JPsurv: Survival Trend Analysis

JPsurv is a statistical tool for analyzing trends in cancer survival by year of diagnosis. It uses joinpoint models to evaluate changes in survival over calendar time, accounting for both year of diagnosis and time since diagnosis. The tool supports analyses of conditional survival and provides options to evaluate and relax proportional hazards assumptions.

Key Features

• Analyze survival trends by year of diagnosis
• Identify changes in trends using joinpoint models
• Estimate conditional survival trends
• Evaluate and relax proportional hazards assumptions
• Support relative and cause-specific survival

Methods Overview

JPsurv fits proportional hazards joinpoint models on the log hazard scale. The hazard of cancer death is modeled as a baseline hazard over time since diagnosis and a multiplicative effect of year of diagnosis and covariates. The year-of-diagnosis effect is represented as connected linear segments, with the number and location of joinpoints estimated from the data.

Data Requirements

Discrete-time survival data in life table format, grouped by time since diagnosis. Supports relative and cause-specific survival. JPSurv can support data exported directly from SEER*Stat and Excel, CSV, or txt files, as long as they adhere to a certain format and contain the necessary variables to fit the JPSurv model.

Outputs

Estimates of survival trends over time, changes in hazard patterns, and conditional survival. In addition, model parameters, joinpoint locations, and trend measures for survival are produced.

What’s New

• Conditional survival trend modeling
• Options to evaluate and relax proportional hazards assumptions

Part of the JPsurv Suite

JPsurv focuses on trend analysis. For survival modeling and cure models, see CanSurv.