Skip to main content
NCBI home page
Canadian Journal of Public Health = Revue Canadienne de Santé Publique logoLink to Canadian Journal of Public Health = Revue Canadienne de Santé Publique
. 2022 Aug 18;113(5):686–697. doi: 10.17269/s41997-022-00665-1

Impact of Ontario’s Harmonized Heat Warning and Information System on emergency department visits for heat-related illness in Ontario, Canada: a population-based time series analysis

Kristin K Clemens 1,2,3,✉, Alexandra M Ouédraogo 2, Britney Le 2, James Voogt 4, Melissa MacDonald 5, Rebecca Stranberg 6, Justin W Yan 7, E Scott Krayenhoff 8, Jason Gilliland 4,9, Cheryl Forchuk 10, Rafique Van Uum 11, Salimah Z Shariff 2
PMCID: PMC9481795  PMID: 35982292

Abstract

Intervention

Ontario’s Harmonized Heat Warning and Information System (HWIS) brings harmonized, regional heat warnings and standard heat-health messaging to provincial public health units prior to periods of extreme heat.

Research question

Was implementation of the harmonized HWIS in May 2016 associated with a reduction in emergency department (ED) visits for heat-related illness in urban locations across Ontario, Canada?

Methods

We conducted a population-based interrupted time series analysis from April 30 to September 30, 2012–2018, using administrative health and outdoor temperature data. We used autoregressive integrated moving average models to examine whether ED rates changed following implementation of the harmonized HWIS, adjusted for maximum daily temperature. We also examined whether effects differed in heat-vulnerable groups (≥65 years or <18 years, those with comorbidities, those with a recent history of homelessness), and by heat warning region.

Results

Over the study period, heat alerts became more frequent in urban areas (6 events triggered between 2013 and 2015 and 14 events between 2016 and 2018 in Toronto, for example). The mean rate of ED visits was 47.5 per 100,000 Ontarians (range 39.7–60.1) per 2-week study interval, with peaks from June to July each year. ED rates were particularly high in those with a recent history of homelessness (mean rate 337.0 per 100,000). Although rates appeared to decline following implementation of HWIS in some subpopulations, the change was not statistically significant at a population level (rate 0.04, 95% CI: −0.03 to 0.1, p=0.278).

Conclusion

In urban areas across Ontario, ED encounters for heat-related illness may have declined in some subpopulations following HWIS, but the change was not statistically significant. Efforts to continually improve HWIS processes are important given our changing Canadian climate.

Supplementary Information

The online version contains supplementary material available at 10.17269/s41997-022-00665-1.

Keywords: Extreme heat, Temperature, Population health, Environment, Public health

Introduction

Extreme heat can have significant negative effects on human health. High temperature has been linked with dehydration and heat stroke (where core temperatures rise to greater than 40°C and cause confusion/unconsciousness) (Simpson & Abelsohn, 2012). In some studies, heat exposure can cause cardiovascular events (from cardiac strain and reduced cardiac output) (Crandall & González-Alonso, 2010), acute kidney injury (from dehydration) and even death (Bobb et al., 2014; Chen et al., 2016; Lavigne et al., 2014). High temperature can be particularly consequential to patients who already live with comorbidities including respiratory disease and diabetes (Kenny et al., 2010).

Those who live in urban areas can also be particularly susceptible to heat (Li & Bou-Zeid, 2013). Compared with rural regions, cities are warmer in air and surface temperature due to the replacement of vegetation with impervious surfaces, an increase in three-dimensional surfaces, modified radiative and thermal properties of built surfaces and anthropogenic emissions of heat, water and pollutants (Oke et al., 2017). This additional warmth (called the urban heat island effect or UHI) can lead to air temperatures that are up to 10°C warmer by night and 1–3°C warmer by day (Oke et al., 2017). During heat waves, UHI can also have synergistic effects on temperature (ambient temperatures can be much higher than they would otherwise be) (Zhao et al., 2018). Moreover, cities can be particularly vulnerable to heat because they house ‘at-risk’ populations including the elderly, those experiencing homelessness and people living with medical comorbidities (Harlan et al., 2013).

Heat Alert and Response Systems (HARS) are public health strategies to help health officials, emergency management, and community and social service providers prepare for and respond to extreme heat events (Alberini et al., 2011). HARS consist of five elements (Government of Canada, n.d. a):

  1. Community mobilization and engagement to determine community needs, identify an agency to coordinate heat responses and identify stakeholders to carry out response actions;

  2. Alert protocols that identify heat-health risks, and activate and deactivate communication and community response plans;

  3. Community response plans which leverage partners and assist vulnerable people;

  4. Communication plans to deliver education about adaptation, connect partners and alert citizens and stakeholders to extreme heat risks; and

  5. Evaluation plans to validate heat-response measures and identify opportunities to improve HARS.

In Canada, the Meteorological Service of Canada, Environment and Climate Change Canada (ECC), supports local and regional HARS protocols through a Heat Warning Program (HWP). Using predefined temperature thresholds, HWP relays heat alerts to public health authorities, and encourages local protection efforts (e.g. staying indoors, opening cooling stations) (Kovats & Hajat, 2008). In Ontario, the HWP is referred to as the Heat Warning and Information System (HWIS). In May 2016, ECC launched a harmonized HWIS in most areas of our province. The harmonized HWIS delivers health-informed, region-specific heat warnings to three geographic areas: Northern Ontario, Southern Ontario and Extreme Southwestern Ontario (Ontario Ministry of Health and Long-Term Care, 2018). Standard heat-health messaging is then communicated to the 36 provincial Public Health Units (PHUs). Alongside municipal and community partners, PHUs then implement local heat response plans. While specific response plans vary with each municipality, efforts can include the communication of clear heat warnings to the public, providing education to at-risk groups and healthcare providers, opening cooling stations, promoting use of air conditioning, adjusting public school programming and providing targeted outreach (City of Toronto, n.d.).

Although there have been studies on the impact of HWP on the public’s awareness of heat warnings, behavioural change following warnings (Angus, 2006; Kalkstein & Sheridan, 2007; Sheridan, 2007) and heat-related mortality (Chau et al., 2009; Ebi et al., 2004; Fouillet et al., 2008; Weinberger et al., 2018), there has been limited focus upon whether HWPs reduce heat-related morbidity (e.g. cardiovascular events, heat stroke, dehydration). Previous evaluative studies that have been conducted have also been marred by methodological issues (Benmarhnia et al., 2019; Nitschke et al., 2016). The use of before and after cross-sectional designs and simple ordinary least squared analyses are unable to capture natural trends in data including seasonality, non-stationarity and autocorrelation (Lagarde, 2012). Also, if models assume that error terms are not correlated (when in fact they are), significant bias can result (Lagarde, 2012). Moreover, there has been little attention on the effectiveness of HWP in urban regions where temperature exposure differs, and limited efforts to understand the value of HWP in at-risk groups including people at risk of homelessness.

The launch of the harmonized HWIS in Ontario presented a natural opportunity to evaluate its impact on heat-related morbidity in urban regions. In this study, we examined whether emergency department (ED) visits for heat-related illness changed following implementation of the harmonized HWIS. We also explored whether effects differed across heat warning regions, and in various at-risk subpopulations. We hypothesized that the implementation of HWIS would be associated with a reduction in heat-related ED encounters.

Methods

Design and setting

We conducted a population-based time series analysis in Ontario, Canada, using administrative health (ICES, formerly the Institute for Clinical Evaluative Sciences), and outdoor temperature data (ECC weather stations). Given cities’ vulnerability to heat, and to align our project with funding from the Canadian Institutes of Health Research’s Healthy Cities Research Initiative, we focused a priori on urban areas. We limited our study period to warmer months (May to September from 2012 to 2018) to exclude the impact of cooler temperatures on morbidity (Chen et al., 2016), while retaining the late spring and summer season when heat alerts are commonly triggered (City of Toronto, 2020). Our study end date was 2018 due to the availability of weather station data.

Ontario has a population of over 14 million residents who have universal access to hospital and physician services. Information on their use of health services is held in administrative databases at ICES. Databases are linked using unique encoded identifiers and analyzed within a privacy and security compliant environment. ICES is an independent, non-profit research institute whose legal status under section 45 of Ontario’s Personal Health Information Act (PHIPA) allows it to collect and analyze healthcare and demographic data without research ethics board approval or informed consent from participants for health system evaluation and improvement. We followed the REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) (Supplementary Materials Table 1).

To sufficiently power our time series analysis (detailed below) and to optimize the number of events per interval, we divided our study period into 2-week intervals for a total of 77: 44 intervals pre-HWIS implementation (May 1, 2016) and 33 intervals post-implementation.

Data sources

We used the Registered Persons Database (RPDB) to capture demographics, vital statistics (e.g. age, sex) and the postal code locations of all Ontario residents. The RPDB holds demographic data for all Ontarians issued a provincial health card (Government of Ontario, n.d. a). We used the National Ambulatory Care Reporting System (NACRS) database (a complete database which captures diagnostic and procedural information acquired during all ED visits in Ontario) to examine encounters for heat-related illness (indicator variable described below) (Government of Ontario, n.d. b).

We also used additional ICES-specific databases including CONTACT to ensure that residents were alive and eligible to receive government-funded healthcare over the study period (i.e. eligible for the Ontario Health Insurance Plan or OHIP). To identify at-risk subpopulations, we used well-validated ICES datasets derived from case definitions of comorbidities including the Ontario Diabetes Database, Chronic Obstructive Pulmonary Disease, Asthma and Congestive Heart Failure databases (Gershon et al., 2009a, 2009b; Lipscombe et al., 2018; Schultz et al., 2013). To identify those at risk of homelessness, we used a validated case definition relevant to those ≥14 years of age. This case definition leverages administrative codes for homelessness (e.g. diagnostic codes for homelessness or inadequate housing, residential status documented to be homeless, admission living setting defined as shelter) in databases including NACRS, the Canadian Institute for Health Information’s Discharge Abstract Database or CIHI-DAD (diagnostic and procedural information from hospital stays), the Ontario Mental Health Reporting System database (data on patients admitted to adult-designated inpatient mental health beds), Home Care Databases (home care services, e.g. nursing) and the National Rehabilitation Reporting System Database (Richard et al., 2019). Compared with a reference standard of the known housing status of a longitudinal cohort of housed (n=137,200) and homeless or vulnerably housed (n=686) individuals, this algorithm has a very high specificity of 99.9% and a sensitivity of 23.2% (Richard et al., 2019). It has been effectively used in several high-impact publications in our province (Liu et al., 2022; Shariff et al., 2022). While the algorithm is unable to comprehensively capture all people experiencing homelessness (as many do not access healthcare services during an episode of homelessness), those identified by the algorithm likely did have recent experience of homelessness.

As our outdoor temperature data source, we used ECC weather stations. Over the study period, there were 105 weather stations located across Ontario. We summarized hourly weather station temperatures as daily maximum values and assigned a daily maximum temperature to each individual in our study using the weather station closest to their postal code region. Maximum temperatures were then averaged for all included individuals during each 2-week study interval.

Population

Our primary population was the general population of Ontario, Canada. Subpopulations included children (<18 years), older adults (≥65 years), adults living with four key medical comorbidities (diabetes, heart failure, chronic obstructive pulmonary disease, and asthma) and those ≥14 years experiencing homelessness. To assess the potential impact of geography on HWIS effectiveness, we examined Ontario’s three heat warning regions (population stratified using postal codes).

For each cohort, we used standard data cleaning to exclude those who died, whose recorded age or sex was missing or invalid, or who were not permanent residents of Ontario prior to the study interval. We also excluded rural residents (i.e. from communities with <10,000 residents) and those not eligible to receive healthcare (i.e. emigrated from the province).

Exposure

The month of harmonized HWIS implementation was our exposure of interest (May 2016). Of note, some PHUs (i.e. Toronto Public Health) piloted the harmonized HWIS during the summer of 2015 in preparation for the Pan-American games (Henderson et al., 2020).

Outcomes

Our primary interest was the rate of ED visits for heat-related illness per 100,000 persons per study interval. We selected heat-related illness as our indicator as these visits are strongly associated with high temperature (e.g. heat stroke) (Sun et al., 2021), and because other health outcomes have triggers that are not always related to heat (e.g. cardiovascular events may be related to high cholesterol and plaque rupture rather than heat exposure). We used International Classification of Diseases 10th revision codes (ICD-10) for heat-related illness. ICD-10 codes were developed by the World Health Organization to promote the monitoring of disease across populations. Following a healthcare encounter, trained coders use structured guidelines to review medical charts and assign them with relevant codes. Administrative coders rely on diagnoses listed by care professionals in the chart to assign codes; they do not interpret reports, diagnostic imaging or laboratory data (Canadian Institute for Health Information, 2016).

ICD-10 codes are then entered into healthcare databases, including NACRS. Codes in the primary position in NACRS represent the main problem, or the diagnosis that the treating healthcare provider felt was most clinically significant (or responsible for the most resource utilization). Patients may have up to 9 additional problems coded during any ED encounter.

As there was no validated ICD-10 coding algorithm for heat-related illness at the time of our study, a subgroup of the authors (JY, KC) applied their clinical expertise and conducted a detailed review of the literature to develop a list of conditions directly associated with heat or a consequence of heat illness (Council of State and Territorial Epidemiologists CSTE, 2016; Harduar Morano & Watkins, 2017; DeGroot et al., 2017). Our final algorithm included codes for heat stroke, heat cramps, heat exhaustion, and heat fatigue, as well as for their consequences, including acute kidney injury, syncope, and volume depletion. All codes are provided in Supplementary Materials Table 2. We defined a patient as having had an ED visit for heat-related illness if they had at least one relevant code assigned to their ED record in any position.

To help contextualize rates, we also provided the demographics, characteristics and healthcare utilization of included patients who presented with an encounter in July 2012 and July 2017. Furthermore, we quantified the number of heat warning events issued by ECC in select cities. We provide the timing of the warnings, and the total number of days the cities were under heat warnings.

Statistical analysis

To examine the impact of Ontario’s HWIS on ED visits for heat-related illness in each study cohort, we conducted an interrupted time series analysis using interventional autoregressive integrated moving average (ARIMA) models with step functions (i.e. intervention variable has the value 0 before the intervention and 1 after) with adjustment for temperature. Interrupted time series analyses are the strongest and most commonly used quasi-experimental design to assess the impact of interventions when randomized controlled trials are not feasible (Jandoc et al., 2015). ARIMA analyses have led to policy and practice changes in Canada (Antoniou et al., 2021; Lapointe-Shaw et al., 2017).

ARIMA regression analysis allowed us to understand whether changes in ED rates (including the level and trend) were explained by implementation of the harmonized HWIS, while accounting for seasonality and residual correlation (i.e. serial dependence of the outcome measure error terms) (Jandoc et al., 2015; Wagner et al., 2002). Before fitting ARIMA models, we assessed the stationarity of each series. We did this by including seasonal explanatory variables in the models. Since our data consisted of 11 intervals per year, each model included 10 seasonal indicator variables. These seasonal indicators accounted for seasonal effects while also achieving stationarity. We also included the mean maximum temperature assigned to included patients during each 2-week interval as a continuous covariate.

After observing stationarity, we then assessed the autocorrelation and partial autocorrelation functions for residual correlation. If residual correlation was observed, autoregressive or moving average terms were added. Seasonal autoregressive and moving average terms were also considered, and lagged across multiples of 11. Diagnostic checks included a test for white noise residuals (Ljung-Box chi-square statistics) as well as normality plots. We present figures of observed rates, along with parameter estimates, 95% confidence intervals (CI) and p values. p values <0.05 were considered statistically significant. The characteristics of included patients were presented descriptively using means (standard deviations or SDs), medians (interquartile ranges or IQR), numbers and percentages. Demographics were provided at the time of ED visits, comorbidities within 5 years prior, and healthcare utilization within the year prior to their visit. All analyses were conducted using SAS version 9.4 (Cary, North Carolina).

Sample size

In time series analyses, there are no fixed limits on the number of data points to include. Power depends upon the distribution of data before and after the intervention, data variability and the strength of the seasonality effect. In this study, we included 44 time periods prior to HWIS implementation and 33 periods after implementation—greater than the recommended minimum of 30 time periods to ensure appropriate statistical power (Moineddin et al., 2003).

Results

Maximum temperatures and heat warning events

Over the study period, maximum daily temperatures as assigned by weather stations varied between 11.95 and 29.95°C. The mean (SD) and median (IQR) distance between the weather station and the resident’s postal code were 11.26 (7.40) and 9.44 (5.90–14.61) km, respectively.

The heat warning events relayed to select cities increased over the study period and are provided in Table 3 of the Supplementary Material. In Toronto, Ontario (the largest urban centre in Ontario), there were 6 heat events triggered between 2013 and 2015 (before harmonized HWIS implementation) and 14 events between 2016 and 2018. Over the study period, the total number of days under heat warnings increased. In Toronto, for example, there were a total of 16 days under heat warnings between 2013 and 2015, and 32 days between 2016 and 2018. Moreover, the mean start date for the first heat warning event of the season was also earlier across most locations from 2016 to 2018 compared with 2013 to 2015.

ED encounters for heat-related illness

A summary of the individuals included across each study year is provided in Tables 4 and 5 of the Supplementary Material. There were 391,310 unique patients with an ED encounter for heat-related illness over the study period. The mean rate of encounters per interval was 47.5 per 100,000 (range of 39.6–60.1). Peak rates were observed between June and July each year. Encounter rates increased over the duration of study and were positively associated with maximum temperature (p<0.001) (Table 1; Fig. 1).

Table 1.

Parameter estimates, standard errors and p values from segmented time series analysis using ARIMA models

Coefficient Upper confidence interval Lower confidence interval p value Interpretation
General population (all ages)
  Intercept (level at time 0) 28.85 24.93 32.77 <0.01 Before the beginning of the observation period the rate of emergency department visits for heat-related illness was 28.85 per 100,000
  Baseline trend (slope) 0.11 0.08 0.15 <0.01 There was a significant interval to interval increase in the rate of emergency department visits before HWIS
  Level change immediately following HWIS 0.20 −1.24 1.64 0.78 The rate of emergency department visits did not change after HWIS
  Trend (slope) change after HWIS 0.04 −0.03 0.10 0.28 The slope of emergency department visits did not change after HWIS
  Maximum temperature 0.47 0.31 0.64 <0.01 There was a significant association between emergency department visit rates and maximum daily temperature throughout the study period
Age <18 years
  Intercept (level at time 0) 24.85 16.77 32.93 <0.01 Before the beginning of the observation period the rate of emergency department visits for heat-related illness was 24.85 per 100,000
  Baseline trend (slope) 0.19 0.08 0.31 <0.01 There was a significant interval to interval increase in the rate of emergency department visits before HWIS
  Level change immediately following HWIS −0.53 −4.86 3.80 0.81 The rate of emergency department visits did not change after HWIS
  Trend (slope) change after HWIS −0.09 −0.31 0.12 0.39 The slope of emergency department visits did not change after HWIS
  Maximum temperature 0.84 0.52 1.17 <0.01 There was a significant association between emergency department visit rates and maximum daily temperature throughout the study period
Age ≥65 years
  Intercept (level at time 0) 103.16 96.77 109.55 <0.01 Before the beginning of the observation period the rate of emergency department visits for heat-related illness was 103.16 per 100,000
  Baseline trend (slope) 0.11 0.02 0.20 0.02 There was a significant interval to interval increase in the rate of emergency department visits before HWIS
  Level change immediately following HWIS 0.77 −2.70 4.24 0.66 The rate of emergency department visits did not change after HWIS
  Trend (slope) change after HWIS 0.14 −0.02 0.31 0.09 The slope of emergency department visits did not change after HWIS
  Maximum temperature 0.49 0.24 0.74 <0.01 There was a significant association between emergency department visit rates and maximum daily temperature throughout the study period
≥18 years with comorbidities
  Level at time 0 (intercept) 57.88 51.33 64.44 <0.01 Before the beginning of the observation period the rate of emergency department visits for heat-related illness was 57.88 per 100,000
  Baseline trend (slope) 0.18 0.12 0.24 <0.01 There was a significant interval to interval increase in the rate of emergency department visits before HWIS
  Level change immediately following HWIS −0.34 −2.75 2.07 0.78 The rate of emergency department visits did not change after HWIS
  Trend (slope) change after intervention 0.09 −0.02 0.20 0.12 The slope of emergency department visits did not change after HWIS
  Maximum temperature 0.73 0.46 1.01 <0.01 There was a significant association between emergency department visit rates and maximum daily temperature throughout the study period
≥14 years with a recent history of homelessness
  Intercept (level at time 0) 90.92 −4.38 186.22 0.07 Before the beginning of the observation period the rate of emergency department visits for heat-related illness was 90.92 per 100,000
  Baseline trend (slope) 1.37 0.05 2.70 0.05 There was a significant interval to interval increase in the rate of emergency department visits before HWIS
  Level change immediately following HWIS −25.03 −76.71 26.66 0.35 The rate of emergency department visits did not change after HWIS
  Trend (slope) change after HWIS 3.07 0.63 5.51 0.02 There was a significant increase in the slope of rates of emergency department visits after HWIS
  Maximum temperature 7.57 3.86 11.27 <0.01 There was a significant association between emergency department visit rates and maximum daily temperature throughout the study period
Extreme Southwestern Ontario (all ages)
  Level at time 0 (intercept) 25.69 17.87 33.50 <0.01 Before the beginning of the observation period the rate of emergency department visits for heat-related illness was 25.69 per 100,000
  Baseline trend (slope) 0.06 0.00 0.13 0.07 There was no significant interval to interval trend in the rate of emergency department visits before HWIS
  Level change immediately following HWIS −1.43 −4.27 1.42 0.33 The rate of emergency department visits did not change after HWIS
  Trend (slope) change after intervention 0.06 −0.06 0.18 0.33 The slope of emergency department visits did not change after HWIS
  Maximum temperature 0.56 0.23 0.90 0.002 There was a significant association between emergency department visit rates and maximum daily temperature throughout the study period
Southern Ontario (all ages)
  Intercept (level at time 0) 27.35 23.36 31.33 <0.01 Before the beginning of the observation period the rate of emergency department visits for heat-related illness was 27.35 per 100,000
  Baseline trend (slope) 0.08 0.05 0.12 <0.01 There was a significant interval to interval increase in the rate of emergency department visits before HWIS
  Level change immediately following HWIS 0.46 −1.00 1.93 0.54 The rate of emergency department visits did not change after HWIS
  Trend (slope) change after HWIS 0.03 −0.04 0.09 0.43 The slope of emergency department visits did not change after HWIS
  Maximum temperature 0.45 0.29 0.62 <0.01 There was a significant association between emergency department visit rates and maximum daily temperature throughout the study period
Northern Ontario (all ages)
  Level at time 0 (intercept) 32.84 20.44 45.25 <0.01 Before the beginning of the observation period the rate of emergency department visits for heat-related illness was 32.84 per 100,000
  Baseline trend (slope) 0.14 0.03 0.26 0.02 There was a significant interval to interval increase in the rate of emergency department visits before HWIS
  Level change immediately following HWIS −3.27 −7.83 1.28 0.16 The rate of emergency department visits did not change after HWIS
  Trend (slope) change after HWIS −0.04 −0.24 0.17 0.74 The slope of emergency department visits did not change after HWIS
  Maximum temperature 0.57 0.04 1.09 0.04 There was a significant association between emergency department visit rates and maximum daily temperature throughout the study period
Open in a new tab

Bold indicates statistical significance at p<0.05

Fig. 1.

Fig. 1

Open in a new tab

Crude rates of ED visits for heat illness per 100,000 persons in Ontario, between May and September, 2012–2018

Characteristics of individuals with an ED encounter for heat-related illness

The characteristics of those who presented with an ED encounter for heat-related illness in July 2012 and July 2017 are included in Table 2. In 2017, there was a slightly higher proportion in the older age group (16.8% aged 65 or older vs. 14.5% in 2012). In 2017, a slightly higher proportion also appeared to have comorbidities (6.7% vs. 5.9% in 2012).

Table 2.

Characteristics of patients with an ED encounter for heat-related illness in July 2012 and July 2017

Characteristic July 2012 July 2017
N=976 N=571
Age
  Mean ± SD 38.37 ± 22.70 38.47 ± 22.67
  Median (IQR) 35 (20–55) 34 (20–56)
   <18 192 (19.7%) 105 (18.4%)
  18–64 642 (65.8%) 370 (64.8%)
  65+ 142 (14.5%) 96 (16.8%)
Sex, female 395 (40.5%) 231 (40.5%)
Neighbourhood income quintile
  1 (lowest) 220 (22.5%) 140 (24.5%)
  2 206 (21.1%) 112 (19.6%)
  3 179 (18.3%) 108 (18.9%)
  4 198 (20.3%) 100 (17.5%)
  5 ≤ 170 111 (19.4%)
  Missing ≤ 5 0 (0.0%)
Forecasting region
  Region 1 45 (4.6%) 32 (5.6%)
  Region 2 872 (89.3%) 483 (84.6%)
  Region 3 59 (6.0%) 56 (9.8%)
Diabetes 101 (10.3%) 61 (10.7%)
Congestive heart failure 25 (2.6%) 13 (2.3%)
Chronic obstructive pulmonary disease 36 (3.7%) 20 (3.5%)
Coronary artery disease 79 (8.1%) 43 (7.5%)
Chronic kidney disease 31 (3.2%) 17 (3.0%)
Asthma 209 (21.4%) 138 (24.2%)
Anxiety/depression 102 (10.5%) 75 (13.1%)
Charlson comorbidity index
  Mean ± SD 0.54 ± 1.22 0.55 ± 1.25
  Median (IQR) 0 (0–1) 0 (0–0)
  0 318 (32.6%) 194 (34.0%)
  1 49 (5.0%) 23 (4.0%)
  2+ 58 (5.9%) 38 (6.7%)
  No hospitalizations 551 (56.5%) 316 (55.3%)
Hospitalizations
  Mean ± SD 0.26 ± 0.74 0.25 ± 0.67
  Median (IQR) 0 (0–0) 0 (0–0)
  0 823 (84.3%) 471 (82.5%)
  1 98 (10.0%) 71 (12.4%)
  2 34 (3.5%) 21 (3.7%)
  3+ 21 (2.2%) 8 (1.4%)
Emergency department visits
  Mean ± SD 1.14 ± 2.97 1.58 ± 4.35
  Median (IQR) 0 (0–1) 0 (0–1)
  0 558 (57.2%) 308 (53.9%)
  1 207 (21.2%) 124 (21.7%)
  2 96 (9.8%) 47 (8.2%)
  3+ 115 (11.8%) 92 (16.1%)
Family physician visits
  Mean ± SD 5.21 ± 6.49 4.90 ± 6.56
  Median (IQR) 3 (1–7) 3 (1–6)
  0 152 (15.6%) 91 (15.9%)
  1 119 (12.2%) 94 (16.5%)
  2 129 (13.2%) 71 (12.4%)
  3–4 200 (20.5%) 101 (17.7%)
  5–6 132 (13.5%) 80 (14.0%)
  7–8 76 (7.8%) 40 (7.0%)
  9–10 45 (4.6%) 28 (4.9%)
11+ 123 (12.6%) 66 (11.6%)
Open in a new tab

Cell sizes <6 were not presented as per ICES privacy

Abbreviations: IQR interquartile range, SD standard deviation

ED encounters in at-risk subpopulations in urban Ontario

Rates of ED visits in at-risk subpopulations are presented in Fig. 1 of the Supplementary Materials. While differences in rates may have been due to age, sex and comorbidities, those aged ≥65 years, those living with medical comorbidities and individuals with a recent history of homelessness had higher rates of ED visits than the general population (mean rate across intervals was 120.4 in those ≥65 years, 85.7 in those living with comorbidities and 337.0 per 100,000 in those with a recent history of homelessness, respectively) (Table 1). In at-risk subpopulations, we also found that encounter rates increased over time, were positively associated with temperature, and peaked in June and July.

Impact of harmonized HWIS on ED visits for heat-related illness

Following implementation of the harmonized HWIS, there was no statistically significant change in ED encounter rates (rate 0.04, 95% CI: −0.03 to 0.1, p=0.278) (Table 1; Fig. 1; Supplementary Materials Table 6a). In some subpopulations, there was a trend to a reduction (e.g. those <18, those with comorbidities, and those living in extreme Southwestern Ontario and Northern Ontario), but the change was not statistically significant (Table 1; Supplementary Materials Fig. 1; Supplementary Tables 6b–6i). While those with a recent history of homelessness appeared to have a higher rate of ED encounters (i.e. higher slope) in the period after the intervention, this was most apparent from 2018 onward, when coding practices for homelessness also changed (ICD-10 codes Z59.0 and Z59.1 became mandatory in 2018 and likely improved the sensitivity of our algorithm).

Discussion

In this large population-based study of urban residents of Ontario, Canada, we used a quasi-experimental design (time series with ARIMA analysis) to examine whether the implementation of HWIS was associated with a change in ED visits for heat-related illness (Lagarde, 2012). Rates of ED visits did appear to decrease slightly in some subpopulations, an encouraging finding with our warming climate and changing demographics (e.g. aging population, more comorbidities) (Government of Canada, n.d. b). However, overall, the changes were not statistically significant, nor were they statistically significant on a general population level.

There have been very few studies to examine the impact of Heat Warning Programs on morbidity. One small, single-centre study in 2008 used a difference-in-difference approach to investigate whether a change in threshold for the New York City (NYC) Heat Emergency Plan was associated with a reduction in heat-related illness. Investigators found that the change did reduce daily rates of heat-related illness by 0.80 (95% confidence interval or CI 0.27 to 1.33) (Benmarhnia et al., 2019). In a study conducted in Australia, investigators examined the impact of their HWP on average daily rates of ambulance calls and ED presentations during heatwaves vs. non-heatwaves in 2009 and 2014 (pre- and post-implementation). It was observed that incident rate ratios for total ambulance calls were lower during heatwaves in 2014 as compared with in 2009 (p<0.1) (Nitschke et al., 2011). The authors also noted a significant reduction in ED presentations for kidney problems following the implementation of HWP. Our study, which used more robust statistical analyses, had findings that differed from these previous evaluations (i.e. overall statistically null effect of HWIS). This could be because our study was conducted in a different region from Australia which has different maximum temperatures during hot months (e.g. in July 2018, maximum temperature was 34°C in Toronto, Ontario, vs. 49°C in some areas of Australia in December 2018). Residents who live in different world regions may have different acclimatization to extreme heat, or different medical/socioeconomic characteristics that impact their heat vulnerability. Furthermore, in different regions of the world with different healthcare funding, healthcare usage on heat warning days could differ (patients might seek more care in the ED rather than seeing a primary care provider for example). Furthermore, HARS processes in Ontario may differ from those in other countries/regions, and our cities and public health units may have different heat-response plans (i.e. NYC’s plans are particularly robust) (Benmarhnia et al., 2019). In addition, compared with prior studies, our evaluation focused on ED encounters as our heat-related indicator, where assessment of ambulance visits may have increased event rates.

Similar to other studies, we found that socioeconomically disadvantaged individuals, particularly those experiencing homelessness, are at increased risk of heat illness (Nicolay et al., 2016). In fact in our study, those at risk of homelessness were at a 10-fold higher risk of an ED encounter than other subpopulations. This is striking given both the increasing threat of climate change and the great prevalence of homelessness in Canada (Gaetz et al., 1980). Furthermore, like other studies, we also found that elderly individuals and those living with medical comorbidities are at risk of heat-related morbidity due to their underlying physiology (e.g. impaired vasodilation) and use of medications including diuretics (Kenny et al., 2010).

The main threat to the validity of interrupted time series analysis relates to time-varying confounding, such as changes in outcome coding, co-interventions, changes in healthcare utilization, or changes in the population under study (Lagarde, 2012). Over our study period, there was a change in the homeless coding algorithm, as well as change ambient temperature, but we did adjust for temperature in our models. Also, we do recognize that ED rates may have varied over time due to changes in the population structure, though we note only minor differences in the demographics and comorbidities of included individuals. To the best of our knowledge, there were no other changes in healthcare utilization or environmental health policy that could have contributed to our study findings over the period of study.

Other limitations include our use of May 1, 2016 as the intervention date as we do recognize that some Ontario communities may have adopted practices earlier (e.g. Toronto). We chose ED visits for heat-related illness as our morbidity indicator, but we recognize that this indicator has limitations. For example, we could not capture events that were not severe enough to require hospital presentation (e.g. use of emergency medical services only) nor symptoms that did not lead to any medical care. Unfortunately, there are no heat surveillance programs available for these types of analysis in the province of Ontario.

Furthermore, we did not examine whether implementation of HWIS reduced healthcare visits for other outcomes that could be related to heat (e.g. acute cardiovascular events) (Lavigne et al., 2014); however, these events can also be due to factors apart from heat exposure including the presence of dyslipidemia, and diabetes.

Also, we used weather station data which does not capture intra-urban climate well (Spangler et al., 2019), but we do not expect that this led to differential misclassification of temperature exposure over the study intervals. We chose to adjust for maximum daily temperature in our analyses, but we could have focused upon other measures including heat index.

Importantly, we conducted our analysis at a provincial level (harmonized HWIS was a provincial effort), and we did not have the statistical power to examine whether effects were different across included cities. We cannot then make inferences on the effectiveness of HWIS across different municipalities and public health units where mitigation efforts might differ (e.g. homeless outreach, home visits to home care patients, or the opening of cooling centres) (Benmarhnia et al., 2019). We also cannot make conclusions about the effectiveness of the different components of HARS (e.g. heat warning thresholds, community response plans).

Nevertheless, our findings are important. Climate change is not only happening, but it is adversely affecting the health of our population. On a population-based level, we did find a slight reduction in ED encounters for heat-related illness in some groups, but the change was not statistically significant. Given the heat-related vulnerability of at-risk groups, we propose that policy makers and PHUs make special efforts to reach and support these populations, particularly those with medical comorbidities and those living with homelessness who had a much higher risk of ED encounters (Berisha et al., 2017; Nicolay et al., 2016). There may be opportunities, for example, for directed education and focused warnings for these groups (Angus, 2006; Mehiriz et al., 2018; Sheridan, 2007). Cities with successful programming for at-risk groups could share community response plans with other municipalities. Furthermore, should future evaluations of HWPs be conducted, it would also be ideal to capture additional heat-health indicators such as ambulance calls for heat-related illness. With more morbidity data and event rates, examining HWP processes and outcomes in specific communities might be possible.

Contributions to knowledge

What does this study add to existing knowledge?

  • While there have been studies on the impact of Heat Warning Programs (HWP) on the public’s awareness of heat, behavioural change following warnings, and heat-related mortality, few large-scale studies have examined the impact of HWP on heat-related morbidity.

  • There has also been little attention on the effectiveness of HWP in urban regions and in at-risk populations (e.g. people experiencing homelessness), and studies have been marred by methodological issues.

  • We conducted a large evaluative study of ECC’s harmonized HWIS in Canada’s most populous province (Ontario, Canada) with a focus on at-risk populations.

What are the key implications for public health interventions, policy or practice?

  • More heat alerts were triggered in Ontario between 2013 and 2018, and many cities spent more days under heat warnings. The implementation of a harmonized HWIS appeared to reduce rates of ED visits for heat-related illness in some subpopulations, but at a provincial level, the change was not statistically significant.

  • Given HWPs are a main policy tool to protect populations against heat, we suggest ongoing efforts to support effective HWP in our communities, with a particular focus on at-risk groups.

Supplementary information

ESM 1 (134.6KB, docx)

(DOCX 134 kb)

Acknowledgements

We thank Marlee Vinegar for aiding with our literature review.

Author contributions

KC conceptualized and designed the study, interpreted the results and drafted the manuscript. AMO acquired the data, performed the analysis and reviewed the manuscript critically. BL conceptualized the study, performed the analysis and reviewed the manuscript critically. JV conceptualized the study, interpreted the results and reviewed the manuscript. MM designed the study, acquired data, performed parts of the analysis, interpreted results and reviewed the manuscript. RS designed the study, interpreted results and critically reviewed the manuscript. JWY interpreted the results and reviewed the manuscript. ESK interpreted the results and reviewed the manuscript. JG helped conceptualize the study, interpreted results and reviewed the manuscript. CF interpreted results and reviewed the manuscript. RV interpreted the results and reviewed the manuscript. SZ conceptualized and designed the study, interpreted the results and reviewed the manuscript critically.

Funding

This study was funded by a Canadian Institutes of Health Research Operating Grant (Data Analysis Using Existing Databases and Cohorts). This study was also supported by ICES, which is funded in part by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). The study was completed at the ICES Western site, where core funding is provided by the Academic Medical Organization of Southwestern Ontario, the Schulich School of Medicine and Dentistry, Western University, and the Lawson Health Research Institute. Parts of this publication are based upon data and/or information compiled and provided by the Canadian Institute for Health Information (CIHI). However the analysis, conclusions, opinions and statements expressed in the material are those of the authors and not necessarily those of CIHI .

Data availability

The dataset from this study is held securely in coded form at ICES. While data sharing agreements prohibit ICES from making the dataset publicly available, access may be granted to those who meet pre-specified criteria for confidential access, available at www.ices.on.ca/DAS.

Code availability

The full dataset creation plan and underlying analytic code are available from the authors upon request, understanding that the programs may rely upon coding templates or macros that are unique to ICES.

Declarations

Ethics approval

ICES is an independent, non-profit research institute whose legal status under section 45 of Ontario’s Personal Health Information Protection Act (PHIPA) allows it to collect and analyze healthcare and demographic data without research ethics board approval or informed consent from participants for health system evaluation and improvement.

Consent to participate

Not applicable

Consent for publication

Not applicable

Conflict of interest

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  1. Alberini, A., Gans, W., & Alhassan, M. (2011). Individual and public-program adaptation: Coping with heat waves in five cities in Canada. International Journal of Environmental Research and Public Health. 10.3390/ijerph8124679 [DOI] [PMC free article] [PubMed]
  2. Angus J. An evaluation of Toronto’s heat watch warning system. University of Toronto; 2006. [Google Scholar]
  3. Antoniou, T., Martins, D., Campbell, T., Tadrous, M., Munro, C., Leece, P., … Gomes, T. (2021). Impact of policy changes on the provision of naloxone by pharmacies in Ontario, Canada: A population-based time-series analysis. Addiction (Abingdon, England), 116(6), 1514–1520. 10.1111/ADD.15324 [DOI] [PMC free article] [PubMed]
  4. Benmarhnia, T., Schwarz, L., Nori-Sarma, A., & Bell, M. L. (2019). Quantifying the impact of changing the threshold of New York City heat emergency plan in reducing heat-related illnesses. Environmental Research Letters, 14(11), 114006. 10.1088/1748-9326/ab402e
  5. Berisha, V., Hondula, D., Roach, M., White, J., McKinney, B., Bentz, D., … Goodin, K. (2017). Assessing adaptation strategies for extreme heat: A public health evaluation of cooling centers in Maricopa County, Arizona. Weather, Climate, and Society, 9(1), 71–80. https://journals.ametsoc.org/wcas/article/9/1/71/1016/Assessing-Adaptation-Strategies-for-Extreme-Heat-A. Accessed 14 Sept 2021.
  6. Bobb, J. F., Obermeyer, Z., Wang, Y., & Dominici, F. (2014). Cause-specific risk of hospital admission related to extreme heat in older adults. JAMA, 312(24), 2659–2667. 10.1001/jama.2014.15715 [DOI] [PMC free article] [PubMed]
  7. Canadian Institute for Health Information. (2016). Classification and coding. Retrieved 19 May 2016, from https://www.cihi.ca/en/data-and-standards/standards/classification-and-coding
  8. Chau, P. H., Chan, K. C., & Woo, J. (2009). Hot weather warning might help to reduce elderly mortality in Hong Kong. International Journal of Biometeorology. 10.1007/s00484-009-0232-5 [DOI] [PubMed]
  9. Chen H, Wang J, Li Q, Yagouti A, Lavigne E, Foty R, et al. Assessment of the effect of cold and hot temperatures on mortality in Ontario, Canada: A population-based study. CMAJ Open. 2016;4(1):E48. doi: 10.9778/cmajo.20150111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. City of Toronto. (n.d.). About Toronto’s Heat Relief Strategy. Retrieved 27 November 2020 from https://www.toronto.ca/community-people/health-wellness-care/health-programs-advice/hot-weather/about-torontos-heat-relief-strategy/#:~:text=The.
  11. City of Toronto. (2020). Alert Statistics. Retrieved 16 October 2020, from http://app.toronto.ca/tpha/heatStats.html
  12. Council of State and Territorial Epidemiologists CSTE. (2016). Heat-related illness syndrome query: A guidance document for implementing heat-related illness syndromic surveillance in public health practice.
  13. Crandall, C. G., & González-Alonso, J. (2010, 199). Cardiovascular function in the heat-stressed human. Acta Physiologica (Oxford, England), (4), 407. 10.1111/J.1748-1716.2010.02119.X [DOI] [PMC free article] [PubMed]
  14. DeGroot, D. W., Mok, G., & Hathaway, N. E. (2017). International Classification of Disease coding of exertional heat illness in U.S. army soldiers. Military Medicine, 182(9), e1946–e1950. 10.7205/MILMED-D-16-00429 [DOI] [PubMed]
  15. Ebi, K. L., Teisberg, T. J., Kalkstein, L. S., Robinson, L., & Weiher, R. F. (2004). Heat watch/warning systems save lives: Estimated costs and benefits for Philadelphia 1995-98. Bulletin of the American Meteorological Society. 10.1175/BAMS-85-8-1067
  16. Fouillet, A., Rey, G., Wagner, V., Laaidi, K., Empereur-Bissonnet, P., Le Tertre, A., … Hémon, D. (2008). Has the impact of heat waves on mortality changed in France since the European heat wave of summer 2003? A study of the 2006 heat wave. International Journal of Epidemiology. 10.1093/ije/dym253 [DOI] [PMC free article] [PubMed]
  17. Gaetz, S., Dej, E., Richter, T., & Redman, M. (1980). The state of homelessness in Canada 2016. Retrieved 3 May 2022 from www.homelesshub.ca. Accessed 14 Sept 2021.
  18. Gershon, A. S., Wang, C., Guan, J., Vasilevska-Ristovska, J., Cicutto, L., & To, T. (2009a). Identifying individuals with physcian diagnosed COPD in health administrative databases. COPD: Journal of Chronic Obstructive Pulmonary Disease, 6(5), 388–394. 10.1080/15412550903140865 [DOI] [PubMed]
  19. Gershon, A. S., Wang, C., Guan, J., Vasilevska-Ristovska, J., Cicutto, L., & To, T. (2009b). Identifying patients with physician-diagnosed asthma in health administrative databases. Canadian Respiratory Journal, 16(6), 183–188. 10.1155/2009/963098 [DOI] [PMC free article] [PubMed]
  20. Government of Canada. (n.d. a). Heat alert and response systems (HARS) across Canada. Retrieved 8 April 2022 from https://www.canada.ca/en/health-canada/services/climate-change-health/heat-alert-reponse-systems.html?msclkid=9e8f6524b4dd11ec9ad2522a171c6792. Accessed 15 Jul 2021.
  21. Government of Canada. (n.d. b). Temperature change in Canada. Retrieved 3 May 2022 from https://www.canada.ca/en/environment-climate-change/services/environmental-indicators/temperature-change.html
  22. Government of Ontario. (n.d. a). Registered Persons Database (RPDB) - Datasets - Ontario Data Catalogue. Retrieved 8 April 2022 from https://data.ontario.ca/dataset/registered-persons-database-rpdb
  23. Government of Ontario. (n.d. b). National Ambulatory Care Reporting System (NACRS) - Datasets - Ontario Data Catalogue. Retrieved 8 April 2022 from https://data.ontario.ca/dataset/national-ambulatory-care-reporting-system-nacrs
  24. Harduar Morano, L., & Watkins, S. (2017). Evaluation of diagnostic codes in morbidity and mortality data sources for heat-related illness surveillance. Public Health Reports, 132(3), 326–335. 10.1177/0033354917699826 [DOI] [PMC free article] [PubMed]
  25. Harlan, S. L., Declet-Barreto, J. H., Stefanov, W. L., & Petitti, D. B. (2013). Neighborhood effects on heat deaths: Social and environmental predictors of vulnerability in Maricopa County, Arizona. Environmental Health Perspectives, 121(2). 10.1289/ehp.1104625 [DOI] [PMC free article] [PubMed]
  26. Henderson, D., Aubin, L., Behan, K., Chen, H., Doyle, H., Gower, S., … Yusa, A. (2020). Developing a harmonized heat warning and information system for Ontario: A case study in collaboration. Canadian Journal of Public Health, 111(3), 426–432. 10.17269/s41997-020-00337-y [DOI] [PMC free article] [PubMed]
  27. Jandoc, R., Burden, A. M., Mamdani, M., Lévesque, L. E., & Cadarette, S. M. (2015). Interrupted time series analysis in drug utilization research is increasing: Systematic review and recommendations. Journal of Clinical Epidemiology, 68(8), 950–956. 10.1016/j.jclinepi.2014.12.018 [DOI] [PubMed]
  28. Kalkstein, A. J., & Sheridan, S. C. (2007). The social impacts of the heat-health watch/warning system in Phoenix, Arizona: Assessing the perceived risk and response of the public. International Journal of Biometeorology. 10.1007/s00484-006-0073-4 [DOI] [PubMed]
  29. Kenny, G. P., Yardley, J., Brown, C., Sigal, R. J., & Jay, O. (2010). Heat stress in older individuals and patients with common chronic diseases. CMAJ, 182(10), 1053–1060. 10.1503/cmaj.081050 [DOI] [PMC free article] [PubMed]
  30. Kovats, R. S., & Hajat, S. (2008). Heat stress and public health: A critical review. Annual Review of Public Health, 29, 41–55. http://www.ncbi.nlm.nih.gov/pubmed/18031221. Accessed 14 Jun 2021. [DOI] [PubMed]
  31. Lagarde, M. (2012). How to do (or not to do) ... Assessing the impact of a policy change with routine longitudinal data. Health Policy and Planning, 27(1), 76–83. 10.1093/heapol/czr004 [DOI] [PubMed]
  32. Lapointe-Shaw, L., Mamdani, M., Luo, J., Austin, P. C., Ivers, N. M., Redelmeier, D. A., & Bell, C. M. (2017). Effectiveness of a financial incentive to physicians for timely follow-up after hospital discharge: A population-based time series analysis. CMAJ, 189(39), E1224–E1229. 10.1503/CMAJ.170092 [DOI] [PMC free article] [PubMed]
  33. Lavigne, E., Gasparrini, A., Wang, X., Chen, H., Yagouti, A., Fleury, M. D., & Cakmak, S. (2014). Extreme ambient temperatures and cardiorespiratory emergency room visits: Assessing risk by comorbid health conditions in a time series study. Environmental Health, 13(1), 5. 10.1186/1476-069X-13-5 [DOI] [PMC free article] [PubMed]
  34. Li, D., & Bou-Zeid, E. (2013). Synergistic interactions between urban heat islands and heat waves: The impact in cities is larger than the sum of its parts. Journal of Applied Meteorology and Climatology, 52(9), 2051–2064. 10.1175/JAMC-D-13-02.1
  35. Lipscombe, L. L., Hwee, J., Webster, L., Shah, B. R., Booth, G. L., & Tu, K. (2018). Identifying diabetes cases from administrative data: A population-based validation study. BMC Health Services Research, 18(1), 316. 10.1186/s12913-018-3148-0 [DOI] [PMC free article] [PubMed]
  36. Liu, M., Richard, L., Campitelli, M. A., Nisenbaum, R., Dhalla, I. A., Wadhera, R. K., … Hwang, S. W. (2022). Hospitalizations during the COVID-19 pandemic among recently homeless individuals: A retrospective population-based matched cohort study. Journal of General Internal Medicine. 10.1007/S11606-022-07506-4 [DOI] [PMC free article] [PubMed]
  37. Mehiriz, K., Gosselin, P., Tardif, I., & Lemieux, M.-A. (2018). The effect of an automated phone warning and health advisory system on adaptation to high heat episodes and health services use in vulnerable groups—evidence from a randomized controlled study. International Journal of Environmental Research and Public Health, 15(8), 1581. 10.3390/ijerph15081581 [DOI] [PMC free article] [PubMed]
  38. Moineddin, R., Upshur, R. E., Crighton, E., & Mamdani, M. (2003). Autoregression as a means of assessing the strength of seasonality in a time series. Population Health Metrics, 1(1), 10. 10.1186/1478-7954-1-10 [DOI] [PMC free article] [PubMed]
  39. Nicolay, M., Brown, L. M., Johns, R., & Ialynytchev, A. (2016). A study of heat related illness preparedness in homeless veterans. International Journal of Disaster Risk Reduction, 18, 72–74. 10.1016/J.IJDRR.2016.05.009
  40. Nitschke, M., Tucker, G., Hansen, A., Williams, S., Zhang, Y., & Bi, P. (2016). Evaluation of a heat warning system in Adelaide, South Australia, using case-series analysis. BMJ Open, 6(7), e012125. 10.1136/bmjopen-2016-012125 [DOI] [PMC free article] [PubMed]
  41. Nitschke, M., Tucker, G. R., Hansen, A. L., Williams, S., Zhang, Y., & Bi, P. (2011). Impact of two recent extreme heat episodes on morbidity and mortality in Adelaide, South Australia: A case-series analysis. Environmental Health: A Global Access Science Source, 10, 42. 10.1186/1476-069X-10-42 [DOI] [PMC free article] [PubMed]
  42. Oke TR, Mills G, Christen A, Voogt J. Urban climates. Cambridge University Press; 2017. [Google Scholar]
  43. Ontario Ministry of Health and Long-Term Care. (2018). A harmonized heat warning and information system for Ontario. Retrieved from http://www.health.gov.on.ca/en/common/ministry/publications/reports/heat_warning_information_system/heat_warning_information_system.aspx. Accessed 4 May 2021.
  44. Richard, L., Hwang, S. W., Forchuk, C., Nisenbaum, R., Clemens, K., Wiens, K., … Shariff, S. Z. (2019). Validation study of health administrative data algorithms to identify individuals experiencing homelessness and estimate population prevalence of homelessness in Ontario, Canada. BMJ Open, 9(10), e030221. 10.1136/bmjopen-2019-030221 [DOI] [PMC free article] [PubMed]
  45. Schultz, S. E., Rothwell, D. M., Chen, Z., & Tu, K. (2013). Identifying cases of congestive heart failure from administrative data: A validation study using primary care patient records. Chronic Diseases and Injuries in Canada. [PubMed]
  46. Shariff, S. Z., Richard, L., Hwang, S. W., Kwong, J. C., Forchuk, C., Dosani, N., & Booth, R. (2022). COVID-19 vaccine coverage and factors associated with vaccine uptake among 23 247 adults with a recent history of homelessness in Ontario, Canada: A population-based cohort study. The Lancet. Public Health, 7(4), e366–e377. 10.1016/S2468-2667(22)00037-8 [DOI] [PMC free article] [PubMed]
  47. Sheridan, S. C. (2007). A survey of public perception and response to heat warnings across four North American cities: An evaluation of municipal effectiveness. International Journal of Biometeorology. 10.1007/s00484-006-0052-9 [DOI] [PubMed]
  48. Simpson, C., & Abelsohn, A. (2012). Heat-induced illness. CMAJ, 184(10), 1170. 10.1503/CMAJ.120492 [DOI] [PMC free article] [PubMed]
  49. Spangler, K. R., Weinberger, K. R., & Wellenius, G. A. (2019). Suitability of gridded climate datasets for use in environmental epidemiology. Journal of Exposure Science & Environmental Epidemiology, 29(6), 777–789. 10.1038/s41370-018-0105-2 [DOI] [PMC free article] [PubMed]
  50. Sun, S., Weinberger, K. R., Nori-Sarma, A., Spangler, K. R., Sun, Y., Dominici, F., & Wellenius, G. A. (2021). Ambient heat and risks of emergency department visits among adults in the United States: Time stratified case crossover study. BMJ (Clinical Research Ed.), 375, e065653. 10.1136/BMJ-2021-065653 [DOI] [PMC free article] [PubMed]
  51. Wagner, A. K., Soumerai, S. B., Zhang, F., & Ross-Degnan, D. (2002). Segmented regression analysis of interrupted time series studies in medication use research. Journal of Clinical Pharmacy and Therapeutics, 27(4), 299–309. 10.1046/j.1365-2710.2002.00430.x [DOI] [PubMed]
  52. Weinberger, K. R., Zanobetti, A., Schwartz, J., & Wellenius, G. A. (2018). Effectiveness of National Weather Service heat alerts in preventing mortality in 20 US cities. Environment International, 116, 30–38. 10.1016/j.envint.2018.03.028 [DOI] [PMC free article] [PubMed]
  53. Zhao, L., Oppenheimer, M., Zhu, Q., Baldwin, J. W., Ebi, K. L., Bou-Zeid, E., … Liu, X. (2018). Interactions between urban heat islands and heat waves. Environmental Research Letters, 13(3), 034003. 10.1088/1748-9326/aa9f73

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ESM 1 (134.6KB, docx)

(DOCX 134 kb)

Data Availability Statement

The dataset from this study is held securely in coded form at ICES. While data sharing agreements prohibit ICES from making the dataset publicly available, access may be granted to those who meet pre-specified criteria for confidential access, available at www.ices.on.ca/DAS.

The full dataset creation plan and underlying analytic code are available from the authors upon request, understanding that the programs may rely upon coding templates or macros that are unique to ICES.

Articles from Canadian Journal of Public Health = Revue Canadienne de Santé Publique are provided here courtesy of Springer

RESOURCES