Modeling Hospital Response to Mild and Severe Influenza Pandemic Scenarios under Normal and Expanded Capacities

ABSTRACT

William Beaumont Army Medical Center conducted quantitative modeling with FluSurge 2.0 (Centers for Disease Control and Prevention) to determine hospital capabilities in responding to patient arrival surges of the Fort Bliss population in mild 1968-type and severe 1918-type influenza pandemics. Model predictions showed that William Beaumont Army Medical Center could adequately care for all intensive care unit (ICU) and non-ICU patients during a mild pandemic, particularly if hospital capacity was expanded using the emergency management plan, excess surge plan, or activation of a contagious disease outbreak facility. For a severe influenza pandemic, model predictions showed that hospital beds, ventilators, and other resources would be exceeded within 2 or 3 weeks. Even at maximal hospital expansion, for a 12-week severe pandemic with a 35% attack rate there would be peak demand for 214% of available non-ICU beds, 785% of ICU beds, and 392% of ventilators. Health care planners and decision-makers should prepare for resource challenges when developing plans for the next influenza pandemic.

Introduction

Epidemics of influenza occur annually around the world, usually between October and March in the northern hemisphere. Pandemics of influenza, which cause substantially more morbidity and deaths than do epidemics, occur infrequently, with three overwhelming pandemics being observed in the 20th century, namely, the 1918 Spanish, 1957 Asian, and 1968 Hong Kong influenza pandemics. The 1918 Spanish flu was catastrophic; it infected an estimated 1 billion people throughout the world and resulted in 21 to 40 million deaths.¹ This pandemic, in addition to being the most severe in the 20th century, distinguished itself by having at least one distinct peak of excess deaths in young adults between 20 and 40 years of age, an age group not normally adversely affected.² The 1957 and 1968 pandemics were mild, compared with the 1918 pandemic, and were typical in the sense that there were excess deaths in infants, elderly persons, and persons with chronic diseases. The U.S. Department of Health and Human Services estimates combined direct and indirect costs of $181 billion for a future “mild” (e.g., 1957- and 1968-type) pandemic with no intervention.³ Experts predict that another influenza pandemic is highly likely, if not inevitable, and prepandemic planning is essential for minimizing influenza pandemic-related morbidity, deaths, and social disruption.

Despite being aware of the catastrophic effects of an influenza pandemic, the current health care system is ill prepared for even a mild pandemic. For example, the short-lived 1999–2000 flu season caused a critical coast-to-coast shortage of acute patient beds, thereby causing ambulance service diversions, severely crowded emergency departments, and long delays in hospital admissions.⁴ These concerns prompted the development by the U.S. Department of Health and Human Services of a national plan for estimating the impact of deaths, hospitalizations, and outpatient visits attributable to pandemic influenza.³ This plan uses estimates based on the Centers for Disease Control and Prevention (CDC) Advisory Committee on Immunization Practices definition of groups at high risk for complications attributable to influenza.^5,6 These estimates are also embedded within the increasingly popular and freely downloadable FluSurge software models issued by the CDC.^7,8

FluSurge is a spreadsheet-based model that allows hospital administrators and public health officials to quickly estimate the surge in demand for hospital-based resources during the next influenza pandemic.⁹ FluSurge has been used to estimate the impact of an influenza pandemic on critical care capacity at various levels, including the national, state, metropolitan, and individual hospital levels. At the national level, FluSurge was used to model the impact of an influenza pandemic in South Korea¹⁰ and on critical care beds in England¹¹ and to estimate clinically ill persons, patients requiring outpatient care, potential hospitalizations, and potential deaths throughout the United States.³ The Department of Health Services at the state level has also used FluSurge for pandemic planning.¹² At the metropolitan level, the developers of FluSurge evaluated hospital capabilities for the Atlanta area by using 1968 mild pandemic parameters.⁸ Using the model parameters within Flu-Surge for both mild and severe pandemics, metropolitan Atlanta hospitals were predicted to have the capacity to sustain a 1968- type outbreak but would be overwhelmed in a 1918-type out-break.^13,14

At the individual hospital level, subject matter experts in a variety of fields have addressed the challenge of increasing hospital capacity in the event of an influenza pandemic. Based on these efforts, an individual hospital has the potential to double or triple overall critical care capacity through administrative, policy, and engineering controls.¹⁵ Considerable promise has been demonstrated with modifications to heating, ventilation, and air conditioning systems, particularly regarding ventilation (i.e., air changes), pressure differentials (i.e., positive versus negative pressure), and air filtration devices. Portable high-efficiency particulate air filters are an effective means of creating patient isolation areas with required negative pressure and room air changes and represent a significantly lower cost alternative, compared with construction of individual rooms or units.^16,17 Modifications to heating, ventilation, and air conditioning systems of commercial or public buildings by using off-the-shelf technologies that are modest in cost and are able to be implemented immediately could reduce risk of exposure to infectious biological agents.¹⁸ Reduction to exposure is critical because hospitals are major venues for contagious disease transmission, as evidenced by the finding that 77% of severe acute respiratory syndrome cases in Canada and 94% of severe acute respiratory syndrome cases in Taiwan resulted from inhospital exposures.¹⁹

The primary objective of this study was to model hospital capabilities in response to patient arrival surges during mild (1968-type) and severe (1918-type) influenza pandemic scenarios. Specifically, we used FluSurge 2.0 to estimate whether hospital resources at William Beaumont Army Medical Center (WBAMC) would be sufficient under four different conditions, namely, normal capacity, emergency management plan (EMP) capacity, excess surge capacity, and activation of a contagious disease outbreak facility in conjunction with the WBAMC EMP. For each of the aforementioned conditions, we estimated the number of days until hospital capacity would be exceeded (or not) with respect to intensive care unit (ICU) beds, non-ICU beds, and ventilators. We also estimated nursing staff requirements based on the increased patient population. These quantitative results provide a useful approximation regarding the effectiveness of the various options for expanding hospital capacity in meeting the challenges of an influenza pandemic.

Methods

This study was approved by the institutional review board at WBAMC (protocol 06/14). Hospital capacity was first assessed with respect to total ICU and non-ICU beds and number of ventilators. As shown in Table I, WBAMC currently has staffing for 104 beds (18 ICU and 86 non-ICU) under normal capacity. During a mass casualty event, the WBAMC EMP provides for hospital expansion up to 130 beds (four additional ICU beds and 22 additional non-ICU beds), which can be accomplished within 24 hours. In addition, WBAMC can provide up to 351 beds for excess surge capacity in a relatively short period of time. Finally, WBAMC has the capability to activate a contagious disease outbreak facility, which is physically separated from the main hospital, thus providing additional ICU (four beds) and non-ICU (24 beds) resources.

TABLE I

VARYING HOSPITAL BED CAPACITY FOR WBAMC

	No.
	C	EMP	EMP + COF	E
ICU beds	18	22	26	22
Non-ICU beds	86	108	132	329
Ventilators	18	22	26	22
Total beds	104	130	158	351

C, normal capacity; EMP, EMP expansion; EMP + COF, EMP and contagious disease outbreak facility; E, excess surge capacity.

The FluSurge 2.0 spreadsheet model was downloaded from the CDC.⁹ Modeling parameters within FluSurge 2.0 were set for both a 1968-type mild pandemic and a 1918-type severe pandemic, assuming a 12-week duration with attack rates of 15%, 25%, and 35%.²⁰ The Fort Bliss beneficiary population was estimated to be 32,925 at 0 to 19 years of age, 85,741 at 20 to 64 years of age, and 24,327 at ≥65 years of age. Because of the expanding Fort Bliss population attributable to base realignment and closure activities, these population values will likely underestimate Fort Bliss populations in the near future. The following assumptions were used for both pandemics: (1) the average length of a non-ICU hospital stay for influenza-related illness is 5 days, (2) the average length of an ICU hospital stay for influenza-related illness is 10 days, (3) the average duration of ventilator usage for influenza-related illness is 10 days, (4) the average proportion of admitted influenza patients needing ICU care is 15%, (5) the average proportion of admitted influenza patients needing ventilators is 7.5%, (6) the average proportion of influenza-related deaths assumed to occur in the hospital is 70%, and (7) the daily percentage increase in cases arriving, compared with the previous day, is 3%. Justifications for these assumptions and for FluSurge 2.0 modeling parameters are available for review.^5,6,8

As model output, FluSurge 2.0 estimated ICU and non-ICU bed requirements, number of patient ventilators, and number of deaths for the above-defined Fort Bliss populations. Estimates of the most likely, minimum, and maximum scenarios were calculated for the entire 12-week duration by using 15%, 25%, and 35% attack rates. For quantifying weekly totals, however, this model plotted only most likely values.²¹ We also estimated nursing requirements during an influenza pandemic. There is considerable variability in minimal nurse staffing ratios at hospitals. Laws and regulations governing these ratios are frequently revised and may not even be enforceable during an influenza pandemic. Because it has been demonstrated that there is a significant increase in the risk of adverse events when ICU nurses handle more than two patients,¹⁵ we assume one ICU nurse per two patients. For non-ICU nurses, we assume a nurse/patient staffing ratio of 1:4. We uniformly applied a 12- hour work shift for all nurses.

Results

Estimates for total WBAMC hospital admissions and total deaths in mild and severe, 12-week, influenza pandemics are presented in Table II. Hospital admissions for a 1968-type mild pandemic range from 140 patients (i.e., minimum scenario at 15% attack rate) to 982 patients (i.e., maximum scenario at 35% attack rate). For a severe 1918-type outbreak, hospital admissions range from 1,149 patients (i.e., minimum scenario at 15% attack rate) to 8,077 patients (i.e., maximum scenario at 35% attack rate). Similarly, predictive ranges of total deaths resulting from a mild pandemic (43–250 deaths) are much lower than those for a severe pandemic (352–2,052 deaths).

TABLE II

ESTIMATES OF TOTAL HOSPITAL ADMISSIONS AND TOTAL DEATHS FOR FORT BLISS POPULATION RESULTING FROM MILD 1968-TYPE AND SEVERE 1918-TYPE INFLUENZA PANDEMICS AT VARYING ATTACK RATES

	No.
	1968-Type Pandemic			1918-Type Pandemic
	15%a	25%	35%	15%	25%	35%
Total hospital admissions
Most likely	321	535	749	2,640	4,400	6,160
Minimum	140	233	326	1,149	1,915	2,680
Maximum	421	702	982	3,462	5,769	8,077
Total deaths
Most likely	67	112	157	553	921	1,289
Minimum	43	71	100	352	587	821
Maximum	107	178	250	879	1,466	2,052

^a

Attack rate.

To evaluate the effectiveness of hospital capacity in handling a surge in patient arrivals, a time series analysis is useful. For each of the pandemic scenarios, we show the weekly values for the number of non-ICU influenza patients (Table III), the number of ICU influenza patients (Table III), and the number of patients requiring ventilators (Table III). When patient demand exceeds available supply for each of these resources is indicated (see Table I for bed/ventilator capacities under each of the hospital conditions).

TABLE III

COMPARISON OF NON-ICU BED, ICU BED, AND VENTILATOR REQUIREMENTS FOR HOSPITAL PATIENTS FOR 1968-TYPE AND 1918-TYPE PANDEMICS WITH 15%, 25%, AND 35% ATTACK RATES AND 12-WEEK DURATION UNDER NORMAL AND EXPANDED CAPACITIES

		No.
	Attack Rate	Week 1	Week 2	Week 3	Week 4	Week 5	Week 6	Week 7	Week 8	Week 9	Week 10	Week 11	Week 12
Non-ICU beds
1968	15%	2	9	17	24	31	35	37	33	28	21	14	8
	25%	4	16	28	39	51	59	61	55	46	35	24	13
	35%	6	22	39	55	72	83	86	77	65	49	33	18
1918	15%	19	78	136 (1, 2, 4)	194 (1, 2, 4)	252 (1, 2, 4)	291 (1, 2, 4)	301 (1, 2, 4)	270 (1, 2, 4)	229 (1, 2, 4)	173 (1, 2, 4)	117 (1, 2, 4)	62
	25%	32	129 (1, 2)	226 (1, 2, 4)	323 (1, 2, 4)	420 (1, 2, 3, 4)	485 (1, 2, 3, 4)	502 (1, 2, 3, 4)	449 (1, 2, 3, 4)	381 (1, 2, 3, 4)	288 (1, 2, 4)	196 (1, 2, 4)	1031
	35%	45	181 (1, 2, 4)	317 (1, 2, 4)	453 (1, 2, 3, 4)	589 (1, 2, 3, 4)	679 (1, 2, 3, 4)	703 (1, 2, 3, 4)	629 (1, 2, 3, 4)	533 (1, 2, 3, 4)	403 (1, 2, 3, 4)	274 (1, 2, 4)	144 (1, 2, 4)
ICU beds
1968	15%	0	2	4	6	8	10	11	10	9	7	5	3
	25%	1	4	7	11	14	17	18	17	15	12	8	5
	35%	1	5	10	15	20(1)	23 (1, 2, 3)	25 (1, 2, 3)	24 (1, 2, 3)	21 (1)	16	12	7
1918	15%	4	18	35 (1, 2, 3, 4)	52 (1, 2, 3, 4)	69 (1, 2, 3, 4)	83 (1, 2, 3, 4)	87 (1, 2, 3, 4)	86 (1, 2, 3, 4)	73 (1, 2, 3, 4)	58 (1, 2, 3, 4)	41 (1, 2, 3, 4)	24 (1, 2, 3)
	25%	7	29 (1, 2, 3, 4)	58 (1, 2, 3, 4)	87 (1, 2, 3, 4)	116 (1, 2, 3, 4)	138 (1, 2, 3, 4)	146 (1, 2, 3, 4)	143 (1, 2, 3, 4)	122 (1, 2, 3, 4)	96 (1, 2, 3, 4)	68 (1, 2, 3, 4)	40 (1, 2, 3, 4)
	35%	9	41 (1, 2, 3, 4)	81 (1, 2, 3, 4)	122 (1, 2, 3, 4)	162 (1, 2, 3, 4)	193 (1, 2, 3, 4)	204 (1, 2, 3, 4)	200 (1, 2, 3, 4)	171 (1, 2, 3, 4)	134 (1, 2, 3, 4)	95 (1, 2, 3, 4)	56 (1, 2, 3, 4)
Ventilators
1968	15%	0	1	2	3	4	5	5	5	4	4	2	1
	25%	0	2	4	5	7	8	9	9	7	6	4	2
	35%	1	3	5	7	10	12	12	12	10	8	6	3
1918	15%	2	9	17	26 (1, 2, 3)	35 (1, 2, 3, 4)	41 (1, 2, 3, 4)	44 (1, 2, 3, 4)	43(1, 2, 3, 4)	37 (1, 2, 3, 4)	29 (1, 2, 3, 4)	20 (1)12	12
	25%	3	15	29 (1, 2, 3, 4)	43 (1, 2, 3, 4)	58 (1, 2, 3, 4)	69 (1, 2, 3, 4)	73 (1, 2, 3, 4)	71 (1, 2, 3, 4)	61 (1, 2, 3, 4)	48 (1, 2, 3, 4)	34 (1, 2, 3, 4)	20(1)
	35%	5	21 (1)	41 (1, 2, 3, 4)	61 (1, 2, 3, 4)	81 (1, 2, 3, 4)	97 (1, 2, 3, 4)	102 (1, 2, 3, 4)	100 (1, 2, 3, 4)	86 (1, 2, 3, 4)	67 (1, 2, 3, 4)	47 (1, 2, 3, 4)	28 (1, 2, 3, 4)

Numbers in parentheses indicate patient bed capacity exceedance for the following conditions: 1, normal capacity; 2, EMP; 3, excess surge capacity; 4, EMP and contagious disease outbreak facility.

For a mild 1968-type influenza pandemic, WBAMC is predicted to have sufficient non-ICU bed capacity regardless of the attack rate (Table III). WBAMC also has sufficient ICU bed capacity for a mild pandemic except when the attack rate is 35% (Table III). Patient demand for ICU beds exceeds the 18 available ICU beds under normal capacity from week 5 until week 9. When EMP and excess surge hospital expansion increases ICU bed capacity to 22 beds, patient demand for ICU beds still exceeds supply from week 6 until week 8. WBAMC is predicted to meet ICU patient bed requirements for all mild influenza scenarios only when using both the EMP and contagious disease outbreak facility. Finally, our model results predict that WBAMC has enough ventilators to meet patient demand under all mild influenza pandemic scenarios (Table III).

All model scenarios for a severe 1918-type influenza pandemic predict that patient demand for hospital resources (i.e., ICU beds, non-ICU beds, and ventilators) would exceed available supply within 2 to 3 weeks (Table III). The maximal surge of patients occurs in week 7 for a 12-week pandemic with a 35% attack rate. During this peak time, even with expanded hospital resources, patient demand is 214% for non-ICU beds (703 patients with only 329 beds), 785% for ICU beds (204 patients with only 26 beds), and 392% for ventilators (102 patients with only 26 ventilators). Consequently, expansion of hospital capabilities with the EMP, excess surge capacity, or activation of a contagious disease outbreak facility is predicted to fall short in meeting the demand for all patient arrivals. It should be noted, however, that excess surge capacity (i.e., 328 non-ICU beds) is predicted to meet the patient demand for non-ICU bed capacity for a 1918-type pandemic with a 15% attack rate (Table III). If the attack rate is increased to 25% and 35%, then even this significant expansion of non-ICU bed capacity is exceeded by week 5 and week 4, respectively.

We provide estimates of non-ICU and ICU nurse staffing requirements in Table IV. The maximal number of non-ICU nurses during a 12-hour shift for a 1968-type mild pandemic with a 35% attack rate is estimated to be 22 nurses during week 7 (Table IV). For a 1918-type severe pandemic with a 35% attack rate, we estimate 176 non-ICU nurses per 12-hour shift during week 7. The maximal number of ICU nurses during a 12-hour shift for a 1968-type mild pandemic with a 35% attack rate is estimated to be 13 nurses during week 7, whereas the estimate is 102 nurses for a severe pandemic (Table IV).

TABLE IV

ESTIMATES OF NURSE STAFFING REQUIREMENTS PER 12-HOUR WORKSHIFT FOR 1968- AND 1918-TYPE PANDEMICS

		No. of Nurses
	Attack Rate	Week 1	Week 2	Week 3	Week 4	Week 5	Week 6	Week 7	Week 8	Week 9	Week 10	Week 11	Week 12
Non-ICU
1968	15%	1	2	4	6	8	9	9	8	7	5	4	2
	25%	1	4	7	10	13	15	15	14	12	9	6	3
	35%	2	6	10	14	18	21	22	19	16	12	8	5
1918	15%	5	20	34	49	63	73	75	68	57	43	29	16
	25%	8	32	57	81	105	121	126	112	95	72	49	26
	35%	11	45	79	113	147	170	176	157	133	101	69	36
ICU
1968	15%	0	1	2	3	4	5	6	5	5	4	3	2
	25%	1	2	4	6	7	9	9	9	8	6	4	3
	35%	1	3	5	8	10	12	13	12	11	8	6	4
1918	15%	2	9	18	26	35	42	44	43	37	29	21	12
	25%	4	15	29	44	58	69	73	72	61	48	34	20
	35%	5	21	41	61	81	97	102	100	86	67	48	28

Non-ICU nursing requirements assume a 1:4 nurse/patient ratio. ICU nursing requirements assume a 1:2 nurse/8 patient ratio and 12-hour work shifts.

Discussion

This modeling study demonstrates that WBAMC has the resources (beds, ventilators, and nurses) to adequately manage a mild 1968-type influenza pandemic for the current Fort Bliss population. Expansion of hospital capabilities, particularly implementation of the EMP and activation of a contagious disease outbreak facility, ensures that a patient arrival surge causes no lapse in ICU and non-ICU patient care for all scenarios. These model estimates, however, assume current Fort Bliss population values and do not account for noninfluenza patients competing for hospital resources. Furthermore, the Fort Bliss military population is increasing on a daily basis, which we must address in future model simulations, and we do not account for a potential influx of populations from El Paso City/County and bordering Juarez, Mexico. Consequently, WBAMC planners should not be confident that these currently favorable model estimates necessarily reflect reality. WBAMC would likely need to expand hospital capacity even during a mild influenza pandemic.

In the event of a severe 1918-type influenza pandemic, hospital beds and other resources will be overwhelmed within 2 or 3 weeks. Patient care will be adversely affected during a severe pandemic. As an Army medical training facility, WBAMC has a large personnel resource pool of nurses and doctors, as well as a civilian and contractor workforce. Because it is impossible to predict illness rates for nurses, we cannot estimate whether WBAMC will be able to maintain recommended nurse staffing levels. Irrespective of whether an influenza pandemic is mild or severe, WBAMC is not an isolated health care facility and will need to pool resources with other El Paso City/County hospitals and clinics. Plans are already being developed for these contingencies.

Conclusions

This comparative modeling study quantifies over time whether WBAMC is predicted to have adequate resources under normal and expanded operating conditions. The FluSurge 2.0 model predictions estimate that WBAMC currently has the resources to handle a mild influenza pandemic without adversely affecting patient care. If necessary, WBAMC is capable of expanding its ICU and non-ICU capacity to meet predicted surges in patient arrivals. In the event of a severe 1918-type influenza pandemic, hospital resources such as ICU beds, non-ICU beds, and ventilators are predicted to be overwhelmed within 2 or 3 weeks, regardless of implementation of hospital expansion options. Decision-makers at all levels should consider these modeling results when developing pandemic influenza response plans.

Footnotes

1

Presented at the 9th Annual Force Health Protection Conference, August 10, 2006, Albuquerque, NM.

Author notes

2

The views expressed in this article are those of the authors and do not reflect the official policy or position of the Department of the Army, the Department of Defense, or the U.S. government.