This scoping review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) checklist as well as general related guidance ( Supplementary File 1 ) (Pham et al., 2014; Tricco et al., 2018). The project was registered with the Open Science Framework (https://osf.io/shku4/) (Foster and Deardoff, 2017).

The scoping review approach is considered particularly relevant and useful for analysing, identifying, and synthesising a range of complex health public concerns, such as the management of relapsing malaria (Munn et al., 2018; Peters et al., 2020). The research question examines the literature that exists on the burden and treatment of recurrent Pv malaria in India, with an emphasis on relapses. Second, identifying causes/determinants of relapses in the Indian context was also addressed. Third, we also investigated the efficacy and safety of primaquine for the radical cure of Pv malaria infections. The knowledge gaps, practical implications, solutions, and future prospects of Pv malaria treatment are also discussed.

The searches were conducted from September to December 2022 and March to April 2023 without any restriction period in six electronic databases, including PubMed, Wiley Online Library, clinicaltrials.gov, the International Clinical Trials Registry Platform, ScienceDirect, ResearchGate and Google scholar. Official websites of local journals and scientific associations were also scrutinised. Search terms were used in combination with Boolean terms (i.e., AND, OR) ( Table 1 ). The full texts of open access publications were retrieved from each electronic database. Principal investigators and editors-in-chief of local journals were contacted by email or phone call to request a full-length paper and/or more details on the relevant studies. The papers were purchased in case of absence of feedback or refusal. Full texts were scrutinised for eligibility in the PRISMA-ScR. Also, the reference list of relevant papers was examined to find more potentially relevant papers.

Only papers published from 1913 to 2022 and written in English were included. Quantitative and interventional studies were considered of interest if they addressed any aspect of recurrent Pv malaria including type (i.e., recrudescence, reinfection, and relapse), prevalence, clinical presentation, biology, determinants, outcomes, diagnostics, prevention, and treatment. The cut-off of 1913 was chosen as the oldest Indian medical journal, i.e., the Indian Journal of Medical Research, started its first releases circa 1913.

We excluded studies i) with a qualitative-based design (narrative review, conference, letter, correspondence, comment), ii) protocol studies, iii) conducted outside India, iv) having evaluated antimalarial drug efficacy against Pv malaria without addressing Pv recurrences, v) having evaluated antimalarial drug efficacy against recurrent Pf malaria, vi) having evaluated presumptive treatments of Pf and/or Pv infections, and vii) having used animal models. Titles and abstracts of potentially relevant papers were evaluated independently by the authors. Some publications were excluded at this step of the screening strategy, while duplicates were removed.

First, one author (L.P.K.F.) initially extracted the relevant information from eligible studies using an Excel standardised form. The data charting process mapped study findings according to the attributes which could be categorised into five groups namely i) study details (first author’s name, year of publication, paper title, collection area, malaria endemicity, state/territory union and year of data collection); ii) design details (setting, study design, study population, parasitological diagnosis method, definition used to identify Pv relapses); iii) drug details (number of treatment arms, number of patient in each treatment arms, number of anti-relapse regimens, comparative control regimen, nature of blood schizonticidal and liver anti-relapse drugs, dose of blood schizonticidal drug, total dose of liver anti-relapse drug, daily dose and duration of anti-relapse treatment, G6PD profile, method used to determine blood level of anti-relapse drug); iv) follow-up details (type, frequency and duration), and v) outcomes (number of recurrent parasitaemia events in each treatment arms, number of recurrent events in each treatment arms, month of recurrent/relapse event in each treatment arms, sociodemographic and clinical characteristics of patients with recurrent event, number of haemolytic anaemia event). The recurrent event included relapse, recrudescence, and reinfection. The design of this extraction form was piloted and refined by all authors.

Data were explored and summarised in congruence with scoping review objectives, along with narrative analysis. The selected evidence based on the source, study characteristics, and major findings were mapped and presented in tabular and/or graphical forms. For instance, the geographical distribution of studies was mapped by Indian states/union territories. Also, pie charts were used to present the distribution of included studies by setting (research institute, health facility, and community), study design (case report, randomised clinical trial – RCT, and prospective study), and study population (children, adults, and general population). The term “not specified” was used if details on one of the above-mentioned items were not mentioned in the included studies. PQ total dose was classified as very low (≤ 2.5 mg/Kg), low (> 2.5 – < 5 mg/Kg) and high (≥ 5 mg/Kg) as described earlier (John et al., 2012). The multiplication of blood stages of early relapses is suppressed by slowly eliminated drugs, and thus, relapses become microscopically patent 3 weeks after QN or artesunate, 3 – 6 weeks after artemether + lumefantrine, and 5 – 7 weeks after CQ treatment (White, 2011; Chu and White, 2021). In this context, we categorised Pv malaria recurrences and/or relapses as frequent (~3 – 7 weeks) and long (~8 – 9 months).

We used the approach proposed by Commons and colleagues to appraise the proportion of Pv recurrences caused by relapse, with slight modifications (Commons et al., 2020). In this approach, the primary outcome was the incidence rate of Pv recurrences over 365 days following a given treatment. The incidence rate ratio (IRR) was calculated as the ratio of the incidence rate of Pv recurrences in the treatment arm with PQ to the ratio of incidence rate of Pv recurrences in the treatment arm without PQ. The minimum proportion of recurrences due to relapse was calculated by subtracting the IRR from 1 (Commons et al., 2020). This approach assumes that PQ has an anti-relapse preventive efficacy of 100% and does not improve the killing effect of asexual blood stage parasites (Commons et al., 2020).

In this section, only RCTs that included PQ in one of the regimen arms, followed up patients actively regardless of duration of follow-up, supervised PQ administration daily, and had an extractable incidence rate were included to assess Pv relapse burden. We used these stringent criteria to reduce any ambiguity about the distinction between reinfection, recrudescence, and relapses and thus give a consistent estimate of the proportion of Pv recurrence attributable to relapse in the Indian context. Also, Pv relapse estimates as calculated in each individual study were extracted and stratified by definition used for relapse, time of follow-up, and PQ dosing.

Two reviewers (L.P.KF. and V.S.) independently assessed the bias risk of the RCTs included to determine Pv relapse burden. The methodological quality of RCTs was evaluated using Joanna Briggs Institute - JBI critical appraisal tools designed for RCTs (https://jbi.global/critical-appraisal-tools). This tool consists of 13 questions on four types of methodology-related bias viz. i) selection, ii) performance, iii) attrition, and iv) reporting. A choice between four answers (“Yes”, “No”, “Unclear” and “Not applicable”) was proposed for each question as per the JBI tool. Thus, the risk of bias was categorised as “low” if the answer was “Yes”, “high” if the answer was “No” and “unclear” if the answer was “Unclear or Not applicable” (Arya et al., 2021). Disagreements between reviewers were resolved through discussion and consensus.

The data used in this review was retrieved from publicly accessible databases. Thus, ethics committee approval was not requested.

A total of 2,106 potentially relevant records were retrieved from electronic databases. Of these records, 1504 were retained after removing duplicates. We excluded 1478 records after analysing titles and abstracts based on the above mentioned exclusion criteria: i) studies conducted outside India (n = 154), ii) studies on recurrent Pf malaria (n = 96), iii) anti-relapse drugs were not tested (n = 7), iv) reviews (n = 5), v) Pv relapses were not addressed (n = 5), vi) studies on presumptive malaria treatment (n = 4), vii) protocols (n = 3), viii) correspondence (n = 1), ix) animal-based studies (n = 1), full-text not found (n = 1), and xi) irrelevant studies (n = 1201) ( Figure 2 ). Twenty-six publications were evaluated for eligibility. We identified three more publications through the bibliographic references listed in the 26 publications, thereby giving a total of 29 potentially eligible publications. Of these, two records were excluded because of evidence of duplicate (i.e., same findings published in different journals). Thus, 27 studies involving ~27,000 Pv-infected patients treated with PQ-based treatment regimens were finally included to analyse Pv recurrences ( Figure 2 ) (Singh et al., 1953; Basavaraj, 1960; Sharma et al., 1973; Roy et al., 1977; Roy et al., 1979; Appavoo et al., 1984; Sinha et al., 1989; Sharma et al., 1990; Prasad et al., 1991; Srivastava et al., 1996; Gogtay et al., 1998; Gogtay et al., 1999; Adak et al., 2001; Dua and Sharma, 2001; Yadav and Ghosh, 2002; Rajgor et al., 2003; Imwong et al., 2007; Saifi et al., 2010; Kim et al., 2012; Rajgor et al., 2014; Pareek et al., 2015; Savargaonkar et al., 2015; Kumar et al., 2016; Savargaonkar et al., 2017; Saravu et al., 2018; Kishore et al., 2020; Gandrala et al., 2022). In order to determine the proportion of Pv recurrences due to relapse, 24 studies were excluded due to three reasons: 21 studies were not designed as RCTs, 2 studies evaluated PQ-based treatment regimens, and recurrence data were not extractable in one study ( Figure 2 ).

Details of the included studies are presented in Figure 3 and Supplementary File 2 . Studies included were published over changes in PQ drug policies in India, with 23.1% of them published from 1960 – 1990, during which Pv infections were treated using CQ given as a single dose followed by PQ at a dose of 0.25 mg/Kg for five days ( Figure 3A ). More than one third (38.5%) of the studies were published from 2010 – present. Most studies were conducted in Maharashtra (n = 6), Delhi (n = 6), Uttar Pradesh (n = 4), and Karnataka (n = 4) ( Figure 3B ). The bulk of studies were conducted in a health facility setting (55.5%), designed as prospective studies (66.4%), and included patients of all ages (57.7%) ( Figures 3C–E ).

CQ was the main blood-stage antimalarial drug associated with PQ regimens, while some studies used artemisinin-based combination therapies (ACT), amodiaquine (AQ), quinine (QN), and pyrimethamine (PYR). The majority of studies (n = 22, 81.5%) evaluated CQ + PQ to prevent Pv recurrences. Other studies associated PQ with other 8-aminoquinolines such as pentaquine, pamaquine and AQ while one study evaluated the ACT + PQ association to prevent Pv recurrences ( Table 2 ). Two studies evaluated the anti-recurrence potential of different PQ monotherapies (i.e., standard and sustained release) (Singh et al., 1953; Pareek et al., 2015). Another study evaluated the anti-relapse effect of bulaquine, a PQ analogue, in patients from Delhi (Adak et al., 2001).

PQ posology greatly varied between studies, with the majority of studies testing very low PQ dose regimens (total dose < 2.5 mg/Kg). PQ was administered daily mostly for 5 days (n = 16, 59.3%), followed by daily administration for 14 days in 10 studies (37%), and 7 days in 3 studies (11.5%) ( Table 2 ). Patients were either followed up actively or passively for durations ranging from 180 to 1460 months. Regarding active follow-up, patients were followed up either fortnightly, weekly or every 4/6/8 weeks.

Light microscopy (LM) was invariably used for Pv detection in patients, while some studies coupled it with polymerase chain reaction (PCR), rapid diagnostic test (RDT), and quantitative buffy coat (QBC). While if G6PD status of patients was not specified or evaluated in the bulk of the study, we noted that recent studies either excluded G6PD-d patients or evaluated G6PD status ( Table 2 ). In addition, two studies determined the blood level of PQ using liquid chromatography – mass spectrometry (LC-MS) and reverse high-pressure liquid chromatography (HPLC) while the technique was not specified in one study (Rajgor et al., 2003; Saravu et al., 2018; Gandrala et al., 2022).

The approach used to analyse recurrent infections was either not specified or inexistent in 34.6% and 26.9% of the included studies, respectively. For the remaining studies, six types of approaches were proposed, and based on patterns of recurrences following the primary infection, genotyping, parasitological evidence, clinical symptoms, and mathematical modelling ( Table 3 ).

A genotyping-based approach relied on molecular amplification and/or sequencing of genetic markers (genes, microsatellites) in order to distinguish between relapse (heterologous infections) and recrudescence (homologous infections). The assumption behind this approach is that recurrent infections are classified as relapses if they are genetically distinct from primary infections. One study from Karnataka used molecular markers such as microsatellites (MS7 and MS10) to analyse Pv recurrence in adults treated with CQ + PQ (Saravu et al., 2018). In an RCT conducted in Maharashtra, genetic markers were associated with a linear regression-based mathematical model to determine the attributable fraction of Pv recurrence caused by relapse in patients treated with CQ alone or in association with PQ for 5 and 14 days ( Table 3 ) (Kim et al., 2012).

Before the advent of molecular methods, earlier studies identified relapses using an approach based on patterns of recurrences, clinical symptoms, and parasitological evidence. In Uttar Pradesh, one study defined relapses as follows: “Patient became positive again, a few weeks after clearance of parasitaemia and in spite of a full course of CQ and PQ, associated with the analysis of periodicity to distinguish relapses from fresh infections” (Sinha et al., 1989).

Few studies analysed demographic characteristics of Pv recurrences and only data was extracted from five community-based studies and one case report conducted in Karnataka, Uttar Pradesh, Gujarat, Delhi, and Orissa states (Basavaraj, 1960; Sinha et al., 1989; Prasad et al., 1991; Srivastava et al., 1996; Yadav and Ghosh, 2002; Savargaonkar et al., 2017), even though several attempts were made to contact the corresponding and supervising authors of unavailable studies. Most Pv recurrences were found in males and those aged > 14 years old. In Karnataka, the proportion of Pv recurrences was higher in males (71.8%) and patients aged > 14 years (53.8%) compared to their female and younger counterparts, respectively (Basavaraj, 1960). Likewise, two studies from Uttar Pradesh reported a higher proportion of males (≥ 66%) and aged > 14 years (≥ 60%) (Sinha et al., 1989; Prasad et al., 1991). In contrast, Yadav and Ghosh found no statistically significant difference in Pv recurrence proportion between males and females in Orissa (Yadav and Ghosh, 2002).

The number of Pv recurrences with regard to treatment arm and latency type is presented in Figure 4 . Recurrences were observed for all arm treatments analysed (CQ, PQ, AQ + PQ, QN + Pentaquine, CQ + PYR, and CQ + PQ). Most Pv recurrences had intermediate and long latency. In patients treated with PQ monotherapy, Pv recurrences had frequent latency, while Pv recurrences had intermediate and long latency in those treated with CQ + PYR ( Figure 4 ).

As stated previously, three studies met inclusion criteria to determine the fraction of Pv recurrences caused by relapse (Rajgor et al., 2003; Kim et al., 2012; Rajgor et al., 2014). Although similarities between the three RCTs were noted for a place (Maharashtra), design (open label RCT), control arm (CQ) and follow-up (active), there were some differences found for population, number of treatment arms, PQ total dosing evaluated, duration of follow-up, parasitological method, and determination of PQ blood level ( Supplementary File 3 ). For instance, patients were followed-up for 180 days in two studies and 450 days in the remaining study. Only one study determined the blood level of PQ in its design. Overall, the risk of bias was low for all evaluation components, with the exception of concealment and blinding to drug treatment ( Supplementary File 4 ).

In the included studies, estimates of the proportion of Pv malaria recurrences that relapsed were based on different approaches (genotyping, mathematical model, temporal pattern of recurrence). Using the Commons and colleagues-based adapted approach, the proportion estimates was ~0 – 60% across the three RCTs included. These estimates were higher than those found based on approaches defined by the authors of each individual study. Using two PCR-based approaches, Rajgor et al. found Pv relapses at proportions of 6% and 2 – 5% in CQ + PQ-treated adults from Maharashtra ( Table 4 ) (Rajgor et al., 2003; Rajgor et al., 2014).

In one study from Delhi, PQ was administered at 0.75 mg/Kg/week for eight weeks in a 23-year-old male with G6PD-d who had presented three episodes of relapses 4, 5 and 6 months after the primary Pv infection. No haemolytic anaemia event was reported in the patient (Savargaonkar et al., 2017).