Post

1) 🇩🇪 There's a new randomised trial of cognitive behavioural therapy (CBT) for Long Covid, from the Johannes Gutenberg University Mainz in Germany. It claims that CBT lowered fatigue but it has the usual fatal limitations that make the results unreliable.

2) First major error: there was no control intervention. The control group was simply put on a waiting list to receive CBT for 16 weeks. This creates an enormous bias so that the results are likely distorted by expectancy and placebo effects.

3) Second major error: all the outcomes were subjective questionnaires while patients nor therapists could be blinded. There was no actigraphy (step counter), exercise or fitness tests, employment rates, healthcare use, or any other objective outcome.

5) CBT also targets cognitions and interpretation of symptoms, and patients have close contact with therapists for multiple sessions. This makes this treatment highly likely to cause distorted results on symptom questionnaires.

6) The third major error is that the authors used 5 primary outcomes and didn't control for multiple comparisons. Only one of these, the fatigue severity scale (FSS), had a p-value (0.045) that was marginally below the significance threshold of 0.05.

7) With correction, however, none of the results would be statistically significant, meaning the difference could simply have occurred by chance. The plot below shows the treatment effect for the FSS scale, which wasn't anything unusual.