Study design and participants

This study analyzed the national ambulance records of the FDMA in Japan from 1 January 2016 to 31 December 2016. These data include all emergency transports throughout Japan except for Tokyo prefecture because fire stations in Tokyo are managed by an organization independent of the national government, and are not included in the FDMA database [13]. The data were collected by EMS personnel, in cooperation with the physicians overseeing the patient’s care. This study included all patients ≧15 years of age suspected to have cerebrovascular diseases by physicians at hospitals, who called an ambulance and were transported to a hospital. Hemiplegia, dysarthria, ataxia or severe headache were classified as cerebrovascular disease-related symptoms. We excluded patients with cardio-pulmonary arrest at the hospital and who had missing data. This study was approved by the Ethics Committee of the International University of Health and Welfare (5-19-46), and the requirement for patient informed consent was waived.

Setting

Japan has an area of 378,000 km² divided into 47 prefectures and the population was approximately 127 million in 2016 [14]. The EMS system in Japan has been described elsewhere [15]. There were 733 fire stations with dispatch centers in 2016; EMS at these fire stations is provided by municipal governments [16]. In most cases, an ambulance has a crew of three providers, including at least one emergency lifesaving technician, a person who has undergone extensive training in the provision of pre-hospital care [17, 18]. Using the protocol established by each municipal fire department, EMS ambulance crews at the scene or emergency dispatchers select an appropriate hospital for emergency care according to medical urgency or the patient’s symptoms.

Data collection

Demographic factors (age and gender), chronological factors (date and time), severity, location, and prefecture were extracted from available data. The severity is classified into 4 categories: dead, severe, moderate and mild [13]. Severe patients are those expected to be hospitalized for over 3 weeks, and moderate patients are those expected to be hospitalized for 3 weeks or less. If patients are not likely to require hospitalization, they are categorized as mild. The location where the emergency occurred is also recorded using the following 5 categories: patient’s home, public area, workplace, road and others. We defined seven geographic regions (Hokkaido-Tohoku, Kanto, Chubu, Kansai, Chugoku, Shikoku, and Kyushu-Okinawa), based on previous studies [19, 20]. To characterize regional (secondary health care area) level effects in these analyses, we obtained data on the following variables at the prefecture level from the Japanese National Survey as well as Japan national physician database provided by Nihon Ultmarc INC: numbers of physicians, female physicians, emergency physicians, and neurosurgeon, population in the area covered by each municipal fire department, the number of elderly people (≧65 years of age), young people (< 15 years of age) per capita income, industry ratio, and mean age of physicians in each area.

Outcome measures

The primary outcome was the difficulty of hospital acceptance of patients suspected to have cerebrovascular diseases [6]. The definition of difficulty of hospital acceptance was the requirement for ≥4 phone calls by EMS personnel to hospitals before obtaining acceptance from destination hospitals, based on reports from the FDMA [2]. The secondary outcome was transportation time from arrival at the scene to arrival at the hospital.

Statistical analysis

Continuous data with skewed distributions was shown as medians and interquartile range (IQR), and categorical data as frequencies and proportions. Bivariate analyses were performed with chi-squared tests for dichotomous variables and the Mann-Whitney U test used for continuous variables. While we use the Mann-Whitney U test to address skewed distribution of some variables, note that the results are almost the same with those obtained using the t-test. For example, the distribution of transportation time is slightly skewed (e.g. skewness = 2.35) but we confirmed that the distribution is sufficiently normal around the mean.

Multivariable analyses were used to assess factors associated with difficulty of hospital acceptance of patients suspected to have cerebrovascular diseases using logistic regression models, and odds ratios (ORs) with 95% confidence intervals (CIs) were calculated. Multivariate linear regression analysis was used to investigate the association between the factors and a reduction in pre-hospital transportation time. We selected covariates based on biological plausibility and previous studies in the multivariable analysis [2, 6, 21]. These variables included age (15–64 years, 65–84 years, ≧85 years), gender (male, female), time of day (daytime [09.00–16.59], night time [17.00–08.59]), day of the week (weekday, weekend), severity (mild, moderate, severe), location (home, public space, workplace, road, others), region (Hokkaido-Tohoku, Kanto, Chubu, Kansai, Chugoku, Shikoku, and Kyushu-Okinawa), month (January, February, March, April, May, June, July, August, September, October, November, December), the number of physicians per population in the area covered by each municipal fire department (<25 percentile, 25–74 percentile, ≧75 percentile), the proportion of emergency physicians to all physicians in the area (<25 percentile, 25–74 percentile, ≧75 percentile), the proportion of neurosurgeon to all physicians in the area (<25 percentile, 25–74 percentile, ≧75 percentile), population in the area, the proportion of older people to all people in the area, the proportion of younger people to all people in the area, per capita income in the area, industry ratio in the area (primary, secondary, tertiary), mean age of physicians in the area, and the proportion of female physicians in the area. In subgroup analyses, we stratified the model according to severity.

Data were analyzed using Stata version 14 (College Station, TX). All tests were two-tailed, and p values <0.05 were statistically significant.

Patient characteristics

Patient characteristics according to the number of phone calls to hospitals by emergency medical service personnel are listed in Table 1. A total of 5283 (2.4%) patients had difficulty obtaining hospital acceptance. There were differences in the groups regarding age, gender, time, day of the week, severity, location, region, number of physicians per population, proportion of emergency physician and neurosurgeon, and proportion of female physician. The mean age of physicians in the area was similar among the groups. The median transportation time was 36 (IQR 29–45) minutes during the night and 34 (28–43) minutes during daytime (Table 2). The Kanto region had the longest transportation times.

Table 1

Basic characteristics and number of phone calls to hospitals by emergency medical service.

	All number of phone calls to hospitals by EMS
	<4		≥4
variables	n = 217643	%	n = 5283	%	p value
Age, median (IQR)	76(65–84)		75(63–83)		<0.001*
Age					<0.001†
15–64	53,721	24.7	1,433	27.1
65–84	118,512	54.5	2,704	51.2
85-	50,693	23.3	1,146	21.7
Gender					0.015†
Male	120,143	55.2	2,937	55.6
Female	102,783	47.2	2,346	44.4
Time					<0.001†
Daytime (9:00–16:59)	84,429	38.8	1,506	28.5
Night (17:00–8:59)	138,497	63.6	3,777	71.5
Day of the week					<0.001†
weekday	159,841	73.4	3,468	65.6
weekend	63,085	29.0	1,815	34.4
Severity					<0.001†
Minor	54,302	25.0	1,202	22.8
Moderate	126,750	58.2	2,959	56.0
Severe	41,874	19.2	1,122	21.2
Location					<0.001†
Home	156,742	72.0	3,848	72.8
Public space	46,191	21.2	946	17.9
Workplace	7,299	3.4	110	2.1
Road	7,325	3.4	176	3.3
Others	5,370	2.5	203	3.8
Region					<0.001†
Hokkaido—Tohoku	30,912	14.2	838	15.9
Chubu	48,256	22.2	1,502	28.4
Kanto	34,844	16.0	366	6.9
Kansai	39,508	18.2	1,511	28.6
Chugoku	17,874	8.2	429	8.1
Shikoku	11,452	5.3	227	4.3
Kyusyu, Okinawa	40,081	18.4	410	7.8
Month
January	21,281	9.8	985	18.6
February	18,108	8.3	604	11.4
March	19,240	8.8	552	10.5
April	17,863	8.2	393	7.4
May	17,479	8.0	393	7.4
June	16,707	7.7	299	5.7
July	16,502	7.6	317	6.0
August	16,499	7.6	292	5.5
September	17,141	7.9	321	6.1
October	18,368	8.4	342	6.5
November	18,606	8.6	372	7.0
December	19,849	9.1	413	7.8
Number of physicians per population					<0.001†
< 25 percentile	59,291	27.2	1,402	26.5
25–74 percentile	100,741	46.3	2,219	42.0
≧ 75 percentile	62,893	28.9	1,662	31.5
Proportion of emergency physicians					<0.001†
< 25 percentile	63,707	29.3	1,353	25.6
25–74 percentile	105,287	48.4	2,479	46.9
≧ 75 percentile	53,932	24.8	1,451	27.5
Proportion of neurosurgeons					<0.001†
< 25 percentile	57,941	26.6	1,310	24.8
25–74 percentile	104,123	47.8	2,780	52.6
≧ 75 percentile	60,862	28.0	1,193	22.6
Mean age of physicians	49.94		50.00		0.319*
Proportion of female physicians	20%		21%		<0.001*
Transportation time (from arrival at the scene to arrival at the hospital), median	8.40		8.65		<0.001*

EMS, emergency medical service; IQR, interquartile range

* Mann-Whitney U test

† chi-squared test

Table 2

Basic characteristics and EMS transportation times.

	transportation times
variables	median	IQR
Age
15–64	34	28–44
65–84	35	29–45
85-	35	28–44
Gender
Male	35	28–44
Female	35	28–44
Time
Daytime (9:00–16:59)	34	28–43
Night (17:00–8:59)	36	29–45
Day of the week
weekday	35	28–44
weekend	35	28–45
Severity
Minor	34	28–43
Moderate	35	28–45
Severe	36	29–46
Location
Home	36	29–45
Public space	33	27–41
Workplace	32	26–41
Road	34	27–44
Others	38	31–48
Region
Hokkaido, Tohoku	36	29–47
Tokai, Chubu, Hokuriku	33	27–42
Kanto	39	32–49
Kansai	35	28–43
Chugoku	36	29–46
Shikoku	34	27–43
Kyusyu, Okinawa	31	25–40
Month
January	36	29–46
February	36	29–45
March	35	28–44
April	35	28–44
May	35	28–44
June	35	28–44
July	35	28–44
August	35	28–44
September	35	28–44
October	35	28–44
November	35	28–45
December	35	29–45
Number of physicians per population
< 25 percentile	37	29–46
25–74 percentile	35	28–43
≧ 75 percentile	32	26–40
Proportion of emergency physicians
< 25 percentile	37	29–47
25–74 percentile	36	29–46
≧ 75 percentile	35	28–44
Proportion of neurosurgeons
< 25 percentile	36	29–46
25–74 percentile	35	29–44
≧ 75 percentile	35	28–44

EMS, emergency medical service; IQR, interquartile range

Risk factors for difficulty of hospital acceptance of patients suspected to have cerebrovascular diseases

In multivariable logistic analysis, a positive association was observed between difficulty of hospital acceptance and night hours (adjusted OR [AOR] 1.54, 95% CI 1.45–1.64), weekend day (AOR 1.32, 95% CI 1.24–1.40), <25 percentile ratio of emergency physicians and neurosurgeons to all physicians (AOR 1.13, 95% CI 1.03–1.23) (AOR 1.36, 95% CI 1.25–1.48), and mean age of physicians in the area (AOR 1.06, 95% CI 1.05–1.07) (Table 3). Conversely, patients 65–84 years of age (AOR 0.86, 95% CI 0.80–0.91), moderate severity (AOR 0.87, 95% CI 0.81–0.94), public space (AOR 0.91, 95% CI 0.85–0.99), and workplace (AOR 0.62, 95% CI 0.51–0.75) were negatively associated with difficulty obtaining hospital acceptance. In addition, there was a high degree of variation in difficulty of hospital acceptance across regions (Fig 2). In multivariate linear regression analysis, similar associations were observed between an increase in pre-hospital transportation time and night hours, weekend days, 25–74 percentile ratio of neurosurgeons to all physicians and mean age of physicians in the area, while the associations were observed between a reduction in pre-hospital transportation time and public space, and workplace. In subgroup analysis, similar associations among the above factors except for public space and 25–74 percentile ratio of neurosurgeons to all physicians were observed.

Fig 2

Regional variation of the odds ratio of number of phone calls to hospitals by emergency medical service personnel.

Table 3

Factors associated with difficulty of hospital acceptance and transportation time among patients suspected to have cerebrovascular diseases.

	All patients
	Number of phone calls to hospitals by EMS		Transportation time
	OR	95% CI	Time Difference	95% CI
Age
15–64	reference		reference
65–84	0.86***	0.80–0.91	-0.17	-0.44–0.09
85-	0.93*	0.86–1.01	-0.60***	-0.96 - -0.24
Female	0.96	0.91–1.02	-0.29***	-0.39 - -0.19
Night	1.54***	1.45–1.64	1.31***	1.03–1.58
Weekend	1.32***	1.24–1.40	0.67***	0.41–0.92
Severity
Mild	reference		reference
Moderate	0.87***	0.81–0.94	0.32	-0.26–0.91
Severe	0.92**	0.84–1.00	0.53	-0.30–1.35
Location
Home	reference		reference
Public space	0.91**	0.85–0.99	-1.31***	-1.66 - -0.97
Workplace	0.62***	0.51–0.75	-2.35***	-2.70 - -2.00
Road	1.00	0.85–1.16	-0.93***	-1.38 - -0.49
Others	1.45***	1.25–1.70	0.78	-0.87–2.44
Region
Hokkaido—Tohoku	0.69***	0.61–0.77	-3.58	-8.37–1.21
Chubu	0.35***	0.31–0.40	-5.69**	-10.33 - -1.04
Kanto	reference		reference
Kansai	1.10*	1.00–1.20	-4.00**	-7.83 - -0.16
Chugoku	0.81***	0.71–0.92	-3.94	-8.97–1.09
Shikoku	0.74***	0.63–0.87	-6.57***	-11.30 - -1.83
Kyusyu, Okinawa	0.22***	0.19–0.26	-8.39***	-13.72 - -3.05
Month
January	reference		reference
February	0.73***	0.65–0.81	-0.34*	-0.71–0.03
March	0.62***	0.56–0.69	-0.91***	-1.27 - -0.55
April	0.47***	0.42–0.53	-1.11***	-1.53 - -0.69
May	0.47***	0.42–0.53	-1.30***	-1.74 - -0.85
June	0.38***	0.33–0.43	-1.60***	-2.10 - -1.11
July	0.40***	0.36–0.46	-1.75***	-2.35 - -1.15
August	0.37***	0.32–0.42	-1.73***	-2.25 - -1.20
September	0.40***	0.35–0.45	-1.41***	-1.98 - -0.83
October	0.39***	0.35–0.45	-1.36***	-1.92 - -0.79
November	0.43***	0.38–0.48	-1.06***	-1.63 - -0.49
December	0.45***	0.40–0.51	-1.03***	-1.54 - -0.52
Number of physicians per population
< 25 percentile	1.09	0.95–1.24	2.81*	-0.16–5.79
25–74 percentile	0.95	0.85–1.04	1.33	-0.91–3.56
≧ 75 percentile	reference		reference
Proportion of emergency physicians
< 25 percentile	1.13***	1.03–1.23	0.86	-0.37–2.08
25–74 percentile	1.17***	1.08–1.26	0.657	-0.87–2.18
≧ 75 percentile	reference		reference
Proportion of neurosurgeons
< 25 percentile	1.36***	1.25–1.48	1.362**	0.10–2.61
25–74 percentile	1.32***	1.22–1.44	1.432***	0.46–2.39
≧ 75 percentile	reference		reference
Mean age of physicians	1.06***	1.05–1.07	0.29***	0.09–0.48
Proportion of female physicians	1.38	0.76–2.53	-2.24	-10.28–5.80

EMS, emergency medical service; OR, odds ratio; CI, confidence interval

* p < 0.10

** p < 0.05

*** p < 0.01

Nagelkerke pseudo R2 = 0.053

Results in context

To our knowledge, this is the first study to document the difficulty of hospital acceptance of patients suspected to have cerebrovascular diseases [6]. A previous study in Osaka, Japan reported that elderly patients, foreigners, unconsciousness, nights and weekends/holidays were associated with difficulty of hospital acceptance at the scene requiring EMS personnel to make ≥5 phone calls to hospitals until the patient was accepted for transport [6]. Another study showed, among older patients, more advanced age, nights, weekend days and gastrointestinal related symptoms were more associated with difficulties of hospital acceptance while patients with cardiac arrest, acute coronary syndrome and stroke-related symptoms were less likely to have difficulties of hospital acceptance [2]. The results of the present nationwide study demonstrate a positive association between chronological factors such as nights and weekends and difficulties of hospital acceptance, consistent with previous studies. In Japan, the number of medical facilities and staff that can treat emergency patients during nights, weekends or holidays is low [6]. In addition, few hospitals in Japan have physicians who work in shifts, and they usually continue to work for long hours regardless of the time of day [2]. Furthermore, this study underscored that a higher mean age of physicians in the area is associated with greater difficulty of hospital acceptance of patients suspected to have cerebrovascular diseases. It might be challenging for senior physicians to work longer hours to accept emergency patients without working in shifts. Therefore, centralization of medical resources such as physicians, other specialized staff, and equipment to prevent unbalanced seniority of physicians and longer work hours, using a dedicated emergency physician model of emergency care which allows emergency physicians to work in shifts, might be helpful to facilitate acceptance of the patients during nights or weekends [21].

There was a high degree of variation in difficulties associated with hospital acceptance across regions. In particular, Hokkaido-Tohoku, Chubu, Chugoku, Shikoku, and Kyushu-Okinawa were negatively associated with difficulties of hospital acceptance compared with Kanto and Kansai which include big cities like Yokohama, Osaka and Kyoto. This is consistent with a prior report which showed that serial rejection of a patient by several hospitals is more frequent in urban areas, where there are many hospitals, but also many patients [4]. To address this issue, a regionalized stroke system which includes establishing primary stroke centers that can deliver intravenous alteplase and better access to those centers is effective [22, 23]. In addition, a law which punishes hospitals if a patient is declined entry to an emergency department for screening and stabilization might be useful, similar to the Emergency Medical Treatment and Active Labor Act (EMTALA) in the United States [24].

Limitations

The present study has several acknowledged limitations. First, we were unable to obtain information regarding in-hospital outcomes and treatment of patients after arrival at the hospital. Therefore, we did not classify cerebrovascular diseases as acute ischemic stroke or cerebral hemorrhage. Second, the actual severity of a patient’s condition might not be reflected by the severity judged at the time when the ambulance was called, because further assessment of the patient’s condition was conducted upon arrival at the hospital. Some patients might deteriorate during transport and be assessed as critically ill on arrival at the hospital. Third, we could not analyze information regarding consciousness, such as Glasgow Coma Scale or Japan Coma Scale because the data were not available. Fourth, we did not consider national holidays, which were included as weekdays. Finally, despite adjusting for potential covariates, we did not exclude other possible residual confounding factors that might affect difficulties associated with hospital acceptance of patients suspected to have cerebrovascular diseases, such as the area’s hospital bed capacity or occupancy rates.