Authors: Hayes,A. J.;Clarke,P. M.;Davis,W.

Publication: Value in Health

Year: 2012

Volume: 15

Issue: 4

Start Page: A191

ABSTRACT:

OBJECTIVES: To develop age and sex-specific risk equations for predicting mortality following major complications of diabetes, using a large linked administrative dataset from Western Australia (WA), (n=13884 patients) and to incorporate these into the UKPDS Outcomes Model. To compare the original and adapted models in predictions of survival and life expectancy following myocardial infarction, stroke, heart failure, amputation and renal failure, and incremental benefits associated with changes in common risk factors.

METHODS: We estimated a multivariate logistic regression model for the probability of death in the year of a complication, and a multivariate semi-parametric survival model (Gompertz) for years beyond the year of the complication. Covariates in the models included the type of complication, comorbidities, sex, type 1 diabetes and age. Using representative input data and clinical risk factors for Australian patients we ran Monte Carlo simulations of the original and adapted models. Parameter uncertainty was evaluated using 1000 bootstrapped coefficients of all model risk equations.

RESULTS: Simulated survival using Australian mortality equations fell inside the 95% confidence interval of observed survival, whilst survival using UK mortality equations fell outside of the interval. The two versions of the model generated differences in life expectancy following specific events; for example life expectancy of a 74 year old following myocardial infarction was 2.74 (95% CI 2.07-3.42) years for UK versus 4.33 (3.85-4.72) years for WA. However there was little impact of using alternative mortality equations on incremental QALYs gained as a result of reducing Hba1c or systolic blood pressure, or on aggregate outcomes of life expectancy for a cohort initially free of complications.

CONCLUSIONS: Mortality following major complications varies across diabetic populations and this can impact on estimates of life expectancy, but it appears to have less impact on incremental benefits of interventions that are commonly used in pharmacoeconomic analyses.

• Listing ID: 4584
• Author/s: Hayes,A. J.;Clarke,P. M.;Davis,W.
• Publication: Value in Health
• Year: 2012
• Volume: 15
• Issue: 4
• Start Page: A191
• Article Keywords: hemoglobin A1c;diabetes mellitus;simulation;mortality;Australia;society;pharmacoeconomics;outcomes research;model;life expectancy;survival;human;risk;United Kingdom;heart infarction;patient;risk factor;logistic regression analysis;death;kidney failure;amputation;prediction;population;systolic blood pressure;heart failure;cerebrovascular accident;confidence interval;Monte Carlo method;insulin dependent diabetes mellitus