A methodological review of population-adjusted indirect comparisons reveals inconsistent reporting and suggests publication bias.

OBJECTIVES: Population-adjusted indirect comparisons (PAICs) were developed in the 2010s to allow for comparisons between two treatments evaluated in different trials while accounting for differences in patient characteristics if individual patient data (IPD) are available for only one trial. Such comparisons are increasingly used in market access applications when a pharmaceutical company compares its new treatment (with IPD available) to another treatment developed by a competitor (with only aggregated data available). This study aimed to describe the characteristics of these PAICs, assess their methodology, and describe the reported results. STUDY DESIGN AND SETTING: Original articles reporting the use of at least one PAIC were searched on PubMed between January 1, 2010 and April 2, 2022. Two reviewers independently selected articles and extracted data. RESULTS: We included 133 publications reporting the results of 288 PAICs. Half of the articles were published on or after May 7, 2020, and 71 (53%) pertained to onco-hematology. The pharmaceutical industry was involved in 130 (98%) articles. Key methodological aspects were reported inconsistently, with only three articles adequately reporting all aspects. A total of 161 (56%) articles reported a statistically significant benefit for the treatment evaluated on IPD. Conversely, only one PAIC significantly favored the treatment evaluated on aggregated data. CONCLUSION: Although the number of published PAICs is increasing, the methodology and transparency need to be improved. Moreover, our study strongly suggests a reporting bias. This situation calls for strengthening guidelines to improve trust in PAIC results and thus their reliability in market access applications.

Diagnosis of chylous abdominal effusions: what is the triglyceride threshold value?

Introduction: Chylous abdominal effusions are serious complications that can be triggered by various aetiologies. The biochemical diagnosis of chyle leakage in ascites or in peritoneal fluid capsules relies on the detection of chylomicrons. Assaying the fluid's concentration of triglycerides is still the first-line tool. Given that only one comparative study has sought to quantify the value of the triglyceride assay for diagnosing chylous ascites in humans, our objective was to provide practical triglyceride thresholds. Materials and methods: We conducted a 9-year, retrospective, single-centre study of adult patients and compared a triglyceride assay with lipoprotein gel electrophoresis for the analysis of 90 non-recurring abdominal effusions (ascites and abdominal collections) of which 65 were chylous. Results: A triglyceride threshold of 0.4 mmol/L was associated with a sensitivity > 95%, and a threshold of 2.4 mmol/L was associated with a specificity > 95%. According to Youden index, the best threshold was 0.65 mmol/L with a sensitivity of 88 (77-95)%, a specificity of 72 (51-88)%, and, in our series, a positive predictive value of 89 (79-95)% and a negative predictive value of 69 (48-86)%. Conclusions: In our series, cut-off of 0.4 mmol/L could be used for ruling-out diagnosis of chylous effusions, while cut-off of 2.4 mmol/L could be used for reasonably confirming diagnosis.