[abstract] DECOMPRESSION SICKNESS (DCS) RISK IS INFLUENCED BY DIVE CONDITIONS AS WELL AS BY DEPTH-TIME PROFILE.

Abstract

BACKGROUND: Understanding the relationship of DCS risk to depth-time exposure requires accurate knowledge of dive profiles, dive conditions, and dive outcomes. This is the purpose of DAN's Project Dive Exploration (PDE), which uses dive computers to continuously record depth-time profiles. MATERIALS AND METHODS: Three dive groups were analyzed: 51,497 warm-water dives from PDE; 6,527 cold-water dives from PDE; and 2,252 U.S. Navy dive trials. DCS risk was predicted for all dives with a probabilistic decompression model calibrated to Navy trials (UHM. 1997;24:275-292). Predicted and observed DCS incidences were then compared for the three groups. Finally, predicted risks were recalibrated to all dives by logistic regression. Possible (but not necessarily known) differences in conditions between groups were represented by separate parameters. RESULTS: The mean observed DCS incidences were 2, 28, and 311 cases per 10,000 dives for warm-water, cold-water, and USN trials, respectively. Probabilistic model predictions were accurate for USN trials but significantly overestimated the observed DCS incidence for warm and cold-water dives (51, 75, and 351 per 10,000, respectively). With recalibration to all dive data, the mean observed and predicted DCS incidences were the same for each dive group (2, 28, and 311 per 10,000, respectively). CONCLUSIONS: The Navy trials were conducted at much higher predicted risks than were the warm or cold-water dives, but even at a given risk, the groups differed significantly: the recalibrated risk of the Navy trials exceeded that of the cold-water dives by a factor of two and exceeded the risk of the warm-water dives by a factor of 30. We conclude: (a) probabilistic models were useful for estimating DCS risk; (b) models may not extrapolate well beyond the data to which they were calibrated; and (c) dive conditions, in addition to dive profiles, may significantly affect DCS risk.