Bernie J. O'Brien

Willingness to Pay: A Valid and Reliable Measure of Health State Preference?

The development of methods to measure willingness to pay (WTP) has renewed interest in cost-benefit analysis (CBA) for the economic evaluation of health care programs. The authors studied the construct validity and test-retest reliability of WTP as a measure of health state preferences in a survey of 102 persons (mean age 62 years; 54% male) who had chronic lung disease (forced expiratory volume <70%). Interview measurements in cluded self-reported symptoms, the oxygen-cost diagram for dyspnea, Short-Form 36 for general health status, rating scale and standard gamble for value and utility of current health state relative to death and healthy lung functioning, and WTP for a hypothetical intervention offering a 99% chance of healthy lung functioning and a 1% chance of death. WTP was elicited by a simple bidding game. To test for starting-point bias, the respondents were randomly assigned to one of five starting bids. All health status and preference measurements except WTP (controlling for income) showed significant (p < 0.05) differences between disease-severity groups (mild/moderate/severe). WTP was significantly (p = 0.01) associ ated with household income, but other health status and preference measures were not. The measure most highly correlated with WTP was standard gamble (r = -0.46). There was no association between starting bid and mean WTP adjusted for income and health status. The test-retest reliability of WTP was acceptable (r = 0.66) but lower than that for the standard gamble (r = 0.82). It is concluded that: 1) large variation in WTP responses may compromise this measures discriminant validity; 2) there is some evidence of convergent validity for WTP with preferences measured by standard gamble; 3) there was no evidence of starting point bias; 4) the test-retest reliability of WTP is comparable to those of other preference measures. Key words: willingness to pay; health state preferences; economics. (Med Decis Making 1994;14:289-297)

In Search of Power and Significance: Issues in the Design and Analysis of Stochastic Cost-effectiveness Studies in Health Care

Application of techniques such as cost-effectiveness analysis (CEA) is growing rapidly in health care. There are two general approaches to analysis: deterministic models based upon assumptions and secondary analysis of retrospective data, and prospective stochastic analyses in which the design of a clinical experiment such as randomised controlled trial is adapted to collect patient-specific data on costs and effects. An important methodological difference between these two approaches is in the quantification and analysis of uncertainty. Whereas the traditional CEA model utilizes sensitivity analysis, the mean-variance data on costs and effects from a prospective trial presents the opportunity to analyze cost-effectiveness using conventional inferential statistical methods. In this study we explored some of the implications of moving economic appraisal away from deterministic models and toward the experimental paradigm. Our specific focus was on the feasibility and desirability of constructing statistical tests of economic hypotheses and estimation of cost-effectiveness ratios with associated 95% confidence intervals. We show how relevant variances can be estimated for this task and discuss the implications for the design and analysis of prospective economic studies.

Probabilistic Analysis of Cost-Effectiveness Models: Choosing between Treatment Strategies for Gastroesophageal Reflux Disease:

When choosing between mutually exclusive treatment options, it is common to construct a cost-effectiveness frontier on the cost-effectiveness plane that represents efficient points from among the treatment choices. Treatment options internal to the frontier are considered inefficient and are excluded either by strict dominance or by appealing to the principle of extended dominance. However, when uncertainty is considered, options excluded under the baseline analysis may form part of the cost-effectiveness frontier. By adopting a Bayesian approach, where distributions for model parameters are specified, uncertainty in the decision concerning which treatment option should be implemented is addressed directly. The approach is illustrated using an example from a recently published cost-effectiveness analysis of different possible treatment strategies for gastroesophageal reflux disease. It is argued that probabilistic analyses should be encouraged because they have potential to quantify the strength of evidence in favor of particular treatment choices.

When Do the "Dollars" Make Sense? Toward a Conceptual Framework for Contingent Valuation Studies in Health Care

There is growing interest in the application of cost-benefit analysis (CBA) as a tech nique for the economic evaluation of health care programs. A distinguishing feature of CBA is that costs and benefits are expressed in the same units of value—typically money. A popular method for estimating money values for health care programs is the use of willingness-to-pay (or accept) survey techniques known as contingent valuation. This paper presents a conceptual framework to help in the interpretation or design of contingent valuation studies in health care. To be consistent with the theory upon which CBA is built, the authors consider what types of questions should be asked of what populations. They conclude that studies undertaking contingent valuation should dis tinguish between compensating variation and equivalent variation, and recognize that respondents can be gainers or losers in utility and therefore should be asked willing ness-to-pay (or accept) questions as appropriate. Current critical-appraisal guidance in the health care literature for CBA is poor and unlikely to offer useful demarcation between good and bad CBA studies. More work is needed exploring whether recently issued guidelines for contingent valuation in environmental damage assessment are applicable to health care studies. Key words: cost-benefit analysis; contingent valu ation ; willingness to pay. (Med Decis Making 1996;16:288-299)

Analysis of uncertainty in health care cost-effectiveness studies: an introduction to statistical issues and methods:

Cost-effectiveness analysis is now an integral part of health technology assessment and addresses the question of whether a new treatment or other health care program offers good value for money. In this paper we introduce the basic framework for decision making with cost-effectiveness data and then review recent developments in statistical methods for analysis of uncertainty when cost-effectiveness estimates are based on observed data from a clinical trial. Although much research has focused on methods for calculating confidence intervals for cost-effectiveness ratios using bootstrapping or Fieller’s method, these calculations can be problematic with a ratio-based statistic where numerator and=or denominator can be zero. We advocate plotting the joint density of cost and effect differences, together with cumulative density plots known as cost-effectiveness acceptability curves (CEACs) to summarize the overall value-for-money of interventions. We also outline the net-benefit formulation of the cost-effectiveness problem and show that it has particular advantages over the standard incremental cost-effectiveness ratio formulation.

Confidence intervals for cost-effectiveness ratios: an application of Fieller's theorem.

Application of cost-effectiveness analysis (CEA) is growing rapidly in health care. Two general approaches to analysis are differentiated by the type of data available: (i) deterministic models based upon secondary analysis of retrospective data from one or more trials and other sources; and (ii) stochastic analyses in which the design of a randomized controlled trial is adapted to collect prospectively patient-specific data on costs and effectiveness. An important methodological difference between these two approaches is in how uncertainty is handled. Deterministic CEA models typically rely upon sensitivity analysis to determine the robustness of findings to alternative assumptions, whereas stochastic (CEA) analysis, as part of prospective studies, permits the use of conventional statistical methods on the cost and effectiveness data for both inference (hypothesis testing) and estimation. This paper presents a procedure for the statistical analysis of cost-effectiveness data, with specific application to those studies for which effectiveness is measured as a binary outcome. Specifically, Fiellers Theorem was used to calculate confidence intervals for ratios of the two random variables of between-treatment differences in observed costs and effectiveness, i.e. the incremental cost-effectiveness ratio.

Efficacy and Cost of Low-Molecular-Weight Heparin Compared with Standard Heparin for the Prevention of Deep Vein Thrombosis after Total Hip Arthroplasty

Convincing evidence exists that heparin prophylaxis reduces the rates of deep vein thrombosis, nonfatal pulmonary embolism, and fatal pulmonary embolism after major surgery [1, 2]. Perioperative heparin prophylaxis has been recommended [3, 4] and is used widely because it is effective, safe, easy to administer, and cost-effective. More recently, several preparations of low-molecular-weight heparin have been developed and approved for clinical use in Europe. Low-molecular-weight heparin is derived from standard heparin by chemical or enzymatic depolymerization and is approximately one third of the molecular weight of standard heparin [5]. Clinical trials in Europe and North America suggest that low-molecular-weight heparin is more effective than standard heparin at preventing deep vein thrombosis after orthopedic surgery without causing increased bleeding complications [6]. In many European countries, low-molecular-weight heparin has become the prophylactic method of choice for patients having major orthopedic procedures. Several preparations of low-molecular-weight heparin have been submitted for approval and are presently under review by regulatory agencies in North America. In addition to overall effectiveness, a critical factor that will probably influence the choice between low-molecular-weight heparin and standard heparin as a primary prophylactic modality will be the relative cost-effectiveness of these two agents. In Europe, low- molecular-weight heparin is more expensive than standard heparin, but its ultimate price in North America is unknown. The cost-effectiveness of low-molecular-weight heparin will depend not only on the relative prices of low-molecular-weight heparin and standard heparin but also on the costs of treating patients developing thrombotic and bleeding complications related to these therapies. To help determine whether low-molecular-weight heparin or standard heparin would be the more appropriate agent for the prevention of deep vein thrombosis after total hip arthroplasty, we compared both the efficacy and the cost-effectiveness of these two agents. To determine valid estimates of the rates of thrombotic and bleeding complications associated with the use of low-molecular-weight heparin and standard heparin after total hip arthroplasty, we did a meta-analysis of the randomized trials directly comparing these two agents. The costs of treating postoperative thrombotic and bleeding complications were estimated using actual patient data from a recently completed clinical trial at our center comparing low-molecular-weight heparin with standard heparin for the prevention of deep vein thrombosis after total hip arthroplasty [7]. Because the North American price of low-molecular-weight heparin is unknown, we estimated its price based on the price ratio between low-molecular-weight heparin and standard heparin in France. Then, in our sensitivity analysis, we compared the cost-effectiveness of these agents over a range of price ratios. Our approach of comparing the efficacy and cost-effectiveness of a novel drug with a standard treatment could be useful for determining the appropriate role of other innovative but often expensive new agents. This type of analysis is especially pertinent if it is done before the price of the drug is established and it is released onto the marketplace. Methods Meta-analysis Articles were obtained from a search of MEDLINE using the key words venous thrombosis, hip arthroplasty, and heparin. All relevant studies were reviewed and the bibliographies were searched for additional articles. Abstracts from recent meetings were also reviewed. When necessary, authors were contacted and requested to provide additional information about study methods. Published randomized, controlled trials directly comparing a low-molecular-weight heparin preparation with standard heparin for the prevention of deep vein thrombosis after hip arthroplasty were eligible for inclusion in the meta-analysis. Potentially eligible studies were excluded if the diagnostic end point of deep vein thrombosis was not made by mandatory venography interpreted by blinded observers or if the study did not use the manufacturers currently recommended dose of low- molecular-weight heparin or standard heparin for the prevention of deep vein thrombosis after hip arthroplasty. Two investigators independently assessed all retrieved articles for eligibility and extracted the results. Disagreements were settled by consensus with the aid of a third independent reviewer. The principal outcome measure of efficacy in the analysis was total deep vein thrombosis. Deep vein thrombosis was subdivided into proximal (involving the popliteal or more proximal leg veins) and distal (isolated to the deep veins of the calf). Evaluable patients in the efficacy analysis included only those successfully completing venography before hospital discharge. The principal safety outcome was bleeding, which was subdivided into major and minor bleeding as was defined by each study. All patients entered in each trial were included in the bleeding analysis. Secondary outcomes of the meta-analysis were symptomatic pulmonary embolism, diagnosed by either high-probability lung scan or pulmonary angiography, and death. We made pulmonary embolism a secondary outcome because its incidence was expected to be low (<1%) and because studies did not distinguish between patients developing isolated pulmonary embolism or pulmonary embolism in combination with deep vein thrombosis. The relative odds of low-molecular-weight heparin versus standard heparin was determined for each of the outcome measures in each trial. Under the assumption of a fixed-effect model, the Breslow-Day test was used to test heterogeneity of outcomes among studies [8]. To combine study results, a common odds ratio was estimated for each outcome using the method of Mantel and Haenszel [9]. A common odds ratio of less than 1.0 favored low-molecular-weight heparin and was statistically significant if the upper bound of the 95% CI was also less than 1.0 [10]. Although the best estimate of relative effectiveness is through odds, we converted odds into rates for use in our economic analysis. This conversion was achieved by first calculating typical event rates in the standard heparin group as weighted averages of observed event rates. The absolute event rates that were expected with low-molecular-weight heparin treatment were then calculated by converting the standard heparin rates to odds, multiplying the odds by the appropriate pooled odds ratio, and then converting the odds back into rates. Cost and Cost-Effectiveness Analyses Two distinct but related analyses were done, both from the perspective of a third-party payer responsible for the reimbursement of all hospital costs. First, we compared the expected costs of managing 1000 patients with either low-molecular-weight heparin or standard heparin after total hip arthroplasty. The cost estimates included the prices of and the administration costs of prophylaxis and the expected costs of managing deep vein thrombosis and bleeding events on either therapy. The expected costs of deep vein thrombosis (proximal or distal) and bleeding (major or minor) were the product of our estimates of the cost of each event and the corresponding event rates from the meta-analysis. Next, a cost-effectiveness analysis was done that quantified the incremental cost of low-molecular-weight heparin for each additional deep vein thrombosis its use averted compared with standard heparin. Costs that were common to both therapy groups or unrelated to the use of deep vein thrombosis prophylaxis were excluded from both analyses. Cost of Prophylaxis The price of low-molecular-weight heparin in North America has not yet been determined. To estimate the cost for the purposes of this analysis, the parent drug companys (Rhone-Poulenc Rorer Inc.; Paris, France) recommended price for the low-molecular-weight heparin, Enoxaparin, in France was obtained and compared to a manufacturers recommended French price of standard heparin (Calciparine, Sanofi-Choay Laboratories; Paris, France). Using this ratio, the price of low-molecular-weight heparin was determined based on the Henderson Hospitals (Hamilton, Ontario) formulary price for standard heparin. It was assumed that all patients received twice daily subcutaneous injections of prophylaxis (30 mg of enoxaparin and 7500 units of standard heparin, respectively) over a 14-day period. Costs of Deep Vein Thrombosis and Bleeding Complications The hospital resource consequences of bleeding and deep vein thrombosis events were estimated by doing a retrospective analysis of patient-specific data from the hospital charts of all 447 patients who were managed in Hamilton, Ontario as part of a randomized trial [7]. (This was one of the studies included in the meta-analysis.) All patients in this study were screened for deep vein thrombosis by venography before hospital discharge, and deep vein thrombosis and bleeding complications were managed at the discretion of the individual practitioners. The direct resource consequences of bleeding events were tabulated from the hospital charts of patients who were judged to have developed bleeding complications during this trial. Costs attributed to bleeding included investigations, consultations, and treatment that directly resulted from the bleeding episode. In cases where bleeding may have resulted in reoperation, rehospitalization, or transfer to the intensive care unit, the cases were reviewed by two experienced clinicians who were blinded to the treatment allocation of the patients. They decided whether the complications and their economic consequences were the result of the bleeding event. To determine whether deep vein thrombosis (proximal or distal) or bleeding [major or minor] events were associated with prolongations in hospitalization we did a multiple regression analysis in which the

Economic Evaluation of Donepezil for the Treatment of Alzheimer's Disease in Canada

BACKGROUND: Donepezil is a new drug recently approved in the United States and Canada for the treatment of Alzheimers disease (AD). We estimated the cost‐effectiveness of donepezil 5 mg daily as an adjunct to usual care in the management of persons with mild‐to‐moderate AD defined as a Mini‐Mental Health State Examination (MMSE) score in the range 10 to 26.

Cost-effectiveness of rhythm versus rate control in atrial fibrillation.

Context Randomized trials show that rate control and rhythm control are similarly effective in the treatment of atrial fibrillation; therefore, economic issues will play a large role in the choice of therapy. Contribution This cost-effectiveness model shows that rate control saves costs compared with rhythm control. Implications From an economic perspective, unless specific clinical factors suggest a benefit of rhythm control for a particular patient, rate control seems to be the preferred strategy for the management of atrial fibrillation. Atrial fibrillation is the most common sustained type of cardiac arrhythmia treated by physicians. Its prevalence increases with advancing age, affecting approximately 5% of those 65 years of age and older and 10% of those older than 80 years of age (1-3). As the U.S. population ages, it is expected that more than 5 million persons will be living with atrial fibrillation by the year 2050 (4). Despite significant advances in the effectiveness of treatments for atrial fibrillation and its associated comorbid conditions, disability and mortality from atrial fibrillation remain high (5-10). The optimal approach to the rhythm management of atrial fibrillation remains unclear. There are 2 main approaches: Rhythm control uses electrical cardioversion, antiarrhythmic drugs, and, sometimes, nonpharmacologic therapies (for example, multisite atrial pacing, maze procedures, or radiofrequency ablation procedures) to maintain sinus rhythm; rate control uses atrioventricular nodal blocking agents (and, if needed, ablation of the atrioventricular junction and pacemaker implantation) for ventricular rate control. Recently, several randomized, controlled studies have compared rate control versus rhythm control. In the largest of these studies, investigators in the Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) randomly assigned 4060 patients with atrial fibrillation (mean age, 70 years) to either rate control or rhythm control (10-12). After a mean follow-up of 3.5 years, mortality did not differ significantly between the groups (hazard ratio for rate control vs. rhythm control, 0.87 [95% CI, 0.75 to 1.01]; P= 0.08), and the rate-control approach was associated with a lower risk for adverse drug effects (12). The results of the other large study were consistent with these findings (13). The RAte Control versus Electrical cardioversion for persistent atrial fibrillation (RACE) study randomly assigned 522 patients with persistent atrial fibrillation after electrical cardioversion to either rhythm control or rate control; the mean follow-up was 2.3 years (13). Patients in both treatment groups received oral anticoagulant drugs. There was a nonsignificant trend toward reduced death or other serious cardiovascular events in patients treated by the rate-control strategy. Consequently, economic factors often play a substantial role in guiding treatment selection. Several authors have examined the incremental cost-effectiveness of rhythm-control versus rate-control strategies for treating atrial fibrillation; however, their studies have been confined to modeling exercises of hypothetical scenarios that lack data on efficacy and resource use from randomized trials (14, 15). This paper reports an economic analysis based on the results of AFFIRM. The objective was to estimate the incremental cost-effectiveness of rhythm-control versus rate-control strategies from AFFIRM. Methods AFFIRM Study Sample AFFIRM included 4060 patients with atrial fibrillation whose treatment was block randomized by center to be either rhythm control or rate control (12). Similar to patients with atrial fibrillation in the general population, most of the patients in AFFIRM were older men (men represented 61% of the sample) with common associated cardiovascular comorbid conditions (history of hypertension [71%], coronary artery disease [39%], and congestive heart failure [9%]). The mean age (SD) for all patients was 69.7 9.0 years, and 75% were 65 years of age or older. The qualifying event was the first episode of atrial fibrillation in 34% of patients and recurrent atrial fibrillation in the remaining 66% of patients. The overriding principles for enrollment of patients in AFFIRM were based on the clinical judgment of the investigator and were as follows: Atrial fibrillation was likely to be recurrent, atrial fibrillation was likely to cause morbidity or death, long-term treatment for atrial fibrillation was warranted, anticoagulation was not contraindicated, the patient was eligible for at least 2 drug trials in both treatment strategies, and treatment in both strategies could be initiated immediately after randomization. Additional information on the design, inclusion and exclusion criteria, and results of AFFIRM are available elsewhere (10-12). The economic analysis described here compares costs and effects of the 2 management strategies among patients enrolled in AFFIRM from the perspective of a third-party payer. The outcome was the incremental cost-effectiveness ratio comparing rhythm control and rate control, measured in dollars per life-year gained. Survival We obtained data on survival from the time of randomization to the end of study follow-up and use of specific health care resources for all 4060 AFFIRM patients. For patients who were lost to follow-up, withdrew from the study, or had incomplete follow-up, all available data were included in the analysis. Patients were censored at withdrawal or loss to follow-up. We derived the within-study mean survival time for each treatment group by using the KaplanMeier product limit estimator to account for censoring during follow-up (16). To obtain an unbiased estimate of mean survival, exposure was truncated at 5.65 years, which was the longest follow-up observed in AFFIRM (17). Resource Use and Costs We estimated costs by multiplying the number of each resource used by its unit cost (18). All unit costs for resources were estimated in U.S. dollars for the year 2002. Price estimates from earlier years were adjusted by applying the Consumer Price Index, Medical Care component (19). For each measure of resource use, 3 different unit costs were derived and considered in separate analyses: a base case for the most likely scenario, a low estimate, and a high estimate. The analysis considered costs of all hospitalizations, cardiac procedures, cardioversion, short-stay and emergency department visits, and medications used to treat atrial fibrillation from the perspective of a third-party payer. Hospital Costs At each follow-up visit during the study, the total number of hospitalized days since the last visit was recorded, along with the primary reason (cardiovascular or noncardiovascular cause) for hospitalization. The mean cost per hospital day was estimated from the Healthcare Cost and Utilization Project (HCUP) statistics for the 1995 HCUP-3 Nationwide Inpatient Sample (20) for Diseases of the Circulatory System, excluding any diagnosis associated with a mean patient age of younger than 18 years, for cardiovascular and noncardiovascular causes. The low and high estimates of the per diem for hospital days were based on the 25th and 75th percentile of mean charges, respectively. The HCUP prices were adjusted to represent costs by using a cost-to-charge ratio of 0.575, which is based on the 2002 estimate from the Centers for Medicare & Medicaid Services (21). In addition, physician charges for subsequent hospital care as a level II visit (Current Procedural Terminology [CPT] [22] code 99232) were applied for each hospital day recorded. In the base case, an average estimate for the physician fee payment for this CPT code was calculated from the 2002 Physician Fee Schedule Payment Amount File National/Carrier for facility-based procedures for all carriers and localities listed in the database (23). In the sensitivity analysis, we used the minimum physician fee across carriers and localities for each procedure as the low cost estimate. We based the high cost estimate on the standard billed charge from the Marshfield Clinic, an ambulatory care facility in Marshfield, Wisconsin. This clinical center recruited most patients in the study and provides an estimate of charges for centers in the United States. Although these estimates are based on data from 1 facility, they are a reasonable estimate for the high-cost scenario, in between billed charges from a teaching hospital and a private clinic. Costs of Cardiac Procedures At each follow-up visit during the study, the number of cardiac procedures (percutaneous transluminal coronary angioplasty, coronary artery bypass graft surgery, pacemakers, valve surgery, ablation) performed since the previous follow-up visit was recorded. No information was available from AFFIRM to describe the number of arteries revascularized during percutaneous transluminal coronary angioplasty interventions or coronary artery bypass graft surgeries. We assumed that only 1 lesion was treated for each percutaneous transluminal coronary angioplasty procedure. We estimated the number of arteries revascularized during bypass surgery as a weighted average from the National Hospital Discharge Survey (NHDS) data set for 2000 (24, 25). We included the costs of the most frequent cardiac procedures in the analysis. Hospital costs include the costs of all facility personnel except physicians. Physician costs consisted of a physician fee for diagnostic and therapeutic procedures, as well as any applicable anesthesia fee. Perfusionist fees for open-heart cardiac procedures were not included because these costs are included in the hospital costs. The analysis included costs for pacemakers and implantable cardioverter defibrillators (ICDs) because they have high unit cost (24, 25). In the base case, the hardware cost (that is, device and electrode or electrodes costs) for the most widely used single-chamber and dual-chamber device was assigned on the

The economic burden of schizophrenia in Canada.

Objective: To estimate the financial burden of schizophrenia in Canada in 1996. Method: Using a prevalence-based approach, all direct health care costs, administrative costs of income assistance plans, and costs of incarceration attributable to schizophrenia were determined. Also included was the value of lost productivity associated with premature mortality and morbidity. In addition to using published papers and documents, direct contact was made with representatives from various provincial and federal programs for estimates of the direct health care and non-health care costs. Results: The estimated number of persons with schizophrenia in Canada in 1996 was 221 000, with equal distribution between males and females. The direct health care and non-health care cost was estimated to be