Why it’s so difficult to count the number of heat-related deaths

In late July, as parts of the southern U.S. approached 30 days of temperatures exceeding 100 degrees, President Biden took to the podium to announce a plan to mitigate the impacts from extreme heat. During this speech, the president reiterated a truth surprising to most people: Heat kills more people in the U.S. than any other weather-related event.

Almost immediately, social media filled with challenges to this premise. Doesn’t cold weather kill more than heat? What about wildfires? People shared different journal articles estimating different figures for deaths or illness to prove their point that their token event kills or injures more people. Nevertheless, by nearly every calculation, in the U.S., heat remains the largest threat to human health than any other weather-related event.

But there was one thing we could all agree on: The already unacceptable number of deaths and injury from climate change is likely undercounted — perhaps as much as a 50-fold underestimation of heat-related morbidity and mortality. In this U.S. this means instead of the estimated 600 officially reported deaths each year from heat, the true number could be as high as 30,000, especially in a year like 2023.

In this age of big data and sophisticated systems for monitoring and tracking, why don’t we have a better and more concrete way of knowing the impact of climate change on health outcomes? The answer may be complicated, but we can do better.

Official counts of heat morbidity and mortality are based on direct impacts, such as heat stroke and heat exhaustion. However, one of the more complicated aspects of extreme heat is that many of the injuries, illnesses, and deaths from heat exposure come from the exacerbation of underlying cardiovascular, respiratory, or renal conditions, or other indirect impacts. If someone is struck by lightning, the cause of death or injury is obvious. The indirect nature of heat impacts, however, means that while heat is often a significant contributing factor to illness or death, it is rarely cataloged as such.

This is because the systems and processes that we have in place for medical coding are not designed to properly consider exposure. They are designed to catalog clinically observable health outcomes. Much of our data on heat-health impacts comes from emergency department visit data. However, when a patient shows up in the emergency department in cardiac distress, emergency physicians are rightfully concentrating on addressing the urgent and life-threatening situation in front of them, not on investigating and determining attribution from external exposures. The job of emergency department personnel is to stabilize the patient. It is the job of other medical professionals, such as a primary care doctor in follow-up visits, to investigate contributing factors.

Additionally, the International Classification of Diseases (ICD) coding system — which is used by medical professionals to collect, process, and classify data — is also the mechanism through which a medical provider receives reimbursement from insurance companies for medical services. The ICD is a global collaborative led by the World Health Organization and allows for the tracking of morbidity and mortality across nations. This means that, potentially, the ICD is one of the most powerful tools to properly quantify the impact of climate change on health. The most recent version, ICD-10, includes specific codes for climate exposures; however, these codes (E-Codes, standing for “external cause of injury”) are not tied to reimbursement. Like Z-Codes (which are used to capture social determinants of health, such as homelessness), E-Codes that are not tied to reimbursement are less likely to be incorporated into a patient’s electronic medical records. There is a lack of financial incentive for health systems to promote or require their use. Approximately half of the states in the U.S. have mandated E-Code reporting; however, there is no national reporting requirement. Moreover, in states that do mandate E-Code reporting, the data are often incomplete, missing, or incorrect due to a lack of uniformity in policies across jurisdictions.

So, while we have public agency partners, like the CDC and NIHHIS, that are working in good faith to help decision-makers at the state and local level understand the impact of heat on their communities, they are severely limited by a health informatics system that has not adapted or modernized to capture the growing health crises due to climate change.

What we are left with to determine the impact of climate change on health is a dense body of scientific research that uses various statistical estimations to calculate excess morbidity and mortality. These are scientifically accepted and appropriate methods, but studies often rely upon different underlying data and approaches, resulting in different estimations and growing confusion over the magnitude of impact. It can also add to the incorrect assumption that there is a lack of scientific consensus.

All of this leads to the ultimate question: How can we do better?

First, we need to improve the education and training for medical providers about the impact of climate change on health outcomes and the need for effective coding. This would mean that at the patient-provider interface, providers are better able to identify climate exposures, educate their patients on mitigation and adaptation strategies, and connect them with available services, such as programs that distribute fans or help pay for energy so people can run air conditioners. It would also mean that providers are more likely to use the available codes associated with climate exposures as secondary or contributing to the primary diagnosis.

Second, we need to consider ways to incentivize the use of E-Codes. The current and more popular fee-for-service system places the focus is on what is observed clinically. However, there are other reimbursement models that incentivize the inclusion of social determinants of health and climate exposures, without undue burden on the provider, by using computer-assisted coding.

And finally, we need to develop standards and protocols for how to estimate excess morbidity and mortality from climate exposures. This would make it easier to compare results from one study to another or from region or nation to another. This is not a small undertaking as these standards would need to be event specific. In other words, heat-specific standards and protocols would be different than those for flooding events.

Modernizing our data systems is really about organizational and behavioral change. The technology is there for us to make the transition to properly classify and quantify the impacts of climate change on our health and well-being. What we need now is the proper training and incentive to act.

Ashley Ward is the director of the Heat Policy Innovation Hub at Duke University’s Nicholas Institute for Energy, Environment, & Sustainability.