Review of Methods used to Estimate Treatment Effects against Relevant Comparators using Evidence from Single-Arm Studies in NICE Single Technology Appraisals

Abstract

OBJECTIVES: The aim of this review of National Institute for Health and Care Excellence (NICE) single technology appraisals (STAs) was to determine how comparisons against relevant comparators have been performed using evidence about new treatments obtained from single-arm studies when individual patient data (IPD) is available only for one study and summary level data were available for the rest of the studies.

METHODS: All the STAs identified within 1st January 2018 to 31 December 2023 in NICE website have been searched to identify relevant STAs involving single-arm studies. Moreover, information was extracted on how prognostic and effect modifier variables have been identified and how survival extrapolation has been conducted in a data extraction form.

RESULTS: A total of 20 TAs were identified where the pivotal study/studies were single-arm study and 16 (76.19%) have used population adjustment methods. All the STAs included in this review were in oncology. Matching Adjusted Indirect Comparison (MAIC) was mostly used population adjustment method (13 out of 20, 61.90%) whereas Simulated Treatment comparison (STC) was the second mostly used method (4 out of 20, 19.04%). More than fifty percent of these appraisals (12 out of 20, 57.14%) have multiple comparisons of treatments with a larger network. The common approach for extrapolation of data beyond the observed period was to fit parametric models with unadjusted IPD of intervention study. Furthermore, identification of variables was mostly done by literature search and clinical expert opinion.

CONCLUSIONS: It is not surprising that adjustment methods like unanchored MAIC and STC were frequently used in NICE STAs with the increase of single-arm studies. These methods make assumptions about the variables included in the analysis which are difficult to satisfy and can produce residual bias.