Integrating Hospitals into Community Emergency Preparedness Planning

Context Recent natural disasters and terrorist attacks have underscored the necessity for health care facilities to integrate their activities with other community response teams. Contribution The investigators developed and administered a nationwide survey questionnaire to assess the existence and character of hospitalcommunity services linkages that facilitated the response to local emergencies. Most responding hospitals conducted community drills; analyzed threat vulnerability; and planned for additional supplies, equipment, and decontamination facilities. Other linkages were less widespread. Cautions Hospital response rate to questionnaires was low, and answers were unverified. Implications Effective coordination of effort requires development of national standards for community preparedness. The Editors Hospital personnel play an important role in disaster response. Their roles vary according to the type of disaster, location, and availability of local resources and can include bioterrorism incident identification, triage and treatment of victims, and promoting accurate and consistent public information. The effectiveness of hospital staff response is greatly enhanced by preevent integration into the community emergency preparedness and response planning process (1). The hospital that establishes linkages clarifies its role and promotes interaction between essential personnel and available community resources that can enhance hospital surge capacity. Recent reports have expressed concern that hospitals are not adequately integrated into community planning. Hospitals are said to be isolated in their planning activities and are possibly the weakest link in emergency response (2-5). To better understand the extent to which hospitals are integrated into community planning, we assessed hospital and community linkages. We also examined the hypotheses that better linkages would be associated with hospitals that perceived themselves to be at risk for a high number of hazards or threats, those located in a community that had experienced an actual disaster, and those that had previously prepared for a major event that required national security oversight (Appendix). During the first 72 hours of a disaster, local agencies are generally the first to respond (6); first responders are often members of the affected community (1, 7). Community preparedness is a complex concept that requires a system-level response because multiple stakeholders are involved and many potential hazards exist. A well-prepared community will have a comprehensive planning process, a thorough emergency operations plan, established response capability, and an ongoing surveillance and notification system for identifying and communicating emergencies. These 4 domains form the conceptual framework for this study (Figure 1), which assessed emergency preparedness and response planning linkages between hospitals and community stakeholders in a national random sample of hospitals. Figure 1. Domains and stakeholders for assessing community emergency preparedness and response linkages. Methods The study methods comprised the following: convening a technical expert panel, developing questions for the survey instrument, developing a hospital sampling strategy, administering the survey instrument and collecting data, and analyzing and interpreting of the results. The Joint Commission on Accreditation of Healthcare Organizations' (hereafter referred to as the Joint Commission) external institutional review board approved the study. To maintain promised confidentiality, no identifiable hospital names or locations were reported. Questionnaire Development The technical expert panel included 12 members with experience across a range of relevant disciplines; the panel comprised a balance of practicing clinicians, academic researchers, and agency emergency preparedness experts. The panel met 4 times, in person and by conference call, to identify topic areas and issues and to review draft questions and pilot results. Following an in-person panel meeting and detailed literature search, the project team drafted a pilot questionnaire that was tested in 9 hospitals. The final version contained 57 items across several topic areas (Table 1). Because the questionnaire was administered before the National Incident Management System was published, it was not entirely consistent with the system's language (Appendix). Table 1. Topics Addressed in the Questionnaire Sample and Implementation Strategy A simple random sample of 1750 hospitals was drawn from the population of all U.S. general medicalsurgical hospitals in the 2003 American Hospital Association database (8) (n= 4863). A 2-phased implementation strategy was used. First, the president of the Joint Commission and the study's principal investigator cosigned invitation letters that were mailed to the chief executive officers of all hospitals in the sample in late January 2004. The invitation letters indicated that participation in the research was entirely voluntary, anonymous, and unrelated to accreditation. In addition, the Joint Commission, the National Rural Health Association, and the American Hospital Association sent listserv messages announcing the project. In the second phase, chief executive officers who replied positively to the invitation provided the name and title of the contact person most familiar with emergency preparedness at their hospital. The questionnaires were mailed to the designated contact person in February 2004. Over the next 8 weeks, nonrespondents received a reminder postcard, personal telephone calls, and a final e-mail message to encourage completion of the questionnaire. Incoming questionnaires were examined for missing or inconsistent information; requests for clarification were sent by e-mail to the contact person. Statistical Analysis Because no widely accepted, predefined model or requirements for linkages existed, we attempted to determine which questionnaire items were most important. The technical expert panel collaborated with the study team during fall 2004 to identify 17 questionnaire items (one of which comprised 6 subitems) that represented minimum basic elements of effective community linkages; these elements were chosen on the basis of face validity, expert opinion, and published literature. To assess reliability, we measured the consistency of an individual's response over time to 3 questionnaire items of varying complexity in a random sample of 52 hospitals. Respondents were contacted by e-mail and were asked to complete 3 follow-up questionnaire items that corresponded with items in the original questionnaire. We calculated the agreement between follow-up items and original responses. Agreement was defined as 2 yes responses or 2 no or don't know responses for each hospital and question over time. In addition, the accuracy of data entry was assessed by calculating agreement from duplicate entry of 12 randomly selected questionnaires. Three demographic factors were used to define groups for comparison: the hospital's average daily census, whether the hospital was located in a Metropolitan Statistical Area (Appendix) or in a rural area, and the duration of the hospital's involvement in community emergency preparedness planning. We also defined groups by 3 risk and experience factors: perceived risk of 6 or more hazards or threats, previous preparation for an event requiring national security oversight, and experience responding to a public health emergency or actual disaster in the community. We used SAS statistical software, version 9.1 (SAS Institute, Cary, North Carolina) for all analyses. For analysis of the 17 basic elements, sampling weights were used to adjust the results for nonresponse (9). To determine the weights, we used logistic regression to estimate the probability that a sampled hospital had completed the survey as a function of bed capacity, accreditation status, location (urban or rural), and region. The estimated response probabilities from this regression were then grouped into 12 weighted adjustment classes so that the number of responses within each class was at least 20 and the units within each class were as similar as possible, based on the estimated probabilities. The inverse of the average predicted probability of response within each weighted adjustment class was then used as the weight. The means and 95% CIs for each of the 17 basic elements, both overall and stratified by the demographic characteristics, were then calculated by using these sampling weights. The association of the basic elements with each of the demographic characteristics was determined by using weighted chi-square tests (in these weighted analyses, PROC SURVEYFREQ and PROC SURVEY MEANS statements were used). A 2-tailed P value of less than 0.05 indicated statistical significance. When interpreting the results, the reader should use caution because the analyses were not adjusted for multiple comparisons. For each of 6 demographic and experience factors, 23 comparisons were made (138 total comparisons). Therefore, approximately 7 comparisons were expected to be significant by chance. The number of missing responses is reported in the text whenever 10 or more responses were missing. Role of the Funding Source The funding source (Agency for Healthcare Research and Quality) was represented by a project officer on the expert panel but had no influence over the design, conduct, and analysis of the study or the decision to submit the manuscript for publication. Results Of the 1750 hospitals that were invited, 678 (39%) chief executive officers responded positively to the letter of invitation. Forty hospitals (2%) declined to participate, 1019 (58%) did not respond to the letter, and 13 questionnaires (<1%) were returned as undeliverable. Of the 678 hospitals that received questionnaires, 575 (85%) were completed. One or more hospitals from 48 states and the District of Columbia respo

H ospital personnel play an important role in disaster response. Their roles vary according to the type of disaster, location, and availability of local resources and can include bioterrorism incident identification, triage and treatment of victims, and promoting accurate and consistent public information. The effectiveness of hospital staff response is greatly enhanced by preevent integration into the community emergency preparedness and response planning process (1). The hospital that establishes linkages clarifies its role and promotes interaction between essential personnel and available community resources that can enhance hospital surge capacity.
Recent reports have expressed concern that hospitals are not adequately integrated into community planning. Hospitals are said to be isolated in their planning activities and are possibly the weakest link in emergency response (2)(3)(4)(5). To better understand the extent to which hospitals are integrated into community planning, we assessed hospital and community linkages. We also examined the hypotheses that better linkages would be associated with hospitals that perceived themselves to be at risk for a high number of hazards or threats, those located in a community that had experienced an actual disaster, and those that had previously prepared for a major event that required national security oversight (Appendix, available at www .annals.org).
During the first 72 hours of a disaster, local agencies are generally the first to respond (6); first responders are often members of the affected community (1,7). Community preparedness is a complex concept that requires a system-level response because multiple stakeholders are involved and many potential hazards exist. A well-prepared community will have a comprehensive planning process, a thorough emergency operations plan, established response capability, and an ongoing surveillance and notification system for identifying and communicating emergencies. These 4 domains form the conceptual framework for this study (Figure 1), which assessed emergency preparedness and response planning linkages between hospitals and community stakeholders in a national random sample of hospitals.

METHODS
The study methods comprised the following: convening a technical expert panel, developing questions for the survey instrument, developing a hospital sampling strategy, administering the survey instrument and collecting data, and analyzing and interpreting of the results. The Joint Commission on Accreditation of Healthcare Organizations' (hereafter referred to as the Joint Commission) external institutional review board approved the study. To maintain promised confidentiality, no identifiable hospital names or locations were reported.

Questionnaire Development
The technical expert panel included 12 members with experience across a range of relevant disciplines; the panel comprised a balance of practicing clinicians, academic researchers, and agency emergency preparedness experts. The panel met 4 times, in person and by conference call, to identify topic areas and issues and to review draft questions and pilot results. Following an in-person panel meeting and detailed literature search, the project team drafted a pilot questionnaire that was tested in 9 hospitals. The final version contained 57 items across several topic areas ( Table  1). Because the questionnaire was administered before the National Incident Management System was published, it was not entirely consistent with the system's language (Appendix, available at www.annals.org).

Sample and Implementation Strategy
A simple random sample of 1750 hospitals was drawn from the population of all U.S. general medical-surgical hospitals in the 2003 American Hospital Association database (8) (n ϭ 4863). A 2-phased implementation strategy was used. First, the president of the Joint Commission and the study's principal investigator cosigned invitation letters that were mailed to the chief executive officers of all hospitals in the sample in late January 2004. The invitation letters indicated that participation in the research was entirely voluntary, anonymous, and unrelated to accreditation. In addition, the Joint Commission, the National Rural Health Association, and the American Hospital Association sent listserv messages announcing the project. In the second phase, chief executive officers who replied positively to the invitation provided the name and title of the contact person most familiar with emergency preparedness at their hospital. The questionnaires were mailed to the designated contact person in February 2004. Over the next 8 weeks, nonrespondents received a reminder postcard, personal telephone calls, and a final e-mail message to encourage completion of the questionnaire. Incoming questionnaires were examined for missing or inconsistent information; requests for clarification were sent by e-mail to the contact person.

Statistical Analysis
Because no widely accepted, predefined model or requirements for linkages existed, we attempted to determine which questionnaire items were most important. The technical expert panel collaborated with the study team during fall 2004 to identify 17 questionnaire items (one of which comprised 6 subitems) that represented minimum basic elements of effective community linkages; these elements were chosen on the basis of face validity, expert opinion, and published literature.
To assess reliability, we measured the consistency of an individual's response over time to 3 questionnaire items of varying complexity in a random sample of 52 hospitals. Respondents were contacted by e-mail and were asked to complete 3 follow-up questionnaire items that corresponded with items in the original questionnaire. We calculated the agreement between follow-up items and original responses. Agreement was defined as 2 "yes" responses or 2 "no" or "don't know" responses for each hospital and question over time. In addition, the accuracy of data entry was assessed by calculating agreement from duplicate entry of 12 randomly selected questionnaires.
Three demographic factors were used to define groups for comparison: the hospital's average daily census, whether the hospital was located in a Metropolitan Statistical Area (Appendix, available at www.annals.org) or in a rural area, and the duration of the hospital's involvement

Context
Recent natural disasters and terrorist attacks have underscored the necessity for health care facilities to integrate their activities with other community response teams.

Contribution
The investigators developed and administered a nationwide survey questionnaire to assess the existence and character of hospital-community services linkages that facilitated the response to local emergencies. Most responding hospitals conducted community drills; analyzed threat vulnerability; and planned for additional supplies, equipment, and decontamination facilities. Other linkages were less widespread.

Cautions
Hospital response rate to questionnaires was low, and answers were unverified.

Implications
Effective coordination of effort requires development of national standards for community preparedness.

-The Editors
Article Hospital Emergency Preparedness Linkages in community emergency preparedness planning. We also defined groups by 3 risk and experience factors: perceived risk of 6 or more hazards or threats, previous preparation for an event requiring national security oversight, and experience responding to a public health emergency or actual disaster in the community.
We used SAS statistical software, version 9.1 (SAS Institute, Cary, North Carolina) for all analyses. For analysis of the 17 basic elements, sampling weights were used to adjust the results for nonresponse (9). To determine the weights, we used logistic regression to estimate the probability that a sampled hospital had completed the survey as a function of bed capacity, accreditation status, location (urban or rural), and region. The estimated response probabilities from this regression were then grouped into 12 weighted adjustment classes so that the number of responses within each class was at least 20 and the units within each class were as similar as possible, based on the estimated probabilities. The inverse of the average predicted probability of response within each weighted adjustment class was then used as the weight. The means and 95% CIs for each of the 17 basic elements, both overall and stratified by the demographic characteristics, were then calculated by using these sampling weights. The association of the basic elements with each of the demographic characteristics was determined by using weighted chi-square tests (in these weighted analyses, PROC SURVEYFREQ and PROC SURVEY MEANS statements were used). A 2-tailed P value of less than 0.05 indicated statistical significance. When interpreting the results, the reader should use caution because the analyses were not adjusted for multiple comparisons. For each of 6 demographic and experience factors, 23 comparisons were made (138 total com-

Table 1. Topics Addressed in the Questionnaire
Hospital and community planning Community-wide training and drills Community plans for additional supplies and equipment Systems for public health reporting and laboratory capacity Community surge capacity plans relating to staffing, transportation, and sharing information about available resources Incident management Communication mechanisms and protocols parisons). Therefore, approximately 7 comparisons were expected to be significant by chance. The number of missing responses is reported in the text whenever 10 or more responses were missing.

Role of the Funding Source
The funding source (Agency for Healthcare Research and Quality) was represented by a project officer on the expert panel but had no influence over the design, conduct, and analysis of the study or the decision to submit the manuscript for publication.

RESULTS
Of the 1750 hospitals that were invited, 678 (39%) chief executive officers responded positively to the letter of invitation. Forty hospitals (2%) declined to participate, 1019 (58%) did not respond to the letter, and 13 questionnaires (Ͻ1%) were returned as undeliverable. Of the 678 hospitals that received questionnaires, 575 (85%) were completed. One or more hospitals from 48 states and the District of Columbia responded (no response was received from hospitals in Delaware and Vermont).
When we compared hospitals in the random sample with general medical-surgical hospitals throughout the United States, we found no significant differences in number of beds (P ϭ 0.78), urban or rural location (P ϭ 0.29), region (P ϭ 0.56), or accreditation status (P ϭ 0.29). Among those invited, there were significant differences in the rate of completed questionnaires by demographic cat-egory ( Table 2). Significantly lower response rates were found among hospitals that were located in rural areas, had fewer beds, were not accredited, and were located in the West North Central region of the country (P Ͻ 0.001 for all variables).
Among responding hospitals, slightly more than half were located in urban areas. The median average daily census across participating hospitals was 87 (interquartile range, 26 to 193). Thirty-five percent were trauma centers. The median number of emergency department visits per day was 70 (interquartile range, 36 to 113). Most (71%) had received $50 000 or less in federal or state funding for preparedness activities in 2003, and 12% received between $50 001 and $200 000; 15% received no funding (n ϭ 516; data missing for 59).
Hospital chief executive officers designated staff members from a wide range of departments as the hospital's primary contact person. Most frequently cited were security (24%), emergency services (17%), administration (13%), emergency management (12%), and facility operations or environmental services (10%). Of these personnel, 19% held positions classified as senior leadership (officer, administrator, or vice president). The median number of staff contributing to the survey per hospital was 4 (interquartile range, 3 to 6). The median time reported to complete the questionnaire was 120 minutes (interquartile range, 90 to 240 minutes).
Regarding the reliability of responses, the consistency corresponded with the item's complexity; 71% agreed on the ability to request and receive laboratory testing of suspected Centers for Disease Control and Prevention category A agents on an around-the-clock basis, 87% agreed on the triage strategy addressed in the community plan, and 79% agreed on hospital involvement in community exercising of the Strategic National Stockpile (SNS) (Appendix, available at www.annals.org). A total of 49 data entry discrepancies were found among 9408 opportunities for error (784 data points multiplied by 12 questionnaires), yielding an overall rate of agreement of greater than 99%.

Community Planning Process, Emergency Operations Plan, and Experience Factors
All simple percentages reported before the results of the basic elements of linkages represent unweighted estimates. Appendix Tables 1, 2, and 3, which present unweighted estimates, are available at www.annals.org. Nearly all responding hospitals reported that their community had a group or committee that was responsible for emergency preparedness planning and response activities. The median number of community entities represented in the groups was 11 (interquartile range, 9 to 13). Most community groups included representation from traditional emergency responders, but few included members from industry or manufacturing, local media, or professional groups. The median number of times per year the community planning group reportedly met in person was 6 (interquartile range, 4 to 12). Of all respondents, 75% reported that the community also had a coalition of health care organizations that coordinated health care emergency planning and response.
Most hospitals reported that their involvement in community planning predated 2001; 27% of respondents reported involvement before 1990. Most (86%) reported that they had conducted a threat and vulnerability analysis in conjunction with other community responders. The Frequency calculated on the basis of unweighted analysis of 575 responses to the question, "Do you perceive your hospital to be at increased risk for any of the following hazards or threats (check all that apply)?" Article Hospital Emergency Preparedness Linkages www.annals.org median number of perceived community hazards or threats was 5 (interquartile range, 3 to 6); hazardous materials, winter storms, and tornadoes were the most commonly reported perceived threats (Figure 2).
Most (86%) hospitals reported using an incident management system; 65% of hospitals reported that their incident management system was developed collaboratively with the local emergency management agency. Nearly three quarters (73%) of the hospitals reported that their communities had crisis communication protocols, and 41% reported being involved in the protocol's development.
Almost all respondents reported that their community had an emergency operations plan that specifically ad-dressed health and medical response. Reports of community plans to expand hospital capacity varied by topic area. More than three quarters of respondents reported that their community plans addressed immunization and prophylaxis for key personnel and the need to expand hospital decontamination capacity. However, only about one half of hospitals reported that the community plans addressed the need to expand hospital capacity to isolate people or support ventilator-dependent patients. Close to one half reported that the community plan addressed mechanisms for tracking patient location and managing a large volume of calls.
Most hospitals reported that their communities had identified nontraditional ways to transport victims to Article Hospital Emergency Preparedness Linkages health care facilities and to provide additional morgue space. Most had identified community-based alternate care sites; however, less than one half reported having formal written agreements with at least 1 site. Similarly, most reported community plans addressed mental health needs of victims and their families, emergency responders, and hospital staff, but few reported having formal written agreements with mental health providers. Fewer than one half (46%) of hospitals reported that they had responded to a public health emergency or actual disaster since 2001; only 16% reported an emergency that substantially challenged the hospital's functional capacity. Twenty-seven percent reported that their community had prepared for a national security event.

Established Response Capability
Unweighted responses to items related to established response capability can be found in Appendix Table 2 (available at www.annals.org). More than three quarters of hospitals had participated in community-wide emergency preparedness and response training since 2001, most often in a classroom setting. The median number of training topics per hospital was 3 (interquartile range, 2 to 4). More respondents (92.4%) reported that their hospitals had participated in community-wide drills or exercises than training; the median number of drills or exercises per hospital between 2001 and 2004 was 4 (interquartile range, 2 to 5). The median length of drills or exercises was 4 hours (interquartile range, 3 to 6 hours). Of the 1567 total reported drills or exercises for which complete time and date information was specified, 78% occurred entirely within a single day shift and only 9% lasted longer than 24 hours. Few of these exercises (36.6%) were unannounced. Most respondents reported that their community emergency operations plan used a formal incident management system and designated where incident management would occur if there was no emergency scene. Regarding communication, fewer than half of the hospitals had around-theclock access to a live voice from a public health department (43%) or volunteer organization (41%) representative.

Surveillance, Reporting, and Laboratory Identification
Fewer than one half of responding hospitals reported that their community plan addressed laboratory testing for individual category A agents in the unweighted analysis (Appendix Table 3, available at www.annals.org). Approximately 52% reported that they could request laboratory testing and receive results around the clock. Of all respondents, 55.6% had a direct electronic link to their state's Health Alert Network and 54.5% reported that their communities had developed standardized epidemiologic forms for case identification.

Overall Perception of Practice
After completing the questionnaire, hospital representatives were asked a single global question in an effort to identify best practices related to linkages: "Do you believe your hospital and community are potential examples of exemplary practice?" Forty-five percent of hospitals responded positively to this question. Table 3 presents the weighted associations between the 17 basic elements of linkages and demographic characteristics. In general, the analysis showed that positive responses on many linkage items were significantly higher among hospitals that were located in urban areas, those with a higher average daily census, and those that became involved in community planning before 2001. However, a higher proportion of rural hospitals reported that their community plan addressed alternate sites of care. There were no significant associations between demographic factors and 3 elements: hospital linkage to the state Health Alert Network, a community plan that addressed creden- Article Hospital Emergency Preparedness Linkages tialing, and completion of a threat and vulnerability analysis in collaboration with the community. There were many significant associations in the weighted analyses between the 17 basic elements (23 items total) and perceived hazards and experience factors ( Table  4). A perceived high number of hazards was associated with 8 items at a P value less than 0.001 and with 7 items at a P value less than 0.05. Previous preparation for a national security event was associated with 17 items at a P value less than 0.001 and with 2 items at a P value less than 0.05. Experience responding to an actual event was associated with 9 items at a P value less than 0.001 and with 7 items at a P value less than 0.05. The high number of significant associations in the bivariate analyses may be related to correlations among the factors.

DISCUSSION
The study accomplished a cross-sectional baseline assessment of community emergency preparedness linkages among hospitals, public health agencies, and traditional first responders in a national sample of hospitals. Our study focused on hospital integration related to 4 domains: the community emergency preparedness and response planning process; the community emergency operations plan; established response capability; and ongoing surveillance, reporting, and laboratory identification.
The range of departments in which hospital contact persons were employed suggests that primary responsibility for hospital emergency preparedness varies widely. Consequently, no single discipline or professional group can be identified as consistently responsible for hospital preparedness, making it difficult for community groups to determine the appropriate hospital liaison. The relatively low percentage of senior leaders completing the questionnaire may indicate insufficient resources or executive attention devoted to emergency preparedness planning.
Having representation from each of the key stakeholders in the community planning process is essential to establishing good linkages. Many communities lacked involvement by media outlets and volunteer organizations, both of which are integral to an effective response (10,11). Some hospitals reported that their communities did not have a crisis communication protocol, an important strategy to provide consistent messages, reduce anxiety levels, and deter concerned persons from seeking care unnecessar- ily (12). Many hospitals also reported no community plans to augment hospital-based surge capacity in the area of pharmaceuticals, supplies, equipment, and isolation. Lack of sufficient plans in these areas could have severe consequences if a serious incident occurs (such as an outbreak of severe acute respiratory syndrome). The Joint Commission accreditation requirements (13) may partially explain why almost all hospitals were engaged in community-wide drills or exercises. Unfortunately, most drills and exercises were short and included only staff on day shifts, which is clearly inadequate for actual emergencies that can occur at any time. The small proportion of respondents involved in community-wide exercising of the SNS is similar to recent reports in which fewer than one half of respondents from state and local public health departments reported that they had exercised their SNS plans (14); only 6 states had adequate capacity to deliver and administer vaccines and antidotes from the stockpile (15). The Centers for Disease Control and Prevention's proposed budget for fiscal year 2006 incorporates an increase of more than 50% for the SNS program (16); however, funding for exercising the SNS was not included (17).
Early detection, identification, and intervention at the local level (6,18), together with ongoing surveillance and reporting, are essential for managing bioterrorist events or infectious disease outbreaks. We have identified opportunities for improving coordination among hospitals, public health departments, and laboratories, including the need for more direct links to the state Health Alert Network, more standardized epidemiologic forms, and around-theclock ability to request tests and receive laboratory results. These findings are consistent with a recent report that two thirds of the states do not use the Internet to collect disease outbreak information, which would cause serious delays in reporting and could potentially impede early warning of disease threats (15).
In general, urban and large hospitals demonstrated similar performance on the basic elements of community preparedness linkages that we identified in this study, but small and rural hospitals varied. This finding is consistent with patterns of preparedness funding in which larger hospitals in urban areas received greater assistance (19). The association between several basic elements and previous experience with coordinating national security events could suggest that such experience is actually a proxy for the overall quality of community preparedness. This might lead one to conclude that the federal coordination involved in preparing for major events is an effective stimulus for promoting linkages and actual preparedness. On the other hand, the association could be indirect; for example, the communities received increased funding that could have caused the improved preparedness. Of interest, other studies also found greater public health preparedness after involvement in a national security event (20,21).
The research team did not expect that nearly half of all respondents would consider their hospital and community to be potential examples of exemplary practice. The high estimation of exemplary practice by respondents might be explained by 1 or more of the following factors: a sincere belief based on evidence or experience that they are well linked, a tendency to overestimate personal capability (22,23), naive beliefs regarding the level of preparedness necessary for an effective response, and the possibility of response bias.
A few previous studies have investigated specific aspects of hospital and community emergency preparedness linkages. Before fall 2001, several studies reported low levels of preparedness linkages between hospitals and communities but substantial improvements thereafter (24 -28). In one study, local and state health officials declared that the most important preparedness enhancements have involved developing strong relationships and connectivity with hospitals and other entities (29).
This study has 4 main limitations. First, the responses reflect only the hospitals' perception of their community linkages. It is unclear whether community stakeholders share these perceptions, and no objective source documents were available for verification. As with all survey research, the results are limited by respondents' knowledge about specific topic areas.
Second, the overall response rate was low, and small and rural hospitals were underrepresented. The weighted analysis was designed to address this issue. It is possible that hospitals that did not respond to the invitation were substantially less integrated into community planning than those who participated. Another potential bias is that respondents may have tended to report better linkages because the questionnaire came from the Joint Commission.
Third, the questionnaire addressed a broad range of topics but did not assess the quality or actual effectiveness of hospital and community linkages in each area. For example, most hospitals reported completing a threat and vulnerability analysis in conjunction with the community, but we do not know if these were comprehensive evaluations that included hazard identification, vulnerability assessment, and risk analysis (30). The selection of the basic elements of community emergency preparedness linkages was an initial attempt to prioritize linkage issues; however, the study team cautions that additional consideration by local and national experts is needed before further use.
Fourth, many specific items in community plans were assessed, but experience has demonstrated that implementation often fails to proceed according to plan. Future studies should use a standardized approach to evaluate drills or event responses so that assessments can be compared with preevent information about linkages and the planning process. Additional research, preferably by a team of multidisciplinary experts, is needed to evaluate the quality of linkage relationships and planning activities and to define a threshold of adequate integration in critical hospital-community interfaces. Periodic reevaluation of linkages would allow policymakers to assess progress over time.
To our knowledge, this study is the first large-scale, national assessment of hospital integration into community emergency preparedness. Overall, responding hospitals reported a substantial degree of integration into community preparedness planning and familiarity with local emergency response plans. At the same time, results related to communication and planning with local public health departments suggest that relationships between hospitals, public health agencies, and other critical community response resources are not adequately robust. The results also suggest a need for greater attention to basic linkage elements among rural hospitals. Rural hospitals have a unique need for collaborative planning regarding surge capacity given severe constraints in equipment and supplies, limited access to medical specialists and additional staff, and an environment that often includes physical proximity to hazardous materials and the threat of agroterrorism (4).
Specific recommendations for enhancing linkages relate to exercises and coordination. Plans are most effective when rigorously exercised. Drills and exercises should be designed to stress the community system-level response over time, and they should address event notification, communication, resource allocation, and patient management. Financial support and other incentives are needed to increase the rigor and scope of community-wide exercises to ensure that response capacity and capability meet the expectations of the U.S. public.
Coordination among health care organizations and between health care and community planning groups should be enhanced. Health care organizations can form coalitions to facilitate a community-wide inventory of medical assets and ensure that they are not each relying on the same constrained community resources. Health care coalitions can also establish a uniform interface with the jurisdiction's incident management system (31). Community planning groups, which are generally better integrated with public health personnel, should reach out to health care providers so that public health departments and providers effectively complement their respective response capabilities. Similarly, health care organizations should clearly understand their role in the community incident command system to ensure compatibility across incident command structures.
Furthermore, the accuracy of both the hospital's perception of its linkages and that of emergency preparedness experts would be greatly enhanced if there were a benchmark for measuring the effectiveness of linkages. For many years, hospitals have desired clearer guidance on expectations for preparedness, such as achievable, objective national standards. The call for national standards is coming from other stakeholders as well, particularly those charged with evaluating the capacity of the public health and safety systems to respond to terrorist events and those responsible for evaluating the impact of substantial funding increases since 2002 (3,15,16,(32)(33)(34)(35)(36)(37). A nationally accepted model is useful for several purposes: establishing accepted expectations for preparedness, tracking and accounting for federal preparedness funds, and determining which programs or areas need improvement.
With or without national standards for linkages, policymakers should consider offering incentives to drive improvements and help counter the current hospital financial environment. Hospitals are continually asked to do more with less (38), and they face the likelihood of additional cuts to reimbursement (39). Although post-2001 funding for hospital preparedness has improved the national state of preparedness, future funding for hospital, state, and local public health preparedness is likely to decrease (17). Unlike public sector organizations, local governmental authorities lack direct control over private health care assets and therefore have less leverage to promote participation (2). There is a natural tendency to shift attention away from preparedness toward more immediate issues as the memory of major events recedes. One must remember that the quality of community linkages is just 1 aspect of preparedness that will influence hospital ability to respond to an event. Other major issues to consider include limited bed capacity when hospitals are full, seriously overcrowded emergency departments, and potential failures related to the civil infrastructure (for example, electricity, water, and fuel).
In conclusion, future events and disasters will require local leadership, emergency responders, hospital staff, and other health care providers to manage a coordinated response. Preestablishing strong linkages for preparedness and response among community stakeholders should be useful for pandemic influenza, earthquakes, and train derailments, as well as terrorist events. Growing evidence suggests that hospital and community linkages have substantially improved in recent years, but more improvement is needed. The process of building and maintaining linkages through collaborative planning must be dynamic. Plans should evolve as people, threats, systems for detection and response, and funding priorities change (40). Complacency must not compromise our national progress toward achieving strong linkages among the entities that are critical to limiting the human impact of future disasters.

APPENDIX Terminology
An event requiring national security oversight, as defined in the National Response Plan (41), is one that, by virtue of its political, economic, social, or religious significance, may be the target of terrorism or other criminal activity. Examples of national special security events include presidential inaugurations, the Olympics, national political conventions, the Super Bowl, and the United Nations General Assembly (42).
The National Incident Management System (43) delineates a standard method for managing any large-scale incident in the United States. All organizations seeking to maintain eligibility for federal preparedness funding will be required to incorporate this method into their preparedness plans. The system's incident command protocol provides terminology and concepts for interface between different organizations during incident response. These concepts should be considered in developing any final model for hospital linkages and should be used when designing future research questionnaires in this area.
The Centers for Disease Control and Prevention's Strategic National Stockpile (SNS) has large quantities of medicine and medical supplies to protect the U.S. public if there is a public health emergency (for example, a terrorist attack, influenza outbreak, or earthquake) severe enough to exhaust local supplies. After federal and local authorities agree that the SNS is needed, medicines will be delivered to any state in the country within 12 hours. Each state has plans to receive and distribute SNS medicine and medical supplies to local communities as quickly as possible (44).
A Metropolitan Statistical Area is a county or group of contiguous counties that contains at least 1 city with a population of 50 000 people or more or a U.S. Census Bureau-defined urbanized area of at least 50 000 people with a metropolitan population of at least 100 000 people. In addition to the county containing the main city or urbanized area, a Metropolitan Statistical Area may contain other counties that are metropolitan in character and are economically and socially integrated with the central counties. In New England, cities and towns (rather than counties) are used to define Metropolitan Statistical Areas (45).

Additional Tools and Related Initiatives
Strategies to achieve compliance with national preparedness standards, if they were to be developed, have been described by the Congressional Research Service, and some have already been implemented (46). In addition, recent national initiatives are likely to facilitate improvements in linkages. These include the National Response Plan (41) and the Homeland Security Presidential Directive 8 (47), which calls for the development of a national preparedness goal and a national preparedness assessment and reporting system. Both the Health Services and Resources Administration and the Centers for Disease Control and Prevention have been working toward establishing goals and indicators for hospital and public health preparedness, respectively (19,48). The Office of the Assistant Secretary for Public Health Emergency Preparedness has issued cross-cutting critical benchmarks that apply both to hospitals funded by the Health Services and Resources Administration and to public health agencies supported by the Centers for Disease Control and Prevention. The Agency for Healthcare Research and Quality has produced several helpful documents and tools related to emergency preparedness planning and response, such as guides for dispensing mass quantities of prophylactic medications to communities, evaluating drills, and developing emergency contact centers (49 -51).