Background
The use of decision aids to promote shared decision making is especially important in prostate cancer treatment decisions, a condition in which personal values and preference regarding risks of treatment side effects influence the patient’s choice [
1‐
3]. The Personal Patient Profile-Prostate (P3P) is a web-based decision support system that has been shown to decrease decisional conflict in men newly diagnosed with localized prostate cancer [
4,
5]. Despite evidence of the value of decisions aids, uptake in the United States remains low and economic value may be an important factor. U.S. health care systems, third-party payers and often, providers, want evidence of economic value and patient acceptance before adopting. There is a lack of information about the direct economic value of decision aids precluding providers’ and payers’ consideration of investments in decision aids [
6]. There are few economic evaluations of decision aids because the downstream cost savings are often unavailable for measurement. Further, for value-sensitive decisions there are often no right or wrong choices based purely on clinical evidence. However, after engaging with a decision aid, each user can ascertain the value of the aid. Willingness-to-pay (WTP) then becomes an ideal framework for direct economic valuation of decision aids from the user’s viewpoint [
7‐
11] Other non-economic measures, such as decision regret and decisional conflict also can act as measures of an optimal decision choice; [
12] these outcomes were part of efficacy testing in our previous clinical trials [
4,
5].
There are three contemporary approaches to the measurement of WTP, all of which have been used to find the maximum price a person is willing to pay for a given quantity of a good: 1) contingent valuation, a stated preference approach in which the consumer is directly asked their WTP for a total good; 2) experimental auction, a revealed preference approach where persons are provided actual money and asked to behave in a market of goods, or observed directly in transactions; and 3) conjoint analysis, a stated preference approach where WTP is derived from ranking, rating or selection of goods alternatives by attributes [
13]. The choice of method is primarily driven by cost, ease of use, whether or not the good is currently available in the market, the desire for pricing or utility, and whether the estimated WTP is likely to mimic actual market place behavior [
14].
The purpose of this study was to estimate clinical trial participants’ WTP for the P3P during the time of decision making for localized prostate cancer care. We also present data on the WTP reported for the control group who received a variety of other ‘usual care’ preparations.
Discussion
The majority of participants in both study groups readily reported WTP for decision preparation, for the P3P intervention and for the various materials provided to participants in the UC group. Participants in both study groups were willing to pay a median value ranging from $25–50, depending on SV. Demographic variables and participants’ baseline stages of decision making predicted various WTP values suggesting differential benefit for individual characteristics.
Participant characteristics related to higher WTP values suggest which patients likely need greater decisional support. Those not married or partnered valued P3P more; men with fewer personal resources may value a decision aid more highly than a man with a partner who can share in the decision-making process. Those who had not yet or just started decision making logically indicated a higher value for decision support. Finally, the finding that participants are willing to pay more for P3P if they prefer to make shared rather than passive decisions is important. This valuation could mean that participants who were not yet ready to engage with a decision aid, had a more private decision-making style or found less value in decision aids. These WTP valuations of P3P can help to focus on future patients who would benefit most from decision aids: men not yet starting decision making, those without a partner/spouse to share in the decision process, and those open to shared decision making. Increased WTP for P3P in both lower and higher incomes, versus moderate incomes, differs from the expected direct association of WTP with ability to pay. Further work could explore detailed financial resources and WTP responses.
The demand curve for WTP demonstrated that the approach performed well in economic terms. The curve flattened at lower prices as the demand became more elastic in the presence of low-cost education substitutes.
In the UC group, the type and extent of preparatory educational materials reported was variable. It was likely more difficult to provide a consistent WTP value for a collection of various educational materials. Our finding that a lower proportion of UC participants, as compared to P3P users, were able to provide a WTP value supports this implication.
The participant characteristics associated with WTP for usual care differed from those for P3P. For UC, those with a lower level of education expressed higher WTP; perhaps perceiving a greater need for decision materials provided (or not) by the consulting physician’s office. In contrast to the P3P group, UC respondents who had already decided were willing to pay more than those who had not yet or just started their decision making. This counter-intuitive result might indicate that participants who made a decision even before meeting the physician were willing to pay more for a diverse collection of non-personalized material, while those unable to make a quick decision with the available material valued such materials less. Patients will always have access to a variety of ad hoc and Internet sources and may have difficulty determining quality.
Limitations in our study are those inherent in many WTP studies: protest responses (nonresponses when unwilling to provide a value or for some zero values), preference uncertainty or the confidence of one’s WTP response, and outliers [
7,
24‐
29]. Non-response, infinity (very high valuations) and zero responses are not uncommon in WTP assessments [
7,
27‐
29] and are difficult to interpret. Non-response is often from a lack of exposure to the product, as was the case in our study, but it also can represent a protest response, an expression of preference that is not reflective of economic assumptions underlying WTP theory [
27,
28]. Halstead, et al. and others [
27,
30] found that censoring protest responses should only occur if the characteristics of the protest responders do not differ significantly from the other respondents. Our comparisons of responders and non-responders found no particular characteristic for bias.
Zero responses also can be either a true zero WTP value or a protest vote. For example, some of the P3P participants’ reasons for giving a zero value, unwillingness to pay for available internet information, suggested a protest WTP response. Although some methods have been described modeling zero values, or protest responses, to be conservative we included all zero WTP responses in our estimates [
7,
27,
31]. Preference uncertainty may be caused by incomplete knowledge about product features or just uncertainty about their own preferences [
32]. WTP values are less valid when preference uncertainty is high, such as for the UC’s undefined materials, as evidenced in part by the 44% not providing a WTP value.
Another limitation, common to WTP studies is the presence of outliers. We had one main outlier in the P3P group who expressed a WTP value 100 times greater than the next highest value. There is no consensus on how outliers should be handled in WTP studies; the method largely depends on how the WTP results will be used. For analysis of characteristics associated with WTP values, removing the outliers provides the most accurate results, as we did here in all our means-based analyses. However, for inclusion in cost-benefit analyses, it is suggested to retain those outliers to represent the full range of the benefit of the intervention. An alpha-trimmed mean can also be considered [
33].
Our results are limited by estimating the value of P3P alone as an adjunct to usual patient education. Although P3P’s value may be additive to usual education, we cannot assume that the value of usual education would remain constant when P3P is also used.
A strength of our study was that we measured and controlled for starting point bias, common but not universal in WTP surveys that use a bidding approach. Similar to our finding that the P3P group starting with a high value was willing to pay about $42 more than those starting low, Stalhammar [
34] found a starting point bias, with a WTP of only $9 vs $38 for a pill taken at meals instead of before meals depending on starting with a low or high amount, respectively. Another study evaluating WTP for treatment of bleeding disorders found a median WTP of only $1500 when bidding started at $500 and $3500 if started at $5000 [
8]. Starting point bias is not fully eliminated by any method that provides a direct WTP value. Future studies can use multiple approaches to explore the effects of different WTP questionnaire methods [
19]. Although conjoint analyses are becoming more popular for measuring patient preferences in health care, their WTP responses are not independent of other measure attributes included in a conjoint approach.
There are very few economic studies evaluating decision aids, and fewer that use WTP for evaluating cost-benefit. One such study used WTP to determine the cost-benefit of the use of a decision support program for patients with breast cancer in a rural community [
18]. This program included a trained facilitator who met with the participant before a physician visit and provided a personalized written “consultation plan” to use during the visit. WTP averaged $150, however the cost of this labor-intensive program was high, resulting in a positive net-benefit only when the most efficient parameters were considered. We anticipate that the P3P program will have a more efficient implementation cost due to its intrinsic automation and be able to demonstrate a positive net-benefit given the high mean WTP value from participants. As WTP for P3P reported here was a demonstration of the perceived value, in future analyses we will combine this with the costs for P3P program delivery across delivery sites to determine the overall value of the P3P program in a net benefit calculation.
The WTP expressed by men making decisions after a diagnosis of localized prostate cancer demonstrated a valued decision aid, tailored to users’ personal preferences and concerns and always available for return visits. Such availability is important, given the frequent reports that patients are often not prepared for cancer consultations [
18,
35], or do not remember the content of these meetings [
10,
36,
37]. The WTP for P3P was instructive, especially when further characterized by influence of demographic characteristics. Our findings suggest that single men or those with modest incomes are likely targets for decisional support. Further, men who are open to a shared decision-making style are likely eager recipients of decision support. Health care systems with limited resources could target groups with the greatest need for decision preparation and provide P3P.