Influence of socioeconomic factors on medically unnecessary ambulance calls

Emergency ambulance system in Japan

In Japan, local governments organize emergency ambulance service as a public service. Anyone can use an ambulance free of charge by making a phone call to "119". Ambulance crews are dispatched for all emergency calls, and all patients attended by a crew are transported to a hospital unless they refuse. Most local governments staff ambulances with emergency life-saving technicians who are trained for cardiopulmonary resuscitation [15, 25, 26].

Study site and study design

The study was conducted in Yokohama, Japan, which has a population of 3,579,133 (2005 census) and covers an area of 434 square kilometers. In Yokohama, the number of ambulances dispatched in 2005 was 1.58 times greater than that in 1995 [16], although the population in 2005 was only 1.08 times greater than that in 1995 [27]. The city's ambulance transport service is unified and managed by the Emergency Medical Division of the Yokohama Fire Bureau.

We conducted an anonymous self-administered questionnaire-based survey targeting 3,600 city residents randomly selected from the city resident registration list; 1,200 city residents were extracted from each of three age groups, 20–39 years, 40–64 years, and 65 years or over. Questionnaires were distributed once and returned in September 2004. Reminders were not sent. Detailed information about sampling for the study has been published elsewhere [28].

Questionnaire

The questionnaire consisted of three parts. The first part included questions on demographic and socioeconomic characteristics such as the respondent's age, gender, and pretax annual household income.

The second part included hypothetical questions pertaining to ambulance use. We applied the dichotomous choice method to estimate the probability of calling an ambulance [29–31]. Three hypothetical nonemergency situations were presented. Scenario 1 represented a situation in which the respondent faces a nonserious condition such as an ankle sprain or bruise on the leg. Scenario 2 represented a situation in which the respondent's child or young relative faces a nonserious condition such as a nose cold. Scenario 3 represented a situation in which the respondent's elderly relative faces a nonserious situation such as lack of transport to a clinic. The influence of user charges was assessed along with the hypothetical questions by presentation of one of eight possible prices ($0, $9.50, $28.50, $47.50, $95.00, $190.00, $285.00, and $475.00; $1 = ¥105) for use of an ambulance. Each respondent was shown only one price randomly selected from the eight prices for each of the three hypothetical scenarios. For each situation, respondents answered "yes" if they would call an ambulance or "no" if they would not. The three hypothetical scenarios were as follows:

Scenario 1: At around 6 a.m., you miss your step and tumble down the stairs. You then develop pain in your legs, but you can walk with difficulty. If you were in this situation, would you call an ambulance?

Scenario 2: One morning, a child who is living with you develops a cough and nasal discharge. You give him/her a drug, which has been kept at home for future use, but the symptoms do not disappear even by 8 p.m. on the same day. If you were in this situation, would you call an ambulance? Even if you do not live with a child, please answer the question as if you do.

Scenario 3: It's around 8 a.m. An elderly person who is living with you needs to go to the hospital to see his/her primary care physician. Normally, someone in your family would drive the elderly person to the hospital, but no one can do so today. If you were in this situation, would you call an ambulance? Even if you do not live with an elderly person, please answer the question as if you do.

The third part included questions on the city's emergency medical system, such as "Have you ever used an ambulance?" "Do you feel hesitant to call an ambulance?" "Are you familiar with the primary emergency medical centers?" "Are you familiar with the emergency medical information center?" and "Do you know of any private patient transport service?"

Analysis of the influence of socioeconomic factors

The probability P of a "yes" response to the dichotomous choice questions was expressed by a probit model [32, 33]:

where "Y" equals 1 if the respondent chose "yes" and 0 if otherwise. Φ is the standard cumulative normal distribution, and β_n is the estimated parameter vector. Price is the hypothetical price of ambulance use and is treated as a quantitative variable. Gender is a dummy variable (male = 1, female = 0). Age is the age group that the respondent belongs to, with 1 representing 20–29 years, 2 representing 30–39 years, 3 representing 40–49 years, 4 representing 50–59 years, 5 representing 60–69 years, 6 representing 70–79 years, and 7 representing 80 years and over. Family is a dummy variable for family structure: living alone was expressed as 1 (other = 0), when the response to Scenario 1 was analyzed; living with children aged 5 years or younger was expressed as 1 (other = 0) when the response to Scenario 2 was analyzed; living with a relative aged 65 years or older was expressed as 1 (other = 0) when the response to Scenario 3 was analyzed. Income is the pretax annual household income, expressed as 1 if less than $19,000, 2 if $19,000 to $37,999, 3 if $38,000 to $56,999, 4 if $57,000 to $75,999, 5 if $76,000 to $94,999, and 6 if $95,000 or more. Car is a dummy variable for possession of a car by which the respondent could visit a hospital or clinic (yes = 1, no = 0). History is a dummy variable for the respondent's past ambulance use (yes = 1, no = 0). Hesitation is a dummy variable for feeling hesitant to use an ambulance (yes = 1, no = 0). E.Center is a dummy variable for knowledge of a primary emergency medical center (yes = 1, no = 0). E.Info is a dummy variable for knowledge of the emergency medical information center, which serves as a directory where individuals can obtain the name of a suitable hospital or clinic (yes = 1, no = 0). P.Trans is a dummy variable for knowledge of a private patient transport service (yes = 1, no = 0).

The marginal effect of each factor, that is, a change in the probability of a "yes" response (for an independent binary variable) or of a one-unit increase (for an independent ordinal variable), i.e. estimated percentage of incremental demands for ambulance use produced by socioeconomic factors, was also computed with the probit model.

Statistical analysis

STATA/SE 8.2 software for Windows (Stata Corp., College Station, Texas, USA) was used for all statistical analyses including probit regression analysis. P values of less than 0.05 were considered statistically significant.

Ethics

The questionnaire-based study was conducted anonymously. No personal information, such as the respondent's name or address, was contained in the returned questionnaires. Analysis of ambulance calls to the Yokohama Fire Bureau to estimate incremental demand was also conducted anonymously. The study was approved by the Committee on the Regional Emergency Medical Service System of Yokohama after ethical aspects of the study were reviewed by the Human Rights Affairs Division, Civic Affairs Bureau, City of Yokohama.

Characteristics of respondents

Number of respondents who stated they would have called an ambulance

Factors influencing unnecessary ambulance calls

In Scenario 1, i.e., when facing a nonserious situation, male respondents were more likely than female respondents to state that they would call an ambulance (p = 0.046); age of the respondent positively influenced the hypothetical decision to call an ambulance (p < 0.001); and respondents who lived alone were also more apt to state that they would call an ambulance (p = 0.015). Possession of a car and hesitation to call an ambulance had significant negative effects in Scenario 1 (p = 0.002 and p < 0.001, respectively). In Scenario 2, i.e., when a child or young relative faced a nonserious situation, male respondents were more likely than female respondents to state that they would call an ambulance (p = 0.001); age of the respondent and a history of ambulance use positively influenced the hypothetical decision to call an ambulance (p < 0.001 and p = 0.004, respectively). Household income, possession of a car, hesitation to call an ambulance, and knowledge of a primary emergency medical center had significant negative effects in Scenario 2 (p = 0.001, p < 0.001, p = 0.027, and p = 0.005, respectively). In Scenario 3, i.e., when an elderly relative faced a nonserious situation, male respondents were more likely than female respondents to state that they would call an ambulance (p = 0.001), and age of the respondent positively influenced the hypothetical decision to call an ambulance (p < 0.001). Possession of a car, hesitation to call an ambulance, and knowledge of a primary emergency medical center had significant negative effects in Scenario 3 (p = 0.018, p = 0.006, and p = 0.013, respectively). There was no evidence of multicolinearity; the sample correlation coefficient between any pair of variables was < +0.70 and > -0.70.

Potential incremental demands for unnecessary ambulance use produced by socioeconomic factors

Marginal effects estimated by the probit model show the potential incremental demands produced by socioeconomic factors (Table 3). Analysis of responses to Scenario 1 indicated that the incremental demands due to age, living alone, nonpossession of a car, and nonhesitation to call were an estimated 3.9%, 5.5%, 4.4%, and 5.4%, respectively, of the unnecessary calls that individuals would have placed for themselves. These four factors together potentially produce 19.2% of the overall incremental demand. From the analysis of responses to Scenario 2, the incremental demands due to age, nonpossession of a car, nonhesitation to call, and lack of knowledge of primary emergency medical centers were estimated at 2.5%, 7.1%, 3.0%, and 4.2%, respectively, of the calls that family members would have made for children in unnecessary situations. The potential incremental demand produced by the four factors together was estimated at 16.8%. Similarly, the incremental demands due to age, nonpossession of a car, nonhesitation to call, and lack of knowledge of primary emergency medical centers estimated from the analysis of responses to Scenario 3 were 1.3%, 2.3%, 2.4%, and 2.4%, respectively, of the unnecessary calls that family members would have made for elderly relatives. The potential incremental demand produced by the four socioeconomic factors together was estimated at 8.4%.

Limitations

Our study was limited in several ways. First, although there are many situations for which an ambulance call is inappropriate, the hypothetical situations presented in the questionnaire were few. Although the socioeconomic factors that were shown in the study to influence the decision to call an ambulance were nearly the same for all three situations, further research that tests various other situations is needed to clarify the socioeconomic factors that consistently influence the decision to call an ambulance, regardless of the situation. Second, this study used hypothetical scenarios to elicit intention regarding ambulance use rather than data from actual calls. With the dichotomous choice method, there can be a discrepancy between a hypothetical "yes" and a real "yes" [41–43]. Respondents may have been inclined to answer "no" to the questions in which nonemergency situations were presented. Third, the study was conducted in a large urban city. The results of the study may not be applicable to rural areas. Fourth, the questionnaire response rate was 60%; thus, the survey data may not be representative of the whole target population. Non-response bias is always a possibility in a self-administered questionnaire-based survey. In our study, elderly respondents aged 65 years or over accounted for 41.4% of the total respondents. Thus, results of the study may have tended to reflect elderly persons' intentions, although, we adjusted for the influence of age by means of multivariate regression analyses. Another potential bias, that persons who call an ambulance easily might tend not to participate in such a survey, is difficult to resolve.