4.0 Discussion

4.1 Summary

When deaths during the 2021 EHE were compared with deaths during previous years, there was a pronounced increase in odds of death for those with schizophrenia. This increase was observed among all deaths and the three EHE subgroups: heat-related (X30) deaths, deaths with pending cause (R99), and non-heat-related deaths (not X30 or R99). We also found that odds of EHE mortality were increased for those with chronic kidney disease and ischemic heart disease among all EHE deaths. While chronic kidney disease was increased among the heat-related (X30) and non-heat-related subgroups (not X30 or R99), ischemic heart disease was only increased in the non-heat-related subgroup (not X30 or R99). Depression and diabetes were associated with increased odds of EHE mortality among heat-related deaths (X30), substance use disorder and chronic obstructive pulmonary disease were associated with increased odds among deaths with pending cause (R99), and ischemic stroke was increased among non-heat-related deaths (not X30 or R99). Having 3 or more chronic diseases was also associated with increased odds of EHE death, though the odds did not increase with an increasing number of diseases. Finally, the odds of EHE mortality were significantly lower among those with angina, hospitalized transient ischemic attack, dementia, and osteoporosis.

4.2 Chronic diseases and increased odds of death during the EHE

4.2.1 Specific chronic diseases

Schizophrenia was the most strongly associated with the odds of EHE death after adjusting for the other 20 chronic diseases. This aligns with a growing number of studies reporting an increased risk of death among those with mental illness during EHEs (J. Liu et al., 2021; Lõhmus, 2018). For example, Hansen et al. (2008) found that the rate of deaths attributed to schizophrenia-type disorders was doubled (Incidence Rate Ratio = 2.08) during heat wave periods (daily max temperature ≥35℃ for at least 3 consecutive days) compared with non-heat wave periods in Adelaide, Australia from 1993-2004. We report an OR of 3.07 [2.39, 3.94] for administrative diagnosis of schizophrenia among all-cause EHE deaths in BC. The relationship between increased risk of death and mental illness is not fully understood and is likely the result of a complex set of interacting factors. Some people with schizophrenia may lack insight into their own health status, and thus may not perceive and respond to overheating (J. Liu et al., 2021). Schizophrenia is also treated with antipsychotic medications that can affect thermoregulation (Lõhmus, 2018). Furthermore, the condition is associated with stigmatization, social isolation, economic marginalization, and coincident substance use disorder, all of which are independent risk factors for EHE mortality (Henderson, McLean, et al., 2022; J. Liu et al., 2021; Raphael et al., 2020; Semenza et al., 1996).
While subgroup analyses showed that people with schizophrenia were at increased odds of EHE mortality regardless of the cause of death, cause-specific associations with depression and substance use disorder may indicate more nuanced links to mental illness. Specifically, the odds of heat-related death (X30) were increased among those with depression. Like schizophrenia, depression is treated with medications that can affect thermoregulation and is associated with stigmatization, social isolation, and lower-socioeconomic status (Lõhmus, 2018). Along with the results for schizophrenia, this suggests that mental illness is an important risk factor for heat-related mortality.
Our results also suggest that substance use disorder was associated with increased odds of EHE mortality among deaths with pending underlying cause (R99). Many of these deaths may be due to the ongoing toxic drug supply crises in BC (BC Coroners Service, 2022b). Substance use disorder has been previously associated with increased hot weather mortality (J. Liu et al., 2021; Wilson et al., 2013). Acute intoxication during EHEs may impair decision-making, and different substances (e.g., cocaine) can affect physiological susceptibility to heat (Bohnert et al., 2010). Long-term substance use is associated with health impacts that may increase the risk of EHE death (Ebi et al., 2021) such as alcoholic cardiomyopathy (Maisch, 2016). Further, mental illness and substance use disorder often co-occur, and may overlap to increase heat-related mortality risks (Brady & Sinha, 2005). For instance, 144 (47%) of all 309 decedents with schizophrenia in this study also had substance use disorder.
The association between chronic kidney disease and increased odds of EHE mortality aligns with previous research. For example, in Adelaide, Australia, hospital admissions for renal disease and acute renal failure were heightened during heat wave periods (A. L. Hansen et al., 2008), and the risk of same-day mortality among hemodialysis patients with end-stage renal disease was increased by 31% [1%, 70%] during EHEs in three USA cities from 2017 – 2019 (Remigio et al., 2019). This is quite consistent with our OR estimate of 1.36 [1.18, 1.56]. Associations between EHE mortality and kidney disease may be linked to increased sweat production, fluid loss, and dehydration during hot weather. For instance, as the body becomes dehydrated, the blood volume decreases putting strain on the kidneys, which may increase the risk of renal damage and failure (Ebi et al., 2021). Although pre-existing kidney problems may exacerbate this risk, chronic kidney disease is also associated with other conditions, such as heart disease, which may also increase the risk of death during EHEs (Ebi et al., 2021; Gansevoort et al., 2013). Indeed, we report increased odds of EHE mortality among those with ischemic heart disease and ischemic stroke. This is unsurprising because cardiovascular conditions are among the most widely recognized risk factors for heat-related mortality (Ebi et al., 2021) and are often referenced in public health messaging about extreme heat (Centers for Disease Control and Prevention, 2017).
Finally, consistent with previous studies, the OR for diabetes was increased among EHE deaths with a heat-related cause of death (X30). For example, a recent meta-analysis reported a pooled risk ratio of 1.18 [1.13, 1.25] for diabetes-related death during hot weather across more than twenty studies (Moon, 2021). This is somewhat lower than the OR of 1.42 [1.08, 1.86] we report for administrative diagnosis of diabetes among heat-related EHE deaths. The association between diabetes and hot weather mortality may be explained by findings that diabetes can affect the physiological capacity to dissipate heat by decreasing cutaneous blood flow and sweating (Kenny et al., 2016). However, these impacts depend on how well the disease is managed, how long someone has lived with diabetes, and their aerobic fitness (Kenny et al., 2016).

4.2.2 Burden of chronic disease

The odds of EHE mortality were higher for those with 3 or more chronic diseases but did not increase markedly for those with a higher number of chronic diseases. Previous research has also found higher risk among those with a higher burden of chronic disease. For example, a study of nearly 200,000 people in Germany found that dementia was associated with higher risk of death during hot weather, and that risk increased among those needing more care (Fritze, 2020). Similarly, Foroni et al. (2007) found that elderly persons living in Modena, Italy during a hot weather period in 2003 were more likely to die if they; (1) had a higher comorbidity score, (2) were more likely to use public health services, (3) took more medications, (4) were more likely to have been admitted to the hospital in the previous year, and (5) were more likely to have cognitive impairments or other disability. Together, these results indicate that individuals with more pre-existing conditions and with poorer overall health may have been at increased risk of death during the 2021 EHE in BC.

4.3 Chronic diseases and decreased odds of death during the EHE

Angina, hospitalized transient ischemic attack, dementia, and osteoporosis were associated with significantly lower odds of mortality during the EHE. This was surprising because some of these factors have been previously linked to increased heat susceptibility (Ebi et al., 2021; Fritze, 2020). For example, as previously mentioned, cardiovascular conditions are among the most widely recognized risk factors for heat-related mortality (Ebi et al., 2021), and dementia was associated with an 11% increase in the risk of death during periods of high ambient temperatures between 2004 – 2010 across Germany (Fritze, 2020). Our results may be inconsistent with other studies due to the case-only design or because these individuals may have received heightened care during the EHE, such as an increased frequency of health checks (Kafeety et al., 2020) and increased use of air conditioning. This is supported by BCCS reports (BC Coroners Service, 2022a) that most heat-related deaths occurred inside of private residences and not in health care settings where the frailest individuals may have been housed. However, more detailed investigations are required to better interpret these results.

4.4 Non-heat-related deaths

The distribution of chronic diseases among the non-heat-related (not X30 or R99) EHE deaths was similar to the typical weather deaths. However, the odds of non-heat-related death during the EHE were still increased for those with schizophrenia, chronic kidney disease, ischemic stroke, and ischemic heart disease, suggesting that deaths with non-heat-related causes were still affected by the EHE. For example, decedents with ischemic heart disease may have been coded with a cardiovascular cause of death. However, exposure to extreme heat increases the oxygen demand of the heart as the body cools itself by redistributing blood flow to the skin. As a result, those with ischemic heart disease are at a heightened risk of experiencing cardiovascular events during hot weather because the disease causes a reduction in blood flow and oxygen delivery to the heart (Ebi et al., 2021). Despite these deaths being related to heat exposure, they may have been coded with non-heat-related causes.
There may also be unidentified heat-related deaths (X30) included in the non-heat-related group. Only deaths reported to BCCS had the opportunity to be attributed to heat exposure, and approximately 50% of deaths during the EHE were not investigated by BCCS (BC Coroners Service, 2021; 2022a). Furthermore, there were an estimated 740 excess deaths during the 8-day EHE (Henderson et al., 2021), but BCCS only attributed 562 deaths to extreme heat during this period, suggesting that the non-heat-related group may include unidentified heat-related deaths. Under-attribution of heat-related deaths in vital statistics data is a widely recognized concern (Henderson, Lamothe, et al., 2022).

4.5 Limitations of this study

This study had important limitations. First, this is a case-only study comparing individuals who died during the EHE with those who died before the EHE. The reported effect estimates cannot be interpreted as risk and protective factors as could be done for a case-control study comparing EHE deaths with EHE survivors. Instead, an increased odds ratio indicates that the prevalence of a specific condition was higher among those who died during the EHE compared with those who died during more typical summer conditions. Second, typical weather deaths were drawn from the 2012-2020 period, and may not reflect the deaths that would have occurred in 2021 in the absence of the EHE. Changes in population demographics and therapeutics may lead to different point prevalence of the chronic diseases we examined over time.
Third, we were not able to fully separate deaths from heat-related and non-heat-related causes during the EHE because of reporting delays between BCCS and BCVSA. However, most studies of EHE mortality cannot do any such subgroup analyses because heat-related deaths often go unrecognized and unattributed in vital statistics data (Henderson, Lamothe, et al., 2022). In comparison, the health impacts of the 2021 EHE in BC were recognized early in the event, prompting the BCCS to remind clinicians to report potentially-heat-related deaths. As a result, many excess deaths during the EHE were attributed to extreme heat (BC Coroners Service, 2022a). This study represents an important advancement over prior work because it was able to examine the prevalence of chronic diseases separately among heat-related deaths, deaths with pending causes, and non-heat-related deaths.
Finally, we were not able to separate the effects of air pollution from those of heat. However, BC experienced long periods of very poor air quality in July 2017 due to extreme wildfire smoke, and there was no substantive increase in excess mortality (Figure 1; Table S1). In nearby Washington state, poor air quality from wildfire events has been associated with a 4.4% increase in all-cause mortality (Y. Liu et al., 2021), which is substantially lower than the 95% increase in daily mortality observed during this EHE. However, the effects of air pollution and heat may also interact to increase the overall risk of death. For example, approximately 2000 of 11000 excess deaths during a 2010 heat wave in Russia were attributed to an interaction between wildfire smoke and high temperatures (Shaposhnikov et al., 2014). During the 2021 EHE in BC, high ground-level O3 concentrations in greater Vancouver (Henderson, McLean, et al., 2022) and higher concentrations of PM2.5 throughout the province (Table S1), likely contributed to the increased deaths with pending cause (R99) among those with chronic obstructive pulmonary disease. While such deaths might eventually be coded with a non-heat-related cause, air pollution concentrations were increased by the hot weather, such that these deaths were still associated with the overall event (Henderson, McLean, et al., 2022).

4.6 Conclusion and future directions

We found that schizophrenia was most strongly associated with the odds of death during the 2021 EHE in BC. Other chronic diseases associated with increased odds of death were chronic kidney disease and ischemic heart disease, as well as a higher overall burden of diseases. The BCCS reported that most heat-related deaths occurred inside private residences, often among those who lived alone in multi-unit buildings without functioning air conditioning (BC Coroners Service, 2022a). While we do not have individual-level information on socioeconomic status, our previous study on this EHE showed a doubling of odds in the most deprived urban communities (Henderson, McLean, et al., 2022).
Together, these details paint a stark picture of those who were likely at the highest risk: people with mental and physical illnesses and disabilities who were economically and socially marginalized. These results, based on relatively rapid case-only analyses, should now be confirmed and extended with more detailed analyses in a case-control design. Our future work will compare individuals who died during the EHE with similar individuals who survived matched on age, sex, geographic region, setting (e.g., long-term care, emergency room, community), and other indicators using the same data platform. This will allow further examination of specific chronic diseases, the burden of chronic disease, and other potential risk factors such as dispensed pharmaceuticals.
As with our first analysis on this EHE (Henderson, McLean, et al., 2022) we undertook this rapid case-only study to generate actionable evidence for public health policy and practice. These studies suggest that some interventions should be directed to those with severe mental illness, finding ways to reach susceptible people in their homes. Public health should partner with organizations that serve such populations to facilitate outreach before, during, and after EHEs (Kafeety et al., 2020). In the short-term, targeted interventions should focus on health checks during hot weather and rapid transition to indoor environments with safe temperatures. Longer-term interventions are also necessary to mitigate the future risks of climate change, including changes to building codes (Crawford, 2022) and strategies that promote passive cooling through construction and urban design (Kenny et al., 2018).