OBJECTIVES: In the absence of head-to-head comparison, non-randomized comparative data may be available separately as IPD for both the investigational and control cohorts. NICE guidelines have suggested several steps to select the appropriate ITC method(s). This review aims to assess the use of these methods in submissions to NICE.

METHODS: All NICE technology appraisals (TAs) published between 2015-2019 with an ITC using IPD on the investigational and control cohorts were included. Information on interventions, indication, pivotal study design, data sources, ITC methodologies, endpoints, criticism, and NICE recommendation were extracted.

RESULTS: Out of 159 reviewed TAs, eight TAs were included. The included submissions were for indications in blood cancers (n=4) and solid cancers (n=4). The pivotal study design was an RCT and non-RCT in four and five TAs, respectively. All included submissions received positive recommendation: three were recommended for routine commissioning and six were accepted within the cancer drug fund. Data sources for comparators came from retrospective studies (n=4), RCTs (n=2) and single-arm studies (n=2). Most of the submitted evidence was based on non-UK studies (n=7). Five TAs used more than one methodology. These methods included covariate adjustment (n=5), propensity score matching (n=4), naïve comparison (n=2), and inverse probability weighting (n=1). Key criticism included differences in baseline characteristics, study design, endpoints definition, treatment duration. Criticisms regarding high uncertainty due to immature data, small sample size and lack of adjustment for potential confounders were also mentioned. When the impacts of these factors were shown to be minimal, by means of sensitivity analyses, ITCs were accepted.

CONCLUSIONS: Despite the availability of methods to adjust the comparison of non-randomized IPD, the use of these methodologies is still challenged given the limitations of real-world data.