Building design and engineering for TB Transmission Control – Two hospitals in Haiti as examples for discussion.
Edward A. Nardell, MD, DSMHI, BWH, PIH, Boston
As you know last week I visited Haiti to look at 3 hospitals (one was for GHESKIO) from a TB transmission control perspective. Eventually I hope to get to our other sites in Rwanda, Lesotho, and Peru, although I have been to several hospitals in Lima early on.
Based on those observations I wanted to begin a dialogue about hospital design for TB transmission control that may lead to some general principles, best practices, and possibly research questions. As some of you know, HIGH has expressed an interest in this and laboratory design issues at clinical/research sites, and a first meeting of a group to discuss these issues is scheduled for Tues, Feb 12, at noon (FXB I believe). In addition, a HSPH course on this subject, generalized to engineering methods to prevent transmission of airborne infections, is planned for July, 2008.
To start the ball rolling I will make some observations about two of our hospitals in Haiti and use them as a focus for formulating questions and points for discussion, consensus, or disagreement:
Example A: Zame Lasante, Cange
First let me share the innovative triage scheme that Paul Farmer implemented to use the availablespace. This 3-compartment scheme is not perfect, but it is much better than what most hospitals in high burden areas are doing, based on my experience.
Most TB patients are treated with DOT in the community, but when hospitalization is needed, there are 3 possible sites within the Cange facility:
1. All smear negative patients, regardless of drug susceptibility and HIV status, can go to the general medical ward under the theory that they are not as likely to transmit as much as sputum sm + patients.
2. All smear positive/HIV-neg patients, regardless of drug susceptibility pattern, can go to the Tom White TB Pavilion, featuring increased natural ventilation and upper room germicidal irradiation (UVGI) fixtures and room paddle fans.
3. All smear positive/HIV-pos patients go to one of the 6 individual isolation rooms with window exhaust ventilation to produce directional airflow into the room from the corridor and UVGI and room paddle fan (I think there were paddle fans).
The limitations of this scheme are that smear-neg. patients can be infectious, especially for HIV-infected persons, and MDR smear positive patients can re-infect smear positive non-MDR patients, even though HIV-neg. Still, it is much better than what is done in most hospitals in resource-limited settings.
Comments on the general medical ward:
1. Like many good primary care and referral hospitals, this one is overcrowded. There were patients on mattresses on the floor the day I was there. In many African hospitals, 2 patients to a bed are not uncommon.
2. Windows are open, but there are fewer on the general medical ward and fewer UVGI fixtures. The paddle fan shown in the photo was not on, probably for comfort only, not as part of a UVGI/air mixing system as in the Tom White Pavilion.
3. A study in Lima showed that about 14% of routine admissions to a general medical ward for a variety of other conditions had unsuspected TB, and many were smear positive, some were MDR. That is in a low HIV setting.A surveillance study for TB prevalence on a wardin Haiti, Lesotho, or Rwanda, would be very useful. According to Bill Pape, 30% of HIV-pos patients with a cough seeking VTChave TB. Thedanger is that HIV patients must also be admitted to general medicine and may be infected by known, smear neg. TB cases, or by unsuspected cases.Increased surveillance for TB is one issue (cough officers, etc.). The same extranatural ventilation and UVGI/paddle fans used on the TB ward should be applied on general medical wards. Less crowding (obviously) would help.
Comments on the Tom White TB Pavilion:
1. This building was designed and built 10 years ago with TB transmission control in mind. It has fenestrated brickwork at the top of the high ceilings for extra ventilation, in addition to large open windows. UVGI is deployed along with ceilingpaddle fans.There are 2 male wards and 2 female wards as well as a room for patients who are disruptive, known as the Cycloserine room, given the potential for that drug to precipitate psychotic behavior.
2. One enters the second floor to a kind-of lobby with bathrooms off of it. It is a generous space, but empty of patients, whereas the patient beds are quite tightly packed in the 2 male and 2 female wards. There is just enough room to walk between the beds. I should have written down the number of beds and the floor areas, but did not. Perhaps we can get that information. I suspect that the pavilion space could be utilized more uniformly, ie, using the unused lobby area for patients, to allow more even distribution. Perhaps there are logistical reasons why that has not been done.
3. UVGI fixtures were too few for the area/volume of patient care area. The gross rule of thumb has been 1 30 W fixture for every 200sq ft. I would estimate (without measuring) that we were half to 1/4 of that rule in the pavilion. There was one fixture in the large 2nd floor entrance which was doing no good, and another in on the first floor as you enter that was burned out. Lamps must be changed on an annual schedule and records kept in a prominent place to be sure this is done. Lamps will appear blue after the end of their useful life, but put out no useful germicidal UV.
4. The danger here is re-infection of drug susceptible TB patients with drug resistant TB strains, even among these non-HIV infected persons. Transmission to staff is also an issue.
Comments on the 6 isolation rooms:
1. These rooms are off a separate single corridor from the general medical ward. Each has a powerful kitchen exhaust fan through the outside wall. The fans are fairly noisy. Exterior fins or louvers on the fan are fully open when the door to the room is open, indicating high flow rates, but they drop down and the fan strains when the door is closed, since there is insufficient air intake. This means that air changes per hour (ACH) is greatly compromised when the door is closed. The door is difficult to close, demonstrating strong directional airflow into the room from the corridor. However, the doors need to be "undercut" by an inch or so to allow better airflow under the door when closed.
2. Having strong directional airflow into each room off the same corridor means that rooms are competing for the room air in the corridor. If one fan is not working, for example, as mentioned below, the other rooms are sucking that room air into the corridor and into the other rooms. The solution is to not have central corridors with neg pressure isolation rooms off of them. Much better is the arrangement in the new Lacolline facility where the entrance to the isolation rooms is essentially outside (it is a covered breezeway) physically separate from the rest of the building. An open window or large vent in that corridor will allow rooms to draw air from outside, not from other isolation rooms.
3. In one isolation room, the window was opened, the exhaust fan was working freely with its external fins wide open, but there wasno negative pressure, the door closed easily, and there was no perceptible directional pressure into the room. I did not do a smoke test, however. Contaminated air from that room could be sucked out into other rooms by the 5 other operative fans.
4. A UVGI lamp and center paddle mixing fan was operative in each room. I believe the fan was blowing downward - more about that later. Note that the UV lamps are not uniformly centered on the wall for optimal UV distribution (see photo).
In summary, ZL Cange has an excellent triage plan and an adequate facility, but it could be improved by less crowding, possibly by investigating the prevalence of unsuspected TB on the general medical ward, by increasing the number of UVGI lamps, and being sure that isolation room windows are closed and the fans operative to achieve directional airflow. Installing UVGI in the corridor leading to the 6 isolation rooms would also make sense.
Example B: the hospital at Lacolline – just constructed and dedicated
This facility was designed by the Haiti team with TB transmission control in mind. The corridors are large and airy, with Haitian decorative iron grills instead of windows at their end. Natural ventilation is increased further by an open courtyard in the center of the hospital. The male, female, and pediatric wards have fewer beds per area than Cange, but presently unoccupied, so it remains to be seen what bed supply and demand situation will evolve. The windows are not the louvered type, but more modern sliding type. The wards are equipped with UV lamps and multiple paddle mixing fans. The fans are single speed (high) and blow downward.
The 4 isolation rooms are off a covered breezeway, not a shared closed corridor, making it unlikely that rooms will compete for the same air as in Cange, with the possibility of cross contamination. They also have a wall exhaust fan, window that can be closed, and a UVGI lamp, which again is not centered in the room. My photo of one of the rooms does not show a paddle mixing fan in the isolation room.
The ambulatory waiting area is a combination of inside and outside (to be vine covered) seats. For a very large space, there are now just 2 UVGI fixtures, and multiple mixing fans.
Comments: This new facility was thoughtfully and beautifully designed with TB transmission control in mind. Of course, I have a few suggestions:
1. The windows are relatively few compared to Cange and other hospitals I have seen and they have sliding windows which effectively limits the opening to half the total area of the window. Louvered windows as in Cange have two benefits by comparison – they open the complete window area, and they provide some protection from rain entering during the rainy season. The sliding windows are more likely to be closed when it rains to keep rain out. The sliding windows are perfect for the isolation rooms where they will be closed.
2. The UVGI fixtures are too few for most of the areas where they are being used, not using the 30W per 200 sq ft rule of thumb. We are working on better application software designed to achieve a minimal average UV in the upper room, but it is not yet ready for use. We concluded the waiting room, for example, could use 3 or 4 more fixtures. Where they are used they should be positioned so as to give as even distribution in the upper room as possible. The best position for a single fixture is usually the short wall so that ray length is maximized. When 2 or more fixtures are used they are staggered to give uniform upper room distribution.
3. The paddle fans appear to be one speed (high) resulting in excessive noise and considerable down-draft which sick patients will not find comfortable. Room air mixing is as effectively done with an up-drawing fan which is generally better tolerated. Patient discomfort often results in the fans being turned off, defeating their purpose. Generally, variable speed reversible fans are ideal, but more expensive. Have your engineer look at the possibility of running all the fans on a rheostat and find out if they can be reversed.
4. The same precautions about being sure the isolation room windows are closed so that the extraction fan works to produce directional airflow applies, except that their doors being off a breezeway makes that directional airflow less pressing.
General design issues prompted by these two examples:
1. General medical wards.
a. Is there an optimal bed to floor area or bed to room volume ratio? Is this by local code, or by more general guidelines?
b. Is there a optimal window area to room volume ratio? Should certain types of windows (louvered) be generally used over sliders?
c. Can we achieve even better ventilation using older fenestrated brick work as in this older building at the GHESKIO Siguenau Hospital?
d. Should we be assessing numbers of room air changes by CO2 dilution (as Rod Escombe has done in Peru) on a number of days as a guide to design under local climatic conditions? Clearly none of this applies to Russia, for example, but Lacolline is probably different than Cange and our sites in Lesotho and Rwanda, and they will vary by season and day, depending on the weather.
e. How much can we depend on natural ventilation? As noted above, it depends.
f. Should we be depending on UVGI? There is laboratory evidence that efficacy drops dramatically at RH > 70%. There is no field evidence. If we use it, should we shoot for more UV than produced by the Lumalier fixtures? Atlantic UV fixtures put out a lot more UV per fixture, but you must buy the lamps from them.
g. How much paddle fan downdraft or updraft is tolerable, and how much fan noise is tolerable? Can we determine paddle fan specifications?
2. Isolation rooms.
a. Can we arrive at specifications for room size and exhaust fan capacity? What noise level is tolerable?
b. How much airflow should occur with the door closed? Other than a smoke test to determine direction, how can we judge how much airflow? CO2 decay?
c. Should we identify an inexpensive CO2 meter and develop a simple protocol using a fire extinguisher as a CO2 source – for use in estimating outdoor air ventilation in resource limited settings?
d. Should isolation room windows be locked so that they cannot be opened, defeating the purpose of the exhaust fan?