Hospital Emergency Surge Capacity: An Empiric New York Statewide Study

Introduction 

Current national policies seek improvement in hospitals’ ability to respond to disasters. Federal agencies recommend that hospitals be able to rapidly accommodate 500 new inpatients per million regional population.1 In mass-casualty events or infectious epidemics, the resources of individual hospitals will be overwhelmed, and coordination of regional resources is essential.2, 3, 4, 5 Federal agencies recommend formulation of plans in which the responses to extraordinary events build on an infrastructure of existing resources in ordinary daily use.4

Out-of-hospital and emergency department (ED) responses to disasters have been planned extensively, but plans for subsequent hospital care have received less attention.6 The ability of hospitals to accommodate new inpatients is crucial for the maintenance of ED functions in a mass-casualty incident. The lack of published data on the balance between projected disaster needs and existing hospital resources makes planning difficult. Licensed hospital beds can be counted according to publicly available information in most states. However, it cannot be assumed that peak hospital capacity stated on administrative operating certificates represents the number of functional beds ready to be used with available staff, supplies, and equipment.7 Data on typical hospital occupancy and its variation are not readily available from public sources of information. Furthermore, published empiric data are not available about the typical number of vacant functional beds available in an emergency or the number of additional beds that can be quickly opened by discretionary reduction of hospital occupancy.

The present study was conducted to empirically describe the typical statewide and regional occupancy of hospitals in New York. Patterns of variation in occupancy were identified. The ability of hospitals to accommodate new patients immediately and by discretionary reduction of occupancy was estimated.

Results 

Daily occupancy for 1996 to 2002 at 242 New York State hospitals was analyzed. The occupancy (peak, mean, and SD) at each hospital is shown, for children (Figure 1A) and adults (Figure 1B). Few hospitals treat as many as 20 children at a time, whereas many treat more than 100 adults at a time. The year, month, and day of the week were each associated with variation in hospital occupancy (P<.05, for pediatric and adult activity). These trends are apparent by visual inspection of daily occupancy data for yearly and seasonal variation (Figure 2A for children; Figure 2C for adults) and for variation with day of the week (Figure 2B for children; Figure 2D for adults). Occupancy tended to decline during successive years of the study period. Peaks and troughs, respectively, were reached in February and August for children and January and August for adults. Weekly variation tended to peak on Tuesdays for children and Wednesdays for adults, whereas the lowest occupancy occurred on Saturdays for both age groups.

View Large Image Download to PowerPoint

Figure 1. Occupancy (numbers of patients) at each New York State hospital (A) for pediatric and (B) adult patients. Large dot: mean; bar: standard deviation; small dot: peak. Hospitals are ordered from lowest to highest mean occupancy.

View Large Image Download to PowerPoint

Figure 2. Statewide total daily hospital occupancy (numbers of patients) (A) for children during January 1996 to December 2002, (B) for children in May and June 2002, (C) for adults during January 1996 to December 2002, and (D) for adults in May and June 2002.

During the study period, statewide peak capacity (sum of 95th-percentile highest occupancies for each hospital) was 2,707 for children and 46,613 for adults. These peak capacities correspond to 703 beds per million children and 3,045 beds per million adults. Throughout the study period, statewide pediatric percent occupancy was 60%, whereas adult percent occupancy was 82%. Fewer beds for new pediatric admissions were typically available, with an average of 268 vacant functional pediatric beds per million children and 555 vacant functional adult beds per million adult population. During periods when hospitals were typically busiest (Tuesdays in February for children and Wednesdays in January for adults), 193 and 328 vacant functional beds per million age-specific population were available for children and adults, respectively. During the least active periods (Saturdays in August), 354 and 733 vacant functional beds per million age-specific population were available for children and adults, respectively. Although population-based vacancy rates among 8 regions vary by a factor of 2 (Table 1), none of the regions could readily accommodate 500 per million children.

Table 1. Average hospital vacancies in the 8 regions⁎ of New York State.

	Region	Average Regional Hospital Vacancies per Million Age-Specific Population
		Children	Adults
	1. Western New York	216	796
	2. Finger Lakes	187	516
	3. Central New York	262	460
	4. New York–Pennsylvania	208	453
	5. Northeast New York	239	486
	6. Mid-Hudson	291	595
	7. New York City	359	588
	8. Nassau–Suffolk	156	465

⁎ Regions are classified as the health service areas of the New York State Department of Health.9

After the September 11, 2001, attacks, statewide occupancy changed insignificantly. However, in the New York City region, a discretionary modification of admissions and discharges resulted in a 9% and 8% reduction in occupancy by September 12 from the September 10 baseline for children and adults, respectively. Because occupancy generally increases from Monday to Wednesday, a slightly larger reduction is evident when this intraweek trend is taken into account. Using a linear regression model, we estimate that on average each New York City hospital reduced occupancy by 1.3 pediatric patients (95% CI 0.27 to 2.28 patients) and 27.7 adult patients (95% CI 20.04 to 35.31 patients) during the week of September 11 relative to the expected occupancy on September 12, 2001. These hospital-level averages correspond to city-level reductions of 88 pediatric patients (10.7%) and 1,938 adult patients (10.6%) below the expected occupancy for September 12, 2001.

Limitations 

The following limitations and assumptions should be considered in interpreting our findings. The analysis is limited to overall inpatient resources. The reported data pertain to the initial transition from ordinary use of hospitals to a disaster response. Subsequent availability of hospital resources in a protracted crisis and analysis of specific hospital services are beyond the scope of this study.

Discussion 

Federal targets for hospital surge capacity are defined in Health Resources and Services Administration guidance to states.1 The recommended surge capacity in the 2004 guidelines was “500 adult and pediatric patients per 1,000,000…above the current daily staffed bed capacity….” Target proportions of pediatric and adult patients and particular disorders were not specified. A newer version of the guidelines specifies a surge capacity response time of 3 hours.15 Because hospital facilities for children and adults are largely distinct and incidents could involve large numbers of children or adults, we evaluated pediatric and adult capacity separately.

The target surge bed ratio of 500 patients per million population was derived from National Disaster Medical System projections, data from established trauma regions and systems in the United States and other countries, and on the basis of expert judgment.1 The targets are intended by federal agencies to define an order of magnitude for projected surge capacity and provide a starting point to develop plans that can be realistically achieved in the immediate future. The federal guidance statement acknowledges that terrorism scenarios can be envisioned that would impose even greater burdens on the health care system.

In New York State, there are fewer total pediatric beds and fewer vacant functional pediatric beds per age-specific population than are typically available for adults. Although this discrepancy is appropriate to the ordinary age-related rates of hospitalization, gaps will often be encountered in trying to accommodate 500 new patients per million age-specific population recommended by federal agencies in a major emergency. The mismatch between projected needs and existing resources is likely to be greatest for children.

In evaluating hospital bed capacity, we assumed that beds routinely used for pediatric or adult care could provide appropriate and satisfactory care in disasters if patient severity and complexity are similar to those of the patients routinely served at each hospital. If patients’ treatment needs in a disaster differ substantially from the ordinary, then emergency capacity is more difficult to estimate. It is also assumed that hospitals remain accessible, transportation is available, information about vacant beds is known, and decisionmaking authority exists to effectively distribute patients to the appropriate vacant bed. In some states, real-time central reporting of bed availability provides a mechanism to match needs and resources.16

Administratively defined hospital capacity may differ from functional capacity,7 and published estimates of peak functional hospital capacity are not available. Therefore, an empiric method was needed to define functional peak. We selected the 95th percentile for occupancy at each hospital to identify an optimistic high capacity that was repeatedly achieved during the 7-year period of observation. Other empiric measures of peak occupancy could be considered; for example, the absolute peak, or 99th percentile achieved at each hospital. However, these very high occupancies would represent singular events, likely dependent on ideal circumstances and not pertinent to disaster scenarios. Assumptions have been selected to provide plausible best-case estimates. If widespread disruption compromises transportation, communication, or facility operations, then surge capacity would be further reduced.

Capacity estimates reported in this study represent the entire study period. Higher occupancy in earlier years of the study period may overestimate present functional peak capacity. Ongoing national hospital resource reduction may be especially problematic for accommodating surges of burn patients.17 If scheduled hospital staffing is lower in summer or on weekends, reflecting lower expected patient loads, then functional peak capacity at these times may be lower than suggested by our data.

Estimates of discretionary reduction of hospital occupancy by more than 2000 patients after September 11, 2001, should be regarded as conservative. During September 11, it quickly became evident that few hospital resources would be needed for inpatients in the New York City region. It is likely that a larger reduction in occupancy could have been achieved, had it been necessary. Surveys of nurse managers and physicians suggest that one third of hospitalized adults could be discharged within 24 hours to accommodate new patients in a major emergency.18 One might expect that the greater the need, the greater the discretionary reduction in occupancy. Although previous reports19, 20 describe reductions in occupancy at individual facilities after September 11, 2001, to our knowledge the present report is the first to provide region-wide data for a postdisaster discretionary reduction in hospital occupancy.

Although patterns of hospital activity in New York State may not be directly generalizable to other states, these population-based data provide an empiric basis for quantitative models to study strategies extending hospital resources in major emergencies. In summary, the data suggest that even under optimistic assumptions, 500 new patients per million age-specific population will often overwhelm existing hospital resources, especially for an incident involving large numbers of children. Further analysis of tradeoffs will be necessary to determine how to better accommodate emergency surges. Baseline resources could be increased. However, costs may be prohibitive, with little gain in the routine function of the health care system. Alternatively, plans could be made to extend the capacity of existing resources in disasters by altering standards of care4 or facilitating assistance by neighboring regions. Finally, the selection of 500 patients per million as a surge target is somewhat arbitrary.1 For each region, local factors may suggest optimal ways to balance needs and resources.

In conclusion, our report of limitations in hospital surge capacity demonstrates a contrast between the matching of normal needs and resources with that in disasters. The data indicate the need for decisionmaking processes, professional clinical guidelines, and a legal basis for the transition between normal hospital operations and disaster responses. The accommodation of surges of patients in disasters may necessitate altering standards of care or facilitating the use of resources from neighboring regions.