The Devil is in the Detail…

This weekend has been a bad time for UK diving with three confirmed fatalities and one missing diver, presumed dead. The incidents all happened in different locations and the only real common factor at this stage was that they were undertaking diving either at the time of the accident or around the time of the accident.

Unfortunately 2014 has been a bad year with 14 fatalities so far (which I have details of), 6 of which were in Ireland in one month over the summer and 4 this weekend just gone. Any fatality is a loss to friends and family and my thoughts go out to all those involved. Those who undertake the sport recognise that there are risks, some more appreciative than others.

The Risk

The mathematical mean for fatalities in the UK from 1998 to 2013 is 16.45 (range 10-24, SD 4.30), this equates to a fatality risk of 0.71:100 000 dives per year based on BSAC membership, statements made as to the proportion of BSAC members as part of the UK diving population and how many dives they have conducted (using 2007 BSAC survey data). However, using stats like this is meaningless unless you know the sort of diving taking place and the experience levels and currency of those involved therefore I am not a great believer in 1:X fatality rate as a measure of the safety of a recreational activity, especially when we have no baseline to compare against.

The Detail

I get asked on a number of occasions “Do I have any more details on causality for these incidents or indeed across the wider incident details that I collect through DISMS ( or through my PhD research?” The simple answer is no because the level of detail recorded in incident reports is very superficial, if they are collected at all.  Forums are great sources but there are biases at play with not all of the facts being provided, especially if they compromise credibility.

I believe that the number of reported incidents is less than 10% of ‘reportable’ incidents occurring. Given this lack of detailed data, trying to ascertain causality and trends is hard, if at all possible. However, that doesn’t mean we shouldn’t collect data, especially non-fatal incident data.

Data Provision

A number of organisations collect incident data but this is in the main confidential and they produce edited versions of the reports (DAN and BSAC for example). Even some of the data that is publicly available cannot be reproduced for copyright reasons. DISMS has been developed to allow the reporter to determine the level of disclosure and I, as the admin, always respect that. I have a long-term goal to try and create a single diving incident data repository which has data feeds from other organisations, something along the lines of ASRS, Aviation Safety Network or Skybrary and I have been in contact with three agencies so far about data provision.  More information as I get it.

Part of my PhD research has been to develop a model which can be used to provide causality attribution and I hope that this can be adopted for such a global database. For example, the BSAC annual report covers broad headings such as DCI, Buoyancy Issues, Fatality etc, but these are too broad to try to determine trends. e.g. a diver may run out of gas, make a rapid ascent and then have DCI. This would likely be classified as a DCI incident as it is the most serious classification. DISMS allows multiple tags to be attributed to each incident (including dataset ownership); this will allow detailed searches to be conducted across multiple datasets e.g. “CCR incidents in BSAC data” or “overhead incidents in DAN data”, or “CCR overhead incidents” across the whole database.

Why Incidents Occur?

My presentation at Eurotek 2014 ( will present some of the evidence from a survey conducted earlier this year. This survey went much further than looking at configuration, depth, time, gases and gender and looked at the decisions divers made and what human factors contributed to their non-fatal incident. There is already evidence from other domains such as aviation and oil & gas that equipment failure is not the major contributory factor in incidents and that human decision making is; there is no reason to believe that SCUBA diving would be any different.

How you can help improve things?

We can learn from other’s mistakes. If you report your errors/mistakes others can learn from you. However, first you need to acknowledge that you have made a mistake or an error in the first place! This is very hard sometimes, especially when face or pride may be lost! The majority of incidents I see are down to ‘violations’ of best practice, and no-one wants to admit to not following best practice, especially when it nearly kills you!

However, if you have had an incident and you fill in a report on DISMS or BSAC Incident report (or whatever your local reporting system is), try to remember to add as many details as you can remember as soon after the incident as possible; there is significant evidence that memory fails as time passes, and the ‘facts’ change too. If you have been involved in an incident, get all the facts down straight away (times, depths, gas, equipment config etc), in effect a mind-dump, because you will forget details. Try not to talk to anyone else about the incident before you do this too (difficult I know) but a group consensus will be different to the views of individuals who were involved.

The best way to improve safety, recognise your limits, acknowledge them and don’t exceed them. Calling a dive because you are not ready or not happy is nothing to be ashamed of. As the popular saying goes “Better to be up on the surface, wishing you were down below, than down below wishing you were on the surface!”

Safe Diving…

3 responses to “The Devil is in the Detail…

  1. Hello as a professional in making decisions from dodgy data I have to say we have no statistical data worth a damn on diving incidents.

    Part of the problem is that incidents are – fortunately – pretty rare which means we are looking at small populations and the statistics involved with rare events are notoriously difficult to deal with.

    The second part of the problem is that the incident reports are self-selecting; we all know of incidents that didn’t get reported: uncontrolled ascents or missed stops with no ill effects for example. So we should probably only look at fatalities as the only group that do get fully reported which makes the rarity even greater.

    The final part of the problem is we know a bit – not much – about a small population of divers and dives but that is a highly biased sample as it is those divers who made it into the incident reports – by definition high risk divers. Now if we knew anything at all about the diving population as a whole we could compare the incident divers to the rest of the population to see if there were any significant drivers – either in the type of diver (age sex qualification fitness dived up etc.) or the type of dives (OC/CCR, planned depth, etc.). But we dont even have an estimate of how many dives take place a year let alone anything more detailed.

    The BSAC “estimate” is a joke – “we asked folk at a few popular diving sites on a few sunny days and extrapolated wildly” which in my book gets you an estimate of the population of divers who dive at those sites on sunny days and nothing more. Unless someone gets a handle on the detail of the diving/diver population as a whole – and maybe that would have to be BSAC as they at least have the breadth to do it – we should simply forget about trying to draw any statistical inferences from the incident reports and do like the airline industry does – look at the incidents in detail and draw conclusions.

    • Peter, thanks for the feedback. I agree with everything you have said apart from

      “Unless someone gets a handle on the detail of the diving/diver population as a whole – and maybe that would have to be BSAC as they at least have the breadth to do it – we should simply forget about trying to draw any statistical inferences from the incident reports and do like the airline industry does – look at the incidents in detail and draw conclusions.”

      because I do not believe that BSAC have the breadth to do it with their membership less than half of the UK diving population (their figures); I also believe that there would be some resistance in them having access to other data. I tried to get a survey of UK diving undertaken but it never materialised due to other pressures and lack of support.

      However, I do agree that statistical analysis on the samples we have is almost worthless but I don’t agree that we can examine incidents/accidents in detail and then draw conclusions especially given the small sample size. The airlines don’t look at incidents in isolation, they collect vast amounts of meta-data from non-incident and incident flights and then use that along with very intensive and costly investigations of incidents and accidents to draw conclusions of how things can be improved. Neither of which is possible from within the recreational diving industry.

  2. Pingback: Lies, damned lies, and statistics… | Cognitas Incident Research & Management·

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s