms 6 months post-infection. ‘Fully vaccinated’ was defined as receipt of 3 or more doses of a COVID-19 vaccine at least 7 days before SARS-CoV-2 infection, and ‘not fully vaccinated’ was two or fewer doses [Reference Woldegiorgis12].
In January 2023, the WA Department of Health sent a follow-up survey via text message to persons with long COVID at 3 months who had consented to further follow-up. The responses from the 6-month survey were linked to those from the 3-month survey and information recorded during an interview conducted at the time of their initial SARS-CoV-2 infection, which included demographics, COVID-19 vaccination status prior to infection, and pre-existing health conditions.
Like the 3-month survey, the 6-month survey solicited information on the presence of 22 specific symptoms (henceforth called ‘solicited symptoms’), as well as healthcare utilization and impacts on work or study. Non-respondents were reminded of the opportunity to participate in the research via a text message. Persons who reported having a SARS-CoV-2 re-infection diagnosed by polymerase chain reaction or rapid antigen tests between the 3- and 6-month surveys were excluded.
Analysis
We compared baseline demographic and health characteristics in people who completed and people who did not complete (did not consent/consented but did not complete) the survey using chi-squared tests.
Among those who completed the 6-month survey, we calculated the number and proportion with persistent long COVID. We described long COVID symptoms reported at 3 months post-infection separately for those with and without persistent long COVID, including the mean number of symptoms and proportions with each of the solicited symptoms.
We used Poisson regression with robust error variance to estimate relative risks (RR), with 95% confidence intervals (CI) of potential predictors of persistent long COVID. To examine the association of symptoms reported at 3 months with persistent long COVID at 6 months, we assessed each symptom individually while treating sex, 10-year age group, region (metropolitan vs. rural), vaccination status, and any significant or long-standing health issues at the time of the acute SARS-CoV-2 infection as potential confounders, using inverse probability weighting (IPW). These potential confounders were selected for inclusion based on our previous work, which found they were associated with long COVID 3 months post-SARS-CoV-2 infection [Reference Woldegiorgis12]. To quantify the risk of persistent long COVID associated with increased overall symptomatology, we used a multivariable model, simultaneously adjusting for potential confounders and the number of solicited long COVID symptoms reported by participants at 3 months.
To examine health trajectories among those with persistent long COVID, we assessed the change in the number and type of symptoms individuals reported at 3 and 6 months post-infection as well as reductions in work or study and health care visits in the month prior to the survey. Statistical analysis was conducted in Stata 15 [23].
Results
Survey response
Of 2,291 participants with long COVID 3 months post-infection, 2,086 (91.1%) consented to follow-up and were sent a text message with a link to the 6-month survey. Of those invited, 1,293 (61.9%) completed the survey. Data for 59 respondents were excluded because re-infection between the 3- and 6-month surveys was retrospectively identified, resulting in a final study population of 1,234 respondents, that is, 60.8% of the 2,027 persons with a single SARS-CoV-2 infection invited to participate (Figure 1).
Figure 1. Flowchart of participants for the long COVID follow-up study, Western Australia, 12-30 January 2023. Notes: 1A total of 11,697 participants completed the three-month survey during the initial study. Of those, 2,291 met the case definition for long COVID three months post-infection. 2Survey not sent due to an administrative error.
Characteristics of those who did not consent to follow-up, and those who consented and completed or did not complete the survey, are shown in Table 1. In general, these three groups were similar, with the exception that survey respondents tended to be older and had received more COVID-19 vaccine doses. Females comprised about two-thirds of all three groups (range: 65.8%–67.4%). Importantly, persons who consented to follow-up and completed the 6-month survey were similar to those who consented but did not complete the survey in terms of the mean number of long COVID symptoms reported at 3 months and the proportion reporting any significant or long-standing health issues prior to their initial SARS-CoV-2 infection.
Table 1. Baseline characteristics of respondents and non-respondents
1 Excludes five persons for whom sex was not known.
Risk of persistent long COVID
Recovery or persistence of long COVID 6 months post-infection
Of the 1,234 study participants, 724 (58.7%) reported the presence of one or more COVID-19 illness-related symptom/s or health issues 6 months post-SARS-CoV-2 infection and hence were classified as persistent long COVID; 510 respondents (41.3%) reported no longer having COVID-19 illness-related symptoms or health issues 6 months post-SARS-CoV-2 and were classified as recovered long COVID.
Comparing long COVID symptoms reported at 3 months between those who recovered and those with persistent long COVID
Participants with persistent long COVID reported a mean number of 7.2 (95% CI, 6.9–7.6; range 1–20) solicited symptoms 3 months post-infection compared to a mean of 5.3 (95% CI, 4.9–5.6; range 1–19) among those who recovered.
There was broad similarity in the pattern of solicited symptoms reported at 3 months post-SARS-CoV-2 infection between the cohort that recovered and those with persistent long COVID (Figure 2). Although the cohort with persistent long COVID had a higher frequency of most solicited symptoms at 3 months, the difference in proportions was usually modest, averaging 9.4% (range 0.9%–18.3%) across the 22 symptoms queried.
Figure 2. Proportion of respondents reporting each solicited symptom at three months, by recovered vs persistent long COVID.
Predictors of persistent long COVID
Of 22 solicited symptoms, 17 were significant predictors of persistent long COVID when adjusted for age, sex, region, vaccination status, and pre-existing health conditions (Table 2). When adjusted for all other solicited symptoms and confounders simultaneously, only tiredness/fatigue (Relative Risk [RR] = 1.27, 95% CI: 1.11–1.47), shortness of breath (RR = 1.15, 95% CI: 1.04–1.26), and cough (RR = 1.10, 95% CI: 1.01–1.20) remained significant independent predictors of persistent long COVID.
Table 2. Long COVID symptoms 3 months post-infection associated with persistent long COVID among 1,234 study participants
1 Applied Inverse Probably weighting (IPW). Confounders used to calculate the IPW include age, sex, region, pe-exiting health issues and vaccine doses.
2 Among women of reproductive age (i.e., aged 18 to <=49 years); the denominator is 290.
After multivariable adjustment for sex, age group, region, and vaccination status, the number of symptoms reported at 3 months (i.e., < 6 vs. ≥6) and pre-existing health conditions at the time of initial SARS-CoV-2 infection were significant independent predictors of persistent long COVID (Table 3) Persons with six or more symptoms had a 42% increased risk (RR = 1.42, 95% CI: 1.29–1.58) and those with pre-existing health conditions had a 18% increased risk ([RR] = 1.18, 95% CI: 1.07–1.29) compared to those without (Table 2). When number of symptoms was treated as a numerical variable, each additional symptom increased the risk of persistent long COVID by 5% ([RR] = 1.05, 95% CI: 1.04–1.06). Age, sex, region of residence, and vaccination status were not independent predictors of persistent long COVID in our model (p > 0.05).
Table 3. Demographic and health factors associated with persistent long COVID among 1,234 study participants
1 Poisson model adjusted for all other characteristics in this Table.
2 Excludes five persons for whom sex was not known.
Changes in symptoms, and health service use and return to work/study at 6 months, among individuals with persistent long COVID
Change in symptoms between 3 and 6 months
Among those with persistent long COVID, the number of solicited symptoms reported remained stable over time. The mean number of symptoms reported at 3 and 6 months post-SARS-CoV-2 infections was 7.2 (standard deviation [SD] = 4.0) and 7.1 (SD = 4.1), respectively, and the median was 7.0 (interquartile range [IQR] = 4–10) at both timepoints.
Figure 3 depicts the proportion of individuals who experienced resolution or development of a given solicited symptom across the 3- and 6-month time points. There was variability in the specific symptoms reported by an individual over time. However, the net change in the overall prevalence of a specific solicited symptom between 3 and 6 months was typically modest (< 5%), with the number of individuals who reported the symptom at 3 but not 6 months being largely offset by the number of individuals who reported the symptom at 6 but not 3 months. The only exceptions to this were cough, which saw a net reduction in prevalence of 13.8% and change in menses, with a net increase of 5.3%.
Figure 3. .Proportion of individuals with persistent long COVID reporting resolution or new onset of a solicited symptom at three and six months after SARS-CoV-2 infection
Health service use
A third (33.6%; 243/724) of those with persistent long COVID reported accessing health services in the month prior to the 6-month survey due to ongoing symptoms of long COVID. This figure was only marginally lower than the percentage (38.7%) accessing health services in the month prior to the 3-month survey [Reference Woldegiorgis12]. The vast majority of the healthcare encounters for those with persistent long COVID were outpatient consults with a General Practitioner (97.1%; 236/243); emergency department visits (10.3%; 25/243) and hospital admissions were uncommon (5.8%; 14/243).
Return to work or study
Among the 559 persons with persistent long COVID who reported working/studying prior to their SARS-CoV-2 infection, 179 (32.0%) had ceased or reduced their hours of work or study due to long COVID at 6 months of follow-up. This figure was higher than the proportion (17.8%) who reported reducing or discontinuing work or study in the 3-month survey [Reference Woldegiorgis12]. Of those with persistent long COVID who reported discontinuing or reducing their work or study hours, 68 (38.0%) had already done so at 3 months of follow-up, but 92 (51.4%) reported fully returning to work or study within 3 months of their initial SARS-CoV-2 illness and then subsequently stopping or reducing their hours at 6 months of follow-up.
Discussion
This follow-up study of persons who had long COVID 3 months post-SARS-CoV-2 infection found that four out of 10 persons reported full recovery from long COVID between 3 and 6 months. Reporting one or more long-standing health issues at the time of the initial SARS-CoV-2 infection and six or more symptoms at 3 months were independent predictors of having persistent long COVID. Examining specific symptoms showed the risk of persistent long COVID was higher for those who reported fatigue, shortness of breath, or cough 3 months post-infection, though for each of these, the attributable increase in risk was modest (i.e., 10–27%).
Importantly, we identified a substantial overlap in terms of the number of solicited symptoms reported at the 3-month time point between the cohort that recovered (mean = 5.3; range 1–19) and those who did not (mean = 7.2; range 1–20). In addition, we observed that the types and frequencies of symptoms reported at 3 months among those with persistent long COVID paralleled, albeit at a higher rate, the symptoms reported among those had who recovered.
Prior to analysing our data, we had anticipated that almost all persons would have some resolution of their symptoms over time and that developing new symptoms more than 3 months after infection would be infrequent. Instead, we found substantial variability in the specific symptoms individuals reported at 3- and 6-month follow-up. Though not anticipated, this observation is not unique to our study population, as considerable heterogeneity in an individual’s long COVID symptomatology over extended follow-up has been reported in other settings [Reference Ballouz20]. Another unexpected finding was that there was essentially no reduction in the mean number of symptoms reported by those with persistent long COVID across the 3- and 6-month time points, with the median number of solicited symptoms reported remaining at 7.0. This suggests that self-reported clinical improvement was very limited over this time frame.
Precise mechanisms to explain why some persons recover completely from long COVID and others have little improvement in their symptomology over time remain elusive [Reference Gheorghita24]. Our data suggest that the existence of medical issues before SARS-CoV-2 infection may play a role. In addition, fatigue that interferes with daily life reported at 3 months was the symptom most strongly associated with persistent long COVID, followed by dyspnoea. Both symptoms are characteristic of myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS), as are ‘brain fog’, sleep disruption, and myalgia/arthralgia, which were reported by a majority of persons with persistent long COVID in our study. There is growing evidence that SARS-CoV-2 infection can trigger ME/CFS. Moldofsky et al. report that persistent symptoms experienced by long COVID patients are similar to those described for ME/CFS and a recent meta-analysis found that approximately half of long COVID patients satisfy a diagnosis of ME/CFS [Reference Moldofsky and Patcai25, Reference Dehlia and Guthridge26]. Emerging work suggests that persistence of SARS-CoV-2 infection and development of chronic autonomic dysfunction may be important in explaining ongoing symptomatology associated with long COVID [Reference Zuo27–Reference Fedorowski and Sutton29]. In addition, there is substantial evidence suggesting a role for psychological mechanisms in development of long COVID, but these constructs have been inadequately explored and warrant further investigation [Reference Engelmann30, Reference Lemogne31].
Our study contributes to efforts to better understand the evolution of long COVID illnesses over time. In France, Servier and colleagues identified ‘3 trajectories’ for individuals with long COVID, that is, highly persistent, rapidly decreasing, and slowly decreasing symptoms [Reference Servier19]. However, their study included participants with suspected COVID-19 disease who lacked laboratory evidence of infection, did not report on vaccination status, and spanned a period when three different SARS-CoV-2 variants circulated. Our study extends observations on long COVID illness trajectories to a cohort of highly vaccinated individuals with lab-confirmed infection caused by a single SARS-CoV-2 variant (Omicron), and we observed two distinct trajectories, which ultimately led to full or minimal recovery at 6 months follow-up.
Our findings also underscore the ongoing burden of long COVID on the healthcare system, with a third of those with persistent long COVID seeking medical attention for their symptoms in the month prior to the 6-month survey. Most participants who sought care attended General Practice, highlighting the critical role of primary care in managing long COVID and thus the need to ensure adequate, ongoing resource allocation to primary care settings [Reference Woldegiorgis12]. In addition, there is emerging evidence to suggest that investment in speciality ‘long COVID clinics’ may be warranted, as early treatment at such clinics has been associated with fewer downstream inpatient stays and reduced mortality [Reference Gong32]. For individuals experiencing long COVID, one of the goals of treatment should be to increase the proportion on a trajectory of rapidly decreasing symptoms. Innovative models of care that can be delivered in primary care settings and focus on patient-led self-care should be prioritized. Given the substantial overlap of symptom clusters between persistent long COVID and ME/CFS, existing ME/CFS care guidelines might be adapted to facilitate standardized diagnosis and management of persistent long COVID [Reference Lapp33–Reference Cortes Rivera35]. The substantive proportion of persons with persistent long COVID in our study who reported symptoms consistent with anxiety and depression suggests that incorporating access to robust mental health services may also be helpful in optimizing recovery [Reference Luo36].
Finally, we found that about a third of those with persistent long COVID were not fully back at work or study 6 months after their initial SARS-CoV-2 infection, including a substantial number of people who had previously fully returned to these activities by 3 months. These data are consistent with other investigations reporting that the impact of the disease on patients’ lives began increasing 6 months after their initial illness and corroborate other studies in documenting the significant economic impact of long COVID for both employees and employers [Reference Tene9, Reference Reuschke and Houston37–Reference Gualano39]. Workers with long COVID face a number of challenges in returning to work, including impairment of cognitive function, decreased physical endurance, mental health issues, and societal stigma [Reference Clutterbuck40, 41]. Ultimately, employers can better retain workers experiencing long COVID by creating supportive policies, which, depending on individual circumstances, may include phased return to work and an option for working from home, alterations to hours, duties, and workload, equipment modifications, and time off for healthcare appointments. Occupational health guidelines supporting sustainable return to work for persons with long COVID are available [Reference Stave, Nabeel and Durand-Moreau42–Reference Rayner, Burton and MacDonald44].
In contrast to some other investigations, our analysis did not identify an association between COVID vaccinations and a reduced risk of persistent long COVID. It must be noted, however, that our study population was not well suited to examine this issue because only 5% of the participants had received fewer than three doses prior to their initial SARS-CoV-2 infection. Given the small size of the unvaccinated comparator group in our study, this result should be interpreted with caution.
This study has several limitations. First, long COVID symptomatology was based on self-reported information, as opposed to clinical observations verified in medical records. This approach is not uncommon for large-scale studies where clinical assessments may not be feasible for all participants and is supported by the fact that, when studied, objective clinical signs of impairment among individuals with persistent long COVID correlate well with self-reported symptoms [Reference Peter45]. Second, despite a reasonably high response rate (60%), there remains a substantial proportion of individuals who did not participate, and it is unknown if the non-respondents had different experiences or outcomes which might affect the results. This concern is mitigated, however, by the fact that respondents were similar to non-respondents in terms of key long COVID symptoms reported at 3 months and the proportion reporting any significant or long-standing health issues prior to SARS-CoV-2 infection, as shown in Table 1. Third, our questionnaire did not capture subjective or objective measures of symptom severity, and this information may be important for assessing clinical predictors of persistent long COVID. Last, our survey did not include questions on some factors associated with long COVID in other settings such socio-economic status, body mass index, ABO blood groups, and smoking [Reference Luo46, Reference Wong47].
There are, however, several strengths to this research. First, Western Australia’s pandemic experience enabled assessment of long COVID following a single laboratory-proven SARS-CoV-2 infection with the Omicron variant, thus eliminating the potential impact of repeat infections, simultaneous circulation of different variants, and ambiguous timeframes [Reference Bloomfield48]. Second, vaccination status for participants was obtained from a mandatory, population-based national immunization register. Third, we assessed longitudinal data on individuals gathered across three different time points, that is, a detailed case investigation that occurred at the time of their initial SARS-CoV-2 infection and follow-up surveys conducted at 3 and 6 months. This approach reduces the potential for bias and provides an opportunity to examine changes in the symptom profile of long COVID in the same individual over time.
In conclusion, longitudinal follow-up among a cohort of individuals with long COVID 3 months post-infection found the risk of persistent long COVID at 6 months was greater for those with pre-existing health issues at the time of infection and ongoing fatigue and respiratory symptoms at 3 months. Importantly, we observed that many highly symptomatic persons fully recovered by 6 months and that there was substantial overlap in long COVID symptomatology at 3 months among those who recovered and those who did not. Of note, the cohort with persistent long COVID reported almost no improvement in overall symptomatology between the 3- and 6-month timepoints. These data suggest that there may be distinct clinical trajectories for long COVID between 3- and 6-month follow-up, that is, full versus minimal recovery, which could have implications for support and management of affected individuals.
Data availability statement
The data underlying this report are available from the WA Department of Health upon request and approval.
Author contribution
Conceptualization: P.E., M.W., R.K., P.A., J.M.; Survey design: P.E., M.W., S.N., G.C., L.B.; Survey administration: P.K.; Drafting and analysis: M.W., P.E.; Review & editing: M.W., L.B., R.K., G.C., S.N., P.K., A.J., J.M., P.A., P.E.
Funding statement
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Competing interests
The author(s) declare none.
Ethical standard
The study received ethics approval from the Western Australia Department of Health Human Research Ethics Committees (PRN: RGS0000005516).
Open access
