Research monitoring unmet needs 2005 hurricanes

Monitoring Unmet Needs Using 2-1-1 During Natural Disasters Sherry I. Bame, PhD, Kay Parker, LCDC, CIRS, Jee Young Lee, MS, Alexandria Norman, MUP, Dayna Finley, MArch, Atmaja Desai, MUP, Abha Grover, MS, Courtney Payne, MUP, Andrew Garza, MHA, Ashley Shaw, BS, Robyn Bell-Shaw, BEd, Tasha Davis, MLA, Erin Harrison, MPSA, Rhonda Dunn, PhD, Pratik Mhatre, PhD, Frank Shaw, BA, Chester Robinson, PhD Background: Hurricanes Katrina and Rita struck the Gulf Coast forcing unprecedented mass evacuation and devastation. Texas 2-1-1 is a disaster communication hub between callers with unmet needs and community services at disaster sites and evacuation destinations.

Purpose: To describe the location and timing of unmet disaster needs collected in real-time through Katrina–Rita disaster phases.

Methods: In 2008 –2010, a total of 25 data sets of Texas 2-1-1 calls from August–December 2005 were recoded and merged. In 2011–2012, analysis was performed of unmet need types, with comparisons over time and location; mapping was adjusted by population size.

Results: Of 635,983 total 2-1-1 calls during the study period, 65% included primary disaster unmet needs: housing/shelter (28%); health/safety (18%); food/water (15%); transportation/fuel (4%). Caller demand spiked on Mondays, decreasing to a precipitous drop on weekends and holidays. Unmet needs surged during evacuation and immediate disaster response, remaining at higher threshold through recovery. Unmet need volume was concentrated in metropolitan areas. After adjusting for population size, “hot-spots” showed in smaller evacuation destinations and along evacuation routes.

Conclusions: New disaster management strategies and policies are needed for evacuation destinations to support extended evacuation and temporary or permanent relocation. Planning and monitoring disaster resources for unmet needs over time and location could be targeted effectively using real-time 2-1-1 call patterns. Smaller evacuation communities were more vulnerable, exhausting their limited resources more quickly. Emergency managers should devise systems to more quickly authorize vouchers and reimbursements. As 2-1-1s expand and coordinate disaster roles nationwide, opportunities exist for analysis of unmet disaster needs to improve disaster management and enhance community resiliency.

(Am J Prev Med 2012;43(6S5):S435–S442) © 2012 American Journal of Preventive Medicine

n 2005, Hurricanes Katrina and Rita struck the Gulf Coast approximately 3 weeks and 250 miles apart, forcing the largest mass evacuation and causing the most-widespread devastation experienced in the U.S. since the 1930s Dust Bowl.1,2 The U.S. was stunned by the devastation of Hurricane Katrina, one of the costliest

storms in American history. Within 1 week of landfall on August 29, 2005, Texas provided shelter to over 250,000 Katrina evacuees who sought refuge in the state.3 Hurricane Rita made landfall on September 24 and laid waste to the northeast Texas coast. Many in the Houston area evacuated prior to Rita turning northward for landfall at Beaumont/Pt. Arthur near the Louisiana border.

From the Department of Landscape Architecture and Urban Planning(Bame, Lee, Norman, Desai, Davis, Payne), Department of Architecture (Finley), Department of Nutrition and Food Science (Grover), Department of Rural Public Health (Garza), Department of Construction Science (A. Shaw), Department of Education (Bell-Shaw), Department of Government and Public Service (Harrison), and Department of Computer Science (Dunn), the Public Policy and Research Institute (Mhatre), and Department of Political Science (F. Shaw), Texas A&M University; 2-1-1 Brazos

Valley and United Way Brazos Valley (Parker), College Station; and Health Policy and Economics Consulting, Inc. (Robinson), Bryan, Texas Address correspondence to: Sherry I. Bame, PhD, Professor, Health & Community Services, Urban Planning Program, Mail Stop 3137, Texas A&M University, College Station TX 77843-3137. E-mail: sbame@tamu. edu.

0749-3797/$36.00 http://dx.doi.org/10.1016/j.amepre.2012.09.002

Background

I

© 2012 American Journal of Preventive Medicine • Published by Elsevier Inc.

Am J Prev Med 2012;43(6S5):S435–S442 S435

S436

Bame et al / Am J Prev Med 2012;43(6S5):S435–S442

In 2004, Texas completed its integrated network of 25 autonomous regional 2-1-1 call centers to provide statewide “24/7/365” coverage. The 2-1-1 call centers throughout the U.S. serve their communities for local, regional, and national disasters, ranging from natural and man-made environmental disasters to coordination of H1N1 information and vaccine availability. The personal 2-1-1 telephone contact provides not only triage and appropriate referral for health and human service needs but also reassurance and sustained support for callers through crisis and recovery.4 Any caller’s request for 2-1-1 help is considered an unmet need. If a caller were able to access services to meet the need(s), she/he would contact the appropriate agency directly. Instead, she/he calls this three-digit number for contact information or referral to available and affordable community and disaster resources.

When Katrina evacuees began arriving in Texas, the State’s Offıce of Emergency Management was able to rely on 2-1-1 Texas Network as a communication hub for non-urgent disaster response for hurricane victims and their hosts, and to support communication among social support services, fırst-responders, and volunteers. Texas’s 2-1-1 Network received a remarkable surge in calls throughout Rita evacuation and recovery phases, with unmet needs compounded by both hurricanes. More than 2 million evacuated as a result of Hurricane Rita, many evacuating for a second time during the 2005 hurricane season.5 Communities throughout Texas hosted evacuees, often depleting local health and social support resources.

Considerable amounts of data exist regarding disaster needs, but little is known about disaster victims who encounter access barriers to available, affordable services and resources. Survey researchers face respondents’ recall biases in attempts to assess disaster victims’ experiences and perceptions. Collected in “real time” during disasters, 2-1-1 caller data present an opportunity to identify unmet needs that avoid respondents’ recall errors. This unprecedented analysis of disaster 2-1-1 call data over time and location enables investigation of total unmet needs during all disaster phases of Katrina–Rita throughout Texas’s 254 counties. Spatial analyses portray volume of unmet needs across disaster sites and evacuation destinations as well as evacuation routes. Also compared are frequencies of common primary disaster unmet needs (i.e., housing/shelter, health/safety, food/water, and transportation/fuel). This empirical study presents a unique opportunity for disaster-related programs and policymakers to investigate location and timing of “hot-spots” of unmet needs encountered before and during a disaster.

Methods The study population consisted of all recorded calls to the Texas 2-1-1 Network of 25 regional call centers from August 1 through December 31, 2005. Following the hurricane season, these call centers submitted copies of their call data to 2-1-1 Texas Network headquarters. This included paper call logs, scattered notes, CDs of electronic data fıles, and summary reports of data lost in the study period. When the Department of Homeland Security funded this study in 2008, all caller identifıers were removed and the data released to the researchers. The multiple data sources and fıle formats were recoded into a consistent format to merge into a single database. The research team validated 100% of this recoding and reformatting to ensure consistency and minimize errors.

Four variables were analyzed in 2011–2012: (1) volume of calls for unmet needs; (2) county where the caller sought help; (3) call date; and (4) “at-risk ratio,” total unmet needs per 100 households using 2005 U.S. Census estimates. Unmet needs were coded verbatim and then empirically aggregated into categories. Need categories were not mutually exclusive, with multiple need types possible per caller. Four primary disaster need categories were tallied to compare unmet need proportions and domains encountering access barriers: housing/shelter, health/safety, food/water, transportation/fuel. Time and spatial analyses used the “call” as unit of analysis for any unmet need. The 5-month study period included (1) 4-week baseline of community needs prior to Katrina landfall on August 29; (2) almost 4-week disaster management for Katrina evacuees in Texas; (3) a 3-day evacuation from Rita and disaster response following September 24 landfall; and (4) 3-month recovery post-Rita.

Two types of spatial analysis were conducted. First, total calls during the study period for each of the 254 Texas counties were mapped.

Strata of call volume were empirically determined by observed breaks in the distribution. Second, volume of calls per county was adjusted by the number of households in that county to get a ratio of calls per 100 households. A ratio of 1.051 call per 100 households (i.e., 1% call rate).

County ratios from 0.5 to 5.0 were considered “normal” as modal number of counties and expected range of call rates (shown in light blue, green, and yellow in Figure 1).

“Cold spots” were counties with 50% or fewer calls than would be expected given the number of households (i.e., ratio5 ,0.5 or ,1 call per 200 or more households: shown in dark blue). “Warm spots” were ratios of more than fıve times greater caller demand for unmet needs than expected (.5 calls per 100 households: shown in orange). “Hot spots” were more than ten times greater than expected demand for help, controlling for population size (shown in red and dark red). The empirical strata of a county’s “at risk” level were based on observed gaps in the distribution of call ratios per county. It should be noted that this was not a statistical risk analysis as caller demographic data were not available for appropriate adjustments by population characteristics.

Results Frequency of Unmet Needs and Primary Disaster Needs A total of 635,983 calls were recorded by the Texas 2-1-1 Network during the study period, with each call considered as any unmet need due to access barriers of being www.ajpmonline.org

Bame et al / Am J Prev Med 2012;43(6S5):S435–S442

S437

Third, 15% of callers needed access to food and/or water. Unmet 7000 food needs included ongoing requests for food 6000 stamps and vouchers as well as emergency re5000 quests for food and water during evacuation and 4000 sheltering. Coded here as well as “health” were re3000 quests for special diets, eating disorders, and 2000 safely storing or sanitizing food and water. Fourth, 1000 only 4% of calls were for unmet transportation or 0 fuel needs. This percentAugust 1 September 1 October 1 November 1 December 1 age was surprisingly low Date in 2005 given the traffıc and transport diffıculties in mass Figure 1. Total 2-1-1 calls per day in Texas, Fall 2005 (8/1/2005–12/31/2005) evacuation from these back-to-back storms, perunaware of appropriate resource(s) or contact information, haps reflecting adequate management by law enforcement, resource(s) not available or not affordable to adequately disaster management and local community resources. meet a caller’s need(s). Four types of primary disaster needs were identifıed in 65% of the calls (Table 1). First, 28% of Longitudinal Patterns of Total Unmet Needs callers had housing and/or shelter unmet needs, not only Variation of daily call volume followed a distinct fınding and managing shelter arrangements but also issues weekly pattern (Figure 1). Calls spiked on Mondays such as helping with utilities and housing rehabilitation. then decreased throughout the week to a precipitous Second, 18% of callers had health and/or safety unmet needs drop on weekends and holidays. This was a consistent including fınding providers in unfamiliar evacuation destipattern except for holidays and the immediate disaster nations; replacing disability equipment and modifıcations; period. Even though 2-1-1 operated “24/7/365,” this as well as help needed to access and fınance medical, dental, high Monday demand may reflect pent-up weekend vision, or mental health care, medications, and therapies. A and holiday needs for help when community health new concern of disaster health management evolved around and social services are typically unavailable. mass evacuation of nursing homes, hospice patients, and At Katrina’s landfall, the number of 2-1-1 Texas calls was homebound ill or disabled. normal baseline as most Katrina victims were not evacuated to Texas until 3 days later when call volume more than Table 1. Primary disaster unmet needs per total 2-1-1 doubled. In the second week post-landfall, calls spiked to calls in Texas, Fall 2005 more than 7000 per day then ranged from 4000 to 6000 calls/weekday until Hurricane Rita’s arrival. As fırstNo. of calls per % of need type per responders withdrew after the immediate aftermath, calls total 2-1-1 calls type of unmet Primary disaster connecting evacuees with disaster resources soared. This (N5635,983) need unmet needs burden on evacuation destinations again surged as Rita Housing/shelter 180,601 28.4 made landfall and affected populations, including Katrina Health/safety 113,862 17.9 victims, had to evacuate. In the third week post-Rita, calls dropped to a new threshold of about 1000 –1500 calls/day Food/water 96,159 15.1 higher than baseline that persisted for the remainder of the Transportation/fuel 25,216 4.0 year except for holidays and weekends. In retrospect, this 415,838 65.4 Suma demand pattern was not surprising and reflected reported a experiences of disaster and evacuation communities Need types are not mutually exclusive; thus, they add to more than 100% of total 2-1-1 calls. elsewhere.

Number of calls per day

8000

December 2012

Hurricane Katrina Hurricane Rita August 29 September 24

Bame et al / Am J Prev Med 2012;43(6S5):S435–S442

S438

spatial analyses. Mapping total calls by Texas county highlights variation in demand for help for Hurricane Rita disaster sites in east Texas as well as evacuation destinations throughout the state (Figure 2 and Table 2). The volume of calls per county was strongly correlated with its population size. In Texas, 70% of calls came from the state’s fıve large metropolitan areas. Houston and Dallas alone ac2-1-1 calls counted for 42% of all 0 Texas calls and counties 1 – <10 with Ft. Worth, Austin, 10 – <100 and San Antonio ac100 – <1000 counted for 28% of all 1000 – <5000 calls.

5000 – <10,500 Thirty-nine urban and 48,000 – <70,000 suburban counties throu100,000 – 140,000 ghout the state received 25% of Texas calls. Twelve Figure 2. Geographic distribution of total 2-1-1 calls by county for unmet needs in Texas, Fall urban counties with pop2005 (8/1/2005–12/31/2005) ulations greater than 150,000 accounted for Spatial Patterns of Total Unmet Needs 15% of all calls, with each receiving between 5000 and 10,500 Spatial analysis of call volume reflected considerable differcalls. Twenty-seven smaller urban and suburban counties ences in demand across Texas counties and strong associahad between 1000 and 5000 calls each, summing to 10% of tions with population size. Only 1% of total calls (n56627) total calls. The remaining 210 rural counties accounted for were from outside Texas, representing 45 other states plus only 5% of total calls. South and west Texas’s 120 rural Puerto Rico and Mexico; half were from Florida (re: Hurricounties totaled less than 1% of all calls during the study cane Wilma) and 30% from Louisiana. This small proporperiod. tion of calls from outside Texas was not included in the Table 2. Frequency strata of total 2-1-1 calls per county in Texas, Fall 2005 Minimum no. of calls/ county

Maximum no. of calls/ county

Number of counties

Percentage of counties

Number of calls

Percentage of calls

100,000

,140,000

2

0.79

244,845

41.73

48,000

,70,000

3

1.18

165,236

28.16

5,000

,10,500

12

4.72

85,901

14.64

1,000

,5,000

27

10.63

59,114

10.07

100

,1000

87

34.25

28,068

4.78

10

,100

79

31.10

3,386

0.53

1

,10

41

16.14

197

0.03

0

0

3

1.18

0

Sum

254

100

586,747

0 100.00

www.ajpmonline.org

Bame et al / Am J Prev Med 2012;43(6S5):S435–S442

S439

map of counties at risk for vulnerable population experiencing unmet needs appears different than mapping the total volume of unmet needs (Figure 3 and Table 3).

Eight strata of adjusted “at-risk” ratios were empirically determined.

Ten counties were highrisk “hot spots” with more than ten times the call demand for help than would be expected given 2-1-1 calls per their population size, ac100 households counting for 59% of 0 Texas 2-1-1 calls (shown >0–0.49 in red and dark red).

0.50–0.99 These “hot spots” were 1.0–2.49 east Texas disaster sites 2.5–4.9 5.0–9.9 and the major evacuation 10.0–13.9 destinations distributed 14.0–18.2 throughout the state, including west Texas cities Figure 3. Adjusted distribution of 2-1-1 calls per population size per county for unmet receiving busloads of needs in Texas, Fall 2005 (8/1/2005–12/31/2005) Katrina victims. Even controlling for populaSpatial Patterns Adjusted by Population Size: tion size, Houston, Dallas, and Austin metropolitan areas Analysis of “At-Risk” Locations were “hot spots” overwhelmed with unmet needs and Less-populous counties typically with fewer available access barriers. community resources may endure a greater burden in Twenty-four counties were “warm spots” at moderate supporting disaster needs than do more-urban counties. risk for supporting unmet community and disaster needs Adjustment for population size was calculated as a ratio at fıve to almost ten times greater than expected call of calls per 100 households. Controlling for urban bias, a volume (shown in orange, Figure 3). Accounting for 27% of calls, these counties were along evacuation routes and adjacent to Table 3. Ratio of total 2-1-1 calls adjusted by population size per county in Texas, Fall 2005 Hurricane Rita’s inland Minimum ratio Maximum ratio path. The sum of these 34 per 100 per 100 Number of Percentage of Number of Percentage of higher-risk counties achouseholds households counties counties calls calls counted for 86% of all calls in the state during 14 18.12 2 0.79 73,380 12.51 the study period.

10 13.99 8 3.15 272,025 46.36 Whereas every county 5 9.99 24 9.45 157,946 26.92 in Texas dealt with evac2.5 4.99 61 24.02 65,619 11.18 uees’ needs, the morerural and remote coun1 2.49 77 30.31 13,917 2.37 ties in south and west 0.5 0.99 43 16.93 3,589 0.61 Texas seemed to meet the .0 0.49 36 14.17 271 0.05 needs of their few evacuees with available commu0 0 3 1.18 0 0 nity resources. The 39 Sum n5254 100 586,747 100 “cold-spot” counties with December 2012

S440

Bame et al / Am J Prev Med 2012;43(6S5):S435–S442

50% less than expected call ratio had reported less than 0.05% of total calls (shown in dark blue and white, Figure 3).

Seventy-one percent of counties (n5181) had “normal” expected call ratios ranging from 0.5 to ,5.0, accounting for 14% of all calls (shown in yellow, green, light blue). This spatial analysis of adjusted call ratios reflecting unmet needs illustrated that the large metropolitan and urban evacuation destinations needed outside assistance and disaster resources and services for evacuees well beyond what they normally could provide. The urban and suburban locations along evacuation routes also needed outside help to support disaster victims. Communities unexpectedly distant from the disaster site, but hosting evacuees, would need further timely assistance.

Discussion The study of 2-1-1 call data during Katrina–Rita disaster phases represents a valuable process to evaluate unmet disaster needs recorded in real time for a broad geographic area over an extended time period. This study captured unmet disaster needs at both disaster sites and evacuation destinations through all phases of these disasters. This unique study enables examination of unmet needs based on real-time data, including timing and location of access barriers to disaster and community support services. The fındings illustrate using 2-1-1 data to plan and monitor effıcient allocation of resources and support services during disasters to enhance recovery of vulnerable populations. The same analytic approach could be applied to unmet medical and public health needs, epidemics, immunization campaigns, and health insurance funding effects on access barriers to healthcare services.

Analysis of calls over time revealed expected differences between the evacuation and landfall phases for the two hurricanes. After relocation of Katrina victims into Texas, Texas 2-1-1 call volume surged for an extended time. After emergency shelters began to close and fırst responders and volunteers left, the number of calls spiked. The volume of calls remained higher than twice the baseline rate as victims turned to local community services that were neither staffed nor budgeted to handle the burden of unmet disaster needs.

Spatial analysis of call volume covered disaster sites and evacuation routes and destinations. Mapping total calls revealed an expected correlation between volume of demand for unmet needs and population size, with 70% of calls requesting help in the fıve large metropolitan areas: Houston, Dallas, Ft. Worth, Austin, San Antonio.

These metropolitan centers were designated as shelter hubs, having greater supply and diversity of services to support evacuees. Other urban counties with moderately high call volumes were those located along evacuation

routes or selected for relocating Katrina victims needing to evacuate yet again. Counties located at Rita’s landfall and inland path had moderately high call volumes, but much greater than expected call rates for unmet needs given their population size.

The next step for 2-1-1 call centers is to develop realtime mapping that tracks changes in demand for resources and enables more-effıcient distribution of resources based on monitoring evolving needs. Refıning spatial analysis by type of unmet need and county, city, or ZIP code over time would enhance effectiveness of disaster management and mitigation. Overlaying spatial patterns of call demand and disaster paths would help to better direct disaster resources and indicate access barriers at shelter and supply distribution points. Maps by unmet need types also could be used alongside WebEOC to monitor distribution of emergency versus nonemergency services.

When controlling for population size, the disproportionate burden of unmet needs in evacuation destinations became evident. “Hot-spot” counties were disaster sites and evacuation destinations where volume of unmet needs was more than ten times greater than would be expected given population size. “Warm spots” were counties along evacuation routes where the number of calls was fıve to almost ten times greater than would be expected. The high-risk counties with a high proportion of the population requesting help indicated scarce resources and access barriers. This adjusted “at-risk” ratio may serve as an easily calculated measure of a community’s resiliency to meet the level of need given resources available for health and social support services. Thus, distribution of disaster resources should take into account not only the total number of disaster victims but also the relative burden on evacuation destinations given availability of resources within the communities involved.

Limitations and Lessons Learned Three study limitations regarding underestimation and bias should be noted. First, number of calls was under-reported during call surges. Staff were often overwhelmed and unable to log calls; calls exceeding capacity were dropped by the phone system. According to 2-1-1 personnel, these missed calls were not expected to differ from calls that had been logged.

Although call volume is underestimated, the fındings were expected to be a valid indicator of overall demand trends.

Second, callers were not representative of the general population; however, this bias could not be tested because caller demographic data were recorded unevenly in less than 15% of calls. Although 2-1-1 Texas services were www.ajpmonline.org

Bame et al / Am J Prev Med 2012;43(6S5):S435–S442

recommended by emergency managers and social support agencies, many Texans still remain unaware of this available assistance. Hence, the study population represented only those who encountered access barriers, knew about 2-1-1, had functional phone service, and were motivated to seek help.

Third, longitudinal analysis fındings were biased by 32% missing date information from four of the call centers that had lost their data and instead submitted a summary report of needs by location. Caller needs were missing randomly in 9% of calls, and approximately 7% of calls missed location data. Fortunately, because of random “roll-over” capability among the 2-1-1 Texas Network, a representative distribution and volume of calls had complete date recorded by other call centers for calls originating from counties with missing call center date information. Thus, it is expected that missing date information would underestimate unmet evacuation needs for two of the fıve state metropolitan areas (i.e., Dallas and Ft. Worth); however, the statewide trends were expected to be representative.

Two important lessons were learned from graphing call demand over time. First, unmet total needs surged for 2–3 weeks post-landfall following the departure of most fırst-responders and volunteers and the closing of disaster shelters. Thus, resources for this immediate phase post-event should be extended. Second, postdisaster call volume remained signifıcantly higher than the baseline period partially because of increased familiarity and use of 2-1-1 during the disasters. More importantly, this greater demand reflected greater local unmet needs compounded with disaster recovery needs. Routine unmet community needs were expected to increase because local resources had been diverted to evacuees. The 2-1-1 call centers reported that the greater threshold of increased unmet needs continued well beyond expected recovery periods, never returning to the pre-hurricane levels. Thus, disaster and community support systems need to plan for extended and compounded recovery periods.

Implications for Disaster Management and Policy Several policy implications emerge from this study. New disaster management strategies are needed for the extended evacuation of large urban populations. Emergency support may be needed for longer periods to meet basic needs for shelter, food, and health care. Short-term needs for supplies and services to enable evacuees to return home and to work may need to be re-evaluated, particularly if there were few habitable homes remaining or fewer jobs were available because businesses were damaged and closed. Delays in restarting utilities, particDecember 2012

S441

ularly in extreme heat and humidity conditions, further threaten public health and reduces the ability of community services to respond adequately. Because early recovery phases had a higher threshold of demand for unmet needs, increased support for evacuation destinations would need to be considered as the disaster victims seek help with temporary or permanent relocation.

Success of planning and monitoring resources to meet disaster needs may be examined by mapping both total calls and adjusted call ratio per population size. Disaster assistance traditionally has been focused on the disaster sites; however, with extended large-scale evacuations, fınancial support should be considered for communities along evacuation routes and at sheltering destinations.

During Katrina and Rita, Texas agencies and programs at evacuation destinations provided services well beyond their routine community needs, adversely straining their tight budgets. Retroactive reimbursement for these efforts was cumbersome, even daunting, particularly for small businesses and agencies. Hence, emergency managers should consider devising systems to quickly authorize vouchers and fee-for-services for local merchants and service providers.

Patterns seen in this study mirrored evidence in the post-disaster debriefıngs and anecdotes by those involved in Texas disaster management during Katrina–Rita. In turn, this study validates that 2-1-1 call data are important for monitoring disasters. Projects are underway to develop a nationwide capability to do this as calls are received. Spatial analysis in real time could promptly target neighborhoods in distress, effıciently organize evacuation of residents with special needs, set up distribution points closest to the high-risk populations, and monitor recovery problems and successes.

Implications for 2-1-1 Systems As 2-1-1 systems evolve, expand, and mature across the nation, much of their growth can be attributed to lessons learned from the study and analysis of 2-1-1 disaster data.

Over the past few years, local and regional emergency management personnel increasingly have used 2-1-1 data for planning purposes, enabling more accurate projections of day-by-day needs of the public, as well as specialized needs of vulnerable populations. Nationally, 2-1-1 systems have used analysis of call data to improve methods of operation to better serve 2-1-1 callers and deliver interventions. For example, after the diffıculty of managing and merging data during Katrina–Rita, Texas developed the KnowledgeBase, a statewide software system to communicate up-to-the-minute disaster information concurrently to all 25 call centers in the state. Texas 2-1-1 Network could serve callers from anywhere in the state faster and with more current information. Using a single

S442

Bame et al / Am J Prev Med 2012;43(6S5):S435–S442

software system enables data to be extracted more easily for reporting and further analysis.

Another practical outcome of 2-1-1’s role in disasters has been improvement of communication technology, along with national partnerships, that allow for a 2-1-1 system in one state to assist another 2-1-1 system in another state or region. After San Diego’s 2-1-1 became adept in handling disaster calls during southern California’s 2007 wildfıres, they were able to assist Louisiana systems when Hurricane Gustav headed for landfall in 2008. This 2-1-1 remote support, along with the vast disaster experience of the San Diego system, allowed the Louisiana public to be served more expeditiously. This was substantial aid to a state whose public’s memory and fear were still very fresh after mismanagement during the 2005 Hurricane Katrina disaster. Throughout evacuation from Hurricane Gustav, San Diego’s 2-1-1 was able to provide specifıc and clearly communicated instructions for callers, resulting in a more effective and safer course of action for communities affected by the disaster.

Collaborations on many levels have transpired as a result of 2-1-1 data analysis. In many regions, 2-1-1s have helped to form and support Voluntary Organizations Active in Disaster (VOADs), bringing nonprofıt and faith-based organizations together to share knowledge and resources throughout the disaster cycle to help their community. The 2-1-1 disaster data assist VOAD partners such as American Red Cross and Salvation Army to plan for mass sheltering and feeding; United Ways to plan for donation collections; and 2-1-1 systems to plan for staff and volunteer scheduling. Local and regional disaster managers increasingly rely on 2-1-1 data to identify location and types of unmet needs throughout disaster phases as well as baseline data to enhance community resiliency. Federal Emergency Management Agency (FEMA) regions are developing collaborative relationships with their local or statewide 2-1-1s for more-

effective screening of unmet needs and follow-up communication of disaster victims. Thus, 2-1-1 disaster data have served as a key element to stimulate development of standardized operational procedures and data collection practices, upgraded communication technology, and new partnerships and collaborations built among 2-1-1 systems across the U.S. Publication of this article was supported by funding from the National Cancer Institute (NCI) and the Offıce of Behavioral and Social Science Research (OBSSR) of the NIH (HHSN261201100469P).

The authors thank Ji Sun Lee and Michael Dunaway (Science & Technology Directorate, Department of Homeland Security, Washington, DC) and Beth Wick, Allen Irby, and Deborah Ballard (Texas I&R Network, Austin TX).

This research was funded by the Department of Homeland Security, Science and Technology Directorate, Division of Human Factors/Behavioral Sciences (2008 –2012).

No fınancial disclosures were reported by the authors of this paper.

References 1. Bustillo M. Houston grumbles as evacuees stay put. Los Angeles Times 2006, Aug 21. articles.latimes.com/2006/aug/21/nation/na-evac21.

2. Lindell M, Prater C. Hurricane Katrina and Rita: scientifıc surprises and policy failures. Presented at AFPCN, French Association for the Reduction of Natural Disasters; 2006, Feb 23. www.afpcn.org/060223Katrina/ AFPCN-060223-Lindell.pdf.

3. Nigg J, Barnshaw J, Torres M. Hurricane Katrina and the flooding of New Orleans: emergent issues in sheltering and temporary housing.

Ann Am Acad Polit Soc Sci 2006;604(1):113–28.

4. 2-1-1 Edmonton, Canada. Disaster response, 2009. www.211edmonton.com/ howwehelp/disasterresponse.php.

5. Berger E. Focus on sheltering: they say inland residents need to be accommodated during a hurricane. Houston Chronicle 2007, Aug 14. www.chron.com/disp/story.mpl/hurricane/ike/4839418.html.

www.ajpmonline.org