Nosocomial Tuberculosis Transmission from 2006 to 2018 in Hospital, China

Background: Strong evidence is lacking to support effectiveness of currently implemented tuberculosis control (TB-IPC) (TB) transmission. This 13-year analysis is the longest follow-up investigation to date to identify risk factors underlying nosocomial TB transmission. Methods: We monitored all staff of Beijing Chest Hospital each year from 2006 to 2018. Age, gender, duration, department, education, income, respirator, ultraviolet, and ventilation were chosen as variables. Univariate cox regression, correlation analysis, and multivariate cox regression were analyzed sequentially. Results: Using multivariable cox regression analysis, variables of income, ultraviolet germicidal irradiation (UVGI), natural ventilation and mechanical ventilation conferred significant protective effects, with odds ratios (ORs) of 0.499, 0.058, 0.003, and 0.015, respectively (P<0.05). Medical N95 respirator conferred an excellent protective effect, with an associated TB infection rate of 0%. Notably, inadequately maintained mechanical ventilation systems were less protective than natural ventilation systems. Conclusions: UVGI, adequate ventilation, and medical N95 respirator use may be protective factors against nosocomial TB transmission. Adequate ventilator system training, monitoring, and regular maintenance are critical for achieving effective TB-IPC.

There has been renewed interest in TB infection control, especially in areas with high TB and HIV prevalence [16,17]. Consequently, several TB infection prevention control (TB-IPC) measures have been proposed, although most are only realistically feasible in high-income countries [18], not in lowincome settings [19]. Meanwhile, other TB-IPC recommendations appropriate to any setting lack strong evidence of effectiveness [20]. Therefore, analysis of risk factors for nosocomial TB transmission is needed to guide best TB-IBC practices going forward, especially regarding use of expensive and cumbersome protective equipment, as recommended for controlling epidemic outbreaks (e.g., Covid-19 coronavirus).
Several previous studies have compared various types of masks or respirators for protective effectiveness against clinical respiratory illnesses. However, those studies did not specifically focus on TB infection, were largely cross-sectional, lacked long-term follow-up [21], and lacked strong evidence of efficacy. Therefore, this study was conducted as a first cohort study following all staff of Beijing Chest Hospital for a period of 13 years with the goal to determine risk factors underlying nosocomial TB transmission.

Methods
Beijing Chest Hospital is a TB diagnostic and treatment center in China. All hospital staff are at risk of exposure to TB patients and/or contaminated materials in spite of hospital utilization of air-cleaning equipment, including recommended ventilation systems and/or ultraviolet lamps. Therefore, as an added safety precaution all workers handling contaminated materials or in contact with TB patients are advised to wear masks.
From 2006 to 2018, physical examinations of hospital staff were performed every year. Case diagnosis was based on clinical signs and radiological, serological, and bacteriological investigations.
Nosocomial TB transmission was detected by a change in infection status from negative to positive, as detected using purified protein derivative (PPD) skin tests or interferon-gamma release assays (IGRAs) with or without positive sputum test results or detection of pulmonary TB lesions via computer radiography (CR) or computed tomography (CT).
To identify potential risk factors of nosocomial TB transmission, a two-step process was implemented as follows: (1) Summarize previous study results; (2)

Result
Each year new employees joined the hospital staff and older employees retired. Table 1 shows staff and TB-IPC characteristics in 2018, with distribution of variables for each category (age, sex, duration, education, income, respirator, ultraviolet, and ventilation) and staff proportions as shown.  Table 2. In the enhanced TB-IPC period, a significant reduction in TB infection incidence rate was observed relative to incidence during the early TB-IPC period, when only limited TB-IPC measures were available.
The prevalence of tuberculous disease was analyzed against variables of age, gender, duration, work department, workers' education, income, respirator, ultraviolet, and ventilation, with univariate cox regression results presented in Table 3. Age, duration, department, income, respirator, ultraviolet, and ventilation were selected as potential predictors of nosocomial TB transmission (P<0.1).
Bivariate correlations were analyzed using Pearson or Spearman tests (Supplement table 1). Age and duration were highly correlated, with a Pearson correlation value of 0.886 (P=0.000). To control the interaction between these two variables, only duration was chosen for the next multivariable analysis.
Correlations between ultraviolet and respirator and between ventilation and respirator were low (0.361 and 0.408, respectively, P=0.000). As ultraviolet and ventilation showed a mid-level correlation (Spearman correlation value of 0.539, P=0.000), they were both selected for multivariable cox regression. No associations were found between other pairs of variables.
Next, multivariable cox regression analyses including duration, income, respirator, ultraviolet, and ventilation were performed (table 4)

Discussion
Hospital staff members working in rooms occupied by TB patients were at significantly higher risk of developing active TB than were other employees [22]. To avoid hospital-based nosocomial TB infections, certain specific TB-IPC measures should be administered, while some previous recommended practices require justification due to weak or indirect evidence of effectiveness [20].
In our study, duration was not shown to be significantly correlated with TB nosocomial transmission (p = 0.061, OR: 0.978, 95%CI: 0.955 to 1.001), in opposition to results of previous studies showing that working less time in the hospital might lead to a lower rate of tuberculin skin test (TST) positivity [23]. To explain this discrepancy, here we postulated that elderly workers with long work experience and greater duration might implement practices that more effectively prevent them from contracting TB than practices used by new staff with less experience and lower accumulated duration.
Gender was previously thought to be a significant predictor of TB infection, as some studies indicated that men had higher TST positivity rates that were possibly due to a higher rate of community exposure among males [23]. By contrast, other studies have demonstrated that some specific female professional groups (nurses and laboratory technicians) had significantly higher rates of TST positivity, possibly reflecting the high proportion of females in the nursing field, a high-risk profession for TB infection. Nevertheless, in this study no significant correlation was observed between gender and nosocomial TB infection.
As previous studies have stressed the role of socioeconomic conditions in TB transmission among health care workers [24], here we evaluated income as a risk factor for TB infection and verified it to be a protective factor such that high-income correlated with lower TB incidence rate as compared to low-income status. In our hospital we note that high-income individuals may occupy more spacious working and living spaces and benefit from higher quality diets and healthier work and rest arrangements than those with low-income status, all of which may be relevant to TB prevention.
In theory, respirators may provide additional protection for health workers, who should use particulate respirators that prevent exposure to TB-causing mycobacteria released during high-risk aerosolgenerating procedures that tend to confer a high risk of TB transmission (e.g. bronchoscopy, intubation, sputum induction procedures, aspiration of respiratory secretions, and autopsy or lung surgery using high-speed devices). In addition, respirator use is especially necessary to protect those providing care to infectious multidrug resistant TB (MDR-TB) and extensively drug resistant TB (XDR-TB) patients or people suspected of having infectious MDR-TB and XDR-TB. However, few studies to date have investigated whether use of particulate respirators is of value to those providing routine TB patient care in situations where administrative and environmental controls are in place [20]. In fact, only weak and indirect accumulated evidence supports the use of particulate respirators for TB infection control, prompting us to conduct this first study to compare mask and respirator use for protection against TB over an extended time period (13 years). Here we demonstrated that use of a medical N95 respirator conferred excellent protection against TB infection, with 0% TB incidence observed with use of this device. The reason for the lack of significance of this result may lie in the small number of cases studied (case number = 0). However, as a limitation of correct N95 respirator use, extensive conditions must be met: 1. Requires training. 2. Requires adherence to best practices.
3. May adversely affect health worker's performance during some procedures. 4. Reduces comfort of health workers. 5. No clear guidance exists on how long the same respirator can be used over time.
Notably, an interesting observation of this study, that gauze and medical mask use might be a risk factor for TB infection, must be emphasized. This finding may reflect the fact that masks do not provide adequate protection and/or that workers wearing gauze and medical masks might engage in closer contact with infected patients or contaminated materials, with further investigation warranted to more fully understand mask-associated risk.
Priority should be given to achieving adequate air change per hour (ACH) values using ventilation systems, although this is not possible in some settings; for example, due to climatic conditions (e.g. in winter or during the night), in some building structures, or in situations where TB transmission would pose a high risk of morbidity and mortality (e.g. in MDR-TB wards). In such cases, a complementary measure incorporating use of upper-room or shielded UVGI devices may be an option, even though the effectiveness of such measures are only supported by weak evidence. For example, one epidemiologic study investigating TST conversion rates in health workers showed no major additional benefit; however, another well-designed animal model study (using guinea pigs) demonstrated that upper-room UVGI could reduce TB infection [20]. Here, air UVGI use was verified to be protective against TB infection, while ultraviolet use only for bed sterilization showed no significant protection. This discrepancy my reflect the fact that ultraviolet irradiation-based measures should be capable of achieving air disinfection equivalent to 10-20 ACH if the air UVGI system has been appropriately designed, installed, maintained, and operated. While ultraviolet irradiation measures are generally suitable for most climates, UVGI use has several weaknesses: 1. Requires expertise in design, installation and testing. 2. Requires maintenance and cleaning (not effective if not well maintained).
3. Requires air mixing to be effective, but no easy-to-use tool exists for measuring equivalent ACH. 4.
Direct UVGI exposure or overexposure results in non-permanent adverse effects (photokeratitis and erythema), while upper UVGI devices are potentially hazardous if improperly designed or installed. In well-designed systems, the principal hazard is inadvertent eye exposure by workers climbing up into the high-UV zone to perform painting, cleaning, and maintenance tasks. As with any engineering control, use of a UVGI device requires proper design, installation, operation, and maintenance [20].
Importantly, UVGI devices do not replace ventilation systems; rather, they should be considered a complementary intervention. However, adequate ventilation in health-care facilities is essential for preventing transmission of airborne infections and is strongly recommended for controlling TB transmission, although such recommendations are based on low-quality indirect evidence lacking quantitative demonstration of impact of adequate ventilation on TB transmission. In our study, two ventilation systems were shown to be protective against TB infection, with mechanical ventilation less protective than natural ventilation. Consequently, the mechanical ventilation system used in our hospital was tested carefully and subsequent results indicated that the filters inside the system needed replacement. Thus, our result aligns with results of previous studies showing that health-care facilities relying on natural ventilation alone could maintain effective ventilation through proper system operation and regular maintenance. For example, simple natural ventilation may be optimized by maximizing the size of window openings and by locating windows on opposing walls; indeed, such systems are capable of achieving ACH above the required minimum of 12 ACH to ensure accelerated decay of droplet nuclei. However, natural ventilation has weaknesses: 1. Difficult to control (depends on wind and temperature). 2. No control over direction of air flow. 3. Sufficient permanent openings (e.g. windows and vents) should be guaranteed to maintain adequate ACH. 4. No easy-to-use tool for measuring ACH. 5. Limited applicability (only suitable in a few locations globally). Therefore, when natural ventilation alone cannot provide sufficient ventilation rates, well-designed, maintained, and operated fans (mixed-mode ventilation) can help maintain adequate air dilution. The threshold for ventilation requirements may vary according to the type of ventilation (e.g. recirculated air versus fresh air). In choosing a ventilation system (i.e. natural, mixed-mode, or mechanical) for health-care facilities, it is important to consider local conditions, such as building structure, climate, regulations, culture, cost, and outdoor air quality. Furthermore, any ventilation system must be monitored and maintained on a regular schedule, with adequate resources (budget and staffing) dedicated to proper maintenance. Indeed, the reason that the mechanical ventilation system in our hospital did not play an effective TB-IPC role reflects the fact that mechanical ventilation without adequate maintenance was less protective than natural ventilation. Our results align with other reports of TB infection in health facilities with faulty or no ventilation systems [20], which also require routine maintenance.
In our hospital, we note that several TB cases were detected in non-TB exposed departments, possibly due to limited TB-IPC measures taken in those hospital spaces. For example, roentgenography rooms are often poorly ventilated spaces, where droplet nuclei may remain suspended in the air for hours or even days [20]. Therefore, preventive measures such as adequate ventilation and ultraviolet-based germicidal control may also be needed in non-TB exposed environments, especially in infectious disease hospitals.

Conclusion
In summary, several variables showed protective effect on nosocomial TB control, such as UVGI, adequate ventilation, and use of a medical N95 respirator, and may be recommended as administrative controls. Adequate ventilation system training, monitoring, and regular maintenance may be critical for TB-IPC. In congregate settings with a high risk of TB transmission, multiple factors should be comprehensively considered to maintain TB-IPC.

Declarations
Ethics Approval and Consent this research has been granted by Beijing Chest Hospital affiliated to Capital Medical University Ethics Committee and complied with Ethical Standards, and informed consent was approved by every respondent.