This three-year study of 543 patients with COPD reinforces earlier findings that both the HADO-score and the BODE-index are good predictors of all-cause and respiratory mortality. However, differences appear between the two when applied to patients with different severity of COPD, as measured by FEV1%. In patients with the most severe disease (FEV1 < 50%), the BODE-index is a better predictor than the HADO-score of all-cause and respiratory mortality, whereas in patient with less severe disease (FEV1 ≥ 50%), the HADO-score is as good predictor of respiratory mortality as the BODE-index.
The HADO-score and BODE-index arose from different patient populations. This is likely to determine their characteristics, qualities, and where each is most appropriately used. The BODE-index originated in a study of a sample of patients (recruited in hospitals), who are generally more seriously ill. In the studies defining these measures, almost 70% of the patients in the BODE-index cohort had an FEV1 < 50%, compared with 34% of those in the HADO-score cohort. In contrast, the HADO-score emerged from outpatient pneumology clinics, where the spectrum of COPD severity is wider, and generally lower, than it is among hospitalized patients. The HADO-score is also simpler to determine, which reflects the difficulty of conducting complex or time-consuming tests in the primary care setting.
Two variables shared by both measures are essential predictors of mortality among patients with COPD: FEV1 [1, 2] and dyspnea. Although their importance has been questioned by some, it has been suggested that dyspnea is likely to be a better mortality predictor than FEV1 , though this has not been always replicated . The different variables in the two measures are the patient's perception of his or her general health and self-reported physical activity in the HADO-score and BMI and the 6MWT in the BODE-index.
In our application of the HADO score, we introduced a single question to gauge a patient's sense of his or her level of health ('In general, would you say your health is...'). This question has proved to have good reproducibility, reliability, and strong concurrent and discriminatory scale performance with an established health status measure (SF-12) . Moreover, it has been shown that this single health-related question can stratify patients with different risks for adverse outcomes such as mortality and the use of health resources . A meta-analysis suggests a strong association between self-perceived health status and mortality, even after adjustment for key covariates such as functional status, depression, and co-morbidity, with individuals reporting poor health having higher mortality than their counterparts reporting excellent health (OR, 1.74; 95% CI, 1.51 to 2.02) .
The BODE-index includes BMI as a variable. It has been shown that BMI < 20 kg/m2 constitutes an independent predictive factor related to COPD mortality . In a study by Chailleux et al., the prevalence of malnutrition in patients with severe COPD who required home oxygen therapy ranged from 23% for men to 30% for women .The prevalence of malnutrition appears to be lower in the Mediterranean area. In a study of 3,126 patients, 50% of whom had a FEV1 < 50%, it was about 7% . In the ambulatory setting, BMI offers limited prognostic value, since most patients seen in this setting have mild or moderate COPD, which generally does not cause malnutrition leading to BMI < 20 kg/m2. On the other hand, among patients with severe COPD, it is probably appropriate to carry out a precise measure of body composition, since it has been observed that lower fat-free mass is associated with higher mortality . Since determining body composition would increase the complexity of obtaining the data for the prognostic index, its use should be reserved for patients with severe disease and should be exclusively handled in the hospital or specialized outpatient clinics.
The assessments of activity used in the two measures are also quite different. Self-reported physical activity, used in the HADO-score, is a behaviour (the activity the patient makes) that is quite simple to evaluate. The 6MWT, used in the BODE-index, reflects exercise capacity, a condition the patient reaches that allows him or her to meet the requirements of daily life (what the patient is able to do) . The 6MWT is more difficult to measure. Both concepts, physical activity and exercise capacity, are indistinctly used in epidemiological studies.
Exercise capacity (as peak oxygen uptake, VO2 ) has been associated with COPD mortality independent of FEV1 and age . Among patients with severe COPD, the distance walked in six minutes has been shown to be an independent and better predictor of mortality than FEV1 . In a recent study, the 6MWT had a stronger association with mortality (HR, 0.996; 95% CI, 0.993 to 0.999; P < 0.01) than did peak VO2 (HR, 0.971; 95% CI, 0.959 to 1.000; P = 0.050) .
Regarding physical activity, Garcia-Aymerich et al. showed in a population-based cohort that among patients with COPD, even slight physical activity (equivalent to walking or bicycling for two hours per week) had important consequences for the course of the disease, decreasing by 30% to 40% the risk of hospitalization due to COPD and the risk of respiratory mortality . Interestingly, a questionnaire, or even a single question, on physical activity have proved to have acceptable associations with VO2 .
Despite the differences, both self-reported physical activity and exercise capacity appear to offer similar information concerning morbidity and mortality. Both are good predictors of mortality .
The differences in the HADO-score and BODE-index as predictors of mortality among patients with FEV1 < 50% is worth noting. Although both appear to behave similarly when analyzed using ROC curves (Figure 1), the multivariate analysis yields several differences. This could be due to the cut-points chosen, or because the HADO-score has only three categories of severity, compared to four for the BODE-index.
One limitation of our study is that the sample was composed almost entirely of men, which makes it impossible to generalize the results to women as well. This reflects the fact that in Spain, smoking was once for men only, and women began smoking relatively recently. It is possible that the use of self-reported physical activity in creating the HADO-score could influence the results, since we did not objectively measure the level of activity by accelerometers. Given the size of the cohort, this would have been a difficult undertaking. The main goal of this work was to study the functioning of both scores (BODE and HADO) in patients with different severity. Apart from using multidimensional scores, which is the main point of this work, there is not any other alternative form of classifying severity of COPD than using FEV1. Nevertheless, this choice has some disadvantages since this division of our sample (FEV1 < 50% vs FEV1 ≥ 50%) reduces the number of deaths in each category and thus the power of the study to detect differences between the BODE and the HADO in their ability to predict mortality. But besides that, we were able to detect differences.
Strengths of this study are the prospective collection of data, up to three years of follow-up of a relatively large cohort of COPD patients covering a wide range of severity of the disease and the evaluation of the utility of two COPD-specific severity scales in relation to robust outcomes, such as mortality.