Real-world outcomes of a multidisciplinary rehabilitation programme for adults with long COVID: a retrospective cohort study.
BMC public health · 2026
Abstract originale
BACKGROUND: Long COVID imposes a substantial and ongoing burden, with persistent symptoms affecting daily functioning and quality of life. Multidisciplinary rehabilitation has been widely implemented in clinical practice; however, evidence describing real-world patient outcomes and service delivery remains limited. This study aimed to describe clinical and patient-reported outcomes among adults participating in a structured multidisciplinary rehabilitation programme for Long COVID, and to explore the association between referral-to-assessment time and changes in outcomes. METHODS: This retrospective cohort study analysed routinely collected clinical data from adult patients who completed a structured multidisciplinary rehabilitation programme at the Long COVID rehabilitation centre in a university hospital in South London (January 2021-December 2023). Seven validated outcome instruments were administered at baseline and discharge: the Work and Social Adjustment Scale (WSAS), Fatigue Severity Scale (FSS), MRC Dyspnoea Scale, Patient Health Questionnaire-2 (PHQ-2), Generalised Anxiety Disorder-2 (GAD-2), EQ-5D-5 L, and 1-minute Sit-to-Stand test (STS-1). Pre-to-post changes were assessed using paired t-tests and Wilcoxon signed-rank tests; the proportion of patients achieving minimal clinically important differences (MCIDs) was determined for each measure. The association between referral-to-assessment time and outcome change was examined using linear correlation and restricted cubic spline (RCS) modelling. RESULTS: A total of 170 participants were included. Statistically significant improvements were observed across all primary domains after adjustment for multiple comparisons, except self-care and pain/discomfort. The largest effect was recorded for physical endurance (STS-1: r = 0.590, p < 0.001), followed by breathlessness (MRC: r = 0.491), fatigue (FSS: r = 0.382), and functional impairment (WSAS: d = - 0.323). MCID thresholds were achieved by 63.5% of patients on the STS-1, 48.8% for breathlessness, and 45.9% for depression. No significant linear association was identified between referral delay and outcome change. Non-linear therapeutic windows were identified in functional and psychological domains using RCS modelling. Patients assessed 100-150 days after referral showed the greatest improvement, with benefits declining thereafter. CONCLUSION: Participants demonstrated improvements in clinical and patient-reported outcomes over time during engagement in a multidisciplinary rehabilitation programme for Long COVID. While causal attribution is limited by the observational design, these findings provide real-world evidence on patient outcomes and highlight the potential importance of timely access to rehabilitation services. Further prospective and controlled studies are needed to evaluate effectiveness and optimise care pathways.
Parole chiave
Leggi il testo integrale originale
Testo estratto dalla pubblicazione: sono esclusi immagini, media, supplementi e parti soggette a diritti distinti.
Real-world outcomes of a multidisciplinary rehabilitation programme for adults with long COVID: a retrospective cohort study
Background
Long COVID imposes a substantial and ongoing burden, with persistent symptoms affecting daily functioning and quality of life. Multidisciplinary rehabilitation has been widely implemented in clinical practice; however, evidence describing real-world patient outcomes and service delivery remains limited. This study aimed to describe clinical and patient-reported outcomes among adults participating in a structured multidisciplinary rehabilitation programme for Long COVID, and to explore the association between referral-to-assessment time and changes in outcomes.
Methods
This retrospective cohort study analysed routinely collected clinical data from adult patients who completed a structured multidisciplinary rehabilitation programme at the Long COVID rehabilitation centre in a university hospital in South London (January 2021–December 2023). Seven validated outcome instruments were administered at baseline and discharge: the Work and Social Adjustment Scale (WSAS), Fatigue Severity Scale (FSS), MRC Dyspnoea Scale, Patient Health Questionnaire-2 (PHQ-2), Generalised Anxiety Disorder-2 (GAD-2), EQ-5D-5 L, and 1-minute Sit-to-Stand test (STS-1). Pre-to-post changes were assessed using paired t-tests and Wilcoxon signed-rank tests; the proportion of patients achieving minimal clinically important differences (MCIDs) was determined for each measure. The association between referral-to-assessment time and outcome change was examined using linear correlation and restricted cubic spline (RCS) modelling.
Results
A total of 170 participants were included. Statistically significant improvements were observed across all primary domains after adjustment for multiple comparisons, except self-care and pain/discomfort. The largest effect was recorded for physical endurance (STS-1: r = 0.590, p < 0.001), followed by breathlessness (MRC: r = 0.491), fatigue (FSS: r = 0.382), and functional impairment (WSAS: d = − 0.323). MCID thresholds were achieved by 63.5% of patients on the STS-1, 48.8% for breathlessness, and 45.9% for depression. No significant linear association was identified between referral delay and outcome change. Non-linear therapeutic windows were identified in functional and psychological domains using RCS modelling. Patients assessed 100–150 days after referral showed the greatest improvement, with benefits declining thereafter.
Conclusion
Participants demonstrated improvements in clinical and patient-reported outcomes over time during engagement in a multidisciplinary rehabilitation programme for Long COVID. While causal attribution is limited by the observational design, these findings provide real-world evidence on patient outcomes and highlight the potential importance of timely access to rehabilitation services. Further prospective and controlled studies are needed to evaluate effectiveness and optimise care pathways.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-29023-3.
Introduction
Long COVID is defined as an infection-associated chronic condition that occurs after severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and is present for at least 3 months as a continuous, relapsing and remitting, or progressive disease state that affects one or more organ systems [1]. Epidemiological data indicate that the global pooled prevalence of Long COVID is approximately 36% [2], with a cumulative incidence of around 400 million individuals worldwide as of 2024 [3]. Beyond the clinical burden, the condition carries a staggering socio-economic cost, with an estimated annual global economic impact of $1 trillion, equivalent to roughly 1% of the global economy [3]. The condition impacts multiple organ systems, causing fatigue, breathlessness, cognitive issues, and psychological distress, which can significantly impair daily functioning and quality of life (QoL) [4, 5]. Many individuals are unable to work or participate socially, and healthcare systems face increased demand for specialised services [6]. In the United Kingdom (UK), the Office for National Statistics (ONS) reported that as of March 2024, approximately 2.0 million people in England and Scotland (3.3% of the population) were experiencing Long COVID symptoms [7], with a significant proportion reporting substantial limitations in their daily activities. The complexity and prevalence of Long COVID highlight the need for coordinated evidence-based rehabilitation strategies.
Multidisciplinary rehabilitation — integrating physiotherapy, occupational therapy, psychological support, and self-management education — has been adopted as a cornerstone of Long COVID care across many health systems, guided by national and international clinical frameworks [8]. Emerging evidence suggests that graded exercise therapy (GET), respiratory physiotherapy, and pacing strategies may improve physical capacity, reduce fatigue burden, and alleviate breathlessness in Long COVID cohorts [9, 10]. However, the existing literature is constrained by several methodological limitations: many existing studies are relatively small and often lack standardised outcome reporting frameworks; studies are predominantly small, limiting comparability and the strength of evidence available to guide practice [11]. Furthermore, most investigations have not employed validated, multidomain outcome instruments — such as the Work and Social Adjustment Scale (WSAS), Fatigue Severity Scale (FSS), MRC Dyspnoea Scale, Patient Health Questionnaire-2 (PHQ-2), Generalised Anxiety Disorder-2 (GAD-2), EQ-5D-5 L, the EQ Visual Analogue Scale (EQ-VAS), or 1-minute Sit-to-Stand test (STS-1)— applied systematically at both programme entry and discharge, thereby limiting the rigour of pre-to-post comparisons and clinical inference [12, 13].
A critical but poorly characterised aspect of Long COVID rehabilitation care is the potential influence of referral-to-treatment delay on clinical outcomes. In established rehabilitation medicine, the principle of early intervention is well recognised: delayed access to rehabilitation following acute illness or injury is associated with greater functional decline, psychological deterioration, and reduced responsiveness to intervention [14]. Yet, in the context of Long COVID, no consensus exists on the optimal timing of rehabilitation initiation, and the relationship between the time from referral to first assessment and subsequent functional and psychological gains has not been systematically examined [15]. This gap is clinically important; given the variability in referral pathways, triage practices, and service capacity across health systems, understanding whether delayed access attenuates rehabilitation benefit — or whether therapeutic gains remain achievable irrespective of delay — has direct implications for service design, patient prioritisation, and resource allocation [6, 16]. Moreover, the real-world effectiveness of structured Long COVID rehabilitation, evaluated using validated instruments across functional, physical, psychological, and quality-of-life domains simultaneously, remains insufficiently characterised in large, routinely collected clinical datasets [11].
To address these evidence gaps, we analysed data from adult Long COVID patients who underwent and completed the Long COVID rehabilitation programme. We aimed to: (i) quantify changes following rehabilitation in functional status, physical performance, psychological well-being, and health-related quality of life using validated outcome measures; (ii) determine the proportion of participants achieving the minimal clinically important differences (MCID) for each outcome measure; and (iii) examine whether referral-to-assessment time is associated with the magnitude of improvement in rehabilitation outcomes.
Methods
Study design
This study was a retrospective cohort analysis of routinely collected clinical data from patients with Long COVID. The participants underwent structured multidisciplinary rehabilitation at the Long COVID rehabilitation centre in the university hospital in South London between January 2021 and December 2023.
Study setting and eligibility criteria
This hospital-based study was conducted at a secondary healthcare institution in South London. The structured multidisciplinary rehabilitation programme typically spanned 12 weeks and was tailored to individual clinical needs. Delivered via onsite and virtual formats, the intervention integrated fatigue management education, psychological support, and, where appropriate, group-based graded exercise or movement therapy to address deconditioning while monitoring for post-exertional malaise (PEM).
The study cohort comprised participants (≥ 18 years) diagnosed with Long COVID who were referred from any of the eight Primary Care Networks and healthcare service clusters operating within the London Borough of Croydon. The diagnosis of Long COVID was established through a multidisciplinary team clinical triage in accordance with NICE guideline [NG188] [8]. This process involved a face-to-face or virtual assessment by a senior clinician (physician or specialist physiotherapist) who utilised the Newcastle screening tool [17] to confirm a history of acute SARS-CoV-2 infection via polymerase chain reaction (PCR), lateral flow device (LFD). Clinical confirmation required the exclusion of alternative underlying pathologies—such as cardiac or pulmonary complications—and the presence of a symptom profile that persisted for at least 12 weeks, ensuring that the cohort met the criteria for Post-COVID-19 Syndrome.
To be included in the final analysis, participants must have completed the full 12-week structured rehabilitation programme with documented baseline and endline assessments. Participants were excluded if they lacked documented referral outcomes, were deemed clinically unsuitable for rehabilitation due to active medical contraindications, or if the study team was unable to establish contact. Further exclusion criteria included declining participation, spontaneous recovery prior to the intervention, failure to complete the programme, non-attendance at the final reassessment, or death during the study period.
Data source, collection and variables
All data were extracted from the long COVID rehabilitation centre patient management files using a structured Excel sheet. Variables extracted included demographic (age, sex and ethnicity), referral sources, referral-to-assessment time (calculated as the number of days between the date of referral and the date of the first rehabilitation session), functional outcomes such as WSAS and STS-1, symptoms severity such as FSS and MRC dyspnoea scale, psychological outcomes such as PHQ-2 and GAD-2, QoL parameters: EQ-5D-5L (domains) and EQ-5D visual analogue scale (VAS). The description and scoring of each variable/instrument are in Supplementary File 1.
Study outcomes and validation
For each outcome measure, a change score (Δ) was computed as:1\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\Delta=\text{Endline Score}-\text{Baseline Score}$$\end{document}
Minimal clinically important differences (MCIDs) were defined based on published literature: the WSAS assessed the impact of symptoms across five life domains (work, home management, social leisure, private leisure, and relationships). Scores range from 0 to 40, with higher scores indicating greater functional impairment [18]. The FSS consists of 9 items, rated on a 7-point Likert scale (total score 1–7), with higher scores indicating greater fatigue impact on daily living [19]. Breathlessness was assessed using the MRC Dyspnoea Scale [20], a 5-point scale in which Grade 1 indicates minimal breathlessness and Grade 5 maximal disability. Depressive and anxiety symptoms were screened using the PHQ-2 [21] and the GAD-2 [22], respectively. Both instruments range from 0 to 6, with higher scores indicating increased symptom frequency and severity over the preceding 2 weeks. The EQ-5D-5 L [23] assessed HRQoL across five domains (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) on a 1–5 scale. Additionally, the EQ-VAS [24] assessed self-rated health on a 0–100 scale, where 100 represents the “best imaginable health.” Physical endurance was measured using the STS-1 test [25], which measures the number of repetitions completed in 60 s; higher counts reflect improved lower-limb strength and exercise capacity.
Statistical analysis
All statistical analyses were performed using R (version 4.3.3). Missing data, which occurred randomly and affected fewer than 20% of cases, were addressed using Multiple Imputation by Chained Equations (MICE). To ensure the stability of the estimates, 20 imputed datasets were generated over 20 iterations. The predictor matrix was optimised using the quickpred function, selecting predictors with a minimum correlation of 0.1 and at least 25% usable cases. Administrative date variables (referral, assessment, and completion dates) were excluded from the imputation model to maintain data integrity. Predictive mean matching was employed for all continuous variables, and the final results were pooled according to Rubin’s rules.
Descriptive statistics
Continuous variables were assessed for normality using the Shapiro–Wilk test. Normally distributed variables were reported as means with standard deviations, while non-parametric variables were presented as medians with interquartile ranges. Categorical variables were summarised using frequencies and percentages.
Change in outcomes in the rehabilitation
Paired-samples t-tests were used to assess outcome changes from baseline to discharge for outcomes that followed a normal distribution, while the Wilcoxon signed-rank test was applied to variables that were not normally distributed. Effect sizes were calculated using Cohen’s d for normally distributed paired data to assess the magnitude of change and rank-biserial correlation (rrb) for non-parametric comparisons. Benchmarks for Cohen’s d (used with t-tests and mean differences): small effect: d ≈ 0.20; moderate effect: d ≈ 0.50; and large effect: d ≈ 0.80 or greater. For correlation coefficients (r), often reported in Wilcoxon signed-rank tests, correlations, and regressions: small effect: r ≈ 0.10; moderate effect: r ≈ 0.30; and large effect: r ≈ 0.50 or greater. The p-values were adjusted using the Benjamini–Hochberg false discovery rate (FDR) method to control for multiple comparisons, with significance set at p < 0.005.
Correlation analyses were conducted to examine the association among changes in functional, psychological, and QoL outcomes. Change scores were calculated by subtracting baseline from endline values; negative scores indicate improvement for measures in which higher scores indicate greater impairment. Pearson’s product–moment correlation coefficients (r) were computed to assess the linear association between change scores across domains. This approach provided a standardised effect-size metric, allowing the interpretation of the magnitude and direction of associations relative to usual benchmarks.
Specifically, r values were categorised as 0.10 (small), 0.30 (moderate), and 0.50 (large). Correlations’ significance was assessed with two-tailed tests; p-values < 0.05 were significant. FDR correction was also used for multiple comparisons. The FDR is a statistical method used in multiple hypothesis testing to control the expected proportion of false positives (Type I errors) among all rejected hypotheses (discoveries) [26, 27].
Referral-to-assessment time association with changes in outcomes
To assess whether referral-to-assessment time was associated with changes in outcomes, univariable associations were first examined using Pearson correlation coefficients for normally distributed variables and Spearman’s rank correlation for skewed variables. When non-linearity was suspected between referral time and outcomes, RCS were used to model the relationship, following Harrell’s approach [28]. Model assumptions, including residuals’ normality, linearity, homoscedasticity, and multicollinearity (variance inflation factor < 5), were evaluated for robustness.
Results
Participants flow
The participant selection process is summarised in Fig. 1. A total of 1269 individuals were referred to the centre during the study period. After excluding duplicates (n = 596), 673 distinct patients were identified and assessed for eligibility. Of these, 503 individuals were excluded for the following reasons: lack of a documented referral outcome (n = 98); completion of the programme without a final reassessment (n = 167); deemed inappropriate for the programme during triage (n = 88); no contact established (n = 63); declined participation (n = 39); recovering (n = 20); failure to complete the programme (n = 20); non-attendance at the assessment (n = 7); and death (n = 1). Consequently, a final cohort of 170 participants who completed the programme and provided both baseline and post-programme outcome data was included in the final analysis.
Referral patterns and demographic profile
Referral volume peaked at 50.7% in 2021, then declined sharply through 2022 and 2023 (Supplementary File 2). Stratification by age group showed that the 55–64-year-old group consistently had the highest number of referrals over three years, indicating that middle-aged and older adults primarily used the service. Female patients outnumbered males, and referrals declined for both over time. White patients were the largest referral group each year, about 42.6% overall, up from 40.6% in 2021 to 50.9% in 2023. Black patients accounted for 20.3%, Asian patients for 16.1%, and other ethnicities for 21.0%.
Baseline and endline outcome distributions
Figure 2 presents paired boxplots with overlaid individual data points for all outcome measures, enabling direct comparison of central tendency and dispersion between programme entry (baseline) and discharge (endline). Functional impairment (WSAS; panel a) showed a downward shift in both the median and interquartile range, indicating improved social and occupational functioning. Fatigue severity (FSS; panel b) showed a clear reduction in the median, interquartile range, and high-end outliers, reflecting a clinically significant reduction in fatigue burden, although inter-individual variability persisted. Breathlessness (MRC dyspnoea scale; panel c) decreased from baseline to endline, consistent with the functional gains observed on the sit-to-stand test.
Depressive symptoms (PHQ-2; panel d) and anxiety (GAD-2; panel e) both shifted towards the lower end of their respective scales at endline, with GAD-2 values clustering in the lower range, indicating consistent mental health improvement across the cohort. For health-related QoL (HRQoL), all five EQ-5D-5 L domains — mobility (panel f), self-care (panel g), usual activities (panel h), pain/discomfort (panel i), and anxiety/depression (panel j) — showed lower severity scores at endline, suggesting broader functional and symptomatic recovery. The EQ-5D-5 L health score visual analogue scale (VAS; panel k) showed an upward shift in the median, reflecting improved self-rated health. Objective physical capacity, assessed by the STS test (STS-1; panel l), showed a substantial increase in repetitions at endline, indicating marked recovery in lower-limb endurance and exercise capacity.
Pre-to-post change analysis
Significant improvements were statistically observed across functional, physical, psychological, and HRQoL domains after rehabilitation (Table 1). The largest effect was observed for physical endurance: the STS-1 test showed a large effect size (r = 0.590, adjusted p < 0.001), indicating a substantial increase in lower-limb functional capacity. Breathlessness (MRC dyspnoea scale; r = 0.491, adjusted p < 0.001) and fatigue (FSS; r = 0.382, adjusted p < 0.001) both improved with medium-to-large effect sizes, confirming that the programme showed significant observed improvements in two of the most prevalent and disabling symptoms of Long COVID. Functional impairment (WSAS) improved significantly on a paired t-test (d = − 0.323, adjusted p < 0.001), reflecting significant gains in occupational and social functioning. Psychological outcomes improved significantly, albeit with moderate effect sizes: depressive symptoms (PHQ-2; r = 0.321, adjusted p < 0.001) and anxiety (GAD-2; r = 0.314, adjusted p < 0.001) both decreased from baseline to discharge. Within the EQ-5D-5 L, usual activities (r = 0.284, adjusted p < 0.001), anxiety/depression (r = 0.274, adjusted p = 0.005), mobility (r = 0.234, adjusted p = 0.009), and the VAS health score (r = 0.191, adjusted p = 0.028) were all statistically significant. Pain/discomfort (r = 0.154, adjusted p = 0.057) and self-care (r = 0.106, adjusted p = 0.116) did not reach significance after adjustments, indicating these domains were less responsive or reflected persistent symptoms.
Table 1Comparison of baseline and endline outcomes following multidisciplinary rehabilitation
Outcome Domain Outcome Measure Statistical Test Effect Size Adjusted p-value
Physical & functional STS Wilcoxon signed-rank r=0.590 < 0.001
WSAS (Functional impairment) Paired t-test d=-0.323 < 0.001
Symptom severity MRC Dyspnoea scale Wilcoxon signed-rank r=0.491 < 0.001
FSS (Fatigue) Wilcoxon signed-rank r=0.382 < 0.001
Mental health PHQ-2 (Depression) Wilcoxon signed-rank r=0.321 < 0.001
GAD-2 (Anxiety) Wilcoxon signed-rank r=0.314 < 0.001
QoL (EQ-5D) Usual Activities Wilcoxon signed-rank r=0.284 < 0.001
Anxiety/Depression Wilcoxon signed-rank r=0.274 0.005
Mobility Wilcoxon signed-rank r=0.234 0.009
Health Score (VAS) Wilcoxon signed-rank r=0.191 0.028
Pain/Discomfort Wilcoxon signed-rank r=0.154 0.057
Self-care Wilcoxon signed-rank r=0.106 0.116
Paired t-tests used for normally distributed data (Cohen’s d); Wilcoxon signed-rank tests for non-normal data (rank-biserial r). p-values adjusted using the Benjamini–Hochberg FDR procedure. WSAS Work and Social Adjustment Scale, FSS Fatigue Severity Scale, MRC (Medical Research Council) Dyspnoea Scale, PHQ-2 Patient Health Questionnaire-2, GAD-2 Generalised Anxiety Disorder-2, QoL (EQ-5D) Quality of life (EuroQOL Five Dimensions Questionnaire), VAS Visual Analogue Scale, STS-1 (1-minute Sit-to-Stand test)
Proportion of patients achieving MCID across outcome measures
MCID analysis showed significant improvements across domains, with varying proportions of patients meeting thresholds (Table 2). Physical capacity showed the highest improvement, with 63.5% meeting the MCID on the STS-1 test (n = 108; 95% CI: 56.1–70.4%), indicating physical endurance as the most responsive domain to rehabilitation. Breathlessness (MRC) was second at 48.8% achieving MCID (n = 83; 95% CI: 41.4–56.3%), followed by depressive symptoms (PHQ-2: 45.9%, n = 78; 95% CI: 38.6–53.4%), functional impairment (WSAS: 43.5%, n = 74; 95% CI: 36.3–51.0%), and anxiety (GAD-2: 41.8%, n = 71; 95% CI: 34.6–49.3%). HRQoL domains showed modest changes. The EQ-5D-5 L score and anxiety/depression reached MCID in 38.2% of participants, while usual activities (36.5%), mobility (33.5%), pain/discomfort (30.6%), and fatigue (27.1%) showed lower rates. Self-care had the lowest rate, with 22.4% meeting MCID (n = 38; 95% CI: 16.7–29.2%).
Table 2Proportion of participants achieving the minimal clinically important difference for each outcome measure
Outcome measure Improved (n) Proportion (%) 95% CI
WSAS 74 43.5 36.3–51.0
FSS 46 27.1 20.9–34.2
MRC 83 48.8 41.4–56.3
PHQ-2 78 45.9 38.6–53.4
GAD-2 71 41.8 34.6–49.3
EQ-5D Mobility 57 33.5 26.9–40.9
EQ-5D Self-care 38 22.4 16.7–29.2
EQ-5D Usual Activities 62 36.5 29.6–43.9
EQ-5D Pain/Discomfort 52 30.6 24.2–37.9
EQ-5D Anxiety/Depression 65 38.2 31.3–45.7
EQ-5D Health Score 65 38.2 31.3–45.7
STS 1-min 108 63.5 56.1–70.4
WSAS Work and Social Adjustment Scale, FSS Fatigue Severity Scale, MRC Medical Research Council Dyspnoea Scale, PHQ-2 Patient Health Questionnaire-2, GAD-2 Generalised Anxiety Disorder-2, EQ-5D-5 L EuroQol Five-Dimension instrument, STS 1-min 1-minute Sit-to-Stand test
These findings show that although the rehabilitation programme was associated with meaningful gains in physical and functional areas for many patients, improvements in QoL and symptom-specific outcomes, such as fatigue and self-care, were less consistent.
Linear association between referral-to-assessment time and outcome changes
No significant linear association was observed between referral-to-assessment time and the magnitude of change across any outcome measure (p > 0.05 for all; Table 3). Correlation coefficients were uniformly small, ranging from − 0.134 to 0.092. For most symptom and functional outcomes — including WSAS (r = − 0.105, p = 0.172), FSS (r = − 0.126, p = 0.101), and EQ-5D-5 L self-care (r = − 0.134, p = 0.081) — negative correlation coefficients suggested greater improvement with longer wait times, though none reached significance. PHQ-2 (r = 0.092) and MRC (r = 0.028) showed negligible positive correlations.
Table 3Correlation between referral-to-assessment time (days) and change scores for patient-reported and functional outcome measures
Outcome measure (change) Correlation (\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\:\boldsymbol{r}$$\end{document}) P-value
WSAS (Functional Impairment) -0.105 0.172
FSS (Fatigue) -0.126 0.101
MRC (Breathlessness) 0.028 0.712
PHQ-2 (Depression) 0.092 0.234
GAD-2 (Anxiety) -0.017 0.823
EQ-5D-5 L Mobility 0.060 0.435
EQ-5D-5 L Self-care -0.134 0.081
EQ-5D-5 L Usual Activities 0.050 0.518
EQ-5D-5 L Pain/Discomfort -0.037 0.636
EQ-5D-5 L Anxiety/Depression 0.002 0.984
EQ-5D-5 L Health Score -0.075 0.329
STS 1-min (Physical Capacity) -0.015 0.850
WSAS Work and Social Adjustment Scale, FSS Fatigue Severity Scale, MRC Medical Research Council Dyspnoea Scale, PHQ-2 Patient Health Questionnaire-2, GAD-2 Generalised Anxiety Disorder-2, EQ-5D-5 L EuroQol Five-Dimension instrument, STS 1-min 1-minute Sit-to-Stand test
The linear associations between referral delay and changes in Anxiety (GAD-2: r = − 0.017) and physical capacity (STS-1: r = − 0.015) were negligible. These results suggest that while no simple linear relationship exists between referral timing and clinical change, the exploratory non-linear analysis reveals a more complex, time-sensitive association between referral delay and patient outcomes.
Non-linear associations between referral-to-assessment time and outcomes
RCS modelling showed non-linear associations between referral delay and outcome change, not detected by linear correlation (Fig. 3). Functional impairment (WSAS) and depression (PHQ-2) followed a U-shaped trajectory: the greatest magnitude of improvement — reflected in the most negative change scores — was observed in patients assessed between 100 and 150 days after referral. Beyond this window, improvement declined progressively, with predicted change scores trending back towards zero for delays exceeding 200 days. A similar pattern was observed in the EQ-5D-5 L usual activities domain. Fatigue (FSS) and physical capacity (STS-1) followed divergent trajectories. The FSS spline trended downward with increasing delay, indicating progressively greater reductions in fatigue with longer waits, while the STS-1 curve rose markedly beyond 200 days. These patterns may reflect natural recovery or survivorship bias—patients who remain in rehabilitation after delays might be a healthier subgroup, not necessarily due to the delay. Conversely, breathlessness (MRC) and anxiety (GAD-2) showed flat splines with narrow confidence intervals across referral times, suggesting that the magnitude of observed improvement for these outcomes was consistent regardless of assessment timing. Confidence Intervals widened at both extremes of the x-axis (< 50 and > 300 days) due to fewer patients at very short or long delays and greater statistical uncertainty.
Discussion
This retrospective cohort study assessed the real-world outcomes associated with a structured multidisciplinary rehabilitation programme for Long COVID among 170 patients at a secondary care centre in South London. Three main findings emerge. First, rehabilitation led to significant, meaningful improvements in functional, physical, psychological, and HRQoL domains, especially in physical endurance and breathlessness. Second, MCID confirmed many patients achieved meaningful gains, mainly in physical capacity and mental health, though fatigue, self-care, and pain were less responsive. Third, no clear linear link between referral delay and outcome was found; RCS modelling showed non-linear patterns, hinting at a therapeutic window for improvements, which decreased after about 150 days.
The significant improvement in physical endurance, shown by the largest effect size in the STS-1 test, indicates that deconditioning — not permanent organ damage — mainly causes physical limitations in Long COVID and can be addressed through structured intervention. This interpretation aligns with evidence linking autonomic dysfunction, skeletal muscle mitochondrial impairment, and post-exertional energy dysregulation as reversible causes of functional decline [4, 29]. The substantial improvement in breathlessness supports respiratory physiotherapy and graded activity in multidisciplinary care, aligning with Gloëckl et al.’s findings of improved exercise capacity and dyspnoea after pulmonary rehabilitation in post-COVID patients [10]. Similarly, the significant improvement in functional impairment (WSAS) aligns with findings from Humphreys et al., who identified structured activity engagement as a key driver of occupational reintegration in Long COVID [11]. The more modest response to fatigue (FSS), despite statistical significance, is consistent with the broader literature on post-viral fatigue syndromes, in which fatigue frequently persists beyond the period of structured intervention and may require adjunctive pharmacological, nutritional, or cognitive-behavioural approaches [30]. Only 27.1% of patients met the FSS MCID, suggesting current rehabilitation models may need fatigue-specific strategies beyond graded activity and pacing. While the overall cohort demonstrated significant improvements, the inherent heterogeneity of Long COVID necessitates a cautious interpretation. A critical consideration in this population is the presence of PEM, a hallmark feature where physical, cognitive, or emotional exertion triggers a disproportionate worsening of symptoms [9, 30]. For the PEM-predominant phenotype, traditional aerobic exercise models—originally extrapolated from other post-viral conditions—may be contraindicated and could potentially lead to clinical relapse. Our findings, particularly the more modest response observed in the fatigue domain, underscore the need for rehabilitation to transition from standardised, time-based protocols toward symptom-contingent pacing. This approach prioritises energy conservation and ensures that functional gains in endurance do not occur at the expense of metabolic stability.
The moderate but statistically significant improvements in depression and anxiety are notable, given the high prevalence of psychological comorbidity in Long COVID, estimated at 20–40% across cohort studies [5, 31]. These gains likely reflect an indirect pathway: improved physical capacity and re-engagement with daily activities may have driven psychological recovery, rather than direct psychological intervention, a mechanism previously described in cardiac rehabilitation literature [32]. The failure of pain/discomfort and self-care domains to meet thresholds aligns with evidence that these dimensions represent entrenched somatic symptoms, possibly due to central sensitisation and altered pain processing, that are less responsive to standard exercise rehabilitation [33, 34]. The lack of a clear link between delay and outcome suggests that rehabilitation benefits aren’t solely a function of timing. This is partly reassuring in resource-limited healthcare settings where waiting times are common. However, non-linear RCS modelling offers a nuanced view: a potential therapeutic window exists for functional and psychological domains, with the greatest improvement in patients evaluated within 100–150 days of referral and diminishing returns beyond that. This pattern aligns with the principle that neuroinflammatory and autonomic dysregulatory changes in Long COVID may become increasingly entrenched with prolonged symptom duration, reducing the modifiability of symptoms over time [15, 35]. The upward STS-1 trajectory at longer delays likely reflects survivorship bias—patients remaining after six or more months may be more resilient, not a benefit of late assessment. These findings are the first to provide a non-linear analysis of referral delay effects in a structured Long COVID rehabilitation cohort using validated tools.
Strengths and limitations
A key strength of this study is the use of established MCID thresholds across a wide range of validated instruments. By moving beyond simple statistical significance (p-values), the MCID analysis confirms that the improvements observed in physical capacity, breathlessness, and depression represent tangible, clinically meaningful gains for a substantial proportion of participants. However, the study is limited by the absence of a control group, which limits conclusions: improvements from baseline to discharge can’t be attributed solely to rehabilitation, as natural recovery and regression to the mean may also contribute. However, consistent improvements across domains, effect sizes, and alignment with controlled trials support validity. Furthermore, this study is subject to significant “completer” or “survivorship” bias. The analysis was based solely on the 170 participants who completed the full 12-week programme and both assessments, representing a subset of the 673 distinct patients referred. It is probable that these individuals were the most motivated, possessed higher health literacy, or had symptom profiles that allowed for consistent engagement. Therefore, these results should not be overgeneralised as a definitive measure of programme-observed improvement for the entire Long COVID population, as the outcomes for those who discontinued the programme or were lost to follow-up remain unknown. Accordingly, we describe these findings as “associated with” clinical improvements rather than as evidence of definitive causal effectiveness. Reliance on self-reported outcomes (WSAS, FSS, PHQ-2, and EQ-5D-5L) risks response bias, especially in populations with fluctuating symptoms. Key confounders like vaccination status, COVID-19 severity, comorbidities, and treatments weren’t systematically recorded, and referral delay analysis may be biased by differences between early and late referrers. A single-centre design limits generalisability. Despite these limitations, the use of a large clinical cohort, validated multidomain instruments, and the analysis of temporal effects support the robustness of the core findings.
Clinical implications and future research
These findings have direct implications for clinical practice and health policy, providing evidence for the funding and structuring of multidisciplinary Long COVID rehabilitation within national care pathways. For physiotherapists and exercise physiologists, the results emphasise that while graded physical activity and respiratory physiotherapy yield consistent gains in physical endurance and breathlessness, these interventions should be integrated with pacing and autonomic conditioning to manage post-exertional malaise effectively. The optimal modality of rehabilitation for Long COVID remains a subject of intense debate within the clinical community. The shift from GET toward ‘pacing’ and ‘autonomic conditioning’ represents an evolution in our understanding of the condition’s pathophysiology. Our multidisciplinary model reflects this transition by integrating physiotherapy with occupational therapy-led pacing strategies. This individualised, symptom-contingent approach aims to provide a safe middle ground: allowing for gradual reconditioning in patients who tolerate it, while enforcing strict activity limits for those with significant autonomic dysregulation or PEM. Clinicians should, therefore, employ rigorous screening to distinguish between these phenotypes, as the ‘one-size-fits-all’ exercise prescription is no longer supported by the emerging evidence base.
Observed clinical outcomes suggest that while graded physical activity and respiratory physiotherapy are associated with consistent, significant gains in physical endurance and breathlessness, persistent issues with pain management and self-care indicate that protocols should be supplemented with targeted pain strategies, occupational therapy, and self-management education for multisystem patients. Furthermore, the identification of a non-linear therapeutic window suggests that services should prioritise early triage and referral within three to five months to maximise recovery potential, although meaningful improvements remain achievable for those with later presentations.
Future care models should continue to refine these distinctions through randomised controlled trials with long-term follow-up. Investigations should specifically focus on phenotype-specific rehabilitation protocols, distinguishing between patients with predominant postural tachycardia syndrome (PoTS), respiratory dysfunction, or cognitive impairment, and examine the safety and efficacy of low-intensity resistance training versus aerobic conditioning across these distinct symptom clusters. Finally, further prospective and controlled studies incorporating formal economic evaluations are needed to evaluate long-term sustainability and the cost-effectiveness of multidisciplinary care models.
Conclusions
In conclusion, this study evaluated the clinical and patient-reported outcomes of structured multidisciplinary Long COVID rehabilitation and determined whether referral-to-assessment delays affected clinical outcomes. Across functional, physical, psychological, and health-related QoL domains, rehabilitation was associated with statistically significant improvements in most outcome measures, with the greatest gains in physical endurance, breathlessness, and mental health. Fatigue, pain, and self-care proved less responsive, indicating that current rehabilitation models do not yet fully address the heterogeneous symptom burden of Long COVID. Exploratory non-linear modelling suggested that the magnitude of functional and psychological improvement may fluctuate depending on the timing of the assessment. Specifically, these findings indicate a potential therapeutic window, as gains appeared to diminish when referral-to-assessment times exceeded approximately five months. These findings support multidisciplinary Long COVID rehabilitation within national care pathways and highlight the need for targeted interventions for patients with refractory fatigue, pain, and self-care issues.
Supplementary Information
References
1.National Academies of Sciences, Engineering, and MedicineLong-term health effects of COVID-19: disability and function following SARS-CoV-2 infection 2024 Washington, DCNational Academies10.17226/27756 39312610 National Academies of Sciences, Engineering, and Medicine. Long-term health effects of COVID-19: disability and function following SARS-CoV-2 infection. Washington, DC: National Academies; 2024. 10.17226/27756 .39312610
2.Hou Y Gu T Ni Z Shi X Ranney ML Mukherjee B Global prevalence of long COVID, its subtypes, and risk factors: an updated systematic review and meta-analysis Open Forum Infect Dis 2025 12 ofaf533 10.1093/ofid/ofaf533 41018705 PMC12461872 Hou Y, Gu T, Ni Z, Shi X, Ranney ML, Mukherjee B. Global prevalence of long COVID, its subtypes, and risk factors: an updated systematic review and meta-analysis. Open Forum Infect Dis. 2025;12:9:ofaf533. 10.1093/ofid/ofaf533 .PMC12461872 41018705
3.Al-Aly Z Davis H McCorkell L Soares L Wulf-Hanson S Iwasaki A Long COVID science, research and policy Nat Med 2024 30 2148 64 10.1038/s41591-024-03173-6 39122965 Al-Aly Z, Davis H, McCorkell L, Soares L, Wulf-Hanson S, Iwasaki A, et al. Long COVID science, research and policy. Nat Med. 2024;30:2148–64. 10.1038/s41591-024-03173-6 .39122965
4.Greenhalgh T Knight M A’Court C Buxton M Husain L Management of post-acute COVID-19 in primary care BMJ 2020 370 m3026 10.1136/bmj.m3026 32784198 Greenhalgh T, Knight M, A’Court C, Buxton M, Husain L. Management of post-acute COVID-19 in primary care. BMJ. 2020;370:m3026. 10.1136/bmj.m3026 .32784198
5.Nalbandian A Sehgal K Gupta A Madhavan MV McGroder C Stevens JS Post-acute COVID-19 syndrome Nat Med 2021 27 601 15 10.1038/s41591-021-01283-z 33753937 PMC8893149 Nalbandian A, Sehgal K, Gupta A, Madhavan MV, McGroder C, Stevens JS, et al. Post-acute COVID-19 syndrome. Nat Med. 2021;27:601–15. 10.1038/s41591-021-01283-z .PMC8893149 33753937
6.Venkatesan P NICE guideline on long COVID Lancet Respir Med 2021 9 129 10.1016/S2213-2600(21)00031-X 33453162 PMC7832375 Venkatesan P. NICE guideline on long COVID. Lancet Respir Med. 2021;9:129. 10.1016/S2213-2600(21)00031-X .PMC7832375 33453162
7.Office for National Statistics, UK. Self-reported coronavirus (COVID-19) infections and associated symptoms, England and Scotland: November 2023 to March 2024. https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/articles/selfreportedcoronaviruscovid19infectionsandassociatedsymptomsenglandandscotland/november2023tomarch2024. Accessed 17 Jun 2026.
8.National Institute for Health and Care Excellence. Overview | COVID-19 rapid guideline: managing COVID-19 | Guidance | NICE. https://www.nice.org.uk/guidance/ng191. Accessed 20 Feb 2026.
9.Gloeckl R, Zwick RH, Fürlinger U, Schneeberger T, Leitl D, Jarosch I, et al. Practical recommendations for exercise training in patients with long COVID with or without post-exertional malaise: a best practice proposal. Sports Med Open. 2024;10. 10.1186/s40798-024-00695-8 .PMC11043268 38658496
10.Gloeckl R Leitl D Jarosch I Schneeberger T Nell C Stenzel N Benefits of pulmonary rehabilitation in COVID-19: a prospective observational cohort study ERJ Open Res 2021 7 108 10.1183/23120541.00108-2021 PMC7957293 34095290 Gloeckl R, Leitl D, Jarosch I, Schneeberger T, Nell C, Stenzel N, et al. Benefits of pulmonary rehabilitation in COVID-19: a prospective observational cohort study. ERJ Open Res. 2021;7:108. 10.1183/23120541.00108-2021 .PMC7957293 34095290
11.Humphreys H Kilby L Kudiersky N Copeland R Long COVID and the role of physical activity: a qualitative study BMJ Open 2021 11 e047632 10.1136/bmjopen-2020-047632 33692189 PMC7948149 Humphreys H, Kilby L, Kudiersky N, Copeland R. Long COVID and the role of physical activity: a qualitative study. BMJ Open. 2021;11:e047632. 10.1136/bmjopen-2020-047632 .PMC7948149 33692189
12.Klok FA Boon GJAM Barco S Endres M Geelhoed JJM Knauss S The post-COVID-19 functional status scale: a tool to measure functional status over time after COVID-19 Eur Respir J 2020 56 2001494 10.1183/13993003.01494-2020 32398306 PMC7236834 Klok FA, Boon GJAM, Barco S, Endres M, Geelhoed JJM, Knauss S, et al. The post-COVID-19 functional status scale: a tool to measure functional status over time after COVID-19. Eur Respir J. 2020;56:2001494. 10.1183/13993003.01494-2020 .PMC7236834 32398306
13.Moss-Morris R Weinman J Petrie K Horne R Cameron L Buick D The Revised Illness Perception Questionnaire (IPQ-R) Psychol Health 2002 17 1 16 10.1080/08870440290001494 Moss-Morris R, Weinman J, Petrie K, Horne R, Cameron L, Buick D. The Revised Illness Perception Questionnaire (IPQ-R). Psychol Health. 2002;17:1–16. 10.1080/08870440290001494 .
14.Stucki G Stier-Jarmer M Grill E Melvin J Rationale and principles of early rehabilitation care after an acute injury or illness Disabil Rehabil 2005 27 353 9 10.1080/09638280400014105 16040536 Stucki G, Stier-Jarmer M, Grill E, Melvin J. Rationale and principles of early rehabilitation care after an acute injury or illness. Disabil Rehabil. 2005;27:353–9. 10.1080/09638280400014105 .16040536
15.Barker-Davies RM O’Sullivan O Senaratne KPP Baker P Cranley M Dharm-Datta S The Stanford Hall consensus statement for post-COVID-19 rehabilitation Br J Sports Med 2020 54 949 59 10.1136/bjsports-2020-102596 32475821 PMC7418628 Barker-Davies RM, O’Sullivan O, Senaratne KPP, Baker P, Cranley M, Dharm-Datta S, et al. The Stanford Hall consensus statement for post-COVID-19 rehabilitation. Br J Sports Med. 2020;54:949–59. 10.1136/bjsports-2020-102596 .PMC7418628 32475821
16.Alwan NA A negative COVID-19 test does not mean recovery Nature 2020 584 170 170 10.1038/d41586-020-02335-z 32782377 Alwan NA. A negative COVID-19 test does not mean recovery. Nature. 2020;584:170–170. 10.1038/d41586-020-02335-z .
17.Al-Husinat L Nusir M Al-Gharaibeh H Alomari AA Smadi MM Battaglini D Pelosi P Post-COVID-19 syndrome symptoms after mild and moderate SARS-CoV-2 infection Front Med (Lausanne) 2022 3 9 1017257 10.3389/fmed.2022.1017257 PMC9573938 36262270 Al-Husinat L, Nusir M, Al-Gharaibeh H, Alomari AA, Smadi MM, Battaglini D, Pelosi P. Post-COVID-19 syndrome symptoms after mild and moderate SARS-CoV-2 infection. Front Med (Lausanne). 2022;3(9):1017257. 10.3389/fmed.2022.1017257 .PMC9573938 36262270
18.Mundt JC Marks IM Shear MK Greist JH The work and social adjustment scale: a simple measure of impairment in functioning Br J Psychiatry 2002 180 461 4 10.1192/bjp.180.5.461 11983645 Mundt JC, Marks IM, Shear MK, Greist JH. The work and social adjustment scale: a simple measure of impairment in functioning. Br J Psychiatry. 2002;180:461–4. 10.1192/bjp.180.5.461 .11983645
19.Krupp LB LaRocca NG Muir-Nash J Steinberg AD The fatigue severity scale. Application to patients with multiple sclerosis and systemic lupus erythematosus Arch Neurol 1989 46 1121 3 10.1001/archneur.1989.00520460115022 2803071 Krupp LB, LaRocca NG, Muir-Nash J, Steinberg AD. The fatigue severity scale. Application to patients with multiple sclerosis and systemic lupus erythematosus. Arch Neurol. 1989;46:1121–3. 10.1001/archneur.1989.00520460115022 .2803071
20.Bestall JC Paul EA Garrod R Garnham R Jones PW Wedzicha JA Usefulness of the Medical Research Council (MRC) dyspnoea scale as a measure of disability in patients with chronic obstructive pulmonary disease Thorax 1999 54 581 6 10.1136/thx.54.7.581 10377201 PMC1745516 Bestall JC, Paul EA, Garrod R, Garnham R, Jones PW, Wedzicha JA. Usefulness of the Medical Research Council (MRC) dyspnoea scale as a measure of disability in patients with chronic obstructive pulmonary disease. Thorax. 1999;54:581–6. 10.1136/thx.54.7.581 .PMC1745516 10377201
21.Kroenke K Spitzer RL Williams JBW The Patient Health Questionnaire-2: validity of a two-item depression screener Med Care 2003 41 1284 92 10.1097/01.MLR.0000093487.78664.3C 14583691 Kroenke K, Spitzer RL, Williams JBW. The Patient Health Questionnaire-2: validity of a two-item depression screener. Med Care. 2003;41:1284–92. 10.1097/01.MLR.0000093487.78664.3C .14583691
22.Sapra A Bhandari P Sharma S Chanpura T Lopp L Using Generalised Anxiety Disorder-2 (GAD-2) and GAD-7 in a Primary Care Setting Cureus 2020 12 5 e8224 10.7759/cureus.8224 32582485 PMC7306644 Sapra A, Bhandari P, Sharma S, Chanpura T, Lopp L. Using Generalised Anxiety Disorder-2 (GAD-2) and GAD-7 in a Primary Care Setting. Cureus. 2020;12(5):e8224. 10.7759/cureus.8224 .PMC7306644 32582485
23.Pickard AS Neary MP Cella D Estimation of minimally important differences in EQ-5D utility and VAS scores in cancer Health Qual Life Outcomes 2007 5 70 10.1186/1477-7525-5-70 18154669 PMC2248572 Pickard AS, Neary MP, Cella D. Estimation of minimally important differences in EQ-5D utility and VAS scores in cancer. Health Qual Life Outcomes. 2007;5:70. 10.1186/1477-7525-5-70 .PMC2248572 18154669
24.Feng Y Parkin D Devlin NJ Assessing the performance of the EQ-VAS in the NHS PROMs programme Qual Life Res 2014 23 3 977 89 10.1007/s11136-013-0537-z 24081873 PMC4287662 Feng Y, Parkin D, Devlin NJ. Assessing the performance of the EQ-VAS in the NHS PROMs programme. Qual Life Res. 2014;23(3):977–89. 10.1007/s11136-013-0537-z .PMC4287662 24081873
25.Bohannon RW Hand-grip dynamometry predicts future outcomes in ageing adults J Geriatr Phys Ther 2008 31 3 10 10.1519/00139143-200831010-00002 18489802 Bohannon RW. Hand-grip dynamometry predicts future outcomes in ageing adults. J Geriatr Phys Ther. 2008;31:3–10. 10.1519/00139143-200831010-00002 .18489802
26.Chen X Robinson DG Storey JD The functional false discovery rate with applications to genomics Biostatistics 2019 22 68 81 10.1093/biostatistics/kxz010 PMC7846131 31135886 Chen X, Robinson DG, Storey JD. The functional false discovery rate with applications to genomics. Biostatistics. 2019;22:68–81. 10.1093/biostatistics/kxz010 .PMC7846131 31135886
27.Rouam S False Discovery Rate (FDR), Encyclopedia of Systems Biology 2013 New York, NYSpringer731 2 Rouam S. False Discovery Rate (FDR), Encyclopedia of Systems Biology. New York, NY: Springer; 2013. pp. 731–2.
28.Harrell FE. Regression Modelling Strategies. Springer Series in Statistics. 2015. 10.1007/978-3-319-19425-7 .
29.Davis HE McCorkell L Vogel JM Topol EJ Long COVID major findings, 439 mechanisms and recommendations Nat Rev Microbiol 2023 21 440 133 46 10.1038/s41579-022-00846-2 36639608 PMC9839201 Davis HE, McCorkell L, Vogel JM, Topol EJ, Long COVID. major findings, 439 mechanisms and recommendations. Nat Rev Microbiol. 2023;21(440):133–46. 10.1038/s41579-022-00846-2 .PMC9839201 36639608
30.Islam MF Cotler J Jason LA Post-viral fatigue and COVID-19: lessons from past epidemics Fatigue: Biomed Health Behav 2020 8 61 9 10.1080/21641846.2020.1778227 Islam MF, Cotler J, Jason LA. Post-viral fatigue and COVID-19: lessons from past epidemics. Fatigue Biomed Health Behav. 2020;8:61–9. 10.1080/21641846.2020.1778227 .
31.Michelen M Manoharan L Elkheir N Cheng V Dagens A Hastie C Characterising long COVID: a living systematic review BMJ Glob Health 2021 6 459 10.1136/bmjgh-2021-005427 PMC8478580 34580069 Michelen M, Manoharan L, Elkheir N, Cheng V, Dagens A, Hastie C, et al. Characterising long COVID: a living systematic review. BMJ Glob Health. 2021;6:459. 10.1136/bmjgh-2021-005427 .PMC8478580 34580069
32.Lavie CJ Ozemek C Carbone S Katzmarzyk PT Blair SN Sedentary behaviour, exercise, and cardiovascular health Circ Res 2019 124 799 815 10.1161/CIRCRESAHA.118.312669 30817262 Lavie CJ, Ozemek C, Carbone S, Katzmarzyk PT, Blair SN. Sedentary behaviour, exercise, and cardiovascular health. Circ Res. 2019;124:799–815. 10.1161/CIRCRESAHA.118.312669 .30817262
33.Thaweethai T Jolley SE Karlson EW Levitan EB Levy B McComsey GA Development of a definition of postacute sequelae of SARS-CoV-2 infection JAMA 2023 329 1934 46 10.1001/jama.2023.8823 37278994 PMC10214179 Thaweethai T, Jolley SE, Karlson EW, Levitan EB, Levy B, McComsey GA, et al. Development of a definition of postacute sequelae of SARS-CoV-2 infection. JAMA. 2023;329:1934–46. 10.1001/jama.2023.8823 .PMC10214179 37278994
34.Clauw DJ Calabrese L Rheumatology and Long COVID: lessons from the study of fibromyalgia Ann Rheum Dis 2024 83 136 8 10.1136/ard-2023-224250 37230736 PMC10850638 Clauw DJ, Calabrese L. Rheumatology and Long COVID: lessons from the study of fibromyalgia. Ann Rheum Dis. 2024;83:136–8. 10.1136/ard-2023-224250 .PMC10850638 37230736
35.Klein J Wood J Jaycox JR Dhodapkar RM Lu P Gehlhausen JR Distinguishing features of long COVID identified through immune profiling Nature 2023 623 139 48 10.1038/s41586-023-06651-y 37748514 PMC10620090 Klein J, Wood J, Jaycox JR, Dhodapkar RM, Lu P, Gehlhausen JR, et al. Distinguishing features of long COVID identified through immune profiling. Nature. 2023;623:139–48. 10.1038/s41586-023-06651-y .PMC10620090 37748514
Come citare questo studio
Daodu LP, Nottingham K, Raste Y, Allgrove JE, Arrigoni FIF, Kayyali R. (2026). Real-world outcomes of a multidisciplinary rehabilitation programme for adults with long COVID: a retrospective cohort study.. BMC public health. DOI: 10.1186/s12889-026-29023-3
Consulta la pubblicazione originale ↗