|Economic assessment of troglitazone as an adjunct to sulfonylurea therapy in the treatment of Type 2 diabetes
|Caro J J, Klittich W S, Raggio G, Kavanagh P L, O'Brien J A, Shomphe L A, Flegel K M, Copley-Merriman C, Sigler C
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 use of the new chemical entity, troglitazone, in combination with sulfonylurea therapy. Troglitazone was administered at daily doses of 200 (T200), 400 (T400) or 600 mg (T600). Sulfonylurea therapy comprised 12 mg glyburide (G12).
Economic study type
Cost-effectiveness analysis and cost-utility analysis.
The study population included a hypothetical cohort of Type 2 diabetic patients who were poorly controlled by maximum doses of sulfonylurea (fasting glucose greater than 7.8 mmol/L or 140 mg/dL). The inclusion criteria specified a glycosylated haemoglobin (HbA1c) level outside the normal range (greater than 5.9%), fasting serum glucose (FSG) greater than 7.8 and less than 16.7 mmol/dL, and fasting C-peptide of at least 0.5 nmol/L. The exclusion criteria were chronic use of insulin, history of ketoacidosis, symptomatic diabetic neuropathy, retinopathy, liver disease, renal disease, anaemia, cardiovascular abnormalities (e.g. class III or IV congestive heart failure), significant hypertension, cancer, or a major vascular medical event within the last 3 months.
The setting was not explicitly stated, but it is presumed to have been primary care. The clinical trial was conducted in the USA.
Dates to which data relate
The dates for the effectiveness evidence were not given, but some data came from studies published between 1986 and 1998. The resources and costs were from 1996, which was the price year.
Source of effectiveness data
The effectiveness evidence was derived from a single study and from completed studies. The single study (Horton et al., see Other Publications of Related Interest) provided a measure of the drug effectiveness. The completed studies were used to introduce complication and mortality rates linked to glycaemic control into an existing epidemiological model.
Link between effectiveness and cost data
The costing was undertaken retrospectively on a sample of patients that was different from those included in the effectiveness study.
No power calculations to determine the sample size were reported. The study sample comprised a group of 552 patients who had poorly controlled Type 2 diabetes and were assigned to one of seven groups, according to the dose of troglitazone. The four groups considered in the present economic evaluation were placebo+G12 (n=79), T200+G12 (n=78), T400+G12 (n=76) and T600+G12 (n=82). The method of sample selection was not described. It was unclear whether some patients refused to participate, or were excluded from the initial study sample. The groups receiving placebo+G12 and combined therapy (T200+G12, T400+G12, T600+G12) comprised approximately 50% females, were aged between 25 and 74 years, were of five racial types, and had an average time since diagnosis of 8 years.
The original trial was described as a randomised, placebo-controlled trial of 52 weeks' duration. Reference to the original article shows that it was carried out at 30 centres in the USA and it was described as double-blind. Of the 552 patients, 318 (58%) completed all 52 weeks of the study, 57% of the G12 group and 71 to 85% of the combination group. Further follow-up has been undertaken at two years.
Analysis of effectiveness
The primary health outcomes used in the economic model were changes in HbA1c and FSG. The baseline comparability of the study groups was not discussed, although the groups were fairly similar in their demographic and clinical characteristics. The analysis of the original clinical trial was conducted on an intention to treat basis.
The change in HbA1c was 0.90 (+/- 0.20) in the placebo+G12 group, -0.70 (+/- 0.20) in the T200+G12 group, -0.91 (+/- 0.20) in the T400+G12 group, and -1.75 (+/- 0.20) in the T600+G12 group.
The change in FSG was 1.3 (+/- 0.4) in the placebo+G12 group, -1.7 (+/- 0.4) in the T200+G12 group, -2.1 (+/- 0.4) in the T400+G12 group, and -3.1 (+/- 0.4) in the T600+G12 group.
The effectiveness analysis suggested that troglitazone added to glyburide was effective in reducing HbA1c and FSG in Type 2 diabetes patients with poor glycaemic control, compared with glyburide alone.
An economic model, based on the troglitazone clinical trials and epidemiological data, was developed by modifying an existing model (Eastman et al., see Other Publications of Related Interest). The original model simulated the health status of Type 2 diabetes patients from diagnosis to death. Using a state transition model, estimates of the effect of adding troglitazone to glyburide on the rates and timing of complications associated with Type 2 diabetes were extrapolated long-term from the clinical trial outcomes. The original model was slightly modified to fit the objective of the present economic evaluation. Two cohorts of 10,000 patients were considered.
Outcomes assessed in the review
Some data used to extend the time horizon of the analysis were estimated from the literature. These were the rates and timing of complications associated with Type 2 diabetes and quality of life adjustments.
Study designs and other criteria for inclusion in the review
A formal review of the literature was not undertaken. The designs of the primary studies were not reported.
Sources searched to identify primary studies
Criteria used to ensure the validity of primary studies
Methods used to judge relevance and validity, and for extracting data
Number of primary studies included
The effectiveness data were derived from 7 primary studies.
Methods of combining primary studies
Investigation of differences between primary studies
Results of the review
Data on complications associated with Type 2 diabetes were not reported. The quality of life values were 0.69 for blindness, 0.59 for end-stage renal disease and 0.80 for leg amputation.
Measure of benefits used in the economic analysis
The summary benefit measure used in the economic analysis was the life-years gained (LYG) in the cost-effectiveness analysis, and the quality-adjusted life-years (QALYs) in the cost-utility analysis. A cost-consequences analysis (CCA) was also conducted. The utility weights used to calculate the QALYs were obtained from the original decision model. The benefits were discounted at an annual rate of 3%. The undiscounted results were also presented.
The quantities and the costs were not reported separately. The direct costs reported were from the perspective of a payer with comprehensive responsibility for health care. The categories of costs included in the economic evaluation were treatment of complications, treatment of diabetes and laboratory tests. The treatment costs were taken from published sources. The drug costs were taken from the 1996 edition of Drugs Topics Redbook. The physician costs were derived from Medicare fee schedules, while frequency of physician contact was based on the American Diabetes Association management guide. The cost of monitoring for liver failure (a side effect of troglitazone with an estimated incidence of 1 in 57,000) and the cost of transplantation and death were included. The unit costs for all tests were from Medicare laboratory fees schedule 1996. A detailed description of the costing of complications is reported elsewhere (O'Brien et al., see Other Publications of Related Interest). In brief, the estimation included the direct cost of treating an event acutely and subsequent treatment. The costing of acute events were averages and came from combining all-payer inpatient databases from 5 states, whereas long-term treatment came from a variety of sources. The costs that accrued after the first year were discounted at 3%. No justification was given for this rate, but it was explored in the sensitivity analysis. The price year was 1996.
Statistical analysis of costs
There was no statistical analysis of the costs.
The indirect costs were not included.
The areas of uncertainty investigated were generalisability of the results and extrapolation from primary data. A one-way sensitivity analysis was carried out. The parameters investigated were population characteristics (age, gender, race), discount rate (3, 0, 5, 10), time horizon (95, 10, 15 years), degree of HbA1c drift, duration of diabetes mellitus since diagnosis, and the cost of monitoring hepatic functioning. Population characteristics involved an analysis of extremes (homogeneous population).
Estimated benefits used in the economic analysis
In relation to the CCA, per thousand patients, the addition of T600 was predicted to result in 95 fewer cases of blindness, 128 fewer cases of end-stage renal disease, and 140 fewer amputations than glyburide alone. The lifetime undiscounted LYG increased from 14.9 with 12G to 16.9 with T600+12G. Similarly, the lifetime undiscounted QALYs increased from 12.6 to 15.0.
At 3% discount, the costs were $73,500 ($14,000 for treatment and $59,500 for complications) for the G12 group and $77,700 ($42,000 for treatment and $35,700 for complications) for the T600+G12 group.
Synthesis of costs and benefits
The costs and benefits were synthesised by calculating the cost per LYG and the cost per QALY gained. Incremental cost-effectiveness ratios were calculated. The value of the undiscounted cost/LYG ratio was $2,100 and the value of the cost/QALY gained was $1,750. The discounted equivalents were $3,158/LYG and $3,066/QALY.
The sensitivity analysis showed that the cost/LYG varied according to the sub-population. It was $14,123/LYG among 65- to 75-year-olds, $3,526/LYG among females, $1,272 /LYG among males and $4,107/LYG among whites, and there were substantial cost-savings amongst all-Hispanic populations. In the model, when the commencement of troglitazone therapy was delayed, the gain in LYG was reduced by 1% for each year of postponement.
If the degree to which glycaemic control deteriorates overtime was assumed not to drift upward for either treatment, then the cost-effectiveness would increase to $5,405 per undiscounted LYG. If the reversal of drift observed with troglitazone continued, the cost-effectiveness ratio would increase to $629/LYG. When the analysis was repeated with a time horizon of 15 years, the cost was $11,870/undiscounted LYG or $16,047/discounted LYG. At 10 years the cost was less than $50,000/LYG.
In the sensitivity analysis, liver functioning monitoring after the first year was doubled and the cost-effectiveness ratio increased by 10%
While the long-term implications have not been clinically established in clinical trials, the results from the model suggest substantial health benefits and reasonable cost-effectiveness with troglitazone. The authors concluded that adding troglitazone to glyburide in Type 2 diabetic patients failing sulfonylurea therapy significantly improved glycaemic control and, in turn, reduced the complications of the disease and overall treatment costs.
CRD COMMENTARY - Selection of comparators
The choice of the comparator was not explicitly justified, but it appears to have represented current drug therapy in the authors' setting. Other treatment strategies, such as diet and exercise modification, were not explored. You should decide whether the comparator represents current practice in your own setting.
Validity of estimate of measure of effectiveness
The analysis modelled effectiveness data from a published randomised controlled trial, which was appropriate for the study question. Double-blinding was also applied and the analysis of the clinical study was conducted on an intention to treat basis. These issues ensure the high internal validity of the analysis. However, it is unclear whether the study sample was representative of the study population since strict inclusion criteria were used. Power calculations were not performed and the methods of sample selection and randomisation were not described. Some data were also derived from the literature, but details on the method and conduct of the review were not reported. The authors acknowledged that three assumptions (glycaemic control reduces complications and increases survival, and troglitazone reduces HbA1C drift) were used, although these were supported in the literature. However, actual epidemiological data were not reported. In general, it was difficult to estimate the validity of the overall effectiveness analysis, as few details of the analysis were provided.
Validity of estimate of measure of benefit
The estimation of benefits was modelled. The quality weights were derived from other studies, but details of the values used were not reported. The choice of QALYs and LYG appears to have been appropriate and simplifies the comparison with the benefits of other health care technologies. Appropriate discounting was performed. The benefit was altered by the length over which the effect was experienced.
Validity of estimate of costs
All the categories of cost relevant to the perspective adopted appear to have been included in the analysis. However, the inclusion of the indirect costs would have been interesting. The costs and the quantities were not reported separately. The authors stated that their cost findings were conservative since they did not include the costs where data were scarce, for example, additional effects such as developing a secondary infection. Statistical tests were not performed and sensitivity analyses were not carried out on the cost/resource estimates. Thus, the cost estimates were specific to the study setting. The price year was reported, as were the sources of the cost data. Discounting was relevant and was appropriately applied.
The authors did not perform extensive comparisons of their findings with those from other studies. The issue of generalisability to patient groups was implicitly addressed in the sensitivity analysis. The authors discussed and justified the choice of some estimates used in the decision model.
Implications of the study
Pending long-term clinical findings, extrapolation by modelling existing data suggests that troglitazone used in conjunction with glyburide yields substantial health benefits with reasonable cost-effectiveness.
Source of funding
Supported by Parke-Davis, Ann Arbor (MI), USA.
Caro J J, Klittich W S, Raggio G, Kavanagh P L, O'Brien J A, Shomphe L A, Flegel K M, Copley-Merriman C, Sigler C. Economic assessment of troglitazone as an adjunct to sulfonylurea therapy in the treatment of Type 2 diabetes. Clinical Therapeutics 2000; 22(1): 116-127
Other publications of related interest
Eastman R, Javitt J, Herman W, et al. Model of complications of NIDDM.I. Model construction and assumptions. Diabetes Care 1997;20:725-34.
Horton ES, Whitehouse F, Ghazzi NM, et al for the Troglitazone Study Group. Troglitazone in combination with sulfonylurea restores glycemic control in patients with type 2 diabetes. Diabetes Care 1998;12:1462-9.
O'Brien JA, Shomphe LA, Kavanagh PL, et al. Direct medical costs of complications resulting from type 2 diabetes in the US. Diabetes Care 1998;21:1122-8.
Subject indexing assigned by NLM
Adult; Aged; Blood Glucose /metabolism; Chromans /adverse effects /economics /therapeutic use; Cohort Studies; Cost-Benefit Analysis; Diabetes Mellitus, Type 2 /complications /drug therapy /economics; Drug Costs; Female; Humans; Hypoglycemic Agents /adverse effects /economics /therapeutic use; Male; Middle Aged; Models, Economic; Randomized Controlled Trials as Topic; Sulfonylurea Compounds /adverse effects /economics /therapeutic use; Thiazoles /adverse effects /economics /therapeutic use; Thiazolidinediones; United States
Date bibliographic record published
Date abstract record published