Tackling Heterogeneity in Indirect Treatment Comparison (ITC): A Comparative Review of NICE Submissions in Oncology Versus Non-Oncology

OBJECTIVES: Heterogeneity in indirect treatment comparison (ITC) is expected, and various advanced statistical methods exist to address it. A previous review evaluated how heterogeneity is assessed in non-oncology submissions to NICE. This study aims to: (1) understand how heterogeneity is assessed in NICE oncology submissions, and (2) compare the approaches to handling heterogeneity in oncology versus non-oncology submissions.

METHODS: We reviewed NICE single technology appraisals (STA) published on oncology between April 2022 and April 2024. Terminated, in-development, and withdrawn submissions were excluded. The findings from oncology submissions were compared to findings from non-oncology submissions from the previous review.

RESULTS: A total of 142 STAs were reviewed, with 70 pertaining to oncology indications. ITCs were performed in 41 out of these 70 oncology STAs, with 21 being anchored comparisons. In these 21, heterogeneity was assumed to be absent or could not be assessed in 5 STAs, and heterogeneous studies were excluded from analysis in 3 STAs. Among the 13 STAs reporting heterogeneity, population-adjusted indirect comparison methods (PAICs), including matching-adjusted indirect comparisons (MAIC; n=3) and propensity score matching (PSM; n=2), were most frequently used. Other approaches included random effects models (n=2), assessment using I2 estimates (n=2), identification of treatment effect modifiers (n=2), and meta-regression (n=1).

The use of PAICs was more frequent in oncology submissions (38%; 5/13) compared to non-oncology submissions (4%; 1/25). Heterogeneity in ITCs is mainly caused by imbalances in treatment effect modifiers. The assessment of heterogeneity was carried out more frequently in oncology submissions compared to non-oncology submissions. However, a higher proportion of oncology submissions did not report any evidence regarding heterogeneity (38%; 8/21) compared to non-oncology submissions (31%; 11/36).

CONCLUSIONS: Compared to non-oncology submissions, advanced methods such as PAICs to assess heterogeneity are more commonly used in recent oncology submissions to NICE.