Jump Model and Comparability Ratio Model

Trend data are often recorded or classified based on certain types of coding systems, and changes in code structure or coding rules over time are not uncommon. For example:

  • ICD-9/ICD-10 cause of death coding changes
  • Coding changes in the staging of cancer

The coding changes could cause a sudden increase/decreases, or "jumps" in data, even though the underlying trends has not changed. Joinpoint models that ignore data jumps due to coding changes may produce biased estimates of trends. Two models are developed to incorporate a sudden discontinuous jump in an otherwise continuous joinpoint model:

  • The Joinpoint-Comparability Ratio model (JP-CR) utilizes a pre-estimated size of the jump from supplementary data
  • The Joinpoint-Jump model (JP-Jump) simultaneously estimates the size of the jump, as well as the changes in trend

The basics of these models, the implementation in software, model selection, and examples are given in the Overview page.

The trend of US mortality (1975-2018) for Melanoma shown in SEER*Explorer is modeled by standard Joinpoint model. With the ICD9/ICD10 coding change, the JP-CR and JP-Jump models can be applied to model the trend. The results are shown in Alternate Explorer Tables for Melanoma.

Suggested Citation

Chen HS, Zeichner S, Anderson RN, Espey DK, Kim HJ, Feuer EJ. The Joinpoint-Jump and Joinpoint-Comparability Ratio Model for Trend Analysis with Applications to Coding Changes in Health Statistics. J Off Stat. 2020;36(1):49-62. doi:10.2478/jos-2020-0003

Last Updated: 05 May, 2021