Cognitive Interventions and Rehabilitation to Address Long-COVID Symptoms: A Systematic Review.

Only three of the 19 studies were randomised trials, and most of the rest had no control group at all. Without a control group, improvement could be recovery over time or the effect of taking part in a programme rather than the treatment itself. 18 of the 19 studies were graded at moderate risk of bias and one at high risk, largely because participants and the people measuring outcomes could not be blinded. Samples were small, from a single patient up to 149 people, and several studies found no significant change on the outcome they were targeting, including fatigue in one rehabilitation programme and cognitive screening, mood and quality-of-life scores in another. Most studies delivered several treatments at once, so none can show which part helped. Most studies did not describe their intervention in enough detail for anyone to repeat it. The authors say they searched only five databases, did not search grey literature, and may have missed relevant papers; publication bias cannot be ruled out because positive findings are more likely to be published.

The authors declared no potential conflicts of interest and said they received no financial support for the research, authorship or publication of the article.

Do not read 'every study showed improvement' as proof that these treatments work. Only three of the 19 studies were randomised trials, most had no control group at all, and 18 of the 19 were rated at moderate risk of bias, so improvement could be natural recovery or simply the effect of being in a treatment programme.