Key properties of a methodologically robust health related quality of life (HRQOL) instrument
| Property | Definition | Method of assessment | |||
|---|---|---|---|---|---|
| Validity | |||||
| Face12 | Measures what it is supposed to measure | Full literature review, expert opinion, patient input (eg focus groups) | |||
| Content12 | Adequately samples most important areas of interest | Pre-testing with item reduction or augmentation | |||
| Construct12 | Relationship between score and a hypothesis of what is being measured | Instrument compared with another marker of illness to determine if it behaves as predicted | |||
| Criterion12 (convergent) | Relationship between new questionnaire and an accepted reference | Instrument compared to an accepted reference measure that evaluates the same or similar features | |||
| Discriminative4 | Instrument can distinguish between two groups of dissimilar patients | QOL scores for patients with different disease severity or different patterns of disease should differ significantly | |||
| Reliability | |||||
| Test-retest4 | Ratio of between patient variation to total variation in score | Patients who remain stable should have little change in QOL scores on repeated measures. Described by intraclass correlation coefficient (ICC) (0–1, 1 perfect agreement) | |||
| Internal consistency4 | Correlation of items within same domain or with the full questionnaire score | Cronbach's alpha coefficient (0–1, 1 excellent) | |||
| Responsiveness4 | Signal to noise ratio of change with time | Patients with clinically important change (improve or deteriorate) should have significant change in QOL score | |||