[Stoves] Malawi Philips stove intervention study and Nikhil's 28 sins of insolence about GACC/WHO (Re: Ron Larson)

Wed Jan 18 18:02:53 CST 2017

Dear Andrew

The leap of logic (my words) is this:

A GBD is a post-death analysis and attribution of what contributed to the deaths of the total population of dead people. No individual was examined to create the attribution, it is a 'population statistic'. When they say all people who died prematurely (pick a number) did so from a variety of contributing causes. They are listed, and the % contribution to the pre-mortality is listed in a table.

It does not mean that anyone in particular died from exactly that combination of contributing causes. They are just attributed. Therefore removing the attributed cause entirely from the current population may have no effect at all on the current population for two reasons: first there may not have been a correct attribution in the first place, and second, we have no idea what the total disease load on the person affected by this intervention will be in future, and how other confounding factors will manifest themselves.

In short, you can’t use a guess about what happened to an entire population in the past to claim something about a particular person in the future. Yet that is exactly what is being advanced by this particular set of emission targets for cooking stoves.

As Nikhil has pointed out, (citing something wise said by Kirk Smith) attribution to cause does not mean avoidance is predictable.

Next, suppose you see it written that '4.3m people died prematurely from factor X in the past year’. That is an attribution meaning ‘they said so’. It does not mean even one of them was directly or proportionately affected and died prematurely. It is a general way of talking about disease loads and causes of premature death.

Claiming, quite clearly, that reducing stove smoke emissions by 90% will definitely lead to a statistically significant reduction in the incidence of particular disease, as Kirk has also frequently done, several times in my electronic presence, is postulating a cause-effect relationship that is not embedded in, nor supported by, the GBD numbers. Investigating the link by the WHO from GBD numbers to risk to mortality to exposure to causes of exposure to stoves to stove emissions to stove emissions rates, we find that the linkage breaks when what starts out as a population statistic of people who are already dead is used to claim particular consequences for particular people in the future. GBD makes no claims about the future. It is 100% about the past – for people born after 1930 who are already dead.

This it is unsupportable to claim that a set of performance tiers, having emission rates per minute or hour, leads to particular exposures (a failure to model well) and therefore to a particular medical result (breathing problems/pneumonia, URT infections and so on).  The latter is a failure to provide medical evidence. GBD is not ‘medical evidence’. It is a public health guidance method. It is plastered not with bandages but warnings about how not to use it, one of those ‘do nots’ is to make claims for future health impact. Why? because the link for stove smoke is not proven, as it is for asbestos and tobacco.

The Malawi study may not have proven anything for or against the argument. It was in a sense, a search for ‘bespoke science’ – seeking proof for an assertion. Nothing wrong with that, but it is not a very helpful suggestion to say, as they did, that maybe the Philips stoves were not clean enough to demonstrate that solid fuels can burn well to reduce the incidence of pneumonia. That is not the appropriate counterfactual. It sounds more like, “We think solid fuels cannot burn cleanly enough to ‘meet WHO IAW guidelines’ and are trying to find a way to prove it.”

An appropriate counterfactual is, “Stove smoke (the attributed cause) is not a detectable cause of pneumonia (the disease) in Malawian village children (the cohort).”  Removing (according to the GACC comparison chart) a large % of the PM from cooking made no difference that could be detected.

It does not prove the stove ‘is not clean enough’ because replacing the traditional stove with LPG also showed no detectable difference.  Does that suggest LPG is also not clean enough? Of course it does. The that is not a reasonable counterfactual either.

There is a problem with the proposal. If there is a problem with the proposal there is a problem with attaining the result (as demonstrated) by the negative result.

Regards

Crispin

+++++++++++

On 18 January 2017 at 22:25, Crispin Pemberton-Pigott <crispinpigott at outlook.com<mailto:crispinpigott at outlook.com>> wrote:

"stove smoke causes pneumonia’ and all that jazz.  Nikhil has laid out clearly the defective leaps of logic underlying this claim,"

I don't consider he did; Nikhil posted anything but clearly and whilst I eventually managed to struggle through the invective and snide comments coupled with insider  sarcasm which was beyond my comprehension I accept he has made some very good points, many of which I was unaware.

I hope he continues to contribute as he plainly has much to say about the broader picture of [stoves]

Andrew

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.bioenergylists.org/pipermail/stoves_lists.bioenergylists.org/attachments/20170119/6695c142/attachment.html>