A probabilistic model of hypobaric decompression sickness based on 66 chamber tests

One consequence of the NASA tissue ratio (TR) model is that calculated probability of decompression sickness [P(DCS)] is constant in tests at different ambient pressures so long as the ratio of P1 N 2 to P2 is the same in each test ; P1N 2 is N 2 pressure in the 360 minute half-time compartment, and P2 is ambient pressure after decompression. We test the hypothesis that constant P(DCS) is better described by TRs that decrease as P2 decreases. Data were from 66 NASA and USAF hypobaric chamber tests resulting in 211 cases of DCS in 1075 exposures. The response variable was presence or absence of DCS while at P2. Explanatory variables were P1N 2 , P2, exercise at P2, (yes or no), time to DCS (failure time), and time to end of test in those without DCS (censored time). Probability models were fitted using techniques from survival analysis. The log likelihood for the two parameter log logistic survival model was -846 with only failure and censored times, -801 when TR [P1N 2 /P2] plus exercise were added, and -663 when modified TR [(((P1 N 2 + C1)/P2)-1) c2 ) plus exercise were added, where c1 and c2 are fitted parameters in the five parameter model. Constant P(DCS) was better described by TRs that decrease as P2 decreases ; a conclusion supported by additional empirical observations, and bubble growth models that are independent of DCS data. Exercise increased the P(DCS) at P2. As a description of decompression dose, the modified TR was superior to TR over a wider range of experimental conditions.