Diabetes Res Clin Pract. 2014 Jan 14. pii: S0168-8227(13)00466-X. doi: 10.1016/j.diabres.2013.12.032. [Epub ahead of print]

A cost-effectiveness analysis of a telephone-linked care intervention for individuals with Type 2 diabetes.

Gordon LG1, Bird D2, Oldenburg B3, Friedman RH4, Russell AW5, Scuffham PA6.

Author information

• 1Centre for Applied Health Economics, Griffith Health Institute, Griffith University, University Dr, Meadowbrook, Queensland, Australia. Electronic address: Louisa.Gordon@griffith.edu.au.
• 2School of Public Health and Preventive Medicine, Monash University, Queensland, Australia; School of Nursing, Queensland University of Technology, Australia.
• 3School of Public Health and Preventive Medicine, Monash University, Queensland, Australia.
• 4Medical Information Systems Unit, Department of Medicine, Boston University School of Medicine, Boston, MA, USA.
• 5Department of Diabetes and Endocrinology, Princess Alexandra Hospital, Woollongabba, Queensland, Australia; School of Medicine, The University of Queensland, Brisbane, Queensland, Australia.
• 6Centre for Applied Health Economics, Griffith Health Institute, Griffith University, University Dr, Meadowbrook, Queensland, Australia.

Abstract

AIM:

To assess the cost-effectiveness of an automated telephone-linked care intervention, Australian TLC Diabetes, delivered over 6 months to patients with established Type 2 diabetes mellitus and high glycated haemoglobin level, compared to usual care.

METHODS:

A Markov model was designed to synthesize data from a randomized controlled trial of TLC Diabetes (n=120) and other published evidence. The 5-year model consisted of three health states related to glycaemic control: 'sub-optimal' HbA1c ≥58mmol/mol (7.5%); 'average' ≥48-57mmol/mol (6.5-7.4%) and 'optimal' <48mmol/mol (6.5%) and a fourth state 'all-cause death'. Key outcomes of the model include discounted health system costs and quality-adjusted life years (QALYS) using SF-6D utility weights. Univariate and probabilistic sensitivity analyses were undertaken.

RESULTS:

Annual medication costs for the intervention group were lower than usual care [Intervention: £1076 (95%CI: £947, £1206) versus usual care £1271 (95%CI: £1115, £1428) p=0.052]. The estimated mean cost for intervention group participants over five years, including the interventioncost, was £17,152 versus £17,835 for the usual care group. The corresponding mean QALYs were 3.381 (SD 0.40) for the intervention group and 3.377 (SD 0.41) for the usual care group. Results were sensitive to the model duration, utility values and medication costs.

CONCLUSION:

The Australian TLC Diabetes intervention was a low-cost investment for individuals with established diabetes and may result in medication cost-savings to the health system. Although QALYs were similar between groups, other benefits arising from the intervention should also be considered when determining the overall value of this strategy.