As recreational divers and professional statisticians, we read with great interest the articles Dive Computers: some facts and fallacies by John Lippmann and DCI and dive computers by Lynn Taylor and Simon Mitchell, in the July to September 2000 issue of Alert Diver (South East Asia - Pacific edition, published by DAN SEAP).
The available statistical information about decompression illness (DCI) is notoriously difficult to interpret; we think both articles have done an excellent job of presenting and discussing these data.
However, we disagree with the Editor on one important point. Studies have reported that a very high percentage (as many as 93.7%) of divers with DCI claimed they were diving within the limits of their dive computers or tables. The Editor notes that this
"... indicates that the algorithms used by the computers to predict DCI are inaccurate."On the contrary, we believe this statistic carries no information about the safety factors of tables or computers. Instead it suggests that at least 93.7% of the general diving population are diving within the limits of their tables or computers, which should be comforting.
[Editor's Note commenting on article by Taylor and Mitchell]
"Safe" dive profiles still carry a risk of DCI, so we should not be surprised to meet a diver who has suffered DCI while diving within the limits of the tables or computer. In a survey of divers with DCI, the percentage who were diving within the limits of tables or computers depends on the percentage of the whole diving population who dive inside the limits, as well as on the relative risks of DCI.
As an extreme and unrealistic example, suppose 100% of all divers stay within the limits of their tables or computers. Then a survey of DCI cases would show that 100% of all DCI cases dived within the limits. Obviously, this does not mean that their computers or tables are inaccurate; it just reflects the fact that even dives within the limits of one's table or computer still carry a risk of DCI.
More realistically, suppose that 99% of all dives are done within the limits, and the relative risk of DCI is ten times greater when diving outside the limits. Then about 90.8% of all cases of DCI will occur amongst divers who dived within the limits. Again this would not indicate a failure of the tables or computers to "predict" DCI.
Amongst cases of DCI, there is an over-representation of people who dived outside the limits, because their DCI risk is higher than for people who dive inside the limits. The fact that 93.7% of all DCI cases dived inside the limits tells us that at least 93.7% of the overall diving population dives inside the limits.
The following table shows the interpretation that can be put on the reported survey of DCI cases, under different assumptions. For example if we assume the risk of DCI is ten times higher for a dive outside the limits, then the report that 93.7% of all DCI cases dived inside the limits tells us that about 99.3% of the diving population stays within the limits.
| risk ratio | 5:1 | 10:1 | 20:1 |
| % of diving population diving within limits | 98.7% | 99.3% | 99.7% |
To determine the reliability of dive computers we would need additional data, allowing us to measure the rates of DCI under different diving conditions. A table or computer should be considered reliable if it correctly predicts the risk of DCI. This is difficult to assess experimentally because of the low rates of DCI, the difficulty of complying with dive plans and ascent rates, the number of environmental variables, individual variability, and sampling bias (different rates of reporting).
It will remain difficult to measure the reliability of dive computers and tables until there is more accurate baseline information about the recreational diving population and its activities. This is why it is vital for divers to support the excellent initiatives and activities of DAN in collecting survey data from "ordinary" dives.
Authors:
Prof Adrian Baddeley
and Dr Berwin Turlach