Testing the psychometric properties of the Minnesota Living with Heart Failure questionnaire

Health-related quality of life (HRQOL) is an important outcome in patients with heart failure. One of the most commonly used instruments to measure HRQOL in this population is the Minnesota Living With Heart Failure Questionnaire (LHFQ). Although the psychometric properties of the LHFQ have been tested, the results do not definitively support the psychometric soundness of the instrument.

To examine the psychometric properties of the LHFQ.

Data from 638 patients with heart failure were used to assess the reliability, homogeneity, representativeness, discriminative ability, and construct validity of the LHFQ before and after deletion of 5 items that showed lack of representativeness and contributed to inadequate factor structure.

Cronbach's alphas for the LHFQ Total and subscales were greater than .80. Interitem correlation coefficients in 17 of the 21 items, item-total correlation coefficients in 20 items, and discriminative ability in all items were acceptable. The total and both subscales of the LHFQ differentiated New York Heart Association functional groups. The Physical subscale was moderately related to the physical measures (the Specific Activity Scale and symptom status), whereas the emotional subscale was weakly related to the measures. The results of item p level testing and factor analysis demonstrated that 7 items were consistently problematic and 5 items were recommended to be deleted. The results of the reliability, homogeneity, and construct validity after deletion of these items demonstrated that the psychometric properties of the LHFQ were improved as a result.

The initial results provided additional support for the reliability and substantial evidence for the validity of the LHFQ. However, the results of item and factor analyses did not fully support the psychometric soundness of several items. The psychometric properties of the LHFQ after deleting these items were improved. These results could provide researchers and clinicians a more useful measure of HRQOL.