Environmental factors and daily suicides: a time series analysis

Introduction

Suicide is an act of self-harm with fatal consequences.^1,2 It is a rising global health problem and is responsible for 50% of all violent deaths.³ The World Health Organization (WHO) estimates that the annual number of suicides worldwide exceeds the combined total of deaths due to homicide and war, despite under-reporting in most countries. This underscores the importance of understanding the biological processes underlying suicide to develop effective prevention strategies.⁴

The pathophysiological processes underlying suicide are complex and multifaceted, involving an interplay of individual, biological, sociological, and environmental factors.^5,6 Air pollution and climate change are recognized as major global issues that have significant detrimental effects on human health. Consequently, air pollution and harsh weather have been linked to several mental health issues such as suicide.^7,8

However, previous studies investigating the influence of environment on suicide occurrence have yielded conflicting results.⁹ Most studies reported higher suicide rates in spring and summer.^10–12

A variety of other meteorological variables, such as sunshine duration, temperature, atmospheric pressure, precipitation, and humidity, have also been examined for their potential association with suicide.^13,14

According to WHO, 24% of current diseases can be attributed to environmental degradation.¹⁵ This underscores the urgency for further research into the complex interplay between environmental factors and mental well-being, particularly regarding suicide risk.^16–18

Despite conflicting findings, research on the link between air pollution and suicide is accumulating. A 2021 meta-analysis encompassing 18 studies found a significant overall association between exposure to air pollutants and suicide, with risk ratios ranging from 1.01 to 1.03.¹⁹ Although the precise mechanisms underlying this association are still under investigation, several hypotheses suggest potential pathways. Both air pollution and high temperatures have been linked to neuroinflammation and central nervous system dysfunction, potentially disrupting serotonin metabolism and impacting mood regulation.¹⁹ These factors, in turn, could contribute to increased vulnerability to suicidal thoughts and behaviours.

However, studies investigating the interactions between air pollution and other environmental factors, such as meteorological variables, remain limited. This highlights the need for further research specifically designed to disentangle these complex relationships and to elucidate the combined impact of environmental factors on suicide risk. By addressing these knowledge gaps, we can develop more effective prevention strategies and mitigate the environmental burden on mental health.

From this perspective, this study was conducted to analyse the association between weather conditions, air pollution, and the risk of suicide.

Methods

A two-time series analysis study investigated the relationship between environmental data and suicide cases, over a 7-year period from January 1, 2010, to December 31, 2016.

This study was carried out in the Sousse region, which is located in the Tunisian East Center and is one of the largest cities in Tunisia. Sousse’s location, at 35° 50′ north and 10° 38′ E, places it within the lower semi-arid bioclimatic zone. The city enjoys the influence of the Mediterranean Sea, resulting in a distinct coastal climate. Temperatures throughout the year remain mild, ranging from an average of 11.2°C in January (the coldest month) to 26.3°C in August (the hottest month). Monthly thermal variations are usually less than 10°C, which contributes to a consistent and comfortable climate.^20,21

The study included all deaths investigated by the forensic department of Farhat Hached University Hospital during the study period. This department serves as the central repository for all deceased individuals in the Sousse Governorate, whose deaths were deemed violent or occurred under suspicious circumstances.

Suicide has been defined as the violent death of a person caused by the voluntary intervention of the person himself to cause his own death and where preliminary data from the judicial inquiry disclosed were in favour of this hypothesis.^1,2

We did not include accidents related deaths, homicides, cases of suicide registered in other governorates than Sousse and suspected cases that were not confirmed by forensic autopsy data and preliminary judicial inquiry.

Two separate synoptic sheets were developed in this study. A medical record including the identification of the victim (age, gender, marital status, origin, and history), the circumstances of the death (date, place, circumstances, means used, and causes), autopsy data, and an environmental data sheet.

Climate data, including temperature, humidity, pressure, and precipitation, were collected from the Monastir-Airport meteorological station in the Sousse region (14 km away).

The temperature was measured under the cover and expressed in degrees Celsius (°C). The relative humidity reflects the amount of water vapor in the atmosphere. It is the ratio of the water vapor pressure actually measured to that measured if the air was saturated at the same temperature. The results were expressed as a percentage (%). Atmospheric pressure is the weight of air on the Earth’s surface and is expressed in hPa (hectoPascal). Precipitation is the amount of rain recorded in millimetres.

For each day included in the study, environmental data were recorded for the event day (J0), one day prior (1-day lag), and two days prior (2-day lag).

Air pollution data were obtained from the National Agency for Environmental Protection’s modelling platform, which simulates pollutant dispersion using mathematical models and specialized software. The platform data used in this study underwent rigorous validation and calibration for two years.

The specific air pollutants measured included carbon dioxide (CO₂), ammonia (NH₃), sulphur dioxide (SO₂), nitrogen oxides (NO_x), volatile organic compounds (VOCs), total suspended particulate matter (TSP), and particulate matter (PM) with aerodynamic diameters less than 10 μm (PM10) and less than 2.5 μm (PM2.5).

In our study, the ARIA Impact software, version 1.8 which is open access software (https://aria-impact.software.informer.com/download/) was launched in the Sousse region. This is a Gaussian Cartesian statistical model. The calculation of the deposition of pollutants on the ground makes it possible to take into account two types of pollutants: passive waste gases and dust sensitive to the effects of gravity.

ARIA Impact software uses Pasquill standard deviations. The calculation assumptions of this model are that the turbulence is homogeneous in the lower layers, the measurement of the site is representative of the entire field of calculation, the density of the pollutants is close to that of the air, and the vertical component of the wind is negligible in front of the horizontal component. These hypotheses allow rapid visualization of the orders of magnitude of pollution in domains of 1 to 30 km.

The pollution indicator was introduced into the model using temperature, relative humidity, wind speed, precipitation, and atmospheric pressure, in addition to the city’s topographical data.

The model must be able to simulate air quality and, thus, the dispersion of pollutants emitted by stationary and mobile sources in different weather situations, including urban convection. The hourly concentrations of each pollutant were calculated and validated.

Subsequently, the hourly concentrations of each pollutant were converted using Excel into the daily average concentrations required for our analytical study.

The data were analysed using SPSS 21.0. Qualitative variables were assessed using frequency analysis, while quantitative variables were evaluated for their means, standard deviations, medians, and extreme values. Paired and independent series means were compared using the Student’s t-test, while the one-factor ANOVA test was employed for comparisons involving multiple means.

Frequency comparisons were conducted using Pearson’s chi-square test. The relationship between the two quantitative variables was investigated using the Pearson’s correlation coefficient.

To assess seasonal effects on suicide, suicide dates were categorized by season. Multiple binary logistic regression was performed for multivariate analysis, with the absence or presence of suicide on a given date as the dependent variable. The significance threshold (p) was set at 0.05 for all statistical tests.

This study was approved by the Ethics committee of Sahloul Teaching Hospital on April the 20^th 2022; under the number HS 16-2022.

Results

A total of 118 suicide cases were identified during the study period. The year 2012 saw the highest number of suicides (19.5%) without a statistically significant difference across years (p=0.14). The sex ratio was 3.53. The average age at death was 33.1±13.9 years. Among the suicide cases, the majority (58.5%) were single and resided in rural areas (61%).

Psychiatric histories were reported in 22% of the cases. Nervous depression was the most common pathology (55.3%). The location and mechanism of most reported suicides were home (44.9%) and hanging (45.8%). Suicide occurred as a result of conflict in 33.1% of the cases, financial difficulties in 18.6%, and occupational problems in 13.6%.

The largest number of suicide cases was observed during the summer season, with 40 cases (33.8%), a statistically significant difference (p=0.021).

Comparing the occurrence of suicide in the autumn with other seasons, a statistically significant decrease in risk was observed during this season with an OR=0.48 (CI 95% [0.28-0.82] and p=0.006.

A comparison of mean temperatures between days with and without suicide revealed significantly higher temperatures on days when suicide occurred (p=0.043). This finding was replicated for mean temperatures one day prior to the suicide events (p=0.03).

Table 1 summarizes the results of comparing the mean meteorological parameters, including mean temperature, humidity, atmospheric pressure, and precipitation, between days with and without suicide.

Comparison of mean PM10 and PM2.5 concentrations between days with and without suicide showed significantly higher levels on suicide days (p<10^-3 for both pollutants). Conversely, lower concentrations of NH₃ and SO₂ were observed on suicide days, although the differences were not statistically significant.

No statistically significant differences were found in the mean concentrations of the other studied air pollutants between the days with and without suicide. The same finding was highlighted by comparing the mean concentrations of air pollutants one day and two days before the occurrence of suicide.

A comparison of the means of the measured air pollution parameters (CO₂, NH₃, SO₂, NO_X, VOC, TSP, PM10, and PM2.5) between days with and without suicide is summarized in Table 2.

After multiple binary logistic regression, autumn and NH₃ decline were independently associated with a decrease in suicide risk (p=0.045; ORa: 0.57 IC95% [0.33-0.98], p=0.009; ORa: 0.24 CI95% [0.08-0.71], respectively), whereas a 1 ppm increase in PM10 was independently associated with a 2.5% increase in the risk of suicide (p<10⁻³; ORa: 1.02 CI95% [1.01-1.03]).

Discussion

According to WHO, suicide represents an annual global mortality incidence of approximately 14.5 per 100,000 inhabitants.²² It is a serious public health issue that deserves to be thoroughly analysed, especially in terms of risk factors.^23,24

In our study, the largest number of suicide cases was observed during the summer (33.8%), with a statistically significant difference (p=0.021). Similarly, a statistically significant decrease in risk was observed during falls (p=0.006).

Similar results to ours were reported by Amiri et al.,²⁵ who observed that the largest number of suicide cases occurred during the summer period, with a statistically significant difference (p=0.006). Other studies have reported peaks in spring and early summer.^11,26 However, to date, results are not yet conclusive, as some studies suggest a higher suicide rate in autumn and summer and a lower suicide rate in winter.^27–30 Others report higher rates in the spring and summer^31–33 or spring and autumn³⁴ or only in the spring.^35–38

The seasonality of suicide was first proposed by scientists in the 19th century and has since been observed in various countries and geographical areas.³⁹ Its underlying causes vary, including daylight exposure, chronobiological aspects of mood disorders, and socio-economic influences.⁴⁰

One of the most compelling hypotheses suggests that serotonin dysfunction plays a crucial role and is likely multifaceted, involving a complex interplay of factors, including temperature, in predisposing individuals to self-injurious behaviour.^41–43 This theory postulates that climatic factors can exacerbate vulnerability of the nervous system, potentially leading to increased rates of suicide attempts in spring and early summer.²⁵ Ultimately, understanding the intricate interplay between climate change, serotonin function, and suicidal ideation/behaviour provides valuable insights into seasonal patterns of suicide.⁴⁴

Similarly, significantly higher temperatures were reported during days of suicide (p=0.043). Several studies have reported similar results.^14,45–49 A recent meta-analysis, including 32 studies, 17 time series (53.1%), and 9 cross-case studies (28.1%) conducted in 26 countries, found a significant association between temperature and suicide with an estimated relative risk of 1,09 despite different approaches to exposure measurements.¹⁹

In Kazakhstan, a 1°C increase in the mean apparent temperature was associated with a 2.1% increase in suicide rates.⁵⁰ Similarly, Brazilian time-series analysis revealed a 2.28% increase in total suicides for each 1°C increase in weekly mean temperatures.⁴⁹ However, Swiss researchers analysing long-term monthly data on suicide and weather found that the association is not driven by hot temperatures but by the absence of cold temperatures, suggesting that the lack of cold plays a more significant role.⁵¹

While the exact mechanism underlying this association remains unclear, the hypothesis implicating the serotonin system is considered the most promising explanation.⁵² Evidence suggests that serotonin deficits (5-HT) are associated with non-fatal and fatal suicide attempts.⁵³ Additionally, research indicates that l-tryptophan, a precursor of serotonin, decreases at high ambient temperatures.⁵⁴ Several studies have suggested that ambient temperature-influenced changes in serotonin function may contribute to impulsivity, aggression, and potentially suicidal behaviour.⁵³

In our study, no relationship was found between suicide and humidity, which is consistent with the results of Lee et al.,³⁵ Deisenhammer et al.⁵⁵ and Kurokouchi et al.¹³

Digon and Bock⁵⁶ were the first to report that a drop in atmospheric pressure is correlated with an increase in suicide rates. This result has not been confirmed in other studies,⁵⁷ including ours.

In our study, no association was reported between suicide rate and precipitation, which is consistent with the results of Asirdizer et al.⁵⁸ and Lee et al.³⁵ In contrast, Inoue et al.⁵⁹ reported that suicide rates increased with heavy precipitation.

In our study, after multiple binary logistic regression, a 1 ppm rise in PM10 was independently associated with a 2.5% increase in the risk of suicide (p<10⁻³;ORa:1.02 IC95% [1.01-1.03]).

In the literature, the results of air pollution and suicide differ from one study to another. This mismatch can be explained by geographical variations, climatic conditions, culture, socioeconomic factors, and suicidal behaviour.⁶⁰

According to a recent meta-analysis, a significant increase was reported for PM2.5, PM10, and NO₂ pollutants, with RR of 1.02 (95% CI:1.00, 1.05), 1.01 (95% CI:1.00, 1.03) and 1.03 (95% CI:1.00, 1.07) respectively. However, no significant association was noted between O₃, SO₂ and CO.¹⁹

Lin et al.⁶¹ found a significant increase in suicides related to PM10, NO₂, and SO₂ exposure. A cross-case study, conducted in Utah, USA, including 1,546 suicide cases, found an increase in the number of suicides related to exposure to PM2.5 and NO₂ on the same day and during the 2–3 days preceding suicide.⁶² These effect-offsets would save time for early action to prevent suicide. Awareness of suicide prevention and special attention should be given to high-risk individuals, such as those with a history of mental illness, on days of high pollution levels.

Kim et al.⁶⁰ reported that a high risk of suicide was associated with higher levels of NO₂, SO₂, and PM10 over several days. However, no association was found between PM2.5 and suicide risk. Merrill RM⁶³ and Liu et al.⁶⁴ reported a significant association between suicide risk and exposure to fine PM2.5.

A recent time-series study conducted in California assessed the association between O₃ and fine particulate matter (PM2.5) and various mental illnesses, including suicide, between 2005 and 2013. The authors observed significant cumulative 7-day associations with suicide for every 10 ppb increase in O₃. This association persisted after adjusting for seasons.⁶⁵

However, According to Astudillo-Garcíaou CI,⁶⁶ no association was noted in their time series study between suicides and exposure to PM2.5, PM10, O₃, NO₂, and SO₂, taking into account meteorological variables (air, temperature, and relative humidity). This result is consistent with that of Fermandez-Nino et al.¹⁸

The neuroinflammatory hypothesis is considered to be the primary explanation for the impact of air pollution on suicide risk. Air particles act as irritants and trigger both systemic and local inflammatory responses.⁶⁷

Several researchers have proposed that high levels of air pollution induce cytokine production, leading to neuroinflammation, dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis, and alterations in neurotransmitter levels through direct and indirect pathways. These pathways, in turn, are potentially implicated in the development of depression, suicidal behaviours, or both.^16,62 Further research investigating the neurophysiological response to air pollutants is crucial to fully understand their impact on suicide risk.⁶⁰

Importantly, air pollution and temperature can be strongly correlated, though the correlation varies depending on the pollutant. This correlation poses a potential confounding factor in disentangling the independent effects of temperature and air pollution on suicide risk. Both these factors are associated with neuroinflammation and serotonin neurotransmission, making it challenging to isolate their individual contributions.¹⁹

This study holds the unique distinction of being the first in Tunisia to investigate the association between weather, air pollution, and suicide incidence. By analysing a 7-year time series of weather and air pollution parameters alongside suicide occurrences, our research offers valuable insights into the potential connections between these factors. This time-series approach is particularly well suited for studying the short-term relationship between environmental parameters and acute health events, providing a more robust framework compared to cross-case studies reliant on comparing two periods with self-referential cases.⁶⁸

While weather data were collected from a station located 14 km away in Monastir, existing research suggests that this distance allows for valid extrapolation to the Sousse region.⁶⁹ However, we acknowledge the inherent limitations of not having access to the detailed microclimate and indoor conditions experienced by each individual. Ideally, weather data should be obtained for each household to analyse the impact of meteorological and seasonal factors on suicide occurrences with greater precision. However, acquiring such data is impractical.

Our study leveraged data from the National Agency for Environmental Protection’s modelling platform. This approach incorporates location-specific geographical features, such as population density and proximity to roads. This provides a more nuanced understanding of small-scale spatial variations in air pollution levels, overcoming the limitations of relying solely on data from fixed monitoring stations. Additionally, the one- to two-day lag applied to suicide data allowed us to assess the cumulative effects of pollution exposure beyond immediate impacts.

Conclusion

Suicide is a common cause of death, worldwide. It is usually associated with psychiatric, psychological, and cultural factors. However, other factors that are less frequently reported, such as the environment, should be considered. Indeed, climate change and air pollution seem to have serious impacts on mental health. Health policies should consider these factors when establishing preventive measures. Thus, health policies should consider all of these factors in future preventive programs.

Ethics and consent

The Ethics Committee of Sahloul Teaching Hospital approved this study on April the 20th 2022; under the number HS 16-2022 and provided a waiver of consent.

Author contributions

All authors contributed to the study conception and design.