Synthesising risk from summary evidence across multiple risk factors




Poster session 2 Thursday: Evidence synthesis - methods / improving conduct and reporting


Thursday 14 September 2017 - 12:30 to 14:00


All authors in correct order:

Shrier I1, Coldtiz GA2, Steele RJ3
1 Centre for Clinical Epidemiology, Lady Davis Institute, McGill University, Canada
2 Institute for Public Health, Washington University in St. Louis, United States
3 Department of Mathematics and Statistics, McGill University, Canada
Presenting author and contact person

Presenting author:

Ian Shrier

Contact person:

Abstract text
Background: Academic authors typically investigate risk factors in isolation. For example, one meta-analysis might report the increased risk of diabetes among smokers, another among those who are physically active, another among those who drink coffee and so on. Patients often want to know what their overall risk is across all factors, either absolute or compared to the general population. In 2004, Kim et al. validated the Harvard Cancer Risk Index formula where risk factors from summary information of reported results could be combined to provide an overall risk. De Vito et al. recently applied the same formula to coronary heart disease in 2015. The formula as published, is only applicable to dichotomous risk factors.

Objectives: To describe the underlying principles of the Harvard Cancer Risk Index formula, and to derive a general formula that allows for risk factors with any number of levels.

Methods: Mathematical and probability theory.

Results: Using summary data on risk ratios comparing participants with and without a risk factor, along with prevalence of the risk factor in the general population, it is possible to calculate the risk ratio for an individual with a level of a risk factor compared to the risk in the general population. Under an assumption of a multiplicative model without interactions, one can obtain the overall risk ratio for any individuals’ pattern of multiple risk factors to the risk in the general population. The absolute risk for the individual can also be obtained if the incidence of the disease in the general population is known.

Conclusions: Although these methods require numerous strong assumptions, they synthesise the evidence into a usable and helpful format that addresses a question many patients ask.