A systematic review of the psychometric properties of self-report research utilization measures used in healthcare

BACKGROUND In healthcare, a gap exists between what is known from research and what is practiced. Understanding this gap depends upon our ability to robustly measure research utilization. OBJECTIVES The objectives of this systematic review were: to identify self-report measures of research utilization used in healthcare, and to assess the psychometric properties (acceptability, reliability, and validity) of these measures. METHODS We conducted a systematic review of literature reporting use or development of self-report research utilization measures. Our search included: multiple databases, ancestry searches, and a hand search. Acceptability was assessed by examining time to complete the measure and missing data rates. Our approach to reliability and validity assessment followed that outlined in the Standards for Educational and Psychological Testing. RESULTS Of 42,770 titles screened, 97 original studies (108 articles) were included in this review. The 97 studies reported on the use or development of 60 unique self-report research utilization measures. Seven of the measures were assessed in more than one study. Study samples consisted of healthcare providers (92 studies) and healthcare decision makers (5 studies). No studies reported data on acceptability of the measures. Reliability was reported in 32 (33%) of the studies, representing 13 of the 60 measures. Internal consistency (Cronbach's Alpha) reliability was reported in 31 studies; values exceeded 0.70 in 29 studies. Test-retest reliability was reported in 3 studies with Pearson's r coefficients > 0.80. No validity information was reported for 12 of the 60 measures. The remaining 48 measures were classified into a three-level validity hierarchy according to the number of validity sources reported in 50% or more of the studies using the measure. Level one measures (n = 6) reported evidence from any three (out of four possible) Standards validity sources (which, in the case of single item measures, was all applicable validity sources). Level two measures (n = 16) had evidence from any two validity sources, and level three measures (n = 26) from only one validity source. CONCLUSIONS This review reveals significant underdevelopment in the measurement of research utilization. Substantial methodological advances with respect to construct clarity, use of research utilization and related theory, use of measurement theory, and psychometric assessment are required. Also needed are improved reporting practices and the adoption of a more contemporary view of validity (i.e., the Standards) in future research utilization measurement studies.