Appendix 2: New York City Pilot Test

In this appendix, the modeling assumptions and output from the New York City pilot test are described. The scenario involved a category 4 hurricane that required evacuation of 24 hospitals and 61 nursing homes.

Modeling Assumptions

Listed below are key assumptions used in the model for the New York City pilot test. In the discussion of the modeling results, we have varied some of these assumptions in order to understand their effect on evacuation times.

Vehicles used in the evacuation

• Four types of vehicles are used to transport patients—buses, wheel chair vans, BLS ambulances, and ALS ambulances. (Notably, we have not included helicopters, which could be used to transport the most critical patients.) The capacities of the vehicles are:

Vehicle Type	Capacity (Patients)
Bus Standard bus Mini-wagon	321 8
Wheel chair van Hydraulic lift Ramp-wagon Ambulette	8 wheel chair patients 3 wheel chair patients 2 wheel chair patients
BLS ambulance	2
ALS ambulance	1

1. Actual capacity is 35 passengers; we assume 3 will be hospital staff and/or guardians in the evacuation.

• The number of vehicles, by type, that are available for the evacuation:

Vehicle Type	Number available	Assumptions
Bus Standard bus Mini-wagon	1,159 2,334	90% of the NYC Department of Education (DOE) fleet (1,288) 90% of the DOE fleet (2,593)
Wheel chair van Hydraulic lift Ramp-wagon Ambulette	337 397 913	90% of the DOE fleet (374) 90% of the DOE fleet (441) 40% of the Paratransit fleet (2,282)
BLS ambulance	245	40% of the volunteer sector fleet (131) 40% of the private sector fleet (482)
ALS ambulance	34	40% of the private sector fleet (85)

Patients to be evacuated

• 24 hospitals and 61 nursing homes are evacuated.
• Evacuating facilities are at 100% capacity. Based on capacity data provided by NYC Office of Emergency Management (OEM) staff, the total number of patients to be evacuated is 24,393—9,885 in hospitals and 14,508 in nursing homes.
• No patients will self-evacuate. All evacuated patients will be transported to another facility on one of the vehicles described above.
• Evacuated patients are divided into four types. Each patient type is assumed to be evacuated on a specific type of vehicle:

Patient Type	Vehicle Used for Patient Transport
Ambulatory patients	Bus
Wheel chair bound patients	Ambulette
Bedridden patients requiring constant medical attention at the BLS level during transport	BLS ambulance
Bedridden patients who need constant medical attention at the ALS level during transport	ALS ambulance

• The percentage of patients requiring each vehicle type for transport is the same at all evacuating hospitals. The percentage of patients requiring bus, wheel chair van, BLS, and ALS transport is based on (1) a census count at New York Presbyterian/Weill Cornell Medical Center on April 4, 2006 and (2) a study of evacuation transport requirements at four hospitals in the North Shore LIJ Health System. We used a weighted average of these two sources, assuming two-thirds of the evacuating hospitals had the New York Presbyterian patient mix and one-third had the North Shore patient mix. The resulting percentages are:

Vehicle Type	Percent of Patients Requiring Vehicle Type (Hospitals Only)
Bus	33%
Ambulette	40%
BLS ambulance	13%
ALS ambulance	14%

• The percentage of patients requiring each vehicle type for transport is the same at all evacuating nursing homes. The percentage of patients requiring bus, wheel chair van, BLS, and ALS transport is based on patient mobility data that NYC nursing homes reported to Centers for Medicare & Medicaid Services (CMS) on the CMS-672 form. The model assumes the following percentages:

Vehicle Type	Percent of Patients Requiring Vehicle Type (Nursing Homes Only)
Bus	27%
Ambulette	70%
BLS ambulance	2%
ALS ambulance	1%

Facilities receiving evacuated patients

• Any hospital or nursing home in NYC that is not being evacuated will accept evacuating patients. "Receiving" hospitals will only accept patients evacuated from hospitals; "receiving" nursing homes will only accept patients evacuated from nursing homes.
• Receiving facilities are at 100% capacity.
• Receiving facilities have 15% surge capacity. That is, the number of patients a facility can accept is equal to 15% of their capacity.
• Patients who cannot be accommodated in NYC receiving facilities are transported to a generic out-of-NYC facility. The travel time from an evacuating facility to the out-of-NYC facility is 90 minutes.
• The model does not consider the capacity of different units within a receiving facility. An evacuated patient can be placed in any bed in the receiving facility. In other words, we assume acuity affects only the transport vehicle requirement, and not the bed requirements in the receiving facility (e.g., an ICU patient in an evacuating hospital can be relocated to a floor room, with additional equipment brought in, at the receiving hospital).

Staff available to carry out the evacuation

• The model does not consider staff availability. It assumes that sufficient staff are available to move patients from their rooms to the facility lobby, load them on the vehicle, transport them to the receiving facility, and move them from the vehicle to their new room. In particular, we assume that:
  • sufficient facility staff are available to move patients from their room to the lobby of the facility, so that vehicle staff do not have to retrieve patients from their room.
  • sufficient facility staff are available to accompany patients in the vehicle to the receiving facility, if required (e.g., high-risk patients or minors).
  • enough drivers, EMTs, and paramedics are available to staff the vehicles 24/7 during the entire evacuation period.

Steps in the evacuation process

• Facility staff move patients from their room to the lobby of the facility. The time required to move patients from their room to the lobby of the facility is not included in the total evacuation time. The model does not calculate the time required to move patients from their room to the lobby, and assumes that facility staff will be able to deliver patients to the lobby so that when an evacuation vehicle arrives at the facility, the patients to be transported on that vehicle are waiting in the lobby.
• Once a vehicle arrives at the facility, patients are immediately loaded onto vehicles. As noted above, patients are waiting in the facility lobby for the vehicle. The model assumes no vehicular congestion at the facilities that would delay loading or unloading. The assumed loading time for vehicles is:

Vehicle Type	Loading Time (minutes)
Bus Standard bus Mini-wagon	30 15
Wheel chair van Hydraulic lift Ramp-wagon Ambulette	30 15 10
BLS ambulance	10
ALS ambulance	20

• Evacuated patients will be transported to the nearest available facility that has capacity. Hospital patients are transported only to hospitals, and nursing home patients only to nursing homes. The model ignores "preferred" receiving facility lists for each evacuating facility (e.g., pairs of facilities with memorandums of understanding [MOUs] regarding evacuation).
• The model sends ALS patients to the nearest facilities. Given that assignment, the model sends BLS patients to the nearest facilities that still have available capacity. Van patients have the next highest priority, and bus patients have the lowest priority.
• All facilities are evacuated simultaneously, depending on vehicle availability.
• Travel times between facilities are based on actual travel times between New York Presbyterian hospitals and other NYC facilities during 2005. Ambulances making these trips averaged 20 miles per hour. In the model, all vehicles are assumed to have the same travel speed. To estimate the travel times between all pairs of facilities, we developed a statistical model that relates the New York Presbyterian travel times and the distances between pairs of facilities (using the facility latitude and longitude) and whether inter-borough travel is required.
• The time required to unload patients from a vehicle is the same as the loading time (see above). The model assumes no vehicular congestion at the facilities that would delay patient unloading.
• Staff at the receiving facility move patients from the vehicle to the patient’s new room. The elapsed time from the curb to the patient’s room is not included in the total evacuation time.

Return to Contents

Modeling Results

Results with baseline assumptions

Using the baseline assumptions described above, the estimated total evacuation time is 54 hours. As noted earlier, the "evacuation clock" starts when the first patients begin loading on vehicles at evacuating facilities and ends when the last patient has been loaded on a vehicle at an evacuating facility. Total evacuation time does not include:

1. Time from when evacuation decision is made to when the first patient is in the facility lobby available for transport.
2. Travel time from "vehicle fleet garages" to an evacuation facility at the beginning of the evacuation process.
3. Time to move the patient from their room to the lobby of the evacuating facility.
4. Time to move patients from the curb at the receiving facility to their room.

Total evacuation time depends entirely on transport of ALS patients. The model predicts that the total time to evacuate ALS patients far exceeds the time required to evacuate BLS, van, or bus patients. Therefore, the total evacuation time is determined by the total time to evacuate ALS patients from facilities (primarily hospitals). Although the majority of patients can be moved in wheelchair vans or buses, the small number of ALS ambulances (34) and the fact that they can only move one patient at a time means that those patients (1,529 of them) take the longest time to evacuate. With a 20-minute loading and 20-minute unloading time, the minimum time for each trip is inflexible and drives the total evacuation time.

The model prioritizes ALS ambulances to go the shortest distance (to limit travel time for those patients), but they make on average 45 round trips during the evacuation (see the table below). BLS ambulances taking two patients each and with a greater fleet of vehicles make only four trips each. Wheelchair vans (hydraulic lifts) can take more people but there is a much greater pool of patients to transport. These vans each make six trips to evacuate. Buses are plentiful and each only needs to make one trip in order to evacuate all ambulatory patients. The number of round-trips required during the evacuation for each vehicle is summarized below:

Vehicle Type	Number of Round-Trips Required per Vehicle
Bus	1
Ambulette	6
BLS ambulance	4
ALS ambulance	45

Using the baseline assumptions, 69% of nursing home patients and 67% of hospital patients would have to be transported outside NYC. We assume that hospitals and nursing homes are at 100% capacity, no one self-evacuates, and facilities have 15% surge capacity. The model prioritizes patients by acuity to ensure that ALS patients travel the least distance and ambulatory patients travel furthest. All ambulatory patients are transported out of NYC along with two thirds of the wheelchair patients. All ALS patients, BLS patients and one third of the wheelchair patients fill the surge capacity of the local facilities.

ALS patient evacuation time under alternative assumptions

Given that the total evacuation time is equal to the total time required to evacuate ALS patients, it is important to determine to what extent assumptions regarding ALS patients and transport can be changed and still achieve the desired maximum total evacuation time of 72 hours. We have made only one-at-a-time changes in assumptions.

Percentage of patients in evacuating facilities requiring ALS transport

The baseline assumption is that 6.3% of all patients will require ALS transport, which includes 14% of hospital patients (i.e., 1,384 patients) and 1% of nursing home patients (i.e., 145 patients). The actual percentage of patients requiring ALS transport will very likely be different in a real evacuation. The graph below shows the time required to evacuate all ALS patients assuming higher percentages of ALS patients. If we assume that 8% of all patients require ALS transport (a re-classification of only 440 patients), the model predicts a 71 hour evacuation time for ALS patients.

[D] Select for Text Description

Because each ALS unit can only transport one patient and because the load and unload time for ALS patients is long relative to travel time, the percentage of ALS patients affects the evacuation time in a nearly linear fashion. In this model the percentage of ambulatory patients was left constant while the percentage of BLS and wheelchair was decreased (with the remainder reclassified as ALS).

Capacity of nearby receiving facilities to accept ALS patients

The model assumes that the highest acuity patients go to the closest facilities, and the lowest acuity patients go to the furthest facilities. As a result, facilities that are the closest to the evacuation zone receive a high percentage of ALS patients compared to other facilities, which of course begs the question of whether they can accommodate that many ALS patients. In the model we can reduce the number of ALS patients that any facility receives by reducing the surge capacity.

[D] Select for Text Description

Reducing the surge capacity reduces the total number of ALS patients each facility must take. ALS patients will still be transported first to nearby facilities, so these facilities will continue to receive only ALS patients. However, the total population of ALS patients will be spread over a greater area. As surge capacity decreases, the length of evacuation time greatly increases, with ALS patient travel time increasing as they are transported to further and further facilities. At 5% surge capacity, evacuation of ALS patients exceeds the 72-hour limit.

Load and unload times

While the baseline assumptions include a 20-minute loading time and 20-minute unloading time for ALS ambulances, the model implicitly assumes that there is no congestion at the ambulance load/unload area. Below, we account for congestion by varying the loading and unloading times—the additional time could be viewed as queue delays at the load/unload area.

[D] Select for Text Description

ALS patient evacuation is very sensitive to increases in load/unload time. For ALS patients who are traveling very short distances, the 20 minute load time and 20 minute unload time is a significant proportion of the total time for evacuation. Doubling the load and unload time quickly brings the total evacuation time from 54 hours to 83 hours, and the 72-hour limit is already exceeded at a 75% increase in load/unload time. Other vehicles for which the load/unload time is a smaller percentage of total evacuation time, and which are not required to make as many round trips, are not affected to the same degree.

Travel times

As noted in Chapter 1, travel times in the model are based on actual travel times experienced in inter-hospital transports to New York Presbyterian hospital. Ambulances traveled an average of 20 miles per hour on these trips. Below, in the model we change the travel time assumptions to simulate more congested conditions.

[D] Select for Text Description

Changes in travel time have the greatest effect on ALS ambulances, because they do multiple round trips during the evacuation (by contrast, each bus requires only a single trip during the evacuation). Slowing travel speed (accomplished in the model by proportionately increasing travel time) gradually increases the amount of time to evacuate ALS patients, but not to the same extent as changes in load time.

Changes in baseline assumptions that result in no patients transported out of NYC

As noted above, the model predicts that 69% of nursing home patients and 67% of hospital patients are transported outside NYC. To keep all patients in NYC facilities, the surge capacity would have to be 45% for hospitals and 48% for nursing homes.

For comparison purposes, changing the assumptions regarding capacity and self-evacuation have the following effects:

• Capacity: If all facilities were at 95% capacity, they would need to reach 38% and 41% surge capacity for hospitals and nursing homes, respectively, to ensure all patients stay in NYC. At 90% capacity, the required surge capacity would drop to 31% and 33%.
• Self-evacuation: If we assume 100% capacity, but assume 5% of individuals self-evacuate, the required surge capacity to keep all patients in NYC facilities drops from 45% to 43% for hospitals and from 48% to 46% for nursing homes. A 10% self-evacuation rate drops it further to 41% and 43%, respectively.
• Using both capacity and self-evacuation adjustments, if we assume 90% capacity and 10% self-evacuation rate, then NYC hospitals and nursing homes would be able to accommodate all patients with 27% and 29% surge capacity, respectively.

Return to Contents
Proceed to Next Section