Introduction
Natural disasters, including hurricanes, cyclones, typhoons, and major floods, affect millions of people worldwide and can have significant mental health consequences.Reference Keya, Leela and Habib 1 Exposure to various disaster-related experiences has been associated with adverse mental health outcomes both in the United States (US) and internationally.Reference Keya, Leela and Habib 1 –Reference Lopez-Cepero, O’Neill and Marrero 12 For example, disaster experiences involving threat to one’s physical well-being or that of loved ones have been linked to depression, anxiety, posttraumatic stress disorder (PTSD), and psychological distress among disaster-exposed populations in Puerto Rico, Vietnam, and the Philippines.Reference Sackey, Stewart and Young 2 –Reference Pollack, Weiss and Trung 4 , Reference Lopez-Cepero, O’Neill and Marrero 12 Findings related to disaster-related property damage or loss are mixed, with some studies reporting associations with mental health conditionsReference Sackey, Stewart and Young 2 , Reference Tracy, Norris and Galea 13 and others finding none.Reference Meadows, Serino and Camacho-Monclova 7 , Reference Lopez-Cepero, O’Neill and Marrero 12 , Reference Jacobs and Harville 14 Although research on the loss of access to basic resources such as food, water, electricity, and transportation following disasters is more limited, studies have linked such disruptions to poor mental health outcomes.Reference Chan and Rhodes 8 –Reference Lopez-Cepero, O’Neill and Marrero 12
One exposure type that has received little attention in the disaster mental health literature is exposure to unsanitary/hazardous conditions. Hurricanes and other flood-related disasters often overwhelm water, sanitation, and power systems, leaving people exposed to sewage, without access to safe drinking water, and reliant on spoiled food. Beyond the immediate distress of living in such conditions, these experiences may affect mental health through several pathways. They may elicit feelings of disgust toward oneself or one’s environment, which have been linked to adverse mental health outcomes among trauma-exposed individuals.Reference Clarke, Simpson and Varese 15 –Reference Badour and Feldner 17 Moreover, studies have found increases in mold exposure, poisonings, wounds, gastrointestinal infections, and skin and soft tissue infections following disasters, which may be a result of unsanitary conditions.Reference Quist, Ebanks and Cowan 18 –Reference Benedict, Jackson and Toda 20 These physical health problems may affect psychological well-being and provide an additional pathway through which unsanitary conditions can affect mental health.Reference Felix, Rubens and Hambrick 21 , Reference Arcaya, Lowe and Asad 22 Despite this, unsanitary and hazardous conditions are rarely examined alongside other disaster exposures, and their mental health consequences remain largely unknown.
Puerto Rico offers a relevant setting for examining these exposures, having faced numerous major disasters over the past decade. In 2017, Hurricane Maria caused widespread destruction across Puerto Rico and contributed to nearly 5,000 excess fatalities.Reference Guerra Velázquez 23 –Reference Chandra, Marsh and Madrigano 25 In the months following the storm, residents faced prolonged disruptions to electricity, food, water, and medical care, and survivors reported having to engage in unsanitary or hazardous practices.Reference Lopez-Cepero, O’Neill and Marrero 12 , Reference Guerra Velázquez 23 , Reference Michaud and Kates 26 , Reference Flores Roque 27 Puerto Rico’s sociopolitical context as a United States territory has contributed to systemic vulnerabilities, including an aging electrical grid, unequal disaster relief funding, and a persistent shortage of health care providers, that deepened the impact of Hurricane Maria on the island.Reference Joseph, Voyles and Williams 28 , Reference McSorley, Rivera-González and Mercado 29 High poverty rates and an ethnically minoritized population may further heighten the island’s risk of poor post-disaster mental health outcomes.Reference Goldmann and Galea 5 , Reference Lowe, Bonumwezi and Valdespino-Hayden 30 This vulnerability underscores the need to understand how disaster exposures, including unsanitary and hazardous ones, affect mental health among this population.
This study examines how different types of hurricane-related exposures are associated with post-disaster psychiatric and substance use disorders in Puerto Rico following Hurricane Maria. Unlike prior research, it includes unsanitary and hazardous conditions as a distinct exposure type. Consistent with studies on other disaster-related exposures,Reference Keya, Leela and Habib 1 –Reference Lopez-Cepero, O’Neill and Marrero 12 unsanitary and hazardous conditions are hypothesized to be associated with a higher likelihood of post-disaster psychiatric and substance use disorders. Although set in Puerto Rico, this study may inform broader efforts to address disaster-related mental health. Understanding how specific disaster-related exposures affect mental health is critical for disaster preparedness, response, and recovery, as it can guide intervention priorities and effective resource allocation in disaster-prone regions worldwide.
Methods
Study Design and Participants
This prospective cohort study uses data from the Mental Health and Anti-Addiction Services Administration (MHAASA) survey and the follow-up Hurricane Maria Cohort Study. The MHAASA survey was a psychiatric epidemiologic survey of a representative sample of 3,062 island Puerto Rico residents ages 18-64 interviewed approximately 1 to 3 years before Hurricane Maria (2014-2016). In-depth details on the MHAASA sampling design can be found elsewhere.Reference Canino, Shrout and NeMoyer 31 The Hurricane Maria Cohort Study was fielded among the same MHAASA cohort approximately 3 to 6 years after Hurricane Maria (2020-2023), with a single follow-up observation per participant. The target population for this study was adults living in Puerto Rico during Hurricane Maria and its aftermath. Participants were excluded from the follow-up study if they were deceased, incarcerated, or had a cognitive impairment that precluded participation. Of the original 3,062 MHAASA participants, 231 (7.5%) were ineligible, resulting in 2,831 eligible participants. Study researchers completed a total of 1,733 interviews from these 2,831 eligible participants (61.2% completion rate). Supplemental Table 1 presents descriptive characteristics of participants with and without follow-up data in the Hurricane Maria Cohort Study.
To be included in the current study, participants had to have completed the follow-up interview. Participants who were missing data on outcomes, hurricane-related exposures, and covariates were excluded (n = 621). The final analytic sample consisted of 1,112 Puerto Rican adults who met these criteria. Supplemental Table 2 presents descriptive statistics of participants who were and were not excluded from the final analytic sample. A flowchart illustrating the derivation of the analytic sample is presented in Supplemental Figure 1. The University of Puerto Rico Medical Sciences Campus Institutional Review Board approved this study (protocol #2290033466R001).
Measures
The Hurricane Maria Cohort Study questionnaire contained all psychiatric and substance use disorders captured in the baseline MHAASA survey, as well as additional measures for demographics and post-disaster exposures. Post-disaster exposure measures were adapted from measures from prior disaster studies.Reference Kessler 32 –Reference Felix, You and Canino 35 The final survey instrument was pilot tested and translated to ensure accurate measurement of constructs in Spanish and English.
Psychiatric and substance use disorders
The World Mental Health Composite International Diagnostic Interview (WMH-CIDI) Version 21 was used to assess 12-month psychiatric and substance use disorders based on Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV) criteria in the baseline MHAASA study and the follow-up Hurricane Maria Cohort Study.Reference Kessler and Ustun 36 , Reference Bell 37 It is considered a widely regarded instrument for identifying psychiatric disorders in epidemiological studies and a more precise tool for assessing mental health disorders than other brief screening tools.Reference Kessler and Ustun 36 , Reference Gelaye, Tadesse and Williams 38 English and Spanish versions of the WMH-CIDI have shown diagnostic concordance with blinded clinical interviewers using the Structured Clinical Interview for DSM IV Axis 1 disorders (SCID) for mood, anxiety, and substance use disorders.Reference Alegria, Shrout and Torres 39 The WMH-CIDI can also be administered by telephone with similar validity as household interviewing.Reference Rohde, Lewinsohn and Seeley 40 , Reference Kessler, Al-Desouki and King 41 At the time of the MHAASA baseline survey, the WMH-CIDI had not been updated to reflect the changes to the DSM in the DSM-IV-TR or DSM-V.
The 12-month presence of the following disorders was examined in this study for both time periods: major depressive disorder, dysthymia, bipolar 1, bipolar II, panic disorder, agoraphobia, general anxiety disorder, alcohol abuse, alcohol dependence, drug abuse, and drug dependence. Disorders were grouped into the following categories based on the WMH-CIDI and prior research: any mood disorder (major depressive disorder, dysthymia, bipolar I and II), any anxiety disorder (generalized anxiety disorder, panic disorder, and agoraphobia), and any substance use disorder (alcohol abuse/dependence, drug abuse/dependence).Reference Canino, Shrout and NeMoyer 31 , Reference Kessler and Ustun 36
Hurricane-related exposures
There were 27 exposures examined that were categorized into 4 subcategories to capture distinct mechanisms for behavioral health outcomes: personal impact experiences, damage or loss of property, loss of utility and essential resources, and unsanitary/hazardous conditions. This categorization of exposures was guided by prior research examining hurricane-related exposure types.Reference Lopez-Cepero, O’Neill and Marrero 12 , Reference Davidson, Price and McCauley 42 Personal impact exposures involve experiences that may cause direct emotional, physical, or social harm to the individual. Property damage exposures capture housing and material loss. Loss of utility or resource exposures capture broader infrastructural failures that disrupt daily functioning. Unsanitary/hazardous conditions capture direct exposure to unsafe post-disaster environments that may compromise physical or psychological health. This categorization allows for more nuanced analysis of how different exposure types relate to psychiatric and substance use outcomes.
Personal impact experiences were measured by 7 binary items: getting injured, fear of dying or injury, concern about the safety of family, separation from loved ones due to relocation, being stranded, lost income, and having a family member, close friend, or neighbor die or be seriously injured during the hurricane. Damage or loss of property was measured by 3 binary items: damage or loss of home, damage or loss of non-sentimental possessions, and damage or loss of sentimental possessions. Loss of utilities and essential resources was measured by 9 items assessing how long participants were without power, drinking water, cellphone signal, food, a place to live, clothes, transportation, money for daily expenses, and medications following Hurricane Maria. Numeric responses, reported in either days, weeks, or months, were all converted to day-units. For participants who reported still being without a utility or resource at the time of the interview (n = 96), the number of days between Hurricane Maria and the interview date was calculated. The median number of days for each item was used to categorize them into binary responses. The median number of days for all loss of utility and essential resource variables was 0 except for power (92 days), drinking water (21 days), and cellphone signal (31 days). This approach has been used in prior research with similar post-disaster exposure measuresReference Lopez-Cepero, O’Neill and Marrero 12 and provides cutoff thresholds that reflect participants’ lived experiences.
Exposure to unsanitary/hazardous conditions was assessed through a single multi-response survey item that listed 8 predefined experiences: drinking dirty water, using the bathroom in public, being exposed to sewage water, bathing in dirty water, not bathing for several days, bleeding through an exposed wound, breaking a bone, and eating expired food. While not previously validated, these items were selected to capture common unsanitary and physically hazardous exposures following the hurricane. Participants were asked to indicate all the conditions that they had experienced. Participants who selected “not applicable” (n = 651) or skipped the item entirely (n = 134) were classified as not having any exposure to unsanitary/hazardous conditions. This decision was based on review of a subsample of interview audio files, which suggested that item nonresponse in these cases reflected the absence of exposure rather than missing data. Each affirmative response was added to calculate the number of exposures for each subcategory.
Covariates
Covariates included age (18-34, 35-49, 50-64, 65+), sex (male, female), education (high school or less, more than high school), marital status (married or living with someone, divorced/separated/widowed, never married), gross monthly household income (less than $2,000, $2,000 or more), employment (unemployed, employed), and presence of any psychiatric or substance use disorder before Hurricane Maria (yes, no).
Statistical Analyses
All analyses were conducted in Stata MP 18. 43 Descriptive statistics and prevalence of each hurricane-related exposure were calculated for the study sample. Logistic regression models were run to examine the associations between accumulated hurricane exposure (and exposure subcategories) and having any mood, any anxiety, or any substance use disorders. Additionally, a second set of logistic regression models was run to examine the associations between accumulated hurricane exposure and exposure subcategories and specific mood, anxiety, and substance use disorders. Models controlled for all covariates. Average marginal effects were calculated for all models. All analyses incorporated survey weights to account for the multistage, stratified multi-cluster sampling design. Weights were constructed using the MHAASA baseline survey weights and adjusted to account for the redistribution of weights from baseline to follow-up and differential non-response by gender, age, and region. A post-stratification adjustment was applied to align the sample with the Puerto Rico adult population distribution by region, gender, and age group based on the 2020 Census.
Results
Table 1 contains the descriptive statistics for the entire sample. The study sample was majority female (53.7%), married or living with someone (59.4%), employed (62.8%), had more than a high school education (76.0%), reported a monthly household income greater than $2,000 (56.7%), and had no pre-hurricane psychiatric or substance use disorder (80.2%). Table 2 contains the prevalence of all hurricane-related exposures. The most prevalent personal impact hurricane exposures were concern about the safety of family (69.1%) and fear of getting injured or dying (54.0%). Approximately 70.8% of participants reported damage or loss of their homes, 47.8% reported damage or loss of non-sentimental possessions, and 17.9% reported damage or loss of sentimental possessions. Among utility and resource losses, the most prevalent were being without drinking water for more than 21 days (46.9%), being without cellphone signal for more than 31 days (39.0%), and being without power for more than 92 days (38.5%). The most common unsanitary/hazardous experiences were not bathing for several days (16.5%), bathing in dirty water (11.1%), and drinking dirty water (10.0%).
Weighted descriptive characteristics of adults living in Puerto Rico during Hurricane Maria in 2017 (n = 1,112)
Table 1. Long description
The table consists of two columns: Characteristic and n (weighted column percent).
* Age: 18 to 34 years at 321 (19.0 percent); 35 to 49 years at 274 (34.4 percent); 50 to 64 years at 365 (32.9 percent); 65 plus years at 152 (13.8 percent).
* Sex: Male at 436 (46.3 percent); Female at 676 (53.7 percent).
* Education: High school or less at 262 (24.0 percent); More than high school at 850 (76.0 percent).
* Marital status: Married or living with someone at 616 (59.4 percent); Divorced/separated/widowed at 207 (18.7 percent); Never married at 289 (21.9 percent).
* Monthly income: Less than $2,000 at 493 (43.3 percent); $2,000 or more at 619 (56.7 percent).
* Employment: Unemployed at 458 (37.2 percent); Employed at 654 (62.8 percent).
* Any pre-hurricane psychiatric or substance use disorder: No at 879 (80.2 percent); Yes at 233 (19.8 percent).
Weighted prevalence of hurricane-related exposures among adults in Puerto Rico (n = 1,112)
Table 2. Long description
The table consists of two columns: Exposure and percent reporting.
1. Personal impact:
- Experienced physical injuries: 1.8 percent.
- Fear of getting injured or dying: 54.0 percent.
- Concerned about the safety of family: 69.1 percent.
- Separated from close family because they or you moved away after Hurricane Maria: 21.6 percent.
- Became trapped: 10.3 percent.
- Lost income: 31.2 percent.
- Family, close friend, or neighbor was injured or died: 9.7 percent.
2. Damage or loss of property:
- Damage or loss of home: 70.8 percent.
- Damage or loss of non-sentimental possessions: 47.8 percent.
- Damage or loss of sentimental possessions: 17.9 percent.
3. Loss of utilities and essential resources:
- Without power for more than 92 days: 38.5 percent.
- Without drinking water for more than 21 days: 46.9 percent.
- Without cellphone signal for more than 31 days: 39.0 percent.
- Without food: 4.7 percent.
- Without a place to live: 3.6 percent.
- Without clothes: 1.2 percent.
- Without transportation: 11.1 percent.
- Without money for daily expenses: 31.6 percent.
- Without medications: 10.5 percent.
4. Unsanitary or hazardous conditions:
- Drank dirty water: 10.0 percent.
- Performed biological necessities in public: 4.2 percent.
- Exposed sewage water: 2.4 percent.
- Bathed with dirty water: 11.1 percent.
- Did not bathe for several days: 16.5 percent.
- Bled through an exposed wound: 1.8 percent.
- Broke a bone: 0.9 percent.
- Ate expired food: 7.9 percent.
Note. Values represent weighted column percents.
Table 3 presents the average marginal effects of hurricane-related exposures on having any mood, anxiety, or substance use disorder. Each additional accumulated hurricane exposure experienced was associated with a 1.1 (P < 0.001, 95% CI: 0.05, 1.6), 1.3 (P < 0.001, 95% CI: 0.8, 1.9), and 2.0 (P < 0.001, 95% CI: 1.2, 2.7) percentage-point increase in the probability of having any mood disorder, any anxiety disorder, or any disorder overall, respectively. Analyses by exposure subcategory revealed that each additional personal impact exposure experienced was associated with a 2.7 percentage-point increase in the probability of having any disorder overall (P = 0.028, 95% CI: 0.3, 5.1). Additionally, each additional loss of utility and essential resources exposure experienced was associated with a 2.3 percentage-point increase in the probability of having any anxiety disorder (P = 0.001, 95% CI: 1.0, 3.7) and a 2.3 percentage-point increase in the probability of having any disorder overall (P = 0.010, 95% CI: 0.5, 4.0). Lastly, each additional unsanitary/hazardous condition experienced was associated with a 1.9 (P = 0.017, 95% CI: 0.3, 3.5) percentage-point increase in the probability of having any mood disorder and a 2.1 percentage-point increase in the probability of having any disorder overall (P = 0.043, 95% CI: 0.1, 4.2). Damage or loss of property was not associated with any of the outcomes, and no associations were found between any hurricane-related exposures and any substance use disorder.
Marginal effects of Hurricane Maria exposures on psychiatric and substance use disorder categories among adults in Puerto Rico (n = 1,112)
Table 3. Long description
The table presents M E (95 percent C I) for four disorder categories across five exposure types.
Column headers: Any mood disorder, Any anxiety disorder, Any substance use disorder, and Any disorder.
Rows by exposure type:
* Accumulated hurricane exposure: Mood 1.1 (0.5, 1.6) triple asterisk; Anxiety 1.3 (0.8, 1.9) triple asterisk; Substance 0.2 (-0.2, 0.5); Any 2.0 (1.2, 2.7) triple asterisk.
* Personal impact exposure: Mood 1.4 (-0.7, 3.4); Anxiety 0.4 (-1.4, 2.1); Substance 0.9 (-0.3, 2.2); Any 2.7 (0.3, 5.1) asterisk.
* Damage or loss of property exposure: Mood -1.0 (-3.8, 1.7); Anxiety 1.2 (-1.4, 3.8); Substance -0.1 (-1.4, 1.2); Any 0.0 (-3.1, 3.1).
* Loss of utility and resources exposure: Mood 1.3 (-0.1, 2.7); Anxiety 2.3 (1.0, 3.7) double asterisk; Substance -0.5 (-1.3, 0.3); Any 2.3 (0.5, 4.0) asterisk.
* Unsanitary or hazardous conditions exposure: Mood 1.9 (0.3, 3.5) asterisk; Anxiety 1.0 (-0.7, 2.7); Substance 0.6 (-0.3, 1.5); Any 2.1 (0.1, 4.2) asterisk.
Notes: Asterisks indicate P-values: asterisk P less than 0.05, double asterisk P less than 0.01, triple asterisk P less than 0.001. M E stands for marginal effect and C I stands for confidence interval.
Notes. Adjusted models adjusted for age (18-34, 35-49, 50-64, 65+), sex (male/female), education (high school or less, more than high school), marital status (married or living with someone, divorced/separated/widowed, never married), monthly income (less than $2,000, $2,000 or more), employment (unemployed/employed), and history of any pre-hurricane disorder (yes/no). All disorders were measured with the WMH CIDI.
*P < 0.05; **P < 0.01; ***P < 0.001.
Abbreviations: CI, Confidence interval; ME, marginal effect.
Table 4 presents the average marginal effects of hurricane-related exposures on specific psychiatric and substance use disorders. Among mood disorders, each additional hurricane exposure experienced was associated with a 0.7 (P = 0.002, 95% CI: 0.3, 1.2), 0.9 (P = 0.001, 95% CI: 0.3, 1.4), and 0.4 (P = 0.009, 95% CI: 0.1, 0.6) percentage-point increase in the probability of having post-hurricane major depressive disorder, dysthymia, and any bipolar disorder, respectively. Among anxiety disorders, each additional hurricane exposure experienced was associated with a 0.7 (P < 0.001, 95% CI: 0.3, 1.0), 0.8 (P < 0.001, 95% CI: 0.4, 1.1), and 0.2 (P = 0.046, 95% CI: 0.0, 0.5) percentage-point increase in the probability of panic disorder, agoraphobia, and generalized anxiety disorder, respectively.
Marginal effects of Hurricane Maria exposures on specific psychiatric and substance disorders among adults in Puerto Rico (n = 1,112)
Table 4. Long description
The table consists of 8 columns representing disorders: Major depressive disorder, Dysthymia, Bipolar I or II, Panic disorder, Agoraphobia, Generalized anxiety disorder, Alcohol abuse or dependence, and Drug abuse or dependence. Each cell contains the M E and 95 percent C I in parentheses.
* Accumulated hurricane exposure: Shows significant positive effects for Major depressive disorder 0.7 (0.3, 1.2), Dysthymia 0.9 (0.3, 1.4), Bipolar I or II 0.4 (0.1, 0.6), Panic disorder 0.7 (0.3, 1.0), Agoraphobia 0.8 (0.4, 1.1), and Generalized anxiety disorder 0.2 (0.0, 0.5).
* Personal impact exposure: No statistically significant effects, with values ranging from minus 0.2 to 1.0.
* Damage or loss of property exposure: No statistically significant effects, with values ranging from minus 1.6 to 1.6.
* Loss of utility and resources exposure: Significant positive effects for Agoraphobia 1.7 (0.8, 2.5) and Generalized anxiety disorder 1.2 (0.2, 2.2).
* Unsanitary or hazardous condition exposure: Significant positive effects for Major depressive disorder 1.9 (0.6, 3.3), Dysthymia 1.7 (0.2, 3.3), and Panic disorder 1.6 (0.6, 2.6).
Notes indicate models were adjusted for age, sex, education, marital status, income, employment, and pre-hurricane disorder history. Significance levels are marked by asterisks for P less than 0.05, 0.01, and 0.001.
Notes. Models adjusted for age (18-34, 35-49, 50-64, 65+), sex (male/female), education (high school or less, more than high school), marital status (married or living with someone, divorced/separated/widowed, never married), monthly income (less than $2,000, $2,000 or more), employment (unemployed/employed), history of any pre-hurricane disorder (yes/no). All disorders were measured with the WMH CIDI.
* P < 0.05; **P < 0.01; ***P < 0.001.
Abbreviations: CI, Confidence interval; ME, marginal effect.
Each additional loss of utility and essential resources exposure experienced was associated with a 1.7 percentage-point increase in the probability of agoraphobia (P < 0.001, 95% CI: 0.8, 2.5) and a 1.2 percentage-point increase in the probability of generalized anxiety disorder (P = 0.017, 95% CI: 0.2, 2.2). Finally, each additional unsanitary/hazardous condition experienced was associated with a 1.9 (P = 0.006, 95% CI: 0.6, 3.3), 1.7 (P = 0.031, 95% CI: 0.2, 3.3), and 1.6 (P = 0.003, 95% CI: 0.6, 2.6) percentage-point increase in the probability of having major depressive disorder, dysthymia, and panic disorder, respectively. Damage or loss of property and personal impact exposures were not associated with any specific psychiatric or substance use disorder. No associations were found between any hurricane-related exposures and alcohol abuse/dependence or drug abuse/dependence.
Limitations
This study is limited by the use of self-report data for all hurricane-related exposures. Data collection occurred between 2020 and 2023 to assess events that took place in 2017, raising the possibility of recall bias. However, prior research suggests that individuals can generally recall personally significant and disaster-related experiences with reasonable accuracy, even years after the event.Reference Wu 44 Additionally, other events, such as subsequent hurricanes, earthquakes, and the COVID-19 pandemic, may have influenced psychiatric and substance use outcomes due to delayed data collection.
Participants who did not respond to the survey item that measured unsanitary/hazardous exposures were classified as unexposed, which may have introduced misclassification bias. This decision was based on a review of a subsample of interview audio files, which indicated that item nonresponse in these cases was consistent with a lack of exposure rather than with missing data. Therefore, the potential for misclassification bias is likely minimal. Moreover, additional analyses revealed no significant differences in outcomes between participants who reported no unsanitary/hazardous exposures and nonrespondents (Supplemental Table 3). Therefore, findings may represent a conservative estimate of the association between these exposures and post-hurricane behavioral health outcomes. Additionally, although the exposure subcategories were theoretically guided by prior research, their reliability and validity have not been analytically tested, and heterogeneity within subcategories may have attenuated the observed associations.
Supplemental analyses showed that participants included in the final analytic sample were younger, more educated, had higher monthly household incomes, and were more likely to be employed compared to those excluded due to missing data, which may limit the generalizability of the study findings. Although survey weights were applied to improve the representativeness of the analytical sample relative to the island Puerto Rican adult population, these sociodemographic differences may represent a source of selection bias. Additionally, given the large number of statistical tests conducted, the possibility of spurious findings cannot be ruled out.
Lastly, although all analyses controlled for history of pre-hurricane disorders, the current analytic approach reflects associations between hurricane exposures and post-disaster disorder status among all participants regardless of baseline disorder status, rather than examining associations with new-onset disorders. A fully longitudinal modeling strategy could provide additional insight into how hurricane exposure may contribute to the development of new-onset psychiatric and substance use disorders and represents an important direction for future research. Despite these limitations, this study provides important insight into how disasters impact psychiatric and substance use disorders in Puerto Rico.
Discussion
This study found that exposure to unsanitary and hazardous conditions, which has received little attention in the disaster mental health literature, was associated with an increased likelihood of post-disaster mood disorders. More broadly, associations between hurricane-related exposures and post-disaster psychiatric disorders differed by exposure type and specific psychiatric disorder, consistent with prior work among the Puerto Rican population.Reference Sackey, Stewart and Young 2 , Reference Meadows, Serino and Camacho-Monclova 7 , Reference Lopez-Cepero, O’Neill and Marrero 12 , Reference Musicaro, Valdespino-Hayden and Lozano 45 This study extends prior research by incorporating unsanitary and hazardous conditions and examining a broader range of diagnostic outcomes beyond depression, anxiety, and PTSD.
Although this study focused on Puerto Rico following Hurricane Maria, the findings may have broader relevance to populations affected by typhoons, cyclones, and flooding events globally. Studies conducted after Typhoon Haiyan in the Philippines, the Great East Japan Earthquake and Tsunami, major flooding events in Bangladesh, and flooding in the United Kingdom have similarly documented elevated rates of depression, anxiety, psychological distress, and posttraumatic stress among disaster survivors.Reference Alibudbud 46 –Reference Mulchandani, Armstrong and Beck 49 Across these settings, prolonged disruption of essential services, displacement, and exposure to unsafe environmental conditions have also been documented following disasters.Reference Rahman, Shobuj and Alam 48 –Reference Rahman, Alam and Shobuj 52 The prevalence of these experiences across different sociopolitical and geographic contexts suggests that the risk factors identified in the present study may extend beyond Puerto Rico and may be relevant to disaster-prone regions worldwide.
This study found that greater exposure to loss of utilities and essential resources was associated with a higher likelihood of post-disaster anxiety disorders. Although research on the mental health consequences of utility and resource loss following disasters is limited, the present findings are consistent with evidence from other international contexts and contribute to a growing understanding of how such disruptions may impact post-disaster psychiatric morbidity. For instance, disruptions to essential services were associated with higher odds of probable PTSD, anxiety, psychological distress, and depression following major flooding in England in 2007 and in 2013/2014.Reference Waite, Chaintarli and Beck 10 , Reference Paranjothy, Gallacher and Amlôt 11 Similarly, interruptions to household utilities were associated with increased odds of having probably PTSD following Cyclone Debbie in Australia in 2017.Reference King, Longman and Matthews 53 Notably, these studies relied on symptom-based screening measures rather than clinical diagnostic assessments for mental health outcomes.
The present study extends this literature by demonstrating similar associations using clinically assessed psychiatric disorders. Differences in outcome measurement may also explain why associations with mood disorders were identified in prior studies but not in the present study. As utility and resource loss is a common consequence of major disasters worldwide,Reference Hertelendy, Dresser and Gorgens 54 these findings suggest that efforts to quickly restore access to essential services in the aftermath of typhoons, cyclones, and floods may yield both practical and mental health benefits for affected populations. Additionally, these findings underscore the need for sustained investment in disaster-resilient infrastructure, particularly in disaster-prone regions like Puerto Rico where pre-existing infrastructure is already fragile and successive disasters compound prior damage.
Greater exposure to unsanitary and hazardous conditions was associated with an increased likelihood of post-disaster mood disorders. Unsanitary and hazardous exposures are often overlooked in the disaster mental health literature, despite being common in the aftermath of hurricanes, typhoons, and flood-related disasters worldwide. Floodwaters, sewage contamination, mold exposure, food spoilage, and limited access to sanitation facilities have been documented following disasters across Asia, Oceania, and Europe.Reference Rocha, Ramos and Crispino 55 –Reference Vardoulakis, Matthews and Bailie 57 The present findings suggest that these environmental exposures may represent an underrecognized pathway linking disaster experiences to subsequent psychiatric morbidity and warrant the inclusion of such exposures in future post-disaster mental health research and needs assessments.
The findings that loss of utilities and essential resources was associated with anxiety disorders while unsanitary and hazardous conditions were associated with mood disorders suggest that different exposures may carry distinct psychiatric consequences. Although replication is warranted, these differential associations provide preliminary evidence that tailored post-disaster mental health interventions may benefit from considering specific exposure types and their mental health consequences.
Property damage or loss was the only exposure type not associated with any post-hurricane psychiatric or substance use disorders. One possible explanation for the null findings is that property damage was so widespread following Hurricane Maria that it may have been seen as less psychologically distressing in a context where such loss was normal.Reference Galea, Brewin and Gruber 58 Indeed, more than 70% of the study sample reported experiencing damage or loss of their home. Additionally, it is possible that experiences involving direct threat to one’s well-being or lack of basic resources may have a more profound psychological impact than material damage or loss. Some studies have found stronger associations between physical adversity exposures and mental health outcomes compared to property damage.Reference Lopez-Cepero, O’Neill and Marrero 12 , Reference Galea, Brewin and Gruber 58
Associations varied further when examining specific psychiatric disorders rather than broader disorder categories. For example, although loss of utilities and essential resources was associated with post-disaster anxiety disorders overall, this relationship was driven by agoraphobia and generalized anxiety disorder. Similarly, unsanitary/hazardous exposures were associated with any mood disorders collectively, but disaggregated analyses revealed that this association was limited to major depressive disorder and dysthymia. This emphasizes the importance of examining how hurricane-related exposures impact various mental health outcomes and not just depression and anxiety. These within-category variations highlight a potential need for post-disaster mental health interventions that are tailored to the full range of psychiatric disorders that may emerge. Accessible and culturally relevant mental health services are needed to adequately address the diverse psychiatric needs of disaster-affected communities.
This study did not find any significant associations between any hurricane-related exposure types and post-disaster substance use disorders. This was unexpected given prior research that indicates individuals may engage in substance use as a coping mechanism following exposure to traumatic events.Reference Alexander and Ward 59 , Reference Khantzian 60 Studies conducted after Hurricane Katrina and the 9/11 World Trade Center attacks documented increases in alcohol consumption, cigarette smoking, and cannabis use among survivors.Reference Flory, Hankin and Kloos 61 , Reference North and Pfefferbaum 62 However, the direction of this relationship is not consistent across studies. For example, among Hurricane Katrina evacuees, higher disaster exposure was associated with decreased illicit drug use.Reference Cepeda, Valdez and Kaplan 63 A study among the Puerto Rican population found declines in injection frequency among people who injected drugs after Hurricane Maria but an increased likelihood of reporting an overdose.Reference Abadie, Cano and Habecker 64 The null findings in the present study may reflect limited access to substances in the severely disrupted post-disaster environment, stigma associated with reporting substance use, or changes in use that did not reach the threshold for diagnosable disorder. Taken together, these results suggest that the relationships between hurricane exposure and substance use outcomes may be complex, and further research is needed to understand post-disaster substance use patterns in Puerto Rico.
Conclusion
To our knowledge, this study is among the first to document associations between unsanitary and hazardous conditions and post-disaster psychiatric morbidity. Utility and resource disruptions and exposure to unsanitary or hazardous conditions emerged as important correlates of post-disaster psychiatric disorders, highlighting the importance of assessing a broad range of disaster-related experiences beyond measures of property damage and personal loss. These findings underscore the value of including diverse exposure domains in post-disaster needs assessments to identify individuals who may be at elevated risk for adverse mental health outcomes and to guide targeted intervention efforts. Although conducted in Puerto Rico following Hurricane Maria, these findings may have broader relevance for hurricane-, typhoon-, cyclone-, and flood-prone regions worldwide, where prolonged disruptions in essential services and exposure to unsafe environmental conditions are common consequences of disasters.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/dmp.2026.10438.
Acknowledgment
This research article was completed by Damaris Lopez Mercado as part of a dissertation requirement at Drexel University. The content of this study in all its parts is the sole responsibility of the authors and does not necessarily represent the official opinion or viewpoint of the Puerto Rico Mental Health and Anti-Addiction Services Administration.
Author contribution
Damaris Lopez Mercado: Conceptualization, formal analysis, writing—original draft, writing—review and editing.
Alexandra C. Rivera-González: Writing—review and editing.
Katyana M. Santiago: Writing—review and editing, project administration.
Jim P. Stimpson: Writing—review and editing, methodology.
Ligia M. Chavez: Writing—review and editing, project administration.
Brent Langellier: Writing—review and editing.
Jonathan Purtle: Writing—review and editing.
Mark Stehr: Writing—review and editing.
Jan M. Eberth: Writing—review and editing.
Glorisa Canino: Writing—review and editing, funding acquisition.
Alexander N. Ortega: Writing—review and editing, funding acquisition, supervision.
Funding statement
Research reported in this publication was supported by the National Institute on Minority Health and Health Disparities of the National Institutes of Health (NIH) under Award Numbers R01MD013866 and R01MD016426. The content is solely the responsibility of the authors and does not necessarily represent the official views of NIH.
Competing interests
None.
Financial disclosure
No financial disclosures were reported by the authors of this paper.
