Misidentification of Plasmodium ovale as Plasmodium vivax malaria by a microscopic method: a meta-analysis of confirmed P. ovale cases

Plasmodium ovale is a benign tertian malaria parasite that morphologically resembles Plasmodium vivax. P. ovale also shares similar tertian periodicity and can cause relapse in patients without a radical cure, making it easily misidentified as P. vivax in routine diagnosis. Therefore, its prevalence might be underreported worldwide. The present study aimed to quantify the prevalence of P. ovale misidentified as P. vivax malaria using data from studies reporting confirmed P. ovale cases by molecular methods. Studies reporting the misidentification of P. ovale as P. vivax malaria were identified from three databases, MEDLINE, Web of Science, and Scopus, without language restrictions, but the publication date was restricted to 1993 and 2020. The quality of the included studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS). The random-effects model was used to estimate the pooled prevalence of the misidentification of P. ovale as P. vivax malaria by the microscopic method when compared to those with the reference polymerase chain reaction method. Subgroup analysis of participants was also performed to demonstrate the difference between imported and indigenous P. ovale cases. The heterogeneity of the included studies was assessed using Cochran's Q and I2 statistics. Publication bias across the included studies was assessed using the funnel plot and Egger’s test, and if required, contour-enhanced funnel plots were used to identify the source(s) of funnel plot asymmetry. Of 641 articles retrieved from databases, 22 articles met the eligibility criteria and were included in the present study. Of the 8,297 malaria-positive cases identified by the PCR method, 453 P. ovale cases were confirmed. The pooled prevalence of misidentification of P. ovale as P. vivax malaria by the microscopic method was 11% (95% CI: 7–14%, I2: 25.46%). Subgroup analysis of the participants demonstrated a higher prevalence of misidentification in indigenous cases (13%, 95% CI: 6–21%, I2: 27.8%) than in imported cases (10%, 95% CI: 6–14%, I2: 24.1%). The pooled prevalence of misidentification of P. vivax as P. ovale malaria by the microscopic method was 1%, without heterogeneity (95% CI: 0–3%, I2: 16.8%). PCR was more sensitive in identifying P. ovale cases than the microscopic method (p < 0.00001, OR: 2.76, 95% CI: 1.83–4.15, I2: 65%). Subgroup analysis of participants demonstrated the better performance of PCR in detecting P. ovale malaria in indigenous cases (p: 0.0009, OR: 6.92, 95% CI: 2.21–21.7%, I2: 68%) than in imported cases (p: 0.0004, OR: 2.15, 95% CI: 1.41–3.29%, I2: 63%). P. ovale infections misidentified as P. vivax malaria by the microscopic method were frequent and led to underreported P. ovale cases. The molecular identification of P. ovale malaria in endemic areas is needed because a higher rate of P. ovale misidentification was found in endemic or indigenous cases than in imported cases. In addition, updated courses, enhanced training, and refreshers for microscopic examinations, particularly for P. ovale identification, are necessary to improve the microscopic identification of Plasmodium species in rural health centres where PCR is unavailable.

Eligibility criteria. Observational studies that reported the prevalence of both P. vivax and P. ovale malaria by microscopy and PCR or molecular methods were screened for the misidentification of P. ovale as P. vivax malaria by a microscopic method. The inclusion criteria were (1) studies reporting the prevalence of the misidentification of P. ovale as P. vivax malaria by a microscopic method and (2) studies using PCR or molecular methods to confirm the Plasmodium species. The exclusion criteria were (1) case reports or case series that reported a small number of patients, which can lead to reporting bias for meta-analysis, (2) studies using both microscopic and molecular methods to identify Plasmodium species where the data could not be extracted, (3) studies carried out on the performance of tests as those tests attempted to develop new techniques or new tests for the detection of Plasmodium species, (4) experimental studies that aimed to explore the new finding related to Plasmodium species, (5) studies with no misidentification of P. ovale to P. vivax, or no P. vivax malaria was observed as those studies did not provide the evidence of the misidentification of P. ovale as P. vivax malaria, (6) review articles, (7) studies without the full text, (8) clinical trials, guidelines, studies using the same participants, and other studies without relevant data.

Study selection and data extraction. Two authors (MK and FRM) selected potentially relevant studies
according to the eligibility criteria. Any discordance in the study selection was resolved by consensus. Data selection from relevant studies was managed using Endnote software X7 (Clarivate Analytics, Philadelphia, USA). Data extraction was also performed by two authors (MK and FRM) and crosschecked by the third author (KUK). The following data were extracted: name of first author, year of publication, study area, years of the study, study design, age range (years), gender (male, %), participants (imported or indigenous), PCR for identified Plasmodium spp., target gene for PCR, number of malaria cases identified by microscopy and PCR methods, number of P. vivax cases identified by microscopy and PCR methods, number of P. ovale cases identified by microscopy and PCR methods, number of misidentifications of P. ovale as P. vivax malaria by microscopy, and number of Scientific Reports | (2020) 10:21807 | https://doi.org/10.1038/s41598-020-78691-7 www.nature.com/scientificreports/ misidentifications of P. vivax as P. ovale malaria by microscopy. The data were extracted to pilot-standardized sheets created using Microsoft Excel 2010 (Microsoft Corporation, Washington, USA) before meta-analyses.
Quality of the included studies. The quality of the individual studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) (Table S2) 16 . The tool was comprised of 4 domains: patient selection, index test, reference standard, and flow and timing. Each domain was assessed in terms of risk of bias and concerns of applicability 16 . The index test was microscopy, while the reference standard was PCR method. The results of the QUADAS of all included studies were summarized in the methodological quality graph and summary.
Outcomes. The primary outcome of the present study was the prevalence of misidentification of P. ovale as P. vivax, and also P. vivax as P. ovale malaria by the microscopic method. The secondary outcome was the performance of the PCR test to identify P. ovale malaria compared to that of the microscopic method. Consent for publication. All authors have read the manuscript and consent to its publication.

Results
Search results. The literature search identified 641 records through three different databases: 219 for MED-LINE, 230 for Scopus, and 192 for Web of Science (Fig. 1). After the removal of duplicate articles, the remaining 430 studies were screened. In total, 164 records were excluded because they were irrelevant. The 266 potentially relevant studies were assessed in detail and 245 studies were excluded for the following reasons: 51 studies evaluated the performance of tests or experimental studies, 43 were case reports or case series, 38 studies used both methods but we were unable to extract the data, 33 had no misidentification of P. ovale to P. vivax or no vivax malaria, 22 had no report of P. ovale malaria, 15 were review articles, 12 were sub-microscopic P. ovale infections, 10 determined the prevalence of P. ovale using only microscopy, 6 determined the prevalence of P. ovale using only the PCR method, 6 were P. ovale positive samples with no misidentification, 4 were studies had no full-text, 2 were clinical trials, 1 was guidelines, 1 was on mosquito surveillance, and 1 study used the same participants. Twenty-one studies  were included because they met the eligibility criteria, and one additional study 38 was selected after reviewing the reference lists of the 20 included studies and review articles. Finally, a total of 22 studies 17-38 were included for qualitative and quantitative syntheses.
Characteristics of the included studies. All characteristics of the included studies can be found in www.nature.com/scientificreports/ The pooled prevalence of the misidentification of P. ovale as P. vivax malaria. The pooled prevalence of the misidentification of P. ovale as P. vivax malaria was estimated using all 22 included studies (Fig. 3).
The performance of PCR to identify P. ovale malaria compared to that of the microscopic method. The performance of PCR to identify P. ovale malaria versus that of the microscopic method was estimated using the random-effects model (Fig. 5).

Publication bias. Visual inspection of the funnel plots demonstrated some small study effects that caused
an asymmetric distribution of studies in the plots between the OR and SE (logOR) (Fig. 6) www.nature.com/scientificreports/ due to publication bias or other factors. The results showed that most of the included studies were located in the significant area of the plot (p < 0.01), indicating that publication bias was the cause of the asymmetric distribution of the funnel plot (Fig. 7).

Discussion
The microscopic method for identifying Plasmodium species is still considered the gold standard method for malaria diagnosis in clinical laboratories. However, its limitation is its low sensitivity to detect malaria parasites that are present with a low parasite density [39][40][41] . The sensitivity of microscopy under optimal conditions is limited to approximately 10-50 parasites/μl of blood 42 . In contrast to microscopy, PCR has the advantage of higher sensitivity and is capable of detecting less than 10 parasites/µl of blood [43][44][45] . In addition to low sensitivity, microscopy also has a low specificity or inability to distinguish the morphologically similar P. vivax and P. ovale malaria even by a well-trained or expert microscopist examining blood films. This is the first systematic review and meta-analysis to quantify the misidentification of P. ovale as P. vivax by microscopic methods. The prevalence of misidentification of P. ovale as P. vivax malaria was high (11%), particularly in P. ovale endemic countries (13%). A previous study suggested that misidentification of P. vivax with P. ovale was likely due to the infection of P. ovale resulting in a low parasite density compared to that of P.  www.nature.com/scientificreports/ falciparum 18 , and the misidentification of P. ovale as P. vivax is more frequent than the misidentification of P. ovale as P. falciparum 29 . A previous study included laboratories in hospitals and demonstrated that participants had much more difficulty identifying P. ovale with 100% failure rates, while difficulty identifying P. malariae (22.5% failure) and P. vivax (21.7% failure) was lower 46 . The difficulty in identifying P. ovale malaria was also observed by a study on laboratories in the United Kingdom 47 . In addition, a survey study of 19 provincial laboratories in China with a total of 168 staff members also demonstrated that P. ovale was likely to be misdiagnosed as P. vivax by microscopy 48 . The external quality assessment (EQA) conducted in Senegal, which was part of the national malaria control program (NMCP), demonstrated the misidentification of a P. ovale slide as P. vivax by experts 49 . Moreover, microscopists participating in post training on the proficiency of laboratory technicians in Plasmodium species identification could misidentify P. vivax as P. ovale malaria 50 . Although the prevalence of misidentification of P. ovale as P. vivax was high, the treatment of these two species with chloroquine and radical cure with primaquine to eliminate the liver stages were similar. Imported malaria in non-endemic countries continues to be reported worldwide as international travel or immigration from endemic zones has increased [51][52][53][54] . Therefore, malaria diagnostic tests with high sensitivity and specificity to identify Plasmodium species among travellers are necessary. Due to the low sensitivity of RDTs to identify P. ovale 27,55,56 , molecular techniques such as PCR are recommended to identify P. ovale cases, although they have been shown to miss some P. ovale cases 57 . Moreover, molecular methods such as semi-nested PCR more accurately detected P. ovale mixed infections than microscopy 28 . With more advances in molecular methods for the detection and identification of malaria parasites, real-time PCR methods using fluorescent labels for detecting and quantifying DNA targets have been developed with high sensitivity and specificity 23,58,59 . Interestingly, multiplex real-time PCR failed to detect P. falciparum and P. ovale mixed infection in a previous study because of the high difference in the ratio between P. falciparum and P. ovale (> 1000:1) 32 . These results suggested that the misidentification of P. ovale as P. vivax in several studies might be caused by bias across the considerable  www.nature.com/scientificreports/ prevalence of the main endemic Plasmodium species, by changes in the morphology of the parasite during specimen storage or treatment, or by very low parasitaemia levels. In addition, there is a strong perception that P. vivax is rare in subtropical Africa 60 . Therefore, the identification bias in retrospective studies of imported malaria, where the patients are considered to have acquired their infection in a subtropical African country, is because there is a clear bias in these cases to identify any non-falciparum or non-malariae case as P. ovale can occur. Thus, the number of P. vivax cases misidentified as P. ovale was quantified. The results showed that the misidentification of P. vivax as P. ovale in imported countries was very low (1%). This result indicated a lower possibility of P. vivax being misidentified as P. ovale (1%) than those of P. ovale being misidentified as P. vivax (11%) in imported cases. The major concern of the misidentification of P. ovale as P. vivax in imported cases should be addressed. Therefore, the molecular method can make a decisive contribution to the identification of a less common Plasmodium species or two Plasmodium species with similar morphologies. Taking into account the high sensitivity and specificity of molecular methods for identifying malaria parasites, molecular methods are labour intensive and have a greater potential for contamination and long turnaround times for routine diagnosis, and are not convenient for use in remote settings. Updated courses and intensified training of microscopic examinations are necessary to improve the microscopic identification of Plasmodium species in rural health centres, where molecular techniques are unavailable.
The present study have limitations. First, limited numbers of P. ovale were identified and reported, as it is a neglected Plasmodium species. Therefore, the pooled prevalence of the misidentification of P. ovale as P. vivax malaria might not represent all misidentification that occurred. Second, several studies performed microscopy and PCR to confirm malaria infection, but could not be included in the present study as the necessary data cannot be extracted and the comparison between microscopy and PCR was either not clearly presented or not provided.

Conclusion
Misidentification of P. ovale infections as P. vivax malaria by microscopic methods are frequent and lead to the underreported status of P. ovale cases worldwide. The molecular identification of P. ovale malaria in endemic areas is necessary to provide data for malaria elimination because a higher rate of P. ovale misidentification was found www.nature.com/scientificreports/ Figure 5. The performance of PCR to identify P. ovale IV: Inverse Variance, CI: Confidence Interval, Event: number of patients with P. ovale, random: random effects model, Total: number of all P. ovale cases, Lower in PCR: the proportion of P. ovale cases detected by PCR was lower than those detected by the microscopic method. Higher in PCR: the proportion of P. ovale cases detected by PCR was higher than those detected by the microscopic method. The performance of PCR to identify P. ovale malaria was analysed using Review Manager 5.3 (The Cochrane Collaboration, London, UK) available at https ://train ing.cochr ane.org/.