Introduction: The fitting of probabilistic decompression sickness (DCS) models is more effective when data encompass a wide range of DCS incidence. We obtained such data from the Air Force Research Laboratory Altitude Decompression Sickness Research Database. The data are results from 29 tests comprising 708 human altitude chamber exposures (536 men and 172 women). There were 340 DCS outcomes with per-test DCS incidence ranging from 0 to 88%. The tests were characterized by direct ascent at a rate of 5000 ft x min(-1) (1524 m x min(-1)) to a range of altitudes (226 to 378 mmHg) for 4 h after prebreathe times of varying length and with varying degrees of physical activity while at altitude.
Methods: Logistic regression was used to develop an expression for the probability of DCS [P(DCS)] using the Hill equation with decompression dose as the main predictor. Here, decompression dose is defined in terms of either the tissue ratio (TR) or a bubble growth index (BGI). Other predictors in the model were gender and peak exercise intensity at altitude.
Results: All three predictors (decompression dose, gender, and exercise intensity) were important contributions to the model for P(DCS).
Discussion: Higher TR or BGI, male gender, and higher exercise intensity at altitude all increased the modeled decompression dose. Using either TR or BGI to define decompression dose provided comparable results, suggesting that a simple TR is adequate for simple altitude exposures as an abstraction of the true decompression dose. The model is primarily heuristic and limits estimates of P(DCS) to only a 4-h exposure.