Heat, street drugs and prescribed ones: two studies find the same vulnerability

Heat-health research usually sorts people by age, housing and chronic disease. Two studies published three days apart in early October sort them a different way: by what is in their bloodstream. One looks at drug overdose deaths across the United States over two decades of summers [s1]. The other looks at prescribed antipsychotics during a single extreme heat event in British Columbia [s2]. They are different designs answering different questions, and they land in the same place.

The national picture

The first study asked whether heat raises drug overdose mortality across the United States, using monthly, county-level data from 1999 to 2020 [s1]. Prior evidence had come from local jurisdictions in the US and Canada; the national picture was undetermined [s1].

The design was observational: a fixed-effect model estimated with linear regression, covering all counties in the continental US during June to September, and accounting for time-invariant county-month factors and state-level time-specific shocks [s1]. Deaths came from the National Center for Health Statistics restricted Vital Statistics mortality data, aggregated to county-level rates [s1]. Heat exposure was the monthly average maximum heat index in degrees Celsius, from the CDC's National Environmental Public Health Tracking Network [s1].

A one-degree Celsius increase in the heat index was associated with an increase in all drug overdose mortality of 0.0098 deaths per 100,000 population (95% CI, 0.0057 to 0.014; P<0.001) [s1]. Significant effects appeared for deaths related to opioids (0.0060 deaths per 100,000; 95% CI, 0.0026 to 0.0095), cocaine (0.0028; 95% CI, 0.0012 to 0.0045) and psychostimulants (0.0028; 95% CI, 0.0015 to 0.0041) [s1].

Those coefficients look tiny because they are per 100,000 people per degree. Scaled to the country, the authors estimate that approximately 150 excess drug overdose deaths occurred per year during the hottest periods because of heat exposure [s1].

The effect was not evenly spread. Larger impacts were observed after 2013, in counties with greater levels of social vulnerability, and in suburban and urban counties [s1].

The individual-level picture

The second study looked at the 2021 Western North America extreme heat event in British Columbia, which the authors describe as unprecedented [s2]. People with schizophrenia are known to have increased mortality during extreme heat events, including that one [s2]. The question was what, among the many things that differ between people with schizophrenia, was associated with dying.

Investigators identified everyone with schizophrenia in British Columbia through an administrative chronic disease registry and linked them to antipsychotic dispensations in the 30 days before the heat event [s2]. They then compared 137 people who died during the event with 57,394 who survived, using multiple logistic regression that also included age, sex, income assistance, comorbidities and indicators of schizophrenia severity [s2].

Dispensation of any antipsychotic was associated with increased mortality during the event (odds ratio 2.43; 95% CI, 1.52 to 4.01) [s2]. The association was driven by haloperidol, clozapine and zuclopenthixol [s2]. Risk was higher still with dispensation of two or more antipsychotics in combination (odds ratio 4.05; 95% CI, 2.41 to 6.98) [s2].

Other variables associated with mortality during the heat event were being on income assistance, age, sex, having a mental health-related emergency department visit, and time since entry into the disease registry [s2].

What neither study can show

Both are observational, and both carry the limits that come with that.

The Addiction study measures monthly average heat against monthly county overdose rates [s1]. It cannot say that any particular death happened because of heat, and a monthly average smooths over the short, intense episodes that kill people. Its estimate of roughly 150 excess deaths a year is a model output, not a body count [s1].

The British Columbia study compares 137 deaths against more than 57,000 survivors [s2]. It adjusts for severity indicators, but people prescribed two antipsychotics differ from people prescribed none in ways no registry fully captures — illness severity, housing, isolation, physical health. Confounding by indication is the standing objection to findings like this, and the authors' own framing is careful: people with schizophrenia are at risk during extreme heat events because of multiple overlapping factors, including antipsychotic medications [s2]. There were also significant non-pharmaceutical risk factors [s2].

Nothing in either paper is a reason for anyone to change or stop a prescribed medication. That decision belongs with a prescriber, and neither study examined what happens when treatment is changed.

Why the pairing matters

Read together, the two studies describe a category of heat vulnerability that heat-action plans have not traditionally centred: people whose thermoregulation, judgement or behaviour is altered by a substance, whether prescribed or not.

The Addiction study's finding that impacts were larger in more socially vulnerable counties, and in urban and suburban ones, points at the same populations that overdose prevention already serves [s1]. The British Columbia study points at a group identifiable in advance from a prescription registry [s2]. Both are, in principle, reachable — the second more precisely than the first.

What to watch

Whether the British Columbia association replicates in another extreme heat event and another health system, and whether the drug-specific pattern — haloperidol, clozapine, zuclopenthixol — holds up [s2]. And whether national overdose surveillance begins to treat heat as a modifier worth reporting alongside supply and treatment access [s1].

Sources

• [s1] Heat exposure and drug overdose mortality in the USA, Addiction, 6 October 2025. https://doi.org/10.1111/add.70191
• [s2] Antipsychotics and other risk factors for mortality among people with schizophrenia during an extreme heat event: a population-based case-control study, Scientific Reports, 3 October 2025. https://doi.org/10.1038/s41598-025-17591-0