The association of malaria infection and gestational hypertension in Africa: Systematic review and meta-analysis

Background The World Health Organisation (WHO) estimates that hypertensive disorders of pregnancy (HDP) contribute 14% to global maternal mortality. HDP encompasses several subcategories, including gestational hypertension (GH) and pre-eclampsia. These two conditions are both characterised by a rise in blood pressure, with an onset from 20 weeks of gestation. They also share some common risk factors. The current definition of pre-eclampsia includes raised blood pressure in the absence of proteinuria, thus presenting the two conditions as a spectrum. In this article, we refer to both conditions as gestational hypertension,which is our outcome of interest. The aetiology of GH is not yet clearly understood. Observational studies have suggested that malaria may be associated with GH. However, the evidence from these small studies has been inconclusive. Having a better understanding of the association between malaria and GH may help inform prevention strategies to reduce maternal and infant mortality and morbidity. Methods In assessing the association between malaria infection and GH we explored open access articles published in the English language on Medline, Embase, WHO GIM and Google scholar. The subject related articles were retrieved and processed according to preferred reporting items in systematic reviews and meta-analyses (PRISMA) guidelines. Search date was 9th week of 2018. Inverse variance weighting method in Revman 5 software (Cochrane Collaboration, London, United Kingdom) was used to aggregate evidence by computing the pooled odds ratio to show the nature and strength of the relationship between malaria and GH. Results Using critical appraisal skills program (CASP) checklist tool we identified four good quality case-control studies. The total sample size was 1281 women out of which 518 were cases. These studies together show malaria is associated with GH with an overall odds ratio of 2.67, 95% confidence interval (CI) = 1.58-4.53. Heterogeneity of the individual studies supported fixed effect modelling assumptions (I2 = 0%). Malaria infectin may have a constant effect on GH across different African populations. The funnel plot did not suggest publication bias however, the four studies involved in the meta-analysis were insufficient to rule it out. Conclusions Our findings provide evidence of an association between malaria infection and gestational hypertension; this underscores the need to control malaria especially during pregnancy. We explored on our research question on the association of malaria infection and GH by searching articles on MEDLINE, WHO Global Index Medicus (GIM), Google scholar, EMBASE databases. We extended the search to include grey literature and hand searching of referenced studies. We used these search terms: Malaria, Placental malaria, Hypertension-pregnancy induced (pre-eclampsia, eclampsia, HEELP). Exposure keywords (malaria or placental malaria) were combined using the Boolean word AND with outcome keywords (pregnancy induced hypertension, pre-eclampsia). These keywords were modified to fit the respective databases. We restricted our search to full articles, English language and human studies with no limits on the year of study. We included observational studies (case-control, cross-sectional and cohort) that explored the association between malaria and GH, and excluded case reports, case series and ecological studies. We also excluded studies that did not ascertain the diagnosis of malaria by laboratory tests. Our final analysis only included final studies that had measured the association between malaria infection and GH while adjusting for confounding factors. Our study population was all pregnant women, the exposure was malaria infection during pregnancy, ie, women diagnosed with malaria during pregnancy, or with post-pregnancy diagnosis of placental malaria. Malaria Infection was diagnosed clinically by signs and symptoms accompanied by a blood test (rapid blood test for malaria or blood slide for microscopy). Malaria infection during pregnancy can invade the placenta to cause placental malaria, which can be confirmed after delivery by examining the placenta tissue. In this study, we considered either malaria infection during pregnancy or placental malaria confirmed after delivery as our exposure of interest. Women free of malaria diagnosis during pregnancy or free from post-pregnancy diagnosis of placental malaria were regarded as the unexposed group. The outcome of interest was women with a diagnosis of GH. The searched papers from the databases were downloaded and managed on Endnote X7 software (Clarivate Analytics, Philadelphia, PA, USA). Duplicates were removed by the software and followed by manual search and removal of duplicates. Titles and abstracts of identified studies were screened for relevance. Those not considered relevant were excluded. Additional papers were searched from grey literature sources such as relevant institutions’ repositories (eg, universities). Reference lists of the selected papers were examined to identify additional papers. We extracted key information from the articles; author, year, study design, country of study, sample size and univariate odds ratios (OR) and confidence interval (unadjusted). Where it was available, multivariable (adjusted) odds ratios and confidence intervals were extracted together with the list of variables used in the adjusted model. A full text review of the selected articles was done to select the eligible articles based on our inclusion and exclusion criteria. Critical appraisal skills programme (CASP) checklist tools were used to guide the quality assessment of the observational studies [19]. We reduced misclassification bias of the exposure on malaria infection by including only those studies which had ascertained the diagnosis of malaria with a laboratory test. We assessed the heterogeneity of the included papers to determine the feasibility of a meta-analysis. We had pre-set to accommodate moderate heterogeneity (I2 = 30%) due to the expected population variability of the exposure outcome association. We had anticipated the association between malaria and GH to vary widely across study populations hence we used the random effects model to estimate the pooled effect in the meta-analysis. Inverse variance weighting was used to compute the overall odds ratio in Revman 5 software (Cochrane Collaboration, London, United Kingdom). A forest plot was generated to display overall study results (Figure 1). We used a funnel plot to assess potential publication bias of the selected papers (Figure 2). Forest plot of the meta-analyzed studies. This forest plot shows the odds ratio of the individual studies and their pooled effect in the association between malaria infection and GH. The individual studies are weighed by the inverse variance method. The overall odds ratio is 2.67 with 95% CI = 1.58-4.53. The Heterogeneity of the studies is given by I2 = 0%. Suggesting strong homogeneity in their results. Funnel plot of the meta-analyzed studies. This funnel plot shows that studies with large and small variance had similar estimates of the effect of malaria infection on GH.