Bridging from the Picker Hospital Survey to the CAHPS Hospital Survey
Published In: Medical Care, v. 46, no. 7, July 2008, p. 654-661
Posted on RAND.org on January 01, 2008
OBJECTIVE: Illustrate an accessible method of bridging data from earlier surveys to the CAHPS(R) Hospital Survey to support hospitals' internal quality improvement efforts. DATA SOURCES/STUDY SETTING: Survey of patients with more than 300,000 annual hospitalizations in a large urban hospital. STUDY DESIGN AND DATA: Six pairs of parallel items from the CAHPS and Picker Hospital Surveys were administered to the same 734 patients. The authors assessed item comparability and applied bridging adjustments to convert old items to predicted scores on the new CAHPS items. PRINCIPLE FINDINGS: Differences in wording, response options, and cut points for problem scores yielded large differences in problem score rates between the Picker and CAHPS Hospital Surveys, requiring bridging formulas. Tetrachoric correlations for 5 of 6 pairs indicated high correspondence (r = 0.71-0.97, P < 0.001) in the underlying constructs assessed by the 2 surveys, validating the use of bridging. Bridged scores contain less information per observation than directly measured new scores, but with sufficient sample sizes they can be used to detect trends across the transition. CONCLUSIONS: Hospitals can use the methodology described here to trend their scores from a previous survey to the CAHPS Hospital Survey with sufficient precision to support ongoing quality improvement efforts. Hospitals should administer an instrument containing pairs of old and new items to enough patients to yield at least 625 completes to measure bridging parameters precisely. Where correspondence is high, old items can and should be replaced by CAHPS items. Important old items with weaker associations with new items may be retained.