The Joinpoint-Jump and Joinpoint-Comparability Ratio Model for Trend Analysis with Applications to Coding Changes in...

Area of Science:

• Biostatistics
• Epidemiology
• Health Informatics

Background:

• Analysis of longitudinal health data is crucial for understanding population health trends.
• Coding system changes (e.g., ICD-9 to ICD-10) can introduce artificial jumps in time-series data, biasing trend estimates.
• Traditional joinpoint models may not adequately address these coding-induced discontinuities.

Purpose of the Study:

• To introduce novel statistical methods for incorporating discontinuous jumps into joinpoint models.
• To improve the accuracy of trend estimation in health data affected by coding changes.
• To provide tools for analyzing underlying continuous trends despite data artifacts.

Main Methods:

• Development of the Joinpoint-Jump model to directly estimate the size of data jumps.
• Development of the Joinpoint-Comparability Ratio model using supplementary data to estimate jump sizes.
• Application of these models to real-world health data, including cause of death and cancer staging.

Main Results:

• The proposed models effectively incorporate sudden coding-related jumps into joinpoint trend analysis.
• Estimates of underlying population trends are less biased when accounting for coding changes.
• Demonstrated utility in analyzing ICD-9/ICD-10 transition data and cancer staging data.

Conclusions:

• The Joinpoint-Jump and Joinpoint-Comparability Ratio models offer robust solutions for analyzing health data trends affected by coding shifts.
• Accurate trend estimation is vital for informing public health policy and interventions.
• These methods enhance the reliability of epidemiological analyses relying on historical health records.