End-of-life imaging spending key to evaluating participation in advanced payment models

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - healthcare spending

A recent study in the American College of Radiology aimed to assess spending patterns on high-cost imaging during the final three months of life.

“We examined the relationship between end-of-life care episodes’ spending trajectories and imaging utilization through both descriptive and inferential analysis,” the authors, led by Timothy Copeland of the University of California, San Francisco, wrote. “Identifying patient spending patterns and understanding their drivers may be an avenue for imaging departments to improve estimations of departmental costs if they agree to provide all imaging services under a capitated payment plan.”

Collected records from an academic cancer center’s radiology department, cancer registry and claims were matched to identify individuals who died between April 2013 and June 2014.

Patients spending patterns in their final year of life were identified using group-based trajectory-modeling. A total of six spending trajectories were identified and recorded, including high persistent, low persistent, early rising, late rising, rise and decline, and early decline.

Analysis of CT, MRI and PET across all trajectories was recorded. Multivariate logistic regressions modeled the likelihood of imaging utilization in the final three months of life, and a sensitivity analysis assessed the impact of spending trajectories on model fit.

  • Membership in the late rising trajectory was the strongest predictor of high-cost imaging in the final three months of life followed by diagnosis 12 to six months pre-mortem.
  • The likelihood of imaging the final three months of life was no different between high persistent and low persistent trajectory patients, despite the heterogeneity between the two patient groups.
  • Sensitivity analysis indicated that spending trajectory improved the prediction of imaging in the final three months of life to a greater extent than temporal proximity to death at the time of diagnosis, which may serve as a proxy for severity and/or complexity. 


“In this study, we demonstrated that clinical measures of severity and patient history of utilization must be considered by hospital administrators in their estimations of oncologic imaging utilization,” the authors wrote. “Furthermore, adopting analytic approaches that consider patients’ spending patterns may aid in evaluating participation in advanced payment models.”