RQ 1: Is there an association between ACEs and level of adulthood household emergency preparedness?

RQ 2: Is there an association between healthcare provider discussions of household emergency preparedness and levels of preparedness?

The ACEs Questionnaire was used to assess the incidence of early life trauma in our participants. The tool was developed by Kaiser Permanente researchers, and research conducted with this tool has found strong, cumulative associations between ACEs and an increased risk of diseases, adult health risk behaviors, and poor health outcomes (27, 28). The current scale contains 17 questions in 10 categories where respondents are scored 1 for ‘yes’ and 0 for ‘no’. Total score can range from 0 to 10.

The Societal and Social Influence domain was constructed using the Connection and Caring domain of the Communities Advancing Resilience Toolkit (CART) (29), home ownership status, years in home, and years in community.

The Big Five Inventory (BFI), a 44-item scale developed by John et al. (30), assesses five personality dimensions of openness, conscientiousness, extroversion, agreeableness, and neuroticism through Likert-scaled responses. The scale has demonstrated acceptable reliability and internal consistencies, ranging between 0.79 and 0.87 (31, 32). In this study, we utilized the questions from the Conscientiousness and Neuroticism dimensions.

The Emotional Reactivity domain was measured using the Neuroticism dimension of the BFI and the Difficulties in Emotional Regulation Scale (DERS-16). The DERS-16 is a 16-item scale adapted by Bjureberg et al. (33) used to measure five aspects associated with difficulties in emotional regulation including nonacceptance of emotional responses, difficulty engaging in goal directed behavior, impulse control difficulties, limited access to emotional regulation strategies, and lack of emotional clarity. Respondents self-rate using five-point Likert-type responses where higher scores indicate higher levels of emotional dysregulation. The DERS-16 demonstrated a high internal consistency in this study (α = 0.92).

Although researchers have found that resilience can predict successful reactions to a range of adversities, resilience itself has been differentially defined (34). Given this, our model investigates related concepts that have been found to affect concern and preparation for future challenges, including grit and emotional regulation. Grit is defined as “perseverance and passion for long-term goals” (35), p. 1087. Kannangara et al. (36) found that university students with higher grit scores demonstrated better mental health and self-control. Matthews et al. (37) and O’Neal et al. (38) both report that grit is among the best predictors of success and adaptive responses to acute and chronic stress. We constructed the Resilience domain using the Grit-S and the Brief Resilient Coping Scale (BRCS).

The Grit-S is an 8-item scale adapted from the original 12-item Grit-O developed to measure the concept of grit. The respondent rates the eight statements using a five-point Likert-type scale. Half of the questions address consistency of interest and the other half address perseverance of effort. Testing of the scale has demonstrated acceptable internal consistency. The four samples used to test the adaptation of the Grit-S demonstrated acceptable reliability with internal consistencies ranging between 0.73 and 0.83 (39).

The BRCS is a 4-item scale developed to measure how well respondents adaptively deal with stress. These traits include coping resources, coping with pain, and psychological wellbeing (40). BRCS scores range between 4 and 20 with lower scores denoting less resilience. Kocalevent et al. (41) found that the BRCS has an adequate internal consistency (α = 0.76).

The construct of Motivation can be viewed as inspiring one to desire to prevent future recurrences of a past traumatic event. If people with ACEs experience this motivation, it is plausible that carefully prepared and packaged interventions can leverage this construct to inspire these people to become prepared for disasters. Motivation to prepare for disasters was constructed using a subscale of questions from the Chapman University Survey on American Fears (42) and the conscientiousness scale from the BFI. The Chapman subscale was used to measure five aspects associated with preparing for disasters including keeping an emergency supply kit, belief that they will personally experience a disaster in the near future, guilt and confidence relating to disaster preparedness, and belief that a disaster triggered by a natural hazard can cause them serious harm. Respondents rated their agreement with the five statements using a four-point Likert scale ranging from strongly agree to strongly disagree.

The construct of Coping determines how coping strategies one learns from early-life trauma affects how one prepares for disasters. There was no valid and reliable instrument to measure this construct in a disaster context. To measure this mediator, we asked participants one item: I feel that I can do something to keep me safe during disasters (yes/no/unsure).

The Household Emergency Preparedness Instrument (HEPI) General Preparedness scale is an international, all-hazards, comprehensive, 30-item instrument created to measure disaster preparedness of households with support for face, content, and criterion validity of the instrument (43–45). The HEPI questions are objective and ask about what the respondent presently owns or does in a dichotomous format. Higher scores on the HEPI indicate higher levels of household emergency preparedness. The minimum score a respondent could receive on the General Preparedness scale is 0 and the maximum score is 40 (zero points for each no response, one point for each yes response, and two points for each supply item stored in an actual disaster kit). Two subscales comprise the HEPI General Preparedness scale: Preparedness Actions and Planning (PAP) and Disaster Supplies and Resources (DSR). These subscales represent basic emergency preparedness knowledge, behaviors, and actions applicable to all households.

Some of these instruments sample emotionally salient domains that could affect responses to subsequent items. We therefore randomized the order of the instruments for each participant to control for any anchoring or response biases of completing one instrument before any others.

The study was approved with exempt status by the Institutional Review Board (IRB) of Hunter College of the City University of New York (protocol #2023-0753-Hunter) on November 29, 2023. Participants provided consent when they enrolled in the study. Participants were not compensated. The data for this study have not been approved to be shared beyond the study team.

This was an internet-based, social behavioral, quantitative, cross-sectional study of adults residing in the United States including Puerto Rico. Our independent variable was the ACEs Questionnaire score. We assumed that the ACEs Questionnaire measured our model’s Debilitation and Early Life Trauma domain. Covariates, all chosen based on previous household emergency preparedness literature, included military status, functionally diverse (i.e., disability) status, predominant language, age, household composition or familial structure (partner/children), ethnic/national origin identity, place (ZIP code, years in current home, years in community), race, gender identity, education, employment, income, risk perception, prior disaster experience (property damage, injury/illness due to a disaster), dependence on medications, dependence on medical equipment, home ownership, and healthcare provider discussed disaster preparedness. Our hypothesized moderators included Societal and Social Influences, Emotional Reactivity, Resilience, and Motivation to Prepare for Disasters. Our hypothesized mediator was Healthy Coping strategies, as measured by the participants’ household emergency preparedness self-efficacy. Finally, our dependent variable was the HEPI General Preparedness score.

Previous research using the HEPI with a similar population (43) suggested that disaster education interventions can generate large (Cohen’s d > 0.8) effects. Assuming α = 0.05 and 1–β = 0.8, for a linear regression with 15 continuous predictors and one interaction term, we estimated needing a sample size of at least 135 participants.

Inclusion criteria were being aged 18 years or older, understanding and reading English, and living in the US. Participants less than 18 years of age were excluded because children and adolescents are generally not responsible for their household emergency preparedness activities. We only included residents of the US or Puerto Rico because disaster-related policies, resources, and needs vary by country and we shared US-based disaster preparedness educational resources with the participants after they completed the study survey.

Potential participants were recruited via ResearchMatch.org, a National Institutes of Health sponsored “free and secure tool that helps match willing volunteers with eligible researchers and their studies at institutions across the country” (46), para. 1. ResearchMatch emailed their registered volunteers our IRB-approved recruitment script on our behalf. The volunteers had the option either to consent for their contact information to be released to the study team or decline participation. ResearchMatch then released the email addresses of the interested volunteers to the study team. We then emailed the potential participants a link to the survey, starting with an internet-based informed consent form. On December 20, 2023, ResearchMatch emailed our IRB-approved recruitment script to 1,498 of their registered volunteers via random selection. ResearchMatch prevents duplicate potential participants within a four-month period. We randomly selected a new batch of potential participants once a week for 16 weeks, concluding sampling on April 1, 2024, and data collection on April 8, 2024.

This study investigated relationships between ACEs, their effects on current dispositions, and the effects of both on actual disaster preparedness. Although an eventual goal in this line of inquiry is to test a structural equation model explicitly examining all of the relationships presented in Figure 1, we did not want to recruit sufficient numbers of participants for that until components of the new model were tested separately.

We conducted such tests through a family of linear regression models predicting total HEPI scores that each added sets of variables related to a respective theoretical domain; we then not only tested whether individual variables significantly predicted disaster preparedness, but also whether that domain significantly contributed to our understanding of preparedness. In addition, there were many moderate correlations between most of the variables; adding related sets of variables provided a theory-based framework for parsing out these relationships to make better sense of them and their impacts.

It is worth noting that the order in which domains are added may affect the extent to which they improve model fits. We chose to add the domains in the order we did for theoretical reasons.

Participants were adults aged between 23 and 88 years (mean = 49.66, SD = 17.07) residing in the US. The number of participants residing in each state is depicted in Appendix A, Figure 2. The majority of participants self-identified as female (n = 224, 72.03%), white (n = 263, 84.57%), non-Hispanic (n = 266, 86.08%), and with education at the baccalaureate (n = 114, 36.66%) or graduate (n = 119, 38.26%) levels. Most participants reported renting their home (n = 201, 64.84%). The average length of time in residence was 10.29 years (SD = 10.43, range 23–88) and time in community was 16.16 years (SD = 10.16, range 0–68). Approximately 65% (n = 203) had a partner and 22% (n = 69) had children living in the home. Most (n = 265, 85.21%) participants reported that either they or someone in their home took medications daily; 22.51% (n = 70) required special equipment; and 13.23% (n = 41) required assistance from others. Only 5.47% (n = 17) reported that they had discussed disaster preparedness with their healthcare provider. Experiencing illness or injury from a disaster, or knowing someone who did, was positively associated with healthcare provider discussion of disaster preparedness (r = 0.21). Complete demographic data of our sample is displayed in Table 1.

Although most (n = 231, 74.28%) participants reported having experienced at least one ACE, the number of participants experiencing more than one ACE quickly diminished; the mean of the 301 ACEs responses was 2.67 and the median was 2, indicating a positive skew of the distribution, and the relative lack of ACEs among this sample.

HEPI General Preparedness scores (the composite of the PAP and DSR subscales) ranged from 3 to 40 (n = 300) with an average of 17.75 (SD = 6.59). There was a lower HEPI PAP mean score (M = 7.73, SD = 4.33) as compared to the HEPI DSR mean score (M = 10.02, SD = 2.99). Although there were many small to medium correlations between most of the variables, correlations between ACEs and the HEPI score were low to very low and not significant (r_{HEPI Composite} = 0.001, p ≈ 1; r_PAP = 0.003, p ≈ 1; r_DSR = −0.014, p = 0.814).

Responses from the Chapman University Survey on American Fears scale were used in the constructs of Motivation and Healthy Coping strategies. Full details of the responses are found in Table 2. The response to the item “I feel confident that I know how to prepare for disasters” was positively associated with composite HEPI scores (r = 0.49). The question also captured the respondent’s perceptions with ‘I feel that I can do something to keep me safe during disasters’ positively associated with composite HEPI scores (r = 0.28).

Inter-item reliabilities (coefficient αs) are presented in Table 3. Except for the Chapman University Survey on American Fears Motivation to Prepare for Disaster subscale (α = 0.56), the instruments demonstrated good internal consistency. Note that the Motivation to Prepare for Disaster subscale was not initially created to measure one, unitary dimension, and that we included each item separately here for all theory-based analyses.

Correlations between the major variables are presented in Table 4 (a fuller matrix of correlations is given in Appendix B, Table 7). As those tables show, there were many moderately-sized correlations between most of the major variables. Discounting the correlations between HEPI total and subscores, Neuroticism had the highest average correlations (mean r = −0.25) with the other variables followed by DERS (mean r = −0.21) and Conscientiousness (mean r = −0.11). Correlations with HEPI composite scores ranged from 0.25 (with Conscientiousness) to −0.31 (with Neuroticism), except for CART Connection and Caring scores, with which it did not correlate well (r = 0.08). ACEs Questionnaire scores had the lowest average correlation (x_Correlations = 0.01), correlating best with DERS (r = 0.33), Neuroticism (r = 0.27), and Grit-S (r = −0.25) scores. Correlations with other variables ranged from <0.01 (with HEPI total and PAP) to −0.18 (with CART).

Table 5 summarizes the changes in how well a model predicted HEPI General Preparedness after the variables of a given domain were added to it. Compared to the base model that only included demographic variables, the model that next included ACEs scores produced a significantly better fit to the data (change in BIC = 8.71, df = 1, p - 0.003). However, the term for ACEs in that second model was itself not significant (β = 0.02, p = 0.734). Insight into this seeming contradiction can be gained from the correlations between ACEs scores and those demographic variables, which ranged from very small (has a partner, gender-fluid, & non-binary gender rs ≈ |0.1|) to medium (income r = −0.32); the average correlation between ACEs and demographic variables was 0.11. ACEs thus shares variance with several demographic variables, and with them creates a significantly better prediction of HEPI General Preparedness than demographics alone.

Table 6 presents the parameters for the individual variables for the final model that contained all of the domains; Appendices B, C, Tables 7, 8 present the results for each of the intermediate models, including which variables were significant in each successive model.

Many of the relationships that were significant alone became attenuated to non-significance when other variables were added to the models. Among the variables that remained the most reliably predictive of the HEPI General Preparedness score were income, feeling that keeping a kit helps, feeling confident in one’s current preparedness, DERS scores, and feeling at risk for experiencing a disaster. Requiring special medical assistance, military service, age, and education were also often important until many other variables were added. These results may suggest that those who have the means to prepare have indeed done so, or are actively concerned about preparations.

Our participants were often older adults, so it is perhaps not surprising that we did not find a direct relationship among the rather few ACEs they reported experiencing and disaster preparedness; however, ACEs did affect the relationships of demographics on HEPI scores, suggesting that future structural equation models (like those proposed in Figure 1) may prove insightful.

Table 5 also shows the results of adding additional domains of variables beyond ACEs. Neither adding Provider Discussion nor Motivation (e.g., experiences with previous disasters) significantly improved the fit of the overall model. In general, the Motivation domain and provider discussions were found to be distractions, drawing away from our ability to understand disaster preparedness. This may prove to be unique to people from demographics similar to our participants or even just these data, but these results suggest the counter-intuitive conclusion that provider discussions need not be immediately prioritized.

All other domains (Social and Societal, Healthy Coping and Health Issues, Emotional Reactivity, and Resilience) did improve our understanding of General Preparedness. Like the ACEs Questionnaire scores, the resilience domain improved the model fit (χ² = 6.66, df = 2, p = −0.36) although neither term was significant (β_Grit-S = 0.08, p = 0.266; β_BRCS = 0.06, p = 0.387). Both Grit-S and BRCS scores correlated with most of the other major variables (rs_Grit-S = 0.19–−0.58, rs_BRCS = 0.22–−0.50), suggesting again that their addition to the model helped primarily through clarifying other relationships (i.e., through partialing out the effects of resilience from other relationships with HEPI scores).

Healthy Coping (as measured by the Chapman University Survey on American Fears items “my kit helps” and “confident I’m prepared”), Health Issues (χ² = 50.26, df = 2, p < 0.001) and the Societal and Social Influences domains (χ² = 26.27, df = 3, p < 0.001) were quite important for understanding inclinations toward disaster preparedness. Their contributions to our understanding were significant and large.

Finally, the Emotional Reactivity domain also played an important role (χ² = 19.34, df = 2, p < 0.001), especially scores on the DERS (β_{Final Model} = −0.17, p = 0.035). DERS and Neuroticism scores both significantly correlated with ACEs Questionnaire scores (rs = 0.33 & 0.27, respectively), with their combined emotional reactivity domain continuing to make a significant contribution even after ACES scores were added.

Our findings differ from studies that found healthcare provider discussions about emergency preparedness can benefit patients’ household emergency preparedness (47, 48). Few participants in our study (n = 17, 5.5%), however, indicated discussing household emergency preparedness with their healthcare provider. This finding is consistent with other studies in which few participants and providers indicate discussing household emergency preparedness together (47, 49, 50). The rarity of such communication limits our ability to detect meaningful effects. Although few participants in our study had discussed household emergency preparedness with their healthcare provider, those who did were more likely to have experienced injury or illness from a disaster or to have more health care needs. This is consistent with research indicating that patients identified at high or medium risk for impact from disasters were more likely to receive education on household emergency preparedness than those with low risk or less complex needs (51).

Lack of time is among the reasons why healthcare providers do not consistently initiate household emergency preparedness education with patients (49–51). In our study, those who did have these conversations were more likely to have medical issues that could be affected by disasters, such as requiring assistance with activities of daily living or needing special medical equipment. This suggests that the conversations were at least started to address these needs, whether or not the providers then took the opportunity to expand upon the topic is unknown. It is possible that household emergency preparedness is more “top of mind” for healthcare providers when working with more complex patients; specifically exploring rationale for how healthcare providers prioritize household emergency preparedness education could be valuable in understanding how and when such interactions take place.

We did offer some insight into those whom may be receptive to these sorts of conversations. Responses to the Chapman University Survey on American Fears reflect interest in preventing what participants generally believe are possible and significant disasters, even if they do not believe themselves to be at serious risk. Responses also suggest both concern about not being prepared and belief that preparations can help. Conscientious participants also tended to be better prepared and to generally present more favorable traits, like higher levels of resilience. The Chapman-measured beliefs were associated with the actual levels of preparedness measured by the HEPI. This was especially true of “I sometimes feel guilty that I have not done enough to prepare for disasters” and “I feel confident that I know how to prepare for disasters.”

Adopting an all-people, all-hazards, all-agencies mindset when viewing the ACEs-HEPI theoretical framework for this study, we recognize the key importance of considering subpopulations that may bear a higher burden of risk, such as children experiencing ACEs and adult ACE survivors. Although the majority of participants (74.28%) in this study reported having endured at least one ACE and low levels of household emergency preparedness, this study offers insight on participant characteristics and preparedness behaviors.

Disaster preparedness interventionists, whether they be healthcare providers, public health personnel, community health workers, faith- or community-based leaders, or emergency mangers, can assist community members with understanding their disaster risks by discussing the disasters that are most likely to occur in the community and the causes and prevention of morbidity and mortality outcomes for a recent local disaster. Even if the community members did not experience direct adverse outcomes from the recent local disaster, they can experience them vicariously through these discussions, which then may motivate them to enhance their own preparedness.

The constructs of Motivation and Healthy Coping strategies, as assessed by the Chapman University Survey on American Fears scale, highlight variability in household emergency preparedness behaviors. Despite feeling confident in knowing how to prepare, many participants (40.26%) expressed feelings of guilt or concern about not having done enough to prepare for a disaster. The COVID-19 pandemic exemplified concern and guilt related to the lack of preparedness on many levels, leading to high mortality and morbidity rates, and profound emotional toll across the country (52). Feelings of guilt can be exacerbated if one survives a disaster and their friends and family do not survive (53).

A potentially useful intervention for improving preparedness is motivational interviewing. This is a therapeutic communication technique used to enhance inherent motivation toward specific client goals by “evoking a person’s reasons, desires, and willingness for change using the client’s own speech as a means of clarifying and strengthening their intent” (54), p. 358. Because feeling guilty about not being prepared enough and feeling confident that one knows how to prepare for disasters both predicated higher preparedness levels in this study, interventionists should consider these characteristics when providing disaster preparedness motivational interviewing interventions. Through this technique, the interventionist strives to have the community member come to the conclusions on their own that they may not be fully prepared to endure disaster conditions, that lack of preparedness may impact loved ones or disaster responders if they must be rescued during dangerous storm conditions, and that now, because they have preparedness educational materials from the interventionist, they know how to prepare and can begin preparations. It is plausible that motivational interviewing interventions will result in motivation to start disaster preparedness planning. It is also important to note that motivating community members to equally endorse measurable preparedness planning and disaster supply stockpiling behaviors fosters overall household emergency preparedness. Endorsement of emergency preparedness behaviors contributes to safer post-disaster self-recovery and may minimize the perceived adversity of the disastrous event.

In a recent observational study of US households looking at motivators for disaster preparedness, Miao and Zhang (55) found that recent disaster experience increased participants’ perceived preparedness self-efficacy and propensity to stockpile supplies and make home emergency plans. Conversely, our participants with recent disaster experience were found to be less prepared than the participants who had not experienced a recent disaster. Further research of the effect of recent disaster experience on preparedness self-efficacy is warranted.

Study participants who reported feeling confident in preparing for disasters reported higher levels of preparedness. Similar to the results in this study, Rao et al. (56) reported that individuals who had high confidence in their personal capacity to respond to a disaster had higher overall levels of preparedness. Overconfidence, however, has been viewed as a cognitive bias that impairs an individual’s ability to safely assess when they may experience a future disaster (57). The American Psychological Association (58) describes overconfidence as “an overestimation of one’s actual ability to perform a task successfully, by a belief that one’s performance is better than that of others, or by excessive certainty in the accuracy of one’s beliefs” (para. 1). In this study, over half of the participants (52.1%) disagreed to the following Chapman University Survey on American Fears item, “I will experience a significant natural or manmade disaster in the near future.” Inaccurate disaster predictions, coupled with a lack of household emergency preparedness behaviors, can have devastating individual- and community-level consequences. Interventionists can enhance preparedness self-efficacy by empowering community members with resources and knowledge on how to best prepare for disasters and expressing confidence in the community members’ abilities to make the necessary preparations.

Finally, the current study suggests that those who have the means to prepare have done so or are actively concerned about preparing. These findings are congruent with McNeill et al. (59) where study participants frequently expressed financial resources are a barrier to emergency preparedness. Resolution of this will require equitable emergency preparedness assistance to those without financial resources to purchase emergency preparedness supplies (59) so that existing inequities are not exacerbated after a disaster (60).

In this study, conscientious participants tended to be better prepared and present more favorable traits, like higher levels of resilience. With older or medically frail community members, a focus on client conscientiousness can aid interventionists in assessing how these community members at disproportionate risk for negative disaster impacts might respond to challenges in health, including their likelihood of being prepared for a disaster inclusive of their healthcare needs. The trait of conscientiousness shapes how individuals will respond to such challenges and impacts their behaviors, decisions, and overall well-being, thereby significantly influencing their resilience and health outcomes.

The trait of resilience can aid an understanding of health trajectories in the midst of disasters and emphasize the need for interventionists to focus on mental health of their clients before, during, and after a disaster, particularly the mental health of community members with a history of ACEs. It is critical for interventionists to have a broad view of what constitutes ACEs. Current work in this area focuses not only on ACEs including abuse, violence, incarceration, and homelessness, but also ACEs including poverty, discrimination, violence, poor housing, and lack of opportunity (61). All of these facets of conscientiousness and resilience collectively impact the health and health trajectory of community members.

Emotional reactivity was one of the strongest indicators of concerns about and preparations for disasters. Similar to results found by Reuben et al. (62) and Grusnick et al. (63), we found that ACEs were positively correlated with neuroticism. Emotional reactivity, including neuroticism, was found by Cloitre et al. (64) and Jirakran et al. (65) to mediate the relationship between ACEs and health outcomes of poor physical health, PTSD symptoms, depression, and suicidal behavior. Neuroticism has also been associated with mental health issues, stress, and loneliness (66, 67). We found that higher levels of emotional reactivity were significantly associated with lower levels of household emergency preparedness, identifying another mechanism that excessive emotional reactivity may predispose one to adverse outcomes.

With the increasing frequency and intensity of disasters and humanitarian crises, efforts to promote resilience need to be prioritized. Without such efforts, incidence and prevalence of anxiety, depression, post-traumatic stress disorder, complicated grief, and suicidal ideation are expected to increase (68). In fact, the number of people who experience disaster-related mental health issues regularly outweigh those with disaster-related physical injuries (69). In order to mitigate the risks to mental health and resilience at the community level, researchers have proposed the formation of community Resilience Coordinating Networks where “multi-sector coalitions use a public health approach to empower residents to use their existing strengths and resources, and form additional ones, to prevent and heal mental health problems and turn adversities into opportunities to pursue innovative solutions” (69), para. 19. In order to mitigate the risk to mental health and resilience at the individual level, the concept of “healthy hope” can be fostered. Healthy hope is described as the belief that individuals can improve their future by consciously choosing significant personal goals, finding strategies to help them achieve those goals, and mobilizing their willpower to adhere to their plans even when challenges occur (70). Encouraging disaster preparations can foster healthy hope that a disaster can be endured safely and with minimal discomfort.

This was a cross-sectional, observational study with self-reported outcomes and retains the limitations of such a design. We determined a priori that an adequate sample size for this study was at least 135 participants. With 311 respondents, we exceeded our desired minimum sample size. Although the sample was of adequate size to find real effects and presents a good geographical representation of the US population, there appears to have been enough self-selection to introduce a bias. Recruitment was conducted from a random selection of a national sample, but potential participants self-enrolled. It is likely that these participants tended to be those who are already more concerned about disaster preparedness. Our sample was more affluent, educated, and urban than the general population, and renters were overrepresented. Affluent households may have additional resources available to undertake preparedness actions. Apartment dwellers may encounter significant barriers to stockpiling supplies because of restricted storage capacity. Additionally, as tenants, renters may have little control over building-level preparedness managed by property owners or staff. Gender and race perspectives may also have been limited in our findings since white women were overrepresented.

Self-report and social-desirability bias are also potential threats to validity. However, to minimize these threats, respondents self-reported anonymously to an internet survey, not a live person, and we assured them that there were no right or wrong answers, potentially decreasing social desirability bias. Possible information bias may have occurred from respondent burden due to the length of the survey. To overcome this bias, we added an “are you still paying attention” question in the survey, which was correctly answered by all of the participants.

A final limitation to consider involves potential recall bias because the participants answered questions about their past experiences with disasters and ACEs.

Our research questions could be examined with a stronger study design, using randomness, confirmation of self-report variables, and prospective data collection. Researchers can also consider using the Social-Ecological Model of Risk and Resilience in future research (71). The socioecological systems in which people exist could be relevant to looking at disaster preparedness, resilience, and ACEs, given the relationship of income and social capital to preparedness levels, and that higher prevalence rates of ACEs are noted among populations with less education or income levels and in socially marginalized groups (72).

The survey did not distinguish between who initiated the healthcare provider/patient discussion of household emergency preparedness, so we do not know if participants perceived themselves to be at greater risk and thus sought out additional information, or if the healthcare provider perceived the participant to be at greater risk and thus prioritized this conversation. Although research has not consistently shown a significant positive relationship between healthcare provider discussion and promotion of household emergency preparedness and their patients’ household emergency preparedness, the rarity of such interactions may affect the ability to detect significance. Additional research into mitigation of barriers to meaningful dialog and educational strategies is warranted so that the potential to enhance health through household emergency preparedness is not overlooked.

Exposure to disaster events should continue to be evaluated as an ACE to determine associations with physical and emotional health outcomes in adulthood. Additional model inquiry with a larger sample can test a structural equation model explicitly testing all of the relationships, strengthening future use of the model.