First we briefly introduce e-health systems and describe an example. Kamaishi City, Iwate Prefecture, which started its e-health system in 1993, was chosen for the survey data, as it has one of the longest records of implementation. In addition, the city’s system is operated by a private medical corporation named Rakuzankai at a monthly charge of 2,500 yen (approximately US$22.73) per family of up to four persons, and a users’ association that conducts many events to promote usage.
Since the system uses the CATV network, transmission charges are zero. The peripheral device used by Kamaishi City ,called
“Urara,” is manufactured by Nasa Corporation. The device is
equipped with memory, an electric sphygmomanometer, an electro-cardiograph, an electric signboard, and a button for answering questions. At the hospital, nurses check transmitted data and report to the doctor as well as to the device’s users whenever they notice abnormal symptoms. The hospital sends monthly reports to all users with comments, which are used for their health management. There are 211 devices, and the total number of users is 409.
Nishi-aizu and Sangawa have almost identical systems based on “Urara,” since they imitated Kamaishi’s system.
Katsurao Village has a different system, using NEC’s
“Sukoyaka-mate” device. Katsurao Village distributes terminals to all families who have person(s) older then sixty-five years old, and thus it has the largest number of users. In Table 1, we compare the systems of four regions. We will see that these four regions yield different results according to the differences in their systems.
Table 1. Comparison of systems
No. of users No. of devices Terminal Network Charges
3.1. Contingent valuation method (CVM)
In order to measure the benefits of services not traded on the market, the following methods are often utilized: (a) travel cost method: (b) replacement costs method; (c) hedonic approach; and (d) CVM. In what follows, we use CVM, which has been recently widely adopted in the fields of Health Economics and Environmental Economics. In CVM, the benefits to users are measured in terms of WTP, which is the monetary amount users are willing to pay for receiving a service.
By asking the WTP of each user, we can then construct the surrogate demand function for the e-health system.
Although CVM and WTP have strong theoretical bases, CVM tends to introduce bias, because it asks for concrete valuation and choice under fictitious circumstances. Care should be taken to clarify what kind of bias is introduced, and to remove this bias (for further discussion on biases, see [5]). .
3.2. Questionnaire
For the surveys, we use questionnaires. Let us take the example of Kamaishi City. We interviewed 348 users of the e-health system questions pertaining to the following: (a) WTP; (b) effectiveness; (c) frequency of usage; and (d) user characteristics such as age, gender, income, education, and health condition. The portion of the questionnaire related to WTP is as follows. We begin by asking whether they would be willing to pay monthly charges of 5,500 yen (US$45) to use the system. If their answer is “yes,” we then ask whether they would be willing to pay 7,500 yen (US$62.5). If they reply “yes”
again to 7,500 yen, their WTP is 7,500 yen. If “no,” then we lower the amount to 6,500 yen (US$54.17). If they
reply “yes” to 6,500 yen, then that is their WTP. If again their answer is “no,” we lower the amount further to 5,500 yen. We repeat this process until their WTP is determined. Out of a total of 348 replies, 291 had no missing value. The distribution of WTP from the survey is as follows: 10,000 yen (16 users), 8,000 yen (1), 7,500 yen (12), 6,500 yen (11), 5,500 yen (62), 4,500 yen (8), 3,500 yen (69), and 2,500 yen (112). The distribution of replies is shown in Table 2. Those of the other regions are shown in Tables 3, 4, and 5.
3.3. Estimation of demand function and WTP Based on the above WTP of each user, we estimate the demand function of the system; more precisely, we estimate the probability of acceptance to amounts questioned and the number of users who will agree to pay.
The functional form of demand to be estimated is assumed to be logistic, namely,
Probability of acceptance = 1- 1/(1 + exp (-α-βlogWTP)).
The probability of acceptance is the ratio of the number of users who reply that they are willing to use the device at the amount of charges provided in the questions. The estimated coefficients α and β are summarized in Table 6, 7,8, and 9 for four regions.
The estimated demand function for Kamaishi’s e-health system is shown in Fig. 1. The average WTP is calculated as the area under this demand function, which results in being 4,519 yen (approximately US$37) per user per month. The demand functions of other regions have the same form.
In sum, WTP of Kamaishi, Nishiaizu, Katsurao, and Sangawa are 4,519 yen, 3,177 yen, 1,640 yen, and 2,955 yen, respectively.
Table 2. Distributions of replies: Kamaishi
WTP (yen) 2,500 3,500 4,500 5,500 6,500 7,500 8,000 10,000
No. of users 112 69 8 62 11 12 1 16
Table 3. Distributions of replies: Nishi-aizu
WTP 0 500 750 800 1,000 1,500 2,000
No. of users 141 4 2 1 54 7 3
WTP 2,500 3,500 4,500 5,500 6,500 7,500 8,000 10,000
No. of users 67 38 1 69 6 13 9 5
Table 4. Distributions of replies: Katsurao
WTP (yen) 0 500 1,000 1,500 2,000 2,500 3,000
No. of users 85 95 113 25 50 3 14
WTP (yen) 3,500 4,000 4,500 5,000 5,500 6,500 8,000 10,000
No. of users 2 1 1 9 2 1 1 7
Table 5. Distributions of replies: Sangawa
WTP 0 500 1,000 1,500 2,000 2,500 3,000
No. of users 93 3 14 38 3 23 5
WTP 3,500 4,000 4,500 5,000 5,500 6,500 10,000
No. of users 6 0 17 1 2 2 2
Table 6. Results of estimation of demand function: Kamaishi
Estimated Standard error t-value p-value α 27.441134 2.0739667 13.231 0.0000 β 3.3033572 0.2474318 13.351 0.0000
Table 7. Results of estimation of demand function: Nishi-aizu
Estimated Standard error t-value p-value α 29.866890 1.9972607 14.954 0.0000 β 3.7547249 0.2382865 15.757 0.0000
Table 8. Results of estimation of demand function: Katsurao
Estimated Standard error t-value p-value α 15.800151 1.8418040 11.775 0.0000 β 2.2491025 0.1794782 12.498 0.0000
Table 9. Results of estimation of demand function: Sangawa
Estimated Standard error t-value p-value α 25.470392 2.5995525 9.798 0.0000 β 3.2403337 0.3281845 9.874 0.0000
WTP 0.0
0.2 0.4 0.6 0.8 1.0
-0.2
2,000 4,000 6,000 8,000 10,000 yen
0 Probability
Figure 1. Estimation of demand function: Kamaishi