Longer response scales improved the acceptability and performance of the Nottingham Health Profile

Abstract

Objective: To test alternative response formats for the Nottingham Health Profile (NHP), in terms of acceptability, score distributions, and measurement properties.

Study design and settings: Randomized trial of four response formats for the NHP: original "yes/no" format, a 3-point similarity format ("applies completely/in part/not at all"), a 5-point intensity format ("completely true" to "completely false"), and a 5-point frequency format ("all the time" to "never"). Respondents were patients discharged from a hospital. We compared scores distributions, reliability coefficients, correlations with dimension-specific numerical scales, and patient ratings of the instrument.

Results: Response rates were similar for the four versions. The original response format had the fewest fully completed questionnaires, and the largest ceiling effects. Internal consistency and test-retest coefficients were acceptable for all versions, but were higher for the two 5-point formats. Correlations reflecting convergent and discriminant validity were higher for the longer response formats than for the original version. The frequency format received the highest ratings from patients, particularly from the sicker and older subgroups.

Conclusions: The psychometric performance and patient acceptability of the NHP can be improved by using a 5-point frequency response format instead of the original dichotomous response format.