Here is another more personal post. It started out as a message for friends alarmed by reports they saw in the general press.
This has become a topic in the general press because it appeared in "The Lancet" with apparently authoritative backing. You will have no trouble finding alarming headlines about Chronic Fatigue Syndrome (CFS) and suicide. You can also find serious errors of interpretation in the general press.
Fortunately, the original article was published as Open Access, so it is easily available.
Here's an analysis by James Coyne, who is also an authority on psychology in public health and statistics:
While I agree with what Coyne says, I don't limit myself to the criticisms he expresses. The central problem is that a cohort of 2,147 patients, given a diagnosis of CFS by a grab bag of criteria, had 17 deaths from all causes over a period of 7 years. Only 5 of these were suicides. A comparable cohort from the general population, matched for age and sex, would presumably have one person 5/6ths dead from suicide in a similar period, though there are important unanswered questions about how to construct this comparable cohort.
Without going any further I want to point out that this research study is based on a number of actual suicides equivalent to those appearing in anecdotal press reports, including those in which suicide was precipitated by awareness the patient would be committed to a locked ward because of presumed psychiatric illness.
This should immediately raise questions about diagnostic accuracy. If the diagnoses were in error in those 5 cases the entire thrust of the paper is simply wrong. People who have already fought in the trenches of battles over diagnostic criteria for "CFS" immediately noticed that 65% of the patients were diagnosed by the Oxford criteria. This is especially important because these allow primary depression with secondary CFS, while ignoring possible primary CFS, as an organic disease, with secondary depression. This makes it completely impossible to control for depression in the study. Had those patients who developed depression been treated on that basis alone, without a diagnosis of CFS, it is possible most suicides could have been prevented. Failure to control for depression in a psychiatric study of suicide risk is a serious blunder.
You could compare suicide rates in MS patients and the general population, and get another alarming statistic. Since several studies have found CFS as debilitating as MS, absent the risk of mortality from progressive disease, it might make more sense to compare suicide rates in MS and CFS, if you did not start out with the innate knowledge (gnosis) that MS is organic and CFS cannot be.
If you want to put these rates in perspective you should compare against suicide rates for brain surgeons or psychiatrists. Older statistics suggest this was 10 times the rate in the general population. Published current rates are harder to find. You might also check statistics for prescription drug abuse among medical professionals. Such professionals are seldom placed in locked wards simply because they belong to a high-risk group.
The authors of this study have a close association with the authors of the PACE study on treatment of CFS, which I've analyzed separately. One claim in that study was that Cognitive Behavior Therapy (CBT) and Graded Exercise Therapy(GET) were safe therapies. The only way I can see that they reached this conclusion is that none of the 640 subjects in the study died. Nothing which indicates that no patient was worse has been published. Questions about how they decided if a patient had been harmed remain unanswered. In reading the publications I have no idea how rates of hospitalization in the study groups compare with anything else, and I can't tell how many patients deteriorated while in therapy. I'm far from alone in being unable to understand how they decided treatment was safe. Multiple research scientists and medical specialists are now asking for detailed data to be released, something the authors have fought vigorously.
When the PACE authors reported in the popular press that 60% of the patients responded to treatment, they failed to mention that 45% of the patients in their control group improved. It is entirely possible 45% improved, 45% declined and 10% remained unchanged, a null result. The difference resulting from those preferred treatments would not be enough to gain approval for an antidepressant. You would have a hard time deducing this from press reports.
Here we get into the magic of selection effects on statistics. The PACE cohort of 640 was selected from an intake of 3,158 patients labeled as CFS patients by other doctors. It should not be hard to select that number of patients who are less likely to die within the next year from such a large group, which takes care of safety. Picking patients more likely to have a spontaneous recovery can give the illusion of effectiveness. Choosing whom to treat is considerably different from having safe or effective treatment. Applying these results to the entire intake on the assumption of undemonstrated diagnostic uniformity is another blunder.
In the study concerning suicide there was a dramatic lack of selectivity. What we then have is a situation where the null hypothesis concerning safety and effectiveness of those preferred therapies does not conflict with either published result. It appears that being part of a study involving therapists can cause patients to modulate symptom reports while the study is going on, but this has absolutely no long-term effect on outcomes. You can see the same modulation of symptom reports in the case of MS patients given CBT, so this does not distinguish CFS from organic disease.
Concern for possible misdiagnoses, and how to deal with them in research, becomes a serious question, but that is a topic for another time.
This has become a topic in the general press because it appeared in "The Lancet" with apparently authoritative backing. You will have no trouble finding alarming headlines about Chronic Fatigue Syndrome (CFS) and suicide. You can also find serious errors of interpretation in the general press.
Fortunately, the original article was published as Open Access, so it is easily available.
Here's an analysis by James Coyne, who is also an authority on psychology in public health and statistics:
While I agree with what Coyne says, I don't limit myself to the criticisms he expresses. The central problem is that a cohort of 2,147 patients, given a diagnosis of CFS by a grab bag of criteria, had 17 deaths from all causes over a period of 7 years. Only 5 of these were suicides. A comparable cohort from the general population, matched for age and sex, would presumably have one person 5/6ths dead from suicide in a similar period, though there are important unanswered questions about how to construct this comparable cohort.
Without going any further I want to point out that this research study is based on a number of actual suicides equivalent to those appearing in anecdotal press reports, including those in which suicide was precipitated by awareness the patient would be committed to a locked ward because of presumed psychiatric illness.
This should immediately raise questions about diagnostic accuracy. If the diagnoses were in error in those 5 cases the entire thrust of the paper is simply wrong. People who have already fought in the trenches of battles over diagnostic criteria for "CFS" immediately noticed that 65% of the patients were diagnosed by the Oxford criteria. This is especially important because these allow primary depression with secondary CFS, while ignoring possible primary CFS, as an organic disease, with secondary depression. This makes it completely impossible to control for depression in the study. Had those patients who developed depression been treated on that basis alone, without a diagnosis of CFS, it is possible most suicides could have been prevented. Failure to control for depression in a psychiatric study of suicide risk is a serious blunder.
You could compare suicide rates in MS patients and the general population, and get another alarming statistic. Since several studies have found CFS as debilitating as MS, absent the risk of mortality from progressive disease, it might make more sense to compare suicide rates in MS and CFS, if you did not start out with the innate knowledge (gnosis) that MS is organic and CFS cannot be.
If you want to put these rates in perspective you should compare against suicide rates for brain surgeons or psychiatrists. Older statistics suggest this was 10 times the rate in the general population. Published current rates are harder to find. You might also check statistics for prescription drug abuse among medical professionals. Such professionals are seldom placed in locked wards simply because they belong to a high-risk group.
The authors of this study have a close association with the authors of the PACE study on treatment of CFS, which I've analyzed separately. One claim in that study was that Cognitive Behavior Therapy (CBT) and Graded Exercise Therapy(GET) were safe therapies. The only way I can see that they reached this conclusion is that none of the 640 subjects in the study died. Nothing which indicates that no patient was worse has been published. Questions about how they decided if a patient had been harmed remain unanswered. In reading the publications I have no idea how rates of hospitalization in the study groups compare with anything else, and I can't tell how many patients deteriorated while in therapy. I'm far from alone in being unable to understand how they decided treatment was safe. Multiple research scientists and medical specialists are now asking for detailed data to be released, something the authors have fought vigorously.
When the PACE authors reported in the popular press that 60% of the patients responded to treatment, they failed to mention that 45% of the patients in their control group improved. It is entirely possible 45% improved, 45% declined and 10% remained unchanged, a null result. The difference resulting from those preferred treatments would not be enough to gain approval for an antidepressant. You would have a hard time deducing this from press reports.
Here we get into the magic of selection effects on statistics. The PACE cohort of 640 was selected from an intake of 3,158 patients labeled as CFS patients by other doctors. It should not be hard to select that number of patients who are less likely to die within the next year from such a large group, which takes care of safety. Picking patients more likely to have a spontaneous recovery can give the illusion of effectiveness. Choosing whom to treat is considerably different from having safe or effective treatment. Applying these results to the entire intake on the assumption of undemonstrated diagnostic uniformity is another blunder.
In the study concerning suicide there was a dramatic lack of selectivity. What we then have is a situation where the null hypothesis concerning safety and effectiveness of those preferred therapies does not conflict with either published result. It appears that being part of a study involving therapists can cause patients to modulate symptom reports while the study is going on, but this has absolutely no long-term effect on outcomes. You can see the same modulation of symptom reports in the case of MS patients given CBT, so this does not distinguish CFS from organic disease.
Concern for possible misdiagnoses, and how to deal with them in research, becomes a serious question, but that is a topic for another time.
