The Effect of Malaria and HIV Co-Infection on Anemia: A Meta-Analysis.

Malaria and human immunodeficiency virus (HIV) infections are globally important public health concerns. The objectives of this study were (i) to determine the prevalence of malaria and HIV co-infections in people living in endemic countries, and (ii) to assess the effect of co-infection on anemia.Studies were searched on electronic databases including PubMed, Embase, Medline, Google Scholar, and African Journals Online. Observational studies, assessing the prevalence of co-infection and reporting its association with anemia, were included. The methodological quality of included studies was assessed using a tool called the risk of bias assessment for non-randomized studies. Heterogeneity among studies was investigated with the I-square test. Pooled prevalence of the co-infection and its 95% confidence interval (CI) were estimated using the random-effect model, reflected on heterogeneity among studies. Summary odds ratio (OR), summary standardized mean difference (SMD), and their corresponding 95% CIs were estimated, as appropriate. Subgroup analysis and meta-regression were performed for robustness of results. Publication bias was assessed by visualization of a funnel plot.Twenty-three studies were included in the present study. Overall, the pooled prevalence of co-infection was 19% (95% CI: 15-23%, I: 98.1%), showing 26% (95% CI: 20-32%, I: 98.7%) in adults, 12% (95% CI: 7-17%, I: 95.0) in pregnant women, and 9% (95% CI: 6-11%, I: 68.6%) in children. Anemia was comparable between the monoinfected and co-infected adults (summary OR: 1.49, 95% CI: 0.93-2.37) and increased by 49% in co-infected pregnant women (summary OR: 1.49, 95% CI: 1.14-1.94). The mean hemoglobin concentration was significantly lower in the co-infected group than the monoinfected group (summary SMD: -0.47, 95% CI: -0.61 to -0.33). The results of meta-regression on the prevalence of co-infection using the publication year and total population as covariates showed the I value remained high implying a de facto random distribution of heterogeneity. An asymmetrical funnel plot indicated the presence of publication bias. Due to heterogeneity of the studies in this review, the results have to be interpreted with caution.The findings of this study suggest that the prevalence of malaria and HIV co-infection, particularly in pregnant women, requires special attention from healthcare personnel. Better understanding of the co-infection is crucial for designing treatment strategies. Future well-powered, prospective designs assessing the interaction between malaria and HIV are recommended.

INTRODUCTION

A substantial progress in malaria control has already been achieved in 55 of the 106 endemic countries in the context of target 6 in Millennium Development Goal. However, malaria remains endemic in all 6 World Health Organization (WHO) regions and the heaviest burden is in the African Region which claimed ∼ 90% of all malaria deaths.[1] Sub-Saharan Africa also accounted for an estimated 25 million adults and children living with HIV/AIDS. Globally, malaria and HIV are the 2 major health problems and together they contribute to >4 million deaths per year.[2] People residing in regions where HIV and malaria are endemic are prone to develop co-infection. As such, there is a growing interest in whether this dual infection intensifies each other and attributes to complications in the clinical spectra and treatment outcomes.

Anemia is a significant consequence of malaria infection in terms of morbidity. Co-infection with HIV could lead to worsening of anemia through nutritional or immunological interactions. Empirical studies on malaria-HIV co-infection and particularly its effects on anemia showed varied results. Some studies reported this dual infection adversely affects maternal hemoglobin (Hb) concentration,[3] whereas others reported differently.[4-6] There is limited information on the collective impact of malaria-HIV co-infection on the Hb levels in general population.[7] Hence, a better understanding of the impact of co-infections could help in prevention and control of malaria. Earlier reviews in this field were either merely narrative[8,9] or not up-to-date.[10] However, since there has been a continuous flow of publications on the co-infection in endemic areas, it is timely to synthesis evidence by a meta-analysis.[11] Taken together, the objectives of the present study were (i) to determine the prevalence of malaria and HIV co-infections in people living in endemic countries, and (ii) to assess the effect of co-infection on anemia.

METHODS

The present study was carried out following the guideline on Meta-analyses of Observational Studies in Epidemiology (MOOSE)[12] and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA)[13] (See supplemental checklist). The protocol of this study is available in the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42015020248).

Study Search

Studies on the co-infection between malaria and HIV up to the end of June 2015 were searched in databases such as Embase Medline, Global Health, Google scholar, Scientific Electronic Library Online (SciELO), and African Journals Online (AJOL). To maximize the search, only 2 subject headers were used: malaria and HIV or AIDS and co-infections. Subsequently, specific terms such as plasmodium and HIV/AIDS were used. Only malaria and HIV were used for the search in SciELO and AJOL. Only studies on humans, published in English language were considered. The references of the identified papers and relevant reviews were reviewed to track any study that had not been captured by the electronic search.

Selection of Studies

Studies were included if they were (i) done on participants co-infected with malaria and HIV, and (ii) observational studies which compared the outcomes of anemia between those who were co-infected with malaria and HIV and those with monoinfection. To be eligible, malaria was confirmed by microscopy of Giemsa-stained blood films or rapid-onsite diagnostic test. Subsequent PCR-based analysis for confirmation of the species was advantageous, if done. Severe falciparum malaria and anemia were defined according to the WHO guideline.[14,15] Anemia was defined as Hb < 13 g/dL in men and <12 g/dL in women, whereas severe anemia as Hb < 7 g/dL in adults and <8 g/dL in children.[15] Placental malaria was diagnosed, based on analysis of placenta blood, either by PCR, by blood smear or both. Anemia defined in terms of low packed cell volume (PCV) or hematocrit (HCT) was accepted. If Hb concentration were specified as <5 g/dL or 11 g/dL in the same group, then it was collectively grouped under Hb < 11 g/dL. HIV was confirmed by an HIV RNA analysis/concentration, CD4+ cell counts, and rapid-onsite immunochromatographic test (RDT). For case-control studies, controls were not to be co-infected either with HIV or malaria, or vice versa. HIV in this study is HIV-1. Studies which did not meet the inclusion criteria were excluded.

Data Extraction and Quality Assessment

Two authors independently screened the citations, retrieved articles potentially eligible for the present study and extracted data from the studies, using a piloted data extraction sheet. Any discrepancies were resolved by consensus. The 2 investigators independently determined the quality of studies following the risk of bias assessment for nonrandomized studies (RoBANS) checklist, which consists of 6 domains.[16]

Statistical Analysis

For the variance stabilizing transformation of the proportions in estimating the prevalence of malaria and HIV co-infection, the Freeman–Tukey double arcsine transformation method was used as described elsewhere[17] and metaprop command in Stata. Details of this Stata command are available elsewhere.[18] Data were pooled using DerSimonian and Laird random-effect model, according to the results of heterogeneity testing.[19] For pooling of effect estimates for the effect of co-infection on anemia, summary odds ratio (OR) and its 95% CI for dichotomous data, and summary standardized mean difference (SMD) and its standard deviation (SD) for continuous data were used. The heterogeneity between these studies was assessed with the I2 test. An I2 value of >50% indicated substantial heterogeneity. For pooling of the results, a more conservative random-effect model was used as heterogeneity was substantial.[20]

Sensitivity Analysis

The prevalence of co-infections was stratified into infants, children, adults, and pregnant mothers. Due to paucity of data, stratification by Plasmodium species (P vivax, P falciparum, and mixed infections), CD4 counts in HIV, parasite density, or by gravidae were not possible. To account for within-study heterogeneity, metaregression on the prevalence of co-infection was performed using the publication year and total population as covariates. As a minimum number of 10 studies were available to investigate the publication bias,[20] a funnel plot was drawn with data of the effect size on anemia. This includes the regression line corresponding to the regression test for funnel-plot asymmetry.[21] Data entry was done in spreadsheet and analysis was with STATA 14.0 (Stata Corp, Txt). This study was approved by the Institutional Ethical Review Board ([BMS1–1/2015 [05]).

RESULTS

The study selection process was shown in the flowchart (Figure 1). The initial search yielded 2109 citations, of which 47 potentially met the inclusion criteria. A final of 23 studies (with 24 datasets) (n = 9159) were included in quantitative synthesis. Around one-third (37.5%) of the included studies were published between 2013 and 2015.

FIGURE 1

PRISMA flow diagram indicating the study selection process.

Characteristics of the Included Studies

The main characteristics of the 23 included studies[7,21-43] were presented in Table 1. Among the included studies, all but one study were carried out in the African countries. The remaining 1 was from Thailand in the WHO South-East Asia region.[42] A summary of the 24 excluded studies were provided[44-63] (see, Supplemental Table 1). Sixteen of the 23 included studies used cohort designs; of them, 13 studies were prospective cohort designs, 2 studies were longitudinal cohort studies, and 1 study was nested-cohort study. Malaria transmission was high in all these African countries where studies in this review were done. Thailand has low malaria transmission potential. Many studies showed low or unclear risk of bias (see, Supplemental Table 2).

TABLE 1

Characteristics of the Included Studies

Prevalence of Malaria and HIV Co-Infection

All 23 studies included in the present review provided data on prevalence of co-infection. Among these, 13 studies reported prevalence of HIV in malaria-infected patients,[22-24,27,28,30,35,37-40,42,43] whereas 10 studies on prevalence of malaria in HIV positives.[7,25,26,29,31-34,36,41]

Overall, prevalence of co-infection among individual studies ranged between 2% and 72% and pooled prevalence of co-infection was 19% (95% CI: 15–23%, I2: 98.1%). The pooled analysis of 13 studies with HIV in malaria yielded a prevalence of 22% (95% CI: 16–27%, I2: 98.5%). The pooled analysis of 10 studies with malaria in HIV-infected individuals yielded a prevalence of 16% (95% CI: 10–22%, I2: 96.8%) (Figure 2).

FIGURE 2

Forest plot showing overall prevalence of co-infections.

Anemia in Co-infection

All studies measured anemia in terms Hb except for 2 which measured in terms of PCVs[32] and hematocrit.[22] Nine studies (with 10 datasets) compared mean Hb levels between those who had dual infection and those who had malaria or HIV infection.[7,25,26,30-32,35,36,40] Thirteen studies compared proportions of anemia between those who were co-infected and those who had monoinfected.[22,24,27-29,33,34,37-39,41-43] Anemia was comparable between the monoinfected and co-infected adults (summary OR: 1.49, 95% CI: 0.93–2.37) and increased by 49% in co-infected pregnant women (summary OR: 1.49, 95% CI: 1.14–1.94) (Figure 3). Only 1 study on children reported that anemia was 5.5 times more in the co-infected group.[28]

FIGURE 3

Overall anemia comparing between the co-infection and the monoinfection groups.

When stratified by infection groups, studies with malaria in HIV showed significant lower level of Hb concentrations in the co-infected group than in the individuals infected with malaria alone (summary SMD:−0.47, 95% CI:−0.61 to −0.33). On further breakdown, this was (summary SMD: −0.33, 95% CI: −0.51 to −0.14) among adults and (summary SMD: −0.66, 95% CI: −0.87 to −0.45) among pregnant women (Figure 4). As expected, a fall in Hb concentration was more pronounced in pregnant women than in the nonpregnant adults.

FIGURE 4

Forest plot indicating the overall effect of malaria in HIV on anemia.

Subgroup Analysis on Prevalence of Co-infection

Studies identified for the current analysis were done in different subgroups. By age groups, prevalence of co-infection was 26% (95% CI: 20–32%, I2: 98.7%) in adults, 12% (95% CI: 7–17%, I2: 95.0) in pregnant women and 9% (95% CI: 6–11%, I2: 68.6%) in children (see, Supplemental Figure 1, which shows prevalence of co-infections stratified by age groups). Only 1 study on infants reported the prevalence of co-infection was 15% (95% CI: 12–19%).[29] The prevalence of co-infection among severe/complicated malaria cases was 28% (95% CI: 14–42%, I2: 94.5%) and 18% (95% CI: 14–22%, I2: 98.0%) among uncomplicated malaria cases (data not shown).

The metaregression plots obtained show that the publication year of the included studies could explain only 4.1% of the heterogeneity (figure not shown) and size of the denominator population of the included studies could attribute to 29.2% of the heterogeneity (Figure 5). The funnel plot of studies assessing effect on anemia showed an asymmetrical plot in the presence of publication bias among studies. This might be a result of the smaller studies producing smaller effect size[21] (Figure 6).

FIGURE 5

Meta-regression plot on prevalence of co-infection and size of denominator population.

FIGURE 6

Asymmetrical funnel plot in the presence of publication bias.

DISCUSSION

The current meta-analysis provides information on the prevalence of malaria and HIV co-infections and its impact on anemia in endemic countries.

Prevalence of Co-infection

As shown, the prevalence of HIV in malaria cases was higher (16–27%) than that of the malaria in HIV-infected patients (15–23%). This implies that the 2 pathogens could interact synergistically in human hosts.[64] Thus, the communities in these regions suffer a double burden of these 2 infections, with the higher burden being the HIV infections. HIV infection could impair immune responses to malaria parasites, leading to a decreased ability to control parasitemia,[27] whereas malaria infection can modulate HIV progression[65] and HIV RNA replication.[29,64,66]

It has been established that both humoral and cellular immune responses play protective roles against malaria infection.[65] During a humoral response the HIV-associated decline in antibody responses to variant surface antigens (VSAs) could partially explain the increased susceptibility of HIV-infected pregnant women to malaria. Hence, HIV infection impairs antimalarial immunity, most notably in immunosuppressed women. This increases the susceptibility of women to malaria during pregnancy. Studies have documented that the placenta provide a unique and hospitable niche for HIV-1 replications.[63,67]

Effect on Anemia

The cause of anemia is multifactorial, involving a number of mechanisms. Both malaria and HIV could individually cause anemia; the present analysis has documented that anemia was comparable between the monoinfected and co-infected adults. Both infections can lead to loss of appetite. Furthermore, advanced HIV/AIDS cases may cause diseases of the gastrointestinal tract, leading to malabsorption of nutrients necessary for the formation of Hb, which subsequently results in anemia.[58] Also, the ability of HIV to infiltrate the bone marrow[68] could accentuate the progression of anemia in co-infected individuals.

It has been well established that the primary target of human plasmodium species is the red blood cells (RBCs). In both vivax and falciparum malaria, parasitized and possibly nonparasitized RBCs are hypothesized to be more fragile than RBCs in noninfected individuals and more prone to damage from shear stress.[69] Malaria causes anemia through increased destruction of infected RBCs as well as uninfected RBCs.[70]

Studies had documented multiple factors that can contribute to ineffective erythropoiesis in severe malaria anemia including (i) hemozoin-induced repression of macrophage function, (ii) secretion of proinflammatory cytokines from macrophages, and (iii) perturbation of erythroblast metabolism with the resultant decrease in RBC output from the bone marrow.[71] Details of these mechanisms are described elsewhere.[70,71] An empirical study showed that HIV-infected multigravidae were less likely to clear infection than primigravidae. The altered immune-recognition in the setting of HIV could have contributed to it.[59] This implies that in malaria endemic regions, anemia could be more pronounced in multigravidae women co-infected with HIV/AIDS.

Study Limitations

A small number of participants in some studies identified for this review may not be adequately powered to reach the statistically significant level between the co-infected group and the mono-infected group. Sampling bias could occur if the participant groups, whether HIV-positive and HIV-negative or malaria positive and malaria negative were recruited at different times of the year. It is also important to note antiretroviral therapy (ART) such as zidovudine may cause anemia in HIV/AIDS patients as well as malaria and HIV co-infected patients. Other confounding factor includes the possibility of co-medication effects and other comorbidity conditions (e.g., tuberculosis, hepatitis B, hepatitis C, diabetes mellitus, renal failure, and chronic liver disease) in the participants in primary studies. These factors should be kept in mind when interpreting the findings. Furthermore the quality of included studies varied because many of these studies were not primarily aimed to assess the effect of anemia in co-infections. This could have attributed to ascertainment bias among these studies. As there was publication bias, findings in the current analysis should be interpreted with caution.

Public Health Implications

Anemia caused by P falciparum remains the frequent indication for blood transfusion in Sub-Saharan Africa.[72] As HIV can affect the course of malaria infection in many ways, screening of HIV for blood transfusion is important. Moreover, routine screening of pregnant women for both HIV and malaria at the first antenatal visits and education on malaria prevention (e.g., insecticide-treated nets and intermittent prophylactic treatment) should be initiated.

CONCLUSION

The findings of this study suggest that the prevalence of malaria and HIV co-infection, particularly in pregnant women, requires special attention from healthcare personnel. Better understanding of the co-infection is crucial for designing treatment strategies. Future well-powered, prospective designs assessing the interaction between malaria and HIV are recommended.