Impact of the active job openings-to-applicants ratio on the number of ambulance dispatches in Japan, 2003-2021: a longitudinal ecological study Authors: Yohei Kamikawa Published: 2024 ABSTRACT ABSTRACTObjectiveTo examine the contribution of the active job openings-to-applicants ratio, a macroeconomic indicator, to the number of ambulance dispatches.DesignLongitudinal ecological study.SettingJapan, between January 2003 and December 2021.ParticipantsAll ambulance dispatches.Primary and secondary outcome measuresThe contribution of the active job openings-to-applicants ratio in a month, adjusted by the number of older people, mean temperature and total population, to the number of ambulance dispatches in that specific month was examined and the primary model was compared to the conventional model considering only the older population, mean temperature, and total population.ResultsThere were 108 724 969 ambulance dispatches during this period. The active job openings-to-applicants ratio was significantly associated with the number of ambulance dispatches (the increase rate of monthly ambulance dispatches for 1% rise in the active job openings-to-applicants ratio, 1.00082; 95% CI 1.00052 to 1.00112). Additionally, the primary model effectively demonstrated better fitness to the actual trend than the conventional model (the quasi-likelihood under the independence model criteria were −2 626 817 720 and −2 626 775 185, respectively).ConclusionsThe number of ambulance dispatches was correlated with the active job openings-to-applicants ratio. Macroeconomic perspectives may be needed to address the issue of increasing ambulance dispatches. STRENGTHS AND LIMITATIONS OF THIS STUDY This study investigated the influence of macroeconomic indicators on the number of ambulance dispatches. This study investigated the influence of macroeconomic indicators on the number of ambulance dispatches. The analysis included all ambulance dispatches in Japan between 2003 and 2021. The analysis included all ambulance dispatches in Japan between 2003 and 2021. The exclusion of covariates beyond the older population, mean temperature and total population may limit the validity of this study. The exclusion of covariates beyond the older population, mean temperature and total population may limit the validity of this study. Focusing solely on data from Japan and limiting the analysis to two macroeconomic indicators—the active job openings-to-applicants ratio and the unemployment rate—may restrict the generalisability of the study’s findings. Focusing solely on data from Japan and limiting the analysis to two macroeconomic indicators—the active job openings-to-applicants ratio and the unemployment rate—may restrict the generalisability of the study’s findings. Introduction Background The annual increase in the number of ambulance dispatches has become a common source of concern in all developed countries.1 This has caused various problems, such as an excessive emergency medical team workload and overcrowding in emergency departments, resulting in higher medical costs, depletion of medical resources and a decline in the quality of medical care, including delays in response and an increase in mortality rates.24 Prior studies have examined factors that contribute to this increase in the number of ambulance dispatches, finding that, typically, an increase in the older population and variations in temperature lead to an increase in the number of ambulance dispatches.510 In Japan, ambulance services are provided free of charge to all residents. The prehospital emergency medical system is managed by local fire departments. As of 2021, there were 724 fire department headquarters, 5302 teams and 65 181 crew members.11 The older population in Japan, aged 65 and over, increased significantly from 23 810 000 (18.7% of the total population) in January 2003 to 36 227 000 (28.9% of the total population) in December 2021,12 and the monthly mean temperature in Tokyo from 2003 to 2021 consistently fluctuated in a cyclical pattern, ranging from around 27°C in the peak of summer in August to around 5°C in the depths of winter in January.13 Therefore, based on the assumption that the increase in the older population and variations in temperature are the main factors contributing to the rise in ambulance dispatches, the number of dispatches in Japan would be expected to increase annually, with some seasonality, as the older population steadily grows each year. However, there was a significant decrease in the number of dispatches between 2006 and 2009, a phenomenon at odds with the expected trend. Notably, the period from 2006 to 2009 coincided with a severe economic recession in Japan, and the trend of the active job openings-to-applicants ratio (AJOAR), a macroeconomic indicator, exhibited a similar trend to the number of ambulance dispatches in the same period and afterwards14 (online supplemental figure 1). The AJOAR is calculated by dividing the number of job openings by the number of job seekers; however, new graduates are excluded, while part-time job seekers are included in the count of job seekers.15 Among macroeconomic indicators, such as the unemployment rate, consumer price index and stock prices, the AJOAR is recognised as one of the best coincident indices for representing the current macroeconomic situation.16 Moreover, an economic study reported that temporary economic growth harmed residents’ health between 1972 and 1991 in the USA.17 Macroeconomic status may be associated with the number of ambulance dispatches, however, no research has investigated this relationship. While several studies have examined the relationship between the number of ambulance dispatches and several microeconomic indicators, such as individual income and type of insurance coverage,5 7 8 18 they cannot explain the nationwide trend. Objectives and impact This study aimed to examine whether the AJOAR correlates with the number of ambulance dispatches in Japan by comparing a conventional model that accounts for only the older population, mean temperature and total population with a primary model that additionally considers the AJOAR. The association between the unemployment rate and ambulance dispatches was also examined. While this study examines two macroeconomic indicators, if these indicators are found to be linked to trends in ambulance dispatches, it could offer an unconventional perspective that addressing macroeconomic conditions could help reduce ambulance dispatches. Methods Study design and setting This longitudinal ecological study considered all ambulance dispatches that occurred in Japan between 1 January 2003 and 31 December 2021. The study protocol was approved by the Research Ethics Committee of University of Fukui (number 20200095). The requirement for written informed consent was waived because of the retrospective observational nature of the study and the inclusion of fully anonymised data from government offices. The study adhered to the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) Statement .19 Patient and public involvement None. Procedures This study compared two models to evaluate the impact of AJOAR on the number of ambulance dispatches. The primary model included AJOAR, the older population and mean temperature, with the total population as an offset variable. In contrast, the conventional model excluded AJOAR. The data used for this study were those published by the Japanese government as of 1 November 2023. For this study, data from the Current Status of Emergency and Rescue Services of the Fire and Disaster Management Agency were used for the number of ambulance dispatches,11 along with data from the Population Estimates of the Statistics Bureau of the Ministry of Internal Affairs and Communications for the estimated number of older people and total population,12 historical weather data from the Japan Meteorological Agency for the mean temperatures,13 data from the Employment Referrals for General Workers of the Ministry of Health, Labour and Welfare for the AJOARs14 and data from the Labour Force Survey of the Statistics Bureau of the Ministry of Internal Affairs and Communications for the unemployment rate.20 Since these data were commonly compiled monthly, monthly data were collected and used in the analysis. Additionally, since the monthly data of AJOAR were publicly available only for the period from 1 January 2003 to 31 December 2021, this period was covered in this study. The data set prepared for this study is also available from the Open Science Framework database.21 Measurements The number of ambulance dispatches in Japan was calculated from the combined dispatch numbers of ambulance cars, trucks, helicopters and other vehicles. Older people were defined as those aged 65 years and over according to previous studies that demonstrated a significant increase in ambulance utilisation by patients aged 65 years and over.6 8 For the mean temperatures, data for Tokyo were used because it is located approximately in the centre of Japan and is the most populous urban area in the country. Additionally, a study has shown a U-shaped relationship between temperature and ambulance use, where ambulance use is lowest when the temperature is 22.5°C; it has also shown that the rate of change in the number of ambulances used varies depending on whether temperatures are higher or lower than 22.5°C.10 Since it is impossible to handle a U-shaped relationship with linear regression, it is necessary to separate the data into two parts, one with an upswing effect and one with a downswing effect. Therefore, an original data set was constructed in which data were set as 0 if the mean temperature was below 22.5°C and as the mean temperature minus 22.5 if the mean temperature was above 22.5°C, which was named the ‘high-temperature effect’. Conversely, a data set in which data were set as 0 if the mean temperature was above 22.5°C and as the mean temperature minus 22.5 if the mean temperature was below 22.5°C, was named the ‘low-temperature effect’. Models for comparison Two regression models were compared to examine the statistical contribution of the AJOAR to the outcome of monthly ambulance dispatches. The primary model consisted of the monthly older population, high-temperature effect, low-temperature effect and AJOAR, based on the assumption that AJOAR in a month has an immediate impact on the number of ambulance dispatches in that specific month. The monthly total population was also included as an offset variable. On the other hand, the conventional model for comparison was constructed with only the older population, high-temperature effect and low-temperature effect, including the monthly total population as an offset variable. Statistical analyses Continuous variables are reported as medians and IQRs. As the primary analysis, a generalised estimating equation (GEE) analysis with a first-order autoregressive covariance structure and the assumption of a log link Poisson distribution was performed on the two models. Robust variance estimators were used for the final consequences. The regression coefficient estimates for the variables, along with their 95% CIs, were calculated and considered significant when the p-values were below 0.05. Additionally, the increase rate of monthly ambulance dispatches for 1% rise in the AJOAR was derived from the regression coefficient estimate to evaluate its practical impact. Goodness-of-Fit tests using Pearson’s χ2 tests were also conducted, and the regression estimates were regarded as fitting the actual values when the p-values were 0.05 or higher. Then, the quasi-likelihood under the independence model criteria (QICs) were compared to determine which model best explained the trend in the number of ambulance dispatches. A model is better than the other when the QIC is lower.22 In the secondary analyses, the graphs of the models were compared with the actual ambulance dispatch trend visually to confirm the outcome of the primary analyses. The research hypotheses were that AJOAR would significantly contribute to GEE and that the primary model would exhibit better fitness than the conventional model, which meant that AJOAR was associated with the trend of ambulance dispatches. For the sensitivity analysis, a GEE analysis was conducted, in which AJOAR was replaced with the unemployment rate, another macroeconomic index, in the primary model. A graph of the model was also generated. The sample size of observation points was estimated to be at least 40 months, 10 times the number of explanatory variables (AJOAR, old population, high-temperature effect and low-temperature effect) in the primary regression model. As the period covered was 228 months, the sample size was found to be sufficient. All the statistical analyses were performed using R software (V.4.4.1; R Foundation, Vienna, Austria). Results Characteristics of study data A total of 108 724 969 ambulance dispatches occurred in the 228 month period from 1 January 2003 to 31 December 2021. Summarising patients by age and sex was not possible because they were completely anonymised and thus could not be aggregated. The median values during the period were as follows: the monthly number of dispatches—476 424 (IQR 432 168 to 511 069); AJOAR—1.04 (IQR 0.78 to 1.21); older population—30 433 000 (IQR 27 447 000 to 34 854 500); mean temperature—16.9°C (IQR 9.2 to 23.2); total population—127 588 500 (IQR 126 947 500 to 127 842 750) and the unemployment rate—4.0% (IQR 3.0 to 4.5). There were no exclusions or missing data for any variable. Main results The GEE analysis for the primary model revealed that AJOAR significantly contributed to the number of ambulance dispatches, as shown in table 1. The regression coefficient estimate for AJOAR was 0.082 (95% CI 0.052 to 0.112), whereas it would be 0.00082 (95% CI 0.00052 to 0.00112) if AJOAR×100 was used instead of simply AJOAR. Based on this finding, the rate of increase in monthly ambulance dispatches was calculated to be 1.00082 (95% CI 1.00052 to 1.00112) for 1% increase in AJOAR, since e0.00082 equals 1.00082. For instance, if AJOAR were to increase by 1% in a month from the median number of dispatches of 476 424, it would be 476 815 (95% CI 476 671 to 476 957); that is, there would be an increase of 391 (95% CI 247 to 533) in the next month. Goodness-of-Fit tests showed no evidence of poor fit for either the primary model (χ2/df=1.00, p=0.24) or the conventional model (χ2/df=1.00, p=0.24). The QIC of the primary model was −2 626 817 720, which was better than that of the conventional model (−2 626 775 185). Table 1Results of the generalised estimating equation analyses for the number of ambulance dispatchesExplanatory variableRegression coefficient estimate (95% CI)QICPrimary model UnadjustedAJOAR0.221 (0.188 to 0.255)*−2 626 306 437 AdjustedAJOAR0.082 (0.052 to 0.112)*−2 626 817 720Older population, 106 people0.019 (0.016 to 0.021)*High-temperature effect, °C0.028 (0.025 to 0.032)*Low-temperature effect, °C−0.008 (−0.009 to −0.006)*Conventional model−2 626 775 185Older population, 106 people0.023 (0.021 to 0.026)*High-temperature effect, °C0.028 (0.024 to 0.032)*Low-temperature effect, °C−0.008 (−0.009 to −0.006)**P <.05.AJOARactive job openings-to-applicants ratioQIC, quasi-likelihood under the independence model criterion Results of the generalised estimating equation analyses for the number of ambulance dispatches P <.05. AJOARactive job openings-to-applicants ratioQIC, quasi-likelihood under the independence model criterion According to the secondary analyses, the primary model effectively reproduced the characteristic trend in the number of ambulance dispatches, with an increase occurring from 2003, a decrease occurring approximately 2008, a subsequent increase occurring beginning in 2010 and a significant decrease occurring from 2020 to 2021 (figure 1). In contrast, the conventional model excluding AJOAR yielded a steady annual increase with some seasonality against the actual trend of dispatches (figure 2). The number of ambulance dispatches and the estimates from the primary model considering the active job openings-to-applicants ratio. The dashed line indicates the actual number of ambulance dispatches, and the solid line indicates the number of ambulance dispatches estimated by a generalised estimating equation analysis considering the older population, mean temperature, total population and the active job openings-to-applicants ratio. The number of ambulance dispatches and the estimates from the conventional model that do not consider the active job openings-to-applicants ratio. The dashed line indicates the actual number of ambulance dispatches, and the solid line indicates the number of ambulance dispatches estimated by a generalised estimating equation analysis considering only the older population, mean temperature and total population. In the sensitivity analysis, the GEE results showed that the unemployment rate also had a significant impact on the number of ambulance dispatches (online supplemental table 1). The estimated regression coefficient for the unemployment rate was −0.026 (95% CI −0.041 to −0.012), with a QIC of −2 626 817 720. The Goodness-of-Fit test indicated no issues with model fit (χ2/df=1.00, p=0.24). The graph of the model also reproduced the characteristic trend of dispatches (online supplemental figure 2). Discussion Findings This study demonstrated the statistical contribution of the monthly AJOAR to the number of ambulance dispatches in the same month in Japan and the better fitness of the primary model considering the AJOAR to the actual trend compared with the conventional model that considered only the older population, mean temperature and total population. These results indicated a correlation between the number of ambulance dispatches and AJOARs. Interpretation Although the increase rate of monthly ambulance dispatches for 1% rise in the AJOAR ratio appears small at 1.00082, it has a significant impact in practice. Specifically, the AJOAR increased by 17% between April 2015 and April 2016, indicating that the number of ambulance dispatches likely increased from 476 720 to an estimated 483 409 (calculated as 476 720×1.0008217). Given that the actual number of dispatches in April 2016 was 487 095, it is estimated that approximately 64.5% of this increase can be attributed to the rise in AJOAR. Similarly, the AJOAR decreased by 47% between April 2008 and April 2009, suggesting a decline in ambulance dispatches from 407 094 to an estimated 391 709 (calculated as 407 094×1.00082−47). Since the actual number of dispatches in April 2009 was 404 562, this decline can be primarily attributed to the decrease in AJOAR. Furthermore, given that other key factors influencing the number of ambulance dispatches, such as the ageing population and seasonal temperature changes, cannot be swiftly altered by human intervention, the findings of this study are significant. This is because the AJOAR can be regulated through monetary and fiscal policies. Additionally, the sensitivity analysis showed that the unemployment rate also had a significant impact on the number of ambulance dispatches. While both AJOAR and the unemployment rate are reliable indicators of macroeconomic conditions, neither is a fully comprehensive measure due to its respective limitations.23 24 Nevertheless, the fact that these different indicators produced similar results suggests that macroeconomic conditions can have a significant influence on the number of dispatches. No prior study has successfully explained the decline in the number of ambulance dispatches in Japan between 2006 and 2009. A study conducted in Yokohama, a city in Japan, assumed that the decrease in the number of ambulance transports that occurred in the city in 2007 was due to a campaign to prevent inappropriate emergency calls conducted during that time.25 However, it is now known that the decrease in the number of ambulance transports in 2007 was a nationwide phenomenon in Japan. As it is unreasonable to assume that a campaign in one city could have affected the entire country, it is doubtful that the campaign activities had a real effect, as referred to in the study. However, the present study clarified that two macroeconomic indicators, AJOAR and the unemployment rate, were associated with the decrease in the number of ambulance dispatches. Indeed, AJOAR decreased from 2006 to 2009, coinciding with a period of declining dispatches14 (online supplemental figure 1). Furthermore, it was unintentionally shown that the macroeconomic indicators were also related to the decrease in the number of dispatches during the COVID-19 pandemic from 2020 to 2021. Dicker et al reported that the number of ambulance dispatches decreased with the onset of the COVID-19 pandemic. The authors suggested that the decrease might have been because of reduced exposure to risks such as trauma, infectious diseases and air pollutants due to reduced opportunities to go outside caused by COVID-19-related lockdowns and avoidance behaviour.26 Since such a decrease in human activity can certainly lead to the atrophy of economic activity in a society,27 the result of the present study was consistent with existing research. While it is common to consider specific social events, such as the COVID-19 pandemic, when explaining the trend in ambulance dispatches, the current study suggests the possibility that certain macroeconomic indicators can account for the nationwide trend of ambulance dispatches unitarily without focusing on particular events, as the impact of such events ultimately manifests in macroeconomic phenomena and indicators. This is the first study to show that macroeconomic indicators are associated with the number of ambulance dispatches. Several studies have examined the relationship between the number of emergency calls and microeconomic indicators, such as individual income and the type of insurance coverage.5 7 18 However, as far as focusing on the economic situation of individuals, it is not possible to derive general measures, as there are so many different reasons why individuals have high or low incomes that measures would have to be ad hoc and on a case-by-case basis. Therefore, one study reached the reluctant conclusion that relying on the ethical motivation of each individual to reduce unnecessary emergency calls was ultimately the only option.3 In this respect, the present study, which shows a novel link between the number of dispatches and macroeconomic indicators, is unconventional. This is because macroeconomic-related phenomena require societal measures, including economic and healthcare policies. This perspective has the potential to pave the way for more effective and realistic recommendations to address the issues of increasing ambulance dispatches. The finding of this study (the number of ambulance dispatches increases as macroeconomic indicators improve) may mean that the preference for health services increases as the economy grows, while this demand decreases as economic growth diminishes. Another hypothesis is that temporary economic growth worsens public health. Although it was not a study of emergency dispatches, one economic study reported by Ruhm revealed that for every 1% decrease in unemployment (ie, economic growth), the mortality rate for trauma and some diseases increased by about 0.5% in the USA. The author proposed the following explanations. First, people spend less time on improving their health during a booming economy, as they want to spend more time on labour. Second, overworking during this time negatively affects individuals’ health. Third, traffic accidents increase because business travel is more frequent.17 Another economic study conducted in Japan reported similar conclusions.28 Nevertheless, Ruhm also indicated that long-term economic growth possibly offsets the adverse events driven by temporary economic growth.17 In other words, an increase in ambulance dispatches during temporary economic growth may be a phenomenon limited in the long-term recession period, from which Japan suffered after the ‘economic bubble burst’ in the early 1990s. Either way, macroeconomic perspectives will be needed to address the issue of increasing ambulance dispatches in the future. However, it is impolitic to claim that economic growth should be curbed to reduce the number of ambulance dispatches citing the results of this study because the insistence ignores human instincts to seek a more affluent lifestyle every day. That is human nature, so no one has the right to stop economic growth. Based on the above considerations, certain measures should be taken to provide means other than calling for an ambulance to address growing health service demands in the future. For example, Patton et al reported that setting up a community emergency medicine outreach team that responds to low-acuity emergency calls successfully achieved a 68% non-conveyance rate.29 As found by a prior study, ambulances are often called inappropriately due to inaccurate knowledge of diseases.30 Thus, the development of smartphone applications that facilitate the identification of health-related questions and provide reliable information can also be a countermeasure to reduce the number of ambulance dispatches.31 Such local administrative enhancements and technological innovations may be effective and realistic solutions. Other solutions should also be explored in the future. Limitations This study has several limitations. First, it did not account for factors beyond the older population, mean temperature, total population, unemployment rate and AJOAR. While other variables contributing to the rise in ambulance dispatches have been identified, they fall into two categories: individual and regional. In the individual category, factors such as a sense of isolation,6 7 32 low income,5 7 8 32 limited educational background,32 a lack of knowledge of health and medical resources7 8 32 and situations that lead to risk overestimation (eg, distress, heavy responsibility and excessive instructions from others)8 32 have been reported. Regional factors encompass air pollutants,33 exciting events,34 population density,35 36 limited hospital accessibility5832 and excessive hospital accessibility.5 6 37 Using these factors to describe nationwide trends poses difficulties because of the wide variance among individuals or regions, and it is impractical to plan comprehensive measures. In contrast, macroeconomic indicators represent nationwide trends and can be manipulated by governments. This study did not take those individual and regional factors into consideration for these reasons. However, in addition to the factors mentioned above, there may be other unknown factors that significantly impact the number of ambulance dispatches. This could affect the comprehensiveness and validity of the study’s results. Despite these limitations, this study is meaningful because the primary model, though simple in structure, successfully captured the complex trends in the number of ambulance dispatches in Japan for the first time. Second, patient characteristics such as age and sex could not be taken into consideration because of fully anonymised data. Third, regional differences in mean temperature were not considered as the temperatures were represented by the measurements from Tokyo. Fourth, it was not investigated for what kind of patients (eg, trauma or not, with low or high income, emergent or non-emergent) ambulance dispatches increased or decreased with macroeconomic change. Fifth, lag effects and seasonality were not considered in the analyses. Finally, the generalisability of the findings should be approached with caution, as the results of this study were based solely on data from Japan and examined only two macroeconomic indicators. Future studies are required to address these problems. Conclusions The number of ambulance dispatches was correlated with AJOAR, and temporary economic growth increased the number of ambulance dispatches. Macroeconomic perspectives may be needed to address the issue of increasing ambulance dispatches. Specific measures should be explored in the future. supplementary material