Measuring Cancer Clinical Trial Understanding

Abstract

Researchers, practitioners, and participants in cancer clinical trials must have a clear understanding of clinical trials if participation in them is to be solicited ethically and effectively. A valid and reliable measure of cancer clinical trial understanding did not exist prior to a 2005 study conducted for the Coalition of Cooperative Cancer Groups. This report outlines a measure derived from that study, discusses the rationale for its component items, examines its psychometric properties, and demonstrates the relationship of this measure to the enrollment decision. Data from national samples of cancer survivors and the general public demonstrate the measure's validity and reliability. Results are discussed as they relate to patient understanding of clinical trials, informed decision making, and health communication processes.

Notes

¹Weighting is a procedure for national samples of adults to assure that variations in response rate across demographic categories do not bias any analyses based on those data. The weighted sample is an accurate reflection of the demographic composition of the U.S. population at the time that the sample was taken. Detailed demographic information is available from the first author upon request.

²The coding guide is available from the first author upon request.

³The coding of a 0-to-10 scale into correct and incorrect inherently involves a judgment about a cut-point. In the literature using the 0-to-10 metric, there is a tendency to classify the highest (or lowest) three points as correct, depending on the direction of the item.

⁴Unlike exploratory factor analysis, which identifies factors based on the correlations found in the data, confirmatory factor analysis is a test of an a priori hypothesized measurement model associated with a set of items. More information about confirmatory factor analysis can be found in Long (1983). More information about LISREL can be found in Jöreskog and Sörbom (1993).

⁵A factor loading may be thought of as the correlation between an item and the underlying dimension, or factor. For a discussion of factor loadings and factor scores, see Long (1983).

⁶Root mean square error of approximation (RMSEA) is one of several widely used fit statistics. A value of .05 or lower is considered a good fit for either a confirmatory factor analysis or a structural equation model. See Schumacker and Lomax (2004) for an introductory discussion; see Hayduk (1987) for a more comprehensive discussion.