Objectives: To identify methodological research on the incorporation of adverse effects in economic models and to review current practice.
Data sources: Major electronic databases (Cochrane Methodology Register, Health Economic Evaluations Database, NHS Economic Evaluation Database, EconLit, EMBASE, Health Management Information Consortium, IDEAS, MEDLINE and Science Citation Index) were searched from inception to September 2007. Health technology assessment (HTA) reports commissioned by the National Institute for Health Research (NIHR) HTA programme and published between 2004 and 2007 were also reviewed.
Review methods: The reviews of methodological research on the inclusion of adverse effects in decision models and of current practice were carried out according to standard methods. Data were summarised in a narrative synthesis.
Results: Of the 719 potentially relevant references in the methodological research review, five met the inclusion criteria; however, they contained little information of direct relevance to the incorporation of adverse effects in models. Of the 194 HTA monographs published from 2004 to 2007, 80 were reviewed, covering a range of research and therapeutic areas. In total, 85% of the reports included adverse effects in the clinical effectiveness review and 54% of the decision models included adverse effects in the model; 49% included adverse effects in the clinical review and model. The link between adverse effects in the clinical review and model was generally weak; only 3/80 (< 4%) used the results of a meta-analysis from the systematic review of clinical effectiveness and none used only data from the review without further manipulation. Of the models including adverse effects, 67% used a clinical adverse effects parameter, 79% used a cost of adverse effects parameter, 86% used one of these and 60% used both. Most models (83%) used utilities, but only two (2.5%) used solely utilities to incorporate adverse effects and were explicit that the utility captured relevant adverse effects; 53% of those models that included utilities derived them from patients on treatment and could therefore be interpreted as capturing adverse effects. In total, 30% of the models that included adverse effects used withdrawals related to drug toxicity and therefore might be interpreted as using withdrawals to capture adverse effects, but this was explicitly stated in only three reports. Of the 37 models that did not include adverse effects, 18 provided justification for this omission, most commonly lack of data; 19 appeared to make no explicit consideration of adverse effects in the model.
Conclusions: There is an implicit assumption within modelling guidance that adverse effects are very important but there is a lack of clarity regarding how they should be dealt with and considered in modelling. In many cases a lack of clear reporting in the HTAs made it extremely difficult to ascertain what had actually been carried out in consideration of adverse effects. The main recommendation is for much clearer and explicit reporting of adverse effects, or their exclusion, in decision models and for explicit recognition in future guidelines that 'all relevant outcomes' should include some consideration of adverse events.