To the best of my knowledge, a systematic review aims to collect and summarize all the published & unpublished literature revolving around a certain topic/sub-topic.
Sometimes, I encounter results in ClinicalTrials.gov which are yet to be published, or abstracts which do not have their full-texts available yet, or conference proceedings which do not include their methodologies in fine detail.
In this case, when the methods section is not addressed appropriately, what tools could be employed to assess the risk of bias/quality of such research types?
Thank you beforehand.
Unfortunately, with only-abstract, it is NOT possible to fully assess the bias. The abstract usually contains up to 300 words, hence full analysis of patient selection, control, intervention, flow & timing is not possible. I would conduct a separate analysis of bias from as much information as it is possible to obtain, following bias-tool guidelines specifically for the study design. However, the limitation of bias assessment MUST be acknowledged in the discussion/bias (best in both) section, and I'd suggest conducting a subgroup analyses, where these studies may be separately analysed from the full-text studies. Greetings
Firstly, try to track down if there are any published journal articles of the same study. You will be suprised how many are not linked to their trial records. Contact the corresponding author and view their profile in ResearchGate to see if any of their journal articles match up to the trial record.
In any case, assess the risk of bias the same as any other study. If you use Cochrane's ROB tool, it has NI (no information) as a possible answer to the signaling questions, so you can use this were neccessary. It is best to use the same risk of bias tool for all studies (of the same type), otherwise you will not be able to compare the results.
I think a 'sensitivity analysis' rather than a 'subgroup analysis' would be more appropriate. This would involve re-running your main meta-analysis, excluding these studies. However, this may or may not be neccessary depending on a variety of factors. For instance, if you do a sensitivity analysis excluding high and medium bias studies, and all studies with limited information are rated as high or medium bias, then another sensitivity analysis specifically excluding these studies would be unneccessary.
Joseph Whittaker Paweł Marek Łajczak
Thank you for your insights.
- Badges
- Science topic
- Similar topics
- Biological Science
- Anatomy
- Tissue