The Validation of Dive Computer Decompression Safety

How do you determine what's safe and what's not?

Over the last two decades, electronic dive computers have replaced decompression tables in most segments of recreational diving. Yet during the same time, the overall incidence of decompression sickness (DCS) does not appear to have changed, dispelling early worries that abandoning tables to dive with computers would result in increased DCS.

There are dozens of dive computer models on the market. Their effective use depends on sound design, quality of manufacturing and safety of their decompression algorithms. Questions about algorithms often overshadow all other safety issues. However, dive computer manufacturers generally do not provide information about their algorithms, their use in particular computer models or their impact on DCS risk. There are at least two reasons for this: 1) dive computers are not regulated; 2) validation of decompression safety is complicated and expensive. Thus, in most cases manufacturers do not have the data necessary to support claims of risk control or risk reduction — an important issue for divers.

The current state of dive computer validation is uncertain, but it is likely that all manufacturers do test dive computers to some extent, to protect both users from injury and themselves from liability. The lack of published material may stem from apprehension regarding a lack of hard data.

Safe dive computers must maintain a future DCS incidence level at or below the current DCS rate. DAN® has observed the current rate to be 0.01–0.04 percent in recreational diving; this rate was determined by data gathered through DAN's Project Dive Exploration (PDE) research project. Human decompression trials based on DCS symptoms require hundreds of repetitions for a single dive profile and thousands for a wide depth-time range. Testing is further complicated by the heterogeneity of recreational divers as well as by the variability of individual responses to decompression. One group of divers may have different outcomes than another, so it is important for computer manufacturers to test divers and profiles representative of the product's expected users.

Historical data from the U.S. Navy, DCIEM, COMEX, Duke University, the University of Pennsylvania and DAN's PDE can be used for retrospective testing. Most historical data pertain to square profiles, with relatively high DCS risk, obtained from the sample data of military or commercial divers. These data are more suitable for calibrating dive computers aimed at technical diving with mandatory decompression stops than for multilevel, no-decompression recreational dives.

Divers and dive magazines sometimes try to evaluate dive computers of interest by benchmarking them against decompression tables or another dive computer they trust. Benchmarking dive computers against decompression tables may help for square profile dives but is of limited value for low-risk multilevel dives. Moreover, there is no dive computer with a published DCS incidence (or risk), and so no dive computer is a valid choice for a benchmark evaluation of another computer in multilevel diving.

Divers who stretch their dive computers to the limits and take lots of DCS hits may over time tell which dive computer is better for them, but that is a risky approach, and their experiences may not benefit other divers.

It is important to recognize that most manufacturers regard the U.S. Navy decompression tables (or other tables with a known or assumed safety record) as a high-end risk benchmark to which they add a safety margin according to their own understanding of DCS risk. Based on no-stop ascent limits, it may be reasonable to assume that one dive computer is safer than another, but that is not necessarily true; the shape of a dive profile may affect decompression safety more than depth and bottom time. For a final evaluation of a specific dive schedule generated by dive computers, some human subject dive trials may be necessary.

Validation of decompression safety may be sped up by using venous gas emboli (VGE) as the outcome of interest in place of DCS. Venous gas bubbles may be detected in divers without DCS symptoms. When bubbles become abundant (high VGE grade), the risk of DCS increases. Decompression trials that use VGE grade as the outcome do not provide the same insights as studies that include actual cases of DCS. However, VGE-based trials require fewer repetitions, may be completed in a shorter time and reduce the DCS risk in their subjects. Properly used, VGE outcome data may make decompression safety validation affordable.

We asked our panel of experts to answer a few basic questions regarding validation of decompression safety.

What are the acceptable measures of decompression safety?

Sergio Angelini: Acceptable measures vary depending on the category of diving. For recreational diving we aim to have zero systematic events, meaning that those few DCS [cases] we do observe are either due to diver error (fast ascent, missed deco stop, etc.) or so-called undeserved hits (those without an apparent explanation).

Karl Huggins: There are three:
  1. The incidence of DCS associated with pushing the dive computer to the limits of its decompression algorithm
  2. Comparison of dive computers against a set series of dive profiles that have known risk outcomes
  3. Approximation of operational DCS incidence from dives performed with the dive computer within observational studies like Project Dive Exploration. However, since the majority of divers do not always push their dive computer to the limit, the estimated operational risk will be lower than the risk associated with pushing the decompression algorithm to its limit.

Martin Sayer: The acceptable measures of decompression safety are those that have a DCS probability close to zero and take into account the differences in individual susceptibility and accidents. However, it's well known that differing models of decompression computers range in how they measure decompression safety as the variation in no-stop times. What is not known is how different computers assess decompression safety and how same-day and/or multiday repetitive dives alter the risk.

Richard Vann: DCS is the primary measure of decompression safety, but DCS severity must be considered as well as DCS incidence. Doppler-detected VGE are a secondary measure of decompression safety.

Is there a "gold standard" of decompression safety, and if there is not, what standard can be used instead?

Huggins: There is no single decompression safety standard in the development of dive computers. In a perfect world, dive computer manufacturers would perform statistically significant human-subject tests to the limits of the dive computer and publish their results. For the alternatives, we need to define a minimum decompression safety validation procedure.

Vann: There is no gold standard. Decompression safety depends on what the user judges to be an acceptable DCS incidence, but acceptable incidence differs for mild and serious DCS. For example, current judgment by the U.S. Navy is that 2 percent is acceptable for mild DCS, while 0.1 percent is acceptable for serious DCS. Commercial divers in the Gulf of Mexico, on the other hand, wish to have not more than 0.1 percent mild DCS and 0.025 percent serious DCS. Users and user groups must decide for themselves what they consider acceptable.

Sayer: There are certainly computer models on the market that give more bottom time than others, but it is often not clear what the consequential penalties are for this within an extended dive series. There are a number of published decompression models that could be used as a "gold standard." However, for comparisons between complex multilevel, multidive algorithms and empirically tested decompression tables, H.V. Hempleman's pressure/root time relationship provides a simple tool.

Angelini: We need to differentiate between the types of diving. Recreational no-stop diving, excluding square dives, meets the acceptable measure as defined in my answer to the question on that subject. Most of recreational diving is controlled by dive computers. In an ongoing study, we compared three major brands over more than 120 nonrepetitive, no-stop profiles and found a remarkable agreement. So we could say that those computers do represent one (and the same) gold standard, as long as a safety stop at the end of the dive is included. We continue testing for repetitive, decompression and square dives.

Please define the minimum decompression safety validation procedures for new dive computers.

Sayer: Standard decompression models used in dive computers are often stated as being "modified" versions, and, in many cases, safety levels may be changed by users. It would be helpful to know what the quantifiable effects of different levels of user-controlled conservatism were. It is unlikely that manufacturers would or could publish this type of information, but at least they should publish test results of standardized pressure/time profiles applied in single and multiple dives as well as multiday series.

Vann: This is a matter of judgment. The traditional method would be to conduct chamber trials under the expected conditions of use, but this is impractical for most organizations due to cost and time constraints. Another method might be to make a series of predetermined man-dives at the predicted safety limits of the algorithm.

How many test dives should be conducted? In the 1940s and '50s, Navy chamber dive trials usually tested each dive profile four to 12 times. Recently this has increased to 50-150 dives per profile. In 1983 the Orca Edge was tested in 100 chamber dives of an unknown number of profiles. To my knowledge, no other dive computer has been tested in even 100 dives. When dive computers were first introduced, many dive physicians believed that the DCS incidence would increase drastically. This did not happen, and there is no evidence of any more DCS for dive computers than for dive tables, although the proportion of AGE among injured divers appears to have decreased. Even so, conducting 100-1,000 open-water dives before marketing would be desirable.

Unfortunately, there is a finite probability of DCS for most dives, and when more dives are conducted, some DCS becomes likely. Probabilistic decompression models calibrated to chamber dive trials and open-water dives will estimate DCS probability and allow individuals and communities to choose dive computers according to their own judgment of what the DCS incidence is acceptable or "safe."

Angelini: We must differentiate between new computers utilizing existing algorithms and those utilizing new algorithms. In the former case, it is simply a matter of ensuring that the computer behaves as intended when exposed to a multiplicity of environmental conditions, dive profiles and all possible settings of the dive computer. New decompression models, on the other hand, must be validated by scientific means and as such require the cooperation of universities, possibly the military and for sure organizations such as DAN.

Huggins: The first step would be to establish an industry standard that all dive computer manufacturers could adhere to. Since the likelihood of the manufacturers doing human subject tests is low, a protocol along the lines of the following would have a better chance of implementation:
  1. Run the decompression algorithm against a set series of dive profiles with known risk.
  2. Run the profiles allowed by the decompression algorithm through an established decompression risk model.
  3. Verify that the dive computer hardware and software run the decompression algorithm properly.
  4. Publish the results of the tests along with the risk estimates from each test profile.

Meet The Experts
Sergio Angelini, Ph.D., is research and development manager at Mares S.p.A. in Rapallo, Italy. Prior to this, he was general manager at Uwatec AG in Switzerland (2003-2008) and director of engineering at Scubapro North America (1998-2003). He was certified in 1992 and has logged more than 2,000 dives.

Karl Huggins, a NAUI Instructor since 1980, has been diving since 1977. He is noted for his work in decompression theory and models, which resulted in the University of Michigan Sea Grant (HUGI) dive tables. One of his most significant contributions to the diving community is as co-inventor of the Edge dive computer, one of the first commercially viable dive computers. Since 1992 Huggins has served as program director of the Catalina Hyperbaric Center at the USC Wrigley Science Center on Catalina Island, Calif. Huggins is the recipient of the 1990 Leonard Greenstone Diving Safety Award and the 1993 DAN/Rolex Diver of the Year award.

Martin Sayer, Ph.D., is head of the U.K. Natural Environment Research Council's National Facility for Scientific Diving. He is currently also head of the Dunstaffnage Hyperbaric Unit in Scotland. Sayer has published a number of peer-reviewed publications about the uses and misuses of decompression computers; he is the principal investigator for a study comparing the performance of present-day dive computers within an operational context. Sayer has been diving for more than 35 years and has logged more than 4,000 dives.

Richard D. Vann, Ph.D., is the vice president of DAN Research as well as an assistant research professor in anesthesiology, safety officer and director of applied research at the Duke Center for Hyperbaric Medicine and Environmental Physiology in Durham, N.C. Vann is responsible for planning and executing the DAN Research programs; he oversees the coordination of DAN Research activities with the Duke Center for Hyperbaric Medicine and Environmental Physiology and actively promotes DAN Research through speaking engagements, presentations to the dive industry and numerous publications related to diving.

© Alert Diver — Summer 2010