More than 47,000 individuals died by suicide and more than 312,000 received care in the emergency department (ED) for self-harm injuries in 2018, part of a decade-long upward trend (CDC, 2021).1 In addition, in 2020, an estimated 52.9 million adults had a mental illness (21.0% of all adults) and 24.3 million (46.2%) had received treatment for mental health services in the previous year (National Institute of Mental Health, 2022). While the structural risk factors of suicidal thoughts and behaviors have been studied extensively (Stack, 2021), much less is known about the short-term triggers of suicide and suicide attempts (Franklin et al., 2017), and it is unclear to what extent smaller shocks trigger these behaviors. The impulse of the moment is an important factor in suicidality (Baca-Garcia et al., 2005; Baca-Garcia et al., 2001; Yen et al., 2004; Zouk et al., 2006). In a study of attempted but not completed suicides, Simon et al. (2001) reported that a quarter of attempts happened within 5 min of the decision being made. This highlights the importance of ephemeral shocks in triggering suicide. However, the challenge in understanding these short-term factors is the lack of high-frequency data on triggers and outcomes. Furthermore, the events and circumstances which trigger suicidality may be endogenous to an individual’s mental health, such as unemployment and relationship break-ups.

We overcome both of these challenges by examining the impact of a small exogenous health shock on suicide. While physical health is a significant predictor of suicide and poor mental health (Stack, 2021), a small shock may exacerbate the mental health burden to a tipping point, triggering episodes requiring emergency care and, potentially, resulting in suicide. We test this hypothesis by examining the impact of seasonal allergies caused by exposure to pollen on the incidence of suicide. We use novel, high-frequency data on seasonal pollen measurements and suicide counts across many counties in the United States to evaluate a plausibly causal relationship.

Though a naturally occurring byproduct of plants’ reproductive cycles, pollen affects human well-being significantly. While itchy eyes, runny nose, and cough seem like minor symptoms, these also affect quality of sleep, mood, and cognitive ability. Pollen allergies are the single largest reason for missed workdays in the United States and are responsible for $4.9 million in estimated losses to employers each year, a higher figure than for health conditions such as hypertension or coronary heart disease (Lamb et al., 2006). Decades of environmental policy have regulated and mitigated man-made air pollution, resulting in a stabilization and, in some cases, declines in the levels of pollutants. Pollen, however, has never been targeted by environmental policy, nor can it be effectively controlled by it. In fact, the pollen season has been increasing in intensity and duration in the past 30 years due to climate change, with plants blooming earlier in the season (Anderegg et al., 2021; D’Amato et al., 2020; Shea et al., 2008; Zhang and Steiner, 2022).

At the same time, seasonal allergies can be treated using over-the-counter medications such as antihistamines and decongestants, which are relatively inexpensive, widely available, and safe to use. As mental health conditions and suicidality have gained attention and their incidence has accelerated in recent years, we evaluate the role of a relatively minor and easily treatable health shock in triggering deaths by suicide.

Prior work using associative study designs in limited geographic areas has identified a connection between pollen and suicide. Stickley et al. (2017) examined suicides in Tokyo; Jeon-Slaughter et al. (2016) examined nonfatal self-directed violence in Dallas, TX; Qin et al. (2011) and Qin et al. (2013) evaluated suicide in two localities in Denmark. All of these studies found positive associations between pollen and suicide. In the United States, at the national level, Messias et al. (2010) associated suicidal ideation with pollen. Postolache et al. (2005) used monthly data from 1995-1998 and found higher pollen levels associated with increased suicides, though these results were not replicable with more recent data (Woo et al., 2012a). A significant limitation in these studies is that they have severely limited external validity; while they evaluate effects at 30 to 100 particles per cubic meter, our data shows that pollen levels range from 0 to many thousands of particles per cubic meter each season.

We build on this prior work by evaluating the impact of seasonal pollen on suicide using datasets of daily pollen measurements and daily counts of suicides from many counties across the United States to understand how a small environmental shock affects suicide. We use a model richly saturated with county and time fixed effects so that we take advantage of shocks due to pollen levels deviating from county-season averages. Our effects, therefore, are pooled average effects on the count of suicides of a change in the quartile of pollen within a county. This method allows us to estimate effects net of adaptation consistent with our previous research (Danagoulian and Deza, 2025), but departing from other related pollen papers (Akesaka and Shigeoka, 2023). The advantage of using this adaptive method is that if individuals endogenously choose where they live as a function of allergies or receive allergy treatment, our identification evaluates their response to an unseasonably high or low pollen day. A limitation of this approach is that quartiles are not comparable across counties and seasons. We address this limitation by checking the robustness of our estimates to alternative definitions of pollen quartiles.

We estimate a Poisson model to account for the relatively low occurrence of suicides at the county level each day, with a rich set of fixed effects. The granular fixed effects include county- and calendar-specific factors which affect suicidality. We also include a rich set of weather controls to account for the interaction of these with pollen production and suicidality.

Our analysis finds a 4.5% increase in deaths by suicide when pollen levels are in the second quartile of measurement, 5.5% increase in the third quartile, and 7.4% increase in the fourth quartile relative to the first quartile. We find that these deaths occur primarily among individuals with known mental health conditions and those who have received mental health treatment in the past. Consistent with other studies, our effects are strongly driven by white men, however, we find unexpected high vulnerability among Black individuals.

We examine the potential spurious nature of our findings by using an alternate data set. Using Google Trends data, we show that searches for terms including depressive thoughts increase by up to 1.7 units as pollen levels rise. We also show that searches for allergy terms rise on high pollen days, confirming that is the mechanism.

This study makes four significant contributions to the literature on suicidality. First, although the structural causes of suicide are well studied, the short-term triggers of suicide are less examined. Our study illustrates the impact of a widespread shock of relatively small magnitude – seasonal allergies – on suicidality.

Second, this study uses data from 34 metropolitan areas across the United States, allowing us to examine regional differences and draw national conclusions. Data scarcity has constrained the previous literature on the relationship between pollen and suicide to opportunistic studies in narrowly defined geographic areas (such as Tokyo, Dallas, or Copenhagen), limiting the generalizability of their findings. Though the data used in this study are not nationally representative, they include a large number of metropolitan areas allowing us to take advantage of variation in place-based predictors of suicidality.

Third, this study uses robust methodology to evaluate a plausibly causal effect between a small health shock and suicidality. While pollution exposure is endogenous to individual activity and choices, changes in pollen levels vary in a manner that is plausibly exogenous. High-quality and high-frequency pollen and suicide data allow us to overcome the challenges faced by most existing studies in this area which control only for demographic characteristics, weather, and calendar weeks; none include sufficiently granular fixed effects to absorb unobserved time- and county-specific seasonal variation. We employ granular time and county fixed effects to account for unobserved factors which may contribute to both pollen and suicidality.

Finally, our study contributes to the literature in environmental economics by highlighting an overlooked cost of climate change. Most estimates project a 1.3% decrease in GDP resulting from a 2.5C increase in temperatures (Tol, 2018). Our analysis provides evidence of a secondary mechanism of climate change on the economy. In view of these mechanisms, this study allows for avenues of adaptation to the effects of climate change through investments in allergy testing and mental health.

The next section describes the impact of pollen on human health and behavior, provides an overview of known social and economic predictors of suicide, and summarizes the literature on air quality and suicidality. The following section describes the data and the identification strategy. The results are presented in the next section, followed by a discussion and conclusion.