Monday, 17 March 2014

Day 216: The one-question questionnaire

There will be early starts and maybe late closings this week at the Montreal Tobacco Trials.

The days have been extended as part of an agreement to take only three more days to finish the industry's critique of the plaintiffs' expert epidemiologist, Jack Siemiatycki.

That leaves one and a half (long) days each for the second (Mr. Kenneth Mundt) and third (Mr. Bertram Price) of the three men hired to kill off the idea that it is reasonable to conclude that if someone has smoked for 5 pack-years and has one of the diseases covered by the Blais class action, then their illness is more likely due to smoking than not.

It's not just the compression of time - nor the fast pace of the questions and answers - that gave today an unusual intensity and concentrated focus.

Ever since Mr. Siemiatycki testified, it's been clear that his numbers and the principles behind them are key to the Blais case, and therefore key to the companies' defence. With that defence winding to a close, this is a make-or-break moment for both sides.

The first of Mr. Siemiatycki's three critics testified last week. Mr. Laurentius Marais testified on behalf of JTI-Macdonald, with erudition and mannered precision, took a fine-tooth comb to the methods used by Mr. Siemiatycki. He pounded especially hard on the idea the way in which confidence intervals were treated by Mr. Siemiatycki's meta-analysis, saying that the result was statistically unreliable and that the results could not be trustedt.  Justice Riordan may have been convinced of the methodological sincerity, but he did not sound satisfied that this criticism helped him address his challenge of finding something reasonably reasonable to apply to this case.

All that to say that it didn't look (to me at least) that Mr. Marais had scored a home run.

The second man up to bat

American epidemiologist, Mr. Kenneth Mundt, was hired by Philip Morris/Rothmans, Benson and Hedges (PMI/RBH), and given essentially the same mandate as Mr. Marais -- i.e. to critique Mr. Siemiatcyki's methods and conclusions.

He is a man with long experience in studying the link between toxins and diseases on behalf of industries that produce toxins and don't want them linked to disease. Now only 54 years old, he has spent about half of  his professional life as a consultant to industries in need of "Applied Epidemiology", and is now a principal at Environ, a one-stop shop for industries in search of expert witnesses and regulatory friends.

Although Mr. Mundt has been hired as an expert in about a dozen tobacco trials, and has testified at "ten or eleven" other trials, this is only his second appearance as a witness in a tobacco suit.

The voir-dire process was no longer than it had to be (the plaintiffs asked few questions and did not oppose his role). Within an hour, Justice Riordan qualified Mr. Mundt as an expert in "epidemiology, epidemiological methods and principles, cancer epidemiology, etiology and environmental and lifestyle risk factors, and disease causation in populations." 

Much of a sameness

Although it is shorter (a mere 27 pages!) and less dense than the report discussed last week, Mr. Mundt covers much of the same ground, and takes a similarly harsh view. He concludes that the numbers produced by Mr. Siemiatycki are "unreliable for their intended purpose, and cannot be scientifically or convincingly substantiated."

And, like the previous witness, Mr. Mundt produced no estimates of his own, nor gave any advice on where more reliable numbers could be found.

But a different line of attack 

Although the message was the same, the messengers and the delivery (of both lawyers and witness) were vastly different today. Different also was the angle of attack on Mr. Siemiatycki.

Mr. Pratte had used open-ended questions to invite Mr. Marais to give bookish answers about profound methodological concerns. Today, Mr. Simon Potter (counsel to PMI/RBH) used very pointed questions to invite Mr. Mundt to say "yes" or "no" to questions that gradually painted Mr. Siemiatycki's report as one which was flawed and misleading.

The goal seemed to be to demonstrate to Justice Riordan that using Mr. Siemiatycki's numbers would not only be unreasonable from a statistical point of view, but would also be unjust. Such a broad-brush approach would cover too many people who should not be considered harmed by cigarettes.

And such different style!

Each of the lawyers in this case has their own distinct style, but that of Mr. Potter is perhaps more overwhelmingly distinct than others'. (If you can remember what over-acting looked like in the years when people still said "thespian", then you might have a flavour of the day).

From an audience perspective, it worked for him today. His points were made memorably - and there was a narrative arch to his message.

He had distilled his points and was careful to not walk over them. If Law and Order were about toxic torts, this could have been television.

One question is never enough

The thread that was used to weave Mr. Mundt's views together was the denouncement of a "one question questionnaire," and the elaboration of the many ways in which such an approach would be a bad route to follow.

The "one question" refers to Jack Siemiatycki's estimate that if someone with the four Blais diseases had smoked a "critical amount" of cigarettes before becoming ill, then their disease could be considered, on a balance of probabilities, to have resulted of smoking. The critical amount -- 5 pack-years, or a lifetime smoking of 36,500 cigarettes or more -- is now part of the definition of the Blais class.

(The "one question" isn't quite accurate! In addition to answering a question about whether or not they had smoked 36,600 cigarettes, a potential class member would also have to say "yes" to whether or not they had a fatal disease.)

In a volley of short questions, and one word answers, Mr. Mundt confirmed the many ways in which this "one question questionnaire" was inadequate to the task.

It would not take into account how many cigarettes per day (intensity) or how many years (duration) or how long ago one quit. Nor did it reflect the age at which one started smoking, the types of cigarettes one smoked or the amount of tar they produced. It provided no information on whether one knew about the risks of smoking, or were exposed to tobacco advertising or tobacco industry positions on harm. Etc.etc. etc.

Yes. Yes. Certainly. Yes. Yes. Then, a change of pace to No. No. Absolutely Not. No.

For much of the morning, Mr. Potter did the heavy lifting. Mr. Mundt had little to do but nod along. (Perhaps he too was lulled by the steady rhythm of the questions - at least once he said "yes" too quickly to an improperly phrased question, which later required Mr. Potter to back-track.)

Special attention was paid to:

*  the benefits of quitting. Mr. Mundt agreed that it came close to that of a never-smoker after a few decades, and with the studies and Health Canada statements to that effect. (Exhibit 30219)

* the shape of the dose-response curve. Mr Siemiatiycki had used a straight line to represent the increase in risk from increased smoking. Mr. Mundt felt it should be "s" shaped. (words for the day: Sigmoidal and Spline)

* the need for "other metrics".  Pack-years were deficient, as they did not take into account the variables in ways people smoked that might affect disease outcomes.

* heterogeneity. Lumping people together, combining study results, lumping cancers together, or averaging risk continued to come under heavy fire.

Past studies 

A non-linear dose
response curve (Exhibit 30223)
A second device used by Mr. Potter to good effect was to contrast the methods used by Mr. Siemiatycki in his expert report with those that he used in previous studies.  The clear inference in the questions, if not in the answers, was that the methods used for court purposes had been chosen to exaggerate the plaintiffs' case.

Among Mr. Siemiatycki's long publication record, Mr. Potter introduced papers where more heterogeneity had been used, more concern for confounding were expressed, non-linear dose response curves had been considered, and smokers who quit a long time ago were considered to be at low risk. (Exhibits 30221, 20023, 30221).

These were all missing from Mr. Siemiatycki's "overly simplistic" expert report, said Mr. Mundt.

It was only when Mr. Mundt was asked to conclude that he spoke at length without additional prompting from Mr. Potter. He delivered a well-constructed and very polished-sounding summary of his concerns. He directly addressed Justice Riordan:

 "I find it (the report) too simplistic and too blunt to be of the use that I think I understand you will put it." 

"I wish to raise before you that this is not indicative of the strongest science. The other papers from Dr. Siemiatycki are a testimony of what he is capable of producing, but has not produced for your purposes in this matter." 

He elaborated his concerns for several minutes, concluding with advice that echoed that given last week by Mr. Marais': "Collectively, these [faults] make it an unreliable summary for your purposes."

Questions from the other side

It was a much tougher André Lespérance who stood to cross-examine Mr. Mundt following a brief mid-afternoon break. The body language was tough, the tone of voice unusually firm and there was no sugar-coating of questions.

Nor was there any of the usual softening-up exchanges. Mr. Lespérance went quickly to a paper that Mr. Mundt had co-authored -- one in which he had employed essentially the same analytic approach that he now criticized in Mr. Siemiatycki. (Exhibit 1705)

Mr. Mundt's recent harsh words when contrasting Mr. Siemiatycki's earlier works heightened the drama of this moment. As did the change in tone by Mr. Mundt, who looked genuinely non-plussed. He looked at the paper as if he had never seen it before. (As third author, is it possible he had not?). He said he had trouble recalling it or the methods used in it.

The paper studied the cause of lung cancer among roofers. Although published in a scientific journal, it had been underwritten by trade organizations representing roofing companies. They were, I presume, pleased at the conclusion that it was roofers' smoking that was the cause of their disease, not their occupational exposure to asphalt.

Mr. Lespérance pointed out that in coming to that conclusion, Mr. Mundt and his co-authors used an even more crude categorization of smoking history than pack-years. Nor did they distinguish between types of cancer or make other adjustments for heterogeneity in their "random effects model."

"Is it good epidemiology?" Mr. Lespérance asked Would it pass muster with Mr. Marais?

"It serves its purposes," said Mr. Mundt opaquely, before conceding that his paper also would also have been vulnerable to criticism by Mr. Marais.

Mr. Lespérance challenged him to address the issues as if he were in a judge's position. "You have somebody working as a roofer, who has smoked and has quit smoking. ...  How are you going to determine whether smoking was the cause of disease? ...."In a smoker you will never know because we can't see - we need to work with probabilities.

Mr. Mundt took refuge in his message box. "You are asking me to speculate on any one individual. I wouldn’t speculate on the specific causes of his disease on those circumstances..." But he also defended his profession and the use of epidemiology in such an instance. "Individual causation may not be knowable" but "epidemiology is the right place to go."  

Mr. Mundt said that science had marched on since that paper was published  in 2007. His views on multiplicativity and pack years had changed because there were now newer techniques and better methods of collecting data. (Mr. Siemiatycki's expert report was written only 2 years later).

Mr. Lespérance showed him the opinion of other prominent epidemiologists who recently reconfirmed support for the use of pack-years. (Exhibit 1707). Over the objections of Mr. Potter, he asked Mr. Mundt to respond to the views of an unidentified proponent of pack-years: "More accurate estimations [of tar dose] can be approximated through the combination of measures such as duration and intensity of smoking like pack years." 

Mr. Mundt's looked quite uncomfortable, and reasonably so as Mr. Lespérance revealed that man behind those words was none other than Mr. Mundt -- and that he had said his as recently as 2010.

It was already past 5:00, and the court soon adjourned. Saved by the bell -- at least temporarily.

The cross-examination of Mr. Mundt will continue Tuesday morning.

This post was back-timed to make indexing consistent