Effect Measure

A lot of things that seem on first glance to be “news” are really just reprints or slight edits of press releases written to tout a commercial product. This is also true of “Newsletters” that charge money for inside news.

Datamonitor is a company that claims to be “the world’s leading provider of online data, analytic and forecasting platforms for key vertical sectors. We help 5,000 of the world’s largest companies profit from better, more timely decisions” (Datamonitor website). Some of the stuff they give away, since I see it and I don’t subscribe to anything they sell. But based on its accuracy, I don’t have much confidence in their premium products. Consider the “news” item about a TB test being marketed by a company called Cellestis. The headline says that the Cellestis TB test has been found effective in a landmark study:

Cellestis International has reported positive results from a landmark study that showed QuantiFERON-TB Gold, a blood test for detecting TB infection, six times more accurate than the conventional tuberculin skin test or TST at predicting which tuberculosis-exposed individuals will go on to develop TB disease.

The results demonstrated from the study indicated that QFT had a predictive value for developing TB disease of 15%, more than 6 times greater than the 2.3% for the TST. (Datamonitor)

Predictive value (which comes in two forms, not one) is not a measure of accuracy because it depends on two other, real measures of accuracy, sensitivity and specificity, together with the prevalence of the condition in the specific population being tested. It is therefore essentially worthless as an accuracy measure, or at best, hard to compare with other methods. There are many ways positive predictive value can be boosted, some of which aren’t very helpful to the patient or the provider.

These kinds of test kits aren’t new, either. So this is just PR. Pure, unadulterated PR. If that’s what the Datamonitor product is, why would anyone pay for it? Just asking.


  1. #1 Stagyar zil Doggo
    February 16, 2008

    Any comments on the QuantiFERON-TB Gold besides the above “news” item? How does its sensitivity and specificity compare to TST/Mantoux or Chest Radiographs?

  2. #2 revere
    February 16, 2008

    Stagyar: I don’t know the exact answer but your question needs to be sharpened. This is claiming the test is better (in both measures) of predicting which TB infected people will be or are active cases, i.e., active disease with shedding of acid fast organisms. That is a fairly stringent requirement. Maybe they have good data on this but the “news” piece doesn’t say what the data are (excuse the pedantry of using data in the plural; it is an affectation of epidemiologists).

  3. #3 Eric
    February 16, 2008

    Having looked at a very limited number of studies on QFT-Gold, if I recall it was very close to the skin test in terms of sensitivity and specificity, but would theoretically be much better as prevalence of TB drops, or in increasingly linked AIDS-TB infection.

    That and it’s less subjective.