Suppose I have patients with these events (at varying/individual times):

  • change in insurance
  • change in Health care provider
  • adverse event
  • late prescription fill
  • missed prescription fill

I also know from prescription fill records when treatment is dropped - this is the terminating event - but not all patients terminate.

Each of these events can happen one or multiple times in any order (or none). I'm treating them as independent (even though the HCP change could be triggered by the insurance change).

What I would like to model is the increase in risk for dropping treatment (over the background risk) based upon the individual events happening to the patient. The important item to capture is the increased risk as each event occurs and hopefully how the order affects the risk. The idea would be to provide a risk evaluation for a given patient as any of these events occur to guide interventions towards patients most at risk over the background risk.

Background risk over time - I would think - can be provided by survival analysis of patients with none of these events but that dropped treatment.

I've looked at hidden Markov models, survival analysis with multiple events and the Fine-Grey model. I'm looking to understand what direction to go from a modeling perspective.

If this is essentially answered elsewhere please point me on...this is my first question so I hope I've provided enough information.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.