Introduction
TB transmission in healthcare facilities is an important public health problem. Occupational TB has been widely documented in the developed world, and is increasingly recognised in low and middle income countries [
1,
2]. In addition to staff, patients and visitors to healthcare facilities are at risk of infection. TB transmission in other institutional settings such as prisons, homeless shelters and schools is also important [
3‐
5].
The dual epidemics of HIV infection and drug-resistant TB conspire to compound the problem of institutional TB transmission. HIV patients latently-infected with TB are more likely to develop active TB disease [
6], and thus cause onward transmission via the airborne route. Indeed in high TB burden countries, a high proportion of HIV cases themselves present with active TB, forming a significant pool of TB infection within waiting areas of HIV clinics. Furthermore, HIV infection increases hospitalisation and attendance at healthcare facilities. The roll out of HIV care may inadvertently promote the risk of airborne TB transmission through the congregation of highly susceptible patients with a high incidence of TB in settings which are often overcrowded, such as anti-retroviral treatment centres [
7]. TB drug resistance may increase overall TB transmission in a number of ways, including diagnostic delay and prolonged periods of treatment. There may be frequent attendances at healthcare facilities whilst drug-resistant TB patients are treated unsuccessfully with first-line anti-TB drugs prior to correct diagnosis. Such inadequately treated patients may be highly infectious [
8]. Nosocomial transmission has been strongly implicated in the emergence of extensively drug resistant TB in South Africa in health care settings where airborne infection control was poor or absent [
9,
10].
Airborne TB transmission depends on a number of factors, including source strength (for example cough frequency of an undiagnosed TB patient) and the number of susceptible persons exposed [
11]. A major determinant of transmission is room ventilation with fresh air, which serves to dilute the concentration of airborne infectious particles. Room ventilation may be provided by mechanical ventilation systems that may also deliver negative pressure. Such systems require specific expertise to design and are expensive to install and maintain, and are inappropriate for many low resource settings where the burden of TB is highest. Natural ventilation by simply opening the windows may provide higher rates of air exchange for little or no cost, but is climate dependant [
12]. TB infection control guidelines recommend 6–12 air-changes/hour (ACH) room ventilation for high risk settings [
13,
14]. Even if such environmental control measures are implemented, they are often limited to areas considered high risk, such as TB wards. However it is untreated TB patients prior to diagnosis, or inadequately treated drug-resistant TB patients, who are likely to be the most infectious [
15]. These patients are commonly found in areas such as emergency rooms, waiting rooms, out-patient clinics, and X-ray departments. These areas are often overcrowded, especially in low resource settings, and are often not the focus of TB infection control efforts.
We studied the effect of simple architectural modifications to existing infrastructure to improve natural ventilation in out-patient consulting rooms and waiting rooms in hospitals in a high TB burden setting, and used mathematical modelling to estimate reductions in the risk of TB transmission to patients and healthcare workers.
Discussion
This study has demonstrated that simple, low cost modifications to existing infrastructure can greatly improve natural ventilation in healthcare settings, in this case in hospital consulting rooms and waiting rooms. Such rooms are likely to contain infectious TB patients and the increased ventilation considerably reduced TB transmission risk to staff and other patients calculated using a standard airborne infection model. These interventions to improve natural ventilation within existing infrastructure were achieved at minimal or modest cost. These findings demonstrate the potential for simple implementation of effective environmental control measures for preventing TB transmission in healthcare settings, which is of particular importance in the context of the roll out of HIV care.
A number of guidelines exist for preventing TB transmission in healthcare facilities, including WHO guidance for low resource settings [
13,
14]. TB infection control involves administrative measures to ensure the prompt diagnosis, isolation and initiation of effective treatment of TB patients; environmental control measures to reduce the airborne concentration of infectious droplets; and personal respiratory protection. TB infection control has often been neglected, especially in low resource settings, and is frequently limited to small areas of healthcare facilities perceived to be at highest risk, such as respiratory isolation rooms. However, it is TB patients who are untreated (prior to diagnosis) or inadequately treated (due to delayed diagnosis of drug-resistant TB) who are likely to be the most infectious [
8,
15,
18,
19]. These types of patients are likely to be found in waiting rooms, out-patient clinics, X-ray departments, and emergency rooms. In a Canadian study, 47% of 250 TB patients made 258 visits to the emergency department prior to diagnosis, and 95% of source case nosocomial infectiousness time occurred in the emergency department [
20]. In this study, new TB patients were documented in overcrowded waiting rooms, where average waiting times were around 3 h. It is likely that these poorly ventilated, overcrowded areas are responsible for considerable TB transmission, as suggested by the airborne infection model.
Administrative control measures such as triage of coughing patients or use of screening algorithms in emergency departments are undoubtedly useful, but their effectiveness is limited [
21,
22]. In a study from Peru, up to 31% of TB patients attending an emergency department had entirely unsuspected TB, and would not have been identified by screening protocols [
23]. A modelling study of the emergence of extensively drug-resistant (XDR) TB in KwazuluNatal predicted that administrative control measures alone would avert just 10% of new XDR-TB cases, compared with 33% if natural ventilation were implemented [
24]. Thus environmental control measures such as room ventilation or upper-room ultraviolet light [
25] that protect against both the expected and the unexpected TB case across healthcare facilities therefore assume increased importance. Mechanical ventilation is expensive, and requires specific expertise in design, installation, and maintenance. Poor maintenance of mechanical ventilation, which may lead to serious adverse consequences such as positive instead of negative pressure, has been widely documented in developed countries, and associated with nosocomial TB outbreaks [
26‐
31]. Maintenance of such systems is even harder in low resource settings, where budget and parts for annual maintenance may be unavailable. Furthermore, high air exchange mechanical ventilation is limited by its cost to high-risk areas such as respiratory isolation rooms. Areas such as waiting rooms or consulting rooms are often ventilated at much lower rates, reflecting odour and heat considerations rather than control of airborne infection. In contrast to mechanical ventilation, natural ventilation, where climate permits, is applicable across most parts of healthcare facilities, including large, overcrowded waiting rooms such as those in this study [
12]. It is difficult to provide consistent directional airflow with natural ventilation, however where there are prevailing winds (as is the case in this study in Lima which is a coastal city) seating arrangement may be adjusted for additional safety. In the two consulting rooms in this study, for example, the chairs for the consulting doctors were located adjacent to the windows open to prevailing breezes, with patients being consulted ‘downwind’ across a desk.
This study has shown that even simple modifications to existing infrastructure can facilitate high rates of natural ventilation. This is important because in many healthcare facilities resources may not be available for new buildings, or logistical difficulties may hinder relocation of a particular hospital service to a more appropriate site. Much higher rates of ventilation may be achieved when infrastructure is specifically designed to facilitate natural ventilation, such as the new respiratory out-patients waiting room in this study, where 54–68 ACH were measured. It is therefore important that TB infection control is prioritized at the design stage when planning new healthcare facilities, allowing optimization of environmental control measures such as natural ventilation, with carefully planned layout of services and direction of patient flow to minimise time spent in overcrowded areas by potentially infectious patients. For example, the siting of a waiting area outside where climate permits is likely to have a beneficial effect on airborne disease transmission control [
32]. These architectural design and patient flow considerations are especially important in the roll out of HIV care, where highly susceptible persons are required to spend often long periods in very overcrowded facilities, sharing airspace with those who are likely to have active TB disease. There exists considerable potential for funding agencies to require that TB infection control be an integral part of TB and HIV control initiatives where new building or remodelling of existing infrastructure is planned. This need is of great urgency when considering the number of new healthcare facilities already under construction around the world as HIV treatment is scaled-up.
Limitations of this study include the assumptions inherent in the airborne infection model, and lack of precision in estimating the prevalence of new TB cases in waiting and consulting rooms owing to the small amount of data collected. However, we believe that the estimates of TB risk are conservative, because only new TB cases were considered to be infectious, and no allowance was made for the possibility of drug resistant TB, where those with undiagnosed drug resistant TB being treated with first line therapy would still be infectious [
8]. However, it is the reduction in TB risk that is the main outcome measure of interest in this study. When the only variable changing in the model pre- and post- intervention is room ventilation, the absolute numbers of infectors, or the exact time spent in the waiting room, have almost no influence on the percentage risk reduction in TB transmission observed. Another limitation is that relatively small numbers of ventilation measurements were made in each room, but this is mitigated by the magnitude of the increases in ventilation following the intervention observed consistently in each room. The carbon dioxide tracer gas technique has limitations, especially in not being able to detect variations in ventilation rates around a room if only one sensor is used [
33], and through the fact that carbon dioxide is produced by room occupants, in contrast to other tracer gases such as SF
6. However carbon dioxide use has the advantages of low cost, practicality, and safety considerations for occupied buildings. Most importantly, any limitations in the technique used could be assumed to apply equally to the pre- and post-intervention measurements, minimising impact on the outcome of interest, the reduction of TB transmission risk.