|Cost-utility analyses of intensive blood glucose and tight blood pressure control in type 2 diabetes (UKPDS 72)
|Clarke P M, Gray A M, Briggs A, Stevens R J, Matthews D R, Holman R 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 compared three independent treatment policies addressed at patients with Type 2 diabetes.
Policy 1 was conventional (mainly through diet) versus intensive blood glucose control with sulphonylureas or insulin.
Policy 2 was conventional versus intensive blood glucose control with metformin in overweight patients.
Policy 3 was less tight blood pressure control (aiming to keep blood pressure at <200/<105 mmHg) versus tight blood pressure control (aiming for <150/85 mmHg), using an angiotensin-converting enzyme (ACE) inhibitor (captopril) or beta-blocker (atenolol).
The conventional blood glucose control policy aimed to maintain a fasting plasma glucose (FPG) concentration of less than 15 mmol/L, without hyperglycaemia symptoms, while trying to achieve the lowest FPG possible with diet alone. The intensive blood glucose control policy aimed for an FPG concentration less than 6.0 mmol/L while patients were treated with short-acting as well as long-acting insulin. The goal was to reach pre-meal glucose concentrations of 4 to 7 mmol/L.
The study population comprised patients aged between 25 and 65 years with newly diagnosed Type 2 diabetes. The patients included in the study had an FPG concentration greater than 6.0 mmol/L. No further inclusion or exclusion criteria were reported in the current study. The authors referred to a separate clinical study (UKPDS Group 1991, see 'Other Publications of Related Interest' below for bibliographic details).
The setting was secondary care. The economic study was carried out in England, Scotland and Northern Ireland.
Dates to which data relate
The effectiveness data were obtained from the UKPDS Group 1991 clinical trial. It was reported that patients in the UKPDS trial were recruited from 1977 to 1991. Quantities of drugs and inpatient resources were derived during the conduct of the trial. Outpatient resources were derived from a cross-sectional survey carried out between January 1996 and September 1997. The utility outcomes were estimated in 1997, while the cost data were derived from sources published between 2002 and 2004. All costs were reported for the price year 2004.
Source of effectiveness data
The effectiveness data were derived from a single study, the United Kingdom Prospective Diabetes Study (UKPDS) (UKPDS Group 1991).
Link between effectiveness and cost data
The costing was carried out retrospectively on the same sample of patients as that used in the effectiveness study.
The authors only reported limited information for this field, referring instead to the clinical paper (UKPDS Group 1991). Initially, 5,102 patients with newly diagnosed Type 2 diabetes, who were aged between 15 and 65 years, were recruited to the study. All patients went through a dietary run-in period after which 4,209 patients were found to have an FPG concentration of 6.1 to 15 mmol/L, without any symptoms of hyperglycaemia. Of these, 3,867 were randomised to either a conventional glucose control policy (n=1,138), an intensive glucose control policy with insulin (n=1,156), or sulphonylureas (n=1,573).
A second group of 342 overweight patients (more than 120% of their ideal body weight) were allocated to an intensive glucose control policy using metformin. These patients were compared with 411 overweight patients who were allocated to the conventional glucose control policy group. The authors also reported that an embedded study was carried out in which 1,148 patients with hypertension were randomised to either a less tight (n=390) or tight (n=758) blood pressure control policy using an ACE inhibitor (captopril) or beta-blocker (atenolol).
The analysis was based on a multi-centre, prospective, randomised intervention trial. The method of randomisation and possible blinding were not reported in the current study. The authors only reported limited information for this field, referring instead to the clinical paper (UKPDS Group 1991). The duration of follow-up and losses to follow-up were not reported in the current study. The time horizon of the model equalled the patients' lifetime.
Analysis of effectiveness
It was not reported whether the analysis was conducted on an intention to treat basis. The primary outcomes were not reported. In addition, it was not reported whether the patient groups were comparable at analysis. Relevant details were given in the clinical paper (UKPDS Group 1991).
Effectiveness results were not reported in the current study. Instead, the reader is referred to a separate clinical paper (UKPDS Group 1991) for relevant details.
No clinical conclusions were provided.
The authors constructed a simulation model, the UKPDS Outcomes Model (UKPDS OM) to estimate time from end of follow-up until the first incidence of diabetes-related episodes and death in order to derive QALYs. Diabetes-related episodes included myocardial infarction (MI), angina, stroke, heart failure, amputation, renal failure and blindness in one eye. The model was characterised as a probabilistic discrete-time illness-death model. Details of the model were described elsewhere (Clarke et al. 2004, see 'Other Publications of Related Interest' below for bibliographic details). The model consisted of an integrated system of parametric equations that calculate the annual probability of any of the diabetes-related outcomes and uses Monte Carlo methods to estimate the incidence of events. The model was based on the assumption that each patient from the end of the trial had the overall mean glycated haemoglobin (HbA1c) and blood pressure levels, thus the patients' hazard was only dependent on the incidence of complications during the trial.
Measure of benefits used in the economic analysis
The author used health utility (QALYs) as measure of benefit in the economic analysis. Type 2 diabetes complications were also included and assigned a health utility value. The health utility values were estimated using the EQ-5D health status questionnaire, which was administered to 3,192 patients who were still participating in the UKPDS in 1997. When estimating utilities, the authors assumed that the incidence of numerous complications had an additive effect on utility. In addition, the same health utility values were assigned to patients in similar health states regardless of the treatment policy. All health utility values were reported. The health benefits were discounted at annual rates of 3.5% and 6%.
The direct health service costs covered treatment costs. Such costs were for metformin, chlorpropamide, glibenclamide, glipizide, Ultralente insulin (1,000-unit vial or 750-unit cartridge), soluble insulin (1,000-unit vial or 750-unit cartridge) and atenolol (25, 50 or 100 mg), as well as other drugs not explicitly listed. Implementation costs covered the clinic visit to a general practitioner, home visit or surgery visit by practice nurse, dietician and hospital eye clinic. Estimated standard practice cost by treatment covered the cost of being conventional primarily on diet or tablets, the cost of conventional insulin treatment, the cost of intensive treatment on tablets, and the cost of intensive insulin treatment. The cost of complications included the hospital costs of fatal MI, nonfatal MI, nonfatal stroke and amputation, and the non-hospital costs of macrovascular and microvascular events. Out-of-pocket expenditures incurred by patients when visiting clinics and the cost of informal care provided by family members were not included in the analysis because of the lack of available data.
The quantities of inpatient health care resources were derived directly from the UKPDS study, while non-inpatient utilisation of health care resources was derived from a cross-sectional survey conducted between January 1996 and September 1997. All costs were derived from published sources. The costs and the quantities were not reported separately, although resource use per patient was recorded. As the costs were incurred during more than 2 years, discounting was appropriately undertaken. The price year was reported.
Statistical analysis of costs
The costs were reported as mean values with standard deviations (SDs) and as mean differences (between conventional and intensive treatment groups) providing the 95% confidence intervals (CIs). No further statistical analyses were undertaken and the unit costs were treated deterministically.
The indirect costs were not included in the analysis.
UK pounds sterling () and Euros (EUR). The conversion rate was 1 = EUR 1.5.
The authors conducted various multi-way sensitivity analyses to investigate the robustness of their results to variability in the data. Anti-diabetic therapy costs, standard practice costs and the costs of complications were varied by +/- 50%. The effects of different assumptions about continuing health effects were also investigated. In particular, the authors investigated the impact on cost-effectiveness ratios when using only the benefits accomplished within the trial, and when the average difference in HbA1c (or blood pressure) between treatment policies was extended beyond the trial. Estimates of QALYs were also varied in a one-way sensitivity analysis. In particular, it was assumed that the mean utility for people without complications was 1.0 (full health), as opposed to 0.79 used in the baseline analysis. A further one-way sensitivity analysis tested the impact of using a discount rate of 6% for cost parameters, as opposed to 3.5% used in the baseline analysis.
Estimated benefits used in the economic analysis
Patients in the intensive blood glucose control policy with insulin and sulphonylureas were estimated to achieve 16.62 (SD=8.35) QALYs (undiscounted), compared with 16.35 (SD=8.36) QALYs (undiscounted) in the conventional therapy group. The difference of 0.27 (95% CI: -0.48 to 1.03) or when discounted 0.15 (95% CI: -0.20 to 0.49) was not statistically significant.
The QALYs of overweight patients assigned to an intensive blood glucose control treatment with metformin did not differ significantly from those for overweight patients in the conventional glucose treatment option.
Patients in the tight blood pressure control policy gained 14.16 (SD=7.81) QALYs, compared with 13.71 (SD=8.00) QALYs for patients in the less tight blood pressure control group. However, the difference of 0.29 (95% CI: -0.26 to 0.85) discounted was not statistically significant.
The total costs were reported per patient:
The mean total discounted costs for conventional and intensive blood glucose control with insulin or sulphonylureas were 14,984 (SD=17,888) and 15,868 (SD=14,465), respectively. The mean difference in costs was 884 (95%CI: -483 to 2,250).
The mean, total discounted costs for conventional and intensive blood glucose control with metformin in overweight patients were 16,941 (SD=23,193) and 15,920 (SD=13,678), respectively. The mean difference in costs was -1,021 (95% CI: -4.291 to 2.249), indicating cost-savings due to intensive therapy.
The mean, total discounted costs for tight and less tight blood pressure control were 15,786 (SD=16,378) and 15,895 (SD=16,025), respectively. The mean difference in costs was 108 (95% CI: -2.347 to 2.563).
Synthesis of costs and benefits
The authors used cost-effectiveness planes. In all three scenarios, the cost-effectiveness planes indicated that the 95% confidence surface expand beyond a single quadrant of the cost-effectiveness plane, making computations of 95% CIs difficult because of the inherent instability of ratio statistics. Therefore, the authors employed different values of the maximum willingness-to-pay for a QALY in order to determine the probability that the methods compared were cost-effective.
When the costs and effects were discounted, intensive blood glucose control had a 10% chance of being cost-saving. At a willingness-to-pay of 20,000 per QALY, there is a 74% chance that the strategy is cost-effective.
There is also a 77% probability that metformin is cost-saving in comparison with conventional policy. At a willingness-to-pay of 20,000 per QALY, metformin has a 98% chance of being cost-effective. On the other hand, tight blood pressure control has a 47% chance of being cost-saving when compared with the less tight control policy, and an 86% chance of being cost-effective when assuming a willingness-to-pay of 20,000 per QALY.
The results of the sensitivity analysis demonstrated that a 50% increase in standard practice costs increased the cost per QALY (discounted) from 6,028 to 10,051, while a 50% reduction in standard practice costs reduced the cost per QALY to 2,251.
The use of lower complications costs increased the cost per QALY. The same effect was applied to tight blood pressure control.
The continuing treatment effect reduced the average cost to 4,209 per QALY. Variations in the utility values did not alter the results.
The authors concluded "although the point estimates of cost-effectiveness fall within the acceptable range, we cannot be confident, at conventional error probabilities, that the modelled interventions are cost-effective".
CRD COMMENTARY - Selection of comparators
A justification was provided for the comparators used. You should decide if these represent valid health technologies in your own setting.
Validity of estimate of measure of effectiveness
The analysis was based on a prospective randomised intervention trial, which seemed appropriate given the study question. The study sample was representative of the study population. However, it was unclear whether the patient groups were comparable at analysis since details of the sample were not provided. It is not possible to comment on the internal validity of the effectiveness results since the authors referred to a separate clinical paper for details of the clinical study. No power calculations were reported, thus it was not possible to ascertain whether the results obtained were due to the intervention or due to chance.
Validity of estimate of measure of benefit
The measure of benefit used was the health utility (QALYs). Survival was derived from a probabilistic discrete-time illness-death model. This was an appropriate approach to take in view of the short-term clinical trial data and the long-term outcome of interest. QALYs were measured using the EQ-5D health status questionnaire.
Validity of estimate of costs
From the perspective of the health care purchaser adopted, it appears that most relevant categories of costs have been included in the analysis. Although some costs were omitted from the analysis, their omission is unlikely to have affected the authors' conclusions. The costs and the quantities were not reported separately, although resource use per patient was recorded. However, this would not enable the analysis to be easily reworked for other settings. The unit costs were treated deterministically, but sensitivity analyses were carried out to assess the robustness of the cost estimates used. No statistical analysis on the quantities was carried out. Discounting and the price year were appropriately reported.
The authors compared their study findings with those of published studies, reporting that any differences were due to different implementation costs. The issue of generalisability of the results to other settings was not directly addressed. The authors do not appear to have presented their results selectively. The study enrolled patients with newly diagnosed Type 2 diabetes and this was reflected in the authors' conclusions.
The authors reported a number of limitations to their study. First, only the costs of diabetic therapy were included in the within-trial period analysis. Second, when the time horizon was extended beyond the trial period, the continuing treatment effect was not taken into consideration in the baseline analysis. The authors examined each of the three interventions separately, even though clinical practice has demonstrated that the interventions can used simultaneously when treating patients. Finally, the parent clinical trial (UKPDS) was first designed many years ago. Since then, standard care for diabetes has progressed and more intense care is now being provided. Therefore, the authors feared that the incremental benefits reported in the current study might be difficult to achieve nowadays.
Implications of the study
The authors did not make explicit recommendations for changes in policy or practice. However, they called for country-specific cost-effectiveness evaluations in cases of highly differentiated health care costs. In addition, since the UKPDS trial has pointed out a linear relationship between HbA1c and risk of complications and between blood pressure and risk of complications, the authors suggested that more research is necessary to investigate the effectiveness and cost-effectiveness of evolved treatment options.
Source of funding
Funded by 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 in the National Institute of Health, USA; the British Heart Foundation; Novo-Nordisk; Bayer; Bristol Myers-Squibb; Hoechst; Lilly; Lipha; and Farmitalia Carlo Erba.
Clarke P M, Gray A M, Briggs A, Stevens R J, Matthews D R, Holman R R. Cost-utility analyses of intensive blood glucose and tight blood pressure control in type 2 diabetes (UKPDS 72) Diabetologia 2005; 48(5): 868-877
Other publications of related interest
UKPDS Group. UK Prospective Diabetes Study VIII: study design, progress and performance. Diabetologia 1991;34:877-90.
Clarke P, Gray A, Briggs A, et al. A model to estimate the lifetime health outcomes of patients with Type 2 diabetes: the United Kingdom prospective diabetes study (UKPDS) outcomes model (UKPDS 68). Diabetologia 2004;47:1747-59.
Subject indexing assigned by NLM
Adult; Aged; Antihypertensive Agents /economics /therapeutic use; Blood Glucose /metabolism; Blood Pressure /physiology; Cost-Benefit Analysis; Costs and Cost Analysis; Diabetes Complications /economics /epidemiology; Diabetes Mellitus, Type 2 /blood /economics /physiopathology; Great Britain; Humans; Hypertension /prevention & Hypoglycemic Agents /economics /therapeutic use; Middle Aged; control
Date bibliographic record published
Date abstract record published