A total of 1,053 articles were retrieved from searches in six databases, including ProQuest (n = 385), Scopus (n = 207), Embase (n = 191), MEDLINE (n = 115), PubMed (n = 112), and Ovid (n = 43). After 423 duplicated articles were removed by automation tools (n = 380) and manual screening (n = 43), the remaining studies (n = 630) were screened according to their relevant titles and abstracts. After non-related articles (n = 518) were excluded, the remaining articles were assessed for eligibility (n = 112). Twelve articles18,19,21,30,31,32,33,34,35,36,37,38 met the eligibility criteria and were included. Further searching in Google Scholar identified four additional studies20,39,40,41 that met the eligibility criteria. No relevant studies were found upon scanning the reference lists of the included studies. Finally, 16 original articles18,19,20,21,30,31,32,33,34,35,36,37,38,39,40,41 were included for review (Fig. 1).
Studies’ characteristics and risk of bias
The majority of studies were published between 2010 and 2023 (75%) and were cross-sectional studies (75%). Most studies were conducted in Africa (56.3%), including Nigeria, Sudan, and Uganda. The remaining studies were conducted in Asia (31.3%), including India and Turkey, Europe (France), and South America (Colombia). Most enrolled individuals infected with P. falciparum malaria (62.5%), and most participants were adults (31.3%). Approximately 50% of the participants in the included studies had symptomatic malaria. All included studies used microscopic examination to detect malaria parasites (Tables 1 and S2). The risk of bias among the included studies was examined using the JBI critical appraisal tools for case–control, cohort, cross-sectional, and experimental studies. The results of the assessment showed that one case–control study lacked exposure period of interest30; four cross-sectional studies lacked the identification and strategy to deal with confounding factors19,21,32,40; and the cohort study had unclear follow-up details34. All studies were included in the review (Table S3).
GPx between malaria patients and uninfected individuals
Sixteen studies compared blood levels of GPx between malaria patients and uninfected individuals18,19,20,21,30,31,32,33,34,35,36,37,38,39,40,41. Based on the finding of these studies, eleven studies demonstrated that GPx blood levels were significantly lower in malaria patients than in uninfected individuals (68.75%)18,19,30,32,33,34,35,36,39,40,41. In contrast, three studies published that blood levels of GPx were significantly higher in malaria patients than in uninfected individuals (18.8%)20,37,38. Finally, two studies showed no difference in blood levels of GPx (12.5%)21,31.
The difference in blood levels of GPx between malaria patients and uninfected individuals was estimated in the meta-analysis of 15 studies that reported quantitative data18,19,20,21,30,31,32,33,35,36,37,38,39,40,41. Results showed diminished blood levels of GPx in malaria patients compared to uninfected individuals (P < 0.01, Hedges’ g: − 4.06, 95% CI − 5.49–(− 2.63), I2: 99.07%, 1,278 malaria patients/627 uninfected individuals, 15 studies, Fig. 2). The meta-regression of publication year, study design, country, continent, Plasmodium species, age groups, and clinical status demonstrated that the publication year, Plasmodium species, and clinical status significantly affected the pooled estimate (P < 0.05, Table S4). Subsequently, subgroup analyses of the publication year, Plasmodium species, and clinical status were conducted.
Following subgroup analysis, the publication year had significant subgroup differences (P < 0.01, Supplementary Fig. 1). Studies published between 2010 and 2023 showed diminished blood levels of GPx in malaria patients relative to uninfected individuals (P < 0.01, Hedges’ g: − 5.83, 95% CI − 7.98–(− 3.69), I2: 99.26%, 11 studies). In contrast, the blood levels of GPx were similar between the two groups in studies published between 2000 and 2009 (P = 0.81, Hedges’ g: − 0.12, 95% CI − 1.14–0.89, I2: 91.21%, 2 studies), and in studies published before 2000 (P = 0.42, Hedges’ g: 0.76, 95% CI − 1.08–2.60, I2: 96.28%, 2 studies).
Subgroup analysis of the Plasmodium species also showed subgroup differences (P < 0.01, Supplementary Fig. 2). Studies that enrolled patients with P. falciparum infections reported diminished blood levels of GPx in malaria patients compared to uninfected individuals (P < 0.01, Hedges’ g: − 3.06, 95% CI − 4.46–(− 1.65), I2: 98.39%, 9 studies). However, studies that enrolled patients with P. falciparum and P. vivax malaria reported similar blood levels of GPx between the two groups (P = 0.08, Hedges’ g: − 6.31, 95% CI − 12.88–(− 0.63), I2: 99.71%, 3 studies), and patients with only P. vivax malaria (P = 0.15, Hedges’ g: − 2.05, 95% CI − 4.83–0.74), I2: 98.64%, 2 studies).
The subgroup analysis of the clinical status also showed subgroup differences (P = 0.01, Supplementary Fig. 3). Studies that enrolled symptomatic malaria patients and did not specify the clinical status of malaria patients demonstrated diminished blood levels of GPx in malaria patients compared to uninfected individuals (P = 0.03, Hedges’ g: − 1.55, 95% CI − 2.91–(− 0.18), I2: 97.99%, 7 studies) and (P < 0.01, Hedges’ g: − 8.01, 95% CI − 11.94–(− 4.09), I2: 99.52%, 6 studies), respectively. However, similar blood levels of GPx between the two groups were shown by studies that enrolled both symptomatic and asymptomatic malaria patients (P = 0.42, Hedges’ g: − 2.60, 95% CI − 8.91–3.70, I2: 99.06%, 2 studies).
GPx between malaria patients with P. falciparum and P. vivax infections
Three studies compared blood levels of GPx between malaria patients with P. vivax and P. falciparum infections37,38,40. The findings of the three studies showed similar GPx levels among patients with P. falciparum and P. vivax malaria37,38,40. Following meta-analysis of the three studies, blood levels of GPx were similar between patients with P. vivax and P. falciparum infections (P = 0.48, Hedges’ g: 0.10, 95% CI − 0.19–0.39, I2: 54.11%, 202 P. falciparum patients/251 P. vivax patients, 3 studies, Fig. 3).
GPx between malaria patients with different levels of parasite density and disease severity
Six studies investigated variations in GPx blood levels between malaria patients with various parasite densities18,19,20,21,36,39. Three studies demonstrated an inverse correlation between blood levels of GPx and parasite density18,20,36. Two studies showed no association between blood levels of GPx and parasite density19,21. In contrast, one study showed elevated blood levels of GPx in patients with moderate parasite density compared to those with low and high parasite densities39.
The leave-one-out meta-analysis showed no impact of individual studies on the pooled effect estimate (P value < 0.05 in each rerun meta-analysis, Fig. 4). When the fixed effects model was applied for comparisons with the random-effects model, the results indicated diminished blood levels of GPx in malaria patients compared to uninfected individuals (P < 0.01, Hedges’ g: − 0.82, 95% CI − 0.96–(− 0.69), I2: 99.07%, 15 studies, Supplementary Fig. 4), indicating that the change of assumption of the statistical model did not affect the stability and robustness of the results. The results of the sensitivity analysis indicated that the results of the meta-analysis were robust.
The asymmetrical distribution of Hedges’ g of individual studies was demonstrated by visualization of the funnel plot (Fig. 5). Results of the Egger’s test showed significant differences (P < 0.01). The publication bias was due to the small number of studies included in the meta-analysis. In addition, the distribution of Hedges’ g of individual studies was inside the significant area of the Contour-enhanced funnel plot (P < 0.05, Fig. 6). Therefore, the heterogeneity of the Hedges’ g from individual studies was the cause of the funnel plot asymmetry.