Table 2. Criteria for Rating Methodological Quality of Multivariate Studies.

The rating criteria for methodological quality is adapted from Chen and Winstein.

	Criteria	Each item received a score of 1 if:
	Internal validity
1 2	Adequate Definitions Outcome measures Predictor measures	Outcome measures and predictor measures were precisely defined.
3 4	Reliable or valid measurements Outcome measures Predictor measures	Studies tested the reliability and validity of measures (outcome and predictor) or referenced the literature on the clinimetric properties of each measure.
5	Blinded tester	Blinded testers were specified.
6	Appropriate time point to capture predictors	Measures of potential predictors were acquired prior to the measurement of the primary outcome.
7	Control of subject dropout	Dropouts during observation specified, characteristics of dropouts did not influence conclusion.
	Statistical validity
8	Control for statistical significance	Rationale for statistical approaches was specified, the relationship between outcome measures and predictor measures was tested for statistical significance.
9	Adequate sample size	At least 10 subjects for each predictor variable examined.
10	Control for multicollinearity	Interaction between two or more predictor measures was tested in the prediction model.
	External validity
11	Identification of wheelchair type	The study identified if the sample was comprised of manual or power wheelchair users, or both. If both, results were stratified by wheelchair type.
12	Specification of inclusion and exclusion criteria	Relevant subject characteristics (eg, age, sex, diagnosis) were specified for patient selection.
13	Description of additional treatment effects during period of observation	Information on rehabilitation treatment was reported.
14	Cross-validation of the prediction model	Prediction models were validated in a second independent group of wheelchair users.
15	Description of clinical meaningfulness	Minimal clinically important differences were considered.