Abstract
Background: Although numerous health-related quality-of-life instruments are available to measure patients' quality of life, few studies have compared these measures directly to determine how they function in the same group of patients.
Objective: The purpose of this study was to empirically compare psychometric properties of the Chronic Heart Failure Questionnaire (CHQ), the Minnesota Living with Heart Failure Questionnaire (LHFQ), and the General Health Survey Short-form-12 (SF-12).
Sample: A convenience sample of 211 patients with heart failure completed baseline questionnaires; 165 patients completed the entire 26-week study.
Methods: Patients completed telephone interviews at baseline and at 4, 8, and 26 weeks after baseline. To compare mode of administration, a subset of patients (n = 173) completed face-to-face and telephone interviews.
Results: Patients reported low-to-moderate health-related quality-of-life overall. Reliability of the three instruments was satisfactory. Responsiveness to changing condition, as evaluated by analysis of variance, receiver operating curve characteristics, and the minimal clinically important difference method, indicated that the CHQ and LHFQ were more responsive to changing conditions than the SF-12. No major differences were noted between the scores of the face-to-face interviews and the baseline telephone interviews. The LHFQ and SF-12 were easier and took less time to administer than the CHQ.
Conclusions: While all three instruments were reliable and valid, the CHQ and LHFQ were more sensitive than the SF-12 in detecting clinically important changes over time.
Heart failure (HF) is a major public health problem within the United States (US) and, designated by some experts as an epidemic, 4.8 million persons within the US have HF with almost 500,000 incident cases reported each year (American Heart Association, 2002;Crofts, Giles, Pollard, Keenan, Casper et al., 1999). Unfortunately, mortality and morbidity rates remain high, with the 12-month mortality rate for patients with HF approximately 15% (AHA). An estimated $23 billion in direct and indirect costs will be spent on HF care in 2002 (AHA). Despite the new and effective therapies being developed for these chronically ill patients, their health-related quality of life (HRQL) remains poor (Bennett et al., 2002;Bennett, Perkins, Lane, Deer, Brater et al., 2001;Riegel et al., 2002;Grady, Jalowiec, Grusk, White-Williams, & Robinson, 1992).
Health-related quality of life is a multidimensional concept comprised of several domains, including physical/biological factors, symptom status, functional status, health perceptions, and overall well-being (Testa & Simonson, 1996;Wilson & Cleary, 1995). Health-related quality of life is an important focus of study because patients' perceptions of HRQL are used increasingly to evaluate healthcare interventions. Both researchers and clinicians should know if the instruments assessing HRQL are valid, reliable, sensitive to the changing condition of patients, and feasible to administer.
Instruments designed to measure HRQL are generally categorized as generic or specific. Generic HRQL instruments are designed to assess general aspects of HRQL among persons with a variety of health conditions. Generic instruments are useful in comparing the HRQL among patients with differing chronic disorders. Also these measures have assessed HRQL within single populations of patients with one disorder (Ware, Kosinski, & Keller, 1994; 1995). However, because generic HRQL instruments are global, they may not detect major problems experienced by patients with specific disorders such as HF.
In contrast to generic instruments, specific HRQL instruments measure the problems and HRQL related to a specific disease or disorder. For example, dyspnea and fatigue are the two primary symptoms of HF patients, and therefore specific instruments for HF focus on these symptoms in assessing HRQL. Specific instruments ideally address specific problems, but they are not useful for comparing HRQL among people with varying conditions. Additionally, the full range of domains comprising a person's HRQL may not be assessed.
In a comparison of the psychometric properties of a generic HRQL instrument and a disease-specific HRQL instrument, Wolinsky and colleagues (1998) administered the generic General Health Survey Short-form 36 (SF-36) (Ware et al., 1994) and the specific Chronic Heart Failure Questionnaire (CHQ) (Guyatt, Nogradi, Halcrow, Singer, Sullivan, & Fallen, 1989) to 670 outpatients with known chronic HF and/or coronary artery disease and found that both instruments were appropriate for these outpatients. However, the CHQ had fewer floor and ceiling effects, better internal consistency reliability, and better construct validity than the SF-36. In evaluating discriminant validity to determine whether the instruments could differentiate among patients with varying disease severity, both the SF-36 and the CHQ discriminated among patients in the five quintiles of Ambulatory Care Groups (a measure of disease severity) although the SF-36 better differentiated disease severity. Because the CHQ had better overall psychometric properties, these authors concluded that using the SF-36 is questionable in studies of group or individual comparisons with patients having HF or coronary artery disease (Wolinsky et al., 1998).
Ni and colleagues (2000) compared the responsiveness of the generic General Health Survey Short-form 12 (SF-12) (Ware et al., 1995) and the specific Minnesota Living with Heart Failure Questionnaire (LHFQ) (Rector, Francis, & Cohn, 1987) among 87 outpatients who were newly enrolled in a university-based, multidisciplinary program for HF. The HRQL was assessed at baseline and at 3 months after baseline. In addition to the two HRQL instruments, patients completed a questionnaire regarding their health status during the 3 months between baseline and follow-up. The mean change scores of both the total SF-12 and the total LHFQ significantly improved over time. The LHFQ was better able to discriminate between patients who improved and those who did not improve than the SF-12, leading the authors to conclude that the LHFQ was more responsive than the SF-12.
In a cross-sectional study, Sneed and colleagues (2001) compared the SF-36 and the LHFQ among 30 patients enrolled in a multidisciplinary HF clinic. The authors found that the physical and emotional subscales of the LHFQ were significantly correlated with the SF-36 physical and mental component summary scales. Based on correlations between the scores of 11 patients with both HF and arthritis, it was concluded that the SF-36 differentiated physical and emotional quality of life better between these two types of conditions. However, the study was limited by the small sample.
Because so few direct comparisons of instruments evaluating HRQL of HF patients have been reported in the literature, it is not known whether instruments such as the SF-12, LHFQ, and CHQ are comparable for use in research or clinical practice. Two of the three studies cited above were cross-sectional, which did not allow for examination of clinically important changes. Therefore, the purpose of this study was to empirically compare the psychometric properties of two specific HRQL measures, the CHQ (Guyatt, Nogradi, Halcrow, Singer, Sullivan, & Fallen, 1989) and the LHFQ (Rector, Francis, & Cohn, 1987), and one generic HRQL measure, the SF-12 (Ware et al., 1995), in an urban population of patients with HF. The LHFQ and the CHQ were chosen for comparison because they are commonly used in HF trials. The SF-12 was selected for comparison because it is widely used and was designed to measure general HRQL rather than disease-specific HRQL. The SF-12 was chosen rather than the SF-36 to minimize respondent burden in answering the questionnaires. The specific aims of the study were to: (a) estimate the reliability of the CHQ, LHFQ, and SF-12, respectively, in a single population of patients with chronic HF; (b) compare the mode of administration (face-to-face and telephone interviews) of the CHQ, LHFQ, and SF-12 in a single population of patients with chronic HF; and (c) evaluate the responsiveness of the CHQ, LHFQ, and SF-12, to detect changes in patients' clinical health status, respectively, in a single population of patients with chronic HF.