STIMULATE-ICP: Largest Long COVID Trial Finds Specialist Care Reduces Fatigue, but Multi-Organ MRI Adds No Benefit

The largest interventional trial ever conducted for Long COVID (search)Search disease has delivered a mixed verdict on two widely deployed components of integrated care: multi-organ MRI scanning and digital rehabilitation. Published in Nature Medicine, the STIMULATE-ICP trial found that specialist Long COVID care itself drove meaningful reductions in fatigue, but adding pre-clinic multi-organ imaging conferred no additional benefit, while a digital rehabilitation app produced only a modest, delayed improvement of uncertain clinical significance.

The cluster-randomized, phase 3 trial enrolled 1,152 adults with Long COVID (search)Search disease—defined as persistent, otherwise unexplained post-COVID symptoms lasting four or more weeks—across six NHS Long COVID clinics in England between August 2022 and August 2024. Participants were allocated via 122 primary care network (PCN) clusters to one of four arms: multi-organ MRI (Coverscan) plus usual care, digital rehabilitation (Living With COVID Recovery app) plus usual care, both interventions, or usual care alone.

Specialist care drives fatigue reduction across all arms

The prespecified primary outcome—mean Fatigue Assessment Scale (FAS) score at 12 weeks—improved across all four groups. Changes from baseline were 3.9 points in the multi-organ MRI arm, 5.3 points in the digital rehabilitation arm, 4.4 points in the combined arm, and 4.5 points in the usual care arm. Overall, the trial observed a 12.6% reduction in FAS from a high baseline, a change the authors note is "commensurate with clinically meaningful reductions in respiratory conditions."

However, after adjusting for baseline FAS, gender, PCN size and deprivation, drug trial participation, and trial site, neither multi-organ MRI nor digital rehabilitation showed a statistically significant benefit over usual care at 12 weeks. Compared with usual care, estimated FAS at 12 weeks for multi-organ MRI, digital rehabilitation, and their interaction differed by 0.18 (95% CI −0.72, 1.09; P=0.691), −0.53 (95% CI −1.42, 0.36; P=0.245), and −0.17 (95% CI −1.06, 0.72; P=0.710), respectively.

"Although not designed to test the NHS ICP's efficacy, we suggest its direct impact," the authors write, noting that 59% of participants experienced a clinically important drop of at least 3 points in FAS by 12 weeks.

Digital rehabilitation: a signal at 24 weeks

At the 24-week secondary endpoint, digital rehabilitation demonstrated a statistically significant improvement in fatigue compared with usual care. The difference in FAS at 24 weeks for digital rehabilitation was −1.14 points (95% CI −2.12, −0.16; P=0.023). Quality of life, measured by EQ-VAS, also improved at 24 weeks with digital rehabilitation, showing a difference of 2.74 units (95% CI 0.07, 5.41; P=0.045).

The authors caution that "this observed difference between treatment groups of 1.14 points on FAS was statistically significant but may not be clinically meaningful," emphasizing that the finding is from a secondary outcome and "should be viewed as hypothesis-generating rather than definitive."

Multi-organ MRI: no incremental value

Of the 437 participants who received multi-organ MRI, 105 (24%) had site-reported abnormal findings. Yet regression analyses restricted to those with a Coverscan showed almost no difference in FAS at 12 and 24 weeks between those with abnormal and normal scans. "Our results do not support routine use of comprehensive multi-organ imaging in Long COVID (search)Search disease at present," the authors conclude, noting that "Coverscan abnormalities do not currently open different therapeutic options, which may change with increased understanding of Long COVID."

A pragmatic design under profound uncertainty

The trial's cluster-randomized design—in which entire NHS clinics, rather than individual patients, were assigned to treatment arms—was both a methodological necessity and a source of inherent limitation. Investigators and participants were unblinded, creating potential for performance and detection bias, particularly given that primary endpoints were patient-reported.

The study population reflected the real-world demographics of Long COVID (search)Search disease: mean age was 48.9 years, 66.4% were female, and 14.4% were from minority ethnic groups. Median symptom duration at enrollment was 394 days (IQR 238–717 days), and participants reported a median of nine Long COVID symptoms at baseline. Common comorbidities included depression (23.4%), allergy (19.7%), asthma (19.2%), and anxiety (14.1%).

The authors acknowledge several limitations, including inability to control for interactions between drug allocation and ICP components for participants co-enrolled in a nested drug trial, variability in intervention delivery across sites, and the challenge of designing a trial in 2021 without fully anticipating changes in viral variants, repeat infections, and vaccine deployment during the recruitment period.

Implications for practice and future research

"Overall, our findings support investment in accessible, specialist care for Long COVID (search)Search disease, and further assessment of the efficacy of digital rehabilitation," the authors write. They call for "future trials of ICP interventions of different designs in different settings" to determine optimal integrated care pathways.

The trial's safety profile was benign: four serious adverse events were reported among four participants in the ICP trial, none related to trial interventions.

Full evaluation of all ICP components, including clinical and health economic data, is expected over the next year, with additional secondary outcomes—including mental health measures—to be reported within 12–24 months.