Tropical Storm Allison rapid needs assessment — Houston, Texas, June 2001

Morbidity and Mortality Weekly Report, May 3, 2002

On June 5, 2001, Tropical Storm Allison made landfall on Galveston Island, Texas. During the next 2 days, the system soaked much of southeast Texas and south-central Louisiana with more than 10 inches of rain as it moved slowly northward. On June 7, the storm made a clockwise loop back to the southwest, bringing even more rain to already drenched areas. The record rainfall caused billions of dollars in flood-related damage and approximately 25 deaths and led to a presidential disaster declaration covering 31 Texas counties (Figure 1) and 28 Louisiana parishes. Harris County; Texas (2000 population: 3,400,578), center of the Houston metropolitan area, was among the hardest hit with some areas receiving up to 37 inches of rain in 24 hours (1) (Figure 2). To evaluate the community's immediate public health needs, the City of Houston Department of Health and Human Services (HDHHS) conducted a rapid needs assessment in the areas most affected by flooding. This report summarizes assessment results, which identifie d increased illness in persons living in flooded homes, suggesting a need for rapid resolution of flood-related damage and the possibility that residents should seek temporary housing during clean-up and repair. The findings underscore the usefulness of rapid needs assessment as a tool to minimize misinformation, identify actual health threats, and ensure delivery of resources to those with the greatest and most immediate need.

Severely flood-affected areas of Houston, identified by the City of Houston Office of Emergency Management, were divided into two sectors: Area A, in the northeast section of the city, comprised 96 census tracts, 148,654 housing units, and 443,356 residents; and Area B, covering the western part of the city, comprised 72 census tracts, 167,158 housing units, and 400,868 residents. Using modified cluster sampling (2,3), HDHHS obtained a random sample of 30 census tracts from each of the two areas by using statistical software (SPSS V.10). One neighborhood from each of these 60 tracts was selected by delineating block groups using 1997 household estimates (housing unit data from the 2000 census were nor yet available), breaking into quintiles the distribution of household population, and selecting the block group with the highest population and concentration of streets. A standardized question-naire was developed to collect information about demographics, extent of home damage, number of residents forced to le ave their homes, reported illnesses and injuries among household members, disruption of public services, and current needs. The goal was to obtain completed questionnaires from seven households in each of the 60 selected census tracts for a total sample of 420 households. Data were analyzed by using SAS for Windows (release 8.01). Data were stratified by census tract for calculation of illness and injury odds ratios for living in a flooded versus nonflooded home.

HDHHS conducted the survey door to door on June 16, 2001, 1 week after the heaviest rainfall caused the worst flooding, and met its goal of 420 completed questionnaires (Table 1). The overall response rate was 59.3%; nonresponses included 257 instances of no one at home, 20 refusals, seven uninhabited households, and four households in which a language barrier prevented communication. Of the 420 households surveyed, 389 (92.6%) were single-family homes, and mean household size was 3.3 persons (range: 0-20 persons). A total of 137 (32.6%) surveyed households had floodwaters in the home; mean floodwater depth was 16 inches (median: 12 inches, range: 1-60 inches), and mean duration floodwaters remained in the home was 36.8 hours (median: 24 hours, range: 1-168 hours). A total of 149 (35.5%) surveyed households reported damage to the home; all but one of these were either habitable (116 [77.9%]) or repairable (32 [21.5%]). Survey participants in 57 (13.6%) households reported spending at least 1 night away from h ome. For some period during or after the flooding, 138 (32.9%) households reported interruption in telephone service, 63 (15.0%) had no sewage service, 61 (14.5%) lost electricity, 44 (10.5%) lost natural gas supply, and 23 (5.5%) had no running water. At the time of the survey, some households were still without telephone service (38 [9.0%]), natural gas (29 [6.9%]), sewage service (21 [5.0%]), electricity (eight [1.9%]), or running water (eight [1.9%]). The most commonly identified needs were mosquito control, pharmacy access, and new household furnishings. Other needs included medical access, shelter, food, home repair, clothing, transportation, drainage assistance, financial assistance, and heavy trash pick-up.

Fifty-four (12.9%) surveyed households reported at least one person with illness that occurred after the onset of flooding. Persons living in flooded homes were significantly more likely than those living in nonflooded homes to report illness; the only specific illness significantly associated with residing in a flooded home was diarrhea/stomach conditions (Table 2). A total of 17 (4.0%) surveyed households reported at least one person injured after the onset of flooding. No significant association existed between likelihood of reporting an injury and living in a flooded home compared with a nonflooded home, nor were any specific injuries so associated.