Prevalence and proportion estimate of asymptomatic Plasmodium infection in Asia: a systematic review and meta-analysis

Asymptomatic Plasmodium infection raises a problem for the persistent transmission of malaria in low-endemic areas such as Asia. This systematic review was undertaken to estimate the prevalence and proportion of asymptomatic Plasmodium infection in Asia. The systematic review was registered at PROSPERO (ID: CRD42022373664). The research followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses. A comprehensive search of five databases, Ovid, Scopus, MEDLINE, PubMed, and Embase, was conducted to identify studies of asymptomatic Plasmodium infection in Asian countries. The pooled prevalence of asymptomatic Plasmodium infection, the pooled proportion of asymptomatic Plasmodium infection among all parasitised individuals, and the associated 95% confidence intervals were estimated using a random-effects model. A total of 916 articles were retrieved, and 87 articles that met the criteria were included in the systematic review. The pooled prevalence of asymptomatic Plasmodium infection among enrolled participants in Southeast Asia, South Asia, and Western Asia was 5.8%, 9.4%, and 8.4%, respectively. The pooled proportion of asymptomatic Plasmodium infection among all parasitised individuals in Southeast Asia, South Asia, and Western Asia was 89.3%, 87.2%, and 64.8%, respectively. There was a low prevalence of asymptomatic Plasmodium infection, but there was a high proportion of asymptomatic Plasmodium infection per all parasitised individuals in different parts of Asia. These results may support and facilitate elimination and control programs for asymptomatic Plasmodium infection in Asia.


Search procedures and study inclusion.
A comprehensive search of five databases (Ovid, Scopus, MEDLINE, PubMed, and Embase) was conducted to identify studies of asymptomatic Plasmodium infection in Asian countries (inception to October 16, 2022). The search strategy includes the terms "malaria OR Plasmodium OR "remittent fever" OR "marsh fever" OR paludism) AND ("asymptomatic infection" OR inapparent OR subclinical OR presymptomatic) AND (Asia OR Asian)". Details of the search strategy in five databases can be found in Table S1. We restricted the search to studies published after 2000. We included cross-sectional studies and cohort studies (at baseline) of asymptomatic Plasmodium infection among participants in Asian countries. Non-research articles, case reports, case series, case-control studies, conference abstracts, and in vitro and in vivo studies were excluded from this study. Two authors (MK and KUK) performed the study selection and data extraction and assessed the risk of bias among the included studies. Disagreement between the two authors was resolved by another author (AM).
The systematic review and meta-analysis were reported following the PRISMA checklist 13 . The risk of bias was assessed using Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) for crosssectional and cohort studies 16 . There were scale ratings to categorise the quality of studies as good, fair, or poor in cases where the percentile of the total score (overall, 22 items and scores) were more than 75%, 50-74%, or less than 50%, respectively 17,18 . Statistical analysis. (1) The pooled prevalence of asymptomatic Plasmodium infection among enrolled participants in all studies, and (2) the pooled proportion of asymptomatic Plasmodium infection among all malaria-positive cases and associated 95% confidence intervals (CIs) were estimated using a random-effects model (DerSimonian and Laird method) 19 . The meta-regression analysis was performed to determine whether covariates affect the prevalence estimate or not. The subgroup analyses of the publication year, parts of the continent, participants' group, age group, and method for malaria detection were performed to explore the source of heterogeneity and demonstrate the differences in prevalence in specific groups of participants. The heterogeneity of the effect estimates was assessed using the Chi-square and I 2 statistics as described previously 20 . Publication bias was not assessed across all included studies because there is no clear definition or consensus regarding what constitutes a positive result in a proportional meta-analysis 21 . Characteristics of the included studies. All details about the characteristics of the included studies are shown in Table S2. Briefly, most of the included studies were conducted in Southeast Asia (60 studies, 69%), followed by South Asia (20 studies, 23%), and Western Asia (7 studies, 8.05%). In Southeast Asia (60 studies), most of the studies were conducted in Thailand (10 studies, 16.7%), Myanmar (9 studies, 15%), Cambodia (8 studies, 13.3%), Malaysia (8 studies, 13.3%), Lao PDR (7 studies, 11.7%), Indonesia (6 studies, 10%), Vietnam (6 studies, 10%), the Philippines (2 studies, 3.33%), the Democratic Republic of East Timor (1 study, 1.67%), and three studies were conducted in multi-sites (1 in the Thailand-Myanmar border/Cambodia, 1 in the Thailand-Myanmar border/Cambodia/Vietnam, 1 in Thailand and Myanmar). In South Asia (20 studies), most of the studies were conducted in India (14 studies, 70%), followed by Bangladesh (4 studies, 20%), Pakistan (1 study, 5%), and Bhutan (1 study, 5%). In Western Asia (7 studies), most of the studies were conducted in Yemen  Quality of the included studies (risk of bias). The quality of the included studies was demonstrated in Table S3. For the overall quality, all 87 studies were high-quality studies (100%). In terms of methodology quality, 86 studies 3,6,7,10,11,[92][93][94][95][96][97][98][99][100][101][102][103] were of high quality, while one 91 was of moderate quality. All studies were included in the meta-analysis.

Proportion of asymptomatic Plasmodium infections per parasitised individuals in Asia.
The pooled proportion of asymptomatic Plasmodium infection among all parasitised individuals when the molecular method was used alone or in combination with other methods was 88.0% (95% CI 86.3-89.7%, I 2 : 98.25%, 70 studies, Fig. 4). Figure 5 depicts the geographical distribution as well as the estimated proportion of asymptomatic Plasmodium infection cases among all parasitised individuals. The meta-regression analysis showed that the country and participants' group affected the proportion estimate, with a P value less than 0.01, but the publication year, being part of the Asian continent, or method for malaria detection did not affect the proportion estimate (Table S4). In addition, the meta-regression analysis showed that the participants' group and country were the most-fitted covariates to explain the heterogeneity of the proportion estimate (adjusted R-squared = 30.76% and 29.19%, respectively). Subgroup analyses of parts of the Asian continent, country, and participants' group were performed using data from studies that used the molecular method alone or in combination with other methods. The subgroup analysis of parts of the Asian continent showed that the highest proportion of asymptomatic Plasmodium infection per parasitised individuals was found in Southeast Asia (89.3%, 95% CI 84.7-91.1%, I 2 : 98.23%, 56 studies), followed by South Asia (87.2%, 95% CI 79.9-94.5%, I 2 : 98.10%, 11 studies), and Western Asia (64.8%, 95% CI 30.6-99.0%, 3 studies) when the molecular method was used alone or in combination with other methods ( Table 2). The subgroup analysis of the country showed that the highest proportion of asymptomatic Plasmodium infection per parasitised individuals in Southeast Asia was found in Thailand/Myanmar (100%, 95% CI 93.5-100%) 67 , followed by the Democratic Republic of East Timor (100%, 95% CI 74.1-100%) 37  www.nature.com/scientificreports/ proportion of asymptomatic Plasmodium infection per parasitised individuals was found in studies that enrolled participants in communities without age information (100%, 95% CI 89.3-100%) 74 , participants that had undergone training in the forest (100%, 95% CI 89.3-100%, 2 studies), and participants in communities/migrant workers (100%, 95% CI 51.0-100%) 100 . Table 2 shows the proportion of asymptomatic Plasmodium infection per parasitised individual in different parts of Asia, countries, and participants' groups when molecular and nonmolecular methods were used.

Discussion
Asymptomatic infection is a disease reservoir that must be considered when determining the malaria risk for malaria-naive individuals and when developing malaria control and elimination strategies in areas transitioning from pre-elimination to elimination, such as in Asia. When the molecular method was used alone or in combination with other methods, the meta-analysis results showed that the prevalence of asymptomatic Plasmodium infection in Asia was approximately 6.3% and varied between 5.8% and 6.7%; meanwhile, the proportion of asymptomatic Plasmodium infection per all parasitised individuals was high at 84.5% and varied between 79.1% and 89.9%. These results indicate the low prevalence estimates of asymptomatic Plasmodium infection in Asian communities where malaria is of low endemicity. The low prevalence estimates of asymptomatic Plasmodium www.nature.com/scientificreports/ infection and a high proportion of asymptomatic cases may be due to the increasing use of highly sensitive methods for the detection of malaria in asymptomatic and submicroscopic individuals. Compared with studies that used molecular methods alone or in combination with other methods, lower pooled prevalence (3.4%) and proportion estimates (72.2%) of asymptomatic Plasmodium infection in Asia were documented when nonmolecular methods such as microscopy, RDT, or a combination of these two methods were used for the detection of malarial parasites. Consistent with the findings from a previous meta-analysis of the diagnostic accuracy and limited sensitivity of microscopy and RDTs for the detection of asymptomatic Plasmodium infections 104 , the pooled results of the present study confirmed that using the molecular method alone or in combination with other methods could increase the chances of detecting asymptomatic Plasmodium infection among participants. In Southeast Asian countries, the subgroup meta-analysis showed the lowest prevalence of asymptomatic Plasmodium infections among participants but the participants harbored the highest proportion estimate of asymptomatic Plasmodium infections as compared to other areas. The high proportion of asymptomatic Plasmodium infection among parasitised individuals in Southeast Asia may be explained by the fact that participants in this area, particularly the adult population, have been exposed to several episodes of Plasmodium infection and have acquired immunity against clinical malaria 105 . A previous study suggested that local transmission causes the rapid acquisition of immunity and a reduction in the frequency of symptomatic infections 106 . As asymptomatic Plasmodium infections are associated with submicroscopic parasite densities, the proportion of sub-microscopic infections is high, which may indicate that malaria control efforts have been successful and that some populations retain some residual immunity to the infection 107 . Although the prevalence and proportion estimates of asymptomatic Plasmodium infection in Southeast Asia were heterogenic, the homogeneity of the proportion estimate of asymptomatic Plasmodium infection was found in studies that used molecular methods in combination with non-molecular methods for the detection of asymptomatic Plasmodium infection in Thailand 3,27,28,35,64,88,92 . In other parts of Asia, such as Western Asia, there were limited studies that demonstrated asymptomatic Plasmodium infection, such as in Yemen, which showed a high prevalence (23%) 24 and in Iran, which showed a very low prevalence (0.2%) as detected by a combination of molecular and non-molecular methods 59 . This result indicated the heterogeneity of asymptomatic Plasmodium infection in these areas. The low prevalence of asymptomatic Plasmodium infection in Iran can be explained by the fact that Iran has almost succeeded in eliminating malaria from the country 59 . Meanwhile, the combination of molecular and non-molecular methods detected the high prevalence of asymptomatic Plasmodium infection in Yemen as compared to the use of non-molecular methods only 24 . Therefore, this result also indicated an increased sensitivity for the detection of asymptomatic Plasmodium infection when molecular methods are used in combination with microscopy or RDT. In South Asia, the prevalence of asymptomatic Plasmodium infection was higher in Bangladesh than in India or Pakistan and was the www.nature.com/scientificreports/ highest in Bhutan. Interestingly, asymptomatic Plasmodium infection was reported in areas where malaria was previously almost eliminated, such as Bhutan, due to the transmission of malaria across the Indian border 100,108 .
The systematic review and meta-analysis demonstrated that the highest prevalence of asymptomatic Plasmodium infection detected by molecular methods alone or in combination with microscopy or RDT was found among adults. This result indicated that asymptomatic Plasmodium infection was more frequently found among adults because members of the adult population in low transmission areas have been exposed to several episodes of Plasmodium infection and have acquired immunity against clinical malaria 105 . Age is one of the factors that contribute to asymptomatic Plasmodium infection 7,79,109 . Nevertheless, a previous study by van Eijk et al. 7 showed that young children (5 years old) were much more likely to have sub-microscopic P. vivax, which can lead to asymptomatic P. vivax infection. The infection with P. vivax may result in faster immunity acquisition, and asymptomatic P.vivax infection can occur at younger ages than P. falciparum infection 109 . The cause of the faster development of clinical immunity to P. vivax is unknown; however, frequent population exposure to hypnozoites may be a factor 110 .
The meta-analysis results indicated that using routine diagnostic methods in combination with the highly sensitive method would detect asymptomatic Plasmodium infection at a higher rate in this region. The highly sensitive tests for detecting asymptomatic Plasmodium infection with low density in monitoring malaria transmission www.nature.com/scientificreports/ intensity in low transmission areas aid in malaria control and elimination efforts. Although PCR is highly sensitive for the diagnosis of low-density parasitemia, it requires well-trained personnel and expensive laboratory facilities and is not suitable for surveys of asymptomatic Plasmodium infection in communities. Additional point-of-care testing is required to determine the true burden of asymptomatic Plasmodium infection in low -transmission settings. In Asia, it is possible that more cases of asymptomatic Plasmodium infection exist in other areas where malaria surveys have never been conducted, particularly in urban regions of Southeast Asia or China where the prevalence of malaria is low. If the malaria surveys using highly sensitive tests are conducted in those regions, the prevalence estimates of asymptomatic Plasmodium infection would be low, but the proportion estimate of asymptomatic Plasmodium infection would be high. Although the prevalence estimates of asymptomatic Plasmodium infection in Asia were low, particularly in Southeast Asia as compared to other areas of Asia, particularly South Asia, a small prevalence of infections was sufficient to restart malaria transmission, which disturbed malaria elimination goals in several countries. The evidence of the meta-analysis results also implies the need for active case detection in low malaria transmission settings as the transmission of malaria takes place locally. In addition, the aggressive prevention and control of infections by strengthening malaria intervention strategies such as the sufficiently wide distribution of insecticide-treated bed nets (ITN) and the coverage of indoor residual spraying (IRS) are crucially needed. Also, the mass treatment of the asymptomatic infection carriers before the transmission season is important to prevent transmission of infection to reservoirs and to other populations. The sensitive detection, aggressive prevention and control, and also mass treatment of asymptomatic infection reservoirs would be able to reduce the number of infections and disrupt malaria transmission 111 . Furthermore, the findings of the current study will serve as a source of up-to-date information for those who aim to conduct further investigation on asymptomatic Plasmodium infection in particular areas of Asia. The present systematic review has limitations. First, some studies did not show the exact prevalence of asymptomatic Plasmodium infection; hence, excluded from the meta-analysis. Second, some studies were excluded because they did not collect clinical information until the patient was positive for malaria by microscopy. Third, the community/participant selection may not have included participants from all age groups, minority ethnicities, and diverse occupations, which may have introduced bias into the asymptomatic Plasmodium infection prevalence estimates. Fourth, using only microscopy might cause false negatives, while using only RDT for Plasmodium detection might cause false positives, which might underestimate or overestimate the prevalence and proportion estimate of asymptomatic Plasmodium infection, respectively. Finally, geographical heterogeneity was a limitation of the study as the frequency of asymptomatic spread varies according to the epidemiological situation, transmission rates, and consequent host immunity.  www.nature.com/scientificreports/

Conclusions
Asymptomatic Plasmodium infection is a disease reservoir that should be taken into account. There was a low prevalence of asymptomatic Plasmodium infection in different parts of Asia, but there was a high proportion of asymptomatic Plasmodium infection among all parasitised individuals. These results may support and facilitate the elimination and control programs for asymptomatic Plasmodium infection in Asia.

Data availability
All data relating to the present study are available in this manuscript and supplementary files.