The epidemiology of tuberculosis in health care workers in South Africa: a systematic review

Background In South Africa, workplace acquired tuberculosis (TB) is a significant occupational problem among health care workers. In order to manage the problem effectively it is important to know the burden of TB in health care workers. This systematic review describes the epidemiology of TB in South African health care workers. Methods A comprehensive search of electronic databases [MEDLINE, EMBASE, Web of Science (Social Sciences Citation Index/Science Citation Index), Cochrane Library (including CENTRAL register of Controlled Trials), CINAHL and WHO International Clinical Trials Registry Platform (ICTRP)] was conducted up to April 2015 for studies reporting on any aspect of TB epidemiology in health care workers in South Africa. Results Of the 16 studies included in the review, ten studies reported on incidence of active TB disease in health care workers, two report on the prevalence of active TB disease, two report on the incidence of latent TB infection, three report on the prevalence of latent TB infection and four studies report on the number of TB cases in health care workers in various health care facilities in South Africa. Five studies provide information on risk factors for TB in health care workers. All of the included studies were conducted in publicly funded health care facilities; predominately located in KwaZulu-Natal and Western Cape provinces. The majority of the studies reflect a higher incidence and prevalence of active TB disease in health care workers, including drug-resistant TB, compared to the surrounding community or general population. Conclusions There is relatively little research on the epidemiology of TB in health care workers in South Africa, despite the importance of the issue. To determine the true extent of the TB epidemic in health care workers, regular screening for TB disease should be conducted on all health care workers in all health care facilities, but future research is required to investigate the optimal approach to TB screening in health care workers in South Africa. The evidence base shows a high burden of both active and latent TB in health care workers in South Africa necessitating an urgent need to improve existing TB infection, prevention and control measures in South African health care facilities. Electronic supplementary material The online version of this article (doi:10.1186/s12913-016-1601-5) contains supplementary material, which is available to authorized users.


Background
Health care workers have an increased risk of acquiring tuberculosis (TB) as they are exposed to TB in their community as well as at their place of work [1][2][3]. In South Africa, workplace acquired TB is an important occupational disease among health care workers. According to the 2006 compensation fund claims, tuberculosis in health care workers, whilst generally underreported, is the third most commonly reported occupational disease in South Africa [4]. A review of the data submitted to the Compensation Commissioner by health care workers regarding hospital acquired infections from January 2007 to December 2009 in Limpopo province (i.e. one of the nine administrative divisions within the country) found that TB was the most common hospital acquired infection, with 47 (83.9 %) of the 56 reported cases of infectious diseases being TB cases [5].
Furthermore, the Human Immunodeficiency Virus (HIV) epidemic has impacted greatly on TB in South African health care workers. The HIV epidemic has increased the patient burden with TB co-infection in health care facilities, particularly at the primary care level. The increased number of sick patients with HIVrelated TB has increased health care workers' exposure to occupationally acquired diseases such as TB. Additionally, 11-20 % of South African health care workers are themselves, HIV-positive [6][7][8]. These HIV-infected health care workers have an increased risk for acquiring TB as well as for progressing from latent TB to active clinical disease [6].
Despite international policy recommendations [9] and national legislative provisions to address workplace acquired TB, generally South African health care facilities do not have adequate or appropriate infection prevention and control measures in place to protect their employees [10] (Dwadwa et al: Health worker access to HIV/TB prevention, treatment and care services in Africa: situational analysis and mapping of routine and current best practices, unpublished). The poorly enforced TB infection prevention and control policies (including administrative, engineering and personal protective equipment measures), the over-crowded health care facilities with insufficient ventilation to allow for appropriate environmental infection control and the high levels of undiagnosed, infectious TB patients, all contribute to health care workers' increased risk of exposure to TB in South African health care facilities.
The success of any TB Control Programme is dependent on health care workers' knowledge and application of appropriate policies and practices. In 2010 there were only 3.97 medical practitioner per 10 000 population and 18.97 professional nurses per 10 000 population in South Africa [11], reflecting the scarcity of health care workers in an overburdened health care system. TB infection amongst health care workers results in absenteeism and sometimes death or disability of health care workers, further weakening the already overburdened health care system. The first step towards managing workplace acquired TB in South African health care workers is to determine the true burden of TB in this population. This systematic review aims to collate all research conducted in South Africa reporting on any aspect of epidemiology of TB in health care workers in South Africa.

Methods
This systematic review is part of a larger project aimed at identifying all scientific evidence on the epidemiology of, and programmatic response to, TB in South African health care workers (http://www.cebhc.co.za/researchkey-outputs/research-evisat/). The larger project collated information on TB epidemiology, TB infection, prevention and control and TB prevention, treatment and care among health care workers in South Africa. Only the results of the epidemiology of TB in health care workers will be presented here.

Criteria for studies for inclusion in this review
Based on previously published definitions, the term "health care worker" was defined as "all people engaged in actions whose primary intent is to enhance health" and included, but was not restricted to, physicians, nurses, allied health personnel, health educators, social workers, midwives, community health workers, laboratory personnel, pharmacists, radiographers, volunteers, orderlies, and health-facility administrators [2,12]. All studies reporting on any aspect of TB epidemiology in health care workers in South Africa were presented in this review.

Search methods for identification of studies
A comprehensive search strategy (see Table 1), without any date or language restrictions, was used to search the following databases: To identify additional research we contacted experts (for example, researchers, health care workers, South African Department of Health employees) working in the field of TB in South Africa and screened the reference lists of eligible studies and review articles.

Data collection and analysis
Three reviewers independently screened the results of the searches (Liesl Nicol, Shaheen Mehtar and Keertan Dheda) and extracted data (Liesl Nicol, Muhammad Osman and Lydia Mudzikati) from eligible studies using a standardised data extraction form. Any disagreements regarding study eligibility or data extraction were resolved by discussion among the review team. The degree of selection bias, detection/information bias and confounding was determined for each included study using a validated risk of bias tool [13]. We planned to summarise the incidence data on TB in health care workers. However, after extracting the data from the eligible studies we noted that only some of the studies reported the 95 % confidence interval or standard deviation around the mean incidence or prevalence values. After consultation with a statistician it was agreed that the high level of heterogeneity between the incidence values would make summation of the data meaningless. Therefore, we decided to present the data in a horizontal bar chart to graphically reflect the incidence of active TB disease in health care workers compared to the general population or surrounding community.

Search results
The electronic search yielded a total of 668 references and searching other resources (for example: screening references of included studies, contacting experts in the field) yielded three additional studies (Fig. 1). A total of 587 references were excluded after the initial screening as the references either did not pertain to health care workers or the studies were not conducted in South Africa. We obtained the full text articles of 85 references, 48 of which were eligible for inclusion in the larger review looking at the epidemiology of and programmatic response to TB in South African health care workers. The reasons for excluding the remaining 37 references are outlined in Additional file 1: Table S1. The data from studies reporting on TB infection prevention and control, TB occupational health and safety, and care of health care workers with TB will be summarised in a future paper.

Description of included studies
We identified 16 studies reporting on various aspects of TB epidemiology in health care workers in South Africa. The detailed characteristics and findings of these studies are presented in Table 2.

Study setting and location
All of the included studies were conducted in publicly funded (government-funded), as opposed to privately funded, health care facilities in South Africa. Studies were conducted in health care facilities located in KwaZulu-Natal and the Western Cape, two of the nine South African provinces (i.e. administrative divisions within the country). We did not find any studies reporting on TB epidemiology in health care workers in the Eastern Cape, Free State, the Northern Cape and the North West province. The included studies provide information on TB epidemiology in health care workers in primary, secondary and tertiary level health care facilities. One study was conducted from a mobile testing unit in a community-based, as opposed to facility-based, setting [6].

Study design and risk of bias
Eight retrospective cohort studies, six cross-sectional studies and two prospective cohort study provided information on the epidemiology of TB in health care workers. Table 3 provides a summary of the risk for selection bias, detection/information bias and confounding present in each of the included studies. The risk of selection bias and detection/information bias was judged to be low in the retrospective cohort studies as these studies extracted data from hospital records or databases using data extraction forms. However, it is well known that there is underreporting of TB cases among health care workers, which may introduce detection/information bias and affect the external validity of the studies. Confounding was judged to be unclear in most of the studies as it is often unclear if the TB was acquired in the community or in the work place because of the high background incidence and prevalence of TB in South Africa. This issue is likely to confound most studies in this setting ( Table 3).

Incidence of active TB disease in health care workers
Ten studies reported on the incidence of active TB disease in South African health care workers. Six of the ten studies reporting on the incidence of active TB disease in health care workers show a higher incidence of active TB disease in health care workers compared to either the general population or people in the surrounding community ( Table 2) (Dwadwa et al: Health worker access to HIV/TB prevention, treatment and care services in Africa: situational analysis and mapping of routine and current best practices, unpublished) [14][15][16][17][18].
Both national surveys reported a higher incidence and prevalence of active TB disease in health care workers compared to the general population (Fig. 2). One national survey of 133 primary health care facilities, conducted between 2006 and 2008, reported a 2-3 times higher incidence of active TB disease in health care workers compared to the general population [15] ( Table 2). The authors noted that only 40 % of surveyed facilities had a TB screening program for health care workers. Many cases, therefore, are thought to go unreported [15]. The other national survey was a situational analysis conducted in 2006 that investigated health care worker access to HIV/TB prevention, treatment and care services in South Africa. This survey reported an active TB disease incidence of 1.13 to 1.47 % in health care workers from various publicly funded health care facilities (Dwadwa et al: Health worker access to HIV/TB prevention, treatment and care services in Africa: situational analysis and mapping of routine and current best practices, unpublished). In contrast, in 2006 the incidence of active TB disease in the South African population was 963 per 100 000 population (approximately 1 %) [19].
Studies conducted in public sector hospitals in KwaZulu-Natal during the time period 1999 to 2010 found that health care workers in these facilities had significantly increased incidence of both drug sensitive and drug-resistant TB disease compared to the surrounding communities ( Fig. 2) [16][17][18]. In contrast to this, studies conducted in the Western Cape province between 2008 and 2011, showed that health care workers from primary, secondary and tertiary care hospitals had a similar or lower incidence of active TB disease compared to the general population at the city and provincial level [14,20].
Four studies reported the incidence of active TB disease for the various occupational categories [16,18,20,21]. While certain categories of health care workers appeared to have higher incidence of active TB disease, only one study reported a statistically significant difference in      Table 2).

Prevalence of active TB disease in health care workers
Two studies provide data on the prevalence of active TB disease in South African health care workers. Active TB disease was less prevalent in health care workers working in health care facilities (1.4 %) [14] compared to those working in the community (5 %) [6]. Both of the studies report a higher prevalence of TB disease in health care workers compared to the prevalence in the surrounding community (population-based prevalence survey reported a TB disease prevalence of 3 %[6]) or general population (national prevalence of active TB disease for 2010 was estimated to be 0.8 % [19]) [6,14].

Incidence of latent TB infection in health care workers
Two studies provided information on the incidence of latent TB infection in health care workers in South Africa. Health care workers from primary and secondary level health care facilities in the Western Cape province had an annual rate of latent TB infection of 38 % (95%CI: 22-56 %, converted from negative to positive TST during 2009-2011) [14]. In Gauteng province, a cohort of health care workers and medical students working in three public sector health care facilities had an annual incidence of latent TB infection of 27 % (25/93 converted from negative to positive TST during 2008-2009) [22].

Prevalence of latent TB infection in health care workers
Three studies reported on prevalence of latent TB infection in health care workers in South Africa. The prevalence of latent TB infection in health care workers ranged from 84 % (TST-positive, health care workers from primary and secondary health care facilities in the Western Cape; 2009) [14] to 48 % (TST-positive, medical students and health care workers from public sector health care facilities in Gauteng; 2008) [23]. Two of the three studies reporting on prevalence of latent TB infection were conducted in the Western Cape [14,24]. In Gauteng, the prevalence of latent TB infection was higher among HCWs than medical students (56.7 % vs. 26.6 %, respectively; p < 0.0001) [23].

Drug-resistant TB in health care workers
Two studies specifically report on drug-resistant TB in health care workers in South Africa. One study reported specifically on the incidence of hospital admissions, to a public referral hospital in KwaZulu-Natal, for multidrugresistant or extensively drug resistant TB disease among health care workers [17]. In this study the annual estimated incidence of hospital admission for multidrug resistant (MDR)-TB among health care workers was more than 5 times that of the adult general population and more than 6 times of that of general population for extensively drug resistant (XDR)-TB (Table 2). A further retrospective case record review reported that ten of the 334 patients diagnosed with (XDR)-TB between 1996 and 2008 in the Eastern and Western Cape were health care workers [25].

Risk factors for occupational TB in health care workers
The findings of the five studies investigating the risk factors associated with active TB disease or latent TB infection (LTBI) in health care workers are presented in Table 4.

Age of health care worker
There wasn't a consistent association across the studies with regards to health care worker age and risk of active TB disease or latent TB infection. Furthermore, the particular age category of health care worker most affected by active TB disease or latent TB infection also differed between studies (Table 4) and only two studies reported a significant difference in the incidence of active TB disease between the various age categories. Ayuk et al. [20] reported that health care workers 40 years or older had a significantly higher incidence of active TB disease

Duration of employment
Both of the studies that assessed the relationship between duration of employment and incidence of active TB disease found no significant association [18,20]. In the one study that assessed the relationship between duration of employment and incidence of latent TB infection, duration of employment of greater than 20 years in a primary level care facilities was statistically significantly associated with an increased risk of latent TB infection (TST positivity; OR: 3.47; 95 % CI: 1.01-11.97) [14]. Wilkinson et al. [21] LOW: Staff TB data was extracted confidentially from the anonymized tuberculosis control programme register.
Incidence of TB disease: LOW (Case ascertainment is known to be complete because tuberculosis treatment cannot be obtained anywhere else in the district and all staff illness episodes are recorded in personnel files) UNCLEAR: Community vs occupational exposure to TB; HIV status of all of the HCWs not known LOW low risk of bias, HIGH high risk of bias UNCLEAR unclear risk of bias a Selection bias refers to systematic differences between baseline characteristics of the groups that are compared or characteristics of those who participate in the study and those who don't. It is important that the group described is representative of the population of interest b Detection bias/information bias refers to systematic differences between groups in how outcomes are determined. Participant's self-reported outcomes are usually associated with a high risk of detection or information bias

Health care worker occupation and work location
One study assessed the relationship between health care worker occupation and incidence of active TB disease and found no significant association for the various occupation categories [18]. No significant association was found between health care worker job category and risk of latent TB infection in the one study that reported on this relationship [23].
Tudor et al. [18] reported that compared to health care workers with no history of working in these areas, health care workers with a history of working in the TB ward (IRR 2.03, 95 % CI 1.11-3.71), paediatric ward (IRR 1.82, 95 % CI 1.07-3.10), the outpatient department (IRR 2.08, 95 % CI 1.23-3.52) and the stores department (IRR 2.38, 95 % CI 1.06-5.34) had a significantly increased incidence of active TB disease.

HIV status
Two studies reported on the relationship between HIV status and the incidence of active TB disease [18,20].  [18,20].
Two studies assessed the relationship between HIV status and incidence of latent TB infection [14,22]. Adams et al. [14] showed that HIV-positive health care workers were less likely to have a positive TST (OR: 0.41, 95 %CI: 0.18-0.92). In contrast, McCarthy et al. [22] did not find a significant association between HIV status and incidence of latent TB infection as measured by either IGRA assay or TST.
Only one study reported on the association between HIV status and the prevalence of latent TB infection [23]. Although HIV-negative health care workers were less likely to have a positive TST (OR 0.28; 95 % CI 0.10-0.74), this association was not significant after adjustment for age, job category and TB knowledge score [23]. presented is an average of the combined incidence of active TB disease in health care workers from best practice and randomly selected health care facilities; all incidence data for general population in the Republic of South Africa was obtained from the World Health Organisation Global TB database [19]; data on the incidence of active TB disease in the surrounding community was obtained from the individual studies; MDR-TB: multi-drug resistant tuberculosis; XDR-TB: extensively-drug resistant tuberculosis In KwaZulu-Natal, Wilkinson et al. [21] reported that 12 out of the 14 TB cases in health care workers tested for HIV were positive. O'Donnell et al. [17] reported that of those tested, 67 % of the health care workers admitted to hospital for MDR-TB or XDR-TB were HIVpositive. Tudor et al. [18] reported that 21 % of the health care workers with TB were HIV-positive. In the Western Cape, Adams et al. [14] reported that 3/5 health care workers newly diagnosed with TB were HIVpositive.

Diabetes
There is limited data from the studies in this review on the association between diabetes and risk of TB disease or infection. Both of the nurses diagnosed with TB in Balt et al. had non-insulin dependent diabetes [26].

TB infection prevention and control (IPC) training
One study assessed the association between training in TB IPC and risk of active TB disease and found that health care workers who reported having had no previous TB IPC training were significantly more likely to acquire active TB disease (OR: 2.97, 95%CI: 1.15-7.71) [20]. Two studies reported on the association between TB IPC knowledge, training and practice and the risk of incident latent TB infection [14,22]. Adams et al. [14] found that having "some training in infection control procedures aimed at self-protection" significantly reduced the odds of latent  [23] found that medical students who had a TB knowledge score > 7 (the median score for all participants), had reduced odds of a positive TST by >70 % (adjusted OR: 0.29; 95 % CI: 0.09-0.98).

Discussion
The majority of the studies reflect a high burden of active TB disease and latent TB infection in health care workers in South Africa. Although ten studies provide information on the incidence of active TB disease, only two studies provide information on the prevalence of active TB disease in health care workers in South Africa. Furthermore, only two studies reported on the incidence of latent TB infection with only three studies reporting on the prevalence of latent TB infection. Only two studies reported on the burden of specifically drug-resistant TB in South African health care workers. Most of the studies were conducted in the provinces of Kwa-Zulu Natal and the Western Cape, with little to no information on the burden of active TB disease or latent TB infection in health care workers employed in health care facilities in the Eastern Cape, the Free State, the Northern Cape and the North West province. Considering the importance of the issue it is surprising that there is relatively little research describing the actual burden of TB in health care workers in South Africa.
Other reviews have shown that the local prevalence of TB and HIV influences the risk and burden of active TB disease in health care workers in those localities [1][2][3]. The background incidence of active TB disease differs between provinces in South Africa, with the background incidence/prevalence of active TB disease in KwaZulu-Natal being one of the highest in the country. This may explain why the incidence of active TB disease was consistently higher in health care workers compared to the community in Kwa-Zulu Natal province and not in the Western Cape province.
Previous reviews have shown that the infectiousness of the patient population and the level of care of the facility, influence the extent of TB transmission to health care workers [1][2][3]. Health care workers at primary care facilities and those working in the community may be more likely to be exposed to infectious, as yet undiagnosed, TB patients compared to health care workers at tertiary care facilities. The studies reporting on prevalence of active TB disease, conducted in the Western Cape at a similar time (2009/2010), support this in that active TB disease was more prevalent in community health workers (5 %) [6] compared to health care workers in primary and secondary health care facilities (1.4 %) (Adams: Prevalence and determinants of TB infection in health care workers, unpublished).
Two studies provide information on the incidence of latent TB infection in health care workers. In both cases the annual incidence of latent TB infection in these South African health care workers is three to four times higher than the incidence previously reported in health care workers in India, another high TB incidence setting [14,27,28].
There is limited data on the various risk factors for acquisition of active TB disease or latent TB infection in health care workers in South Africa (Table 3). HIV status and training in TB infection, prevention and control were the only two factors for which there were consistent associations. While HIV-positivity was significantly associated with increased odds of acquiring active TB disease, it was not consistently associated with increased odds of acquiring latent TB infection. The treatment of HIV and latent TB infection in health care workers will need to be explored in future studies to understand the impact on the development of active TB disease in HIVinfected health care workers.
In four studies training in TB infection prevention and control was significantly associated with decreased odds of acquiring active TB disease [20] and latent TB infection [14,22,23]. This finding emphasizes the importance of on-going TB infection prevention and control training among all health care workers in the facility. Considering these findings and that all employees of the health care facility are at risk of acquiring TB it can be emphasized that all employees of the health care facility should receive training in TB infection, prevention and control.
Previous reviews reported that the specific work locations and occupational categories of health care workers are associated with a higher risk of acquiring TB disease [1][2][3]. The studies in our review that investigated these relationships showed somewhat similar findings. In one study there was no significant association between health care worker occupation and incidence of active TB disease, however, specific work locations (TB ward, paediatric ward and the outpatient department) were significantly associated with increased risk of acquiring active TB disease [18]. In the other studies certain occupational categories of health care workers appeared to have higher incidence of active TB disease, however, only one of the four studies reporting on this found the difference in incidence rates between the various occupational categories to be statistically significantly different [20]. Nevertheless, it was startling to see the high incidence of active TB disease in housekeeping staff, paramedical staff and hospital security staff [16,20]. Housekeeping and hospital security staffs, in particular, are unlikely to receive training in appropriate TB infection prevention and control, providing a possible explanation for the high burden of TB in these individuals.

Conclusions
There is relatively little research on the epidemiology of TB in health care workers in South Africa, despite the importance of the issue. Only two studies provided information on the annual incidence of latent TB infection in health care workers in South Africa. In addition to this, there is a dearth of information on the epidemiology of TB in health care workers from Eastern Cape, Free State, the Northern Cape and the North West provinces of South Africa. The lack of comprehensive information on the true burden of TB in health care workers will have implications on the implementation of preventive therapy policies.
To determine the true extent of the TB epidemic in health care workers in South Africa, screening for active TB disease (with identification of drug resistant TB strains) should be conducted in all health care facility employees, in all health care facilities on a regular basis. Future research should investigate the optimal approach, in terms of cost-effectiveness, feasibility, practicality and usefulness, to TB screening for active TB disease in health care workers in South Africa.
Finally, the evidence base shows a high burden of both active and latent TB in health care workers in South Africa, necessitating an urgent need to improve existing TB infection, prevention and control measures in South African health care facilities. Further to this training in TB infection, prevention and control be provided to all health care facility employees, including nonclinical staff.