Introduction

Malaria is a prevalent parasite disease in humans, caused by many Plasmodium species, including P. falciparum, P. vivax, P. ovale, P. malariae, and P. knowlesi1. Notwithstanding substantial progress in malaria management, drug-resistant parasites present a formidable obstacle2. Approximately 2 billion people worldwide are at risk of malaria, resulting in 1.5 to 2.7 million fatalities per year3.

Pakistan continues to bear a significant portion of the difficulty in combating malaria, with an estimated one million cases annually4. The devastating floods in Pakistan in 2022 exacerbated malaria spread and damaged healthcare services and livelihoods. This year’s (2024) estimate indicates an eightfold increase in cases in Pakistan from 2021 to 2023, rising from approximately 506,000 to 4.3 million5. The majority of malaria cases in the 2022 outbreak were diagnosed as P. vivax which is caused by mosquitoes’ bite, and owing to its propensity to bite in the mornings and outdoors, is especially resistant to insecticides and other preventative efforts6. P. vivax malaria is more challenging to diagnose than P. falciparum due to the lower quantity of parasites normally found in the bloodstream of an affected person. P. vivax has the potential to induce illness weeks to years following exposure, as a result of the activation of dormant hypnozoites residing in the liver. The delayed onset of symptoms often occurs long after returning from endemic areas, which can obscure clinical suspicion for malaria and consequently delay diagnosis7,8. RDTs exhibit reduced sensitivity for P. vivax compared to P. falciparum, which heightens the likelihood of overlooked diagnoses, especially in instances of low parasitemia9,10. Since the hypnozoite phase cannot be detected using current diagnostic testing, it often leads to a missed detection of asymptomatic carriers and recurring infections until an infection is detectable through the presence of blood-stage parasites11.

P. vivax had traditionally been viewed as a “benign” form of malaria; however, an increasing amount of literature is challenging this view for many researchers in the field of malaria research, citing increased numbers of reported life-threatening conditions or mortality associated with the P. vivax species. A systematic review and meta-analysis which included studies from India over the last twenty years found that the pooled percentage of patients suffering from severe vivax malaria (SVM) was 29.3%12. This finding aligns with reports from other endemic regions in Latin America yet is higher than those observed in Eastern Africa13,14,15.

Pakistan’s national guidelines for the treatment of Plasmodium vivax malaria advocate a three-day regimen of chloroquine (CQ) to eradicate blood-stage parasites, succeeded by a 14-day administration of primaquine (PQ) to address dormant liver forms (hypnozoites) and avert relapse16. The World Health Organization (WHO) has approved Artemether/Lumefantrine (AL) as an artemisinin-based combination therapy (ACT) for the treatment of uncomplicated malaria by any plasmodium species17. AL has shown efficacy rates surpassing 90% at Day 28 in multiple clinical studies conducted in diverse endemic environments18,19,20. AL is commonly employed to manage clinical malaria cases where Plasmodium species remain undetermined through diagnostics, as well as in instances of mixed-species infections involving P. vivax and P. falciparum17. Conventional chloroquine treatment is no longer efficacious in many regions due to the high prevalence of chloroquine drug-resistant P. vivax20. The initial case of vivax resistance was identified in Pakistan in 2015 in a pregnant woman, and it was treated by switching to a combination of artemether and lumefantrine21.

The distinctions among relapse, recrudescence, and reinfection are critical in understanding how malaria causes recurrence once an individual has recovered. Relapse results from the re-emergence of malaria as a result of the re-activation of previously “dormant” liver stage Plasmodium parasites (hypnozoites) e.g., P. vivax or P. ovale22. These “dormant” parasites can remain within the liver for extended time frames, i.e., weeks to months, at which point they could generate another blood stage infection during the later stages of an initial recovery from malaria. Recrudescence occurs when malaria returns as a result of the continued existence of blood stage plasmodium parasites remaining in an individual who was treated for malaria but not sufficiently to eliminate all of the parasites23. This often results from insufficient or incomplete therapy and is not attributed to liver-stage reactivation. Reinfection occurs when a person contracts malaria anew from a different plasmodium parasite after recovering from the initial infection, frequently as a result of inadequate immunity or continued exposure to infected mosquitoes23.

Since the onset of the malarial epidemic in 2022, Pakistan has encountered numerous challenges, including the rise in drug resistance and the scarcity of certain drugs, particularly chloroquine and primaquine24,25. Numerous patients encounter challenges in obtaining chloroquine, leading many physicians to prefer AL over CQ. On the other hand, difficulty in obtaining primaquine results in malarial relapse25. Therefore, the aim of this research was to devise a study to evaluate efficacy of chloroquine vs. artemether/lumefantrine with or without primaquine for the treatment of Plasmodium vivax malaria.

Methods

The time frame for this study was September 2023 to August 2024, at Dr. Ziauddin University Hospital in Karachi, Pakistan. The study began in September 2023, as did the recruitment process that continued until February 2024. All participants were followed up for a total of 6 months and the final follow-up for all participants was completed in August 2024.

Inclusion and exclusion criteria

Participants were included in the study based on their age (18–65 years), gender and diagnosis of P. vivax malaria as an outpatient, emergency department or in-patient within the time frame of the study. Participants were eligible to be included in the study if they had received a positive malaria diagnostic test for P. vivax. This was determined by either light microscopy of thick/thin blood smear samples or by rapid diagnostic tests (RDTs) using the same venous blood sample26. Rapid Diagnostic Tests (RDTs) used in the current study were able to detect one or more of the following proteins: Histidine-Rich Protein II (HRP2); for the detection of P. falciparum and Plasmodium lactate dehydrogenase (pLDH); to specifically identify P. falciparum and/or P. vivax. RDTs utilize immunochromatographic lateral flow technology and comprise a nitrocellulose wick enclosed in a dipstick format27. One end of the test strip has labeled antibodies and a reagent to lyse erythrocytes; a blood sample (1 to 20 mcL) and buffer are introduced, facilitating the migration of the liquid along the strip through capillary action, accompanied by the labeled antibodies. The strip includes a test line (binding antibody that attaches to the parasite antigen, if present) and a control line (bound antibody that binds the migrating-labeled antibody to verify sufficient flow). The normal development duration is 15 to 20 min. All participants provided written informed consent before enrollment in the study.

The exclusion criteria for this study included patients who tested positive for mixed malarial species on microscopy or rapid diagnostic tests, those with co-infections, such as dengue or typhoid fever, and individuals with a history of malaria within six months of the presentations. Additionally, pregnant females and patients with glucose-6-phosphate dehydrogenase (G6PD) deficiency were excluded for participation in the study28. Furthermore, patients who had recently used antimalarial medications for self-medication before enrollment were excluded.

Ethical approval

Before this research began, Dr. Ziauddin University Hospital in Karachi, Pakistan’s Research and Ethical Committee issued ethical approval, which has the reference number R-62/ZU; it also complies with the ethical principles outlined in the Declaration of Helsinki.

Sample size

A power analysis was performed utilizing a 95% confidence interval (CI) and an 80% power level, which are commonly employed in clinical studies to guarantee the statistical strength of the findings. The calculation for the required sample size to identify a 10% difference, ensuring a power of 80% and a significance level of 0.05, was conducted using the appropriate formula for comparing two proportions. The analysis indicated that a necessary sample size of around 300 participants (150 in each group) was needed, considering possible dropouts and loss to follow-up. To enhance the study’s generalizability and minimize the likelihood of Type II error (i.e., not identifying a true effect), we included a total of 354 participants.

Study drug administration

After enrolling eligible participants, patients were administered either CQ 600 mg on Day 1, 600 mg day 2, and 300 mg on day 3 or AL 80/480 mg twice daily for three days in patients with weight ≥ 35 kg18, depending on the attending physician’s preference, cost considerations, and contraindications to treatments. This non-randomized methodology mirrors actual clinical practice and facilitates the comparison of therapy efficacy. The chosen medication was administered according to the structured protocol19,29. All doses of CQ and AL were supervised by attending staff nurses and duty doctors for hospitalized patients, whereas only the initial doses for outpatients were supervised. PQ was administered with a dose of 0.5 mg/kg daily for 14 days30 after screening for G6PD deficiency. In hospitalized patients, PQ administration was supervised, whereas outpatient patients self-administered the medication. The same treatment that was allocated at enrollment was administered in the event of malaria recurrence29.

After receiving treatment, patients were observed for 60 min to ensure their safety and to monitor any adverse reactions or vomiting episodes. Patients who vomited within the first 30 min, received a full dose of the prescription again. For those who vomited between 30 and 60 min, a half dose was administered.

Treatment and follow-up

Treatment follow-up is shown in Fig. 1. Participants in the study were required to attend daily evaluations for the first three days, followed by weekly assessments until day 28. After that, they were required to attend monthly evaluations for an additional five months as long as their initial presentation took place in the outpatient department. Those who were admitted to the hospital had weekly follow-up appointments for one month, followed by monthly appointments for the next five months. Additionally, patients who experienced symptoms suggestive of malaria were requested to return to the clinic. Those who developed P. vivax malaria again after 14 days of treatment were treated with the same medication prescribed at the time of enrollment. Follow-up appointments were scheduled based on the initial schedule31. At the end of the follow-up periods, those who did not receive primaquine were also given full fourteen-day regimen of primaquine.

Clinical procedures

A series of medical check-ups and blood tests including complete blood count with peripheral blood films were carried out on days 1, 2, 3, 7, 14, 21, and 28, followed by monthly evaluations for a period of 5 months. During each visit, any adverse reactions to the medication and other medications currently being taken were recorded, and the levels of hemoglobin (Hb) were measured on days 3, 7, 14, 21, and 28 for consecutive assessments31. The primary safety goal of this study was to quantify the frequency of mild and severe adverse events (AEs) in patients, compute the fractional variation in hemoglobin (Hb) levels, ascertain the percentage of patients whose Hb levels decreased by more than 25% between day 7 and baseline, and assess the percentage of patients who had anemia (Hb levels below 100 g/l) on 3rd day. A Hb value that exceeded the baseline measurement on day 28 was designated as the benchmark for hematological recovery32.

G6PD status was determined using a fluorescent spot test. The hemoglobin concentration was assessed using the HemoCue Hb 301 device. A fluorescent spot test, which determines whether blood samples exhibit fluorescence when exposed to ultraviolet light, was used to evaluate G6PD deficiency23. Bright fluorescence (qualitative) indicates adequate enzyme activity in a normal G6PD outcome. A deficient result indicates no fluorescence, which suggests a significant risk of hemolysis with primaquine treatment, whereas an intermediate deficit manifests as decreased fluorescence. Primaquine was not administered to patients with intermediate or inadequate findings in order to avoid hemolytic consequences. Blood film examinations were conducted by two microscopists with the appropriate certification. Each slide was analyzed separately by microscopists. If the blood smears resulted in inconsistent findings, such as discrepancies in species identification, or disagreement on the positivity of the smear, a third microscopist re-evaluated the smears. Parasite density was calculated by averaging the two closest counts obtained by the microscopists28. Parasite density was determined using the formula: parasites/µl = (number of counted parasites × 8000) / number of counted WBCs. Two qualified microscopists conducted independent examinations of blood films as described previously.

Classification of treatment outcome

The primary goal of this research was to evaluate the frequency of P. vivax relapse 28 days after the initial malaria episode. This was accomplished by comparing the treatment outcomes of the two groups: one receiving AL with PQ and the other receiving CQ with PQ. Secondary results, such as fever resolution, parasite clearance, cumulative risk, and recurrence rate, were also examined at the end of the study32.

Statistical analysis

Analyses were conducted in the enrolled cohort of eligible participants who received one of the four antimalarial regimens at baseline. The primary time-to-event endpoint was time to first Plasmodium vivax recurrence, defined as the interval from completion of antimalarial therapy to parasitologically confirmed P. vivax infection occurring ≥ 14 days after treatment. Time-to-event analyses were performed using Cox proportional hazards regression to estimate hazard ratios (HRs) and 95% confidence intervals (CIs); unadjusted Cox models are presented as the primary estimates in the main manuscript. Kaplan–Meier methods were used to summarize recurrence-free survival, and cumulative incidence of recurrence at Day 28 and at the end of follow-up was calculated as 100 × (1 − S(t)). Incidence rates were calculated as recurrent P. vivax events divided by total person-years of observation, and incidence rate ratios were estimated using negative binomial regression. Continuous variables were compared using non-parametric tests (Kruskal–Wallis for comparisons across treatment groups; Wilcoxon–Mann–Whitney for pairwise comparisons when applicable), and categorical variables were compared using the χ² test or Fisher’s exact test, as appropriate. A two-sided P value < 0.05 was considered statistically significant33.

Baseline balance and weighting

Propensity scores were estimated using a gradient-boosting model including age, weight, temperature, hemoglobin, sex, and G6PD status. Inverse probability of treatment weights (IPTW) were derived to explore confounding control, and covariate balance before and after weighting was assessed using absolute standardized mean differences (ASMDs) (Supplementary Table S1 and Figure S1). IPTW-adjusted analyses were performed as sensitivity analyses; where residual imbalance remained (notably baseline temperature and hemoglobin).

Treatment resistance

Treatment failure/resistance was defined as persistent or increasing parasitemia beyond 72 h after treatment initiation requiring a change in antimalarial therapy18.

Results

Baseline characteristics

This prospective observational study was conducted at Dr. Ziauddin University Hospital, Karachi, Pakistan, between September 2023 and August 2024. Of 1,100 individuals screened, 354 participants met eligibility criteria and were enrolled (Fig. 2). Baseline characteristics for the four treatment cohorts—chloroquine (CQ) monotherapy (n = 98), artemether–lumefantrine (AL) monotherapy (n = 84), CQ plus primaquine (CQ + PQ; n = 93), and AL plus primaquine (AL + PQ; n = 79)—are shown in Table 1. Male participants comprised 55.6% to 73.8% of each cohort. Median age (IQR) was 44 (32–60) years in the CQ group, 47 (32–60) years in the AL group, 55 (34–61) years in the CQ + PQ group, and 44 (27–60) years in the AL + PQ group. Baseline temperature and hemoglobin distributions showed some between-group variation, and most participants had normal G6PD status; intermediate results were observed in a small number of participants in the CQ and AL monotherapy groups.

Fig. 1
Fig. 1
Full size image

Study design and follow-up strategy for participants with Plasmodium vivax malaria. Schematic representation of the prospective observational cohort study conducted at Dr. Ziauddin University Hospital, Karachi, Pakistan, enrolling adult patients aged 18–65 years with confirmed Plasmodium vivax malaria between September 2023 and February 2024 (n = 354). Participants received antimalarial treatment according to routine clinical practice, resulting in four treatment groups: chloroquine (CQ), artemether-lumefantrine (AL), chloroquine plus primaquine (CQ + PQ), and artemether-lumefantrine plus primaquine (AL + PQ). Scheduled follow-up included daily assessments during the first 3 days, weekly visits through Day 28, and monthly evaluations up to 6 months. Suspected recurrences (defined as symptoms occurring ≥ 14 days after completion of initial therapy) were evaluated with repeat parasitological testing and, if confirmed, treated using the same regimen assigned at enrollment. This figure depicts the follow-up framework and does not present treatment outcomes.

Fig. 2
Fig. 2
Full size image

Flow diagram of participant screening, enrollment, and treatment allocation. Flow diagram summarizing screening, eligibility assessment, exclusions, and enrollment of adults aged 18–65 years with parasitologically confirmed Plasmodium vivax malaria in a prospective observational cohort at Dr. Ziauddin University Hospital, Karachi, Pakistan (September 2023 to February 2024). Of 1,100 patients screened, 354 met inclusion criteria and were enrolled for follow-up. Treatment was assigned according to routine clinical practice and physician discretion (non-randomized) to one of four regimens: chloroquine (CQ) alone, artemether–lumefantrine (AL) alone, chloroquine plus primaquine (CQ + PQ), or artemether–lumefantrine plus primaquine (AL + PQ). Exclusion criteria included mixed-species malaria, co-infections, pregnancy, glucose-6-phosphate dehydrogenase (G6PD) deficiency, malaria within the prior 6 months, or antimalarial use before enrollment. The enrolled cohort (n = 354) was followed for recurrence and clinical outcomes for 6 months. Abbreviations: Chloroquine (CQ); chloroquine + primaquine (CQ + PQ); artemether-lumefantrine (AL); artemether-lumefantrine + primaquine (AL + PQ); glucose-6-phosphate dehydrogenase (G6PD).

Table 1 Fundamental attributes of the population.
Full size table

Outcomes on day 28 and conclusion of Follow-up period

A total of 31 recurrent Plasmodium vivax infections occurred within the first 28 days. Additional recurrences were observed after Day 28, resulting in end-of-follow-up recurrence counts of 27/84 (32.2%) in the AL group, 15/98 (15.3%) in the CQ group, 5/79 (6.3%) in the AL + PQ group, and 1/93 (1.1%) in the CQ + PQ group (Fig. 3).

Fig. 3
Fig. 3
Full size image

Distribution of recurrent Plasmodium vivax infections by treatment group during follow-up. Bar charts display the observed number of participants with and without recurrent P. vivax infection in four treatment groups: chloroquine (CQ; n = 98), artemether–lumefantrine (AL; n = 84), chloroquine plus primaquine (CQ + PQ; n = 93), and artemether–lumefantrine plus primaquine (AL + PQ; n = 79). Recurrence was defined as parasitologically confirmed P. vivax infection occurring ≥ 14 days after completion of initial therapy. Panels summarize recurrence status assessed at Day 28 and at end of follow-up (6 months). Bars represent observed participant counts within each treatment group; no error bars or confidence intervals are displayed, and the figure is descriptive. Abbreviation: CQ; Chloroquine, CQ + PQ; Chloroquine + Primaquine, AL; Artemether/lumefantrine, AL + PQ; Artemether/lumefantrine + Primaquine.

Table 2 summarizes recurrence outcomes at Day 28 and at the end of follow-up. By Day 28, recurrence was highest in the AL group (17/84, 20.2%) and lowest in the CQ + PQ group (1/93, 1.1%). The corresponding Kaplan–Meier cumulative incidence at Day 28 was 24.5% (95% CI 14.3–34.7) for AL and 1.5% (95% CI 0.0–4.4) for CQ + PQ. By the end of follow-up, recurrence remained highest in AL (27/84, 32.2%) and lowest in CQ + PQ (1/93, 1.1%), with Kaplan–Meier cumulative incidence estimates of 39.1% (95% CI 26.7–51.4) and 1.5% (95% CI 0.0–4.4), respectively. Overall, primaquine-containing regimens (CQ + PQ and AL + PQ) showed lower observed recurrence proportions and lower cumulative incidence estimates than the corresponding monotherapies.

Table 2 Treatment outcomes and recurrence through day 28 and end of follow-up (unadjusted analysis).
Full size table

Time-to-event comparisons are presented in Table 3. In unadjusted Cox proportional hazards models, recurrence hazard was higher with AL monotherapy than CQ monotherapy at Day 28 (HR 2.683, 95% CI 1.223–5.887; p = 0.014) and by the end of follow-up (HR 2.281, 95% CI 1.186–4.389; p = 0.013). Comparisons of monotherapy versus the corresponding primaquine-containing regimens suggested lower hazards of recurrence with primaquine, although the magnitude and statistical significance varied by comparison and time point (Table 3). There was no evidence of a difference between AL + PQ and CQ + PQ at Day 28 (p = 0.152) or by the end of follow-up (p = 0.091) in unadjusted models. These time-to-event estimates are unadjusted and should be interpreted as associations.

Table 3 Unadjusted Cox proportional hazards models for recurrence.
Full size table

Fever defervescence

Among the 155 participants who were febrile at baseline (CQ, n = 40; AL, n = 45; CQ + PQ, n = 30; AL + PQ, n = 40), 94.8% (147/155) were afebrile within 24 h of treatment initiation. Fever persistence on Day 1 differed across regimens (CQ 12.5% [5/40], AL 0% [0/45], CQ + PQ 3.3% [1/30], AL + PQ 5.0% [2/40]; p = 0.022) (Table 4). Fever persistence on Days 2 and 3 was uncommon and did not differ significantly between groups (p = 0.563 and p = 0.70, respectively) (Table 4).

Table 4 Fever persistence during the first 3 days of therapy among participants febrile at enrollment.
Full size table

Safety

No serious adverse events were observed. The most frequently reported adverse events were headache (88/354, 24.9%), vomiting (24/354, 6.8%), abdominal pain (16/354, 4.5%), and nausea (12/354, 3.4%). The proportion of participants reporting these events did not differ significantly across treatment groups (all p > 0.05) (Table 5).

Table 5 Treatment-related adverse events.
Full size table

Discussion

Despite important advances in malaria prevention and treatment, the emergence and spread of antimalarial drug resistance remains a major barrier to malaria control and elimination efforts2. Pakistan continues to carry a substantial malaria burden, with approximately one million cases annually, and Plasmodium vivax is the dominant species, accounting for roughly 79% of reported malaria infections34. Chloroquine (CQ) has historically served as the backbone of therapy for P. vivax in Pakistan; however, even in recent reviews, systematic evidence confirming CQ-resistant P. vivax from Pakistan remains limited24. In routine practice, CQ has therefore continued to be widely used for the management of P. vivax infection and, at times, for empiric treatment of suspected malaria35. Notably, a single case report from Pakistan described a pregnant patient with recurrent P. vivax parasitemia following CQ exposure who was subsequently treated with artemether–lumefantrine (AL)21. While such reports raise concern, isolated clinical cases cannot establish population-level resistance in the absence of standardized efficacy follow-up, drug level assessment, and molecular characterization.

In this cohort, primaquine-containing regimens were associated with substantially lower observed recurrence within 28 days compared with the corresponding monotherapies. Specifically, recurrence by Day 28 occurred in 10/98 (10.2%) of participants receiving CQ alone and 17/84 (20.2%) receiving AL alone, compared with 1/93 (1.1%) receiving CQ + PQ and 3/79 (3.8%) receiving AL + PQ. These findings are broadly consistent with prior clinical studies from other endemic settings, including Ethiopia, where Day 28 recurrence after CQ has been reported in the range of 14–22%36. In our study, CQ monotherapy showed a lower observed recurrence proportion than reported in some settings; however, recurrence after CQ should not be interpreted as definitive evidence of chloroquine resistance, because molecular genotyping and drug level assessments were not performed and relapse, reinfection, and recrudescence could not be distinguished. Future studies incorporating molecular characterization and pharmacokinetic evaluation are needed to clarify the mechanisms underlying recurrent parasitemia in this setting.

In time-to-event analyses, AL monotherapy was associated with a higher hazard of recurrence than CQ monotherapy in unadjusted Cox models at Day 28 (HR 2.683, 95% CI 1.223–5.887; p = 0.014) and by the end of follow-up (HR 2.281, 95% CI 1.186–4.389; p = 0.013). Because treatment allocation was not randomized and may reflect physician preference, access, and adherence considerations, these comparisons should be interpreted as associations. Our findings regarding continued CQ activity are consistent with reports from Pakistan (e.g., Lower Dir and Swat), which describe susceptibility of P. vivax to CQ33. In routine practice, CQ may be preferentially used due to lower cost and broader availability, particularly among resource-limited populations, and inappropriate empiric antimalarial use or poor adherence to prescribed regimens may contribute to selection pressure over time. Finally, P. vivax strains in South Asia—including southern Pakistan—have been described as predominantly frequent-relapsing (“tropical”) phenotypes, with relapses often occurring within 3–6 weeks after the initial episode, which may contribute to early recurrence patterns in real-world cohorts22.

There are limited data in Pakistan regarding the effectiveness of artemether-lumefantrine (AL) for treating Plasmodium vivax malaria. A hospital based cohort in Nowshera by Muhammad et al. reported beneficial short term outcome using AL and noted no significant treatment failures were reported in their cohort at day 2828. However, in our cohort we found a 20.2% (17/84) recurrence rate after AL monotherapy at day 28, with an increasing trend of recurrences during longer follow up; with 32.2% (27/84) showing recurrences by the completion of follow-up. There could be several reasons why these two cohorts have different results; differing lengths of follow-up and study designs could account for some of this discrepancy. It is possible there were real world issues that affected how well each patient followed through with their prescribed regimen, differences in how patients received care, and other clinical factors that impacted which regimen was prescribed to each patient that were not captured in the study. To help compare our findings with previous research we made comparisons to previous studies. Senn et al. reported very low early clinical failure rates for AL in P. vivax (0.2%–2.2%) over short follow-up intervals19, however, Yohannes et al., reported higher failure/recurrence proportion at Day 28 for AL (approximately 19%), when compared to CQ (7.5%)37, which supports the pattern seen in our study (recurrence at Day 28 was greater for AL monotherapy compared to CQ monotherapy, 20.2% vs. 10.2% respectively). More importantly, the recurrences in our study were substantially less in regimens containing primaquine compared to their respective monotherapies; CQ + PQ had a 1.1% (1/93) recurrence at Day 28 and 3.8% (3/79) for AL + PQ at Day 28 and similar recurrence rates persisted throughout the follow up. Our findings support the findings that primaquine is critical for preventing subsequent episodes of P. vivax malaria and are in accordance with previously published treatment guidelines.

This extended follow-up period provided an opportunity to evaluate the frequency of P. vivax recurrences as well as when they occurred, i.e., during routine outpatient visits for treatment episodes where adherence to the treatment regimen and supervision could be variable. We found that recurrences were occurring at a higher rate after unsupervised treatment episodes, but because this was not a randomized comparison and we did not randomly assign subjects to receive supervised versus unsupervised treatment, these results are subject to interpretation based on potential confounding by indication and other unmeasured variables. Interestingly, when we evaluated the frequency of recurrences after receiving a first regimen vs. a subsequent regimen in the same group of patients undergoing blood-stage therapy with AL or CQ, there was not a statistically significant difference, indicating that factors beyond the blood schizonticide used may play a role in determining the frequency of recurrence. A plausible alternative explanation is that variability exists among individuals in their ability to adhere to the 14 day primaquine regimen in routine care, where it has been documented in prior studies that adherence rates to unsupervised primaquine treatment are less than 30%, and an Indian study demonstrated significantly lower levels of recurrent parasitemia in individuals who received directly observed primaquine treatment vs. those who received unsupervised treatment38. Conversely, another study conducted in Pakistan showed no statistical differences between supervised and unsupervised primaquine administration, which suggests that the association between supervision, adherence and recurrences may be influenced by various contextual and system-level variables39.

While primaquine (PQ) had been administered at Day 2 of treatment, the decline in hemoglobin (Hb) values primarily occurred during the period of enrollment through Day 3 and subsequently the Hb levels improved. The timing of this hemoglobin decrease suggests it may have been more directly related to the acute malaria event and early phase of treatment, as opposed to the effect of primaquine alone; however, since no data were collected to assess the degree of hemolysis or plasma concentrations of primaquine, a medication effect could still potentially occur particularly in those who were more susceptible. In agreement with the majority of the literature regarding the effects of P. vivax on the hematologic system, a surveillance study conducted in Papua, Indonesia found a mean hemoglobin level of 9.53 g/dl among patients infected with P. vivax, and an increased adjusted odds ratio for severe anemia (< 5 g/dl) when comparing to patients without malaria (aOR 1.87)40. In our cohort, treatment-emergent adverse events were similar between regimens that included PQ and those that did not, supporting tolerability of PQ-containing regimens in this setting.

We could find no statistically significant differences in recurrence rates for patients receiving CQ + PQ compared to those receiving AL + PQ at Day 28 (p = 0.152), and at the end of the study (p = 0.091) (Table 3). Rather than supporting the idea that CQ + PQ and AL + PQ were equivalent, our results indicate that within the limitations of this observational cohort and the limited number of recurrent episodes, clinical outcomes in terms of recurrences appeared to be relatively comparable among the two combinations that contained primaquine. The potential practical implications of this finding include support for flexibility in selecting either CQ or AL as the blood schizonticide when primaquine is selected as the antimalarial drug used for radical cure due to variability in drug availability and/or differences in patient-specific factors that can affect the choice of drugs.

On the other hand, the limited number of recurrences in both combination therapy arms limit the precision of our analysis and the large confidence intervals of the estimated hazard ratios suggest considerable uncertainty about the size of any possible differences between the two combinations. Therefore, additional research with larger sample sizes and standardized follow-up and adherence assessments will be required to provide definitive evidence comparing CQ + PQ and AL + PQ and to assess if either combination provides a clinical benefit in standard practice.

Several study constraints exist. First, this research employed physician discretion as opposed to randomization to determine treatments. Consequently, physician discretion may have introduced both selection bias and residual confounding into the data, and consequently comparative findings in the manuscript should be viewed as associations. While supplementary analyses evaluated confounding via gradient boosting models (GBM) propensity score estimates and inverse probability of treatment weights (IPTW), the GBM model did not completely eliminate differences in variables at baseline (e.g., temperature, hemoglobin); therefore, residual confounding is likely to continue affecting comparison of regimens. The researcher also did not record inpatient versus outpatient enrollment status as an analytical variable in the database; therefore, the researcher could not accurately count inpatient vs. outpatient participants nor compare results across settings of care. Second, the researcher cannot differentiate between recurrence, reinfection and recrudescence since genotypic analysis was not performed; therefore, the researcher classified all recurrent malaria events as “recurrent,” resulting in a composite measure as opposed to individual causes. Third, the researcher did not obtain measures of antimalarial drugs (chloroquine, artemether-lumefantrine and primaquine), thus making it impossible to directly assess adequacy of antimalarial drugs and patient compliance; in addition, artemether lumefantrine is administered in six doses, over three days while chloroquine is given for a much shorter period of time and may result in different levels of patient compliance; this may affect the accuracy of the results for each regimen, particularly among out-patients, and may result in bias when comparing artemether lumefantrine to chloroquine. In real-world settings, inadequate adherence has been associated with reduced effectiveness of AL-based regimens41.

G6PD screening was performed using a fluorescent spot test, which does not quantify enzyme activity or capture all variants, potentially affecting assessment of eligibility and risk stratification for primaquine therapy. Finally, variability in follow-up and outpatient attendance may contribute to reporting bias for recurrence outcomes, and the absence of blinding (patients and clinicians were aware of the prescribed regimen) may have influenced symptom reporting and clinical decision-making during follow-up.

The study has several strengths. This was a prospective observational cohort with structured follow-up through Day 28 and approximately 6 months, with recurrence defined using parasitological confirmation and standardized microscopy procedures. The inclusion of four commonly used regimens (CQ, AL, CQ + PQ, and AL + PQ) allowed comparison of primaquine-containing strategies across two blood-schizonticide backbones in a real-world setting. We also prespecified an assessment of baseline confounding using GBM-based propensity scoring and evaluated covariate balance using absolute standardized mean differences (ASMDs), reporting these diagnostics transparently in Supplementary Figure S1 and Supplementary Table S1.

Conclusion

In this observational cohort, chloroquine (CQ) monotherapy was associated with a lower hazard of recurrence than artemether–lumefantrine (AL) monotherapy over both short- and extended follow-up. Primaquine (PQ)-containing regimens were associated with markedly lower recurrence than the corresponding monotherapies, reinforcing the importance of radical cure strategies in reducing recurrent Plasmodium vivax episodes in this setting. Because treatment allocation was not randomized and molecular genotyping and drug level measurements were not performed, these findings should be interpreted as associations and cannot distinguish relapse from reinfection or recrudescence; recurrence after CQ should therefore not be taken as definitive evidence of chloroquine resistance. Nonetheless, our data support continued emphasis on PQ use in national treatment strategies to prevent recurrence. As Pakistan considers implementation of single-dose tafenoquine to improve adherence compared with a 14-day PQ regimen, further implementation and effectiveness studies will be needed to evaluate feasibility, safety (including G6PD screening strategies), and real-world impact. Ongoing surveillance incorporating molecular methods and pharmacokinetic assessments remains important to monitor susceptibility patterns and optimize first-line therapies for P. vivax malaria in Pakistan.