Saturday, December 22, 2012

Port Waratah Coal Service Cancer Cluster

After the forum on the 8th, I received an emailed question about the PWCS cancer cluster.  Cancer clusters are often difficult to interpret.   I have pasted the essence of my response to the question below.
The question asked - how cluster of cancers identified at the Port Waratah Coal Services be "statistically significant" - but also be described as "due to chance"?

The National Cancer Institute in the US provides excellent guidance on how to approach cancer clusters which may provide a useful background to the issue:

While the cluster of multipe cancers among the Kooragang PWCS workers  were "statistically significant" I personally consider "statistical significance" as  a screening step to discern if the occurrence of cancer clustering is more than you might expect and therefore worthy of further investigation - NOT as evidence that there is a local cause. It confirms clustering but not a common exposure.

The PWCS study primarily identifies a clustering of screening detected cancers. Some of them for e.g. the prostate cancers might never have been detected or have never caused illness unless the screening occurred.  You can see the huge disparity between diagnoses and deaths in the Hunter in the graph in this blog post. The blue line shows the massive changes in diagnoses based on PSA testing while the red lines, the deaths due to prostate cancer are very flat  - to me it says that at a group or population level total diagnoses are not predictive of real prostate cancer rates in the group or population.  I know this is hard to believe but diagnosed rates of prostate cancer are only loosely and variably related to actual cancer rates - again the attached graph displays this concept.  I know there is a delay between diagnosis and deaths but you can see that the huge spike in prostate cancer diagnosis in the early '90s (due to increased PSA testing) has not resulted in increased deaths even 15 years later.

PWCS has engaged groups to do more prostate cancer screening of their workers so the rate of prostate cancer diagnosis will now increase even higher I expect.  This needs to be understood because if they run the data again in 12 months it will look as if the problem has got worse.

So, getting back to the issue as to whether there is a common cause/exposure at Kooragang.  The next step after establishing an increase is to look for a common cancer - usually occupational exposures are VERY specific in the cancers they cause.  e.g. aniline dyes and bladder cancer, potassium dichromate and respiratory cancers.  See this NIOSH case study to see how the peak occupational health institute in the US approaches cancer clusters:

So the Kooragang cancers were a mix of cancers, and mostly screening related and this really casts doubt on a work site cause.  Most investigations would stop there and probably should.

So if you thought there was an increased rate of cancer the next step is to look for a biologically plausible exposure for even one of these cancers - not likely to be an exposure that could cause all of them, cancers tend to have different disease triggers and promotion pathways.

So if you look at a coal loader environment you are primarily looking at coal dust - so you would be looking for cancers associated with coal dust.  So if we turn to the peak international agency that defines what is or isnt a carcinogen - the International Agency for Research on Cancer IARC is the place to go. See here:   They have reviewed multiple studies some of which show a relationship with cancer some of which do not. 

IARC are not able to find conclusive evidence that coal dust is a carcinogen in the setting of undergound mines where the levels of coal dust have been hundreds of times higher than you would expect in a coal loader setting. So that makes it very unlikely that it could be a carcinogen in this setting. Of course, it may be that IARC changes their mind on coal dust as a carcinogen over time with new evidence. Usually occupational cancer agents are easy to identify when the list of "high chemical exposures" are reviewed.  I know there is to be continued review of the PWCS site - it will be interesting to see if that can identify a plausible biological exposure pathway.

People have expressed concern about process water used on site - it is hard to imagine a plausible biological pathway for this water, they are not drinking it and the level of exposure from aerosolisation and spray in the air would be incredibly small.  I understand they are further examining the process water.

When it comes down to it, the most plausible explanation is a cluster of increased screening, it is impossible to say it is  for certain, but I would put it way at the top of the list - we know this happens, its plausible.  There has been no investigation of differential screening practices between the two groups - were there lunchroom conversations about  screening or did some of the workers share the same general practice? (many people ask their friends if they can recommend a doctor and some practices support PSA sceening some do not). It is interesting that the follow up investigation is via a hygienist and not to rule out the most plausible explanation - enhanced screening.  I guess the rationale behind that is to rule out a chemical exposure - the exposure most feared, rather than confirming enhanced screening and diagnosis as the cause of the cluster. However, confirming enhanced screening as the cause of the cluster might put a lot of peoples mind at rest.  As mentioned previously,  I expect there will now be an increase in prostate cancer diagnoses at both PWCS sites due to the screening programs which may increase worker anxiety.

No comments: