Cost-effectiveness of intense insulin treatment after acute myocardial infarction in patients with diabetes mellitus

Transcription

1 European Heart Journal (2000) 21, doi: /euhj , available online at on Cost-effectiveness of intense insulin treatment after acute myocardial infarction in patients with diabetes mellitus Results from the DIGAMI study B. Almbrand 1, M. Johannesson 2,B.Sjöstrand 3, K. Malmberg 1 and L. Rydén 1 1 Department of Cardiology, Karolinska Hospital, Stockholm, Sweden; 2 Centre for Health Economics, Stockholm School of Economics, Stockholm, Sweden; 3 SBL Vaccine AB, Stockholm, Sweden Aims The aim of the present analysis was to estimate the cost-effectiveness of intense insulin treatment after acute myocardial infarction in patients with diabetes mellitus based on the results of the Diabetes Mellitus Insulin Glucose Infusion in Acute Myocardial Infarction (DIGAMI) study. In this study 620 patients with diabetes mellitus and acute myocardial infarction were randomized to intense insulin treatment (insulin group) or to serve as controls given standard antidiabetic therapy. Mortality was significantly reduced in the insulin group. Methods and Results The cost-effectiveness ratio was estimated as the incremental cost per life-year and qualityadjusted life-year gained of intense insulin treatment. The incremental costs were estimated as the difference in health care costs and indirect costs (labour production) during the first year of follow-up plus the future costs of increased survival. The life-years gained were based on the 5-year long-term follow-up experience and an assumed annual 20% mortality risk for all patients thereafter. The health care costs were Euro 975 higher in the insulin group during the first year of follow-up, mainly due to a longer period of initial hospitalization related to the institution of multidose insulin. The estimated discounted gain in life-years of the insulin treatment was 0 94 years without and 0 66 with quality of life adjustment, respectively. The cost per lifeyear gained by intense insulin treatment was Euro and the cost per quality-adjusted life-year gained was Euro Thus the estimated cost-effectiveness ratios were relatively low. Conclusion The results of the DIGAMI study indicate that intense insulin treatment after an acute myocardial infarction in patients with diabetes mellitus has an acceptable level of cost-effectiveness. (Eur Heart J 2000; 21: ) 2000 The European Society of Cardiology Key Words: Myocardial infarction, diabetes mellitus, survival, cost-effectiveness, insulin. See page 700 for the Editorial comment on this article Introduction Diabetes mellitus is an important risk factor for cardiovascular disease. Mortality after a myocardial infarction is considerably higher for patients with diabetes mellitus than for patients without diabetes mellitus, both in the acute phase and during long-term follow-up [1 4]. Thus it is important to develop more effective treatment strategies, particularly since diabetic patients account for up to 20% of the total number of individuals Revision submitted 12 July 1999 and accepted 22 July Correspondence: Björn Almbrand, MD, Department of Cardiology, Karolinska Hospital, S Stockholm, Sweden X/00/ $35.00/0 admitted to coronary care units for suspect myocardial infarction [1]. The DIGAMI (Diabetes Mellitus Insulin Glucose Infusion in Acute Myocardial Infarction) study tested whether improved metabolic care by means of intense insulin therapy could improve survival. In all, 620 diabetic patients were randomly assigned to intense insulin treatment (insulin group) or served as controls (control groups). The intensive insulin treatment consisted of an insulin glucose infusion for at least 24 h followed by subcutaneous multidose insulin for at least 3 months. One year mortality was significantly reduced from 26% to 19% [5]. During long-term follow-up (mean 3 4 years; range years) the mortality difference 2000 The European Society of Cardiology

2 734 B. Almbrand et al. Death rate Patients at risk Years in study Control Infusion Figure 1 Actuarial mortality curves during long-term follow-up in the patients receiving insulin glucose infusion and in the control group in the DIGAMI study (figure reprinted with kind permission from the British Medical Journal [6] ). increased further, see Fig. 1. The overall mortality was 33% in the infusion group compared to 44% in the control group [6]. Thus, the DIGAMI study demonstrated that intense insulin treatment after myocardial infarction improves survival among patients with diabetes mellitus. Before a treatment strategy is recommended for substantial patient populations it is important to verify that it is cost-effective, i.e. provides good value for money compared to other health improving measures [7,8]. The aim of this analysis is to estimate the cost-effectiveness of intense insulin treatment after acute myocardial infarction in patients with diabetes mellitus based on the results of the DIGAMI study. Methods Control Infusion RR = 0 72 [ ] P = Absolute reduction 11% The cost-effectiveness was estimated as the incremental cost per life-year gained and quality-adjusted lifeyear gained of intense insulin treatment compared to standard therapy without intense insulin treatment as derived from the DIGAMI control group [7,8]. The analysis was carried out from a societal perspective including health care costs as well as indirect costs (decreased labour production) [8]. These costs were estimated in detail for the first year of follow-up. Since no data were available as regards resource consumption for subsequent years it was assumed that the cost per patient year was the same for both groups thereafter. It has recently been argued that the difference between total consumption and total production during an increase in length of life should be added to the costs in cost-effectiveness analysis [9 11]. We therefore included these future costs. According to current practice [12] both costs and effectiveness were discounted at 3%. All costs were estimated in 1999 Swedish Crowns, but the results are presented in 1999 prices in Euro, based on the exchange rate in June 1999 (Euro 1=SEK 8 70). Costs for antidiabetic therapy The cost of therapy for the diabetic condition was divided into the cost for the insulin glucose infusion in the insulin group and the cost for oral antidiabetics and subcutaneous insulin in both groups. Use of oral therapy and subcutaneous insulin was recorded for each patient in the trial as a basis for calculating the treatment costs. The official Swedish retail prices for insulin were used [13]. Since no detailed data were available as regards insulin consumption after the first year of follow-up, it was assumed that the cost of insulin did not differ between the two groups during the remaining period of follow-up. Costs for the initial hospitalization The costs for the initial hospitalization were divided into costs for hospital days, costs for thrombolytic therapy and costs for pacemakers. Data on the quantities of these resources were collected in the trial. The cost of thrombolysis was based on the official Swedish retail price of streptokinase, and a cost of Euro 115 was used [13]. The cost per hospital day of Euro 521 and the cost of a pacemaker of Euro 4116 was based on cost estimations at the participating centres in the DIGAMI study. Health care costs from discharge to 12 months The number of hospital days, the number of PTCA and CABG procedures, and the number of outpatient physician visits were recorded in the trial from discharge to 12 months of follow-up. The unit costs for these resources were based on cost estimations at the participating centres in the DIGAMI study. The following unit costs were used: Euro 211 per hospital day, Euro 5207 per PTCA, Euro per CABG, and Euro 157 per outpatient physician visit. Labour productivity during the first year of follow-up The working status of each patient was recorded at the 3 and 12 month follow-up visits. The working status was estimated as the percentage of full time work for each patient. The labour production was estimated for the period from discharge until 3 months and the period from 3 months to 12 months. The average working status during the first of these periods was assumed to be an average of the working status at discharge (assumed to be 0 for all patients) and 3 months. The average working status during the second period was assumed to be an average of the working status at 3 and 12 months. To estimate the labour production the working status

3 The DIGAMI study 735 was multiplied by the average value of labour production of a Swedish worker (Euro per year) [14]. Future costs during the increased length of life How to deal with costs that arise due to an increase in the length of life (i.e. costs during the added years) is one of the oldest controversies in this field [7 12]. The solution to the controversy about costs in added years is that the difference between total consumption and total production in added years should be included as a cost in cost-effectiveness analysis [8 11]. We therefore include these full future costs of added years in the analysis. The total consumption can be divided into the nonhealth care consumption and the health care consumption. As annual non-health care consumption we used the average in Sweden for the population above 65 years of age (Euro ) [10]. For the annual health care consumption we used the average health care costs in the intense insulin group and the control group during the first year excluding the costs of the initial hospitalization and the cost of the insulin glucose infusion. This was divided by the survival time during the first year to yield an annual health care consumption of Euro The total annual consumption per added year was thus assumed to be Euro To estimate the annual production in added years we used the average working status of the insulin and control group patients alive at the 12 months visit (14% of full working capacity). Multiplying this by the annual production of a full time worker yielded an annual production of Euro Accordingly we used an annual consumption minus production of Euro ( ) as the future cost per added year. Life-years gained The number of life-years gained was defined as the increase in life-expectancy as a result of intense insulin treatment compared to the treatment in the control group. Data on long-term survival in the trial population has been reported in detail elsewhere [5,6]. In brief the patients were followed-up up to 5 6 years (mean 3 4 years). The actuarial mortality curves in the two groups from randomization to 5 years of follow-up are shown in Fig. 1. The life-years gained during years 0 5 were estimated as the area between the survival curves. To estimate the total number of life-years gained it is also necessary to know the survival after the 5th year. In the base-case analysis we assumed a constant annual morality risk of 20% for the survivors at 5 years. This yields a life-expectancy of 5 years for a person alive at 5 years [15 16]. Furthermore it was conservatively assumed that the annual mortality risk after the 5th year would be equal for patients in the insulin and the control groups. We also varied the mortality risk after the fifth year in a sensitivity analysis (see below). Quality-adjustment of life-years gained To overcome the disadvantage of using only life-years gained as the effectiveness measure, paying no attention to the health status during the gained life-expectancy, quality-adjusted life-years gained were also used as an effectiveness measure. Quality-adjusted life-years were constructed by weighting years of life with a weight between 0 (dead) and 1 (full health) corresponding to the health status in the year. Since no specific information on quality of life was collected in the DIGAMI trial, it was assumed that quality of life did not differ between the two treatment groups. A recent study estimated the average quality weight in the Swedish general population to be about 0 80 in the age-group [17]. According to other studies a myocardial infarction reduces the quality weight by about 0 10 [18 19]. A quality weight of 0 70 ( ) was therefore used in the present study to adjust life-years for quality of life. Since the 0 70 quality weight is an approximation it was varied in a sensitivity analysis (see below). Sensitivity analysis Very few studies have included the full future costs of added life-years and for comparability the costeffectiveness was therefore estimated without these costs. Since some studies only include health care costs we estimated the cost-effectiveness with only such costs included. This estimation was performed both with and without health care costs of added years of life. One estimation was included in which the labour production after the first year of follow-up was assumed to be zero (due to the high mean age of this patient population). This increases the annual consumption minus production in added years from Euro to Euro From Figure 1 it is evident that the number of patients at risk during long-term follow-up decreases over time, increasing the uncertainty of the survival curves. Cost-effectiveness was therefore calculated with a similar annual mortality in both groups after 3 or 4 rather than 5 years. Another uncertain variable is the annual mortality after 5 years of follow-up. This was varied between 10% and 100% in a sensitivity analysis. The quality of life weight was varied between 0 8 and 0 6 and finally, we varied the discount rate for both cost and effectiveness between 0% and 5%, and in one analysis costs were discounted at 3% but effectiveness not at all. Statistical analysis Differences between the groups in quantity of resources consumed and costs during the first year of follow-up were tested for significance. For continuous variables mean values were compared by two-tailed independent samples t-test [20]. The independent samples t-test is robust for non-normality if the hypothesis of equal

4 736 B. Almbrand et al. Table 1 Use of medical resources during the initial hospitalization and up to one year after discharge for the two treatment groups, standard deviations within brackets Item Infusion group Control group P-value Initial hospitalization Hospital days per patient (11 5) 8 88 (9 1) Thrombolytic therapy (%) Pacemaker (%) From discharge to 12 months Hospital days per patient (25 6) (23 9) Number of PTCA (%) Number of CABG (%) Outpatient physician visits per patient 6 07 (4 8) 5 02 (4 5) Table 2 Mean health care costs per patient during the first year of follow-up for the two treatment groups, standard deviations within brackets, 1999 prices in Euro Cost Infusion group Control group Difference P-value Insulin Insulin-glucose infusion Oral antidiabetics 15 (39) 38 (66) 23 <0 001 Subcutaneous insulin 216 (182) 138 (205) 78 <0 001 Total 236 (177) 176 (197) 60 <0 001 Initial hospitalization Hospital days 5342 (5994) 4625 (4728) Thrombolytic therapy 54 (57) 55 (57) Pacemaker 161 (800) 79 (564) Total 5557 (6043) 4759 (4779) From discharge to 12 months Hospital days 2321 (5391) 2338 (5039) PTCA 221 (1133) 249 (1188) CABG 1121 (3180) 1125 (3369) Outpatient physician visits 955 (757) 789 (705) Total 4618 (6836) 4501 (6972) Total during first year (9709) 9436 (8741) variances cannot be rejected [20]. The variance only differed significantly between samples for one variable (the cost of oral antidiabetics) and in that case the non-parametric Mann Whitney test was used for comparisons. For discrete variables a contingency table chi-square test was used [20].AP-value below 0 05 was considered statistically significant. Results Health care consumption during the first year of follow-up The use of medical resources during the initial hospitalization and from hospital discharge to 12 months is reported in Table 1. Although not statistically significant the number of hospital days during the initial hospitalization was 1 4 more in the infusion than in the control group (P=0 098). The number of outpatient visits was higher in the infusion group (P=0 005). It should be noted that the proportion of patients undergoing PTCA and CABG was rather low in both treatment groups. As long as the number of PTCA and CABG does not differ between the treatment groups a higher proportion of patients undergoing PTCA and CABG would, however, not affect the cost difference between the groups. The health care costs during the first year of follow-up is presented in Table 2. The overall cost of insulin was Euro 60 higher in the infusion group (P<0 001). The overall costs during the initial hospitalization were Euro 798 higher in the infusion group (P=0 068). Apart from

5 The DIGAMI study 737 Table 3 Mean working status and labour production per patient during the first year of follow-up for the two treatment groups, standard deviations within brackets, 1999 prices in Euro Infusion group Control group Difference P-value Working status among all patients % of full time work at 3 months 8 01 (24 8) 8 28 (26 1) % of full time work at 12 months (29 2) (39 3) Working status among survivors % of full time work at 3 months 9 11 (26 3) 9 81 (28 2) % of full time work at 12 months (31 9) (33 3) Labour production (Euro) Labour production 0 3 months 273 (849) 284 (901) Labour production 3 12 months 2110 (5556) 2074 (5657) Labour production 0 12 months 2383 (6339) 2358 (6483) the difference in costs for outpatient visits, cost differences were small and insignificant from discharge to 12 months. The average health care costs during the first year of follow-up were Euro per patient in the infusion group and Euro 9436 in the control group, i.e. a difference of Euro 975 (P=0 188). Dividing the net cost by the discounted gain in lifeexpectancy yielded a cost per life-year gained of Euro from the intensive insulin treatment. Dividing the net cost by the discounted gain in quality-adjusted life-years yielded a cost per quality-adjusted life-year gained of Euro Labour production during the first year of follow-up There was no significant differences in labour production between the two groups (Table 3). The total labour production during the first year of follow-up was Euro 2383 in the infusion group and Euro 2358 in the control group, i.e. a difference of Euro 25 (P=0 961). Life-years gained The estimated life-expectancy measured from randomization was 6 41 years in the infusion group and 5 26 years in the control group, i.e. a difference of 1 15 years. Of this increase in life-expectancy 0 43 years was the difference in the survival curves for years 0 5 in Fig. 1, and the rest was due to more patients being alive in the infusion group at the end of 5 years. If life-years were discounted at 3% the gain in life-years decreased to If this gain in life-years was adjusted for quality of life, the gain decreased further to 0 66 quality-adjusted life-years. Cost-effectiveness The net incremental cost per patient in the infusion group was estimated as the increased health care costs of Euro 975 during the first year of follow-up (from Table 2), minus the increased labour production of Euro 25 (from Table 3), plus the future costs due to increased survival of Euro This yielded a net incremental cost of Euro per patient in the infusion group. Sensitivity analysis The result of the sensitivity analysis is shown in Table 4. If the future costs of increased life-expectancy were excluded the cost-effectiveness ratio would decrease substantially to Euro 1000 per life-year gained and Euro 1500 per quality-adjusted life-year gained. If only health care costs were included (including health care costs in added years) the cost per life-year gained would be Euro 6600 and the cost per quality-adjusted life-year gained Euro If only the health care costs in the first year of follow-up were included the results would be similar to the example without future costs. If the estimation of future costs was based on the labour production being none after the first year of follow-up the costeffectiveness ratio would increase slightly compared to the baseline estimate. If it was assumed that the mortality rate was the same for both groups after 3 or 4 years of follow-up rather than 5 years, the estimated gain in discounted life-years would be decrease from 0 94 years to 0 68 years and 0 69 years. This, however, would have little effect on the cost-effectiveness ratio. Since the major part of the net cost is the future cost of added life-years this cost would decrease when the gain in life-years decreased. For the same reason the variation in the annual mortality risk after the 5 years of follow-up had little effect on the cost-effectiveness ratio, although it caused a variation in the discounted gain in life-years between 0 40 years (100% annual mortality risk after five years) and 1 35 years (10% annual mortality risk after 5 years). Varying the quality of life weight between 0 6 and 0 8 led to a variation of the cost per quality-adjusted life-year gained between Euro and Euro

6 738 B. Almbrand et al. Table 4 Sensitivity analysis of the cost per life-year and quality-adjusted life-year gained, 1999 prices in Euro Cost per life-year gained Cost per quality-adjusted life-year gained Baseline estimate Costs Future costs excluded Only health care costs included Only health care costs in year 1 included Labour production 0 after year Long-term mortality differences The same mortality rate after 3 years The same mortality rate after 4 years Annual mortality rate after 5-years 10% % Quality of life weight Discount rate Costs 0%, effectiveness 0% Costs 5%, effectiveness 5% Costs 3%, effectiveness 0% =not applicable. Varying the discount rate between 0% and 5% had little effect on the result. This is because the discounting reduced both the gain in life-years and the future costs of added life-years. If the effectiveness was not discounted at all while costs were discounted by 3% the cost per life-year gained would decrease to Euro and the cost per quality-adjusted life-year gained decreased to Euro Discussion Previous reports from the DIGAMI study revealed that intense insulin treatment improves survival during longterm follow-up [5 6]. It can be estimated that lifeexpectancy will increase by 1 15 years in the infusion group (0 94 years with 3% discounting). Since the gained life-years cannot be expected to be in full health they were adjusted for quality of life. The estimated gain in quality-adjusted life-years was 0 80 (0 66 with 3% discounting). An increased life-expectancy will be associated with a cost for society that equals the difference between total consumption and total production during the added years of life. It has recently been advocated that such future costs should be included in costeffectiveness analysis [9]. A failure to include these costs will, for instance, discriminate against programmes that improve quality of life of elderly patients [9 11]. We therefore included future costs in the analysis. Adding the future costs to the costs of the first year of follow-up (Euro 950) led to an incremental cost per patient in the infusion group of Euro To enable comparisons with previous studies that have not included future costs we also estimated the cost-effectiveness ratios without future costs. The incremental cost was divided by the discounted increase in life-expectancy and quality-adjusted lifeexpectancy, which yielded a cost per life-year gained of Euro and a cost per quality-adjusted life-year gained of Euro To be able to interpret whether the estimated cost-effectiveness ratios are high or low we need to know how much society is willing to pay to gain a life year or quality-adjusted life-year. The only sector in Sweden where this price has been explicitly stated is in cost benefit analyses of road investments, where a price of Euro 1 5 million per life saved is used [21]. The Euro 1 5 million per life saved used in economic evaluations of road investments implies a price of about Euro per life-year gained and about Euro per quality-adjusted life-year gained (at a 3% discount rate). To make these figures useful as a benchmark for the health care field they should be adjusted for differences in the treatment of taxes in economic evaluations of road investments and health care in Sweden. With these adjustments the price decreases to about Euro per life-year gained and about Euro per qualityadjusted-life-year gained. Using the same value of increased survival as in road investments in Sweden, intense insulin treatment is highly cost-effective, since the estimated cost-effectiveness ratios are far below this benchmark. It is also common to compare the costeffectiveness ratios of alternative uses of health care resources to implicitly decide what society is willing to pay per life-year or quality-adjusted life-year gained. In

7 The DIGAMI study 739 most studies, treatments with cost-effectiveness ratios below Euro are considered to be highly costeffective, while those above Euro are considered not to be cost-effective [22 28]. Although the cost per quality-adjusted life-year gained is slightly above the Euro benchmark it is well below the Euro , suggesting that the treatment is costeffective. A problem in comparing the cost effectiveness ratio in this study with other studies is that most other studies did not include the full future costs. If the future costs are not included the cost per life-year gained decreases to Euro 1000 and the cost per quality-adjusted life-year gained decreases to Euro The only Swedish study that has included the full future costs is a costeffectiveness analysis of treatment of mild hypertension [10]. In that study the cost per quality-adjusted life-year gained was of the same magnitude as in DIGAMI for middle-aged and older men and women with mild hypertension. For younger men and women with mild hypertension the cost per quality-adjusted life-year gained was considerably higher than in DIGAMI. It was considered that the treatment was cost-effective among middle-aged and older men and women with mild hypertension [10]. Accordingly it may be concluded that the results of the economic analysis of the DIGAMI study indicates that intense insulin treatment after acute myocardial infarction in patients with diabetes mellitus is cost-effective. References [1] Malmberg K, Rydén L. Myocardial infarction in patients with diabetes mellitus. Eur Heart J 1988; 9: [2] Stone P, Muller J, Hartwell T et al. The effect of diabetes mellitus on prognosis and serial left ventricular function after acute myocardial infarction: contribution of both coronary disease and diastolic left ventricular dysfunction to the adverse prognosis. J Am Coll Cardiol 1989; 14: [3] Karlson BW, Herlitz J, Hjalmarson A. Prognosis of acute myocardial infarction in diabetic and non-diabetic patients. Diabet Med 1993; 10: [4] Barbash GI, White HD, Modan M, Van dwerff F. Significance of diabetes mellitus in patients with acute myocardial infarction receiving thrombolytic therapy. J Am Coll Cardiol 1993; 22: [5] Malmberg K, Rydén L,EfendicSet al. A randomized trial of insulin-glucose infusion followed by subcutaneous insulin treatment in diabetic patients with acute myocardial infarction (DIGAMI study): effects on mortality at 1 year. J Am Coll Cardiol 1995; 26: [6] Malmberg K for the DIGAMI (Diabetes Mellitus Insulin Glucose Infusion in Acute Myocardial Infarction) Study Group. Prospective randomised study of intensive insulin treatment on long term survival after acute myocardial infarction in patients with diabetes mellitus. Br Med J 1997; 314: [7] Drummond MF, O Brien B, Stoddart GL, Torrance GW. Methods for the economic evaluation of health care programmes, second edition. Oxford: Oxford Medical Publications, [8] Johannesson M. Theory and methods of economic evaluation of health care. Dordrecht: Kluwer Academic Publishers, [9] Meltzer D. Accounting for future costs in medical costeffectiveness analysis. J Health Econ 1997; 16: [10] Johannesson M, Meltzer D, O Conor RM. Incorporating future costs in medical cost-effectiveness analysis: implications for the cost-effectiveness of the treatment of hypertension. Med Decis Making 1997; 17: [11] Johannesson M, Meltzer D. Some reflections on costeffectiveness analysis. Health Economics 1998; 7: 1 7. [12] Weinstein MC, Siegel JE, Gold MR, Kamlet MS, Russel LB for the Panel on Cost-Effectiveness in Health and Medicine. Recommendations of the Panel on Cost-Effectiveness in Health and Medicine. JAMA 1996; 276: [13] Läkemedelsinformation AB. FASS 1998, Pharmaceutical specialities in Sweden. Stockholm: Läkemedelsinformation AB, [14] Statistics Sweden. Statistical yearbook of Sweden. Stockholm: Statistics Sweden, [15] Beck JR, Kassirer JP, Pauker SG. A convenient approximation of life expectancy (the DEALE ): I. Validation of the method. Am J Med 1982; 73: [16] Beck JR, Kassirer JP, Pauker SG. A convenient approximation of life expectancy (the DEALE ): II. Use in medical decision-making. Am J Med 1982; 73: [17] Lundberg L, Johannesson M, Isacson DGL, Borgquist L. Health-state utilities in a general population in relation to age, gender, and socioeconomic factors. Eur J Public Health 1999; 9: [18] Tsevat J, Goldman L, Soukup JR, Lamas GA, Connors KF, Chapin CC, Lee TH. Stability of time-tradeoff utilities in survivors of myocardial infarction. Med Decis Making 1993; 13: [19] Glasziou PP, Bromwich S, Simes JR for the AUS-TASK group. Quality of life six months after myocardial infarction treated with thrombolytic therapy. Med J Australia 1994; 161: [20] Newbold P. Statistics for business and economics. 3rd edn. Englewood Cliffs, New Jersey: Prentice-Hall, [21] SAMPLAN. O} versyn av samhällsekonomiska kalkylvärden för den nationella trafikplaneringen Stockholm. SAMPLAN, [22] Laupacis A, Feeny D, Detsky AS, Tugwell PX. How attractive does a new technology have to be to warrant adoption and utilization? Tentative guidelines for using clinical and economic evaluations. Can Med Assoc J 1992; 146: [23] Martens LL, Rutten FF, Erkelens DW, Ascoop CA. Cost-effectiveness of cholesterol-lowering therapy in the Netherlands. Am J Med 1989; 87 (Suppl. 4A): 54S 58S. [24] Tosteson AN, Rosenthal DI, Melton LJ, Weinstein MC. Cost-effectiveness of screening perimenopausal white women for osteoporosis: bone densitometry and hormone replacement therapy. Ann Int Med 1990; 113: [25] Goldman L, Weinstein MC, Goldman PA, Williams LW. Cost-effectiveness of HMG-CoA reductase inhibition for primary and secondary prevention of coronary heart disease. JAMA 1991; 265: [26] van Hout B, Bonsel G, Habbema D, Vandermaas P, Decharro F. Heart transplantation in the Netherlands: costs, effects and scenarios. J Health Econ 1993; 12: [27] Johannesson M, Borgquist L, Jönsson B, Lindholm LH. The cost effectiveness of lipid lowering in Swedish primary health care. J Intern Med 1996; 240: [28] Johannesson M, Jönsson B, Kjekshus J, Olsson AG, Pedersen TR, Wedel H. Cost-effectiveness of simvastatin treatment to lower cholesterol levels in patients with coronary heart disease. N Engl J Med 1997; 336: