Prevalence of HIV infection among female sex workers in the Eastern Mediterranean Region countries: a systematic review

Introduction

Human immunodeficiency virus (HIV) is still an important public health issue, even though three decades has passed since its emergence. Currently, nearly 36.7 million people live with HIV infection around the world. In 2016, approximately 2.1 million new cases of HIV infections were identified [1]. According to the World Health Organization (WHO), only 19.5 million patients (52%) out of 36.7 million HIV-infected individuals are treated with antiretroviral drugs [2]. One way of HIV infection transmission is sexual contact, and female sex workers (FSWs) are among the at-risk groups [3]. FSWs are usually marginalized and deprived of many things in their living environments, and are perceived as criminals. Therefore, they are more vulnerable to HIV infection [3]. According to the WHO, FSWs are 13.5% more likely to be infected with HIV than other women in the age of fertility. In Asia, this probability reaches nearly 30% [4].
In low- to medium-income countries, the prevalence of HIV infection is estimated to be 12% among FSWs [5]. In the Eastern Mediterranean (EM) countries in 2016, approximately 230,000 people lived with HIV infection, with 21,000 new cases of HIV infection identified, and the prevalence rate was 0.1% among adults [2]. A study indicated that the probable world prevalence of HIV infection ranged from 0.3% in the Middle East and North Africa to 29.3% in sub-Saharan Africa [6]. In 2012, the prevalence of HIV infection was reported 5% in the Middle East and North Africa, in which the infection was mainly transmitted by FSWs [7, 8]. In Morocco, a review study showed that the prevalence of infection was 2% among FSWs, and FSWs played a major role in the transmission of HIV infection in comparison with intravenous drug addicts and homosexual men [9]. Accord­ing to previous studies, FSWs are exposed to many risk factors, such as having various sex partners, and not using or rarely using condoms [10]. In Libya, the prevalence of HIV infection among FSWs was 15.7%, and it was mainly due to their high number of sex partners and not using condoms [11]. Studies have reported that the frequency of using condoms is higher among customers who pay for sex than those who do not pay for it. Moreover, the prevalence of HIV infection is higher among FSWs who were intravenous drug addicts compared with those FSWs who used non-intravenous drug [12, 13]. Such individuals are usually infected by a common syringe or unprotected sex [14]. Many review studies have been conducted on the prevalence of AIDS in different parts of the world [15-17]. However, no review has been performed on FSWs in the EM countries. The aim of this systematic review was to investigate the prevalence of HIV infection among female sex workers in the EM countries.

Material and methods

Search strategy

In this review, articles published on the prevalence of HIV infection among FSWs in the Eastern Mediterranean (EM) countries until September 10, 2020, were searched. The information was retrieved from both electronic databases and WHO reports [8]. Appropriate keywords were searched in the Web of Science, Scopus, PubMed, EMBASE, Ovid, Google Scholar, and IMEMER (Index Medicos for Eastern Mediterranean Region) databases. First, the words “Prevalence”, “Incidence”, and “Frequency” were searched by using an ‘OR’ operator. In the second step, “Human Immunodeficiency Virus”, “Immunodeficiency”, “Virus”, “Human”, “Human Immunodeficiency Viruses”, “AIDS Virus”, “AIDS Viruses”, “Acquired Immune Deficiency Syndrome”, and “Acquired Immunodeficiency Syndrome Virus” were searched with ‘OR’ operators. In the third step, “FSWs” was searched. In the fourth step, the names of 21 Eastern Mediterranean countries were searched through an ‘OR’ operator. Finally, the searches of previous steps were joined using ‘AND’ operators to narrow down the results. Additionally, references at the end of each paper were verified. Conferences abstracts were also searched in Scopus, SID, and Embase databases.

Eligibility criteria

The retrieved papers were evaluated, and duplicate papers were deleted by checking the names of authors, publication year, place of study, and sample size. Then, the titles and abstracts of papers were screened according to the inclusion and exclusion criteria to remove irrelevant papers. Later, the entire texts of remaining papers were evaluated. Table 1 demonstrates the flowchart of papers’ selection. The inclusion criteria for the papers included in this review were studies conducted in EM countries, which reported the prevalence of HIV infection among the population, including FSWs who had sex in return for money, drugs, or goods, and their HIV infection had been diagnosed by laboratory tests. Papers that had reported HIV infections based on self-reports, and review papers were deleted. The texts of the papers were evaluated according to STROBE checklist [18]. The items of STROBE checklist were ranked zero (poor), one (medium), or two (good). The poor-quality papers included articles with a total score of 30 or lower. The medium-quality group consisted of the papers with a score between 31 and 35, and the good-quality group included the papers with a score of 36 or higher.

Classification and analysis of published articles

Key information was extracted from the included papers. A test of heterogeneity was performed before meta-analysis. In order to test heterogeneity, 2 test and I2 test were used along with Galbraith chart. In the c2 test, a small p-value indicated a significant difference in the results of included studies. The I2 test was introduced by Higgins, and it shows what percentage of differences in study results are due to heterogeneity among the results rather than sampling. Generally, values higher than 50% showed significant heterogeneity among the studies. We assessed the potential for publication bias by using Egger’s bias test for regression asymmetry [19]. The Galbraith chart graphically indicates if the results are homogenous by plotting each study’s log transferred effect size divided by its SE (Z score) on y-axis and the inverse of SE on x-axis [20]. After observing substantial heterogeneity across the studies, meta-regression was performed to identify potential sources of heterogeneity for years of data collection, and sample size. Sub-group analysis was performed according to years of data collection categories (2010 and before, after 2010), laboratory methods (DBS [dried blood samples] and serum groups), and sample size categories (387 and lower, over 387). Stata version 11 (Stata Corp., College Station, TX, USA) was used to analyze the data.

Results

In this study, 156 papers were retrieved by searching the databases, conference abstracts, reference lists, and reports published by the WHO. Eventually, 21 papers were selected for the review (Figure 1). Table 1 shows the information extracted from these 21 papers, in which 17,205 FSWs were investigated. The results of quality assessment revealed that 5 (23.8%) studies were of good-quality [11, 13, 21-23], 11 (52.4%) were of medium-quality [5, 24-33], and 5 (23.8%) were of poor-quality [34-38]. The study designs of 16 papers were cross-sectional [11, 22, 24, 25, 27, 28, 31-40], and the rest of them were national surveys for AIDS [13, 21, 23, 26, 41]. Out of 21 Eastern Mediterranean countries, only 8 countries had conducted studies on the prevalence of HIV infection among FSWs. The majority of these studies were conducted in Iran [13, 23, 24, 26-28, 34, 35] and Pakistan [22, 29, 30, 36-38]. Moreover, 13 papers were exclusively about FSWs [13, 21, 23-25, 27-29, 31, 33-35, 40], and the rest of them classified FSWs as a sub-group of the population under study [11, 32, 36-38, 41]. Nearly 16 papers had been published in the last 10 years [11, 13, 21-24, 26-28, 30, 33-35, 38-40]. In 11 studies included in this systematic review, the average age of FSWs was 30 and lower [11, 21, 22, 24, 25, 31, 33, 38-41]. FSWs had different education levels, and more than a quarter of FSWs had elementary or junior high school education in 11 papers [11, 13, 21, 23, 24, 26, 27, 31, 33, 35, 37]. Moreover, in other studies, more than half of FSWs were illiterate [22, 25, 30]. The rest of the papers had not reported education levels [28, 34, 36, 38, 39].
The onset age of sexual contact had not been reported in 12 studies [5, 13, 21, 23, 24, 26, 30-36]. Studies reported that the mean age of FSWs at first sex was after 15 years [13, 23, 31, 36], before 18 years [21, 34, 35], at 18 years [24], 22 ± 5.6 years [30], 23 ± 5.1 years [33, 40], or 25 years [26]. The average time, in which women worked as FSWs ranged from 2 to 6 years in five studies [21, 26, 27, 32, 36, 40]. Some other papers reported the frequency of genital infectious diseases [11, 22, 24, 25, 34-36, 38], candidiasis and cervicitis [27], genital ulcers [23, 33, 40], and pelvic inflammation [36] among FSWs. In several papers, FSWs never or rarely used condoms during sexual contact [24, 28, 36]. However, the consistency of using condoms among FSWs were as high as 65% in Pakistan, 64% in Lebanon, and 62% In Iran [22, 27, 32]. Condom use among regular and non-regular clients was reported to be respectively, 92% and 98% in Lebanon [32], and 63.4% among non-regular clients in Libya [11] and 57.1% in Iran [13]. A study from Sudan reported that 36% of FSWs used condoms for all clients [31], and in one study from Kerman, Iran, not using condoms was reported to be 17% and 22% with paying and non-paying partners, respectively [24]. Studies in Egypt and Sudan have mentioned that the main reasons for not using male condoms were not knowing where to obtain condoms [25], client refusal, not liking it, using another contraceptive method, not thinking about it, and partner’s refusal [31, 33]. The most common reported reason for starting to work as FSWs was poverty and financial problems [22]. The prevalence of drug use was high in the included papers and it ranged from 5% in Pakistan [30] and Sudan [21], to 91% in Iran [34] and 96.2% in Egypt [33]. The drugs abused included Iranian crack or heroin (in 16%), opium with methamphetamine (in 16%), methamphetamine (in 25%), methamphetamine with alcohol (in 13%) in Iran [27], and only alcohol in 4.7% [40], to 14% in Afghanistan and Iran, respectively [23]. In these studies, intravenous drug abuse ranged from 0.4% in Pakistan [22] and Afghanistan [40], to 38% in Iran [26]. One study in Iran reported that 11.6% of intravenous drug-addicted FSWs used non-sterilized syringes [13]. However, no statistics were reported on the use of drugs and alcohol in three other studies [11, 25, 32]. In all the studies, laboratory methods were used to identify HIV infections, and six studies used DBS [13, 26, 29, 31, 34, 41]. The rest of the papers used direct blood sampling methods to diagnose HIV infection. The prevalence of HIV infection ranged from 0% to 16% in the 21 papers reviewed. The prevalence of HIV infection was zero in three papers published from Iran [24], Lebanon [32], and Egypt [33]. The results of Egger’s test for the funnel plot were statistically significant (p < 0.001), which means that publication bias was likely to occur (Table 2).
Heterogeneity of the reported prevalence was very high among the studies (I2 = 89.9%, Q = 199.77, df = 11, p < 0.001), and therefore a meta-analysis was not applicable to perform [42]. Moreover, the studies were plotted on Galbraith chart (Figure 2). Some study results were out of the 95% CI, and this showed high heterogeneity. It was neither reasonable nor logical to remove specific studies to make the remaining results homogeneous for a meta-analysis. The result of meta-regression demonstrated that no significant relation was observed between the sample size and the prevalence of HIV, but there was a significant relation between the year of research and the prevalence of HIV (Table 3).
Authors used “visual binning” in SPSS v. 22, and divided the studies into sub-groups according to the years of data collection. The studies were divided into two groups (2010 and before, and after 2010), and in each category, heterogeneity was assessed. Heterogeneity in the 2010 and before group was I2 = 86.2%, Q = 87.23, df = 12, p ≤ 0.001, and in the after 2010 group, it was I2 = 90.7%, Q = 75.30, df = 7, p < 0.001 (Table 4). Therefore, meta-analysis was not done in these sub-groups because of high heterogeneity. The studies were also divided according to the laboratory methods used, into DBS and serum groups. In the DBS method group, heterogeneity was high and I2 = 92.1%, Q = 63.85, df = 5, and p < 0.001. Heterogeneity in the serum method was also high and I2 = 89.23%, Q = 120.71, df = 13, and p < 0.001 (Table 4). Therefore, meta-analysis was not performed in these sub-groups due to high heterogeneity. The studies were divided according to the sample size, into 387 and lower and more than 387. In the 387 and lower group, heterogeneity was high and I2 = 85.82%, Q = 70.53, df = 10, and p < 0.001. Heterogeneity in the more than 387 group was high and I2 = 91.54%, Q = 106.34, df = 9, and p < 0.001 (Table 4). Again, meta-analysis was not done in these sub-groups because of high heterogeneity.

Discussion

The reported prevalence of HIV infection varied differently in the studies on female sex workers in the EM countries, ranging from 0% to 16%. The majority of FSWs had never or rarely used condoms during sexual contact. Also, genital infectious diseases, candidiasis, cervicitis, genital ulcers, warts, or itching and pelvic inflammation were reported among FSWs. The prevalence of drug use was high, and heroin, opium, methamphetamine, and alcohol were the drugs with the highest use among FSWs. In the Eastern Mediterranean (EM) countries, only three countries, including South Sudan, Djibouti, and Somalia, have generalized epidemics; the remaining countries have an estimated HIV infection prevalence of below 1% in the general population, with either low-level epidemics or HIV infection epidemics concentrated in at-risk of HIV infection populations, and with limited spread to the general population [43].
Findings of our study showed that the prevalence rate of HIV infection among FSWs in the EM countries is different. Review studies in Europe (from 2005-2011) showed that there was a wide variation in HIV infection among FSWs at city level, and HIV infection prevalence among FSWs was highest in Eastern Europe in 2006, 2007, and 2009, within high HIV infection prevalence countries, such as Russia and Ukraine [12]. Also in the US (during 1984-2007), the prevalence of HIV infection among FSWs ranged from a minimum of 0.3% (1996-1998) to as high as 32.1% (1992-1994), and the pooled prevalence was 17.3% [42]. A review from Latin America and the Caribbean (from 1986 to 2010) showed that the median of HIV infection prevalence was 2.6% (IQR: 0.6-4.2) among FSWs [15].
HIV infection surveys among FSWs in the eleven Eastern Mediterranean (EM) countries demonstrated that the prevalence rate was below 4% in all countries, with the exception of Somalia (5.2%, 2007), Iran, and Djibouti [43]. The HIV infection prevalence among FSWs in Iran revealed an HIV infection prevalence of 4.5% in 2011 [13], whereas it was 2.6% in 2007; in Morocco, it was 2.7% in 2010, and 19.7% in 2007 and 15.4% in 2009 in Djibouti [44]. However, data on HIV infection prevalence among FSWs is unavailable from several low- or high-income countries, including Bahrain, Iraq, Kuwait, Libya, Oman, Qatar, Saudi Arabia, South Sudan, Syria, and United Arab Emirates (UAE), and the existing evidence do not portray the real picture of HIV infection in this region [43, 44]. On the other hand, HIV infection surveillance for evaluating HIV infection trends were performed in only four countries, included Djibouti, Iran, Morocco, and Pakistan. These countries have developed functioning HIV infection surveillance programs with national-level coverage. However, in the other 10 countries, HIV infection surveillance has been insufficient [43]. A review from the Middle East and North Africa showed a very low prevalence of HIV infection among FSWs in Afghanistan, Egypt, Jordan, Lebanon, Pakistan, Tunisia, and Yemen [45]. Despite the overall relatively low prevalence of HIV infection among FSWs, commercial heterosexual sex networks appear to be a cause of HIV infections in some countries, probably because of the larger size of these networks compared with other high-risk groups, such as homosexual men [45]. For example, in Morocco, commercial heterosexual sex networks contributed to about half of the HIV infection incidences, despite a rather low prevalence (2%) of HIV infection among FSWs [21].
Studies have demonstrated the significance of sex work in the ongoing transmission of HIV infection and other sexually transmitted infections (STIs). In the EM countries, genital infectious diseases, candidiasis, cervicitis, genital ulcers, warts or itching, and pelvic inflammation were reported among FSWs. A review study done in Europe indicated that the prevalence of gonorrhea was reported to be 5% or less across the region in FSWs, with the exception of Georgia (12-18%), and the prevalence of chlamydia was just over 20% in 2004-2006 [12]. Another review study from China reported that the prevalence of active syphilis ranged from 0.8% to 12.5%, herpes from 29.7% to 70.8%, chlamydia from 3.9% to 58.6%, gonorrhea from 2.0% to 85.4%, and trichomoniasis from 7.1% to 43.2% among FSWs [46].
Our review showed that the usage of condoms during sexual contact among FSWs was rare, and condom use among regular and non-regular clients was different. Several studies in African countries [47-49], China [50,51], various Asian countries [52,53], Latin America [54,55], and Europe [56] have shown that FSWs use condoms less frequently with their steady sex partners than with other clients [57]. The studies included in this review demonstrated that the prevalence of drug use among FSWs was high. A review showed that in Iran, over 70% of FSWs had ever used drugs, and around 15% had ever injected drugs. Moreover, older age, longer duration (> 5 years) of involvement in sex work, and a history of alcohol drinking were associated with a life-time history of drug injection among FSWs [58]. In Iran, about 15% of FSWs reported injecting drugs, and these were over three times more likely to be infected with HIV compared with FSWs who never injected drugs [13].
It is possible that most HIV infections among FSWs in Iran are due to drug injection rather than sexual transmission [45]. Drug injection was the main cause of HIV infection among female sex workers in Europe, and the prevalence of HIV infection was higher among FSWs intravenous drug abusers. The prevalence of syphilis was highest among samples taken from FSWs in Eastern Europe [42]. Putting FSWs under surveillance is critical because their behaviors are usually considered illicit, and they are frequently stigmatized, deprived from health services, and exposed to sexual diseases and violence [12]. Between 1989 and 2007, less than 1 percent of HIV infection tests performed in the region were for these key populations [59]. In the EM countries, condoms are unpopular and rarely used. Reasons for such low use are the fact that some men do not believe they are at risk of STIs or HIV infection, they do not like the feeling of condoms, and FSWs are worried about their clients reactions to their suggestion of using a condom. Some believe condoms may be hazardous to use, some are embarrassed about buying and using condoms, or lacking the skills to use them [60].
Some of the essential ways of controlling HIV infection and AIDS include raising awareness in the society through training, consulting, and changing behaviors [61]. According to the WHO, the only effective way for controlling AIDS is training, especially among vulnerable groups [62]. Therefore, comprehensive knowledge about HIV infection and AIDS as well as using condoms should be seriously and actively provided to FSWs [63]. Studies have also shown that health training plans designed according to health belief models, can be effective in preventing AIDS [62, 64]. One of the other educational methods for behavior change is theory of planned behavior (TPB), which considers intention as the main determinant of behavior. Studies have suggested that TPB can be used in education programs for HIV prevention [65], and it is necessary to provide free education and condoms to FSWs [62, 64].
In Tunisia, some forms of commercial sex are legal, and condoms are made available within regulated establishments. Around half of female sex workers in Morocco and Tunisia reported using a condom with their most recent client. However, in general, condom use is still largely tabooed in the region [59]. Four countries, including Pakistan, Egypt, Lebanon, and Tunisia, have provided HIV infection prevalence estimates for FSWs [66]. Different harm reduction programs, such as consistent condom use during high-risk sex and providing clean syringes, have been implemented in many EM countries, such as Egypt, Morocco, Tunisia, Bahrain, Iran, Lebanon, and Oman [66]. In Egypt, harm reduction interventions targeting PWID (people who inject drugs) and FSWs were established in 2008, and included education about safe sex and safe injection, HIV infection counseling, testing using rapid test kits, medical services for the management of sexually transmitted infections, and distribution of behavior-change booklets and brochures, needles, and condoms free of charge [67]. MENA region has had the lowest rate of antiretroviral treatment (ART) globally (11%) [68]. In the UNAIDS 2017 and global AIDS update reports in 2019 [69], none of these countries reported using ART for FSWs. Although some countries, such as Bahrain, Djibouti, Iran, Egypt, and Tunisia, have used ART for patients based on the latest WHO guidelines, which are CD4+ counts under 500 cells/mm3; but Iraq, Qatar, and Yemen did not report ART use [70].
There was a high-rate of heterogeneity among the papers included in this study. The difference in studies might be due to the difference in sampling, demographic characteristics of the participants, place of study, type of risky behaviors, and application of different laboratory kits for HIV diagnosis. Previous reviews have also reported high heterogeneity [6, 42]. In this review, the included papers, except for one paper from Pakistan [36], were conducted between 2004 to 2016. During this time period, different methods might have been used to investigate the prevalence. However, meta-regression showed no relation between the prevalence and the year of research. Most of the studies included in this review were conducted in Iran (8 papers) [13, 23, 24, 26-28, 34, 35], Pakistan (6 papers) [22, 29, 36-38, 41], and Sudan (2 papers) [21, 31]. There was only one study conducted on the prevalence of HIV infection among FSWs in Afghanistan, Egypt, Lebanon, Libya, and Somalia. In the other 13 countries of the EM, such as Bahrain, Djibouti, Iraq, Jordan, Kuwait, Morocco, Palestine, Oman, Saudi Arabia, Syria, Tunisia, UAE, and Qatar, no studies were conducted. However, it is necessary to survey female sex workers in these countries in regard to HIV infection, risky behaviors, problems in performing diagnostic tests, antiretroviral therapy coverage, and HIV prevention programs.
The present review had several limitations. First, the included studies were different in regard to sample size, study implementation dates, and sampling strategies. The prevalence rates reported from Pakistan, Afghanistan, and Lebanon were from before 2008, but the prevalence rates reported for Iran and Libya were after 2010. Moreover, the high heterogeneity of these studies excluded undertaking a pooled analysis. Also, it was not possible to evaluate geographical diversity due to small number of studies and lack of information from countries, such as Bahrain, Iraq, Jordan, Kuwait, Qatar, Saudi Arabia, Oman, Syria, Tunisia, UAE, Yemen, Morocco, and Djibouti. National studies were only conducted in a few countries of that region [13, 21, 26, 30, 35, 38], and other studies were limited to only some cities [11, 22-25, 27-29, 31-34, 36-38, 40]. Another limitation was that different HIV infection diagnosing tests had been used in the included studies, which makes the results even more incomparable. The specimens collected in the studies were different (plasma, serum, whole blood, DBS, and oral fluid), and this depended on the logistics, populations, sites, HIV infection testing strategies, and algorithms. In this review, the majority of tests were performed based on serum and DBS samples.

Acknowledgments

We would like to thank Prof. Maryam Okhovati, who help us in searching the databases.

Conflict of interest

The authors declare no conflict of interest with respect to the research, authorship, and/or publication of this article.

References