Infection control and tuberculosis among health care workers in Viet Nam, 2009-2013: a cross-sectional survey

We aimed to assess time trends in the annual notification rates (2009-2013) of TB among HCW in these provincial and national facilities. Also, we explored potential associations between TB notification rates and different TBIC indicators reported.

To get better insight in the burden of TB among HCW employed in public TB facilities connected to the National TB Program (NTP) in Vietnam, NTP has developed four main TBIC indicators:

Based on these indicators, a short questionnaire was designed in English and translated in Vietnamese. It included additional questions on the characteristics of the TB patients among HCW, the duration of validity of the TBIC plan, the presence of a TBIC focal person and provision of personal protective equipment (PPE) to HCW working in the culture and/or DST section of the microbiology department and HCW serving in MDR-TB wards. Annually, between 2010 and 2013, the form was sent to all national and provincial TB facilities, requesting the administration department of the hospital to complete the form. Information was collected on the period 2009-2013 by the TBIC focal person, who was usually part of the infection control committee overseeing implementation of IC policies in the hospital. Minor changes were made in the form over the years (see Results section).

In Vietnam, Culture/DST rooms in the TB laboratories all use negative pressure. MDR-TB wards in principle use natural ventilation, but if there are less than 12 air changes per hour, natural ventilation is combined with an exhaust fan and ultraviolet germicidal irradiation (UVGI).

A HCW was defined as anyone occurring on the payroll of the health facility in the specific year of reporting. This includes staff working in any clinical department, in paramedical sections (such as pharmacy), laboratories, and administrative and other support sections of the health facility.

TB was defined as active TB, and includes pulmonary (new and retreatment) and extra-pulmonary TB. Since reporting was anonymous we could not check whether these patients were notified and/or bacteriologically confirmed, and included any patient reported by the facility.

A TBIC plan is a written protocol for the prompt recognition, separation, provision of services, investigation for TB and referral of patients with suspected or confirmed TB disease in the health facility [12]. The TBIC plan was considered valid if the validity period included the reporting period. Since the TBIC plans were not available to the researchers, no content checks were conducted.

A TBIC focal person is the person in the health facility who is responsible for coordination of all TBIC activities in the health facility.

The project was approved by the Research Board of the National Lung Hospital in Hanoi. Only data that had been routinely collected by each health facility were requested. Data on HCW were collected in an aggregated manner. No personal information about HCW was requested. Therefore, no personal informed consent was obtained. All health facilities received a code before analysis. All data were analyzed in such a way that the information presented does not provide sufficient information for the identification of persons.

Notification rates were calculated by dividing the summed number of HCW with TB disease in a specific year by the summed number of HCW working in the health facilities in the same year. It was assumed that each HCW worked in the health facility for a full year. The rates were multiplied by 100,000 to obtain a notification rate per 100,000 HCW per year. Exact binomial 95 % confidence intervals (CIs) were calculated around each annual rate. Time trends were assessed applying random effects negative binomial regression analysis on the annual notification rates between 2009 and 2013 after specifying the data as panel (time series) data.

Health facilities were grouped to assess health facility characteristics as follows: (national) referral centers for TB (n = 4; including the three national TB hospitals and one provincial TB hospital acting as a tertiary referral center), provincial TB hospitals (n = 43; including all but one provincial hospitals for TB and lung diseases), and provincial preventive centers (n = 20; including 18 preventive medicine center for social diseases control and prevention and two TB stations).

Multivariate regression models were applied to assess the association between the health facility specific 5-year notification rate and health facility characteristics. Because of the (bimodal) distribution characteristics of the 5-year TB notification rates (see Fig. 1), different multivariate regression models were applied. First, a backward stepwise logistic regression model was built in which the dependent variable was either 1 (any TB cases reported between 2009 and 2013) or 0 (no TB cases at all between 2009 and 2013) to account for the 42 % of health facilities reporting no cases between 2009 and 2013. All variables for which the p-value was less than 0.2 were kept in the model. To assess the associations between notification rates and facility characteristics in those facilities from which cases were reported, the TB notification rate was first log-transformed and this transformed notification rate was analyzed using simple linear regression models. Here, backward stepwise linear regression was applied leaving all variables for which the p-value was less than 0.2 in the model. Time trends were calculated in- and excluding health facilities that did not provide data for all of the years 2009-2013. A p-value smaller than 0.05 was regarded as statistically significant.

The number of health facilities responding to the survey varied between 57/67 (85 %) in 2012 and 2013 and 64/67 (96 %) in 2011 (Table 1). The number of health facilities reporting at least one TB case for the respective year of reporting decreased between 2009 and 2013: from 18/61 (30 %) to 9/54 (17 %) (p = 0.02).

The notification rate of all types of TB dropped from 559 per 100,000 HCW per year in 2009 to 338 per 100,000 HCW per year in 2013 (Fig. 1, left panel).

A time trend in TB notification rates was calculated using data from 47 facilities that reported data for all five years. In these facilities, the notification rates dropped significantly from 593 (95%CI, 388-869) per 100,000 HCW years in 2009 to 197 (95%CI, 98-352) per 100,000 HCW years in 2013, (p = 0.02; beta-coefficient -0,2, 95%CI, -0.36 to -0.03; Fig. 1, right panel). We repeated this analysis excluding four health facilities with a reported increase of >100 % in health staff within one year. This did not change the results: the decrease remained statistically significant (p = 0.04; data not shown).

Overall, 56 % of the reported TB cases were registered new pulmonary TB cases, 6 % were previously treated pulmonary TB cases, and 38 % were registered with extra-pulmonary TB. The distribution remained more or less stable over the years.

Of a total of 104 (95 %) TB cases for whom the working position was recorded, 30 (29 %) were reported among staff employed at TB wards, 24 (23 %) among staff from other (non-TB) clinical wards, 10 (10 %) among microbiology laboratory staff, six (6 %) among staff working at the MDR-TB ward, and 34 (33 %) among staff working in other positions. It remains unknown how the number of TB cases reported relates to the number of staff members per working position, as staff counts per working position were not included in the data collection form for feasibility reasons.

In the period of 2009-2011, 48 of 64 health facilities (75 %) reported having a TBIC plan available; this plan was reported to be available by 45 of 56 facilities (80 %) reporting over the period in 2012-2013 (Table 2).

It is not known which TBIC plans had been based on (inter)national guidelines, and how many of these had been made with input from TBIC experts.

The proportion of facilities reporting to have an appointed TBIC focal person was 83 % (52/63 facilities) for 2009-2011 and 84 % (47/56 facilities) for 2012-2013, and 73 % taking only the 52 health facilities into account that reported on this indicator for both periods (Table 2).

N95 respirators were provided in the majority of microbiology laboratories providing culture (with or without DST) and health facilities with an MDR-TB ward (Table 2). The proportion of such facilities having PPE available for their staff increased between 2009-2011 and 2012-2013 (from 61 % to 90 %, p = 0.02 for culture/DST laboratories, and from 67 % to 100 %, p = 0.01 for MDR-TB wards), assuming those reporting are representative for all facilities with an MDR-TB ward and/or a culture/DST laboratory.

Only six health facilities had not implemented any TBIC measures over the reporting period; these include one TB unit in a provincial hospital and five preventive medicine centers. Together, these facilities reported only one extra-pulmonary TB case in five years’ time. Having only limited data on TBIC measures available, we did not find an effect of any improvements in TBIC measures (defined as newly implementing a TBIC plan or appointing a TBIC person or starting the use of PPE in MDR-TB wards and culture laboratory sections) on the TB notification rates (data not shown). Among the facilities with MDR-TB wards and/or culture sections, the 5-year notification rates seemed lower in those with reported provision of N95 respirators (354/100,000 HCW-years, 95 % CI 277-445) than in those without (448/100,000 HCW-years, 95 % CI 180-921), but this difference was not statistically significant.

The notification rates were lower in those facilities that had no TBIC focal person compared to facilities with a TBIC focal person (Kruskal-Wallis ranksum test p-value, p = 0.08 for 2009-2011 and p = 0.04 for 2012-2013).

Table 3 presents notification rates by facility type and geographical zone (all reports of the health facilities received from 2009-2013 included). The notification rate was near-to-significantly higher in the referral centers compared to provincial health facilities (p = 0.06), with a conditional maximum likelihood estimate (CMLE) of the rate ratio of 1.5 (95 % CI 0.99-2.2). The TB notification rate in Southern health facilities was higher than in Northern, and statistically significantly higher than in Central facilities (p = 0.002; CMLE rate ratio 4.9, 95 % CI 1.7-20). The notification rate in the North was also higher than in the Central zone (p = 0.02; CMLE rate ratio 3.7, 95 % CI 1.3-14.7) (Table 3).

Among the provincial TB hospitals, there was a significant inverse linear association between the number of HCW employed and 5-year TB notification rate (p = 0.001; β-coefficient = 0.8, standard error = 0.2; Fig. 2).

Facilities with an appointed IC focal person were 7.6 times more likely to report any TB cases than facilities that had no focal person (Table 4). Compared to health facilities in Central Vietnam, facilities in North Vietnam were 10 times more likely to report a TB case (Table 4).

Over the reporting period (2009-2013), the rate of all types of TB reported dropped from 559 per to 338 per 100,000 HCW-years. Including only the 47 health facilities reporting for five consecutive years, the notification rates dropped from 593 to 197 per 100,000 HCW years (test for trend, p = 0.02). While this is a marked decline, in 2013 the notification rate among HCW was still almost twice as high as in the general population (which was 111/100,000 in 2013 [5]). This may in part reflect better access to screening and awareness about the occupational risk of TB transmission, but will also reflect the higher risk of getting TB among HCW, which is consistently reported in almost all published literature on this topic [1, 2, 4, 13]. The most recent systematic review reported a relative risk (RR) of active TB disease for HCW compared to the general population of 3.7 (95 % CI 2.9-4.5) with RRs ranging from 1.2 to 14.7 depending on the setting [1]. It should be noted however that these RRs were not corrected for age and sex, since this information is often not available. Between 2009 and 2013, there was a statistically significant decline (p = 0.04) in the TB notification rates among staff working in TB laboratories. This decline may be the result of increased protection and awareness of staff, the availability of biosafety cabinets and improved laboratory practices as a result of continued training activities. Though a decline was also seen among staff working in MDR-TB departments, this was much weaker and failed to reach statistical significance. Notably, the proportion of health facilities providing PPE to staff increased from 61 % in 2011 to 79 % in 2013 for staff in MDR-TB wards (p = 0.17) and from 67 % to 92 % (p = 0.048) for staff in the culture/DST section of the microbiology laboratories. It should be noted that the number of TB patients found in non-TB related departments remained relatively high, and this is in line with results found by an earlier review [2]. Also it should be noted that the available data (all age/sex classes) shows no decline over time in notification rates among the general population during the study period.

Only a few studies have measured the effect of TBIC measures on the TB burden among HCW [7, 8, 10, 14‐16] and from three of these studies, a decrease in latent TB infection was reported, but not a decrease in TB disease. One cross-sectional study only reported about TB disease [10]. A weak association was found between reported IC measures (especially environmental measures) and TB disease incidence, but it disappeared in the multivariable analysis. About 75 % of health facilities included in our report had a TBIC plan and an appointed TBIC focal person. These proportions remained stable over time. Presence of a TBIC focal person in the facility generally was associated with a higher probability of reporting any TB cases (adjusted Odds Ratio = 7.6 95 % CI 1.2-47.7). Similarly, presence of a TBIC plan was associated with finding one or more TB-cases among staff members. This may be the effect of less active case finding and reporting from facilities without a TBIC focal person as compared to facilities with a TBIC focal person. The alternative explanation may be that those facilities with more TB cases among HCW may be encouraged to appoint a TBIC focal person (reverse causality). The information that we collected was probably not specific enough to detect any effect of TBIC measures. Further, confidence intervals are wide since we conducted ecological analysis, with one data point for each health facility. Presence of a TBIC plan alone likely does not contribute to reduced transmission of TB or increased case finding within a facility. This information was not available from the routine reports unfortunately and additional data should be collected about TBIC practices.

Although there is evidence that implementation of IC measures reduces the burden of TB [2, 4], few studies evaluated the impact of TBIC measures in low and middle income countries (LMIC). One study found no effect on TB disease incidence [14] and two showed a reduction in tuberculin skin test conversion rates in HCW [7, 8, 15]. In LMIC, even low-cost strategies are rarely implemented [3, 15, 17] though recently IC has gained more attention and more is being done to protect HCW from TB infection and disease.

Provincial TB hospitals reported lower TB notification rates (329/100,000 HCW-years) when compared to other types of facilities (notably national/regional hospitals with a notification rate of 484/100,000 HCW-years). This may be explained by presence of more difficult-to-treat TB patients and conduct of higher-risk procedures in the larger hospitals, such as bronchoalveolar lavage and pneumectomy, and culture and DST in laboratories, but also by better awareness among HCW about the risks of TB transmission leading to better adherence to annual screening procedures, and more resources facilitating reporting of TB cases. However, more data are needed to learn which of these explanations contributes most.

The probability of reporting any TB cases was lower in Central than in North Vietnam (where two of the three regional and the national TB hospital are located). This association may reflect the underlying TB burden in the general population. The TB prevalence survey found a lower prevalence in Central Vietnam as compared to North and South Vietnam (p < 0.05) [18], and the data also match the overall TB notification data of NTP [19].

The data presented in this report was collected routinely from all provincial and supra-provincial facilities reporting to the NTP in Vietnam. To our knowledge, this information is rather unique and Vietnam is one of the few countries in the region collecting any indicator data on TBIC practices and on TB cases among HCW. However, routine collection of data on just a few indicators understandably has its limitations as much less information can be collected than is usually done in research settings.

First of all, this concerns indicators aggregated by health facility and data can thus not be linked to individual characteristics such as sex, age, working experience, and job type as has been done by others [10]. Also, comparison with the notification rates in the general population was impossible since the age and sex composition of our study population was unknown. Local laws prescribe aggregate data collection of TB among HCW. Moreover, TB is highly stigmatized in Vietnam, also among HCW, and therefore it was considered inappropriate to collect details that could lead to identification of HCW.

The data only include TB cases known to the health facility administration. In Vietnam, TB is still stigmatized [20] and this may cause HCW to seek diagnosis and treatment elsewhere. On the other hand, annual health screening is mandatory and includes chest X-ray imaging. It is known that X-ray screening is a sensitive method to detect pulmonary TB, also in the absence of TB symptoms [21]. This mandatory system also implies that these notification data were derived from active case finding, which is not comparable to the passive case finding strategies used in Vietnam to diagnose TB.

TB infection (measured by tuberculin skin test or Quantiferon test-conversion) better reflects recent transmission of Mycobacterium tuberculosis than TB disease notification since progression to TB disease may take several years and depends on other factors such as immune status of the host [22]. It may take time until changes in transmission will be measurable as changes in notification. Besides that, in countries with a generalized TB epidemic, transmission is likely to occur also in community settings [23‐25].

We analyzed reported information that was not checked by the researchers on site. For example, we did inquire whether a TBIC plan and a focal person were available in the health center and whether N95 respirators were being distributed to the staff. However, though this provided some information, this does not necessarily mean that plans are implemented as they should and respirators are (always) used appropriately when guidelines prescribe so, especially if appropriate training on the importance of TBIC and its practical implementation is lacking.

This effort is a repetitive, low-workload activity, recording information on a minimum number of TBIC indicators, meeting the WHO recommendation to specifically include notifications of TB among HCW. The minimum dataset did not include the number of staff and TB patients in each job category, completeness of HCW screening and notification, occurrence of MDR-TB, and whether it concerned nosocomial or community-based transmission. To enable assessment of the implementation of TBIC measures, routine on-site monitoring is needed. To help countries set up a standard routine TB monitoring system among HCW a guide was developed by an international group of experts under TB CARE I [26]. To assess the effect of TB IC measures on nosocomial TB transmission, sophisticated studies would be needed, including careful assessment of potential epidemiological links and genotyping of TB strains isolated from patients using advanced molecular typing methods, preferably whole-genome sequencing [27].

Finally, presently there are no standardized ways for measuring occupational risk of TB infection and disease in high TB incidence populations (such as the Vietnamese population), or for assessing the implementation of TBIC measures [10]. Repeated surveys could be used to measure and compare infection rates in healthcare workers [28].