Disaster preparedness among women with a recent live birth in Hawaii, a cross-sectional study, results from the Pregnancy Risk Assessment Monitoring System (PRAMS), 2016

Data Collection

PRAMS is a state, population-based sample of women who recently had a live birth. State birth certificate files are used to select approximately 200 women each month in Hawaii. The surveys are self-reported and follow the systematic PRAMS methodology; previously described by Shulman et al. (2018). Women receive the survey approximately two months after delivery. Contact is made initially by mail and then by phone. In Hawaii, PRAMS is only offered in English. Those who participate receive a $10 gift card to a local food market. Consent by mail is implied by returning a completed questionnaire; verbal consent is provided by phone. The protocol is IRB reviewed and approved by the CDC and the Hawaii State Department of Health.

Variables

The eight pre-tested, standardized PRAMS emergency preparedness questions are listed in Table 1 (Centers for Disease Control and Prevention 2016). Data from the Hawaii birth certificate included maternal age, maternal education, marital status, urban or rural residence, maternal race, and participation in Special Supplemental Nutrition Program for Women Infants and Children (WIC) during pregnancy. Age was collapsed into five categories: 19 years or less, 20 to 24 years, 25 to 29 years, 30 to 34 years, and 35 years or greater. Education was categorized as less than high school diploma, high school diploma or equivalent, some college, and baccalaureate degree or higher. The urban or rural residence variable was established by county, and Maui was classified as rural according to the 2012 Census classification. Race was categorized by the racial categories specific to Hawaiian birth certificates: White, Native Hawaiian, Filipino, Japanese, Other Pacific Islander, and other. Anyone selecting Native Hawaiian or part-Hawaiian was grouped into the Native Hawaiian race. Other Pacific Islanders included those selecting the category, and Samoans and Guamanians. All other races were categorized as “other”, consistent with methods described by Sorenson et al. (2003).

Information obtained from PRAMS included current health insurance at time of survey, total income, and family size. Insurance status was categorized into three options: private, public, or none. Individuals that indicated having both private and public insurance were categorized as having private insurance. Military insurance was captured as a separate option on the survey and categorized as private for the analysis. Total income captured the reported household income 12 months prior to the birth, and family size was defined as number of individuals living on the total income in the 12 months prior to the live birth. Family sizes of five or greater were collapsed into a single category. A variable categorizing income as a percent of the federal poverty level (FPL) was developed from total income and family size, compared to the January 2016 poverty guidelines for Hawaii issued by the Federal Register of the Department of Health and Human Services (Department of Health and Human Services 2016).

Four FPL groups are reported as less than or equal to 100%, 101-185%, 186-300%, and greater than 300% FPL, consistent with federal programing eligibility and Hawaii PRAMS trend report (Hawaii 2019; Department of Health and Human Services 2016).

Statistical Analysis

The sample is weighted to be representative of the state by accounting for sampling stratification, nonresponse, and noncoverage. The sample is stratified initially by county (Honolulu, Maui, Hawaii, and Kauai) and within Honolulu county, by birthweight.

Factor analysis was performed to assess underlying factors among the eight emergency preparedness behaviors. The dataset was converted to a tetrachoric correlation matrix, and factor analysis was performed using the maximum likelihood estimate and varimax rotation. The overall measure of sampling adequacy was 0.91 suggesting factor analysis could be used for these data (Kaiser 1974).

Data analysis was performed in SAS-callable SUDAAN 11.0 (SAS v9.4) to account for weighted data and complex survey methods. Multivariable logistic regression models were used to estimate predicted marginal prevalence ratios to assess characteristics of women associated with emergency preparedness behaviors. White was the referent category for race, otherwise the subcategory with the greatest proportion of women was used as the referent category. Variables included in the model were established a priori from existing literature and included age, education, marital status, urban or rural residence, race, insurance status, family size, FPL, and use of WIC during pregnancy.

Study Characteristics

In 2016, 1,999 women were contacted for PRAMS in Hawaii and 1,076 responded (weighted response rate 56.3%). The final sample for this analysis included 1,010 women; 66 were excluded for not answering any of the emergency preparedness questions. Table 2 presents the demographic characteristics of the 1,010 women included in this analysis.

The largest proportion of respondents for each demographic characteristic were aged 30-34 (31.2%), had completed a baccalaureate degree or higher (34.7%), were married (63.4%), lived in an urban area (72.3%), were Native Hawaiian race (30.3%), had private insurance at the time of the survey (69.1%), had a family size of two (30.4%), had an income greater than 300% FPL (26.5%), and did not use WIC (65.0%).

Preparedness Behaviors

Among all women, 79.3% (n=826) reported at least one preparedness behavior, and 11.2% (n=96) reported all eight. Having emergency supplies was most commonly reported with 63.0% (95% CI: 58.7-67.1%) of women having enough supplies at home for at least seven days, followed by 41.3% (95% CI: 37.1-45.6%) having an evacuation plan for their child(ren). Less than 40% of women reported other preparedness behaviors: having an emergency meeting place (37.8%, 95% CI: 33.7-42.1%); practicing what to do in case of a disaster (31.6%, 95% CI: 27.7-35.8%); having a plan to keep in touch with family if separated (38.7%, 95% CI: 34.5-43.0%); having an evacuation plan to leave home (36.6%, 95% CI: 32.6-40.9%); having copies of important documents (31.8%, 95% CI: 27.9-36.0%); and having emergency supplies to take during an evacuation (34.9%, 95% CI: 30.9-39.2%). (Table 3)

Among 20.7% (n=184) of women reporting zero preparedness behaviors, significant differences were observed in urban or rural residence and race. Twenty-three percent of women living in an urban area and 16% of women living in a rural area reported zero preparedness behaviors (p=0.019). Of the six races reported, Japanese women had the highest proportion reporting zero preparedness behaviors, while Other Pacific Islanders had the lowest proportion reporting zero preparedness behaviors. Reporting of zero preparedness behaviors did not vary by education, marital status, insurance, family size, income, or WIC participation. (Table 4)

We assessed factor loading plots and determined the eight preparedness behaviors could be described by three factors (Online Resources 1-3). The first factor, having emergency plans, captured women that responded yes to at least one of the behaviors: emergency meeting place, practiced what to do, plan to keep in touch, plan for themselves to evacuate, or an evacuation plan for their child(ren). The second factor included women who responded yes to having copies of important documents, and the third factor, included women that had emergency supplies for at least seven days and/or had emergency supplies prepared if they had to leave quickly.

Within the sample, 59.8% (95% CI: 55.5-64.0%) of women had emergency plans; 31.8% (95% CI: 27.9-36.0%) had copies of important documents; and 66.8% (95% CI: 62.5-70.8%) had emergency supplies (Table 3). Having emergency plans varied by race; among Other Pacific Islanders, 79% reported an emergency planning behavior compared to 50% of Japanese women. Emergency planning also varied by family size and percent of the FPL. More than 70% of women with a family size of four or greater reported having emergency plans, while 53.5% (95% CI: 45.5, 61.3) of women with a family size of two reported these behaviors. Among women with an income of 186-300% FPL, 70.7% (95% CI: 60.6, 79.1) reported having emergency plans, whereas 50.3% (95% CI: 60.6, 79.1) of women with an income more than 300% FPL reported emergency planning behaviors. Demographic characteristics among women reporting having copies of important documents varied significantly by race and FPL. Having copies of important documents was most commonly reported among Native Hawaiians (44.5%, 95% CI: 36.9, 52.4). In contrast, 11.9% (95% CI: 6.0, 22.4) of Japanese women reported having copies of important documents. A greater percent of women living at 101-185% FPL (38.5%, 95% CI: 30.3, 47.4) had copies of important documents compared to 22.5% (95% CI: 16.5, 29.8) of women living at more than 300% FPL. Reporting of emergency supplies varied by urban or rural residence. Sixteen percent of women living in a rural residence (15.8%, 95% CI: 13.1, 18.9) and 23.6% (95% CI: 18.1, 27.9) of women living in an urban residence had emergency supplies. (Table 4)

Models

The results of the multivariable analysis are presented in Table 5. Results from women 19-years-old and younger, Other Pacific Islanders, and those with no insurance at the time of survey, should be interpreted with caution as their sample size was small (30-59 respondents). Education and marital status were not associated with any preparedness factors. The prevalence of completing at least one planning emergency preparedness behavior differed significantly by race, family size, and poverty level. Other Pacific Islanders were 45% more likely to report having emergency plans compared to White women (aPR 1.45, 95% CI: 1.09-1.93). Women with a family size of one were 33% (aPR 0.67, 95% CI: 0.46-0.96) less likely to report having emergency plans compared to women with a family size of two. In contrast, having emergency supplies is more common among women with a family size of four (aPR 1.31, 95% CI: 1.06, 1.61) and five or more (aPR 1.33, 95% CI: 1.04-1.70) compared to those with a family size of two. Compared to women with an income above 300% FPL, those with an income between 186 and 300% FPL were 28% (aPR 1.28, 95% CI: 1.03-1.60) more likely to have emergency plans.

Having copies of important documents significantly differed in the adjusted models by residence, race, insurance status, family size, and WIC participation. This behavior was less common among women living in a rural residence compared to women living in an urban residence (aPR 0.75, 95% CI: 0.60, 0.95), and among women that used WIC during pregnancy compared to those that did not (aPR 0.72, 95% CI: 0.53, 0.98). Native Hawaiian (aPR 2.16, 95% CI: 1.45-3.24), Filipino (aPR 1.69, 95% CI:1.06-2.69), and Other Pacific Islanders (aPR 2.02, 95% CI: 1.10-3.71) were more likely than White women to have copies of important documents. Additionally, having copies of important documents was more common among women with public insurance compared to women with private insurance (aPR 1.42, 95% CI: 1.04, 1.93), and among women with a family size of four compared to women with a family size of two (aPR 1.50, 95% CI: 1.02, 2.20).

Significant differences were noted by age, race, family size, and FPL in the adjusted model for emergency supplies. Women 19-years-old and younger were 33% more likely to have emergency supplies compared women 30-34 years old (aPR 1.33, 95% CI: 1.14-1.55). Having emergency supplies was also more likely among Other Pacific Islanders compared to White women (aPR 1.44, 95% CI: 1.18-1.76), and women with a family size of five or more, compared to those with a family size of two. Women with an income at or less than 100% FPL were 26% (aPR 0.74, 95% CI: 0.57-0.97) less likely to have emergency supplies compared to women with an income more than 300% FPL.