Wednesday, 12 March 2014

Day 215: Headlong into year 3! .....

The Montreal tobacco trials passed another milestone today. It was two years ago today that courtroom 17.09 was filled to the rafters with those who had come to see the opening of this historic event.

Today, there was an over-flow crowd on the lawyers' benches ahead of the bar, but the only non-participants in the audience were the three regular note-takers: your two humble bloggers and a lawyer who represents the insurance company that might be on the hook for the legal fees for Imperial Tobacco and Rothmans, Benson and Hedges.

The occasion did not go unmarked, however. As everyone was settling into theirs seats, Justice Riordan greeted the room: "Before we start I would like to wish everyone a happy anniversary."  

In a courtroom, every anniversary is a paper anniversary.

A two-day interval of no-confidence

The beginning of the week was spent reviewing the criticisms of the statistician hired by JTI-Macdonald regarding the plaintiffs' experts' estimate of the number of Quebecers who suffered from the diseases covered by the Blais side of the Blais-Létourneau twinned class action suits.

Mr. Laurentius Marais did not hold back!

For two long days he had elaborated on the many and varied faults he found in the report of Montreal epidemiologist, Jack Siemiatycki. (Mr. Symiatycki's expert report is Exhibit 1426.1; Mr. Marais' expert report is Exhibit 40549).

His conclusions were unambiguous, and there was no sugar-coating. He saw no statistical validity to Mr. Siemiatycki's estimates, and said that nothing reliable could be produced from his methods. In getting to this point, he had also thrown a lot of mud at the many component pieces of the meta-analysis conducted by Mr. Siemiatycki.

Mr. Marais is only the first. Two other experts have been hired to beat-up on Mr. Symiatycki's report, and by yesterday evening it was already beginning to look a little bruised.

As they left the court-room last night, the plaintiff lawyers looked more than a little pensive. They had only one day (today!) of cross-examination of this witness. How best to wipe the mud off the Siemiatycki report and put the air back in its tires?

Mr. André Lespérance and Mr. Bruce Johnston were the lawyers assigned to the cross-examination, and when they returned today they looked more confident. As they frequently do, they put themselves in a good-cop/bad-cop sequence. Mr. Lespérance would politely try to elicit admissions, and Mr. Johnston would less gently challenge Mr. Marais' credibility.

The KISS (keep it simple, stupid) principle

Mr. Marais is a man with an obvious fondness for precision. Yesterday, I had felt that he was willing to focus on precise details on the trees to obscure the forest, but in his answers today I recognized no attempts to divert attention from the facts he was being asked by the plaintiffs to acknowledge.

Mr. Lespérance was soon able to use this perhaps unexpected candour to good effect.

He spent the morning inviting Mr. Marais to acknowledge the condition of the forest of population-level effects. The witness, after rephrasing the question into more precise language, seemed happy to acknowledge the central health issues behind the Blais case.

A) The "vast majority" of lung cancers are caused by smoking

Of the thousands of exhibits that are now part of the trial, many hundred were presented by JTI-Macdonald's counsel, including a heavy number of scientific publications. Around 100 scientific articles were introduced as exhibits to support Mr. Marais' report.

Beginning with the reliance material Mr. Marais had himself selected, and then with other papers entered as evidence by JTI-Macdonald, Mr. Lespérance drew attention to the conclusions of many scientists that by far the greatest number of lung cancers were caused by smoking.

Mr. Marais was quick to agree that there was a "consensus"  that the "vast majority" of lung cancers were caused by smoking. He volunteered, without being asked, that the range of was "somewhat north of 75%" and went up to 80% to 90%. He did not quarrel with the many reports shown to him that gave the higher figure.


B) Relative Risk translates into population level effects

As if he had not spent two days critiquing calculation methods, Mr. Marais was surprisingly willing to confirm the relatively simple arithmetic operations presented to him to translate calculated relative risks from smoking into estimates of how much disease could be attributed to tobacco.

What would be the percent of excess cases in a population in which there had been a 20 fold increase in risk? "Then it follows that 19 out of 20 of those cases [of exposed individuals] are additional cases because of the exposure - 95%"


C) "2" is an important threshold value for Relative Risk 

Earlier this week, Mr. Marais had laid as a foundation of his critique what he saw as the misuse of estimates of risk when the range defined by the confidence interval fell below the value of "2" (2 is the value where risk is doubled, with the consequence that the "excess" risk exceeds the background risk.)

But when Mr. Lespérance showed him a review of risk of cancer from radon (Exhibit 1699), Mr. Marais allowed that even when the confidence interval fell below 2, one could make conclusions about "general causation."  

He acknowledged that his concerns about specific causation did not apply to general causation. "It is quite different from saying [it] is not a cause of lung cancer in general."


D) 3 to 5 pack years doubles the risk of lung cancer - at least according to one local study

Mr. Siemiatycki had chosen to conduct a meta-analysis in order to expand the evidentiary foundation of his conclusions about lung cancer and smoking. (He was the author of one of only a few studies on Canadian populations).

It was this meta-analysis prepared for his expert report that had come under such strong attack.

But what if the court relied only on a study conducted about Montrealers? (The only study on the court record is one also done by Mr. Siemiatycki, in 1995. (Exhibit 1428)

Mr. Marais had re-calculated the results of the study (Exhibit 40549C). He was asked to look at this work, and how many pack years his re-calculations showed were necessary before the risk of lung cancer had doubled. Did they not show that "the critical amount would be between 3 and 5 years," Mr. Lespérance suggested.

Mr. Marais agreed. "That would be the implication of these data." 


The other shoe

Bruce Johnston's questions, which took the second half of the day, were focused on providing Justice Riordan with reasons to put less confidence in Mr. Marais' testimony.

He contrasted the ethical norms of the American Statistical Association (Exhibit 1697) to the use of statistics by tobacco companies in trials (Exhibit 1701), and to a report of Mr. Marais's anticipated testimony at a previous trial (Exhibit 1704). He connected Mr Marais' seemingly unsurmountable methodological concerns to a strategy articulated by one his other client companies of setting an impossible level for proof.  (Exhibit 1702R)

Mr. Marais came out of this wrestling match pretty much unscathed. The same cannot be said for his clients.

One of the issues at this trial is whether it was ever reasonable for the companies to maintain that it had not been proven that smoking caused cancer. Mr. Marais, whose high standards for statistical proof had already been made clear, explained why he had reached this conclusion over 20 years ago.

It was not just the "many epidemiological studies of different kinds which demonstrated and documented over decades the strength of the association,"  or the strong consensus of experts, he said. The kicker for him was that no one had ever provided an alternative explanation, or revealed "the hidden factor" of another cause. "There was lots of opportunity for it to have been discovered."

Leaving the judge empty-handed

The three industry witnesses who have reviewed Mr. Siemiatycki's report have two things in common: they think the plaintiffs' epidemiologist's numbers are wrong, and they don't have any right numbers to propose. The strategy seems to be to leave the judge with nothing to work with.

Justice Riordan's questions at the end of the day suggested this is not a position he wants to be in.

"You have come in and you have criticized at length the steps in the method that is applied in trying to fulfill a mandate of coming to a number of sufferers of disease that are likely caused by smoking. Did you do any calculations yourself, after seeing the weaknesses in [Mr. Siemiatycki's] case? Did you estimate the number of people with lung cancer likely to be have caused by smoking?"

"I did not perform an independent study, no."

The judge asked whether the number of pack years required to give a relative risk of 2 could be produced, and Mr. Marais replied that it could only be done with a new study involving thousands of people.

Judge Riordan clarified whether with existing "published information, is it possible to do this?"  Mr. Marais was definitive in his answer. No. "I do not see a method that I could recommend to the court based on the information that I am aware of as I stand here today." 

Seconds later, Mr. Marais was being wished a safe trip home.

The second related witness, Mr. Kenneth Mundt, was expected to testify tomorrow. Instead, for unexplained reasons, the schedule was changed. He and Bertram Price have been rescheduled for the first half of next week.

It feels like a cliff-hanger: Will the companies use the long weekend come up with a new plan and respond to Justice Riordan's clear hints for some numbers to work with?

The trial resumes on Monday.