Comment on "negative emotions and acute cardiovascular responses to laboratory challenges"

Abstract

In their provocative article, Feldman et al. (1) suggest that emotion may not be the primary pathway by which stress affects cardiovascular (CV) functioning. They reanalyze data from several laboratory-based cardiovascular reactivity studies and show that individual changes in CV (systolic blood pressure [SBP] and diastolic blood pressure and heart rate) functioning have a negligible to moderate correlation (between .13 and .34) with individuals' reports of changes in negative emotion (NE). The authors conclude that negative emotions probably play a small role in physiological responses to acute laboratory stressors (1). The issue addressed in the article is ~ important one. There is a strong tendency in behavioral medicine research to assume that the physiological effects of stress are cognitively mediated. This assumption is so strong that emotion/mood data are collected primarily as a manipulation check, to verify or document that the laboratory task is indeed stressful, and have only rarely been used to test the assumed mediating role of emotions. Despite my sympathy with the conclusion of their article, I have reservations about the analysis which led to these conclusions. Specifically, the computation of the correlation of change in negative emotion with change in CV functioning I begins by subtracting out of both sets of change scores the respective mean levels of change associated with each task. Tables 2 and 4 of the article show that the tasks (i.e. stressors) were associated with substantial increases in the CV measures, while Tables 3 and 4 show the same for negative emotion. These concurrent increases in both the CV measures and negative emotion are essentially lost in the correlation of change scores. The remainder of this comment will illustrate, conceptually and empirically, the potential problem of ignoring the mean change in both measures. Consider a hypothetical task that was associated with an average increase of 15 mm Hg in systolic blood pressure and an average increase of 1 point on a 4-point negative emotion scale. Further suppose that this was a very finely tuned task and there were no individual differences (i.e. all individuals respond identically) in both the SBP and NE response. In this case, there is no correlation between the change scores. Yet, clearly the task had a large effect on both SBP and NE, and thus SBP and NE must be related to each other. In one sense, the association is perfect: every time NE increases by 1 point, SBP increases by 15 mm Hg. What we cannot disentangle in this example, or most of the others