Higher temperatures increase emergency department visits for mental illness and suicide rates


Economists at the University of Massachusetts Amherst and California Polytechnic State University have found that higher temperatures increase emergency department visits for mental illness, suicide rates and self-reported days of poor mental health.

Further, the researchers found no evidence of effective adaptation among American populations to these negative effects of heat.

They also warn their findings suggest that both warming local climates and increases in extreme heat events will contribute negatively to population mental health.

In a new study published in the Journal of Health Economics, Jamie Mullins and Corey White report that their estimates indicate that a 1-degree Fahrenheit increase in mean monthly temperature results in a 0.48% increase in the monthly mental health-related emergency department (ED) visit rate and a 0.35% increase in the monthly suicide rate.

“Consistently across outcome measures, we find an increasing, quasi-linear relationship between temperatures and mental health,” they write.

“Mental health appears to deteriorate with increased temperatures across the range of temperatures considered.”

Noting that “the consistent, increasing nature of our estimates across the full range of temperatures suggests that both warming temperatures and more frequent, high-temperature events will harm mental health everywhere such changes occur and irrespective of current local climate conditions,” Mullins and White also sound the alarm for the potential impact that the warming climate may have on mental health.

They also present evidence that heat-disturbed sleep serves as an active channel in the identified temperature-mental health relationship.

Mullins, assistant professor of resource economics at UMass Amherst, and White, assistant professor of economics at Cal Poly, examined ED visits for diagnoses related to mental health in California from 2005-16, and suicides in the U.S. from 1960–2016, as well as daily temperatures for the U.S. over the whole study period.

Ultimately, they found that not only did the incidence of negative mental health outcomes increase in response to additional hot days, but that they decreased in response to additional cold days.

For ED visits in California, the authors’ estimates imply that relative to a day in the 60–70 F range, one additional day less than 40 F leads to a 0.39% decrease in the monthly mental health ED visit rate, and that one additional day greater than 80 F leads to a 0.30% increase, or 0.43 fewer and 0.33 more mental health ED visits per 100,000 residents, respectively.

Across the U.S., the economists’ estimates imply that one additional day below 30 F leads to a 0.43% decrease in the monthly suicide rate, and one additional day above 80 F leads to a 0.24% increase, or 0.0044 fewer and 0.0025 more suicides per 100,000 residents, respectively.

They also found no evidence that any populations have effectively adapted to the negative mental health effects of high temperatures; instead the temperature relationship is stable across time, baseline climate, air conditioning penetration rates, accessibility of mental health services and other factors.

They report that their estimates are also quite stable across regions with different climatic norms.

Mullins and White posit that temperature-disturbed sleep is an active mechanism for their observed changes in mental health, and they provide several pieces of evidence to support this notion.

In particular, they show that the character of the relationship between sleep disturbances and hotter temperatures closely mirrors that between negative mental health outcomes and hotter temperatures.

“While it is possible that temperature independently affects both sleep and mental health in a similar manner,” they write, “we argue that this is not likely to be the case as other research documents a strong link between poor sleep and measures of mental health that are unrelated to temperature.

“Understanding exactly how to address temperature-induced changes in mental health requires understanding the mechanisms through which the impacts operate. We provide evidence suggesting that sleep is one of the primary mechanisms.

A direct policy recommendation stemming from our research is for mental health providers to ensure patients get adequate sleep during periods in which sleep is likely to be disturbed (such as a heat event).”

Mortality and morbidity follow a seasonal pattern [1]. Seasonal patterns have been observed for completed suicides in many parts of the world with peaks in spring and early summer [2].

Schizophrenia [3] has been reported to follow a seasonal pattern, but there is a lack of evidence regarding seasonality patterns for admissions due to schizoaffective disorders and psychosis [3,4].

Recently other mental health outcomes have been shown to be associated with high ambient temperatures, such as hospital admissions due to any mental disorder [5], and higher rates of hospital admissions due to mania [6,7], and involuntary psychiatric admissions [8] have been reported during warmer and sunnier months.

Much of the literature about mental health and temperature pertain to heat waves, and admissions for mental and behavioural disorders increase during heat waves, more so in the elderly [9].

The increase in the number of mental hospital admissions during heat waves is apparent also when stratifying by diagnoses such as organic mental illnesses; dementia; mood (affective) disorders; neurotic, stress-related, and somatoform disorders; disorders of psychological development; and senility, as well as anxiety, behavioural and personality disorders [10,11].

Furthermore, mental and behavioural emergency department visits increase during heat waves for schizophrenia and mood disorders [12].

Relatively few studies have reported mental health outcomes in the cold season [4]. Associations with cold extremes have been observed for hospital admissions for dementia, senility and neurotic disorders [10,13], whereas another study observed no association [12].

Lower winter temperatures were associated with increased mortality risks in individuals younger than 65 years with a history of mental illness or drug abuse in a Swedish register study [14].

Considering global climate change, understanding the effects of climate on human mental health is of great interest to public health [15].

Heat and mental health are not well-studied in colder climates, and mental health issues are a leading cause of illness and costs to society [16].

Determining environmental factors which aggravate mental illness and symptoms of mental illness could help prevention and planning efforts within the health care system in which resources are limited.

The aim of the current study was to investigate short-term effects of high and low ambient temperatures on the worsening of mental illness by studying the association between the daily mean temperature and the daily number of psychiatric emergency department visits in a northern climate.


We found increased psychiatric emergency room visits at temperatures at the 95th percentile in the warm season, for both lags 0–3 and lags 0–14. In the cold season, for lags 0–14 and lags 0–21, temperatures at the 5th percentile were associated with an increased number of PEVs, however, not to a statistically significant extent.

The study setting where the clinic is located is a temperate Swedish west coast climate with relatively warm winters and cool summers, frequent rains and predominant westerly winds. Compared to other study regions in the body of literature in this field Gothenburg has moderate temperatures with the 99th percentile temperature at 21 °C.

A study from Shanghai [5] found that mental disorder hospital admissions increased at a threshold of 24.6 °C, which is near the highest temperature observed in the current study. However, a register-based study from Denmark, which has a similar climate to Gothenburg, observed significant correlations between higher temperature and the risk for admissions for manic state as well as by season, meaning that more patients were admitted in a manic state during the summer months [7].

In an Israeli study, the risk for admission due to schizophrenia was increased with a higher monthly temperature, but the temperatures were much higher than in the current study at ranges from 14 to 34 °C [3].

A common feature for these studies is that they assumed a linear association between the exposure and outcome across a temperature range, whereas several studies have shown that temperature health effects often occur at extreme temperatures in a u-shaped curve [5,19].

Incidentally, Page and colleagues observed a near-linear association between ambient temperature and suicide in men and violent suicide [24].

Furthermore, considering the temperature differences between study settings, it is not certain that the biological mechanisms that govern the observed effects are the same, although humans use a variety of physiological, technical, and behavioural adaptation methods to offset climate exposure [25] and it is well-established that optimal temperature (at which the health outcomes are minimised) varies across countries and populations [19].

Although cold weather effects are reported in relatively few studies of mental health outcomes, Peng et al. [5] observed an increased risk for mental disorder admissions in cold weather at longer lags. Furthermore, Orru and Åström reported increased mortality due to assault in cold weather, suggesting an increase in agitation, but the results did not reach statistical significance [26].

Our results provide tentative, albeit non-significant, evidence of an effect during extremely hot summer days, with the number of PEVs increasing with 9% on such days. In previous heat wave studies, temperatures reached or exceeded 35 °C for three or more consecutive days [9,11]. However, the Chicago heat wave studies cite a high heat index [27,28], as high air humidity was a significant factor in the severity of the event; however, the daily maximum temperature still exceeded 35 °C during four days.

Other studies have reported thresholds for negative effects on mental or psychosocial emergency department visits at 22.5 °C in Quebec, Canada, [29] and 26.7 °C in Adelaide, Australia, for mental health admissions and mortality in elderly with mental illness [10].

Not only hospital admissions due to mental disorders increase during heat waves, but also those with pre-existing mental illness are a risk factor for dying in heat waves [7,10,13,14,15]. Furthermore, the use of, and non-compliance to, psychotropic drugs such as antidepressants, and anti-Parkinson’s disease medication have been identified as risk factors of heat stroke deaths and heat-related mortality [30,31].

The mechanisms which underlie heat-related increases in mental hospital admissions are not yet fully understood. The major monoamine metabolites of serotonin and dopamine are affected by meteorological factors such as heat [32]. Psychiatric medications such as SSRI, and neuroleptics have an association with heat intolerance [18].

The risk estimates from the current study of mental health outcomes follow a similar trajectory to risk estimates for somatic illness and mortality, with high risks associated with cold temperature occurring at longer lags, and with heat at shorter lags [14,33,34]. The size of the risk estimates found in the current study are moderate compared to those found for heat wave mortality, where individuals with a pre-existing psychiatric illness had OR 3.61 (95% CI, 1.3–9.8) of dying in a heat wave, and OR 1.90 (95% CI, 1.3–2.8) for individuals who took psychotropic medications. However, in those studies, somatic illness most likely accounted for the main effect or mediated the association [30].

An epidemiological study such as the current one cannot prove causation, and we are thus unable to determine if there is a direct physiological effect of high or low temperature on mental illness, or if the observed association is mediated by other factors.

Three examples could be;

(1), symptoms of underlying or pre-existing somatic disease may worsen in hot or cold temperatures; an example of this is respiratory disease, a leading cause of years of life lost in the population of mentally ill [35],

(2) socioeconomic circumstances such as homelessness,

(3) weather-associated changes in behaviour, such as spending more time outside when it is warm increasing exposure to heat [36].

A strength of the current study is that the data are based on administrative-register data from a single clinic serving an entire region, minimizing the risk of bias from misclassification. Health care seeking behaviour is unlikely to be biased by socioeconomic status compared to many other countries as fees are minimal. However, the open-door policy of the clinic can mean that not all registered visits are emergencies which may introduce random errors into the data, but is unlikely to bias the results.

A further consequence of the different policies in reaching psychiatric care units might mean that the results are difficult to replicate, but, since the outcome is a proxy for worsening in health, the relative increase should be generalizable to populations in areas with similar climate.

The exposure in the current study is the ambient temperature measured at a central location. The measuring station does not correctly reflect the exposure of all individuals in the study.

However, we assume a similar distribution across the study area, as our interest lay in the change between days with high and low temperatures. We, therefore, consider the exposure measurement adequate for the present study.

Due to missing temperature data, we had to replace data with that from a nearby measuring station, but as these data were during a period of moderate temperatures, we do not believe they skewed the results.

We did not adjust for daily air pollution concentrations, as it is a mediator in the causal pathway [37]. We have previously shown that increasing levels of air pollutants seem to be another risk factor for increased PEVs [18].

In our study, we did not stratify the number of daily PEVs by diagnoses due to lack of statistical power, but also because the accuracy of diagnoses given in the emergency ward may not be optimal as mental health diagnostics can be a time-consuming process [38]. Following the same argument, we did not stratify on other diagnoses than mental health diagnoses, even though increased rates of admission during heat waves have been reported on a diagnosis-specific level [10,11].


In this register-based study, we observed statistically significant associations between extreme temperatures and PEVs in a temperate climate setting. Our results should be verified by others, but there is increasing evidence that individuals with psychiatric disorders should be considered a susceptible group requiring special attention in heat warning systems.

More information: Jamie T. Mullins et al. Temperature and mental health: Evidence from the spectrum of mental health outcomes, Journal of Health Economics (2019). DOI: 10.1016/j.jhealeco.2019.102240

Journal information: Journal of Health Economics
Provided by University of Massachusetts Amherst


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.