Despite being effective in the primary and secondary prevention of sudden cardiac arrest, the cost-effectiveness of implantable cardioverter–defibrillator (ICD) therapy remains debated. We attempted to estimate the value ICD recipients place on their ICD device. We used the contingent valuation method to evaluate the willingness to pay (WTP) and the cost benefit of ICD therapy in an unselected population of 237 recipients. A hypothetical scenario was presented to patients in which at the end of their current ICD no public reimbursement for the replacement would occur. Patients were asked to indicate their out-of-pocket WTP for a replacement ICD using a close-ended question format. Seven different “take-it-or-leave-it” bids were randomly varied and assigned to patients. Median WTP was calculated with nonparametric methods, and multiple logistic regression models were generated to identify factors associated with WTP. Only cost of the device was considered. Median WTP was estimated at CAN $4,125, which corresponds to 21% of the cost of the device (CAN $20,000). In multiple logistic regression analysis, a higher bid (odds ratio [OR] 0.95, 95% confidence interval [CI] 0.91 to 0.99, per CAN $1,000 increase) was associated with a lower WTP, whereas a higher gross family income (OR 2.3, 95% CI 0.9 to 9.0) and higher education (OR 2.2, 95% CI 0.9 to 5.1) were associated with a trend for higher WTP. In conclusion, ICD recipients would be willing to pay a substantial amount for a replacement ICD. Considering the expensive price of the device, ICD recipients value favorably the benefits provided by the ICD.
Implantable cardioverter–defibrillator (ICD) therapy decreases the risk of sudden cardiac arrest in secondary (patients already resuscitated from sudden cardiac arrest) and primary (patients at risk for sudden cardiac arrest) prevention. However, ICD therapy remains expensive; 1 unit costs around CAN $18,000 to CAN $22,000, with the most sophisticated models costing up to CAN $35,000. When considering procedure and follow-up costs associated with the device, the total cost reaches $68,836. Despite being borderline cost-effective, with incremental cost-effectiveness ratios ranging from $38,389 to $150,000 per life year gained, ICD therapy is reimbursed by the majority of insurance plans and third-party payers in most Western countries. The contingent valuation approach is an alternative preference elicitation analysis that attempts to assess, in monetary values, all benefits arising from a given therapy. This may be achieved by offering patients a hypothetical market setting in which the therapy could be purchased and then determining their maximum willingness to pay (WTP; i.e., the highest price they would be willing to pay) for such a therapy or intervention. The theoretic advantage of the contingent valuation approach is the evaluation of all health- and nonhealth-related benefits of an intervention at the same time, all in a monetary unit. The contingent valuation approach may therefore provide a complementary point of view in our assessment of the economic value of ICD therapy from a patient’s perspective. The aim of this project was to measure the value that ICD recipients assign to the replacement of their ICD units.
Methods
From June 2005 to November 2007, all patients followed in the Centre Hospitalier de l’Université de Montréal ICD clinic were invited to participate in the study. The only exclusion criterion was an inability to understand written French. Of the 378 patients who were approached, 237 (63%) agreed to participate. This cross-sectional study protocol was approved by the hospital’s scientific and ethics committees, and all patients provided written informed consent.
We used a self-administered questionnaire, delivered in person by trained research personnel, and instructions on how to respond. The questionnaire could be answered in the ICD clinic or at home and returned by mail. The questionnaire was used to evaluate demographic data, health-related quality of life (QOL), and maximum WTP. The questionnaire also included questions regarding patients’ perspectives on the value of their ICDs using Likert scales. Medical charts were reviewed for co-morbidities and clinical indication for implantation. Detailed ICD charts were also reviewed, specifically looking for shocks or antitachycardia therapy.
To assess a possible relation between WTP and self-reported QOL, patients were asked fill out questionnaires including the Symptom Checklist-90-R (SCL-90-R), which contains questions from several domains of psychological functioning such as depression, anxiety, and somatization, and the 36-item Medical Outcome Study Short-Form Health score (SF-36). In addition to QOL, we assessed patients’ utility score with the EuroQOL questionnaire.
We used a contingent valuation method to assess each patient’s WTP to have the ICD replaced in case the battery reached the end of its usefulness. We used a closed-ended question in a “take-it-or-leave-it” format in which individuals were asked if they were willing to pay a specific price for the therapy presented. To evaluate the monetary value of ICD therapy, each patient was asked the following question: “The situation presented to you is hypothetical. It is presented to evaluate the amount of money you would be willing to pay for an ICD. Imagine that your ICD battery does not work anymore, needs to be replaced, and that the government only pays for 1 ICD unit per person. Would you be willing to pay $X to have it replaced? (Circle 1 answer): (1) yes, (2) no.”
The price (X) was changed among 7 possible bids ranging from CAN $5,000 to CAN $35,000, with CAN $5,000 increments in a ±CAN $15,000 spread around the current price of CAN $20,000. We did not inform patients of the actual price of an ICD. Patients who refused to answer the WTP scenario questions or gave an “I don’t know” response were excluded from analyses.
Because the study was performed within the context of the Canadian publicly funded health care system, we used the perspective of the patient as a hypothetical copayer, limiting the costs to the device itself, excluding procedure-related and future care costs. This perspective was chosen to mimic copayment for drugs imposed by public insurance drug plans, payments that can amount to high values especially for senior citizens.
Continuous variables are described as means ± SD and categorical variables as frequencies. Non-normally distributed continuous variables, such as QOL scores, were compared with the Wilcoxon rank-sum test. We used nonparametric methods to estimate the median WTP as described by Kriström. To identify predictors of WTP, we first tested univariate associations with all potential correlates and then used multiple logistic regression to identify independent predictors. Price of the bid, gross family income, having been shocked by an ICD, previous ICD replacement, and reason for ICD implantation (primary vs secondary prevention) were hypothesized a priori to influence WTP and were forced into the multiple logistic model regardless of their statistical significance. We used forward and backward model selection procedures separately to identify other statistically significant variables (p <0.05) among other demographic characteristics, co-morbidities, and patients’ QOL dimension scores. A 2-tailed p value <0.05 for the Wald test was used as a criterion for statistical significance for all tests. To avoid multicollinearity of SCL-90-R, SF-36, and utility scores (by the EuroQOL) in preliminary analyses, separate models were fit, each with only 1 among the highly correlated scores. Then, the final model included the score that provided the best fit, i.e., the lowest model deviance. The results of the final multivariable model were presented in adjusted odds ratios (ORs) for each predictor variable with 95% confidence intervals (CIs). Analyses were performed using SAS 9.1.3 (SAS Institute, Cary, North Carolina).
Results
Of the 237 patients who answered the questionnaire, almost all (236) filled it at home, and only 1 in person at the ICD clinic. Thirteen patients (5%) refused to answer the WTP questions and 4 patients (2%) answered with a response of “I don’t know”; they were excluded from the WTP analyses. Patients underwent their ICD implantation on average 27 months before answering the questionnaire. Responders’ demographic data are presented in Table 1 .
| Age (years) | 64 ± 11 |
| Men | 185 (84%) |
| Caucasian ethnicity | 216 (98%) |
| Marital status | |
| Married or living as married | 145 (66%) |
| Divorced or separated | 28 (13%) |
| Widowed | 24 (11%) |
| Never married | 21 (10%) |
| Missing | 2 (0.9%) |
| Employment | |
| Currently working | 36 (16%) |
| Unemployed or retired | 183 (84%) |
| Missing | 1 (0.5%) |
| Annual household income | |
| <$30,000 | 96 (44%) |
| $30,000–$60,000 | 61 (28%) |
| $60,000–$90,000 | 26 (12%) |
| >$90,000 | 17 (7.7%) |
| Missing | 20 (9.0%) |
| Education level | |
| Elementary | 37 (17%) |
| High school | 82 (37%) |
| College | 50 (23%) |
| Bachelor degree | 35 (16%) |
| Master or doctorate degree | 12 (5.5%) |
| Missing | 3 (1.3%) |
| Heart failure | 111 (64%) |
| Previous myocardial infarction | 177 (86%) |
| Previous coronary artery bypass grafting | 73 (38%) |
| Indication for implantation | |
| Primary prevention | 125 (57%) |
| Secondary prevention | 90 (41%) |
| Missing | 5 (2.3%) |
| Time since implantation (months) | 27 ± 22 |
| Having been shocked by an implantable cardioverter–defibrillator | 50 (23%) |
| Missing | 3 (1.3%) |
| Previous implantable cardioverter–defibrillator replacement | 49 (22%) |
A large percentage of patients did not know whether they had received shocks or not (44%). Of the 23% of patients who received shocks, most did not know if they had been lifesaving or not (58%). Actually, only 4 patients received inappropriate shocks. Eighty-five percent of patients indicated that they were more afraid of dying since the ICD implantation, and these data were similar when comparing patients treated for primary prevention of sudden cardiac arrest (88%) to those who already presented a malignant ventricular arrhythmia (78%, p = 0.25). Although most patients (63%) said they valued their QOL higher than their life expectancy, 84% said they would be worried if their ICD had to be removed. Patients who presented a malignant ventricular arrhythmia (90%) were in greater fear of the consequences of the removal of their ICD compared to patients who received their ICD for primary prevention of sudden cardiac arrest (79%, p = 0.02).
With the price range we used, most patients (75%) were not willing to pay the proposed bid in the WTP question. Assuming that all patients would be willing to have their ICD replaced if there were no out-of-pocket cost, a linear extrapolation of the WTP, analyzed by nonparametric method, determined a median WTP amount of CAN $4,125 ( Figure 1 ).
Mean scores of several dimensions of QOL, anxiety, and depression assessed with the SCL-90-R and SF-36 questionnaires and patients’ utility score measured with the EuroQOL are presented in Table 2 . Compared to the 165 patients who were unwilling to pay, the 55 patients who were willing to pay reported similar scores in anxiety, somatization, SF-36 Mental Component score, and utility scores. However, patients who were willing to pay the proposed amount showed a trend for being less depressed (p = 0.054) and had a better SF-36 Physical Component score (p = 0.06).
| Variable | Willing to Pay | Unwilling to Pay | p Value |
|---|---|---|---|
| (n = 55) | (n = 165) | ||
| Symptom Checklist-90-R ⁎ | |||
| Anxiety score | 0.31 (0.00–0.50) | 0.40 (0.00–0.60) | 0.49 |
| Depression score | 0.56 (0.08–0.92) | 0.75 (0.23–1.08) | 0.05 |
| Somatization score | 0.56 (0.25–0.83) | 0.67 (0.25–0.92) | 0.23 |
| 36-Item Medical Outcome Study Short-Form Health score † | |||
| Mental Component scale | 50.54 (40.90–58.49) | 49.78 (42.51–58.39) | 0.69 |
| Physical Component scale | 42.09 (37.23–50.80) | 38.76 (30.00–47.40) | 0.06 |
| EuroQOL | |||
| Utility score | 0.79 (0.69–1.00) | 0.75 (0.69–1.00) | 0.48 |
⁎ Scores range from 0 to 4, where 4 is the worst possible score; a score lower than 1 represents patients with mild symptoms.
† Scores range from 0 to 100, where 100 is the best possible score, and are scaled to population means of 50 ± 10.
Results of univariate logistic regression models for predicting binary response to the WTP question are listed in Table 3 . Patients’ answer to the WTP question were influenced by higher gross family income (≥CAN $60,000 vs <CAN $60,000 or missing) and higher education (college or higher vs lower). Price of the bid was associated with a lower probability of answering yes to the WTP question. Better physical health-related QOL as assessed by the SF-36 Physical Component score was associated with a marginally significant trend for higher probability of answering yes to the WTP question. A lower depression score (less depressive symptoms) based on the SCL-90-R was associated with a trend for a higher probability of answering yes to the WTP question. No other variables were found to be predictive of the answer patients gave to the WTP question, including whether they were “married or living as married,” which could indicate a more complex relation to decision-making regarding family spending.
