Preparedness for Natural Disasters Among Older US Adults
Preparedness for Natural Disasters Among Older US Adults
The HRS is a nationally representative cohort study that surveys the social, economic, and health characteristics of Americans aged 50 years and older. Supported by the National Institute on Aging and the Social Security Administration, it explores issues related to work characteristics and the retirement process, changes in labor force participation, the evolving social and economic status of US families, and their health transitions over time. Biennial surveys have been conducted since 1992. In addition to the core survey items administered to all respondents, each survey wave contains a set of "modules." These modules are administered to randomly chosen subsamples of the survey population, each module comprising approximately 10% of the respondents in that wave. A detailed description of HRS methods and questionnaires is contained on its Web site (http://hrsonline.isr.umich.edu). In the 2010 survey, one of the modules, with 21 questions (with a 97.0% response rate), was administered concerning disaster preparedness. Nursing home and other institutional residents and persons aged younger than 50 years (generally younger spouses of the target respondents) were excluded from the analysis, yielding a final sample size of 1304. The rate of item-specific missing data was less than 1.5% for all questions used in this analysis.
Preparedness module questions were assembled from previous local surveys, previously published relevant articles, disaster preparedness technical reports, and discussions with geriatricians and others familiar with emergency situations. We derived health and demographic variables used in this analysis from core HRS questionnaires.
Demographic variables included race/ethnicity, gender, date of birth, and educational attainment. Economic status was represented by annual household income. Individuals were designated as living alone if no other person resided in the household and if the respondent reported that he or she was married but not residing with his or her spouse. Body mass index (BMI) was calculated with the formula weight in kilograms divided by the square of height in meters. Type of housing was classified into 1-family house, apartment, or townhouse; multifamily unit; mobile home; or "other." Overall self-rated health status was classified as excellent or very good, good, or fair or poor. Functional status was assessed with ADLs and IADLs—ADL dysfunction was assessed by self-reported difficulties on 5 tasks: bathing, eating, dressing, toileting, and transferring; IADL dysfunction included difficulties with managing money, managing medications, preparing meals, going shopping, and using a telephone.
We assessed 18 disaster-preparedness indicators in this study. Some questions assessed residential preventive measures, such as having a smoke or fire detector and whether it was tested in the past year, if natural gas supply is used and knowing how to turn it off, and if multiple exits existed in place of residence in case of blockage in an event of emergency. Other items assessed household member disaster preparation activities and individual efforts exerted by the respondent or his or her household's members in disaster preparation. These activities included having a specific disaster plan, written or otherwise, on what to do in case of disaster; being able to receive emergency information by having a battery-operated radio; and having assembled a 3-day supply of food, water, medications, and other necessities.
We calculated an 18-point summary score by summing up the yes-or-no answers for all the disaster preparedness indicators asked in the module. We gave each "yes" answer 1 point and each "no" a 0 for the first 16 questions. We scored related (follow-on) questions 18 and 19, and 20 and 21 each as 1 question (18 and 19: 0 for yes and no, or 1 for yes and yes or no and not applicable; 20 and 21: 0 for no and no, and 1 for either answered as yes). We reverse-scored 2 items (6 and 14) so that a "yes" answer was in a positive direction and a "no" answer was in the negative direction. Lower scores indicated worse preparedness, higher scores better preparedness. We assessed reliability for the preparedness summary score by using Cronbach α (0.61).
We queried respondents on being aware of programs or organizations that work to help prepare people for the possibility of disasters and whether they were registered in any community program or in any medical or other organization that would offer help in the event of a disaster. Also, we asked them if they had participated in an educational program offering a lecture or discussion and had read any materials that familiarized them with signs of disasters and preparation for disasters before the events occurred. Items on respondents' readiness for evacuation included availability of friends or relatives within 50 miles who could offer emergency shelter, having means of transportation or being able to secure one if needed, and whether individuals were able to quickly evacuate their residences without the help of others. We further assessed the reasons for slow exit. We queried about hearing impairment that precluded hearing warning sirens, as well as the use of medical devices requiring electricity. Finally, we queried participants about previous experience with a natural disaster.
Analytic files of HRS data prepared by the RAND Center for the Study of Aging provided processed data items and sampling weights. We compared derived variables with raw questionnaire data before use in analysis. Outcomes included the individual disaster preparedness questions and the preparedness summary score (continuous and classified by 0–9, 10–12, or 13–18). Covariates of interest included age (continuous and grouped by 50–64, 65–69, 70–79, or 80–98 years), gender, race/ethnicity (Black, White, or other), education in years (0–9, 10–11, 12–13, or > 13), marital status (married, partnered, or single), BMI (< 25, 25–29, or ≥ 30), self-perceived health status (excellent or very good, good, or fair or poor) living alone (yes or no), type of residence (house, mobile home, multiunit dwelling, or other), ADL limitations (none, 1–2, 3–4, or ≥ 5) and IADL limitations (none, 1, or ≥ 2).
Frequencies and percentages in Table 1 were unweighted and included 1304 respondents. We used weighted analyses for population estimates in Table 2 and Table 3 and included 1225 respondents, who had HRS-provided sampling weights. We generated unadjusted frequencies and percentages by using categorical data procedures that produced estimates of population proportions, population proportions, and their standard errors. We generated unadjusted and adjusted population P values, least squares means, and standard errors by using regression analysis methods for weighted sample survey data. We generated global P values with the F-test statistic and we assessed comparisons between categories with the t test. Adjustment variables included categorical age, gender, race, categorical education, and categorical income. We also assessed all 2-way interactions among covariates. We performed analyses with SAS version 9.2 (SAS Institute Inc, Cary, NC). We assessed reliability for the preparedness summary score by using Cronbach α (0.60; 95% confidence interval [CI] = 0.56, 0.63).
Methods
The HRS is a nationally representative cohort study that surveys the social, economic, and health characteristics of Americans aged 50 years and older. Supported by the National Institute on Aging and the Social Security Administration, it explores issues related to work characteristics and the retirement process, changes in labor force participation, the evolving social and economic status of US families, and their health transitions over time. Biennial surveys have been conducted since 1992. In addition to the core survey items administered to all respondents, each survey wave contains a set of "modules." These modules are administered to randomly chosen subsamples of the survey population, each module comprising approximately 10% of the respondents in that wave. A detailed description of HRS methods and questionnaires is contained on its Web site (http://hrsonline.isr.umich.edu). In the 2010 survey, one of the modules, with 21 questions (with a 97.0% response rate), was administered concerning disaster preparedness. Nursing home and other institutional residents and persons aged younger than 50 years (generally younger spouses of the target respondents) were excluded from the analysis, yielding a final sample size of 1304. The rate of item-specific missing data was less than 1.5% for all questions used in this analysis.
Preparedness module questions were assembled from previous local surveys, previously published relevant articles, disaster preparedness technical reports, and discussions with geriatricians and others familiar with emergency situations. We derived health and demographic variables used in this analysis from core HRS questionnaires.
Demographic variables included race/ethnicity, gender, date of birth, and educational attainment. Economic status was represented by annual household income. Individuals were designated as living alone if no other person resided in the household and if the respondent reported that he or she was married but not residing with his or her spouse. Body mass index (BMI) was calculated with the formula weight in kilograms divided by the square of height in meters. Type of housing was classified into 1-family house, apartment, or townhouse; multifamily unit; mobile home; or "other." Overall self-rated health status was classified as excellent or very good, good, or fair or poor. Functional status was assessed with ADLs and IADLs—ADL dysfunction was assessed by self-reported difficulties on 5 tasks: bathing, eating, dressing, toileting, and transferring; IADL dysfunction included difficulties with managing money, managing medications, preparing meals, going shopping, and using a telephone.
We assessed 18 disaster-preparedness indicators in this study. Some questions assessed residential preventive measures, such as having a smoke or fire detector and whether it was tested in the past year, if natural gas supply is used and knowing how to turn it off, and if multiple exits existed in place of residence in case of blockage in an event of emergency. Other items assessed household member disaster preparation activities and individual efforts exerted by the respondent or his or her household's members in disaster preparation. These activities included having a specific disaster plan, written or otherwise, on what to do in case of disaster; being able to receive emergency information by having a battery-operated radio; and having assembled a 3-day supply of food, water, medications, and other necessities.
We calculated an 18-point summary score by summing up the yes-or-no answers for all the disaster preparedness indicators asked in the module. We gave each "yes" answer 1 point and each "no" a 0 for the first 16 questions. We scored related (follow-on) questions 18 and 19, and 20 and 21 each as 1 question (18 and 19: 0 for yes and no, or 1 for yes and yes or no and not applicable; 20 and 21: 0 for no and no, and 1 for either answered as yes). We reverse-scored 2 items (6 and 14) so that a "yes" answer was in a positive direction and a "no" answer was in the negative direction. Lower scores indicated worse preparedness, higher scores better preparedness. We assessed reliability for the preparedness summary score by using Cronbach α (0.61).
We queried respondents on being aware of programs or organizations that work to help prepare people for the possibility of disasters and whether they were registered in any community program or in any medical or other organization that would offer help in the event of a disaster. Also, we asked them if they had participated in an educational program offering a lecture or discussion and had read any materials that familiarized them with signs of disasters and preparation for disasters before the events occurred. Items on respondents' readiness for evacuation included availability of friends or relatives within 50 miles who could offer emergency shelter, having means of transportation or being able to secure one if needed, and whether individuals were able to quickly evacuate their residences without the help of others. We further assessed the reasons for slow exit. We queried about hearing impairment that precluded hearing warning sirens, as well as the use of medical devices requiring electricity. Finally, we queried participants about previous experience with a natural disaster.
Analytic files of HRS data prepared by the RAND Center for the Study of Aging provided processed data items and sampling weights. We compared derived variables with raw questionnaire data before use in analysis. Outcomes included the individual disaster preparedness questions and the preparedness summary score (continuous and classified by 0–9, 10–12, or 13–18). Covariates of interest included age (continuous and grouped by 50–64, 65–69, 70–79, or 80–98 years), gender, race/ethnicity (Black, White, or other), education in years (0–9, 10–11, 12–13, or > 13), marital status (married, partnered, or single), BMI (< 25, 25–29, or ≥ 30), self-perceived health status (excellent or very good, good, or fair or poor) living alone (yes or no), type of residence (house, mobile home, multiunit dwelling, or other), ADL limitations (none, 1–2, 3–4, or ≥ 5) and IADL limitations (none, 1, or ≥ 2).
Frequencies and percentages in Table 1 were unweighted and included 1304 respondents. We used weighted analyses for population estimates in Table 2 and Table 3 and included 1225 respondents, who had HRS-provided sampling weights. We generated unadjusted frequencies and percentages by using categorical data procedures that produced estimates of population proportions, population proportions, and their standard errors. We generated unadjusted and adjusted population P values, least squares means, and standard errors by using regression analysis methods for weighted sample survey data. We generated global P values with the F-test statistic and we assessed comparisons between categories with the t test. Adjustment variables included categorical age, gender, race, categorical education, and categorical income. We also assessed all 2-way interactions among covariates. We performed analyses with SAS version 9.2 (SAS Institute Inc, Cary, NC). We assessed reliability for the preparedness summary score by using Cronbach α (0.60; 95% confidence interval [CI] = 0.56, 0.63).