Trial By Error: My 2011 Exchange with White et al about Case Definition

By David Tuller, DrPH

My first exchange of views with the PACE authors involved the issue of case definition–the criteria used to identify the illness they  called chronic fatigue syndrome. This exchange took place courtesy of The New York Times, not long after The Lancet published the results of the PACE trial. In March, 2011, the Times ran a piece about the role of disease criteria in epidemiology, in which I analyzed the PACE trial’s use of the problematic fatigue-based Oxford criteria. (This was actually my second piece involving PACE; the first, a news story, was a piece of crap because I took the study at face value, having never heard of it before the findings were announced.)

After my case definition story ran, I received an e-mail from Professor Michael Sharpe, requesting a correction and–if I remember accurately–offering to send me a copy of the paper. Professor Sharpe’s complaint was that I had not mentioned the PACE trial’s sub-group analyses of participants who had met two other case definitions. I forwarded Professor Sharpe’s note to my editor, who agreed to run a letter from the PACE authors, along with my response. I thought it would be interesting to revisit that exchange, so I’ve posted it below.

Meanwhile, case definition continues to be a major source of debate and contention in the community of people–patients, advocates, researchers–interested in this domain. The issue also gets entangled with the back-and-forth on the name or names used for the illness/disease as well as the core symptom of “post-exertional malaise” or “post-exertional neuro-immune exhaustion” or “exertion intolerance” or whatever.

In these discussions, the distinctions between clinical and research case definitions often get overlooked. As a general rule, clinical case definitions are broader because you want to make sure everyone who might have the condition gets medical care, even if they are eventually determined to be suffering from something else. Research case definitions can be tighter because you want a more homogeneous group to reduce the possibility that extraneous factors might muddy the clarity of the results.

For example, a clinical case definition for an illness might include people with co-morbidities that can generate overlapping symptoms. However, the parallel research case definition might exclude these patients, since their presence could make it harder to interpret the findings.

I’m not going to address specifics of the various criteria in this post. For now, here are a few helpful case definition resources for ME, ME/CFS, etc:

#MEAction recently posted “Demystifying the Diagnostic Criteria for ME and Related Disease,” a thorough account of six key case definitions.

MEAdvocacy, which supports the 2011 ME International Consensus Criteria, has a helpful side-by-side comparison of that case definition with the one from the Institute of Medicine (now the National Academy of Medicine), among other resources.

Last August, the US ME/CFS Clinician Coalition released a set of diagnostic and treatment guidelines based on the IOM/NAM case definition.

Last year, I posted a Q-and-A with Leonard Jason, a professor of psychology at DePaul University, who has published many studies about case definition in this domain. In our conversation, he expressed concerns about use of the IOM case definition in research because it does not exclude those with primary psychiatric disorders.


On March 4, 2011, The New York Times published “Defining an Illness is Fodder for Debate,” my story on case definitions in epidemiology. (The same story ran in the print edition four days later.) On March 14, the news site published the following exchange:

To the Editor:

In “Defining an Illness Is Fodder for Debate” (March 8), your reporter David Tuller correctly noted that the way an illness is defined can often determine what is found in studies of it. He also suggested that this problem of definition had limited the interpretation of our trial of treatments for chronic fatigue syndrome, published recently in The Lancet. That is not the case.

The patients in this trial had a disabling chronic illness in which fatigue was their main symptom and for which no alternative had been found; that is the definition of the syndrome used in Britain. But we also assessed trial participants to see if they met two other definitions of the illness that are favored by some scientists.

We found that both cognitive behavior therapy and graded exercise therapy, when added to specialist medical care, were most effective not only in the whole sample but also in the participants who met these alternative criteria. In addition, these treatments were the most effective whether or not a patient was depressed, a not uncommon accompaniment to this chronic and misunderstood illness. The trial also found that these treatments were safe so long as they were provided by appropriate therapists trained to help patients with chronic fatigue syndrome.

So to Mr. Tuller’s question “Does the evidence from that study prove that these strategies would help patients identified as having chronic fatigue syndrome through very different criteria?,” the answer is “Yes, it does.” Patients and their doctors now have robust evidence that there are two safe treatments that can improve both symptoms and quality of life, however the illness is defined.

P.D. White

T. Chalder

M. Sharpe

London and Edinburgh

The writers are the principal investigators of the PACE trial.


David Tuller replies: The article asked whether findings among a population defined by one set of criteria would apply to populations defined by “very different criteria.” In this study, all participants were first defined, identified and selected not by different criteria but by the same criteria, the so-called Oxford criteria used in Britain. Subgroups within that already screened population who also meet secondary criteria are not easily compared to patients who have not been screened, since an unknown number who met the secondary criteria might not have met the study’s criteria for inclusion. The gold standard for making comparisons across groups of patients identified by three varying case definitions would be a study with three completely separate cohorts, not one large sample with two embedded subgroups.






Comments are closed.

Scroll to Top