Comparator choice in clinical trials of cancer drugs and recommendation categories in National Comprehensive Cancer Network (NCCN) guidelines: a retrospective cohort analysis
- 1Faculty of Law, University of Zurich, Zurich, Switzerland
- 2Department of Health Sciences and Technology, ETH Zurich, Zurich, Switzerland
Abstract
Background: Comparative evidence is important to determine whether new drugs provide added benefit over existing treatments. Ideally, such evidence is generated using appropriate comparators reflecting the standard of care. The National Comprehensive Cancer Network (NCCN) guidelines classify recommendations according to the level of evidence and panel consensus. Category 1 reflects high-level of evidence and categories 2A, 2B and 3 lower-levels of evidence.
Objective: We evaluated how the choice of a comparator was reflected in the NCCN recommendations.
Methods: We conducted a retrospective cohort analysis of cancer drugs approved by the US Food and Drug Administration between 2010 and 2024. For each indication, we identified all populations and assessed whether their corresponding efficacy trials used appropriate, inappropriate or no comparators. We extracted NCCN recommendation categories and compared them across comparator groups. We conducted statistical inference via bootstrap CIs.
Results: Our cohort included 110 indications comprising 176 populations. 56/110 (51%) indications had at least one population tested with an appropriate comparator, 35/110 (32%) had at least one population using an inappropriate comparator, 30/110 (27%) had at least one population based on a single-arm trial and 7/110 (6%) had at least one population for which no data was available. The average proportion of populations classified in NCCN category 1 was 6 percentage points higher with an appropriate than with an inappropriate comparator (95% CI −11 to 25), and 43 percentage points higher than with single-arm trials (95% CI 28 to 58).
Conclusion: Approximately half of the indications recommended in the NCCN guidelines had at least one population tested with an appropriate comparator. The average proportion of NCCN category 1 recommendations did not differ significantly between populations tested with an appropriate and inappropriate comparator but was significantly lower for populations supported by single-arm trials. Greater transparency regarding comparator choice could improve clinical decision-making.
Related
- Quality-recommendation paradox in adult hypothyroidism guidelines: a systematic appraisal of methodological quality, recommendation consistency, and citation impact
- Clinical Comparator Dependence of Inflammatory and Hematologic Biomarker Discrimination for Sepsis: A Matched Retrospective Study Using MIMIC-IV Emergency Department Data
- 2026 Asia Pacific Heart Rhythm Society (APHRS) Guidelines (Part IV) on Comprehensive Lifestyle Risk Factor and Comorbidity Management in Atrial Fibrillation.
- Clinical guidelines for the diagnosis and management of pediatric fecal incontinence: A comprehensive review and treatment algorithm
- Data Completeness and Recorded Advanced HIV Categories in a Regional Registry Extract from Northwestern Romania: A Retrospective Audit of 192 Records and Diagnosis Trends, 2013–2023
BibTeX
@article{glaus2026comparator,
title = {{Comparator choice in clinical trials of cancer drugs and recommendation categories in National Comprehensive Cancer Network (NCCN) guidelines: a retrospective cohort analysis}},
author = {Camille E G Glaus and Cyrill Meier and Simon Hediger and Lara Welti and Kerstin Noelle Vokinger},
journal = {BMJ Open},
year = {2026},
month = oct,
doi = {10.1136/bmjopen-2026-122939},
url = {https://bmjopen.bmj.com/lookup/doi/10.1136/bmjopen-2026-122939},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Comparator choice in clinical trials of cancer drugs and recommendation categories in National Comprehensive Cancer Network (NCCN) guidelines: a retrospective cohort analysis
AU - Camille E G Glaus
AU - Cyrill Meier
AU - Simon Hediger
AU - Lara Welti
AU - Kerstin Noelle Vokinger
PY - 2026
DA - 2026/10/28
JO - BMJ Open
DO - 10.1136/bmjopen-2026-122939
UR - https://bmjopen.bmj.com/lookup/doi/10.1136/bmjopen-2026-122939
AB - Background: Comparative evidence is important to determine whether new drugs provide added benefit over existing treatments. Ideally, such evidence is generated using appropriate comparators reflecting the standard of care. The National Comprehensive Cancer Network (NCCN) guidelines classify recommendations according to the level of evidence and panel consensus. Category 1 reflects high-level of evidence and categories 2A, 2B and 3 lower-levels of evidence. Objective: We evaluated how the choice of a comparator was reflected in the NCCN recommendations. Methods: We conducted a retrospective cohort analysis of cancer drugs approved by the US Food and Drug Administration between 2010 and 2024. For each indication, we identified all populations and assessed whether their corresponding efficacy trials used appropriate, inappropriate or no comparators. We extracted NCCN recommendation categories and compared them across comparator groups. We conducted statistical inference via bootstrap CIs. Results: Our cohort included 110 indications comprising 176 populations. 56/110 (51%) indications had at least one population tested with an appropriate comparator, 35/110 (32%) had at least one population using an inappropriate comparator, 30/110 (27%) had at least one population based on a single-arm trial and 7/110 (6%) had at least one population for which no data was available. The average proportion of populations classified in NCCN category 1 was 6 percentage points higher with an appropriate than with an inappropriate comparator (95% CI −11 to 25), and 43 percentage points higher than with single-arm trials (95% CI 28 to 58). Conclusion: Approximately half of the indications recommended in the NCCN guidelines had at least one population tested with an appropriate comparator. The average proportion of NCCN category 1 recommendations did not differ significantly between populations tested with an appropriate and inappropriate comparator but was significantly lower for populations supported by single-arm trials. Greater transparency regarding comparator choice could improve clinical decision-making.
SN - 2044-6055
SN - 2044-6055
ER -