Association between ambient temperature and genitourinary emergency ambulance dispatches in Japan: A nationwide case-crossover study

Background: Although the effects of temperature on genitourinary morbidity and mortality have been investigated in several countries, it remains largely unexplored in Japan. We investigated the association between ambient temperature and genitourinary emergency ambulance dispatches (EADs) in Japan and the modifying roles of sex, age, and illness severity. Methods: We conducted a time-stratified case-crossover study with conditional quasi-Poisson regression to estimate the association between mean temperature and genitourinary EADs in all prefectures of Japan between 2015 and 2019. A mixed-effects meta-analysis was used to pool the association at the country level. Subgroup analyses were performed to explore differences in associations stratified by sex, age, and illness severity. Results: We found an increased risk of genitourinary EAD associated with higher temperatures. The cumulative relative risk (RR) at the 99th temperature percentile compared with that at the 1st percentile was 1.74 (95% confidence interval (CI) = [1.60, 1.89]). We observed higher heat-related RRs in males (RR = 1.89; 95% CI = [1.73, 2.07]) than females (RR = 1.56; 95% CI = [1.37, 1.76]), and in the younger (RR = 2.13; 95% CI = [1.86, 2.45]) than elderly (RR = 1.39; 95% CI = [1.22, 1.58]). We found a significant association for those with mild or moderate cases (RR = 1.77; 95% CI = [1.62, 1.93]), but not for severe or life-threatening cases (RR = 1.20; 95% CI = [0.80, 1.82]). Conclusion: Our study revealed heat effects on genitourinary EADs in Japan. Men, youth, and mild-moderate illnesses were particularly vulnerable subgroups. These findings underscore the need for preventative measures aimed at mitigating the impact of temperature on genitourinary emergencies.


Introduction
Climate change is a public health crisis that has been affecting the global population. 1 Recent studies have shown increased interest in the effects of ambient temperature on morbidity and mortality from various diseases. 2,35][6][7] However, much less is known about the impact of daily temperature on other conditions such as kidney diseases despite the annualized rate of change of kidney dysfunction significantly increasing by 0.35% over the years 1990-2019. 8lthough empirical evidence has indicated possible associations between nonoptimal temperature and kidney diseases, the findings have been inconsistent.Some studies have reported an increased risk of acute kidney injury (AKI), [9][10][11] hemodialysis initiation, 12 or mortality from renal diseases [13][14][15] associated with high temperatures, while others have found no such association. 16So far, relatively few studies have examined the impacts of both hot and cold temperatures, and most have focused solely on the influence of heat. 1,9,13,15Furthermore, previous studies have tended to focus on specific genitourinary conditions, with few investigating all genitourinary conditions simultaneously.
Moreover, limited studies have been conducted to further identify particularly susceptible subpopulations that are at higher risk of nonoptimal temperature-related genitourinary conditions.These points emphasize the importance of conducting a comprehensive analysis of nonoptimal temperature-related genitourinary morbidity and mortality.Previous nationwide studies in Japan have examined the associations between daily temperature and renal cause-specific mortality using different temperature indexes and have mainly focused on the relationships between temperature and genitourinary mortality. 17,18To our knowledge, our study is the first globally to investigate the effects of daily mean temperature on genitourinary morbidity using national emergency ambulance dispatches (EADs) registries, a sensitive acute health indicator.
In the present study, we investigated the associations between ambient temperature and genitourinary EADs over the full temperature spectrum for the 47 prefectures in Japan from 2015 to 2019.Subgroup analyses were further conducted to examine the modifying effects of sex, age, and disease severity.

Data collection
We used EAD data for all prefectures in Japan from the Fire and Disaster Management Agency of the Ministry of Internal Affairs and Communication. 19Daily EAD data for genitourinary causes were collected for the 47 prefectures of Japan from the period of 1 January 2015 to 31 December 2019 (except for Tokyo prefecture; data were available from 1 January 2016 to 31 December 2019).All genitourinary causes of EADs were coded according to the International Classification of Diseases, tenth revision (ICD-10: codes N00-N99).Each genitourinary EAD case recorded information regarding sex, age group (17 years and younger, 18-64 years old, and 65 years and older), and severity of illness. 20The illness severity was classified into six categories (mild, moderate, severe, life-threatening, death, or other) and was documented by the physician who performed the initial assessment of the patient upon their arrival at the hospital. 21e collected daily mean temperature and relative humidity data for the study period from representative weather stations maintained by the Japan Meteorological Agency in each of the 47 prefectures.Daily mean dew point temperatures were calculated from the mean temperature and relative humidity. 22,23

Statistical analysis
We conducted a two-stage meta-analysis to investigate the association between mean temperature and genitourinary EADs at both the country and prefecture levels.All modeling and statistical computations were performed using R statistical software (Version 4.2.1;R Development Core Team) with the R packages dlnm, 24 mixmeta. 25n the first stage, we used a time-stratified case-crossover study design with a conditional quasi-Poisson regression model to estimate the association between mean temperature and genitourinary EADs in each prefecture.The seasonality and longterm trend were controlled by a time stratum in the model that matched case-control days by selecting the same day of the week within the same month and year. 26To describe the nonlinear and delayed associations, the model incorporated a distributed lag nonlinear model using a cross-basis function for temperature and lag. 27Specifically, the exposure-response association was modeled with a quadratic B-spline with two internal knots placed at the 33rd and 66th percentiles of the prefecture-specific temperature distribution.The lag-response association was modeled with a natural cubic spline with an intercept and two internal knots equally spaced in the log scale.The lag period was extended to 14 days.The number and placement of knots were determined based on the quasi-Akaike information criterion.We also included a natural cubic spline of a 3-day moving average dew point temperature with three degrees of freedom and a binary variable to control for the public holidays.We adjusted for dew point temperature instead of relative humidity as a time-varying confounding factor in the model because dew point temperature is an absolute measure of humidity, thereby making it a reliable measure unlike relative humidity, which depends strongly on temperature and can therefore cause increased diurnal variation. 28,29Statistical analysis results without adjusting for dew point temperature are also generated for comparison.
In the second stage, we applied a mixed-effects meta-analysis to pool the prefecture-specific coefficients obtained from the first stage to obtain the nationwide estimates.We included prefecture-specific mean temperature and temperature range as meta-predictors to address between-prefecture variability that can be explained by differences in temperature distributions.Residual heterogeneity was tested and then quantified by the multivariate extension of the Cochran Q-test and I-square statistic. 30,31This meta-regression model was subsequently utilized to derive the best linear unbiased prediction (BLUP) of the cumulative exposure-response associations across all prefectures.The analytical strategy using BLUP enables prefectures with limited daily genitourinary EAD counts or short series, potentially associated with imprecise estimates, to leverage information from larger populations with similar characteristics. 30The prefecture and pooled curves were then scaled by reentering them on the minimum risk temperature percentile (MMTP) of the temperature distribution of each prefecture and at the country level, defined as the minimum of the exposure-response curve.The search for the minimum risk temperature was limited between the 1st-99th percentile range. 32We also summarized the results by computing the cold-and heat-related cumulative risk ratios (RRs) at the 1st and 99th percentiles from these curves, using the MMTP as the reference.Country-pooled lag-response relationships for heat and cold were also derived from the recentered estimates using prefecture-specific MMTPs derived from the BLUPs. 31ubgroup analyses were performed stratified by sex, age (younger, <65 years; older, ≥65 years), and two illness severity groups (mild to moderate, and severe to life-threatening) using the same two-staged modeling explained above. 20o evaluate the robustness of our findings, we conducted several sensitivity analyses.Initially, we altered the number of knots from two, located at the 33rd and 66th percentiles, to three knots placed at the 25th, 50th, and 75th percentiles and the 50th, 75th, and 90th percentiles for modeling the exposureresponse relationship.For the lag-response relationship, we changed the number of knots from two to three equally spaced knots and the lag period from 14 to 7 days.This study is exempt from ethics approval since the analysis was based on anonymized data.

Results
Table 1 shows the summary statistics for genitourinary EAD at the country level.A total of 639,994 genitourinary EADs were included in the study.The daily mean number of genitourinary EADs was 7.5 (Table 1).Of all genitourinary EADs recorded for the study period, over half of the cases (57.9%) were males, and approximately half of the cases (50.7%) were from the older age category.The majority (96.3%) were from the mild to moderate severity category.Table 2 shows the distribution of prefecture-level meteorological variables.The mean ambient temperature during the study period was 16.0 °C and varied considerably from 9.6 °C to 23.8 °C.The mean dew point at the country level was 10.1 °C and ranged from 3.6 °C to 18.7 °C.(Supplemental Figure 1; http://links.lww.com/EE/A262displays a distinct and consistent seasonality for total genitourinary cases and daily mean temperature.We found that a higher number of genitourinary cases occurred at high temperatures.The summary statistics of genitourinary causes of EADs and weather variables for each prefecture are shown in Supplemental Tables 1 and 2; http://links.lww.com/EE/A262).
Figure 1 shows the pooled exposure-response curve between mean temperature and genitourinary cause of EADs.Results from the analysis of heterogeneity are illustrated in Supplemental Table 3; http://links.lww.com/EE/A262, with a comparison of statistics from the simple multivariate random-effects metaanalysis and meta-regressions with a single or both metapredictors.The pooled estimates from the multivariate meta-regression model indicated low heterogeneity at the country level (I-square statistic = 10.0%;multivariate Cochran Q-test for residual heterogeneity: Q = 195.5195,P = 0.15).The association between temperature and all genitourinary causes of EADs was nonlinear, and the minimum risk temperature was identified at the 1st percentile of temperature.Heat-related risks of genitourinary causes of EADs increased as mean temperature increased.A significant heat risk was observed with a RR of 1.74 (95% confidence interval [CI] = 1.60, 1.89) at the 99th percentile compared with the MMTP (Table 3).The pooled lagresponse curve for all genitourinary causes of EADs at the country level showed a delayed effect for heat and lasted for up to 5 days (Supplemental Figure 2; http://links.lww.com/EE/A262).
A slight difference for heat-related RRs was observed in models with and without adjusting for dew point temperature, although with consistent patterns (Supplemental Table 5; http://links.lww.com/EE/A262).Sensitivity analyses showed that the estimates were robust against alternative specification for the knots of temperature and lag period (Supplemental Table 6; http://links.lww.com/EE/A262).The prefecture-specific exposure-response associations obtained in the first stage (without BLUP) were also analyzed, and the association estimates and exposure-response curves changed only slightly (Supplemental Figure 3; http://links.lww.com/EE/A262).

Discussion
While there have been previous studies that have investigated the associations of ambient temperature with numerous health outcomes such as hospital admission rates, hemodialysis initiation, and mortality from AKI or chronic kidney disease (CKD) to our knowledge, there have been no studies that have examined the associations using EADs.The goal of this study was to assess the associations between ambient temperature and EADs due to genitourinary causes in Japan.Additionally, the study investigated these associations among different subgroup populations based on sex, age, and illness severity.Overall, we found that there are heat-related effects on EADs due to genitourinary conditions at both the country level and the prefecture-level.In addition, we found evidence of risk increase associated with heat exposure among the male, younger (<65 years old) patients, and those with mild or moderate cases of genitourinary conditions.
4][35] However, our results differ from a study in Thailand that found no associations between heatwaves and renal failure mortality 36 and from studies that have shown that the risk of hemodialysis initiation or mortality due to ESRD was higher in the cold, winter seasons. 37,38Additionally, our results are inconsistent with another study that concluded that winter was an independent risk factor for all-cause in-hospital mortality for AKI 39 and a recent time-series study that looked at 204 countries and found that the overall CKD death burden associated with lower temperatures was higher than that associated with warmer temperatures. 40ur subgroup analysis results have some consistencies and inconsistencies with previous studies as well-for example, our findings that men and younger populations are at a higher risk of genitourinary conditions were not quite the same as a previous study in Queensland, Australia that concluded that men and elderly are more vulnerable to heat effects causing genitourinary diseases. 41While the reason behind this difference remains unclear, it may suggest that the elderly population in Japan has better protection against heat, whether it be through preventative healthcare measures, genetics, or living environments; alternatively, this difference may indicate that the younger population in Japan has a different mobility pattern compared with that in Australia that puts them at an increased risk of exposure.Furthermore, our results differ from another study in South Korea that found that cold spells are associated with hospital admission and mortality due to AKI, especially in the elderly population. 42However, our findings share similar findings with another study in Taiwan that concluded that the risk of hospitalization due to CKD in warmer temperatures was higher in the male population. 43hile the exact mechanism for the relationship between ambient temperature and genitourinary EADs is unknown, a possible explanation is that heat in general adds stress to the kidneys due to causes such as dehydration.In hot temperatures, significant stress is added to the renal system, particularly because dehydration leads to lowered blood pressure, which would in turn lead to decreased kidney function and increase chances of kidney injury/damage. 1,44This explanation would make sense in the setting of our findings not revealing statistically significant cold effects.The percentiles of the temperature for centering and RR calculation. b The percentile of minimum risk temperature, identified between the 1st and 99th percentiles of temperature at the country level. c The cold and heat risks are the RRs at the 1st (cold) and 99th (heat) percentiles of temperature.CI indicates confidence interval; MMTP, minimum risk temperature percentile; RR, risk ratio.
To our knowledge, our study is the first to investigate the association between ambient temperature and genitourinary EADs in Japan.One of the strengths of the current study is the use of emergency data, a sensitive indicator that better reflects the acute and nonfatal effects of nonoptimal temperatures compared with other health indicators such as mortality and hospitalization.Another strength of the study is the 5-year time scale and well-covered spatial resolution for all 47 prefectures, allowing for a large sample size and thus the strong statistical power to examine the association at both prefecture and country levels.Third, the current study investigated the entire temperature spectrum within the context, including both heat and cold effects.Moreover, the application of the time-stratified case-crossover design with a distributed lag nonlinear model allows us to assess the short-term and delayed impact of temperature on genitourinary morbidity precisely.In the meantime, our results demonstrated the need to pay additional attention to adjustment for other time-varying confounders such as absolute humidity when estimating heat-related effects, as the time-stratified case-crossover design might not fully capture the time variations in the humidity exposure.This contrasts with most studies that have focused solely on either heat or cold.Last, we also conducted subgroup analyses to further identify susceptible subpopulations and the potential role of age, sex, and illness severity in explaining the differences.
Some limitations of our study include the fact that our study only looked at the 47 prefectures of Japan, therefore the results may not be generalizable to the populations of other countries.Second, there are potential misclassifications of the meteorological and EAD data that could bias the estimates given that the exposure level was measured from representative weather stations in cities, and the primary diagnosis of each EAD case was recorded with a single ICD code.Although the initial assessments performed by the physicians are standardized across Japan, there remains possibility of over-or underestimation of temperature-related EAD risks given that the diagnoses and disease severities were determined from a preliminary exam by the first responders, rather than them being a definitive diagnosis.Additionally, due to the low proportion of severe and worse genitourinary conditions of EAD (3.7%), our results should be interpreted with caution when comparing with other studies assessing more fatal conditions.Further research on specific disease categories is also necessary to properly understand how heat affects genitourinary health.Moreover, our current research investigated all genitourinary conditions grouped together and did not conduct any subgroup analyses based on subtypes of genitourinary conditions; future studies may benefit from such subgroup analyses, but further data collection would be needed to obtain a sufficient sample size.Finally, our study only conducted subgroup analyses based on sex, age, and severity category, but future studies could benefit from the inclusion of other variables, such as race-however, given that the ethnic composition of the population of Japan is 99% Japanese and 0.6% other, 45 race was not included in our current study.S3; http://links.lww.com/EE/A262 for the corresponding numeric data.
In conclusion, our study found that the risk of genitourinary EADs increases with higher temperatures, with particularly notable effects on the young, male population who have mild to moderate genitourinary cases.These findings highlight the importance of heat adaptation strategies in preventing the occurrence of genitourinary EADs, especially in vulnerable populations.
25th and 75th are percentiles.a The summary statistics were calculated based on the prefecture-level averages of each meteorological variable.SD indicates standard deviation.

Figure 1 .
Figure 1.A, Exposure-response curve with 95% empirical CI (shaded gray) and (B) Lag-response associations for emergency ambulance dispatch (EAD) over the extended lag period of up to 14 days with 95% CI (shaded gray) in Japan, estimated by using a conditional Poisson model adjusting for seasonality, longterm time trend, the day of week, holiday, and dew point.Vertical lines indicate the location for the minimum risk temperature (1st) and RR calculations for cold (1st) or heat (99th).CI indicates confidence interval; RR, relative risk.

Figure 2 .
Figure 2. Spatial map of prefecture-specific relative risks (RRs) of heat for all genitourinary emergency ambulance dispatch (EAD).MMT(P)s varied by prefecture, and the locations for RR calculations were the 99th percentiles of each prefecture's mean temperature.See TableS3; http://links.lww.com/EE/A262 for the corresponding numeric data.

Figure 3 .
Figure 3. Pooled cumulative RRs with 95% CIs (vertical bars) for heat-related genitourinary emergency ambulance dispatch (EAD) in subgroup analyses by sex, age (younger, <65 years of age; older, ≥65 years of age), and severity groups, estimated by using a conditional Poisson model adjusting for seasonality, long-term time trend, the day of week, holiday, and dew point.The heat risks are the RRs at the 99th percentile of temperature, compared with the respective minimum risk temperature percentile.CI indicates confidence interval; RR, relative risk.

Table 1 .
Descriptive statistics for genitourinary emergency ambulance dispatches (EADs) from 2015 to 2019 in Japan a

Table 2 .
Summary statistics for meteorological variables from 2015 to 2019 at country level

Table 3 .
Pooled RRs of genitourinary emergency ambulance dispatch (EAD) in JapanPooled cumulative RRs with 95% CIs for genitourinary EADs for overall and subpopulations, estimated by using a conditional Poisson model adjusting for seasonality, long-term time trend, day of week, holiday, and dew point. a