Abstract
Objectives. We describe the demographics of the decedents from the tornado outbreak in Alabama on April 27, 2011; examine the circumstances of death surrounding these fatalities; and identify measures to prevent future tornado-related fatalities.
Methods. We collected information about the decedents from death certificates, disaster-related mortality surveillance, and interview data collected by American Red Cross volunteers from the decedent’s families. We describe demographic characteristics, circumstances and causes of death, and sheltering behaviors before death.
Results. Of the 247 fatalities, females and older adults were at highest risk for tornado-related deaths. Most deaths were directly related to the tornadoes, on scene, and trauma-related. The majority of the deceased were indoors in single-family homes. Word of mouth was the most common warning mechanism.
Conclusions. This tornado event was the third deadliest in recent US history. Our findings support the need for local community shelters, enhanced messaging to inform the public of shelter locations, and encouragement of word-of-mouth warnings and personal and family preparedness planning, with a special focus on assisting vulnerable individuals in taking shelter.
Tornadoes are one of the most deadly natural disasters in the United States and continue to be a major public health concern in the Midwestern, South Central, and Southeastern states.1 Approximately 1200 tornadoes are detected in the United States each year, on average killing 60 to 65 people and injuring 1500 people annually.2 Since 2007, the United States has used the Enhanced Fujita (EF) scale to categorize tornadoes on a scale of 0 to 5, using wind speed estimates based on structural and tree damage.3 Although EF-4 (wind speed of 166–200 mph) and EF-5 (wind speed of > 200 mph) tornadoes make up fewer than 1% of all tornadoes detected, they are responsible for 70% of all tornado-related deaths.2
On April 23, 2011, the National Weather Service forecasted an upcoming storm. This forecast was followed by 351 tornadoes that swept through the Southeastern United States from April 25 to 28, 2011, causing widespread damage and resulting in 338 fatalities in Arkansas, Mississippi, Alabama, Tennessee, and Georgia.4 The deadliest day was April 27, 2011, when a record number of 62 tornadoes, including 8 EF-4 and 3 EF-5 tornadoes, struck Alabama, resulting in the third deadliest tornado event in the United States since 1950.5 In Alabama, these tornadoes resulted in more than 200 fatalities, 2000 injuries, and $4.2 billion in property damage.5 The tornadoes struck Alabama in 3 waves on April 27—starting at 4:01 am, 11:15 am, and 2:40 pm—affecting 35 counties and causing deaths in 19 counties.6 The last wave was the most destructive, destroying the entire infrastructure of the rural towns of Hackleburg and Phil Campbell, and severely affecting cities including Tuscaloosa and Birmingham. The last wave did not end until 9:50 pm. On that day, several tornadoes traveled more than 100 miles, with the total path for all tornadoes covering a distance of 1206 miles.6
Although not foreseen to be so destructive, the storm was forecasted days ahead. The average time between public notification and tornado touchdown on April 27 was 2.4 hours for tornado watches and 22 minutes for tornado warnings, exceeding the target time of 15 minutes set by the National Weather Service.5,7 Yet, the storm resulted in mass fatalities. The objectives of our study were to describe the decedents, examine the circumstances of death, and identify measures to prevent future tornado-related fatalities.
METHODS
The case definition included all deaths directly or indirectly related to the tornadoes on April 27, 2011, in Alabama.8 A death was defined as directly related if it was caused by the environmental forces of the disaster (e.g., strong wind) or by the direct consequence of these forces (e.g., flying debris). A death was defined as indirectly related if it was caused by unsafe or unhealthy conditions generated by the disaster (e.g., hazardous roads) or a loss or disruption of usual services (e.g., a power outage). We collected information about the deceased through 3 different sources.
Data Sources
American Red Cross’s casework records of families of the deceased.
Immediately after the tornadoes on April 27, 2011, the Red Cross attempted to contact all affected households with a tornado-related decedent to provide emotional and financial support. These visits were documented in casework records if the families accepted Red Cross support. We identified 159 deaths from these records, which captured information on household demographics and circumstances of death, obtained during semistructured interviews.
Disaster-related mortality surveillance.
The Red Cross routinely conducts mortality surveillance during disasters, using information from multiple sources. Medical examiners and coroners were the main information sources (n = 210; 89.4%), followed by family members (n = 92; 39.1%). We identified 235 deaths from this surveillance system, which captured individual demographic information and circumstances of death such as date, location, and cause of death.
Death certificates.
We obtained death certificates for all 247 deaths in which the tornado-related injury occurred in Alabama. We obtained death certificates by specifically looking up deaths identified through the other 2 sources (n = 235). We also searched vital records for deaths with International Classification of Diseases, 10th Revision9 code X37 (death by cataclysmic storm) occurring from April 27 to December 31, 2011 (which we believe captures the majority of tornado-related deaths) and identified 12 additional deaths. All death certificates were obtained from Alabama, except for 4 that were obtained from Tennessee and Mississippi (where the deaths occurred after the initial injury in Alabama). Information captured from death certificates included occupation and education, physical address at the time of injury and death, and the official date and cause of death. The death certificate served as our gold standard because it contained the most accurate information for many of the variables, specifically demographics, cause of death, and address of injury and death.
Data Compilation and Coding
We combined information from these sources into a master database. We coded the Red Cross interview narratives from the Red Cross casework record into 81 variables that included categories such as the location of the deceased during the tornado (e.g., rural or urban, indoor or outdoor), the type of warning received (e.g., warned by others, siren, or television), behavior after receiving warning (e.g., action taken, hideout location), the mechanism of death, and social (e.g., living alone) and physical vulnerability (e.g., having a physical disability), which are characteristics that may decrease the individual’s ability to respond to or recover from a disaster. On the basis of preestablished definitions, 2 reviewers independently coded the variables for all deaths with Cohen κ coefficients ranging from 0.91 (95% confidence interval [CI] = 0.84–0.97) to 1. A third independent reviewer resolved coding discrepancies.
Geocoding and Mapping of Fatalities and Tornado Paths
We were able to obtain the decedents’ location of injury during the tornado for 237 individuals and geo-coded these using ArcMap 10.0 mapping software (ESRI, Redlands, CA). Where an exact match could not be mapped, the decedent’s location was plotted to the center of the zip code. We obtained tornado path and severity data from the National Weather Service and overlaid these on the locations of deaths to obtain an overview of the fatalities and the corresponding tornado. We classified the decedents’ location during the tornado as urban or rural according to the 2010 Census definition.10
Data Analysis
To calculate approximate risk ratios, we conducted a study similar to a retrospective cohort study. We approximated the population at risk from the deadly tornadoes by identifying the census blocks within a 1.5-mile radius of each of the 12 deadly tornadoes (those that killed at least 1 person) using Geographic Information System and 2010 Census data (figure available as a supplement to the online version of this article at http://www.ajph.org). Using ArcMap 10.0, we identified 14 335 census blocks (containing 300 142 residents) that were within or overlapped the 1.5-mile radius. We used this population to calculate the approximate risk ratio of tornado-related death for the demographic variables available in the 2010 Census. We conducted a sensitivity analysis and expanded the definition of the population at risk to census blocks within 3 miles and 5 miles from 1 of the deadly tornadoes. However, we found no significant change in the risk ratio (RR) estimates and 95% CIs. We used SAS version 9.3 (SAS Institute, Inc, Cary, NC) to calculate frequencies and percentages, stratified analyses for race and location of death, and Epi Info 7 (Centers for Disease Control and Prevention, Atlanta, GA) to calculate RRs and 95% CIs.
RESULTS
We identified 247 deaths related to the tornadoes that struck Alabama on April 27, 2011. Of the 237 individuals whose location of injury was known, 206 (86.9%) were within 1.5 miles of 1 of the 12 deadly tornado paths. Of the 12 deadly tornadoes, 10 were EF-4 or EF-5 tornadoes and were responsible for the majority of the deaths (n = 221; 89.5%). EF-4 and EF-5 tornadoes represented 17.7% (11 out of 62 tornadoes) of all tornadoes on April 27.
We calculated approximate RRs and found that females were 40% more likely to suffer a tornado-related death than males (RR = 1.39; 95% CI = 1.08, 1.80; Table 1). The mean age of decedents was 50.7 years, with the youngest decedent being aged 4 days and the oldest being aged 97 years. The 5 to 14 and 15 to 24 years age groups were at the lowest risk (RR5–14 = 0.51; 95% CI = 0.28, 0.91; RR15–24 = 0.49; 95% CI = 0.28, 0.84) when compared with the 45 to 54 years age group. The risk of death increased with age, reaching statistical significance at age 65 years or older; the 85 years and older age group was at the highest risk and nearly 4 times more likely to die (RR ≥ 85 = 3.85; CI = 2.05, 7.26). When examining race, we found that White individuals’ risk of death was nearly twice that of Black individuals (RR = 1.71; CI = 1.23, 2.40). Stratification by race revealed that Black decedents were younger, with a mean age of 41.2 years (95% CI = 33.5, 48.8), compared with a mean age of 52.8 years for White decedents (95% CI = 49.6, 56.1; P = .004). Of the 247 decedents, 29 (11.7%) were younger than 18 years. For those aged 18 years or older, 40 (18.3%) had more than a high school education, 156 (71.6%) were employed, and 39 (17.9%) were homemakers. At least 104 (42.1%) individuals had 1 or more physical or social vulnerabilities, which include being aged 65 years or older (n = 84, 34.0%), being aged younger than 5 years (n = 10, 4.0%), living alone (n = 10, 4.0%), having a disability (n = 10, 4.0%), and being pregnant (n = 1, 0.4%). Forty-two (75.0%) of the 56 households for which we had household income information had an annual household income of less than $35 000.
TABLE 1—
Demographics of Decedents in the Alabama Tornado Outbreak, April 27, 2011
| Variable | Decedents, No. (%) | Population, No. (%) | RR (95% CI) |
| Areaa | |||
| Rural | 173 (70.1) | 144 671 (48.2) | 2.90 (2.18, 3.87)b |
| Urban | 64 (25.9) | 155 471 (51.8) | 1.00 (Ref) |
| Unknown | 10 (4.0) | 0 (0.0) | … |
| Gender | |||
| Female | 146 (59.1) | 152 756 (50.9) | 1.39 (1.08, 1.80)b |
| Male | 101 (40.9) | 147 386 (49.1) | 1.00 (Ref) |
| Age, y | |||
| < 5 | 10 (4.0) | 17 841 (5.9) | 0.68 (0.34, 1.37) |
| 5–14 | 16 (6.5) | 38 244 (12.7) | 0.51 (0.28, 0.91) |
| 15–24 | 20 (8.1) | 49 561 (16.5) | 0.49 (0.28, 0.84) |
| 25–34 | 19 (7.7) | 36 753 (12.2) | 0.63 (0.36, 1.09) |
| 35–44 | 21 (8.5) | 38 870 (13.0) | 0.65 (0.38, 1.12) |
| 45–54 | 36 (14.6) | 43 536 (14.5) | 1.00 (Ref) |
| 55–64 | 41 (16.6) | 36 395 (12.1) | 1.36 (0.87, 2.13) |
| 65–74 | 43 (17.4) | 22 215 (7.4) | 2.34 (1.50, 3.64)b |
| 75–84 | 28 (11.3) | 12 648 (4.2) | 2.68 (1.63, 4.39)b |
| ≥ 85 | 13 (5.3) | 4079 (1.4) | 3.85 (2.05, 7.26)b |
| Race | |||
| White | 204 (82.6) | 213 093 (71.0) | 1.71 (1.23, 2.40)b |
| Black | 41 (16.6) | 73 424 (24.5) | 1.00 (Ref) |
| Other | 1 (0.4) | 13 625 (4.5) | … |
| Unknown | 1 (0.4) | 0 (0.0) | … |
| Ethnicity | |||
| Non-Hispanic | 244 (98.8) | 290 697 (96.9) | 2.64 (0.85, 8.25) |
| Hispanic | 3 (1.2) | 9445 (3.1) | 1.00 (Ref) |
| Highest education attained | |||
| Aged < 18 y | 29 (11.7) | ||
| < grade 12 | 56 (22.7) | ||
| Grade 12 or equivalent | 86 (34.8) | ||
| > grade 12 | 40 (16.2) | ||
| Unknown | 36 (14.6) | ||
| Occupationc | |||
| White collar | 83 (33.6) | ||
| Blue collar | 73 (29.6) | ||
| Homemaker | 39 (15.8) | ||
| Student | 10 (4.0) | ||
| None or never worked | 9 (3.6) | ||
| Aged < 18 y | 29 (11.7) | ||
| Unknown | 4 (1.6) | ||
| Physical or social vulnerabilities | |||
| Yes | 104 (42.1) | ||
| Pregnancy | 1 (0.4) | ||
| Lives alone | 10 (4.0) | ||
| Aged < 5 y | 10 (4.0) | ||
| Aged ≥ 65 y | 84 (34.0) | ||
| Disabled, not specified | 10 (4.0) | ||
| > 1 vulnerability | 11 (4.5) | ||
| ≥ 65 y and lives alone | 7 (2.8) | ||
| ≥ 65 y and disabled | 4 (1.6) | ||
| No or unknown | 143 (57.9) | ||
| Total | 247 (100.0) | 300 142 (100.0) | |
Note. CI = confidence interval; RR = risk ratio. The sample size was n = 247. Totals may not add to 100% because of rounding.
Classified according to 2010 Census definition.10
Indicates statistically significant data.
Coded according to 2000 census definition, and classified by categories developed by NIOSH.11
Mechanism and Cause of Death
The majority of the deaths were attributed directly to a tornado (n = 235; 95.1%), and 12 (4.9%) deaths were indirectly related to a tornado (Table 2). Of the direct deaths, 112 decedents (47.7%) were struck or cut by debris, 81 (34.5%) were thrown, and 59 (25.1%) had more than 1 mechanism of injury leading to their deaths. Of the 12 indirect deaths, 7 (58.3%) were related to a power outage. These 7 included 4 deaths caused by house fires, 2 deaths related to medical conditions (e.g., dependent on oxygen concentrator machine, lack of refrigeration for insulin), and 1 death was the result of a fall during the blackout. Causes of death were mostly trauma (n = 223; 90.3%) or trauma-related (n = 17; 6.9%; Table 2). At least 55 (22.3%) of the deceased sustained some form of head injury; for 39 (15.8%), head injury was listed as the official cause of death. For the majority of the deceased, the date of death or body recovery was the day of the tornado (n = 212; 85.8%). The last recorded death related to the tornadoes occurred on September 16, 2011, from trauma-associated pulmonary thromboembolism.
TABLE 2—
Circumstances of Death for Decedents in the Alabama Tornado Outbreak, April 27, 2011
| Death Circumstances | No. (%) |
| Mechanism of death | |
| Directa | 235 (95.1) |
| Struck or cut by debris or objects | 112 (47.7) |
| Thrown | 81 (34.5) |
| Crushed | 44 (18.7) |
| Trapped in rubble | 4 (1.7) |
| Unknown | 54 (23.0) |
| > 1 direct mechanism | 59 (25.1) |
| Indirect | 12 (4.9) |
| House fire after tornado | 4 (33.3) |
| Medical complication or injury owing to electricity failure | 3 (25.0) |
| Complications posttornado injury | 2 (16.7) |
| Myocardial infarction | 2 (16.7) |
| Premature birth | 1 (8.3) |
| Cause of death (death certificate) | |
| Trauma | 223 (90.3) |
| Unspecified | 90 (36.4) |
| Multiple, unspecified body regions | 84 (34.0) |
| Head or head and torso | 39 (15.8) |
| Chest | 10 (4.0) |
| Trauma associated | 17 (6.9) |
| Cardiopulmonary arrest | 6 (2.4) |
| Mechanical asphyxia | 5 (2.0) |
| Complication after trauma | 2 (0.8) |
| Internal hemorrhage | 1 (0.4) |
| Pulmonary hemorrhage | 1 (0.4) |
| Pulmonary thromboembolism | 1 (0.4) |
| Respiratory distress syndrome | 1 (0.4) |
| Other causes | 7 (2.8) |
| Smoke inhalation | 4 (1.6) |
| Acute myocardial infarction | 1 (0.4) |
| Alzheimer’s disease | 1 (0.4) |
| Diabetic ketoacidosis | 1 (0.4) |
| Date of death or body recovery | |
| April 27, 2011 (date of tornado) | 212 (85.8) |
| April 28, 2011 | 20 (8.1) |
| April 29, 2011 | 3 (1.2) |
| April 30, 2011 | 1 (0.4) |
| May 2011 | 7 (2.8) |
| June 2011 | 3 (1.2) |
| September 2011 | 1 (0.4) |
Note. The sample size was n = 247.
Each decedent may have > 1 direct mechanism of death.
Location During the Tornadoes
We examined the decedents’ location of injury during the tornadoes. Individuals who were in rural areas (n = 173; 70.1%) were nearly 3 times as likely to die as those in urban areas at the time of the tornadoes (n = 64; 25.9%; RR = 2.90; 95% CI = 2.18, 3.87; Table 1). The majority of the deceased were indoors during the tornado (n = 228; 92.3%), with 133 (53.8%) in single-family homes and 51 (20.6%) in mobile homes (Table 3). Of the deceased, 165 (66.8%) were in homes that were reported as being completely destroyed. One hundred eighty-six (75.3%) individuals were in their own home; 36 (14.6%) were in the home of a family member, friend, or neighbor; and 4 (1.6%) died at work (which for 3 was also their home). For the deceased who were indoors and whose location was known, the most common locations were the bathroom (n = 18; 7.3%), basement or underground shelter (n = 10; 4.0%), bedroom (n = 10; 4.0%), and hallway (n = 10; 4.0%; Table 3). Four (1.6%) decedents were outdoors at the time of the tornado, and 11 (4.5%) were in vehicles. The majority of decedents died on the scene (n = 214; 86.6%), and a minority survived the initial impact but died en route to a hospital (n = 3; 1.2%) or in a hospital (n = 27; 10.9%; Table 3).
TABLE 3—
Location of Decedents During the Alabama Tornado Outbreak, April 27, 2011
| Location | No. (%) |
| General location | |
| Indoor | 228 (92.3) |
| Single family home | 133 (53.8) |
| Mobile home | 51 (20.6) |
| Apartment | 15 (6.1) |
| Church | 1 (0.4) |
| Factory | 1 (0.4) |
| Hospice care | 1 (0.4) |
| Unknown | 26 (10.5) |
| Outdoor | 4 (1.6) |
| Vehicle | 11 (4.5) |
| Unknown | 4 (1.6) |
| Location within building | |
| Bathroom | 18 (7.3) |
| Basement/underground shelter | 10 (4.0) |
| Bedroom | 10 (4.0) |
| Hallway | 10 (4.0) |
| Living room | 4 (1.6) |
| Garage | 4 (1.6) |
| Kitchen | 3 (1.2) |
| Central room | 2 (0.8) |
| Office space | 1 (0.4) |
| Stairwell | 1 (0.4) |
| Not in building | 19 (7.7) |
| Unknown | 165 (66.8) |
| Location of death | |
| On scene | 214 (86.6) |
| In hospital | 27 (10.9) |
| En route on ambulance | 3 (1.2) |
| Other/unknown | 3 (1.2) |
| Company during tornado | |
| With other deceased and survivors | 65 (26.3) |
| With other deceased | 61 (24.7) |
| With other survivors | 52 (21.1) |
| Alone | 19 (7.7) |
| Unknown | 50 (20.2) |
Note. The sample size was n = 247.
Individuals who survived the initial impact but who later died in a hospital were younger, with a mean age of 37.9 years (95% CI = 27.4, 48.4) compared with a mean age of 52.1 years (95% CI = 48.9, 55.2) for those who died on scene or en route to a hospital (P = .004).
Warning and Behavior
At least 102 (41.3%) of the deceased received some form of warning of the approaching tornado (Table 4). This information was self-reported by survivors in semistructured interviews; thus, we are unable to determine for the others whether a warning was received. Furthermore, many died alone or with others who also died (n = 80; 32.4%); therefore, little information on the circumstances of death and behavior is known for these individuals. Of the known warnings received by decedents, 26 (25.5%) individuals were warned by another person and 15 (14.7%) heard a siren. The exact warning type is unknown for 55 (53.9%) of the individuals warned. For 14 (5.7%) decedents, there was no warning except seeing or hearing the approaching tornado.
TABLE 4—
Warning Received and Protective Measures Taken by Decedents Before Death in the Alabama Tornado Outbreak, April 27, 2011
| Warnings and Protective Measures | No. (%) |
| Warning received | |
| Yesa | 102 (41.3) |
| Warned by another person | 26 (25.5) |
| Heard siren | 15 (14.7) |
| Warned through television | 6 (5.9) |
| Warned by earlier tornado | 3 (2.9) |
| Warned through radio | 1 (1.0) |
| Unspecified | 55 (53.9) |
| > 1 warning | 7 (6.9) |
| No | 15 (6.1) |
| Unknown | 130 (52.6) |
| Protective behaviors | |
| Yes, after hearing formal warning and seeing storm coming | 68 (27.5) |
| Yes, after hearing formal warning | 59 (23.9) |
| > 1 protective measure employed | 26 (38.2) |
| Hid out or sought shelter: bathroombc | 16 (23.5) |
| Hid out or sought shelter: bathtubb | 9 (13.2) |
| Hid out or sought shelter:basementbc | 10 (14.7) |
| Hid out or sought shelter: hallwaybc | 10 (14.7) |
| Hid out or sought shelter: closetbc | 6 (8.8) |
| Hid out or sought shelter: other’s house | 3 (4.4) |
| Hid out or sought shelter: other’s garage | 3 (4.4) |
| Hid out or sought shelter: other’s trailer | 3 (4.4) |
| Hid out or sought shelter: central roombc | 2 (2.9) |
| Hid out or sought shelter: kitchen pantry | 2 (2.9) |
| Hid out or sought shelter: next to wall | 1 (1.5) |
| Escaped to seek shelter: via car | 5 (7.4) |
| Escaped to seek shelter: unspecified | 2 (2.9) |
| Covered with: total | 10 (14.7) |
| Covered with: pillowsb | 5 (7.4) |
| Covered with: blanketsbc | 3 (4.4) |
| Covered with: mattressbc | 1 (1.5) |
| Covered with: person | 1 (1.5) |
| Held others | 22 (32.4) |
| Hid under/behind: total | 4 (5.9) |
| Hid under/behind: overturned couch | 2 (2.9) |
| Hid under/behind: tablebc | 1 (1.5) |
| Hid under/behind: large building (in vehicle) | 1 (1.5) |
| No, heard warning | 9 (3.6) |
| No, did not hear or unknown whether heard warning | 7 (2.8) |
| Unknown | 163 (66.0) |
At least 68 (27.5%) individuals took some action they perceived as protective. Of these, 59 (86.8%) took action after hearing a warning (Table 4). Forty-four individuals (64.7%) took shelter in a location recommended by the Centers for Disease Control and Prevention or the National Oceanic and Atmospheric Administration. Ten (14.7%) people covered themselves with items such as pillows and blankets. Twenty-six (38.2%) individuals took more than 1 measure they perceived as protective. At least 11 individuals knew they were in the path of a tornado but took no action; all 11 were indoors in their own home. Twenty-six individuals (10.5%) changed location after they received a warning or had seen the storm coming. Of these, 6 people went to mobile homes (23.1%), which are known to be unsafe structures during tornadoes.
DISCUSSION
The April 27, 2011, tornado outbreak was one of the deadliest tornado disasters recorded in US history. The number and severity of the tornadoes resulted in a record high number of tornado-related deaths in Alabama. Although the strength of a tornado is a well-documented risk factor for injury and death, human factors also contribute to the public health outcome of such storms.14,15 Known risk factors include being in a mobile home, a vehicle, or outdoors; being older than 60 years; not seeking shelter; and being unfamiliar with warning terminology.14–21 In this event, we found that the majority of the decedents were indoors and in single-family homes, with only a minority being in mobile homes during the tornadoes. This finding contrasts with previous events and the national average for tornado-related fatalities seen from 2001 to 2005, where 57% of the fatalities occurred in mobile homes and 26% occurred in single-family homes.1 Because we do not know the total number of mobile homes and single-family homes in the exposed areas, we were unable to calculate risk ratios. However, this event was significantly more severe than previous tornado outbreaks; therefore, single-family homes that might sustain only minor damage in weaker tornadoes were severely affected.
The majority of decedents died on scene, consistent with previous tornado events,15,17,20,22,23 suggesting that considerable investment in preparedness is critical in preventing tornado-related deaths. Many people, when warned, took some form of protective action. However, many died in locations considered safe according to the current Centers for Disease Control and Prevention and National Oceanic and Atmospheric Administration recommendations,12,13 including basements and hallways, which have historically been considered the first and second safest sheltering locations.15,18,20 However, given that we do not have data on how many people survived by sheltering in basements and hallways, we can only conclude that these locations were not safe for the individuals who died, not that these locations are generally unsafe. Instead, this finding suggests that in the most violent EF-4 and EF-5 tornadoes, even locations historically considered safe may not guarantee survival. In a review of this event to assess the next steps to make this community more tornado resilient, the Tornado Recovery Action Council has recommended that the state explore the implementation of a standardized statewide building code and enforcement policies.6 Although this would likely improve building structures, the costs of implementing these codes and policies may create tremendous economic burden on a community recovering from a disaster; therefore, the cost-effectiveness and practicality of this implementation should be assessed with caution and perhaps in comparison with the cost-effectiveness of installing tornado-safe rooms and underground shelters.
Recommendations
Our findings support recommendations after previous tornado events to provide more local community shelters and to inform the public of the location of these community shelters.15,16,17,19,20,22 In addition, we believe that promoting word-of-mouth warnings through campaigns such as “hear a warning, tell a friend” may help disseminate warnings and motivate family and friends to take action. However, it is important to also emphasize that individuals in the path of the tornado seek shelter before warning others. Finally, our findings reiterate the importance of having a personal or family preparedness plan, as recommended by the Centers for Disease Control and Prevention and the Federal Emergency Management Agency for all disasters.24,25
On the basis of our findings, we believe the personal or family preparedness plan should emphasize access to warnings, access to safe shelters, assistance for older adults and vulnerable populations in seeking shelter, and preparedness for power outages. We found that nearly half of the deceased received 1 or more forms of warning, with the main form being a warning by another person, followed by sirens. We recommend increasing access to warnings through peer warnings, which may help warn people in rural areas in which no sirens may be nearby. In this event, we found that the majority of the decedents died on scene, were cut or struck by debris, or were thrown. Part of the personal or family preparedness plan should incorporate identification of safe shelters, preferably underground shelters, so everyone knows exactly where to go after receiving warning. Several people died in vehicles or outdoors, locations that are unsafe during a tornado,15,18,20,21 and others stayed in mobile homes and refused to take action. Complacency and refusal to take action may stem from a lack of trust in the warning system because of previous false alarms, as suggested by the Alabama Tornado Recovery Action Council, and may be partially overcome by promoting peer warnings.6 These observations once again emphasize the importance of identifying safe shelter ahead of time and taking every warning seriously.24,26
Consistent with existing literature, we found that older adults were at increased risk for tornado-related deaths.14–16,18,19,21 Preexisting health issues, reduced mobility, increased vulnerability to trauma, and increased likelihood of living in older houses associated with older age can all contribute to the higher likelihood of tornado-related fatality.15,18,27 To protect this high-risk group, as suggested by previous studies, current recommendations can focus attention on the older adult population and other vulnerable groups and encourage making plans for family and friends to take extra precaution to warn, check on, and assist these vulnerable groups in seeking shelter or in considering evacuation out of the area if given sufficient warning.14,16,18 Our data showed that deaths indirectly related to the tornadoes were most frequently related to power outages. Raising public awareness of injuries and deaths that can occur during the resulting power outage may prevent these indirect deaths. Officials can encourage individuals to prepare for power outages by having a preparedness kit that contains battery-powered flashlights and a radio, as well as adequate food and water for several days. In addition, individuals dependent on electric-powered medical devices or refrigerated medications should be encouraged to have contingency plans ready.
Conclusions
We used multiple data sources to understand the circumstances surrounding deaths from the tornadoes that struck Alabama on April 27, 2011. In particular, the novel use of Red Cross data for public health research shows the importance of continued collaboration of different agencies and organizations to share information for public health benefits. The population described here may not be a representative sample of the US population; therefore, generalization of the findings to the rest of the United States should be exercised with caution. On the basis of our findings, we recommend a concerted effort at the government, community, and individual levels. We believe that providing community shelters for those without safe rooms or basements, promoting peer warnings, and encouraging a comprehensive personal or family preparedness plan are of utmost importance to preventing tornado-related deaths in future events. Many of these recommendations support recommendations after past events and parallel those of the National Weather Service and the Tornado Recovery Action Council in Alabama. We have also seen encouraging progress toward implementing some of these recommendations. In DeKalb County in Alabama, within 6 months of the tornadoes, Federal Emergency Management Agency funds were allocated to build 7 new community storm shelters, and more than 600 individual Federal Emergency Management Agency applications were filed to build reinforced safe rooms.6 A combination of all such efforts will better prepare the community for future tornadoes and reduce tornado-related death.
Acknowledgments
The material in this article represents original work that has not been previously published in whole or in part in a peer-reviewed journal. A portion of the data were published as part of a general overview of the 4-day tornado outbreak across 5 Southeastern states in the July 20, 2012, issue of the Centers for Disease Control and Prevention’s Morbidity and Mortality Weekly Report (MMWR). The MMWR is not peer reviewed and is the Centers for Disease Control and Prevention’s primary vehicle to issue timely public health recommendations.
We thank the Integrative Care Team volunteers, who collected the American Red Cross data; Catherine Molchan, Melissa Morrison, Judy Moulder, and Joseph Roth, who provided death certificate information; Greg Carbin, John DeBlock, and Gary Goggins for providing the National Weather Service data; and Cynthia Robinson for coding the occupational data.
Human Participant Protection
All human participants in this study were deceased; therefore, this work was exempt from review by the institutional review board.
References
- 1.Ashley WS. Spatial and temporal analysis of tornado fatalities in the United States: 1880–2005. Weather Forecasting. 2007;22(6):1214–1228. [Google Scholar]
- 2.National Weather Service. Thunderstorms, tornadoes, lightning: Nature’s most violent storms—a preparedness guide. Available at: http://www.nws.noaa.gov/om/severeweather/resources/ttl6-10.pdf. Accessed February 23, 2012.
- 3.Enhanced F Scale for tornado damage. NOAA Storm Prediction Center. Available at: http://www.spc.noaa.gov/efscale/ef-scale.html. Accessed February 23, 2012.
- 4.Centers for Disease Control and Prevention. Tornado-related fatalities—five states, Southeastern United States, April 25-28, 2011. MMWR Morb Mortal Wkly Rep. 2012;61(28):529–533. [PubMed] [Google Scholar]
- 5.National Weather Service. Service assessment—the historic tornadoes of April 2011. Available at: http://www.nws.noaa.gov/os/assessments/pdfs/historic_tornadoes.pdf. Accessed January 30, 2012. [Google Scholar]
- 6.Tornado Recovery Action Council of Alabama. Cultivating a state of readiness—our response to April 27, 2011. Available at: http://tracalabama.org/wp-content/uploads/2012/01/TRAC_Report.pdf. Accessed January 30, 2012. [Google Scholar]
- 7.National Weather Service—GPRA performance targets for FY 2007 president’s budget. National Oceanic and Atmospheric Administration. Available at: http://www.nws.noaa.gov/cfo/program_planning/doc/final%20FY07%20PB%20GPRA%20Table%20-%20Hurricane%20actual%204-4-06.pdf. Accessed June 21, 2012.
- 8.Combs DL, Quenemoen LE, Parrish RG, Davis JH. Assessing disaster-attributed mortality: development and application of a definition and classification matrix. Int J Epidemiol. 1999;28(6):1124–1129. doi: 10.1093/ije/28.6.1124. [DOI] [PubMed] [Google Scholar]
- 9.International Classification of Diseases, 10th Revision. Geneva, Switzerland: World Health Organization; 1990. [Google Scholar]
- 10.Urban area criteria for the 2010 Census. Fed Regist. 2011;76(164):53029–53043. [Google Scholar]
- 11.US Census Bureau. Census of Population and Housing 2000: Alphabetical Index of Industries and Occupations and Classified Index of Industries and Occupations. Washington, DC: US Department of Commerce, Economics and Statistics Administration; 2000. [Google Scholar]
- 12.Emergency preparedness and response—during a tornado. Centers for Disease Control and Prevention. Available at: http://www.bt.cdc.gov/disasters/tornadoes/during.asp. Updated December 23, 2003. Accessed December 13, 2011.
- 13.Tornado safety. NOAA Storm Prediction Center. Available at: http://spc.noaa.gov/faq/tornado/safety.html. Accessed March 22, 2012.
- 14.Lillibridge SR. Tornadoes. In: Noji EK, editor. The Public Health Consequences of Disasters. New York, NY: Oxford University Press; 1997. pp. 228–244. [Google Scholar]
- 15.Carter AO, Millson ME, Allen DE. Epidemiologic study of deaths and injuries due to tornadoes. Am J Epidemiol. 1989;130(6):1209–1218. doi: 10.1093/oxfordjournals.aje.a115449. [DOI] [PubMed] [Google Scholar]
- 16.Centers for Disease Control and Prevention. Tornado disaster–Kansas, 1991. MMWR Morb Mortal Wkly Rep. 1992;41(10):181–183. [PubMed] [Google Scholar]
- 17.Centers for Disease Control and Prevention. Tornado-associated fatalities–Arkansas, 1997. MMWR Morb Mortal Wkly Rep. 1997;46(19):412–416. [PubMed] [Google Scholar]
- 18.Daley WR, Brown S, Archer P et al. Risk of tornado-related death and injury in Oklahoma, May 3, 1999. Am J Epidemiol. 2005;161(12):1144–1150. doi: 10.1093/aje/kwi142. [DOI] [PubMed] [Google Scholar]
- 19.Eidson M, Lybarger JA, Parsons JE, MacCormack JN, Freeman JI. Risk factors for tornado injuries. Int J Epidemiol. 1990;19(4):1051–1056. doi: 10.1093/ije/19.4.1051. [DOI] [PubMed] [Google Scholar]
- 20.Glass RI, Craven RB, Bregman DJ et al. Injuries from the Wichita Falls tornado: implications for prevention. Science. 1980;207(4432):734–738. doi: 10.1126/science.207.4432.734. [DOI] [PubMed] [Google Scholar]
- 21.Sugimoto JD, Labrique AB, Ahmad S et al. Epidemiology of tornado destruction in rural northern Bangladesh: risk factors for death and injury. Disasters. 2011;35(2):329–345. doi: 10.1111/j.1467-7717.2010.01214.x. [DOI] [PubMed] [Google Scholar]
- 22.Centers for Disease Control and Prevention. Tornado disaster—Pennsylvania. MMWR Morb Mortal Wkly Rep. 1986;35(14):233–235. [PubMed] [Google Scholar]
- 23.Centers for Disease Control and Prevention. Tornado disaster—Illinois, 1990. MMWR Morb Mortal Wkly Rep. 1991;40(2):33–36. [PubMed] [Google Scholar]
- 24.Emergency preparedness and response—tornadoes: being prepared. Centers for Disease Control and Prevention. Available at: http://www.bt.cdc.gov/disasters/tornadoes/prepared.asp. Updated June 22, 2012. Accessed June 25, 2012.
- 25.Make a plan. Federal Emergency Management Agency. Available at: http://www.ready.gov/make-a-plan. Accessed June 25, 2012.
- 26.Centers for Disease Control and Prevention. Tornado disaster—Texas. MMWR Morb Mortal Wkly Rep. 1988;37(30):454–456, 461. [PubMed] [Google Scholar]
- 27.Mandelbaum I, Nahrwold D, Boyer DW. Management of tornado casualties. J Trauma. 1966;6(3):353–361. doi: 10.1097/00005373-196605000-00005. [DOI] [PubMed] [Google Scholar]