A modeling approach to derive baseline risk estimates for GRADE recommendations: Concepts, development, and results of its application on a guideline.

Article type
Morgano GP1, Wiercioch W1, Anderson DR2, Brożek JL1, Santesso N1, Xie F1, Cuker A3, Nieuwlaat R1, Akl EA4, Darzi A1, Yepes-Nuñez JJ5, Exteandia-Ikobaltzeta I1, Rahman M6, Rajasekhar A7, Rogers F8, Tikkinen KAO9, Yates AJ10, Dahm P11, Schünemann HJ12
1Department of Health Research Methods, Evidence and Impact, McMaster University
2Department of Medicine, Dalhousie University
3Department of Medicine and Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania
4Department of Internal Medicine, American University of Beirut
5School of Medicine, Universidad de los Andes
6Lillian S. Wells Department of Neurosurgery, University of Florida
7Division of Hematology and Oncology, Department of Medicine, University of Florida
8Trauma and Acute Care Surgery, Penn Medicine Lancaster General Health
9University of Helsinki and Helsinki University Hospital
10Department of Orthopedic Surgery, University of Pittsburgh Medical Center
11Department of Urology, University of Minnesota
12Department of Medicine and Department of Health Research Methods, Evidence, and Impact, McMaster University

Baseline risks are required to calculate absolute effect estimates, which are essential elements of evidence summaries produced for guideline panels. Systematic reviews of prognostic observational studies are scarce and the available estimates are often not directly applicable to patient-important outcomes. In some contexts, guideline panels revert to using surrogates to estimate baseline risks but this approach may introduce bias in the estimates of anticipated absolute effects.


To develop an approach to model baseline risks for patient-important outcomes prioritized for recommendations when only baseline risks for surrogate outcomes are available.


This study was part of the American Society of Hematology (ASH) guidelines for the management of venous thromboembolism (VTE). The McMaster University GRADE Centre and the ASH guideline panel for the prevention of VTE in surgical patients developed a modeling approach based on explicit assumptions about the distribution of symptoms, anatomical location, and severity of VTE events.


We applied the approach to derive modeled estimates of baseline risk. These estimates were used to calculated absolute measures of anticipated effects that informed the discussion of the evidence and the formulation of 30 recommendations. The approach increased transparency and reduced potential error in the decision-making process.


Our approach can assist guideline developers facing a lack of information about baseline risk estimates that directly apply to outcomes of interest. It also addresses potential bias of over- or underestimating absolute anticipated effects of interventions that can result from the use of surrogate data.

Patient or healthcare consumer involvement:

Patients representatives were included in the guideline panel and contributed to the development of the model assumptions.