DEVELOPMENT AND VALIDATION OF A NORWEGIAN MODEL FOR COLORECTAL CANCER

Purpose:

Colorectal cancer (CRC) is a leading cause of cancer deaths in most industrialized countries. The aim of this study was in a health care perspective to (i) describe a CRC model to capture lifetime costs and survival and (ii) validate the model.

Method:

We applied a semi-Markov model with 70 health states and tracked age and time since specific health states (using tunnels and three-dimensional data matrix). The model parameters were based on follow-up of 2,049 CRC-patients at Oslo University Hospital (OUS)(survival), the National Patient Register (resource use), published literature and expert opinion. The model followed patients diagnosed with CRC at age 70 until death. The model can address a range of CRC-related themes (a general model) and is adaptable to other populations.

Result:

The model was validated for face-, internal-, cross- and external validity. The validation showed a satisfactory match with other models and empirical estimates for both cost and survival time, without any preceding calibration of the model.

   In a cross validation, 10-year overall survival, weighted for stage and estimated by the model differed by 11.5 days (0.38 months) compared with the OUS data. The costs of our model were 0.3% lower than the prediction by an Irish model (3.0%, 1.3%, -3.6% and 1.2% lower in the Irish model for stages I, II, III and IV, respectively).

   For external validation, we compared relative survival estimated by the model with patients monitored by the Cancer Registry of Norway which contains a complete set of data for CRC patients in Norway. The model predicts 3.9% higher relative survival than national data during the first year, and 0.9%, 5.6% and 5.6% lower relative survival 5, 10 and 15 years after diagnosis, respectively.

   In another external validation, the model was compared to empirically estimated (“model-free”) total costs based on a Norwegian population study (National Patient Register). Accounting for differences in assumptions, the model estimate was 3.1% higher than the model-free estimate.

 Conclusion:

The Norwegian CRC model predicts well stage specific survival and costs and may be valid for estimating health and resource consequences of CRC policies.