Abstract

There is little current evidence on the cost of cancer in the era of (expensive) stratified medicines. Even less is known about the relationship between pharmaceuticals and medical services, emergency presentations and hospitalisations, and particularly whether the different types of care are substitutes or complements. This paper estimates the current expenditure of delivering different types of cancer care, compares the determinants of expenditure, and attempts to identify inter-relationships, using a large linked dataset: Cancer 2015. Cancer 2015 is a prospective longitudinal population-based molecular cohort study in Victoria, Australia. The cohort data (clinical, genomic, patient-demographics and treatment intention) have been linked to administrative reimbursement data (Commonwealth and State health care resource use and cost).

We estimate the bi-monthly cost of each type of health care resource use (pharmaceuticals; medical services, which includes general practice visits, pathology tests and radiology; emergency presentations; hospital admissions and day case visits) for a two year period. A panel regression for total health expenditure is estimated and a fractional multivariate logit model is used to estimate the determinants of the various shares of health expenditure, before predicting out the effects on total expenditure using the estimated parameters.

The results offer insight regarding forward budget planning, not only do they provide estimates of the various shares of health expenditure, but also the effect of possible policy and practice changes on both total health expenditure and which budget bears the cost.