|Cost-effectiveness analysis of intensive blood-glucose control with metformin in overweight patients with Type II diabetes (UKPDS No 51)
|Clarke P, Gray A, Adler A, Stevens R, Raikou M, Cull C, Stratton I, Holman R
This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.
The study investigated the use of metformin (500 to 2,550 mg/day as required) as an agent for intensive blood-glucose control in overweight patients with Type II (non-insulin dependent) diabetes.
Economic study type
The study sample was drawn from a population comprising overweight patients (120% above the ideal bodyweight; body mass index, BMI, approximately 25.6 kg/m2) aged between 25 and 65 years, who were newly diagnosed with Type II (non-insulin dependent) diabetes-mellitus and had a fasting plasma glucose greater than 6 mmol/L on two occasions. The patients recruited to the study had a mean age of 53 (standard deviation, SD=9) years and a mean BMI of 31.7 (SD=4.9) kg/m2.
The setting was secondary care. The economic study was carried out in the UK.
Dates to which data relate
The dates when the effectiveness evidence was collected were unclear, although patients were recruited to the study between 1977 and 1991. Some resource use data were collected between 1996 and 1997. The price year was 1997.
Source of effectiveness data
The effectiveness data were derived from a single study.
Link between effectiveness and cost data
The costing was carried out prospectively on the same sample as that used in the effectiveness analysis.
No power calculations to determine the sample size were reported. A total of 5,102 patients were recruited into the study. Following this (from the first 15 centres), 753 patients started the trial. Of these 342 overweight patients were allocated to receive the glucose control policy with metformin and 411 overweight patients were allocated to receive the conventional dietary-based treatment.
The study was based on a multi-centred, randomised, clinical controlled trial. The median duration of follow-up was 10.7 years. The method of participant allocation or blinding was not reported.
Analysis of effectiveness
The primary health outcome in the analysis was the gain in life expectancy derived from within-trial effects, measured using a simulation model. The analysis was conducted on an intention to treat basis. It was unclear whether the groups were comparable at analysis in terms of demographics and prognostic features. However, the authors referred to another study (see Other Publications of Related Interest) from which this information might be derived.
From the model, the mean life expectancy was 21.3 years (95% confidence interval, CI: 20.5 - 22.0) in the conventional treatment group and 22.3 years (95% CI: 21.6 - 23.0) in the metformin group, from the date of allocation to the groups.
The authors concluded that intensive blood-glucose control that involved metformin was more effective (in terms of life expectancy gains) than conventional dietary-based treatment.
A simulation model was used to estimate the time from end of follow-up to death. This was considered appropriate given the chronic nature of the disease.
Measure of benefits used in the economic analysis
The authors' intention was to carry out an incremental analysis to calculate the cost per life-year saved. The benefit measure was derived from the clinical study as reported in the effectiveness results.
The health care (purchaser) costs included those relating to treatment, visits to a nurse or general practitioner (GP), and the costs of treating diabetic complications. The resource use data for medication were derived from routine data. Hospital admissions and length of stay were recorded and coded from routine clinic visits. The costs of unusually expensive hospital stays were reported separately, to avoid bias in the results. The costs were derived from a variety of sources, mainly UK national statistics and from the participating trial centres. Patient survey data was the source for non-inpatient health care resource use (including those relating to contact with GPs, nurses, podiatrists, opticians, dieticians and other hospital clinics over the preceding 4 months). The unit costs (based on published estimates) were assigned to each of these service items and the total annual cost calculated. A Tobit model was used to explore relationships between resource use and patient characteristics.
The costs and the resource quantities were reported separately. Discounting was applied at both 3% and 6% to reflect differences in US and UK guidelines, respectively. Adjustments were made for the additional costs incurred in relation to protocol-driven deviation from normal clinical practice.
Statistical analysis of costs
The data were presented as descriptive statistics (mean, SDs and 95% CIs).
In line with the perspective chosen, the indirect costs were not included.
Uncertainty in the results was investigated by looking at both the variability in the data and the generalisability of the results. In terms of non-hospital resource use, a Poisson-based model was used to repeat the analysis of numbers of visits in each non-inpatient category. For costs, the effects of alternative visit patterns (other than standard practice) to health professionals were considered in the sensitivity analysis. Following a simulation model exercise for the primary outcome (life expectancy), a non-parametric bootstrap process was used to estimate uncertainty. In this process, metformin and conventional treatment patients were sampled with replacement from the study cohort.
Estimated benefits used in the economic analysis
The mean difference between the treatment groups was 1.0 year (95% CI: 0.0 - 2.1).
When a 6% discount rate was applied, the difference was 0.4 years (95% CI: 0.0 - 0.8).
When a 3% discount rate was applied, the difference was 0.6 years (95% CI: 0.0 - 1.2).
The (undiscounted) mean cost per patient was 8,165 (SD=10,904) in the conventional treatment group and 7,883 (SD=7,086) in the metformin group. The mean cost-difference was 281 (95% CI: -1,578 - 1,015).
When discounted at 6%, the mean cost was 5,893 (SD=7,989) in the conventional treatment group and 5,635 (SD=4,582) in the metformin group. The mean cost-difference was 258 (95% CI: -1,171 - 655).
When discounted at 3%, the mean cost was 6,878 (SD=9,215) in the conventional group and 6,607 (SD=5,613) in the metformin group. The mean cost-difference was 271 (95% CI: -1,345 - 801).
Although the cost of complications was included in the analysis, the potential adverse effects relating to the use of metformin were not specified.
Synthesis of costs and benefits
Given that the metformin group was associated with cost-savings and higher effectiveness (the dominant strategy), the authors provided a cost-effectiveness acceptability curve. The purpose of this was to assist clinical decision-making by illustrating the relationship between a ceiling cost-effectiveness ratio and the probability of cost-effectiveness below that chosen ratio. Using a 6% discount rate, the results showed a 95% probability that the cost-effectiveness of intensive blood glucose therapy incorporating metformin was less than 1,600 per life-year gained and a 71% probability that metformin would produce cost-savings in comparison with conventional treatment. The sensitivity analysis revealed that even when the cost of metformin in the UK was doubled, and the number of health professional visits increased in line with US levels (using a Poisson model), metformin would remain cost-effective.
Intensive glucose control incorporating metformin is both cost-saving and more effective in overweight patients with Type II diabetes.
CRD COMMENTARY - Selection of comparators
The choice of the comparators was justified on the basis that metformin had been the favoured treatment option in comparison with conventional dietary-based intervention in a recent prospective diabetes study (see Other Publications of Related Interest). You should decide if this represents a widely used technology in your own setting.
Validity of estimate of measure of effectiveness
The analysis was based upon a randomised, clinical controlled trial, which was appropriate given the study question. The limited details of the study sample revealed that mean age and BMI were higher than that recorded in the study population, thus potentially limiting the generalisability of the results to that population. Although the intention to treat analysis was a strong point, the internal validity of the study could not be fully ascertained by the information given. Power calculations were not presented, nor were details of the method of randomisation, blinding, loss to follow-up and comparability of the groups at analysis. However, these might be available in the preceding UK Prospective Diabetes Study (see Other Publications of Related Interest). Appropriate statistical analyses were undertaken to account for potential biases and confounding factors.
Validity of estimate of benefit:
Although the authors intended to present a summary measure of benefit (cost per life-year saved), the dominance of the metformin strategy dictated that a cost-effectiveness acceptability curve would be more useful to decision-makers in clinical practice.
Validity of estimate of costs
A thorough exploration and breakdown of the direct costs relevant to the UK health care perspective were included. In addition, the authors appropriately considered the costs of complications arising from the target condition, although potential adverse effects of the treatment drug were not specifically reported, in terms of either cost or resource use. The emphasis on standard practice assessment (e.g. removing protocol-driven costs and resource use from the analysis) will contribute positively to the generalisability of the results. The costs and the quantities were reported separately, thus enhancing the reproducibility of the study in other settings. Resource quantities obtained within the trial (and by comparison with a standard practice setting) were appropriately tested in a sensitivity analysis, as were costs (taken from a variety of published national statistics). Discounting was necessary given the chronic nature of the disease, and this was usefully calculated at rates appropriate to different countries. The price year was reported, which will aid any future reflation exercises.
The authors compared their findings with those from other studies which were in agreement, in terms of evidence of effectiveness. The authors acknowledged that some analysis on the side effects of the treatment drug would have been useful for assessing compliance rates.
Implications of the study
The authors considered metformin to be a favourable primary pharmacological therapy for overweight individuals with Type II diabetes. The extent to which these study findings are generalisable to the specified study population, in terms of patient characteristics and severity of condition, will require careful consideration.
Source of funding
Supported by grants from the UK Medical Research Council, the British Diabetic Association, the UK Department of Health, the National Eye Institute and the National Institute of Digestive, Diabetes and Kidney Disease (NIH, USA), the British Heart Foundation, Novo-Nordisk, Bayer, Bristol Myers Squibb, Hoechst, Lilly, Lipha and Farmitalia Carlo Erba.
Clarke P, Gray A, Adler A, Stevens R, Raikou M, Cull C, Stratton I, Holman R. Cost-effectiveness analysis of intensive blood-glucose control with metformin in overweight patients with Type II diabetes (UKPDS No 51) Diabetologia 2001; 44(3): 298-304
Other publications of related interest
UKPDS Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet 1998;352:837-53.
Subject indexing assigned by NLM
Adult; Aged; Antihypertensive Agents /economics /therapeutic use; Blood Glucose /analysis; Blood Glucose Self-Monitoring /economics /statistics & Body Mass Index; Cost-Benefit Analysis; Costs and Cost Analysis; Cross-Sectional Studies; Diabetes Mellitus /blood /drug therapy /economics; Diabetes Mellitus, Type 2 /blood /drug therapy /economics; Diet, Diabetic; England; Great Britain; Humans; Hypertension /complications /drug therapy /economics; Hypoglycemic Agents /economics /therapeutic use; Insulin /economics /therapeutic use; Metformin /economics /therapeutic use; Middle Aged; Northern Ireland; Obesity; Patient Education as Topic; Scotland; numerical data
Date bibliographic record published
Date abstract record published