Introduction
Natural disasters, particularly hurricanes, can damage physical infrastructure and harm mental health, and systematic reviews report increased suicidal behavior—ideation, attempts, and death by suicide—after disasters, with female sex, pre-existing mental illness, low income, disrupted access to care, and weakened social support emerging as consistent risk factors.Reference Jafari, Heidari and Heidari1, Reference Kõlves, Kõlves and De Leo2 These conditions are especially pronounced in resource-limited settings, and all were intensified in Puerto Rico after 2 Category 5 hurricanes struck within 15 days in September 2017: Hurricane Irma on September 5 and Hurricane Maria on September 20, the latter devastating the island’s infrastructure and medical services and producing prolonged, near-total loss of power and telecommunications.
Although these risk factors are well established, far less is known about how suicidal behavior evolves at the population level in the months after a catastrophic hurricane, particularly in Puerto Rico. Prior disaster mental-health studies have relied largely on surveys or clinical samples; for example, a large survey of Puerto Rican youth documented substantial distress after Hurricane Maria,Reference Orengo-Aguayo, Stewart and de Arellano3 and emergency-department visits for mental illness increased after Hurricane Sandy.Reference Lin, Lu and Justino4 Few studies have used crisis-hotline records—a continuous, centrally collected, near-real-time source—to track suicidal ideation and attempts across an entire affected population, and to our knowledge none has characterized these outcomes before and after Hurricanes Irma and Maria.
We therefore used 7 years of crisis-hotline surveillance to describe temporal patterns of suicidal ideation and suicide attempts surrounding the 2017 hurricanes, analyzing suicide-related calls to Línea PAS (Primera Ayuda Psicosocial), the hotline operated by Puerto Rico’s Administration of Mental Health and Anti-Addiction Services (ASSMCA), from 2015 to 2021. This descriptive study was designed to establish whether a population-level signal was present and to identify the most affected groups; we hypothesized that the hurricanes would be followed by a marked, temporary increase in suicide-related calls. As climate change increases the frequency and intensity of tropical cyclones,5 such evidence is increasingly relevant to disaster preparedness.
Methods
Design and Data
We conducted a retrospective, descriptive analysis of secondary, de-identified records from Línea PAS, a 24/7 crisis hotline operated by ASSMCA. The dataset comprised all suicide-related calls over 84 consecutive months (January 2015-December 2021), with 2015 as a pre-disaster baseline. Each call was classified as suicidal ideation or suicide attempt and stratified by sex and predefined age bands (0–17, 18–20, 21–40, 41–59, 60–64, 65–74, and ≥75 years) per ASSMCA standards, with primary attention to adults (≥18 years). Duplicate or incomplete records were excluded, and classification and monthly aggregate counts were cross-validated against ASSMCA metadata and annual totals. The records contain no individual or geographic identifiers, so repeat callers cannot be distinguished and localized events—such as the January 2020 earthquake sequence in southwestern Puerto Rico—could not be examined.
Analysis
All analyses were descriptive; no multivariable, adjusted, or causal models were fitted. For each month we computed ideation calls, attempt calls, and their sum (total suicide-related calls), tabulated annually (Table 1) and for the three-month post-disaster window (November 2017–January 2018). Because total hotline volume varied greatly across years, absolute counts were the primary measure. Each monthly count was treated as a realization of a Poisson process; counts are reported with exact (chi-square) Poisson 95% confidence intervals and sex proportions with Wilson score 95% confidence intervals. Temporal patterns were displayed as a time series with September 2017 (Hurricane Maria landfall; month 33) as the index event (Figure 1). We chose a descriptive approach deliberately: our aim was first to establish whether a population-level signal was present, and the aggregate monthly data—together with the telecommunications-driven undercounting in September-October 2017—preclude the assumptions required for individual-level or formal interrupted time-series modeling. Analyses used R; the expanded statistical methodology and software details are given in the Supplementary Methods, and the complete annotated workflow and all stratified displays are provided as Supplementary Material. The study was approved by the San Juan Bautista School of Medicine Institutional Review Board (EMSJBIRB-8-2021).
Annual suicide-related calls to Línea PAS, Puerto Rico, 2015–2021, by component (ideation, attempts), sex, and total hotline volume. Counts are number of calls. “% of all calls” is total suicide-related calls divided by all hotline calls that year; this proportion is sensitive to overall hotline demand (note 2020)
Table 1. Long description
Data table titled “Calls received by Línea PAS stratified by suicidal ideation, suicide attempt, sex, and age group, 2015–2021.” The columns are the seven calendar years 2015 through 2021; within each year, two sub-columns give the number of suicidal-ideation calls and the number of suicide-attempt calls, each shown as a count with, in parentheses, its percentage of that year’s total hotline calls. The rows are organized in three blocks: by sex (female, then male); by age group (0–17, 18–20, 21–40, 41–59, 60–64, 65–74, and ≥75 years); and two summary rows giving total suicide-behavior calls and total hotline calls for each year. Key patterns: total suicide-related calls ranged from 15,699 to 32,814 per year and were highest in 2017 (32,814), the year of Hurricanes Irma and Maria; women made more ideation and attempt calls than men in every year; and total hotline volume rose steeply over the period, peaking in 2020 (922,797 calls), which lowers the percentages for that year. Because total hotline volume varies greatly across years, absolute counts are the primary measure.
Monthly frequency of suicide-related calls to Línea PAS, January 2015-December 2021 (84 months): total, suicidal ideation, and suicide attempts. Shaded bands are exact Poisson 95% confidence intervals for each monthly count. The vertical dashed line marks September 2017 (Hurricane Maria landfall); total suicide-related calls peaked at 4,548 in January 2018. Horizontal dashed lines indicate the pre-hurricane baseline (January 2015-August 2017) mean + 2 SD for total calls (3,853) and for attempts (822). X-axis, calendar year; Y-axis, number of calls per month.
Figure 1 Long description
Line chart showing the monthly number of suicide-related calls to the Línea PAS crisis hotline in Puerto Rico from January 2015 to December 2021. Three labeled series are plotted—total suicide-related calls, suicidal-ideation calls, and suicide-attempt calls—each with a shaded 95% confidence-interval band. A vertical dashed line marks September 2017 (Hurricane Maria landfall). Call volume dips in September–October 2017, then rises sharply to a peak between November 2017 and January 2018, reaching a maximum of 4,548 total calls in January 2018, before returning toward baseline by February 2018. Ideation closely tracks the total, and attempts are lower but rise concurrently. Horizontal dashed lines mark the pre-hurricane baseline (January 2015–August 2017) mean plus two standard deviations for total calls (3,853) and attempt calls (822). The x-axis is calendar year; the y-axis is number of calls per month.
Results
Across the 84-month period, Línea PAS recorded between 15,699 and 32,814 suicide-related calls per year (Table 1). Absolute suicide-related calls were highest in 2017—the year of Hurricanes Irma and Maria—with 32,814 calls, followed by 2016 (28,363). Women made more ideation and attempt calls than men in every year (Supplementary Figure S3). Total hotline volume rose sharply over the period, from 83,442 calls in 2015 to 922,797 in 2020, largely reflecting COVID-19-pandemic demand; consequently, suicide-related calls as a proportion of all calls fell over time and are interpreted with caution, with absolute counts emphasized throughout.
The monthly series shows the pattern most clearly (Figure 1). Call volume fell sharply in September-October 2017 (months 33–34), then rose steeply to a pronounced peak between November 2017 and January 2018 (months 35–37), reaching the series maximum of 4,548 total calls in January 2018; ideation tracked the total closely, and attempt calls showed a smaller, concurrent rise. Volume returned toward baseline by February 2018. When stratified by age (Supplementary Figures S1–S2), the increase in ideation and attempts was most evident among working-age adults, particularly the 21–40 and 41–59 groups, whose monthly counts rose well above their pre-hurricane baseline. No comparable island-wide increase appeared around the January 2020 earthquake or the March 2020 onset of COVID-19 restrictions.
During the 3-month peak (November 2017-January 2018), Línea PAS received 13,345 suicide-related calls (10,045 ideation; 3,300 attempts). The 21–40 age group reported the most calls for both ideation (2,050) and attempts (827), followed by the 41–59 and 65–74 groups. Women accounted for 7,244 calls (54.3%) versus 6,101 (45.7%) for men (Supplementary Table S1).
Discussion
Using 7 years of crisis-hotline surveillance, we found that suicide-related calls to Línea PAS rose sharply about 2 months after Hurricanes Irma and Maria, peaked from November 2017 through January 2018, and returned toward baseline by February 2018. The increase spanned both sexes and all adult age groups and was most pronounced among working-age adults. The apparent decline in calls during September-October 2017 most likely reflects loss of telecommunications rather than a true reduction in distress: by early October 2017 roughly 79% of cell sites were out of service, and more than half remained nonfunctional at the end of October,6 so the true onset of the increase may have preceded the measured peak.
The delayed rise likely reflects several overlapping, non-exclusive processes rather than a single mechanism. First, part of the apparent delay is probably a measurement artifact, as the loss of telecommunications likely suppressed call volume during the acute phase, so that the measured increase partly tracked the gradual restoration of cellular and telephone service. Second, help-seeking may have been delayed: in the immediate aftermath, survivors may have been occupied with securing shelter, water, food, and safety, and may have contacted crisis services only once the most urgent needs were partially met. Third, the psychosocial burden may have been cumulative rather than instantaneous, accruing over weeks of prolonged outages, displacement, economic loss, and interrupted care. These explanations are complementary, and the observed 2-to-3-month interval is most plausibly the joint product of delayed restoration of communications and genuine delayed and cumulative psychosocial responses.
Independent vital-statistics data are consistent with a real increase during this window. Compared with the same months in 2016, recorded suicide deaths in Puerto Rico rose by roughly 106% in November (16 to 33) and 67% in December (15 to 25), and average monthly suicides rose from about 23 during January-August 2017 to about 27 over the subsequent three months.7 Most suicide deaths in Puerto Rico occur among men,7 whereas women more often report nonfatal suicidal behavior and seek help; the female preponderance of calls alongside rising, male-dominated suicide deaths is therefore expected and underscores the need for targeted outreach to men during recovery.
The increase coincided temporally with the island’s prolonged loss of electricity. Hurricane Maria caused catastrophic damage and an official death toll of 64 that later excess-mortality analyses placed in the hundreds to several thousand,Reference Kishore, Marqués and Mahmud8 and utility restoration was slow—about 40% of customers still lacked power four months after landfall, with near-complete restoration only by spring 2018.Reference Shermeyer9 Suicide-related calls returned toward baseline over the same period during which power was progressively restored. Prolonged outages disrupt income, daily routines, and continuity of care and have been linked to adverse mental-health outcomes,Reference Klinger, Landeg and Murray10 and emergency visits for mental illness rose after Hurricane Sandy.Reference Lin, Lu and Justino4 These observations are ecological and temporal; they are consistent with, but cannot establish, a contribution from infrastructure disruption.
This study has limitations. It captures only individuals who contacted Línea PAS, omitting cases seen at poison control, emergency departments, or psychiatric facilities, so total suicidal behavior is underestimated; the unit of analysis is the call, not the individual, so repeat callers may be counted more than once. Loss of telecommunications after Maria almost certainly produced artifactual undercounting in September–October 2017, complicating any before–after comparison, and post-disaster out-migration altered the underlying population denominator. Crisis-hotline surveillance also reflects who is aware of the service, who can reach it, and who chooses to call—factors that changed as public awareness, accessibility, and behavioral-health services expanded between 2015 and 2021, so some of the variation in call volume likely reflects changing utilization rather than the true incidence of suicidal behavior. Comparisons over time, especially across the pandemic, should be read with this in mind. Finally, as a descriptive, ecological study, it cannot establish causation or adjust for confounders.
Conclusion
In a seven-year surveillance dataset, suicide-related calls to Puerto Rico’s crisis hotline rose markedly after Hurricanes Irma and Maria, peaked while much of the island remained without electricity, and returned toward baseline as power was restored. Although these ecological data cannot prove causation, they are consistent with a substantial post-hurricane mental-health burden and a possible contribution from prolonged infrastructure disruption. Crisis hotlines generate continuous, near-real-time records of population distress and can be incorporated into disaster preparedness and recovery as a component of behavioral-health surveillance—used to set pre-event baselines, detect emerging increases, and target resources—provided their own power and telecommunications are hardened so that surveillance is not lost when distress is greatest. Rapid restoration of utilities and attention to sex differences in help-seeking should be prioritized in climate-vulnerable regions.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/dmp.2026.10475.
Author contribution
RL: Drafting; critical revision; final approval. SH: Drafting; critical revision; final approval. RV: Drafting; final approval. RT: Methodology; formal analysis; investigation; critical revision; final approval. YIN: Conception and design; data acquisition; critical revision; final approval.
Funding statement
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Competing interests
None.
