Evaluating the strengths and limitations of structured modified rankin scale validation studies - A systematic review

Variations in performance of structured modified Rankin Scale (mRS) questionnaires in people with stroke within different cultural and language settings have not been systematically assessed. We systematically reviewed all studies of structured mRS questionnaires compared to in-person unstructured mRS evaluation scores.

We searched PubMed and Web of Science for studies that evaluated structured mRS questionnaires against in-person unstructured mRS evaluations among people with stroke. Studies were analyzed for country of investigation, mRS self-assessment modality, care setting (inpatient vs outpatient), concordance between structured and unstructured mRS scores, and risk of bias.

We identified 13 structured mRS questionnaire validation studies across all continents except Africa. The telephone mRS questionnaire was most commonly used. Average time between unstructured and structured assessments was 14 days. The weighted kappa agreement between structured and in-person unstructured scores ranged from moderate to very good (K_w range: 0.56-0.90), though unweighted kappa scores were significantly lower (K range: 0.27-0.68). Discrepancies between structured and in-person unstructured scores tended to result from patients at the scale extremes. Patients with good mRS outcomes (mRS ≤2) tended to rate themselves better on structured questionnaires than their clinicians rated them on unstructured evaluations while those with poor outcomes (mRS ≥3) tended to rate themselves worse than their clinicians. Major limitations across studies included sampling bias in inpatient settings, time delays between in-person unstructured and structured mRS assessments, and differences in the formatting of unstructured mRS assessments.

While structured mRS questionnaire validation studies have generally displayed good agreement with in-person unstructured outcomes, significant discrepancies exist for patients with very poor and very good outcomes. Future studies should limit time between mRS assessments and better understand reasons for differences in structured and unstructured assessments. Further, data on validity of structured mRS tools are lacking from much of the world, especially Africa.

1K01TW011771-01A1 (Saylor).