Hepatitis C virus viremic rate in the Middle East and North Africa: Systematic synthesis, meta-analyses, and meta-regressions

Objectives To estimate hepatitis C virus (HCV) viremic rate, defined as the proportion of HCV chronically infected individuals out of all ever infected individuals, in the Middle East and North Africa (MENA). Methods Sources of data were systematically-gathered and standardized databases of the MENA HCV Epidemiology Synthesis Project. Meta-analyses were conducted using DerSimonian-Laird random-effects models to determine pooled HCV viremic rate by risk population or subpopulation, country/subregion, sex, and study sampling method. Random-effects meta-regressions were conducted to identify predictors of higher viremic rate. Results Analyses were conducted on 178 measures for HCV viremic rate among 19,593 HCV antibody positive individuals. In the MENA region, the overall pooled mean viremic rate was 67.6% (95% CI: 64.9–70.3%). Across risk populations, the pooled mean rate ranged between 57.4% (95% CI: 49.4–65.2%) in people who inject drugs, and 75.5% (95% CI: 61.0–87.6%) in populations with liver-related conditions. Across countries/subregions, the pooled mean rate ranged between 62.1% (95% CI: 50.0–72.7%) and 70.4% (95% CI: 65.5–75.1%). Similar pooled estimates were further observed by risk subpopulation, sex, and sampling method. None of the hypothesized population-level predictors of higher viremic rate were statistically significant. Conclusions Two-thirds of HCV antibody positive individuals in MENA are chronically infected. Though there is extensive variation in study-specific measures of HCV viremic rate, pooled mean estimates are similar regardless of risk population or subpopulation, country/subregion, HCV antibody prevalence in the background population, or sex. HCV viremic rate is a useful indicator to track the progress in (and coverage of) HCV treatment programs towards the set target of HCV elimination by 2030.


Introduction
Viral hepatitis is ranked as the 7 th leading cause of mortality worldwide [1], with nearly half of this mortality attributed to hepatitis C virus (HCV) [1]. Despite its global burden, the Middle East and North Africa (MENA) remains the most affected region [2,3]. With the advent of direct-acting antivirals (DAAs) to treat and cure HCV infection [4], a global target was set to eliminate HCV infection by 2030 [5,6].
A key feature of HCV natural history is that not all infected persons develop chronic infection [7][8][9]. While infected persons pass through a stage of acute infection for few months, and develop antibodies against HCV infection, a proportion of them spontaneously clear the infection and becomes HCV antibody (Ab) positive but HCV ribonucleic acid (RNA) negative [7][8][9]. The remainder of infected persons become chronic carriers of the infection and persist as HCV Ab positive and RNA positive [7][8][9]. For a given population, the proportion of chronically infected individuals (HCV Ab positive and RNA positive), out of all ever infected individuals (HCV Ab positive regardless of RNA status), defines the HCV viremic rate for this population [10].
Assessing and understanding the HCV viremic rate is critical for biological, epidemiological, and public health consequences. The HCV viremic rate provides a measure of HCV spontaneous clearance rate and its determinants, and how this rate may vary by population [11]. The HCV viremic rate furnishes also a direct measure of the likelihood that a member of a specific population is chronically infected, as well as an indirect measure of the risk of HCV reinfection in this population [11]. It is further essential for estimations of the number of HCV chronic carriers in different populations and countries, and consequences for resource allocation and development of screening and treatment programs. The HCV viremic rate will also play an increasingly important role in assessing and monitoring the progress in (and coverage of) HCV treatment programs in different populations, as we forge ahead towards HCV elimination by 2030.
The HCV viremic rate has been assessed through numerous studies in different populations globally, but its measures show extensive variability across studies [12][13][14][15][16][17][18][19][20][21]. The HCV clearance rate, which is strongly linked to HCV viremic rate [11], has been also assessed in multiple prospective cohort studies [7,[22][23][24], but its measures also show wide variation across studies [11]. To our knowledge, no study have yet been conducted to provide an overall pooled estimate and subgroup pooled estimates for the HCV viremic rate that factor the wide diversity of studies for this measure. No study has also investigated the sources of heterogeneity in available HCV viremic rate measures.
Against this background, we aimed in the present study to provide pooled estimates for the HCV viremic rate, overall and for different risk populations and different countries of the MENA region. We also aimed to investigate the sources of heterogeneity in available measures in MENA. These quantitative assessments were based on a comprehensive and standardized database of systematically gathered HCV viremic rate data.
This study was conducted as part of the MENA HCV Epidemiology Synthesis Project, an ongoing effort to characterize HCV epidemiology and inform public health research, resource allocation, policy, and programing priorities in MENA [11-21, 25, 26].

Data sources
We retrieved studies reporting HCV RNA prevalence measures strictly among HCV Ab positive individuals from the MENA HCV Epidemiology Synthesis Project databases. These databases consist of 2,543 studies reporting HCV Ab prevalence among 52,598,736 participants, 47 studies reporting HCV Ab incidence among 29,600 participants, and 338 studies reporting HCV genotypes among 82,257 participants. The retrieved HCV RNA prevalence measures were nearly always extracted from studies whose main outcome measure was HCV Ab prevalence in some specific population. HCV RNA prevalence was a secondary outcome of these studies.
The HCV Synthesis Project databases were compiled through systematic reviews of the literature [12-17, 19-21] that were informed by the Cochrane Collaboration handbook [27], and reported as per the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [28]. The literature searches were conducted using broad search criteria with no language or year restrictions (S1 and S2 Boxes), and were based on international databases (PubMed and Embase), regional databases, national databases, and the MENA HIV/ AIDS Epidemiology Synthesis Project database [29,30]. Separate searches were also conducted for the non-indexed literature consisting of public health reports and routine data reporting. The flowcharts summarizing the searches can be found in the previous publications [12][13][14][15][16][17][19][20][21]. The PRISMA checklist for the present study can be found in S1 Fig. The definition of MENA for this project consisted of 24 countries including: Afghanistan, Algeria, Bahrain, Djibouti, Egypt, Iran, Iraq, Jordan, Kuwait, Lebanon, Libya, Mauritania, Morocco, Oman, Pakistan, Palestine, Qatar, Saudi Arabia, Somalia, Sudan, Syria, Tunisia, the United Arab Emirates (UAE), and Yemen.

Study selection and classification
All studies reporting a measure of HCV viremic rate were included, provided the sample size was !10. The HCV viremic rate was defined as the proportion of HCV Ab positive and RNA positive individuals out of all ever infected (i.e. HCV Ab positive) individuals in the sample. The overall sample size was replaced by stratified measures whenever this was possible while maintaining a subsample size !10. The viremic rate measures were classified based on the perceived risk of HCV exposure, as informed by existing classifications [2,[12][13][14][15][16][17]19-21], as follows: 1. General populations: populations at a low risk of being exposed to HCV infection such as blood donors, healthy adults, healthy children, and pregnant women, among others. Those referred to in included studies as "general populations" were labeled as other general populations to avoid confusion with the name of this category.
2. Populations at intermediate risk: populations at an intermediate risk of being exposed to HCV infection such as health care workers, diabetics, and prisoners, among others).
3. Populations at high risk of healthcare-related exposures: populations at a high risk of being exposed to HCV infection due to a medical condition that requires frequent injections or blood transfusions such as hemodialysis, thalassemia, and hemophilia patients, among others.
ranges: before 2000 and 2000 and thereafter. The sample size variable was categorized as !50 and <50. The sampling method variable was categorized as probability-based sampling and non-probability-based sampling.
For the year of data collection variable, we imputed the missing observations using the median of the results of the subtraction of the year of data collection from the year of publication. A sensitivity analysis using the imputed and the non-imputed observations revealed no impact on the statistical significance of the variable.

Scope of evidence
We identified 178 measures for HCV viremic rate among 19,593 HCV Ab positive individuals (Table 1). These measures included 81 in general populations, 20 in populations at intermediate risk, 51 in populations at high risk of healthcare-related exposures, five in PWID, eight in populations with liver-related conditions, and 12 in special clinical populations. One study was considered as "mixed" since the sample included a mix of general populations and an intermediate risk population [40].
There were data on HCV viremic rate in 16 out of the 24 MENA countries (Table 1). Egypt contributed the largest number of data points (n = 89), and the majority of these were from studies in general populations.

HCV viremic rate
HCV viremic rate varied across and within the risk populations with a broad range of 9-100% and a median of 68.8% ( Table 2). The overall pooled mean HCV viremic rate (across all data points) was 67.6% (95% confidence interval (CI): 64.9-70.3%).
By sampling method (Table 4), the pooled mean HCV viremic rate was 58.7% (95% CI: 49.4-67.6%) for studies using probability-based sampling but not at the national level, 68.7% There was (overall) evidence for strong heterogeneity in HCV viremic rate in all the different meta-analyses with generally a p-value <0.0001 (Tables 2-4). The I 2 for the pooled estimates indicated that the vast majority of the variation was due to true variation in HCV viremic rate across studies rather than chance (generally I 2 >>50%). The prediction intervals were generally very broad confirming substantial variation in measured HCV viremic rate across studies. Forest plots for the meta-analyses by risk population can be found in S2 Fig. Table 5 displays the results of the univariable meta-regressions to identify the predictors of HCV viremic rate and sources of between-study heterogeneity. None of the hypothesized predictors were statistically significant (p-value >0.05), and none were eligible for inclusion in the final multivariable model (p-value >0.1). Therefore, no multivariable meta-regression was conducted.

Predictors of HCV viremic rate and sources of heterogeneity
Though no variables were significantly predictive of HCV viremic rate, there were notably trends of lower viremic rate for females and children (Table 5), and trends of higher viremic rate for populations with liver-related conditions and for populations with high (>50%) HCV Ab prevalence.

Discussion
Through a comprehensive analysis of systematically extracted data, we investigated HCV viremic rate in the MENA region. We found that about two-thirds of HCV Ab positive individuals HCV viremic rate is defined as the proportion of chronically infected individuals (HCV Ab positive and RNA positive), out of all ever infected individuals (HCV Ab positive regardless of RNA status). Accordingly, it is closely linked to HCV clearance rate, defined as the proportion of people who spontaneously clear their infection-that is the proportion of people who clear their acute infection and do not become chronically infected [11]. Our results then imply that over 30% of infected individuals spontaneously clear their infection, a higher proportion than that estimated in prospective cohort studies of about 25% [7,23]. While prospective studies provide a direct approach to estimating the clearance rate, there are known methodological limitations and potential biases that may lead to underestimation of clearance rate [7,11,22]. Our results, using an independent methodology from that of prospective studies, suggest that one-third of infected individuals spontaneously clear their infection.
In planning for this study, our implied hypothesis was that we will identify several predictors of higher HCV viremic rate. PWID and populations at high risk of healthcare-related exposures may have a weaker immune system and are at a higher risk of HCV reinfection, therefore should have a higher viremic rate. Female sex is associated with higher spontaneous clearance rate [7,23,43], and therefore we expected the viremic rate among men to be larger than that among women. We further expected a higher viremic rate in populations with higher HCV Ab prevalence, as HCV Ab prevalence can be seen as a proxy for the risk of repeated HCV exposures. Lastly, we expected a higher viremic rate in populations with liver-related conditions, since the presence of these conditions could be indicative of chronic HCV infection. Nevertheless, none of these hypothesized effects were identified as statistically significant in our meta-regression analyses. Though there was a trend of lower viremic rate in women-only studies, and trends of higher viremic rate in populations with liver-related conditions and populations with higher (>50%) HCV Ab prevalence, none of these trends reached statistical significance. These results suggest that either some of these effects may not be present as originally hypothesized, or that the effect size of these effects was not large enough to be a This mean is the mean of HCV Ab prevalence in the study population from which the HCV viremic rate was extracted. b Q: The Cochran's Q statistic is a measure assessing the existence of heterogeneity in effect size. c I 2 : A measure that assesses the magnitude of between-study variation that is due to differences in effect size across studies rather than chance. d Prediction interval: A measure that estimates the 95% interval in which the true effect size in a new study will lie.
https://doi.org/10.1371/journal.pone.0187177.t004 detected in our sample of 178 viremic rate measures, or that the heterogeneity in the effect size lowered the power of the analysis to detect these differences. Though we could not identify any significant predictor of HCV viremic rate, the viremic rate varied widely across studies. This may suggest that much of this variation could be due to random effects, such as those related to the complex laboratory methods used in assessing the viremic rate. Assessment of HCV viremic rate requires a two-test algorithm, for HCV Ab and HCV RNA, and the diagnostic assays and protocols can vary from one study to another. Different assays, whether for HCV Ab or for HCV RNA, may also have different sensitivities and specificities, which can impact the estimated HCV viremic rate [44,45]. Even small random errors in assessing the denominator (HCV Ab positive cases), or the numerator (HCV RNA positive cases), can lead to large variation in calculated viremic rate.
Another source of random errors in calculated HCV viremic rate could be sampling variation as the "effective" sample size in viremic rate studies (the number of HCV Ab positive cases), tend to be small ( Table 1). The number of HCV Ab positive cases is most often a subsample of the original study sample size-the original study sample size is the number of individuals recruited in the original study whose serostatus could be HCV Ab negative or HCV Ab positive. The median size of the (sub) sample of HCV Ab positive cases in included studies was only 50. With a viremic rate of 67.6% (as was the pooled estimate), this small median sample size leads to a wide confidence interval (95% CI: 53.3%-80.5%). This highlights how (sub) sample size could be a major cause of the observed variation. The large variations in HCV viremic rate across studies (Table 1), but the small variations in the pooled mean HCV viremic rates (Tables 2-4), suggest caution against using the highly variable and possibly error-prone study-specific viremic rates in estimations of the number of HCV chronic carriers in different populations and countries, instead of the stable pooled means. We advocate here for the use of one standardized HCV viremic rate, say the overall pooled mean estimated in this study (Table 2), in ongoing chronic HCV infection estimations -such as the global estimations being conducted for the World Health Organization [3]. We further advocate for the use of pooled means, rather than study-specific estimates, for assessing and monitoring the progress in (and coverage of) HCV treatment programs, as we forge ahead towards HCV elimination by 2030. Of notice here that this progress monitoring will require repeated population-based measures of HCV antibody positivity and HCV RNA positivity in the same population, with sufficiently large sample sizes to assess statistically the trends in HCV viremic rate.
Our study has several limitations. The availability of data varied by risk population and country, and we did not identify any HCV viremic rate data for eight MENA countries. The number of studies was limited for some risk populations-only five studies were identified for PWID, and these were mostly conducted among PWID with access to prevention programs. Sample size varied across studies, and the sampled risk population may not have been representative of the wider risk population in the country. Despite these limitations, we identified a substantial volume of viremic rate data in MENA that facilitated the conduct of different types of analyses, thereby generating informative inferences.

Conclusions
Two-thirds of HCV Ab positive individuals in MENA are chronically infected with HCV infection, implying that over 30% of infected individuals spontaneously clear their infection. Though there was extensive variation in the study-specific HCV viremic rates, the pooled mean viremic rates were similar regardless of risk population or subpopulation, country or subregion, HCV Ab prevalence in the background population, or sex. These findings argue for the use of one standardized HCV viremic rate, such as the overall pooled mean viremic rate provided in this study, in estimations of the number of HCV chronic carriers in different populations and countries. These findings also highlight the utility of using the pooled mean viremic rate as an indicator to track the progress in (and coverage of) HCV treatment programs in different risk populations and countries, as viral hepatitis treatment programs are established and/or expanded with the ultimate target of HCV elimination by 2030.