The Use of Regression Equations to Estimate Peak Work Rate in People with COPD

Keywords :

• COPD
• rehabilitation
• exercise
• regression analyses

Dear Editor,

We read with interest the recent paper by Sillen et al (1) regarding the use of regression equations to estimate the peak work rate (Wpeak) achieved during an incremental cycle ergometry test (ICET) in patients with chronic obstructive pulmonary disease (COPD). The authors should be commended on the large sample available for the analyses. However, we believe their findings may warrant some further interpretation and discussion, particularly given the clinical importance of these results.

First, the new regression equation presented in the paper included the measurement of fat free mass (FFM) as an independent variable. Previous regression equations to estimate Wpeak were developed on the premise of allowing clinicians who work in minimally resourced pulmonary rehabilitation programs (PRP) to estimate Wpeak using common measurements such as height, weight and six-minute walk distance (6MWD) (2–4). Thereafter, the estimate of Wpeak could be used to prescribe an initial cycle-based exercise training intensity. Data from surveys across Australia (5), Canada (6) and the United Kingdom (7) suggest that FFM is rarely measured as part of clinical PRPs. Although we accept that clinical PRPs, which do not have access to bio-electrical impedance, can derive measures of FFM using skin fold anthropometry, measures of FFM obtained using different methods may not be interchangeable (8).

We also considered that, as that the regression equation included FFM and body weight, it was possible that the variance in FFM was largely explained by body weight. Nevertheless, as multi-collinearity diagnostic statistics were not reported, it was not clear the extent to which FFM contributed independently and meaningfully to the estimation of Wpeak. Taken together, it appears that in order to use the equation published by Sillen et al (1) measures of FFM, derived using bio-electrical impedance are required. This precludes its use in standard clinical PRPs that operate using minimal resources.

Second, the characteristics of the participants in the study by Sillen et al (1) were different to those who participated in the previous studies that developed regression equations to estimate Wpeak (2–4). Of note, those in the study by Sillen et al (1) were, on average, younger and heavier. Further the mean 6MWD reported in the study by Sillen et al (400 m) (1) was considerably less than that reported by Hill et al (464 m) (2) and Luxton et al (508 m) (3), and greater than that achieved by Kozu et al (325 m) (4). Importantly, the sample in the study by Kozu et al (4) was Japanese rather than Caucasian. Some Asian ethnicities are known to have differences in body composition when compared with Caucasians, such as less muscle mass for a given body mass index (9). Given these differences between the samples, it is not surprising that the previously published regression equations did not yield precise estimates of Wpeak in the study by Sillen et al (1).

Finally, and perhaps most importantly, the authors concluded that regression equations were unable to accurately estimate Wpeak for the majority of patients with COPD and therefore ‘the Wpeak needs to be assessed in order to prescribe a patient-tailored training load during cycle ergometry training’ (1). This implies that the measure of Wpeak is a single, stable and repeatable measure for each patient with COPD. A previous study indicated that 99% of the differences in Wpeak achieved during two ICET performed 1 to 2 weeks apart under identical conditions was ± 15 W (10). When tests were performed on four occasions, each separated by one month, the mean intra-subject variability in Wpeak was 7 W (11). This variability is greater than that deemed to be acceptable in the current study (i.e. ± 5 W). The measurement of Wpeak is also dependent on the test protocol used (12, 13). Even if we assume that Wpeak is a single, stable measure for each patient, which can be used to prescribe cycle-based training at an optimal load, many patients appear to be unable to tolerate the initial training work rate (14) and international pulmonary rehabilitation position statements recommend intensity is titrated according to symptoms of dyspnea or fatigue (15).

Although offering full cardiopulmonary exercise testing to all patients referred to PRP might be possible in some countries, for large countries with areas of low population density such as Australia and Canada, patients often live several hundreds of kilometres away from a facility with access to the equipment and expertise needed to conduct such an assessment. For PRP that operate in these regional and remote areas, rather than guessing an initial cycle-based exercise training work rate, a more scientific approach would be to; (i) select the regression equation that was developed in a study sample with characteristics that are most consistent with those patients referred to the program, (ii) use this equation to estimate Wpeak and initial training intensities and, (iii) titrate work rate (up or down) according to symptoms. Even if such an approach does not result in ‘high-intensity’ training for all patients, data from a recent Cochrane review suggests that the difference in gains made following training at high vs. low intensities by patients with COPD during pulmonary rehabilitation may not be significant (16).

Declaration of Interest Statement

The authors have no conflict of interest with the content of this letter. The authors are responsible for the writing of this letter.