Residual analysis in the determination of factors affecting the estimates of body heat storage in clothed subjects

Body heat storage can be estimated by calorimetry (from heat gains and losses) or by thermometry [from changes (Delta) in mean body temperature (<(T)over bar (b)>) calculated as a weighted combination of rectal (T-re) and mean skin temperatures (<(T)over bar (sk)>)]. If an invariant weighting factor of T-re and <(T)over bar (sk)> were to be used (for instance, Delta<(T)over bar (b)> = 0.8 . Delta T-re + 0.2 . Delta<(T)over bar (sk)> under hot conditions), body heat storage could be over- or underestimated substantially relative to calorimetry, depending on whether the subject was wearing light or protective clothing. This study investigated whether discrepancies between calorimetry and thermometry arise from methodological errors in the calorimetric estimate of heat storage, from inappropriate weightings in the thermometric estimate, or from both. Residuals of calorimetry versus thermometric estimates were plotted against individual variables in the standard heat balance equation, applying various weighting factors to T-re and <(T)over bar (sk)>. Whether light or protective clothing was worn, the calorimetric approach generally gave appropriate estimates of heat exchange components and thus heat storage. One exception was in estimating latent heat loss from sweat evaporation. If sweat evaporation exceeded 650 g . h(-1) when wearing normal clothing, evaporative heat loss was overestimated and thus body heat storage was underestimated. Nevertheless, if data beyond this ceiling were excluded from the analyses, the standard 4:1 weighting matched calorimetric heat storage estimates quite well. When wearing protective clothing, the same 4:1 weighting approximated calorimetric heat storage with errors of less than approximately 10%, but only if environmental conditions allowed a subject to exercise for more than 90 min. The best thermometric estimates of heat storage were provided by using two sets of relative weightings, based upon the individual's metabolic heat production (M over dot in kilojoules per metre squared per hour): {4 - [(M over dot - x over dot) . x over dot (-1)] . 2}:1 for an initial, thermoneutral environment and {4 + [(M over dot - x over dot) . x over dot (-1)] . 5}:1 for a final, hot environment; the optimal value of x over dot lay between 450 and 500 kJ . m(-2) . h(-1). We concluded that the accuracy of thermometric estimates of heat storage can be improved by modifying weighting factors of T-re and <(T)over bar (sk)> according to the environment, type of clothing, and metabolic rate.