A change-point regression approach for estimating no observed adverse effect level from systematic review.

Abstract

Systematic reviews can be used not only to evaluate the efficacy and usefulness of a drug or food ingredient, but also as a safety assessment method. One of the aims of safety assessment is to estimate the no observed adverse effect level and the lowest observed adverse effect level. However, no methodology to statistically estimate the no observed adverse effect level from systematic review results has yet been reported. Estimation of the no observed adverse effect level involves a search for the dose above which adverse events occur is even exploration of the thresholds in dose response. To search for the dose above which adverse events occur, we examined an estimation method using the weighted change-point regression model, which includes the weights of each study used for systematic reviews in the model. This model could be applied to safety data of an omega-3 study in the form of a systematic review. We demonstrated that the dose response to omega-3 intake regarding adverse events had a threshold value and that the no observed adverse effect level could be estimated using the developed model.