Search

To systematically compare the PEDro scale and the Jadad scale when applied to the stroke rehabilitation literature.A literature search of multiple databases was used to identify all trials from 1968 through 2002. Each article was reviewed and assigned quality scores according to PEDro and Jadad criteria. Quality scores for both scales were compared using descriptive statistics. The correlation between the scales was estimated using the Pearson product moment correlation coefficient.272 randomized controlled trials were retrieved and subjected to quality scoring to both the PEDro scale and the Jadad scale. Mean scores (with standard deviation) for the PEDro and Jadad scales were 5.78 (1.4) and 2.46 (1.1), respectively. The Pearson coefficient determined the PEDro and Jadad scales to be significantly correlated (r = .59, P.01). Although significant, the correlation was not deemed to be very strong. When applied to physical or rehabilitation therapy studies only, the scales were slightly less correlated (r = .49, P.01) than among drug-based studies (r = .52, P.01).In the stroke rehabilitation literature, where double-blinding studies are often not possible due to the nature of the interventions, breaking down the levels of blinding and accounting for concealed allocation, intention-to-treat, and attrition is important. Accordingly, the PEDro scale provides a more comprehensive measure of methodological quality of the stroke literature.

Objectives To apply a tool that purports to differentiate between efficacy and effectiveness studies to stroke rehabilitation trials and to evaluate its applicability and reliability. Study Design and Setting Three raters developed item operational definitions before independently applying the seven-item scale to 151 randomized controlled trials (RCT), published during or after 1997, that evaluated either a pharmacologic (P, n = 78) or a nonpharmacologic (NP, n = 73) intervention. Inter-rater reliability was assessed for both individual items and total scores, separately for P and NP trials. Results Item inter-rater reliability (multiple-rater kappa) ranged from 0.00 (95% CI [confidence interval]: −0.13, 0.13) to 0.85 (95% CI: 0.73, 0.98) and from 0.21 (95% CI: 0.08, 0.34) to 0.79 (95% CI: 0.66, 0.92) for P and NP RCTs, respectively. For the total score (dichotomized), kappa values were 0.43 (95% CI: 0.31, 0.56) and 0.51 (95% CI: 0.37, 0.64) for P and NP trials, respectively. Conclusions The tool provides a solid foundation on which to base further discussion of the differential criteria of efficacy–effectiveness trial design. Scale items should be properly operationalized depending on the research question of interest and evaluated for reliability before the scale is used for definitively judging a given study's design or the external validity of its results.

Objective The objective of this two-phase study was to assess the adequacy of the reporting of concealed allocation (CA) in randomized controlled trials (RCTs) evaluating interventions associated with stroke rehabilitation. Study Design and Setting In phase I of the study, 50 RCTs included in a systematic review were selected to establish agreement between two raters. Two investigators determined if the method described to conceal the randomization schedule was adequate, inadequate, or not reported. In phase II, using a larger sample size ( n = 165), the differences in the proportion of studies with and without adequate CA are reported for two comparisons: (1) pharmacological vs. nonpharmacological trials and (2) multicentered vs. single-site studies. Results In both phases I and II, CA was described adequately in one-third of all studies sampled. The agreement between raters was 88% ( к = 0.79; 95% confidence interval: 0.65, 0.94). No significant differences in the adequacy of reporting for CA were found with respect to study type (pharmacological vs. nonpharmacological), whereas multicentered trials reported adequacy of CA more frequently. Conclusion Although concealment of group allocation is an important feature of trial design, it was frequently not reported in many RCTs associated with stroke rehabilitation.