I am also fairly certain that she was informed in her pursuit of her MPH about sampling and analytical bias, particularly bias caused through operator error - which is why we like lots and lots of 'n' as well as replicate samples. As a Ph.D. candidate (PhDc), she should have had more exposure to this concept than us regular ol' MPHs. So it is a bit difficult to defend her sampling plan and methodology which she and Wilma Subra used to formulate their conclusion which was submitted to the citizens of the Town of DISH, Texas.
So here I am, once again, having to ignore all of that potential error and bias (scientific) and make an assumption as to its validity. The TCEQ had to acknowledge the same thing in their memo regarding Wolf Eagle Environmental's sampling results. So, as it stands, I must assume that the data obtained on August 17, 2009, is an accurate representation of what a citizen of DISH, Texas is exposed to.
Because a comparison of these collected values were made against values established by the TCEQ as 'safe' (ESL), and the reporting by both Ms. Rich and Ms. Subra that there were 45 incidents where the values obtained from analysis of seven different sample locations exceeded these safe values, the confidence in the data they collected and analyzed is paramount. Unfortunately, it is not, Still, it is all we have, and the cat has been let out of the bag regarding the potential health impact that could come about because of these values reported along with how this information was laid out in their conclusions.
So how does one work through this? What should have taken place on August 17, 2009. were seven "instantaneous" samples collected at least twice during the day. Analysis costs money, so the fewer the samples the less the cost. But, this also brings in bias, which is why researchers always want more data. Had instantaneous samples been collected instead of leaving the valve open for 24 hours, a more clear comparison to a "safe" value could have been made because ESLs and AMCVs are based on one hour sample collections in terms of impact from exposure.
Instead we are presented with analytical results that show contaminant ppb concentrations that were collected over a 24 hour period. Although I am perfectly within my right to take this 24 hour value and divide it by 24, by doing so, I will dilute any spikes that may have happened. This is why the TCEQ monitors these contaminants on an hourly basis. Because you may exceed a threshold during one hour period which would be diluted by all the other hours collected. Averaging is important, but so is knowledge of when - or if - thresholds are being exceeded during certain times of the day.
The seven samples - had they been instantaneous - would have been a snapshot of that particular time and day. Even had they purposely biased the sample by choosing the most ideal conditions, it would have been of more use than a sample collected over a 24 hour period when trying to look at potential risk for a one hour period. To counter this problem with Wolf Eagle Environmental's methodology, the TCEQ in their memo made the assumption that the 24 hour values obtained reflected a one hour sampling event that was diluted by 23 hours of air that contained no contaminants. Why? Because instantaneous samples are a snapshot, 24 hours are an average. It is just as possible that this scenario happened as it was that the levels were uniform throughout the day at which case they could have been divided by 24 (my way).
To determine the potential 1-hour maximum concentrations of all reported chemicals, the reported 24-hour concentration was multiplied by 24. This calculation conservatively assumes that the chemical had a maximum concentration for one hour and was not detected for the remaining 23 hours.And therein lies the problem with how we determine risk. We make assumptions most always on the side of 'worst case' or upper bound limits just to be on the safe side. This compounded conservatism gives us values that are extremely low so that we health and safety types can reasonably tell you "it's safe."
If you read Wolf Eagle Environmental's final report as well as Wilma Subra's summary report they never state explicitly that there is a health concern. The TCEQ, however, does make a statement on health concerns in the Barnett Shale stating explicitly "there are no immediate health concerns from air quality in the area." What you need to realize here is that we public health types cannot say, for certain, that there are, or will never be, a health concern. Why? Because we lack data on humans. So we use available data on animals and models, and subjective health questionnaires to help us make a reasonable guess as to the level of concern.
And it is because of this guessing, and the need to be 'sure,' that levels are set so low. And it is also because of this guessing that those who have an agenda or see anecdotal evidence of cause and effect, feel free to guess as well. And all of this leads to confusion, concern, and fear for the people that are downwind from the event.
But there is a difference between guessing using scientific principles and methods and the reliance on anecdotal evidence, incorrect sampling, improper comparisons, conjecture, and implausibility. In other words, public health safety guessing is not just guessing. There is a method behind the madness, and it has to do with accepting a model that is built around a a flawed premise and compounded conservatism. It's all about the reference value (ReV) and the fact - as my Risk Assessment Professor - Dr. Robin Autenrieth - so succinctly informed me "the people demand a number."
Next Post: Air Quality in the Barnett Shale - Part 6: Cumulative Risk & ESL Development
.
No comments:
Post a Comment