wait a moment

Comparison of Continuous and Discrete Measurements of Dyspnea During Exercise in Patients With COPD and Normal Subjects: Statistical Analysis Considerations

Both linear, and power2, function equations have been used to summarize the relationships between physiologic variables during exercise and the perceptual response of dyspnea. In simple terms, both approaches involve the matching of one continuum to another continuum. In 1992, Killian and colleagues2 published results of perceived magnitude of dyspnea and leg effort during cycle ergom-etry in 460 normal subjects using power function equations. However, the data shown in Figure 2 of an individual patient with COPD clearly illustrate that linear relationships can also be used to describe such results. Based on correlation coefficients as an indicator of goodness of fit, the physiologic/perceptual relationships in our subjects were similar for linear and power function analyses, except that linear regression for VE-breathlessness fit the data significantly better than did the power function. Moreover, numerous investigators” have used linear regression to report their data. For these reasons, we used linear function analyses to summarize the findings in this report.

Test-Retest Reliability
Both the continuous and discrete methods were highly reliable over time in these populations of elderly subjects. Previous investigators have demonstrated satisfactory to excellent reliability of the slope of a physiologic variable and dyspnea (measured by the discrete method) in patients with asthma or COPD on repeat testing. Harty et al reported good “reproducible measurements in repeated tests” in six normal subjects who used the continuous method. In a previous study, we observed excellent test-retest reliability in healthy, college-aged subjects with both the discrete and the continuous methods at 2- to 3-day intervals, further

Comparison of Continuous and Discrete Methods
One major difference between the two methods for measuring breathlessness during exercise is that the continuous method resulted in significantly more ratings than the discrete method. The high number of ratings with the continuous method can be an important advantage because patients with reduced exercise capacity may only be able to give a small number of ratings with the discrete method. Consequently, the slopes of the linear function may become difficult to calculate with any confidence or certainty, especially if the data points are nonmonotonic (do not always increase) with progressive exercise. In our previous report, we also found that the slopes of work-dyspnea were similar between the two methods in separate studies of 14 healthy women and in 14 healthy, college-aged subjects.