Synthesis of continuous and time-to-event outcomes is often complicated by the use of multiple outcome scales and heterogeneous reporting of outcomes across trials. Simple methods of evidence synthesis for clinical effectiveness can fail to account for these issues and result in a reduction of the evidence base, which can be further reduced at the cost-effectiveness stage as common outcome measures, such as standardized mean differences, cannot easily be incorporated into the economic decision model. Recent methodological advances for synthesizing continuous and time-to-event outcomes aim to include a greater proportion of the available evidence base within a single coherent analysis.

To assess the statistical methods commonly used in health technology assessment (HTA) and establish whether recent advances in synthesis methods have been adopted in practice, we conducted a review of HTA reports and guidelines published in the United Kingdom (UK) between 1 April 2018 and 31 March 2019 reporting a quantitative meta-analysis (MA), network meta-analysis (NMA) or indirect treatment comparison (ITC) of at least one continuous or time-to-event outcome.

Forty-seven articles were considered eligible for this review. Fifty-one percent of eligible articles reported at least one continuous outcome and 55 percent at least one time-to-event outcome. Twenty-nine articles reported NMA or ITC and twenty-seven reported MA of a continuous or time-to-event outcome. Forty articles included a decision model, of which twenty-seven incorporated evidence from a synthesis of a continuous or time-to-event outcome with eleven informed by a single trial (despite synthesis being conducted).

Uptake of methods to include a greater proportion of the available evidence base within a single coherent analysis in UK HTA reports has been slow. Evaluating health technologies using an evidence-based approach often results in better outcomes for patients. Therefore, HTA analysts and decision modelers must be aware of the expanding literature for synthesis of continuous and time-to-event outcomes and appreciate the limitations of simpler approaches.