Shifting focus to phenotypes: the evolution of long COVID rehabilitation

Extract

The “rehabilitation” face of long COVID has changed. In 2020, “post-COVID sequelae” were largely recognised in populations after severe acute illness involving intensive-care-acquired weakness and organ impairment [1, 2]. Foundational interim guidance focused on adapting established pulmonary rehabilitation for these hospitalised cohorts [3]. Meanwhile, the millions of people with long COVID in the non-hospitalised population were largely hidden from healthcare and research studies [4].

Shareable abstract

Modern long COVID rehabilitation must go beyond improving exercise capacity. It must serve as behavioural and physiological mediator enabling the “adaptive middle ground”: a precise zone of symptom-titrated activity to navigate sustainable recovery. https://bit.ly/4nXML2v

Long COVID phenotypes

The “rehabilitation” face of long COVID has changed. In 2020, “post-COVID sequelae” were largely recognised in populations after severe acute illness involving intensive-care-acquired weakness and organ impairment [1, 2]. Foundational interim guidance focused on adapting established pulmonary rehabilitation for these hospitalised cohorts [3]. Meanwhile, the millions of people with long COVID in the non-hospitalised population were largely hidden from healthcare and research studies [4].

Over the last 5 years, clinical services for long COVID increasingly see patients who after mild acute infection experience a multisystem “long tail” of complex symptoms including fatigue, breathlessness, pain, cognitive dysfunction, dysautonomia, and post-exertional symptom exacerbation (PESE) (also referred to as post-exertional malaise (PEM). The precise prevalence of PESE is unknown, but one meta-analysis estimated a pooled prevalence of 25% (95% CI 17–36%; 10 studies, n=4076) in adults with long COVID, though the certainty of the evidence remains low due to the heterogeneous assessment methods used [5]. These patients often remain “invisible” to traditional diagnostic pathways because standard clinical investigations (e.g. pulmonary function tests) are frequently normal despite profound disability. There are different phenotypes of long COVID in terms of recognisable patterns of symptom presentation, functional impact, and behavioural coping [6], and phenotypic rehabilitation approaches are needed [7].

The evolution in patient presentation to healthcare services requires an evolution in our therapeutic approaches. Exercise-based rehabilitation can improve short-term exercise capacity after COVID-19 hospitalisation, whether delivered face-to-face or remotely [8]. However, the evidence regarding quality of life and symptom burden following rehabilitation programmes is mixed. Some studies have reported meaningful improvements in perceived health status and fatigue [9], while others have found no significant group-level change [10]. The inconsistency may reflect differences in intervention design, but also methodological challenge of measuring quality of life in a multisystem condition, as generic instruments (e.g. EQ-5D and SF-36) may not capture the nuanced symptom fluctuations that are central to long COVID recovery. The improvements in exercise capacity may not necessarily lead to recovery of symptoms and programme content needs to be further tailored to the evolving long COVID phenotypes, in an era where the majority are non-hospitalised.

Short-term wins but long-term needs

The systematic review and meta-analysis by Chakraverty et al. [11] in this issue of ERJ Open Research provides an essential empirical foundation for long COVID rehabilitation. Their analysis of >1400 adults (including both hospitalised and non-hospitalised cohorts) from 12 studies demonstrates that structured, multidisciplinary rehabilitation yields substantial short-term gains. Patients achieved clinically meaningful improvements, most notably an average increase of 102 m in the 6-min walk distance, alongside significant reductions in dyspnoea and anxiety scores. In addition to these improvements in clinical outcomes, molecular science may be revealing the biological “why” behind these successes. The benefits of rehabilitation may extend beyond the cardiovascular and muscular systems, as individually titrated exercise training in adults with long COVID may drive beneficial changes in gene expression, positively reprogramming immune and metabolic responses to physiological stress [8, 12].

Despite these short-term benefits, Chakraverty et al. [11] also highlight a critical gap in current evidence: educational content and delivery varied widely across studies, and its contribution to outcomes, particularly long-term sustainability, remains unclear. While physical interventions clearly boost short-term capacity, the role of education as an active therapeutic ingredient has not been systematically evaluated. Without standardised educational content that teaches patients to recognise physiological limits and apply pacing strategies to better manage PESE/PEM, there is a risk that functional gains will diminish once supervised programmes end. Importantly, long-term effects remain limited and inconsistent across the reviewed studies, underscoring an urgent need for follow-up trials designed to assess sustained effects.

This challenge is not new to rehabilitation science. For example, in patients with COPD, exercise-based pulmonary rehabilitation reliably improves short-term exercise capacity, but benefits commonly diminish within 12 months [13]. Future trials of long COVID rehabilitation must learn from these experiences and move beyond demonstrating short-term efficacy to be specifically designed with long-term follow-up (≥12 months) and include pre-stratification by PEM severity, standardised patient-centred outcome measures, and behavioural endpoints alongside physical ones. It is essential to embed self-management education and maintenance strategies from the very beginning, rather than treating supervised exercise as the endpoint. Tele-rehabilitation offers a promising platform for precisely this transition.

Tele-rehabilitation: feasibility and the challenge of progression

For long COVID, tele-rehabilitation potentially improves access as it addresses an important physiological issue; for patients with significant PESE/PEM, the physical and cognitive demands of travelling to a clinic, waiting in crowded spaces and recovering from an in-person session may be enough to trigger PESE/PEM. Tele-rehabilitation eliminates this challenge entirely by enabling precise, home-based titration of activity within each patient's energy envelope. The study by Ferreira et al. [14], also in this issue, addresses this need by exploring tele-rehabilitation for a more contemporary cohort where 78% were never hospitalised. The trial randomised 132 participants to either an individualised 8-week tele-rehabilitation programme and usual care, or supplemented usual care. Both groups showed comparable changes in the primary endpoint (functional mobility assessed by a questionnaire). However, the intervention group showed improvements in several secondary outcomes, including fatigue, dyspnoea, pain/discomfort, and perceived health status. All these secondary outcomes were patient-reported, which is particularly meaningful in a population whose profound functional disability is frequently undetectable by standard objective investigations such as pulmonary function or exercise tests. Adherence was high, with 96% of scheduled sessions attended. Further, no serious adverse events occurred, which is a critical safety signal. Importantly, 39% of intervention participants were unable to progress their exercise prescription through one or more FITT (frequency, intensity, time and type) principles due to symptom burden. These findings underscore both the feasibility and the challenges of tele-rehabilitation in a predominantly non-hospitalised cohort with persistent symptoms including PESE/PEM.

Ferreira et al. [14] utilised the DePaul Symptom Questionnaire (DSQ-PEM) to screen for PEM, yet the reliability of this tool remains a subject of intense debate. Recent studies [15, 16] provided a sobering critique that while the tool is sensitive, it is not very specific, leading to a high rate of potential “false positives” or mischaracterisations. Clinicians must therefore use the DSQ-PEM not as a definitive diagnostic tool, but as a starting point for clinical dialogue. A best-practice proposal for long COVID exercise training recommends stratifying prescriptions by PEM severity (none versus mild/moderate versus severe) using PEM screening (assessed by a clinical interview combined with the DSQ-PEM) [17]. The proposal also recommends shifting the focus toward pacing/energy management when more severe PEM is prevalent.

A precision rehabilitation approach must also be mindful of patients with a manifest diagnosis of myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) whereby PESE/PEM is a shared hallmark [18]. However, the severity and physiological thresholds often differ between long COVID and ME/CFS, with long COVID patients without an ME/CFS diagnosis experiencing less pronounced PEM episodes [18]. For patients at the severe end of the spectrum, clinicians should emphasise autonomic stabilisation and strict energy conservation before considering any increase in physical activity.

Looking beyond the treadmill: navigating the “adaptive middle ground”

The next frontier of long COVID recovery is not solely focused on increasing exercise capacity, but involves refining the behavioural and digital delivery of rehabilitation. Both Chakraverty et al. [11] and Ferreira et al. [14] demonstrate that whilst multidisciplinary rehabilitation can improve short-term outcomes, translating these gains into sustained recovery requires addressing the specific behavioural coping strategies and symptom phenotypes that many patients present with.

The avoidance-endurance challenge

According to the avoidance-endurance model, a biopsychosocial framework [19], chronic disability often arises from two maladaptive (counterproductive) behavioural extremes:

• Fear–avoidance behaviour: driven by the anticipation of a “crash”, these patients remain inactive, triggering a downward spiral of secondary deconditioning.

• Endurance-related “warrior” behaviour: these patients ignore physiological warning signs in order to maintain their pre-illness identity, often leading to PESE/PEM.

In practice, a patient with a fear–avoidance phenotype may require careful reintroduction to activity with explicit cognitive reassurance to decouple movement from anticipated harm. In contrast, a patient with an endurance-related coping phenotype requires structured activity ceilings, pacing education, and physiological feedback to override the drive to push through symptoms. Therefore, both phenotypes require fundamentally different therapeutic conversations.

A large German multicentre study by Kupferschmitt et al. [20] (n=721) revealed that 68% of adults with long COVID enter rehabilitation with one of these maladaptive coping strategies: 22.3% with fear–avoidance behaviour, and 45.5% with endurance-related behaviour. However, these phenotypes are not fixed. In the same study, rehabilitation decreased endurance-related behaviours in 32% and fear–avoidance behaviours in 37% of the individuals [20].

The Behavioural Activity Cockpit

To navigate the “long tail” of recovery, education should serve as the primary stabiliser. Therefore, we propose the Behavioural Activity Cockpit (figure 1), a practical framework for this education, to empower patients to find the right balance between pacing strategies and traditional exercise and to steer away from maladaptive “warrior” or “avoidance” traps. Within this model, the patient acts as the driver, while clinicians and therapists serve as co-pilots who guide the journey rather than dictating the pace.

FIGURE 1.

The Behavioural Activity Cockpit for long COVID rehabilitation. This conceptual framework illustrates the dual-monitoring required for sustainable recovery. Gauge (a) (physical activity level) represents the physiological spectrum, navigating between energy conservation (pacing) and traditional symptom-titrated exercise. Gauge (b) (coping strategy) visualises the behavioral coping strategies. Red zones indicate maladaptive extremes: fear–avoidance (leading to secondary deconditioning) and the “warrior” mentality (risking post-exertional symptom exacerbation (PESE), post-exertional malaise (PEM) or crashes). The “adaptive middle ground” (green zone) represents the therapeutic window where recovery is maximised.

By integrating tele-rehabilitation, this framework can be delivered directly into the patients’ home, facilitating a seamless transition from supervised intervention to sustainable, self-managed pacing and activity. This could ensure that functional gains are not temporary spikes in performance, but lasting improvements built on a foundation of patient safety and self-literacy.

Nevertheless, the integration of tele-rehabilitation must explicitly address health inequalities. Unlike older rehabilitation populations, such as those with COPD, long COVID predominantly affects working-age adults, who generally have a higher level of digital literacy. However, the digital divide in long COVID is not primarily generational: lower socioeconomic status, culture, language, and severity of cognitive impairment need additional focus to achieve equitable access [21]. To ensure that the benefits of tele-rehabilitation reach those who need them most, it is essential to implement hybrid models that combine in-person and remote components alongside targeted digital navigation support.

Furthermore, while organisations such as the World Health Organization and NICE (National Institute for Health and Care Excellence) have provided valuable “living” management guidance, there remains a notable absence of definitive, evidence-based clinical practice guidelines specifically around the multicomponent rehabilitation for the non-hospitalised long COVID cohorts. In the absence of such standardised frameworks, the mission of modern long COVID rehabilitation must go beyond improving exercise capacity, and serve as a behavioural and physiological mediator that equips patients to find the “adaptive middle ground”: a precise zone of symptom-titrated activity where recovery finally outpaces relapse.