DCS--Playing the Odds

[boulderjohn]

In a recent article about decompression sickness (DCS), a well known diver called DCS "a statistical inevitability." Although he did not explain what he meant by that phrase, I have read similar statements before. Most people mean that because statistics show that a certain number of dives yield a certain number of DCS cases, as an individual diver's total dive count grows, so does the statistical likelihood that the next dive will be one that will put him or her in the chamber. That argument is often coupled with another common observation--since many cases of DCS (some say half) occur to people diving within limits of a decompression algorithm, staying within your limits does not protect you from that inevitability. As common as those arguments might be, they are not valid. They reveal a misunderstanding of the nature of statistics, probability, and DCS itself.

Statistical inevitabilities are associated with events that are identical or nearly identical to one another. Toss a coin 1,000 times, and the results should be nearly half heads and half tails. That is not at all true, however, when the events in that series are not that similar. To illustrate this, let's look at an argument that is structurally similar to the DCS argument. Every year in the United States, homeowners suffer from heart attacks after shoveling snow. If we were to take the average number of those heart attacks and divide it by the average number of homeowners, we should be able to identify the probability that any homeowner will get a heart attack shoveling snow, right? Of course not, because....

1. Homeowners in Florida are less likely than homeowners in Minnesota to be out shoveling snow.
2. A 25-year old homeowner in excellent health is less likely to suffer a heart attack than a 74-year old with coronary artery disease.

I would argue that there is no likelihood that a fit 25-year old Floridian will suffer a heart attack after shoveling snow in the coming year, regardless of what those statistics indicate.

In or Out of Limits?
Let's say that the oft-quoted statistic that half of DCS cases occur to people diving within limits is accurate. People seem to think this indicates that diving within limits makes no difference, but that difference is enormous. At least 95%, possibly more than 99%, of all divers dive within the limits of whatever decompression system they are using, meaning that this huge percentage is providing only half of the DCS cases, while the tiny group that is violating those limits is providing the other half. It also means that the statistical likelihood of getting DCS, which is incredibly small, is at least half of that small percentage if you stay within those limits. If the percentage of DCS cases per dive is only 0.02%, as I have heard many times, that means people diving within their limits only have a rate of 0.01%. That sounds small to people who do not understand math, but it is actually an enormous difference.

There is also a big difference between divers who are inside those limits. Compare a diver who does one dive in a day to a depth of 10 feet for 20 minutes to a diver who does 5 dives a day, with each one to the very edge of no decompression limits. All those dives are within limits, but they are certainly not equal in their risks. Those limits cover a wide range of possibilities, from absolutely, positively safe to right on the edge of danger. It is like the difference between army generals relaxing in the officer's lounge and a platoon in search of enemies on the front line--both are in the war zone, but the danger is not equal.

Those limits themselves have a great variation. Let's look at a diver who dives to 100 feet using the U.S. Navy tables, staying at that depth for 24 minutes. That diver has stayed within no decompression limits according to that system, but that dive has exceeded the limits of the PADI tables by 4 minutes, and the diver should be doing an emergency decompression stop and staying out of the water for a minimum of 6 hours. Technical divers doing decompression stops might be amazed to learn how very little testing has been done on some of the algorithms currently being used. It is very possible that some popular systems are putting divers much closer than they would like to DCS every time they dive, and they would have no way of knowing that. In some cases, being within a given system is actually no more than being a continuing part of an experiment in decompression.

Not All Divers Are Alike
The scientists who developed the PADI tables tested divers using Doppler bubble imaging to detect bubbles in the blood. They were very surprised to see what a difference there was between divers. Some developed bubbles much more easily than others. For reasons that are not at all understood, some divers seem to have a much greater likelihood for developing DCS than others. One condition that is believed by some to put divers at greater risk is a PFO, a naturally occurring hole in the heart that allows bubbles in the venous blood to pass through to the arterial blood without first going through the lungs. The degree to which that is true is debatable, but what is not debatable is that for whatever reason, some people are simply more likely to get DCS than others.

Summary
Unlike flipping a coin, a dive being performed by one diver is not statistically equal to other dives being performed by other divers. Thus, the likelihood of getting DCS varies greatly from one diver to another and from one dive to another. Fit divers who keep their dive profiles well within accepted limits are by no means on a statistically inevitable path toward DCS.

 

[BRT]

In a recent article about decompression sickness (DCS), a well known diver called DCS "a statistical inevitability." Although he did not explain what he meant by that phrase, I have read similar statements before. Most people mean that because statistics show that a certain number of dives yield a certain number of DCS cases, as an individual diver's total dive count grows, so does the statistical likelihood that the next dive will be one that will put him or her in the chamber. That argument is often coupled with another common observation--since many cases of DCS (some say half) occur to people diving within limits of a decompression algorithm, staying within your limits does not protect you from that inevitability. As common as those arguments might be, they are not valid. They reveal a misunderstanding of the nature of statistics, probability, and DCS itself.

Statistical inevitabilities are associated with events that are identical or nearly identical to one another. Toss a coin 1,000 times, and the results should be nearly half heads and half tails. That is not at all true, however, when the events in that series are not that similar. To illustrate this, let's look at an argument that is structurally similar to the DCS argument. Every year in the United States, homeowners suffer from heart attacks after shoveling snow. If we were to take the average number of those heart attacks and divide it by the average number of homeowners, we should be able to identify the probability that any homeowner will get a heart attack shoveling snow, right? Of course not, because....

1. Homeowners in Florida are less likely than homeowners in Minnesota to be out shoveling snow.
2. A 25-year old homeowner in excellent health is less likely to suffer a heart attack than a 74-year old with coronary artery disease.

I would argue that there is no likelihood that a fit 25-year old Floridian will suffer a heart attack after shoveling snow in the coming year, regardless of what those statistics indicate.

In or Out of Limits?
Let's say that the oft-quoted statistic that half of DCS cases occur to people diving within limits is accurate. People seem to think this indicates that diving within limits makes no difference, but that difference is enormous. At least 95%, possibly more than 99%, of all divers dive within the limits of whatever decompression system they are using, meaning that this huge percentage is providing only half of the DCS cases, while the tiny group that is violating those limits is providing the other half. It also means that the statistical likelihood of getting DCS, which is incredibly small, is at least half of that small percentage if you stay within those limits. If the percentage of DCS cases per dive is only 0.02%, as I have heard many times, that means people diving within their limits only have a rate of 0.01%. That sounds small to people who do not understand math, but it is actually an enormous difference.

There is also a big difference between divers who are inside those limits. Compare a diver who does one dive in a day to a depth of 10 feet for 20 minutes to a diver who does 5 dives a day, with each one to the very edge of no decompression limits. All those dives are within limits, but they are certainly not equal in their risks. Those limits cover a wide range of possibilities, from absolutely, positively safe to right on the edge of danger. It is like the difference between army generals relaxing in the officer's lounge and a platoon in search of enemies on the front line--both are in the war zone, but the danger is not equal.

Those limits themselves have a great variation. Let's look at a diver who dives to 100 feet using the U.S. Navy tables, staying at that depth for 24 minutes. That diver has stayed within no decompression limits according to that system, but that dive has exceeded the limits of the PADI tables by 4 minutes, and the diver should be doing an emergency decompression stop and staying out of the water for a minimum of 6 hours. Technical divers doing decompression stops might be amazed to learn how very little testing has been done on some of the algorithms currently being used. It is very possible that some popular systems are putting divers much closer than they would like to DCS every time they dive, and they would have no way of knowing that. In some cases, being within a given system is actually no more than being a continuing part of an experiment in decompression.

Not All Divers Are Alike
The scientists who developed the PADI tables tested divers using Doppler bubble imaging to detect bubbles in the blood. They were very surprised to see what a difference there was between divers. Some developed bubbles much more easily than others. For reasons that are not at all understood, some divers seem to have a much greater likelihood for developing DCS than others. One condition that is believed by some to put divers at greater risk is a PFO, a naturally occurring hole in the heart that allows bubbles in the venous blood to pass through to the arterial blood without first going through the lungs. The degree to which that is true is debatable, but what is not debatable is that for whatever reason, some people are simply more likely to get DCS than others.

Summary
Unlike flipping a coin, a dive being performed by one diver is not statistically equal to other dives being performed by other divers. Thus, the likelihood of getting DCS varies greatly from one diver to another and from one dive to another. Fit divers who keep their dive profiles well within accepted limits are by no means on a statistically inevitable path toward DCS.

Great post!

 

[MargaritaMike]

Fit divers who keep their dive profiles well within accepted limits are by no means on a statistically inevitable path toward DCS

Since you have done all of these extrapolations, which are impressive by the way, and have drawn conclusions from them, I have only one question. What is the definition of "Fit".

Cheers - M²

and

 

[KenGordon]

In a recent article about decompression sickness (DCS), a well known diver called DCS "a statistical inevitability." Although he did not explain what he meant by that phrase, I have read similar statements before. Most people mean that because statistics show that a certain number of dives yield a certain number of DCS cases, as an individual diver's total dive count grows, so does the statistical likelihood that the next dive will be one that will put him or her in the chamber. That argument is often coupled with another common observation--since many cases of DCS (some say half) occur to people diving within limits of a decompression algorithm, staying within your limits does not protect you from that inevitability. As common as those arguments might be, they are not valid. They reveal a misunderstanding of the nature of statistics, probability, and DCS itself.

Statistical inevitabilities are associated with events that are identical or nearly identical to one another. Toss a coin 1,000 times, and the results should be nearly half heads and half tails. That is not at all true, however, when the events in that series are not that similar. To illustrate this, let's look at an argument that is structurally similar to the DCS argument. Every year in the United States, homeowners suffer from heart attacks after shoveling snow. If we were to take the average number of those heart attacks and divide it by the average number of homeowners, we should be able to identify the probability that any homeowner will get a heart attack shoveling snow, right? Of course not, because....

1. Homeowners in Florida are less likely than homeowners in Minnesota to be out shoveling snow.
2. A 25-year old homeowner in excellent health is less likely to suffer a heart attack than a 74-year old with coronary artery disease.

I would argue that there is no likelihood that a fit 25-year old Floridian will suffer a heart attack after shoveling snow in the coming year, regardless of what those statistics indicate.

In or Out of Limits?
Let's say that the oft-quoted statistic that half of DCS cases occur to people diving within limits is accurate. People seem to think this indicates that diving within limits makes no difference, but that difference is enormous. At least 95%, possibly more than 99%, of all divers dive within the limits of whatever decompression system they are using, meaning that this huge percentage is providing only half of the DCS cases, while the tiny group that is violating those limits is providing the other half. It also means that the statistical likelihood of getting DCS, which is incredibly small, is at least half of that small percentage if you stay within those limits. If the percentage of DCS cases per dive is only 0.02%, as I have heard many times, that means people diving within their limits only have a rate of 0.01%. That sounds small to people who do not understand math, but it is actually an enormous difference.

There is also a big difference between divers who are inside those limits. Compare a diver who does one dive in a day to a depth of 10 feet for 20 minutes to a diver who does 5 dives a day, with each one to the very edge of no decompression limits. All those dives are within limits, but they are certainly not equal in their risks. Those limits cover a wide range of possibilities, from absolutely, positively safe to right on the edge of danger. It is like the difference between army generals relaxing in the officer's lounge and a platoon in search of enemies on the front line--both are in the war zone, but the danger is not equal.

Those limits themselves have a great variation. Let's look at a diver who dives to 100 feet using the U.S. Navy tables, staying at that depth for 24 minutes. That diver has stayed within no decompression limits according to that system, but that dive has exceeded the limits of the PADI tables by 4 minutes, and the diver should be doing an emergency decompression stop and staying out of the water for a minimum of 6 hours. Technical divers doing decompression stops might be amazed to learn how very little testing has been done on some of the algorithms currently being used. It is very possible that some popular systems are putting divers much closer than they would like to DCS every time they dive, and they would have no way of knowing that. In some cases, being within a given system is actually no more than being a continuing part of an experiment in decompression.

Not All Divers Are Alike
The scientists who developed the PADI tables tested divers using Doppler bubble imaging to detect bubbles in the blood. They were very surprised to see what a difference there was between divers. Some developed bubbles much more easily than others. For reasons that are not at all understood, some divers seem to have a much greater likelihood for developing DCS than others. One condition that is believed by some to put divers at greater risk is a PFO, a naturally occurring hole in the heart that allows bubbles in the venous blood to pass through to the arterial blood without first going through the lungs. The degree to which that is true is debatable, but what is not debatable is that for whatever reason, some people are simply more likely to get DCS than others.

Summary
Unlike flipping a coin, a dive being performed by one diver is not statistically equal to other dives being performed by other divers. Thus, the likelihood of getting DCS varies greatly from one diver to another and from one dive to another. Fit divers who keep their dive profiles well within accepted limits are by no means on a statistically inevitable path toward DCS.

What does your last line actually mean?


Btw apparently it has been snowing

It Snowed in Florida and People Lost Their Minds

I have a relative who had a heart attack at age 32. As far as I know he was not in Florida nor shovelling snow. He was fit though.

Special precautions are now taken to screen UK football players as so many were falling over dead. You can be extremely fit and still have an undiagnosed heart issue.

 

[Neilwood]

@boulderjohn - excellent description of it. Even within NDL diving there is an option to "pad" the safety margin by sticking a bit further from the limit - someone who hugs the NDL within 1 min at all depths is going to be at more risk than someone who never gets within 5 mins of the NDL on the same algorithm.

Another thing that I think is often overlooked is ascent rate. A diver can be well within the numbers for the particular algorithm they have chosen (stayed well away from NDL) but if they aren't careful about ascent rate it can cause off gassing at an unsafe rate (particularly closer to the surface where pressure differentials increase). Ascending at 10m/min might be safe at 30-20m but it is not as safe at 5-0m where a slower ascent would be a better idea.

 

[boulderjohn]

What does your last line actually mean?

My last line was:

"Fit divers who keep their dive profiles well within accepted limits are by no means on a statistically inevitable path toward DCS."

It is a simply contradiction of the idea of "statistical inevitability," the idea that everyone will get DCS if they dive enough times, a belief that assumes that everyone's chances of getting it are about even.

There was a ScubaBoard debate about this maybe 5-6 years ago. A frequent poster, long time diver, and sometime official with a major diving agency said essentially that in pretty much so many words. He came up with the statistic that DCS occurs about once in every 2,000 dives, a number I believe is dramatically wrong to begin with. He said that therefore, according to the laws of probability, anyone with 2,000 dives who has not yet had DCS is defying the odds. In response, I made the argument I did at the beginning of this thread, and he responded by saying I was an idiot who did not understand statistics. If the average of all dives is 1 out of 2,000, then, by golly, all divers have a 1 in 2,000 chance of getting DCS on every dive.

 

[KenGordon]

But does that inform anyone at all?

For the sake of argument let’s say that there are two sorts of diver, those that will get bent one time in 10000 benign dives and those that will never get bent by those dives.

Now let’s take a new, innocent SB member just out of AOW. Which group are they in?

Now, let’s assume that there is one group with a chance on one in 20k. How should the new AOW change their behaviour vs the other assumption?

This is not like shovelling snow where you can say there is no snow so no risk.

 

[tursiops]

If the average of all dives is 1 out of 2,000, then, by golly, all divers have a 1 in 2,000 chance of getting DCS on every dive.

This whole discussion should be using Bayesian statistics rather than frequency statistics. You are effectively using Bayesian; the unnamed official was using frequency.

 

[jeff h]

How should the new AOW change their behaviour vs the other assumption?

For starters be fit, hydrated, avoid fatigue and follow conservative profiles and decompression. Minimize physical injures and calcium buildups. I think the statements being made are more about physiology and fitness level than actual in water experience. In other words, a fit diver is probably less likely to get bent than an unfit diver. So if DCS occurs ever 1/2000 dives, and two unfit diver get bent, then they've just made the quota for 4000 dives.

So if 1/2000 is the number for the entire diving population, all sizes, shapes and fitness levels, then what is the number for fit divers?

 

[KenGordon]

For starters be fit, hydrated, avoid fatigue and follow conservative profiles and decompression. Minimize physical injures and calcium buildups. I think the statements being made are more about physiology and fitness level than actual in water experience. In other words, a fit diver is probably less likely to get bent than an unfit diver. So if DCS occurs ever 1/2000 dives, and two unfit diver get bent, then they've just made the quota for 4000 dives.

So if 1/2000 is the number for the entire diving population, all sizes, shapes and fitness levels, then what is the number for fit divers?

How is that different to how he should behave in the other scenario?