Unambiguous results from randomized controlled trials show that treatment with statins for patients with coronary artery disease (CAD) are both clinically effective (2;18) and leads to potential savings of health services (13). The higher the CAD event risk, the stronger is the evidence that lipid-lowering therapy is cost-effective (16). Despite this finding, there are still a not-negligible gap between what is achieved in clinical practice and what should be achieved according to scientific evidence-based goals for secondary prevention in patients with CAD, even for patients in high-risk groups (1). The reason for this gap is probably that there are several barriers and complexities between evidence-based guidelines and anticipated behavior change in real clinical practice (5;9). Keys to successful cost-effective implementation of new evidence could be educational methods focusing on how and when to use the content of guidelines in the local context and content of the target physicians, and facilitation of the process of local consensus building (8). However, educational intervention studies with patient-related end points are scarce. Studies measuring the cost-effectiveness of educational/implementation methods are even more scant. Salkeld et al. state that the cost-effectiveness of general practice-based lifestyle interventions is yet to be demonstrated (19).
We have shown that Case Method Learning (CML) for general practitioners (GP) resulted in a significant reduction of lipid levels in patients with coronary artery disease (CAD) to a degree that—according to current knowledge—should decrease mortality and morbidity in CAD (2;14). This study is an economic evaluation of the costs and effects of this intervention.
METHODS
Patients
The patient registry of Stockholm County Council allows identification of all inpatients and outpatients visiting the hospital clinics. The Department of Medicine at Södertälje Hospital, located in the southern most part of Stockholm County, Sweden, provides planned and emergency health care for acute and elective cardiac patients in a catchment area of approximately 95,000 habitants. Thus, we could identify all these patients, during the preceding year, with a diagnosis of CAD (ICD-9 code 410-414). We identified 429 patients [les ] 70 years of age with such a diagnosis and scrutinized their medical records.
Criteria for a confirmed diagnosis of CAD in the patient record were as follows: (i) A diagnosis of angina pectoris, either by objective criteria based on coronary angiography, or pathologic findings on exercise test or stress test, or a clinical assessment based on typical angina symptoms at exercise with or without electrocardiographic (ECG) evidence of possible or definite ischemia. (ii) A diagnosis of myocardial infarction based on either World Health Organization criteria (21) or on unequivocal ECG findings.
A total of 323 patients fulfilled the inclusion criteria. We had identified, thus, all the patients in the population with a recent need for specialist care. We invited all of these patients to participate in the main study (January 1995); 68 patients refused to participate, leaving 255 patients to be included in the main study (79 percent of the identified unselected population). Hypertension was considered to be present if systolic blood pressure was >140 mm Hg, if diastolic blood pressure was >90 mm Hg, or if antihypertensive medication were used. Diabetes mellitus was defined as fasting plasma glucose of [ges ]7.0 mmol/L, a self-reported history of diabetes mellitus, or treatment for diabetes.
The patients were randomly assigned into two groups according to which group the physician responsible for their care belonged to. The intervention group (I) for a GP participating in the CML seminars and the control group (C) for a GP working at a primary health-care center only receiving the local practice guidelines. A specialist group (S) representing patients treated by a specialist in cardiology or internal medicine served as an external control group. A total of 220 patients completed the 2-year study.
Intervention
New evidence-based guidelines were presented to all participating physicians. The physicians in the intervention group also participated, together with a locally well-known facilitator (cardiologist), in recurrent interactive CML seminars at their own primary health-care center. Details of the intervention in the randomized controlled educational study have been published previously (14).
Costs
Data were gathered prospectively during the 2-year study period. Direct costs of health-care resources (utilization of lipid-lowering drugs) were measured. Resources attributable to cholesterol-lowering drugs were recorded, and changes from baseline were calculated. Lipid-lowering drugs were recorded both as defined daily doses (DDD) and as actual cost. Cost of the educational intervention was based on time consumption in hours and corresponding salary cost for attendance of the GPs, and preparation, travel, and seminar time and corresponding salary cost of the facilitator. Mean salaries for GPs and the mean salary of hospital-based consultants—for the facilitator—in Stockholm County Council were used in the cost analysis. Payroll taxes were included. All costs are calculated on the basis of Swedish prices in 2002 and converted to US$ at the 2002 exchange rate (1 US$=9.5 SEK [Swedish Crowns]).
Quality of life
EuroQol 5D Index (EQ-5D) is a quality of life instrument well-established for use as weight in calculation of quality-adjusted life years (QALYs). The respondents may classify their present health status in five dimensions—mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each question can be answered in three levels—no problems, moderate problems, and severe problems. The UK EQ-5D index tariff was used to obtain single index values between 0 (dead) and 1 (full health) for the QALY calculations (6). EQ-5D is an easy and rapid generic instrument, well-validated and found to be reliable in different cultures and in different diseases (3;4;17).
Ethical Considerations
All participating patients gave written informed consent. The study complies with the Declaration of Helsinki and was approved by the ethics committee of Karolinska Institute at Huddinge University Hospital.
RESULTS
Baseline characteristics of the primary health-care centers and general practitioners are shown in Table 1. Baseline characteristics of the patients are shown in Table 2. Patients in the intervention group had their LDL cholesterol reduced by 0.5 (confidence interval [CI], 0.1–0.9) mmol/L compared with controls (p<.05; Figure 1). No change did actually occur in the control group. LDL cholesterol in the specialist group—the external control group—decreased by 0.6 (CI, 0.4–0.8) mmol/L. Consumption of DDD of lipid-lowering drugs increased by 0.2 (CI, 0.04–0.3) in the primary care intervention group and 0.3 (CI, 0.2–0.5) in the specialist group (p<.03; Figure 2) but did not change in the primary care control group 0.0 [CI, −0.1–0.1].
Decrease in low-density lipoprotein cholesterol (mmol/L) at 2 years compared with baseline in the intervention (I), control (C), and specialist (S) groups. Means and 95 percent confidence intervals.
Increase in defined daily doses (DDD) of statins at 2 years as compared with baseline in the intervention (I), control (C), and specialist (S) groups. Means and 95 percent confidence intervals.
Costs
The costs of CML—seminar time of GPs and travel, planning, and seminar time of the facilitator—in the intervention group were US$ 3.3 per treated patient during the 2 years of the study (Table 3). Costs of lipid-lowering drugs were US$ 106 during the second study year in the intervention group (Table 4) as opposed to US$ 52 in the control group. This finding means that the increase in costs of lipid-lowering drugs was US$ 47 higher per patient and year in the primary care intervention group than in the primary care control group. Total costs for lipid-lowering drugs and CML in the intervention group were US$ 154 during the 2 years and US$ 87 in the control group. As a comparison, the costs of lipid lowering were US$ 307 in the external specialist control group. The additional costs of the educational intervention did only represent 2 percent of the drug cost (Tables 3 and 4).
Quality of Life
Quality of Life according to EQ-5D was 0.80 (CI, 0.75–0.85) in the intervention group at baseline and did not change during the study (Table 4). In the control group, EQ-5D was 0.79 (CI, 0.71–0.86) at baseline and 0.76 (CI, 0.71–0.86) at 2 years (not significant). The EQ-5D in the specialist group at baseline was slightly lower than for patients treated in primary care (p=.048). In the specialist group, EQ-5D increased (p=.03) from 0.72 (CI, 0.69–0.76) to 0.76 (CI, 0.73–0.80). There were no significant differences between the groups at the end of the study.
DISCUSSION
We have shown that the use of CML improves the quality of care of the patients at only a marginal additional cost of 2 percent of the drug cost. It can be argued that the effect of the CML strategy is that more patients suitable for lipid lowering actually start and continue treatment. The extra cost for achieving this end is the cost of the educational program that amounts to US$ 3.3 per patient. What is shown in earlier studies is that secondary prevention is cost-effective (12;13). Only adding a small investment cost for the educational program should not change this conclusion.
To expand a bit further on this, the cost-effectiveness of the CML strategy compared with the control could be viewed in terms of the 4S-study (2). In the 4S-study, the patients LDL cholesterol was reduced by 1.7 compared with control. The average follow-up time in that study was 5.5 years. The increase in life years was estimated at 0.377, whereas the increase in the number of discounted life years (5 percent discount rate) amounted to 0.240 (13).
LDL cholesterol in the present study was reduced at a fraction of 0.29 (0.5/1.7) compared with the 4S-study. A total of 0.5 mmol/L may sound a rather marginal lipid lowering compared with the 1.7 mmol/L in the 4S-study. There are, however, several aspects strengthening the argument that the lipid lowering in the present study is of clinical relevance. The 4S-study was a randomized controlled trial with the intention to give all patients in the active group statin treatment. In the present study, the intention was to give all physicians in the active group education with the purpose to increase statin treatment in their CAD patients. The mean lipid lowering is calculated in the whole group of patients and not only in the subgroup who actually got treatment (the proportion of patients treated with statin increased by 20 percent in the intervention group). This finding means that the patients—who actually had statin treatment instituted during the study—had at least the same lipid lowering as shown to be of clinical significance in the 4S-study. Furthermore, it is known that the clinical benefit from statin treatment is related to baseline risk rather than to actual lipid levels (20). The patients in the present study all had a high baseline risk.
We can assume that the increase in life years in this study is equal to the increase in life years found in the 4S-study multiplied by the fraction of 0.29. Thus, the increase in the number of life years comparing CML with the control for a period of 5.5 years is 0.11 (0.29*0.377), whereas the increase in the number of discounted life years is 0.07 (0.29*0.24). Adjusting the life years with the quality of life during these years (0.8) results in an increase in the number of QALYs of 0.0875 (0.29*0.377*0.8), whereas the increase in the number of discounted QALYs is 0.0557 (0.29*0.24*0.8).
The increase in costs during the study period is equal to US$ 3.3 (education)+US$ (59.2–12.7), which is the annual increase in costs for lipid-lowering drugs during the 2 year study period. The total cost increase during the same follow-up period as in the 4S-study would be equal to 3.3+5.5*(59.2–12.7), which amounts to US$ 256. The corresponding discounted cost increase is US$ 233. From a societal perspective, also costs in added years of life should be included (15). The costs in added years of life are defined as the difference between annual production and consumption in different age groups. Assuming an extra annual cost in added years of life for patients in this study of US$ 16,000 will result in an extra costs per patient of US$ 1,120 (0.07*16,000), because consumption exceeds production in gained life years (7). Thus, the total extra discounted costs of CML strategy compared with the control is equal to US$ 1,353.
The resulting discounted cost per gained QALY is then equal to US$ 24,300=1,353/0.0557, which is below what is generally accepted as the value of a gained QALY. The value of a gained QALY is usually stated to be approximately US$ 60,000 (10;11). Thus, it is indicated that the CML-supported lipid lowering is a cost-effective strategy in secondary prevention of CAD in primary care.
CONCLUSIONS
We have shown that CML for physicians in primary care is a cost-effective educational method to implement new evidence in the local context and content of the target physicians. CML resulted in a lipid lowering at a degree that, according to the results in the 4S-trial (2), should decrease mortality and morbidity (14).
Policy Implications
The main finding in the present study is an increase in the quality of care—at only a marginal increase in the cost—by use of the well-established educational method CML. Our findings should probably warrant the use of CML to decrease quality gaps in the care of other major chronic diseases.
CONTACT INFORMATION
Anna Kiessling, MD, PhD (anna.kiessling@kids.ki.se), Karolinska Institute at Danderyd University Hospital, Karolinska Institute; Consultant and Head, Centre for Clinical Education, KKC, Danderyd University Hospital, SE 182 88 Stockholm, Sweden
Niklas Zethraeus, PhD (henz@hhs.se), Assistant Professor, Centre for Health Economics, Stockholm School of Economics, Sveavägen 65, 113 83 Stockholm, Sweden
Peter Henriksson, MD, PhD (pethen@ki.se), Professor, Karolinska Institute at Danderyd University Hospital, Karolinska Institute; Consultant, Department of Internal Medicine, Danderyd University Hospital, SE 182 88 Stockholm, Sweden
The study was supported by grants from Stockholm County Council and Karolinska Institutet. Assistant Professor Magnus Johannesson Stockholm School of Economics is gratefully acknowledged for invaluable advice concerning the planning of the study. We also thank Katarina Patzelt Brandt for skillful management of interviews and careful handling of case report forms. A.K., N.Z., and P.H. have no conflicts of interest to declare.
