You are raising interesting points. Assuming that all included studies in the systematic review adequately fit the pre-defined review question elements (PICOMT), then it would be such a waste if exclusion will be done in the assessment just because some of the studies do not provide "meta-analyzable" data. I think a compromise will be to assess the evidence quality of the "meta-analyzable" and "non-meta-analyzable" studies separately but in the same report. Forcing all of the studies in a single GRADE assessment may not be worth the complication and effort.

Please consider the alternatives to the standard GRADE summary of findings tables mentioned in this trial protocol: Article Two alternatives versus the standard Grading of Recommendati...

Good luck!

Bruna Barcelos So I have performed my systematic reviews. However meta-analysis was not feasible due to Heterogeneity. I did GRADE for all sub groups anyway.

You can do meta-analysis and this does not invalidated GRADE, because GRADE will provide a recommendation for that evidence. I suggest if you have a group of studies that you can perform meta-analysis, you will group these studies and evaluate their GRADE together and for the others where meta-analysis is not possible, you can group these studies according to their effectiveness or type of interventions and you evaluate the GRADE for all of them.

It is not mandatory to GRADE the evidence just when meta-analysis is performed. You can use GRADE for studies included in the narrative review as well.

I am happy to help if you need further assistance.

If you want to read more how I did mine please check my thesis from the School of Nursing, USP on the link below:

Thesis Intervenções para a redução dos erros de identificação do pa...

I think you should perform GRADE analysis for only quantitative synthesis, in the situation you performed quantitative synthesis after did qualitative synthesis. But if you only performed qualitative synthesis, you can include all of the studies in the GRADE analysis.

These are the things I have used previously;

I judged the imprecision based on the optimal information size criteria and confidence interval. To rate downgrade the imprecision;

· If an OIS criterion was not met, I rated down for impression unless the sample size was very high (at least 2000 participants for the dichotomous outcome and 400 for continuous outcome).

· If the OIS criteria were met and the 95 % confidence interval excludes no effect (i.e. CI around RR exclude 1.00 for the dichotomous outcome and MD exclude 0.00 for continuous outcome), I judged it as adequate precision.

· If the OIS met and confidence interval overlaps no effects (i.e CI includes RR of 1.00 for dichotomous and MD exclude 0.00 for continuous), I rated down if the CI fails to exclude important benefit or important harm.

Hello Dear

I have had very similar experience. You can find this meta-analysis with GRADE for the level of evidence

https://journals.lww.com/ajpmr/Fulltext/2018/11000/Effectiveness_of_Physiotherapy_Interventions_on.4.aspx

Dear Bruna,

You seem to have inadvertently stumbled upon one of the many imperfections in research. If a system allows us to 'do a few systematic reviews in just a few weeks' then I think its fair to say something is probably wrong with the system. I published the attached photo (2015) to depict what we are actually doing in a meta-analysis of RCTs. Apparently all RCT's are valued as 'equal and the same' although some clearly differ from others (as do apples and oranges; 'apples and pears' in Dutch). In GRADE - introduced by Gordon Guyatt, who also coined the term 'evidence-based medicine' - a superbly executed study can be downGRADEd because it wasn't randomized, while a badly conducted study is appraised higher just because it's an RCT. A systematic review (or meta-analysis) is a label that should be exclusively used for reviews where ALL types of evidence and ALL types of research are considered and appreciated. Please read:

Charlton, B. The Zombie science of evidence-based medicine: a personal retrospective. A commentary on Djulbegovic, B., Guyatt, GH  Ashcroft, RE (2009). J Eval Clin Pract 15, 930–934 (2009).

Ioannidis, J. P. A. The Mass Production of Redundant, Misleading, and Conflicted Systematic Reviews and Meta-analyses. Milbank Q 94, 485–514 (2016).

Mercuri, M. & Baigrie, B. S. What confidence should we have in GRADE? J Eval Clin Pract 24, 1–7 (2018).

I hope this helps

Ron Legerstee

Hello Dear,

You can Grade the studies to assess their quality and plan a sensitivity analysis including the study you excluded to see if their exclusion had a significant impact on the pooled effect you have obtained with high quality studies.

Also consider that not all systematic reviews can be meta analyzed. As you stated the data must be comparable to be included. As you mentioned that you have few studies I will not stress on the importance of sensitivity analysis for subgroups, so what I think you can do, is to see if the meta-analysis is feasible and if not, conduct a systematic review only.

https://handbook-5-1.cochrane.org/chapter_9/9_7_sensitivity_analyses.htm#:~:text=The%20process%20of%20undertaking%20a%20systematic%20review%20involves%20a%20sequence%20of%20decisions.&text=A%20sensitivity%20analysis%20is%20a,that%20were%20arbitrary%20or%20unclear.