The Importance of Proximity to Death in Modelling Community Medication Expenditures for Older People: Evidence from New Zealand

Background: Concerns about the long-term sustainability of health care expenditures (HCEs), particularly prescribing expenditures, has become an important policy issue in most developed countries. Previous studies suggest that proximity to death (PTD) has a significant effect on total HCEs, with its exclusion leading to an overestimation of likely growth. There are limited studies of pharmaceutical expenditures in which PTD is taken into account.
Objective: This study presents an empirical analysis of public medication expenditure on older individuals in New Zealand (NZ). The aim of the study was to examine the individual effects of age and PTD using individual-level data.
Methods: This study uses individual-level dispensing data from 2008/2009 covering the whole population of medication users aged 70 years or older and resident in NZ. A case–control methodology was used to examine individual cost and medication use for a 12-month period for decedents (cases) and survivors (controls). A random effects two-part model, with a Probit and generalized linear model (GLM) was used to explore the effect of age and PTD on expenditures. Results: The impact of PTD on prescription expenditure is not as dramatic as studies reporting on acute and/or long-term care. The 12-month decedent-to-survivor mean expenditure ratio was 1.95; 2.09 for males and 1.82 for females. The additional cost of dying in terms of prescription drugs decreases with age, with those who die at 90 years of age or older consuming fewer drugs on average and having a lower mean expenditure than those who died in their 70s and 80s. The following variables were found to have a decreasing effect on the mean monthly prescription expenditures: a reduction of 2.2 % for each additional year of age, 4.2 % being in the Maori ethnic group, and 7.8 % for Pacific Islanders. Increases in monthly expenditure were associated with being a decedent 32.1–62.6 % (depending on month), being of Asian origin 16.2 %, or being a male 12.6 %.
Conclusions: Given the variance reported between survivors and decedents, future projections should include PTD in their models to improve accuracy. Policies targeted at reducing expenditures should not focus on age but on ensuring appropriate and cost-effective prescribing, particularly towards the end of life.