Plasmodium vivax and mixed infections are associated with severe malaria in children: a prospective cohort study from Papua New Guinea.

BACKGROUND: Severe malaria (SM) is classically associated with Plasmodium falciparum infection. Little information is available on the contribution of P. vivax to severe disease. There are some epidemiological indications that P. vivax or mixed infections protect against complications and deaths. A large morbidity surveillance conducted in an area where the four species coexist allowed us to estimate rates of SM among patients infected with one or several species. METHODS AND
FINDINGS: This was a prospective cohort study conducted within the framework of the Malaria Vaccine Epidemiology and Evaluation Project. All presumptive malaria cases presenting at two rural health facilities over an 8-y period were investigated with history taking, clinical examination, and laboratory assessment. Case definition of SM was based on the World Health Organization (WHO) criteria adapted for the setting (i.e., clinical diagnosis of malaria associated with asexual blood stage parasitaemia and recent history of fits, or coma, or respiratory distress, or anaemia [haemoglobin < 5 g/dl]). Out of 17,201 presumptive malaria cases, 9,537 (55%) had a confirmed Plasmodium parasitaemia. Among those, 6.2% (95% confidence interval [CI] 5.7%-6.8%) fulfilled the case definition of SM, most of them in children <5 y. In this age group, the proportion of SM was 11.7% (10.4%-13.2%) for P. falciparum, 8.8% (7.1%-10.7%) for P. vivax, and 17.3% (11.7%-24.2%) for mixed P. falciparum and P. vivax infections. P. vivax SM presented more often with respiratory distress than did P. falciparum (60% versus 41%, p = 0.002), but less often with anaemia (19% versus 41%, p = 0.0001).
CONCLUSION: P. vivax monoinfections as well as mixed Plasmodium infections are associated with SM. There is no indication that mixed infections protected against SM. Interventions targeted toward P. falciparum only might be insufficient to eliminate the overall malaria burden, and especially severe disease, in areas where P. falciparum and P. vivax coexist.

Introduction

Most of the research and published literature on malaria focuses on P. falciparum and much less on P. vivax [1,2]. This focus is due to the very high burden of mortality attributed to the falciparum species in Africa [3]. However, there is growing evidence that P. vivax is responsible for a significant burden of disease worldwide accounting for half of all malaria cases in Asia and Latin America [4]. As the term “benign tertian malaria” implies, vivax malaria is usually an uncomplicated disease that runs a benign course and is rarely fatal. This clinical paradigm has been challenged recently [5] by numerous reports of symptoms and signs of severe disease, and even deaths due to P. vivax monoinfections [6-9]. However, most of the published literature consists of case reports or small descriptive clinical series lacking denominators. The relative contribution of P. vivax versus P. falciparum to severe morbidity has not been properly assessed, except for one study in Thailand that found very little severe disease and no death due to P. vivax [10]. This Thai study suggested, rather, a protecting effect of vivax, as had a study from Vanuatu [11,12]. It is only recently that severe disease due to P. vivax has received more attention with a recent report from a hospital in northeastern Indonesian New Guinea (Papua) where they found 36 severe malaria (SM) cases among 1,135 P. vivax infected patients (3.2%) [13].

The malaria epidemiology in Papua New Guinea (PNG) offers the opportunity to investigate in the same population the respective contribution of the different Plasmodium species in terms of morbidity, and severe disease in particular. Indeed, it is one of the few countries where all four Plasmodium species coexist at sufficient prevalence to allow proper comparisons. The aim of the present clinico-epidemiological study was to assess the proportion of cases presenting at health facilities with severe manifestations of malaria and associated P. vivax parasitaemia, to compare this proportion to that of P. falciparum in the same setting, and to investigate the potential of mixed P. falciparum and P. vivax infection to protect against severe disease.

Methods

Study Area and Population

The Wosera lies in the East Sepik Province of PNG. Transmission of malaria is perennial, but the distribution of rainfall shows some seasonality with 65% falling during the so-called wet season occurring from October to April. Malariometric surveys conducted in the area showed an overall Plasmodium prevalence rate of 60% in the early 1990s decreasing to 35% by 2002. The reduction was from 38% to 22% for P. falciparum, 20% to 10% for P. vivax, 16% to 4% for P. malariae [14,15], and was probably related to a gradual increase in the use of insecticide treated nets and a change in treatment policy.

Health Services

The morbidity surveillance was conducted in two health facilities, Kunjigini Health Subcentre in the South Wosera and Kaugia Health Subcentre in the North Wosera, within the framework of the Malaria Vaccine Epidemiology and Evaluation Project [16]. Both health facilities are staffed with nurses and have beds to admit severe cases for a short period of time. Limited laboratory tests are available (light microscopy for malaria, haemoglobin photometer, and urine dipstick). The local referral centre is Maprik Hospital, a 110-bed government hospital. Recourse to health services is usual in case of fever, especially in children, but referral is very uncommon, due to transport problems and lack of qualified staff in the hospital. First-line treatment policy for uncomplicated malaria (UM) was a 3-d regimen of oral amodiaquine in children < 20 kg, or chloroquine in those > 20 kg; treatment for severe or resistant malaria included a 5-d regimen of intramuscular quinine combined with a single dose of oral sulfadoxine-pyrimethamine (SP). In 2000, the first-line treatment changed to a combination of amodiaquine or chloroquine (3 d) plus SP (single dose) for UM, and the second-line to artesunate (7 d) plus SP (single dose on day 3). The 28-d failure rates with the regimens used in the area during the study were ∼20% for P. falciparum and ∼0% for P. vivax [17,18].

Morbidity Surveillance

Patients were prospectively recruited if they were diagnosed on clinical grounds only as having malaria (presumptive diagnosis). The diagnosis was based on a history of fever without obvious symptoms or signs of another disease. These presumptive malaria cases were further investigated by a research nurse. After recording demographical information, closed questions were asked on the characteristics of the morbid episode (history of fever, headache, cough, dyspnea, vomiting, diarrhoea, abdominal pain, and convulsions). A standardized physical examination was then performed (axillary temperature, level of consciousness, respiratory rate, chest indrawing, and Hackett's grading of spleen). A blood sample was taken for parasitological examination by microscopy and haemoglobin measurement (for details see Genton et al. [19]).

Laboratory Tests

Parasitology.

Blood films were stained with 4% Giemsa and examined for 100 microscopic thick film fields prior to being declared negative. In positive films, parasite species were identified and densities were recorded as the number of parasites per 200 white blood cells (WBC). This procedure was followed for each species. Densities were converted to parasites per microliter of blood, assuming 8,000 WBC per microliter. A routine quality control procedure was performed [20]. Briefly, 10% of randomly selected slides and all those that had a density between 1 and 5 parasites per 200 WBC (for any of the four species) were reread, in a blinded fashion, by a supervisor microscopist. When the results on a batch of ∼1,000 slides did not reach 75% agreement on positivity/negativity, species, and density (including a margin of error that increased with increasing density), the entire batch was reread and the same quality control applied again.

Haemoglobin concentration.

Haemoglobin level was determined using the HemoCue photometer (Ängelholm).

Ethical approval and informed consent

The morbidity surveillance in the Wosera area received ethical approval from the Medical Research Advisory Committee (MRAC) within the framework of the larger Malaria Vaccine Epidemiology and Evaluation Project [16]. Patients or patients' guardians were asked for oral informed consent by the research nurse. This mode of consent was approved by the MRAC, and considered appropriate for this setting where many people cannot read and write. Moreover, the procedures performed (systematic history taking, clinical examination, and diagnostic test for malaria) provided patients with a laboratory-confirmed diagnosis of malaria and consequently with better management of the illness.

Data Analysis

Case definitions.

A UM case was defined as a presumptive malaria patient (see above) associated with documented Plasmodium asexual blood-stage parasitaemia by light microscopy.

A SM case was defined using the standard criteria established by the World Health Organization (WHO), except that we also included species other than P. falciparum, and we did not take into account a parasite density >250,000/μl since we expected lower densities in P. vivax than in P. falciparum infections [21]. Because clinical and laboratory facilities were limited in the setting we were working in, we could only assess a subset of criteria defining SM. Thus, our definition included a patient with presumptive malaria associated with documented Plasmodium asexual blood-stage parasitaemia, and with at least one of the following: history of more than one convulsion in the last 24 h; coma (patient unable to localize a painful stimulus); respiratory distress (high respiratory rate [> 50 for children 0 to < 2 mo of age and > 40 for others including adults] and chest indrawing or shortness of breath); and haemoglobin < 5 g/dl. Since there is ongoing debate about whether it is appropriate or not to include patients with associated symptoms or signs (comorbidity), we also report on a more restricted definition excluding patients with associated cough and/or diarrhoea, two features that may or may not be related to malaria [22]. The latter symptoms were usually mild since, if significant, the patients would not have been diagnosed as presumptive malaria.

Analysis.

We analysed the proportion of SM cases by species (P. falciparum monoinfection, P. vivax monoinfection, and mixed P. falciparum and P. vivax infections) among all malaria cases (UM and SM) recorded in the morbidity surveillance from 1997 to 2004. We stratified our analysis by age group, since we expected different effects in individuals with different level of pre-existing immunity. We focused most of the analyses in the 0- to 5-y age group, since in highly endemic areas of PNG vivax malaria and SM, in particular, are encountered mainly in young children [23,24].

The chi square test was used (i) to compare rates of SM between the groups of malaria cases infected with P. falciparum, P. vivax, or mixed species; (ii) to investigate seasonal variation of SM rates; and (iii) to compare the distribution of clinical and laboratory features defining SM between groups of SM patients infected with P. falciparum, P. vivax, or mixed species. The proportions of cases with each of these features among P. falciparum, P. vivax, or mixed species SM are illustrated by Venn diagrams (created using free software that draws three set area proportional graphs) [25]. A chi square for trend was used to investigate the pattern of SM over the 8-y period.

Parasite densities were compared between severe and uncomplicated malaria as well as between species using log-transformed data and t-tests.

Data management was performed using the Foxpro software and data analysis Stata version 6. Since only a subsample of our patients in the morbidity surveillance were also included in the demography surveillance database (different villages), no systematic follow-up was possible.

Results

A total of 73,620 patients were registered in the 8-y period of the morbidity surveillance from 1997 to 2004. Among those, 17,201 cases were diagnosed as presumptive malaria cases and investigated further by the research nurse. A total of 9,537 of those patients (55%) had asexual Plasmodium parasitaemia. Among those, 6,886 (72%) had pure P. falciparum, 1,946 (20%) P. vivax, 328 (3%) P. malariae, 27 (0.3%) P. ovale, and 350 (4%) mixed infections (314 P. falciparum and P. vivax; 21 P. falciparum and P. malariae; ten P. vivax and P. malariae; three P. falciparum and P. ovale; one P. malariae and P. ovale; and one P. falciparum, P. vivax, and P. malariae). The age-specific parasite prevalence and species distribution among all presumptive malaria cases is shown in Figure 1. The highest parasite prevalence was found in children aged 5 to < 10 y (peak age 7 y). On the other hand, the highest proportion of P. vivax was seen in children 2 to < 5 y (peak age 2 y), with up to 20% parasitaemic.

Figure 1

Age-Specific Prevalence of Parasitaemia among Presumptive Malaria Cases

Among parasitaemic individuals with all data available (7,759), 6.2% (95% confidence interval [CI] 5.7%–6.8%) fulfilled the case definition of SM. Table 1 details the proportion of SM cases associated with P. falciparum, P. vivax, and mixed P. falciparum and P. vivax infections by age group using two different case definitions: standard and restricted (excluding patients with cough or diarrhoea). Children aged < 2 y were more likely to have SM (standard definition) than those aged 2 to < 5 y (odds ratio [OR] 2.2, 95% CI 1.8–2.7) and those aged 5 to < 10 y (OR 6.4, 95% CI 4.7–8.6); for individuals infected with P. vivax, the ORs were of 2.8 (95% CI 1.8–4.4) and 7.5 (95% CI 3.5–15.7), respectively; for individuals infected with P. falciparum, the ORs were of 2.0 (95% CI 1.5–2.6) and 6.0 (95% CI 4.2–8.5); for individuals with mixed infections, the ORs were of 3.2 (95% 95% CI 1.3–8.0) and 32.1 (95% 95% CI 3.2–319, Fisher's exact < 0.001). The decrease in risk of severe disease was linear with age for all patients with P. falciparum, P. vivax, or mixed infections (Figure 2). The decrease in risk was significantly more pronounced for mixed (OR 0.52/y, CI 0.38–0.72) compared to single infections (OR 0.72/y, 95% CI 0.68–0.76, likelihood ratio [LR] χ2 = 4.7, degree of freedom [df] = 1, p = 0.03), but similar for P. falciparum (OR 0.73/y, 95% CI 0.68–0.74) and P. vivax single infections (OR 0.65/y, CI 0.56–0.77, likelihood ratio [LR] χ2 = 2.5, df = 1, p = 0.11).

Table 1

Proportion of SM Cases Associated with P. falciparum, P. vivax, and Mixed P. falciparum and P. vivax Infections by Age Group Using Two Case Definitions: Standard and Restricted (Excluding Patients with Cough or Diarrhoea)

Figure 2

Age-Specific Risk of Severe Disease with P. falciparum, P. vivax, and Mixed Infections in Children <10 y

Figure 3 shows the proportion of SM in children < 5 y infected with any species (except P. ovale because cases were too few) or mixed infections. Children < 5 y infected with both P. falciparum and P. vivax infections (mixed) were more likely to present with SM than those who were infected with P. falciparum only (17% versus 12%, OR 1.5, 95% CI 1.0–2.4), or P. vivax only (17% versus 9%, respectively, OR 2.1, 95% CI 1.3–3.5). In children < 2 y, almost one-third (29%) with mixed infections had SM. In this age group, the proportion of SM among P. vivax infections was close to that among P. falciparum infections (14% versus 18%, respectively, for the standard definition, 4% versus 6% for the restricted definition) (see Table 1).

Figure 3

Mosaic Plot of SM by Species in Children < 5 y

Horizontal, species; vertical, proportion of SM.

The proportion of SM cases (< 5 y) presenting with each type of clinical manifestation or laboratory feature defining SM is shown schematically in Figure 4 by Venn diagrams for the group of SM infected with P. vivax, P. falciparum, or mixed species. P. vivax SM and mixed SM presented more often with respiratory distress than did P. falciparum SM (61% versus 41%, χ2 = 9.85, p = 0.002 and 67% versus 41%, χ2 = 5.88, p = 0.015, respectively), but less often with anaemia; the latter being only significant for the comparison between P. vivax and P. falciparum (19% versus 41%, χ2 = 14.57, p = 0.0001 and 29% versus 41%, χ2 = 1.36, p = 0.2). Neurological manifestations tended to be less frequent in mixed SM than in P. falciparum or P. vivax SM (8% versus 23% and 26%, respectively), but none of the differences were significant, probably due to the small numbers. Comparing the rate of each neurological sign between P. vivax and P. falciparum SM, fits were as frequent in both species (22% for both, χ2 = 0, p = 0.96); as well as coma (4% versus 2%, respectively, χ2 = 1.25, p = 0.26).

Figure 4

Proportion of SM Cases with Each of the Main Defining Clinical or Laboratory Features among Children <5 y Infected with P. vivax, P. falciparum, and Mixed Species (Venn Diagram)

More generally, among all children < 5 y with parasitaemia, the prevalence of anaemia was 5.1% in patients infected with P. falciparum, 1.6% in those with P. vivax, and 5.3% in those with mixed infections; respiratory distress was found in 5.5% of P. falciparum patients, 5.1% of P. vivax patients, and 10.3% of those with mixed infections; and repeated convulsions were encountered in 2.5% and coma in 0.2% of P. falciparum patients, in 2% and 0.3%, respectively, of P. vivax individuals, and in 1.1% and 0%, respectively, of those with mixed infections.

To investigate whether the increased risk of SM in children infected with two species was due to a higher overall parasite burden, we compared densities in mixed infections versus those in monoinfections. As shown in Figure 5 (restricted to children < 5 y), total parasite densities were indeed significantly higher in children with mixed infections than in those with P. falciparum or P. vivax monoinfections, whether all cases (t = −2.41, df = 2,968, p = 0.02 and t = −11.57, df = 1,372, p < 0.01, respectively) or the severe ones only (t = −2.38, df = 283, p = 0.02 and t = −5.39, df = 108, p < 0.01, respectively) were considered. Densities tended to be lower in the severe P. falciparum and P. vivax cases than in the uncomplicated ones; the reverse was true in cases with mixed infections. In the latter, the average proportion of P. falciparum density/total parasite density was 83% in the UM and 87% in SM.

Figure 5

Total Parasite Densities in Uncomplicated and SM by Species (Children <5 y)

Line is median, boxes are 25th and 75th percentile, and whiskers are upper and lower adjacent values.

Age was the prominent risk factor for presenting with SM, but the season and year of attendance also had a significant effect. Malaria (parasitaemic) patients attending during the wet season were more likely to have SM than those attending during the dry season (OR 1.3, 95% CI 1.1–1.6), and this effect was significant for P. vivax (OR 1.9, 95% CI 1.2–2.9), but not for P. falciparum (OR 1.1, 95% CI 0.9–1.4), or mixed infections (OR 0.8, 0.4–1.9). Over the years, there was a significant decrease in the rate of SM (OR 0.8, 95% CI 0.7–0.8), with a clear step observed from 2000 onward, probably as a consequence of the new treatment policy that was introduced in that year (adding SP to amodiaquine or chloroquine).

Discussion

The present study shows that in a rural area of PNG where all four species of Plasmodium coexist, P. vivax is responsible for 21% of all SM cases, P. falciparum for 71%, and mixed P. vivax and P. falciparum infections for 5%. A feature of this clinico-epidemiological report is the simultaneous presence of P. falciparum as an internal control for P. vivax, so that the relative proportions of severe disease can be determined, regardless of its definition. The proportion of patients < 5 y who presented with SM among all P. vivax infected children was as high as 9%. The corresponding value for P. falciparum was 12%. Since the evaluators and the defining criteria for uncomplicated and severe malaria were strictly the same, we can be confident that the relative values (P. vivax versus P. falciparum) are correct. Moreover, the absolute value for the proportion of severe cases among all P. falciparum-infected children (12%) was lower than those found in numerous other studies of clinical malaria, the main reason being that our study took place in a peripheral health facility and not in a referral hospital as most, if not all, other studies [26-28]. It is clear that this absolute value depends entirely on the criteria used to define SM and on the population from which the cases are selected. Since we were working in a setting that did not allow us to measure certain parameters included in the WHO definition, such as glycemia or creatininemia, we probably slightly underestimated the true proportion of severe malaria caused by both species. Whichever definition is used, the relative contribution of P. vivax to the overall burden of SM morbidity is considerable in this region, and especially in children <2 y, where the proportion of P. vivax parasitaemic children presenting with severe disease reached a worrying rate of 14% (versus 18% for P. falciparum). These data on severe vivax malaria are in line with those reported in a joint paper from a hospital in Papua, Indonesia, also published in this issue of the journal [28]. In that setting, severe disease was present in 23% of the patients with malaria, with rates similar for P. falciparum and P. vivax (22%), but significantly higher in patients with mixed infections (34%). These two observational studies, although both originating from the same region of the world, add to the growing realization that P. vivax is not benign. They are the first studies, to our knowledge, to incorporate denominators that allow assessing the relative magnitude of the problem when compared to that of P. falciparum.

Although all types of malaria complications observed with P. falciparum were also encountered with P. vivax, their distribution was different: severe cases of P. vivax were less likely than those of P. falciparum to present with anaemia, but more likely to suffer from respiratory distress. These observations are understandable since we know that P. vivax malaria patients have lower densities than P. falciparum patients, and therefore the likelihood of having profound anaemia is less, even if other immunological factors may contribute to red cell destruction. The proportion of severe P. vivax cases with anaemia was much less in our study (19%) than in Tjitra's (86%) [28]. Their hypothesis of P. vivax resistance contributing largely to the high prevalence of anaemia in their setting is plausible since we had 0% P. vivax resistance in the Wosera during the study period [17,18] versus 65%–95% in theirs [29,30]. Our finding of increased risk of respiratory distress in vivax patients when compared to falciparum patients is in line with recent data from Anstey et al. who showed a more severe alveolar-capillary dysfunction in vivax than falciparum malaria [31]. These observations are consistent with a greater inflammatory response to a given parasite burden in P. vivax relative to that in P. falciparum, and suggest that P. vivax infected erythrocytes may sequester within the pulmonary microvasculature [31]. Severe vivax malaria is thus an emerging recognized entity, and the long-standing belief that P. vivax is a nonsequestering parasite might need to be revisited [31,32].

Contrary to the study in Thailand that reported lower frequencies of SM in dual P. vivax and P. falciparum infections [10,33], we found that patients with mixed infections were more likely to present with SM than those infected with a single Plasmodium species. Our data are in line with case series from India [34], Malaysia [35], and Indonesia [28]. The mechanism for the higher proportion of SM among patients with mixed infections might be a higher overall parasite load. Indeed we found that severe cases with mixed infections had higher parasite densities than those with single P. falciparum and, even more so, P. vivax infection. This increased severity does not formally contradict the possible protective role of asymptomatic P. vivax infection against a subsequent P. falciparum morbid episode. In a study conducted in the same area as the present one [36], asymptomatic individuals infected with P. vivax, P. malariae, or mixed infections were less likely to have a morbid episode with P. falciparum during follow-up. This may mean that indeed non-falciparum species could protect against the risk of morbid episode because of P. falciparum, but, once the disease is established, having mixed infection makes the patient sicker, owing to the higher parasite density, as found in our study. This hypothesis must be balanced against the observation that, also in our series, SM cases with single P. falciparum or P. vivax infection did not harbor higher parasite densities than uncomplicated cases. This has been repeatedly found in different settings. Parasite density has not been identified as a risk factor for death in one of the largest series of malaria cases in Africa [37].

Our study is limited by the fact that parasite species were identified by microscopy, and it is known that this method largely misses minority infections [15,38,39]. Thus, mixed infections may have been largely underestimated, and a more sensitive detection method may have led to different results. While this limitation may be valid, it is believed that, in general, the infections that make the patient sick are those of high density, and that these are the ones that are identified by microscopy, as in the present study; those with very low densities are unlikely to contribute much to the acute phase, and even less to severe manifestations.

Another potential limitation is the accuracy of malaria diagnosis. In a high transmission area, where the prior probability of a young child being parasitaemic is high, the diagnosis of clinical malaria is not straightforward. We know from comparable settings in Africa that sepsis is commonly misdiagnosed as malaria, or that pneumonia can be associated with malaria. Since no blood cultures and no chest radiographs were performed, we cannot be certain that these children did not have a disease other than malaria. However, this potential bias is inherent to all studies of SM, even those that include thorough clinical assessment and sophisticated laboratory procedures. In intervention studies, a parasite density threshold is advised to increase the specificity of the diagnosis. When using a P. falciparum density threshold that should have led to over 90% sensitivity and 90% specificity for a diagnosis of malaria in this area and age group (< 5 y) [40], we found that 74% of the P. falciparum UM and 71% of the P. falciparum SM were above 4,000 parasites/μl. When applying the same statistical methodology to P. vivax (I. Mueller, personal correspondence), 71% of the P. vivax UM and 66% of the P. vivax SM were above 1,000 parasites per microlitre. We can thus postulate that at least two-thirds of the cases were “true” malaria. Even if some of our cases actually had alternative/additional diagnoses, the finding of severe disease with an associated P. vivax parasitaemia cannot be disregarded, as in P. falciparum cases [41]. These children should still be managed as SM cases, whether illness is due to P. falciparum, to P. vivax, or to another cause indistinguishable from malaria [42,43]. This approach is supported by a recent retrospective analyses of hospital records in Jayapura, West Papua, Indonesia, which found risk of fatal outcomes among SM patients was comparable between those with falciparum versus vivax malaria [13]. Epidemiological studies such as the present one should be used as a notice for clinicians working in areas where P. vivax is present, and as a signal for researchers to further investigate the pathogenesis of P. vivax malaria and the potential associated comorbidities, such as sepsis, in the same way as has been done for P. falciparum [44].

Why is P. vivax malaria only now recognized as a potential serious threat to life, although the parasite has been present for centuries? One of the reasons may be that SM has always been, and almost exclusively, associated with cerebral complications, and the latter are rare in P. vivax. The prejudice to only consider P. falciparum as potentially severe was therefore difficult to overcome. It is only in the 1990s that other clinical manifestations and laboratory parameters have been identified as indicators of poor outcome [26]. Anaemia and respiratory distress, for example, have been recently included in the case definition of SM, and these signs are encountered frequently in P. vivax malaria. There is therefore increased awareness to recognize manifestations that may be less apparent than coma, but still harbor poor prognosis. Also, the relatively recent establishment of large morbidity surveillance systems in different epidemiological settings has improved the ability of researchers to capture unexpected events or trends.

The finding that a significant proportion of SM morbidity is due to P. vivax calls for interventions that do not target P. falciparum parasites only, at least in areas where several Plasmodium species coexist. In areas with multiple species, new interventions should be assessed for their effectiveness on overall malaria morbidity, so that potential interactions, especially compensatory effects on other species, can be identified early enough, before widespread implementation.