TABLE 1.
METHODS USED TO ASSESS THE PSYCHOMETRIC ADEQUACY OF HRQoL AND PRO MEASURES
| Measurement Property | Method of Assessment | Description | Considerations |
|---|---|---|---|
| Reliability | Test-retest reproducibility | Stability of scores upon readministration when no change has occurred in the concept of interest | Most important type of reliability for instruments used in clinical trials |
| Internal consistency | Extent to which the items consistently measure the same underlying construct (e.g., Cronbach's coefficient alpha). | Internal consistency alone is not sufficient evidence of reliability | |
| Inter-interviewer agreement | Agreement between responses when an instrument is administered by different interviewers | Only relevant to instruments administered by an interviewer | |
| Validity | Content, face validity | Completeness, relevance, and comprehensibility of items in assessing the concept of interest | Difficult to assess quantitatively, requiring use of qualitative methods |
| Construct validity | |||
| Convergent validity | Correlation with measures of related constructs in the hypothesized manner, consistent with the proposed conceptual framework | Referent measures may be previously validated PROs or non-PRO measures | |
| Divergent, discriminant validity | Lack of correlation with measures that are intended to be different or conceptually distinct | ||
| Known groups validity | Ability to distinguish groups that are expected or known to be different with regard to the concept of interest | ||
| Predictive validity | Ability to accurately predict future health status or other relevant outcomes | ||
| Responsiveness (longitudinal validity) | Calculation of responsiveness statistic (e.g., effect size) | Ability to detect changes in the measured concept over time, usually in response to a specific intervention or known change in health | Responsiveness of an instrument may depend on the time interval |
| Interpretability | Minimal important difference (MID) | Smallest difference in score that is considered meaningful to patients and/or clinically relevant | MID must be determined by triangulation of different methods, and may vary for different subpopulations |
Definition of abbreviations: HRQoL = health-related quality of life; PRO = patient-reported outcome.
Adapted from FDA Guidance for Industry on Patient-Reported Outcome Measures (4).