19 October 2010

Schizophrenia Blood Test Enters Market

A commercial schizophrenia blood test was released earlier this month. The test is not FDA approved (the company that makes it asserts because it falls in an exemption from the FDA approval process). It is being marketed to psychiatrists and takes about a week from receipt of a blood sample to produce a result.

Accuracy and Limitations

The false positive and false negative rates are significant for patients who don't score in the extremes of their "biomarker" test (a term that conceals more than it shows as a result of its vagueness). Eyeballing the data, the test is pretty unequivocal for those in the top 60% of the range of results for people otherwise diagnosed as schizophrenic and for those in the bottom 40% of the range of people otherwise diagnosed as non-schizophrenic, and provides suggestive evidence in many other cases. The limited accuracy of the test makes it worrisome because "biological" tests are often given undue weight by decision makers. Also worrisome is the fact that it isn't clear what is going on in the false negative and false positive cases. What is spoofing it? This is purely a black box statistical kludge, so we don't know. For example, it isn't clear if the scores are stable over the course of the condition.

The test does not appear to be a DNA marker test, based on the journal article research supporting its use. So, for example, it isn't clear if the test would have any predictive power in the case of someone with a family history of schizophrenia who is not currently symptomatic. Currently, it is being marketed as a tool for "the diagnosis of recent-onset schizophrenia." This no doubt helps its accuracy, as false negatives are far less likely in people who have already shown some symptoms.

The implication is that the false positive rate in the general public might be far higher than in the people who actually are given the test (a problem shared by almost all medical diagnostic tests).

The disclaimer to the test results addresses these concerns stating:

The test is an aid in the diagnosis of patients with symptoms of recent onset schizophrenia who do not have co-morbid conditions such as diabetes, severe in ammation, or autoimmune disease. The results are not intended to be used as the sole means for clinical diagnosis or patient management decisions. This test is not indicated for screening asymptomatic individuals.

Paternity Tests Compared

An interesting comparison is the likelihood of a finding of non-paternity in paternity tests done because paternity is disputed (about 30% of fathers tested in this context are not actually genetic parents) compared to those were there is a high confidence of paternity (about 1.7% of fathers tested in this context are not actually genetic parents).

Paternity tests are generally required to show a very high probability of paternity to be prove paternity in court (a 97% probability in Colorado), but no fixed standard regarding what level of certainty constitutes a positive result has been set for most other scientific tests, and a high probability is appropriate to require in paternity tests because intermediate indeterminate results are quite rare due to the high accuracy of DNA based paternity testing. Discarding some of the best available evidence might be inappropriate where no other biologically based tests are available.

Is It Useful Medically or Forensically?

This test doesn't appear terribly useful as a medical diagnostic test, because schizophrenia is a mental condition that is not known for being particularly difficult to diagnose in a medical setting. It was one of the first widely diagnosed mental conditions because the symptoms are so distinctive, has a very strong tendency to run in families, and has a characteristic age of first appearance of symptoms (generally in young adulthood). It is also far less obvious than the company would suggest that a slightly earlier diagnosis is particularly helpful from a treatment perspective. This is a condition with no cure, just an adult lifetime of symptomatic treatment.

One can imagine, however, that it might be both used and abused in a forensic setting where an insanity defense is claimed and there is a powerful incentive to prove or disprove the existence of schizophrenia in a person with a powerful incentive to distort the diagnosis process. As a new test, it would require expert testimony, but this wouldn't have much economic cost because expert testimony is required anyway in every insanity defense case. It has not be the subject of any court decisions. But, at face value, it would appear to meet the federal evidentiary standard of Daubert, and most state evidentiary standards follow that standard or are more lenient in allowing expert testimony. A negative or inconclusive result could be abused by prosecutors to show an absence of a legitimate insanity defense when schizophrenia is actually present, as a significant number of cases appear to be false negatives, possibly because there may be multiple biochemically distinct forms of schizophrenia. But, a definitive range result could be very persuasive for an otherwise skeptical jury and for prosecutors reluctant to reach a plea bargain otherwise.

Even a marginally useful or scientifically dubious test, like a polygraph, can be useful out of court, to the extent it is believed to work by those receiving it, because people who believe they will test positive may behave differently than those who do not, when asked to take the test.

No comments: