Example

“In this trial, which compared standard of care with monthly IPTp [intermittent preventive treatment in pregnancy] with sulfadoxine–pyrimethamine with monthly IPTp with dihydroartemisinin–piperaquine and monthly IPTp with dihydroartemisinin–piperaquine plus azithromycin, the use of dihydroartemisinin– piperaquine was associated with reductions in clinical malaria, malaria infection detected by microscopy during pregnancy, and placental malaria at the time of birth. However, despite these reductions in malaria, the risk of adverse pregnancy outcomes (composite primary endpoint) was significantly lower in the sulfadoxine– pyrimethamine group than in both dihydroartemisinin–piperaquine groups . . .

 

“In a previous meta-analysis a modest, but nonsignificant, reduction of 17% in the number of adverse pregnancy outcomes favouring dihydroartemisinin– piperaquine versus sulfadoxine–pyrimethamine was reported . . . These findings led WHO [World Health Organization] to recommend larger definitive studies to determine whether IPTp with dihydroartemisinin–piperaquine improves adverse pregnancy outcomes compared with sulfadoxine– pyrimethamine in areas with high sulfadoxine–pyrimethamine resistance. Our results are consistent with those from a third trial in eastern Uganda, which showed that monthly IPTp with dihydroartemisinin– piperaquine did not decrease the number of adverse pregnancy outcomes among livebirths or reduce fetal loss compared with monthly IPTp with sulfadoxine–pyrimethamine . . . All of these trials, including ours, show that dihydroartemisinin– piperaquine is the superior antimalarial, resulting in 40–90% fewer malaria infections during pregnancy [196]."

 

Explanation

A good discussion section of a completed randomised trial should start with a brief summary of the trial results, balancing both benefits and harms of the intervention. The discussion sections of scientific reports are often filled with rhetoric supporting the authors’ findings486 and provide little measured argument of the pros and cons of the study and its results. Indeed, some discussion sections can be overly optimistic when discussing the trial findings (interpretation bias; spin) [33, 487-489]. Authors need to guard against such behaviours, as they diminish the rigour of the scientific effort and may result in a loss of trust by readers.

 

Readers will also want to know how the trial results relate to those of other randomised trials. This can best be achieved by including a published systematic review in the results or discussion section of the report [490, 491]. This might be an easy ask for some authors as a systematic review might have been part of the rationale for conducting the trial. However, for others, such synthesis may be impractical, and quoting and discussing any existing systematic reviews of similar trials may be more appropriate. One recent estimate suggests that nearly 80 systematic reviews are published daily [492]. Discussing trial findings in the context of results from any systematic reviews will help readers assess whether the results of the randomised trial are similar to those of other trials in the same topic area and whether participants are similar across studies. Reports of randomised trials have often not dealt adequately with these points [490]. Several methods to address the issue of setting new trial findings within the context of previous research have been proposed [491]. Where conducting a new updated systematic review is not practical, adding the new trial result to the previous systematic review is a much simpler alternative.

 

We recommend that at a minimum, the discussion should be as systematic and objective as possible and be based on a comprehensive search, rather than being limited to studies that support the results of the current trial [493, 494].