Applying GRADE guidance for choosing the right question in a systematic review supporting guideline development: a case study

Article type
Authors
Kawano L1, Zeraatkar D2, Tramujas L3, Agoritsas T4, Vandvik P5, Guyatt G6
1Magic Evidence Ecosystem Foundation, Oslo, Norway; Hcor Research Institute, Hcor Hospital, Sao Paulo, Brazil
2Magic Evidence Ecosystem Foundation, Oslo, Norway; Department of Health Research Methods, Evidence, and Impact, McMaster University, Hamilton, Canada
3Hcor Research Institute, Hcor Hospital, Sao Paulo, Brazil
4Magic Evidence Ecosystem Foundation, Oslo, Norway; Division General Internal Medicine, University Hospitals of Geneva, Geneva, Switzerland
5Magic Evidence Ecosystem Foundation, Oslo, Norway; Institute of Health and Society, Faculty of Medicine, University of Oslo, Oslo, Norway
6Magic Evidence Ecosystem Foundation, Oslo, Norway; Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Canada
Abstract
Background:
Every systematic review (SR) team faces the crucial choice of deciding on a broad or narrow question. Although GRADE offers guidance for the decision, challenges remain. We therefore offer a case study to illustrate challenges and possible solutions.

Methods:
A team of methodologists with expertise in the GRADE approach participated in the guideline development process addressing proactive therapeutic drug monitoring (TDM) of biologic drugs for inflammatory bowel disease (IBD), inflammatory arthritis, and psoriasis. The guideline panel faced controversy in defining the breadth of the question with respect to the population of interest. The methods team guided them through the process of making their decision.

Results:
The guideline aimed at addressing the impact of TDM in the three disease groups. The rationale behind proactive TDM (the intervention) is that it allows early detection of anti-drug antibodies and thus facilitate treatment adjustment to achieve optimal disease control. The panel was concerned that the risk of developing anti-drug antibodies differs across disease groups and TDM would be more effective in conditions in which that risk is higher. Some panel members therefore preferred formulating separate recommendations for each disease group while others preferred formulating a single recommendation.
Applying GRADE guidance, methodologists identified the key issue in deciding on a single versus multiple questions: is the biology consistent with a sufficiently high probability of a similar effect thus suggesting a sufficiently high probability that the magnitude of the effect will be similar across the range of disease populations considered.
Because the biology was uncertain, the methodologists suggested examining the available data (see figure). Results provided no support for a subgroup effect (effect appeared similar across each of the three groups), supporting formulating a single recommendation for all three groups.

Conclusion: GRADE provides a formal structure to address the choice of a broader or narrow question and recommendation. The example highlights how, when uncertainty remains regarding the consistency of effect across subgroups, conducting a formal subgroup analysis can resolve the issue.