The research team leveraged multi-payor claims data to build a cohort of near-continuously insured adults aged 50 to 74 who, as of December 2020, had no prior diagnosis of colorectal cancer and had not undergone a total colectomy. A 10-year lookback period was used to establish insurance enrollment and screening history. Rather than relying on county-level averages, the investigators created bespoke small areas at the intersection of counties and ZIP codes, allowing aggregated claims data to be linked through patients’ residential ZIP codes and counties.

Statewide, the findings showed consistently high screening coverage over the study window: from 2012 to 2020, between 79.8% and 81.2% of continuously insured screen-eligible adults aged 50 to 75 were up to date with colorectal cancer screening. Endoscopy remains the most common modality by which screen-eligible adults achieve up-to-date status. In 2020, at the small-area level, the mean prevalence of being up to date was 79.6%, with a standard deviation of 9.5 percentage points, indicating that while the state average was high, individual communities varied considerably around that mean.

A key methodological step involved testing whether standard regression assumptions held for the geographic data. Diagnostic tests of an Ordinary Least Squares model revealed a significant degree of spatial dependence in the error terms, with a Moran’s I statistic of 0.1540 and a p-value below 0.00001. Neighboring small areas tended to have similar residuals, meaning a conventional model that ignores geography would produce misleading inference. The researchers therefore turned to a spatial statistical framework designed explicitly for areal data.

The team fit a Conditional Autoregressive (CAR) Bayesian generalized linear model with a logistic link function, employing Leroux priors to account for spatial autocorrelation among the small areas. The spatial model estimated an inverse relationship between the minimum distance to access an endoscopy facility and the prevalence of being up to date with colorectal cancer screening, with a point estimate of −0.0645 and a credible interval spanning −0.1156 to −0.0103. Because the credible interval excludes zero, the association is unlikely to reflect random variation alone.

The study’s scope and limitations are important for interpreting the results. The cohort was restricted to continuously insured adults, so the estimates do not represent uninsured residents or people with gaps in coverage. The analysis captures screening status as of December 2020 and reflects the insured population of North Carolina rather than the full adult population. In addition, the reported association between distance and screening prevalence is an ecological, area-level relationship; it describes patterns across communities rather than individual-level behavior, and the observational design does not establish that distance itself causes changes in screening uptake.

Even with those caveats, the study demonstrates how routinely collected claims data can be used to map cancer screening coverage at a fine geographic scale. Small-area estimation can identify communities where screening lags despite strong statewide averages, and the measured link between endoscopy access and up-to-date prevalence gives health planners a geographic signal to examine when directing screening resources across North Carolina.

Subject of Research: Small-area estimation of colorectal cancer screening prevalence and endoscopy access in North Carolina

Article Title: Leveraging claims data-based small-area estimation to evaluate endoscopy access and the area-level prevalence of being up-to-date with colorectal cancer screening among screen-eligible adults in North Carolina

Article References: Planey, A. M., Jackson, B. E., Kuo, T.-M., Reich, B. J., Wheeler, S. B., Reuland, D. S., Ganguly, A. P., Baggett, C. D., Lund, J. L., & Reeder-Hayes, K. (2026). Leveraging claims data-based small-area estimation to evaluate endoscopy access and the area-level prevalence of being up-to-date with colorectal cancer screening among screen-eligible adults in North Carolina. Cancer Causes & Control, 37(11), 185. https://doi.org/10.1007/s10552-026-02272-8

Image Credits: AI Generated

DOI: 10.1007/s10552-026-02272-8

Keywords: colorectal cancer screening, endoscopy access, small-area estimation, claims data, Bayesian spatial model, North Carolina, health geography, spatial autocorrelation, Cancer Causes & Control, screening prevalence, Leveraging, claims