The review, registered prospectively in PROSPERO under number CRD420261405899, set out to answer a question that previous literature had largely skirted. Earlier systematic reviews of stroke rehabilitation guideline implementation had concentrated on the quality of the guidelines themselves, on clinicians’ awareness of them, or on the barriers and facilitators professionals perceive when trying to follow them. What those reviews did not do was pool direct measurements of whether individual patients actually received recommended assessments, screenings, or therapy. The new analysis focused specifically on patient-level or chart-level adherence, excluding studies that reported only facility-level indicators, clinician opinions, or qualitative accounts of implementation barriers.

To find eligible studies, the researchers searched PubMed/MEDLINE, Scopus, Web of Science, ProQuest, and Google Scholar from database inception through September 2026, screening the first 200 Google Scholar records by relevance and manually checking the reference lists of included studies. From 923 records identified, 445 duplicates were removed, leaving 478 for title and abstract screening. After 358 were excluded as irrelevant, 120 full texts were assessed and 114 were excluded, most commonly because their data were unsuitable or not extractable at the patient level. Six studies ultimately met the inclusion criteria, all of them non-randomized by design requirement, and all restricted to English-language publications because translation resources were unavailable.

The six studies came from Qatar, Canada, and New Zealand and spanned inpatient, outpatient, and community rehabilitation settings. Ajimsha and colleagues audited 127 patient charts at an inpatient stroke rehabilitation centre in Qatar against the Physical Therapy After Acute Stroke guideline, finding overall physical therapy adherence of 70.9 percent, or 90 of 127 charts. McClure and colleagues reviewed 123 charts from a Canadian inpatient stroke rehabilitation facility and found that 102 patients, or 82.9 percent, had documented completion of a formal cognitive screening instrument such as the MoCA or MMSE, with modified or incomplete versions not counted as complete screens. That was the highest adherence estimate in the synthesis.

At the other end of the range, MacKenzie and colleagues examined 70 charts from a Canadian outpatient stroke rehabilitation clinic and found that not a single patient had undergone depression screening with a standardized tool before a standardized screening form was implemented, an adherence rate of 0.0 percent. Auger and colleagues reported that sexuality-related rehabilitation services were provided to only 5 of 15 patients, or 33.3 percent, at the baseline of a Canadian pilot implementation study in inpatient rehabilitation. Thompson and colleagues, conducting a prospective service audit of a community rehabilitation team in Wellington, New Zealand, found that 17 of 46 patients with complete timing data, or 37.0 percent, received face-to-face community rehabilitation contact within seven calendar days of hospital discharge, in line with New Zealand Stroke Network recommendations.

The sixth study, by Yan and colleagues, took a different methodological approach. Rather than relying on chart documentation, it used direct objective measurement during monitored physical therapy sessions across multiple Canadian inpatient rehabilitation sites. Of 166 medically stable adults assessed at the midpoint of their rehabilitation, only 30, or 18.1 percent, achieved at least 20 minutes of activity at aerobic intensity during the monitored session, the target set by stroke rehabilitation guidelines. The authors of the review note that this estimate reflects a single monitored session rather than performance across the entire rehabilitation episode, an important limitation when interpreting what the figure means for a patient’s full course of therapy.

Combined, the six studies contributed a denominator of 547 patients or patient charts, with individual sample sizes ranging from 15 to 166. Because the outcome was a single-group proportion and one study reported zero events, the analysts logit-transformed the proportions with a continuity correction of 0.5 added to event and non-event counts, then pooled them using a random-effects model estimated by restricted maximum likelihood in Stata version 17. The pooled logit estimate was −0.59 (95 percent CI −2.09 to 0.91), which back-transforms to the headline adherence rate of 35.6 percent. Heterogeneity was extreme: the between-study variance τ² was 3.21, I² was 97.6 percent, H² was 41.66, and Cochran’s Q test with five degrees of freedom yielded 136.46, p < 0.001. In practical terms, the included studies were not estimating a single uniform adherence construct at all.

The review’s authors are unusually explicit about what this heterogeneity means for interpretation. The six studies assessed six different care processes, physical therapy adherence, cognitive screening, depression screening, sexuality-related rehabilitation, timely community rehabilitation contact, and aerobic-intensity therapy, in different settings with different adherence definitions. The care-process domains were not selected a priori; they simply reflected what the eligible studies had reported. Consequently, the pooled estimate has limited clinical interpretability and should be treated as a broad implementation-performance indicator rather than a measure of adherence to any single recommendation. A leave-one-out sensitivity analysis underscored the instability: removing the McClure cognitive screening study lowered the pooled estimate to roughly 26.7 percent, while removing the MacKenzie zero-adherence study raised it to about 49.3 percent, with all confidence intervals remaining wide.

Methodological quality, appraised with the Joanna Briggs Institute checklist for studies reporting prevalence data, revealed further reasons for caution. Common concerns included single-centre sampling, reliance on retrospective chart documentation, incomplete or selectively available records, and small samples. Ajimsha and colleagues included only 127 of 216 initially identified records, raising questions about coverage, and Auger’s baseline estimate rested on just 15 patients. Thompson’s prospective audit used consecutive sampling, but complete timing data were available for only 46 of 50 participants. A visual inspection of the funnel plot suggested asymmetry, but with only six studies the authors stress that funnel plots and formal publication-bias assessments have limited reliability, and an exploratory Galbraith plot showed substantial dispersion consistent with the clinical and methodological diversity of the estimates.

What the analysis does establish, within its limits, is that patient-level delivery of guideline-concordant stroke rehabilitation care appears suboptimal and highly variable across the small body of eligible evidence. Adherence was relatively higher for established, routinely embedded processes such as cognitive screening and structured physical therapy, and lower for less embedded domains including depression screening, sexuality-related rehabilitation, timely community follow-up, and aerobic-intensity therapy, though the authors note this pattern is a hypothesis that future studies should test directly. The review did not evaluate whether adherence levels were associated with subsequent functional or recovery outcomes. The authors argue that the findings support routine patient-level audit of stroke rehabilitation care rather than reliance on the mere existence of guidelines, and they point to implementation strategies drawn from prior literature, standardized documentation templates, electronic prompts, audit and feedback, staff education, local clinical champions, and clear multidisciplinary referral pathways, as candidate approaches for closing the gap between what guidelines recommend and what patients actually receive.

Subject of Research: Patient-level adherence to guideline-recommended care processes in adult stroke rehabilitation

Article Title: Delivery of selected guideline-based care processes in adult stroke rehabilitation: a systematic review and meta-analysis

Article References: Aljohani, M. M. A., & Alhusayni, A. I. (2026). Delivery of selected guideline-based care processes in adult stroke rehabilitation: a systematic review and meta-analysis. Frontiers in Medicine, 13, Article 1963143. https://doi.org/10.3389/fmed.2026.1963143

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1963143

Keywords: stroke rehabilitation, guideline adherence, systematic review, meta-analysis, quality of care, clinical practice guidelines, cognitive screening, depression screening, physical therapy, evidence-practice gap, implementation science, PROSPERO