Health-care workers’ occupational exposures to body fluids in 21 countries in Africa: systematic review and meta-analysis

Abstract Objective To estimate the lifetime and 12-month prevalence of occupational exposure to body fluids among health-care workers in Africa. Methods Embase®, PubMed® and CINAHL databases were systematically searched for studies published between January 2000 and August 2017 that reported the prevalence of occupational exposure to blood or other body fluids among health-care workers in Africa. The continent-wide prevalence of exposure was estimated using random-effects meta-analysis. Findings Of the 904 articles identified, 65 studies from 21 African countries were included. The estimated pooled lifetime and 12-month prevalence of occupational exposure to body fluids were 65.7% (95% confidence interval, CI: 59.7–71.6) and 48.0% (95% CI: 40.7–55.3), respectively. Exposure was largely due to percutaneous injury, which had an estimated 12-month prevalence of 36.0% (95% CI: 31.2–40.8). The pooled 12-month prevalence of occupational exposure among medical doctors (excluding surgeons), nurses (including midwives and nursing assistants) and laboratory staff (including laboratory technicians) was 46.6% (95% CI: 33.5–59.7), 44.6% (95% CI: 34.1–55.0) and 34.3% (95% CI: 21.8–46.7), respectively. The risk of exposure was higher among health-care workers with no training on infection prevention and those who worked more than 40 hours per week. Conclusion The evidence available suggests that almost one half of health-care workers in Africa were occupationally exposed to body fluids annually. However, a lack of data from some countries was a major limitation. National governments and health-care institutions across Africa should prioritize efforts to minimize occupational exposure among health-care workers.


Introduction
Worldwide, health-care workers risk occupational exposure to blood-borne pathogens through contact with human body fluids. Although about 60 blood-borne infectious pathogens have been identified, including Epstein-Barr virus, most occupation-related, blood-borne infections are due to hepatitis B virus (HBV), hepatitis C virus (HCV) and human immunodeficiency virus (HIV). 1,2 However, other blood-borne pathogens still pose a risk: for example, in the 2013-2016 Ebola virus disease outbreak, over 890 health-care workers were infected, with a case fatality rate of 57%. 3 Occupational exposure can occur through percutaneous injury (i.e. a needle or sharp object penetrates the skin), mucous membrane exposure (e.g. of the eyes, nose or mouth) and non-intact skin exposure. Percutaneous injury accounts for 66 to 95% of occupational exposures to blood-borne pathogens. 4 Little is known about the global burden of percutaneous injury among health-care workers. However, a 2005 report estimated that worldwide more than 3 million occupationrelated percutaneous injuries occur annually. 4 Moreover, about 40% of HBV and HCV infections and 2.5% of HIV infections in health-care workers were due to percutaneous injuries. 5 Hence, each year, percutaneous injury resulted in around 66 000 HBV infections, 16 000 HCV infections and 1000 HIV infections, which together caused about 1100 deaths as well as substantial disability. 4 More than 90% of these infections occurred in developing countries, particularly in Africa, where infection is more prevalent and adherence to standard precautions can be poor. 5 Given the severe consequences of blood-borne infections, many high-income countries have established surveillance systems to monitor exposure to body fluids in health-care settings. 6 These systems help inform policy-makers for reducing the risk of transmission of blood-borne pathogens. In many African countries, such systems are not available and, consequently, exposure to body fluids is rarely monitored. Furthermore, occupational exposure of health-care workers in Africa is generally underreported and poorly documented -one Nigerian study found that up to 97% of exposures were not reported. 7 The true incidence of blood and body fluid exposure in Africa is, therefore, uncertain. The 2005 report estimated that the incidence of sharps injuries in individual health-care workers in Africa was 2.10 per annum. 4 However, the authors based the estimate on survey findings from eight African countries and did not include data on laboratory technicians or other auxiliary health-care workers. Moreover, the authors obtained the data in hospitals and may not be representative of the diverse range of health-care settings in the continent. A Congolese study found an annual prevalence of occupational exposure to body fluids among health-care workers of 44.9%, with an average of 1.38 exposures per health-care worker per year. 8 A Burundian study reported an annual prevalence of 67.6%, with an average of 2.7 exposures per health-care worker per year. 9 Body fluid exposure in African health-care workers Asa Auta et al.
Here we conducted a systematic review of observational studies to estimate the prevalence of occupational exposure to blood and body fluids among health-care workers in Africa, because a continent-wide estimate would help increase awareness of such exposure and prompt preventative measures.

Methods
We searched the Embase®, CINAHL and PubMed® databases on 1 September 2017 for original research articles published between January 2000 and August 2017 that reported the prevalence of occupational exposure to blood or other body fluids among health-care workers in Africa. The following search terms were combined with others using Boolean operators: "occupational exposure", "accidental exposure", "blood", "body fluid", "blood-borne pathogens", "health-care workers", "health workers", "health personnel" and "Africa" (Box 1; available at: http://www.who. int/bulletin/volumes/95/12 /17-195735). Additional articles were identified by checking reference lists and by Google and Google Scholar searches. There were no language restrictions. The research protocol was registered in the PROSPE-RO international prospective register of systematic reviews (CRD42017054288).
For this review, we considered occupational exposure to body fluids to occur through percutaneous injury, mucous membrane exposure, non-intact skin exposure and bites. We included studies that reported the lifetime or 12-month prevalence of occupational exposure through at least one of these routes. Health-care workers included all paid and unpaid individuals working in a health-care setting who could be exposed to infectious materials, including blood and body fluids. Hence, we included studies that involved doctors, nurses, laboratory technicians, auxiliary healthcare workers or students undertaking clinical training or gaining experience in health-care settings. In addition, we included studies if they were observational studies with either a cohort or cross-sectional design. We excluded case reports, case series, case-control studies, qualitative studies, studies with fewer than 100 participants and, because of historic underreporting in Africa, studies that reviewed reported cases of blood and body fluid exposure. Two reviewers independently screened studies against inclusion and exclusion criteria (kappa for inter-rater agreement: 90.8%). Discrepancies were resolved by consensus.
The quality of each study was assessed and the risk of bias was judged using eight parameters, modelled largely on the Joanna Briggs Institute's critical appraisal framework for prevalence studies: the sampling frame, sample size, sampling strategy, detailed description of research setting and population, response rate (adequate if 60% or higher), reliability of the instrument used, recall bias (12 months or shorter) and statistical analysis methods -failure to satisfy each parameter was scored as 1. 10 The risk of bias was classified as either low (total score: 0 to 2), moderate (total score: 3 or 4) or high (total score: 5 to 8).
Two reviewers extracted data from the studies and entered them into Microsoft Excel v. 16.0 (Microsoft Corporation, Redmond, United States of America). The data included: (i) author; (ii) year of publication; (iii) study country; (iv) sample size; (v) response rate; (vi) recall period; (vii) prevalence of blood and body fluid exposure; (viii) prevalence of percutaneous injury; (ix) prevalence of mucous membrane and non-intact skin exposure; (x) prevalence of blood and body fluid exposure by health staff category; and (xi) the proportions of cases due to needle-stick injury, splashes, cuts and bites. Any discrepancy was resolved by consensus.
Two countries, Egypt and Libya, are included in WHO's Eastern Mediterranean Region, but were classified as African for the purposes of this analysis.

Data analysis
We categorized studies by whether they measured lifetime or 12-month prevalence and by the type of blood and body fluid exposure considered: (i) all types, including percutaneous injury and mucous membrane exposure; or (ii) percutaneous injury only. Generally, we estimated lifetime prevalence using data from studies that reported the proportion of participants exposed to body fluids at any time during their career. Twelve-month prevalence was estimated using data from studies that reported the proportion of participants exposed to body fluids in the preceding 12 months. We derived pooled prevalence estimates of blood and body fluid exposure by random-effects meta-analysis based on the DerSimonian-Laird approach. 11 We assessed the robustness of our findings in sensitivity analyses that excluded studies with a high risk of bias.
Interstudy heterogeneity was assessed by Cochran's Q, which gives values for X 2 and P, and the percentage of the total variation across studies due to heterogeneity was estimated using Higgin's I 2 statistic. 12 The causes of heterogeneity were explored in subgroup and metaregression analyses. We considered the covariates: (i) geographical region; (ii) type of health-care facility; (iii) study period; (iv) sampling procedure (i.e. random versus convenience sampling); (v) sample size; (vi) proportion of doctors; (vii) proportion of nurses; (viii) proportion of laboratory staff; and (ix) the risk of bias classification. Only those covariates found to be significant at P < 0.10 were included in the multivariate model. In addition, the pooled prevalence of blood and body fluid exposure in different categories of health-care worker were derived in stratified analyses and the relative risk of occupational exposure between groups was determined by pooling data using a random-effects model. We performed all statistical analyses using Stata version 13.1 (StataCorp LP., College Station, USA).
The 12-month prevalence of all types of occupational exposure to blood and body fluids ranged from 17.0% (95% CI: 15.3-18.7) in a Kenyan study 20 to 67.6% (95% CI: 61.4-73.8) in a Burundian study (Fig. 4). 9 The estimated pooled 12-month prevalence was 48.0% (95% CI: 40.7-55.3). Regional pooled estimates ranged from 33 In Fig. 6, the slopes of the fitted lines suggest that the 12-month prevalence of both all types of exposure to blood and body fluids and of percutaneous injury decreased only gradually over the study period. The estimated pooled 12-month prevalence for studies published between 2010 and 2017 was 47.3% (95% CI: 41.5-53.1) for all types of exposure and 33.7% (95% CI: 28.2-39.2) for percutaneous injury (Table 2). These estimates were comparable to the overall estimated pooled 12-month prevalence for all types of exposure and percutaneous injury, which were 48.0% (95% CI: 40.7-55.3) and 36.0% (95% CI: 31.2-40.8), respectively.
Overall, substantial heterogeneity was observed among the studies for the estimated 12-month prevalence of all types of exposure to blood and body fluids (X 2 : 1816.5; P < 0.001; I 2 : 98.7%) and of percutaneous injury only (X 2 : 780.9; P < 0.001; I 2 : 96.5%). Meta-regression analysis showed that, of all the covariates explored in the bivariate analyses, only geographical region had a P-value less than 0.10: (P = 0.0874) and geographical region explained 17.6% of the between-study variation in the estimated 12-month prevalence of percutaneous injury.

Subgroup analyses
As many of the studies included disaggregated data, we were able to estimate: (i) the pooled 12-month prevalence of occupational exposure to blood and body fluids by job category; and (ii) the relative risk of all types of exposure to blood and body fluids or of percutaneous injury between various demographic groups, which were distinguished, for example, by job category, gender, years of working experience or receipt of training on prevention of blood and body fluid exposure (details available from the corresponding author). The estimated pooled 12-month prevalence of exposure to blood and body fluids for medical doctors (excluding surgeons), nursing staff (including midwives and nursing assistants) and laboratory staff (including laboratory technicians) was 46.6% (95% CI: 33.5-59.7), 44.6% (95% CI: 34.1-55.0) and 34.3% (95% CI: 21.8-46.7), respectively. Moreover, when data on percutaneous injuries were included, there was no significant difference in the risk of all types of occupational exposure between these job categories: the relative risk (RR) was 1.108 (95% CI: 0.926-1.326) for doctors versus nursing staff, 1.267 (95% CI: 0.733-2.193) for doctors versus laboratory staff and 1.332 (95% CI: 0.947-1.874) for nursing staff versus laboratory staff. Nor was there a significant difference in risk between males and females (RR: 0.886; 95% CI: 0.692-1.133).
In addition, when data on percutaneous injuries were included, there was no significant difference in the risk of all types of occupational exposure between health-care workers with 5 years or less working experience and those with more than 5 years (RR: 0.999; 95% CI: 0.831-1.202). In contrast, healthcare workers who worked 40 hours or more per week were significantly more likely to be exposed than those who worked fewer hours (RR: 2.221; 95% CI: 1.001-4.926). Six studies reported on health-care workers who had received training on infection prevention and oc-

Discussion
We found a high lifetime and 12-month prevalence of occupational exposure to blood and body fluids among healthcare workers in Africa: about two thirds were exposed during their entire career and almost one half were exposed each year. Most exposure was due to percutaneous injury, which had an estimated 12-month prevalence of 36.0%. Direct comparison of our findings with those in other continents was difficult because of a lack of similar, continent-wide systematic reviews and meta-analyses. Nevertheless, the high prevalence of percutaneous injury among healthcare workers in Africa has serious implications because most occupational exposure to blood-borne viruses, such as HBV and HIV, occurs via this route. This can have implications for the exposed health-care worker's health, the transmission of blood-borne viruses to patients and the availability of scarce human resources for health care in Africa.
We found a variation in healthcare workers' exposure to blood and body fluids across Africa. Occupational exposure to blood and body fluids and percutaneous injury were consistently more frequent in northern Africa and less frequent in southern Africa. The reason for these regional differences is not clear. One possible explanation is that blood and body fluid exposure was underreported in some studies, which is likely. Alternatively, our findings may reflect regional differences in the level of knowledge of occupational exposure or in adherence to standard precautions.
Our meta-analysis found that the 12-month prevalence of blood and body fluid exposure differed little between various professions and there was no significant difference in risk. A critical appraisal of the literature showed that these figures may have been influenced by differences in study methods and in the categorization of health-care workers, but most discrepancies observed were linked to the underreporting of blood and body fluid exposure. 75,76 In contrast, we found that the risk of blood and body fluid exposure was higher among health-care workers who had received no training on infection prevention, which is unsurprising because training improves knowledge and preventive practice. Furthermore, the risk of occupational exposure was also increased among staff who worked more than 40 hours per week. The acute shortage of health-care workers in Africa may, therefore, have contributed to the present findings. 4 Inadequate staffing often results in a high patient-to-staff ratio, which may in turn lead to staff having to work longer hours to bridge gaps in personnel. 77 Although longer hours can bring additional rewards for health-care workers, levels of stress and fatigue can increase, which may result in overworked staff becoming less alert and more susceptible to exposure to blood and body fluids. 77 Our findings Fig. 4. Meta-analysis, 12- Body fluid exposure in African health-care workers Asa Auta et al.
may therefore indicate the need not only to promote the safety and well-being of existing health-care workers in Africa, but also to address the acute shortage of health-care workers across the continent. Our study also highlights the need to step-up efforts to reduce occupational exposure to blood and body fluidsparticularly via percutaneous injury -among health-care workers in Africa. Percutaneous injury could be prevented by practical interventions such as safety engineered devices, including needleless intravenous systems, auto-disable syringes and blunt suture needles. However, our findings suggest that it may be more cost-effective to address factors contributing to increased exposure in the continent, such as a lack of training and long work hours. Regular in-service training for health-care workers could help promote standard precautions for preventing the transmission of bloodborne infection, such as hand hygiene, the use of personal protective equipment and techniques for minimizing the manipulation of sharps, including the avoidance of needle recapping. In addition to training health-care workers, a holistic strategy is needed to address the acute shortage of health-care workers in the continent and to monitor staff workload. Furthermore, standard precautions could be supplemented by educating health-care workers to take responsibility for their own health and safety and for that of others who may be affected by their actions at work. Finally, governments should provide policies and support systems for the surveillance, reporting and management of occupational exposure to blood and body fluids among health-care workers. This study has some limitations. First, the cross-sectional design of the studies reviewed does not allow causal relationships to be established. Second, because the studies reviewed were based on self-reported retrospective data, they may be prone to recall and social desirability biases. Therefore, it is likely that exposure was underreported in many studies. Third, our review included single or limited reports from some countries and many reports concerned regional studies that were not nationally representative of the study countries. These factors may affect the generalizability of our findings. Furthermore, our review would have benefited from the inclusion of studies from Guinea, Liberia and Sierra Leone, where there was substantial transmission of Ebola virus infection among health-care workers during the recent outbreak. However, no studies of the prevalence of occupational exposure to blood and body fluids among health-care workers in these countries have been published. Future research in these countries should investigate occupational exposure to blood and body fluids and the circumstances in which it occurs to inform policy and practice. Nevertheless, our study provides an insight into the burden of occupational exposure to blood and body fluids among health-care workers in Africa and could prompt the development of appropriate policies, systems and processes in the continent. ■ Competing interests: None declared.