NMAsurv: An R Shiny application for network meta-analysis based on survival data

Area of Science:

• Biostatistics
• Health Informatics
• Medical Research Methodology

Background:

• Network meta-analysis (NMA) is crucial for comparing multiple interventions across diverse trials, particularly for medical applications.
• Traditional NMA for survival data relies on the proportional hazards (PH) assumption, synthesizing only hazard ratios (HRs).
• Advanced non-PH NMA methods exist but require significant programming expertise, limiting their adoption.

Purpose of the Study:

• To introduce NMAsurv, a user-friendly R Shiny tool designed for survival-data-based NMA.
• To enable researchers with limited R programming experience to conduct complex survival NMA.
• To provide an intuitive platform for various NMA functions, including model building and assumption testing.

Main Methods:

• Development of NMAsurv, an R Shiny application accessible via a web interface.
• Support for both pseudo-individual participant data and aggregated data inputs.
• Implementation of a point-and-click interface for network plotting, PH assumption testing, and NMA model estimation (including fractional polynomial, piecewise exponential, parametric, Cox PH, and generalized gamma models).

Main Results:

• NMAsurv offers effortless execution of survival NMA, including survival and HR plot generation.
• The tool accommodates various NMA models and data input types.
• Demonstration of NMAsurv's utility through a real-world NMA example.

Conclusions:

• NMAsurv significantly lowers the barrier to entry for performing advanced survival NMA.
• The tool empowers a wider range of researchers to conduct sophisticated analyses of time-to-event data.
• NMAsurv facilitates robust comparative effectiveness research in medicine and other fields utilizing survival analysis.