The Mimika district lies on the southern coast of Papua in Eastern Indonesia (Figure 1), covering an area of 21,522 square-kilometres with 12 sub-districts and 85 villages. The area is largely forested with both coastal and mountainous areas. Each year a total of approximately 5.5 metres of rainfall is recorded with peaks in July to September and December (unpublished data). At the last census in 2004, the population in the lowlands was reported as 130,000. One hospital, the Rumah Sakit Mitra Masyarakat (RSMM) in the town of Timika, services the whole district and is the only hospital available for the lowland population. Due to the presence of a local mine, there is economic migration, with the local population increasing by an estimated 16% per year. This has resulted in the diverse ethnic origin of the local population, with highland Papuans, lowland Papuans and non-Papuans all resident in the region. Healthcare for the population is provided by the public clinics of the local ministry of health, the Public Health Malaria Control programme (PHMC) of the mine, the RSMM hospital and the private sector.

Malaria transmission is perennial, but restricted to the lowland area where it is associated with three mosquito vectors: Anopheles koliensis, Anopheles farauti and Anopheles punctulatus 1112. Entomological inoculation rates vary between 1 and 4 infected bites per year (unpublished data). Bed net coverage is estimated to be approximately 40%. In view of the high number of infections in non-immune patients, local protocols recommend that all patients with patent parasitaemia at any level are given antimalarial therapy. At the time of the study local treatment guidelines advocated chloroquine plus sulphadoxine-pyrimethamine for P. falciparum and chloroquine monotherapy for non-falciparum malaria. An assessment of local treatment regimens in 2004 highlighted that the day-28 cure rate of chloroquine monotherapy was less than 35% for patients with P. vivax and that of chloroquine plus sulphadoxine-pyrimethamine was only 52% for patients with P. falciparum 7. Local and national guidelines also recommend that patients with P. vivax parasitaemia over 1 years old, receive 14 days unsupervised treatment with primaquine, however adherence to and effectiveness of this regimen in this setting is not known.

Data was collected from routine health facilities surveillance data and from a cross-sectional survey.

Data from health facilities were collated to determine the number of patients treated for laboratory-confirmed malaria between January 2004 and December 2005. The established routine malariometric surveillance network comprised 14 primary health clinics and the RSMM hospital. Four primary health clinics were funded by the local ministry of health and ten were funded by the public health malaria control (PHMC) programme of the local mine. The total number of malaria cases reported were equally divided between the hospital, the PHMC clinics and the local ministry of health clinics.

At all of these facilities malaria treatment guidelines require that treatment should be preceded by microscopic confirmation of thick or thin blood film. Parasite species was assessed by surveillance facility microscopists from Giemsa-stained thick blood films and peripheral parasitaemia determined from the number of parasites per 200 white blood cells and quantified as 1+, 2+, 3+ or 4+. A thick smear was considered negative on initial review if no parasites were seen in 100 high power fields. A thin smear was also examined to confirm parasite species and used for quantification if parasitaemia was greater than 200 per 200 WBC. All laboratories in the surveillance network provide weekly aggregated data on the number of slides read, and the species reported according to four age categories (less than 1 year, 1 to 4 years, 5 to 9 years and 10 or more years). The surveillance microscopists participate in ongoing quality assurance, training and monitoring. In 2004 the microscopy service at the hospital was assessed for accuracy, and a random sample of 1200 positive slides was selected for rereading by an independent microscopist with more than 10 years' experience. In total, slides were available in 97% (1,158/1,200) cases with concordance between the readings of 89% (1,032/1,158). Of the 126 discordant slides 18 (14%) were negative on second rereading, 46 (37%) monoinfections were reread as mixed species, 47 (37%) mixed infections were reread as monoinfections. A further 6 slides with P. falciparum infections were reread as P. vivax, and 9 slides with P. vivax infections were reread as P. falciparum.

In the hospital, data were collected from individual patient records. For those attending the hospital outpatient clinic, information of the hospital number, date of visit, age and sex of the patients and diagnosis, including the species of infection if malaria, were registered on computerized hospital records (Q-Pro_; Jakarta, Indonesia). All patients admitted to hospital for more than 12 hours were reviewed by a member of the onsite research team, and details of their visit documented on a standard report form which included, for this study, the hospital record number, data of admission, age and sex of the patient 9.

A cross-sectional survey, was conducted during July to December 2005, to determine the demographic breakdown of the local population, the population prevalence of malaria and anaemia, and document the treatment-seeking behaviour and costs incurred by febrile illness. The survey methodology and results will be described in more detail separately. For the treatment seeking survey, the sample size required to determine the true prevalence of households with someone with a recent history of fever was based on an estimated prevalence of 40% and population size of 140,000; a multiplication factor of 2 was used to take into account the design effect due to cluster sampling. A sample size of 800 households was required to achieve an estimate of prevalence with greater than 95% confidence 13.

Households were chosen by three-stage cluster random sampling, according to WHO guidelines 14. Household members were defined as anyone who lived under one roof, ate from one kitchen and had resided in the study area for at least six months. For each member, demographic information and fever history were recorded using a standardized questionnaire. When the member was present, weight and height were documented and a finger prick of blood taken for blood film examination. Axillary temperature was recorded and respondents defined as symptomatic if they reported history of fever in the last 24 hours or had an axillary temperature of 37.5°C or above. All household members who reported fever in the last months were asked to complete a questionnaire on treatment seeking behaviour.

Blood film examination was carried out for all the facility surveys. The peripheral parasitaemia was calculated assuming a white cell count of 7,300 μl-1 ¹ 7. All positive blood films and 10% negative blood films were rechecked in the National Institute of Health Research and Development reference laboratory in Jakarta, and results that differed from those obtained in the unit laboratory were reviewed by the two head laboratory technologists. Microscopically confirmed malaria cases were treated according to Indonesia Ministry of Health guidelines. Children and adults found to be anaemic either on clinical examination or after determination of haemoglobin levels were provided iron supplementation according to local guidelines. Other common ailments detected clinically were also treated.

Data on patients diagnosed with malaria were single entered into Epi Info 6 (U.S. Centers for Disease Control and Prevention, Atlanta, Georgia, USA) and data from the household surveys double entered and validated using EpiData 3.02 software (EpiData Asoociation, Odense, Denmark). Analysis was performed using SPSS for Windows (vs 15 SPSS Inc, Chicago, Illinois, USA) and STATA statistical software (Version 8, Stata Corp. TX). The Mann-Whitney U test or Kruskal-Wallis method were used for nonparametric comparisons, and Student's t-test or one-way analysis of variance for parametric comparisons. For categorical variables, percentages and corresponding 95% confidence intervals (95% CI) were calculated using Wilson's method. Proportions were examined using χ² with Yates' correction or by Fisher's exact test. Pyrogenic thresholds were defined using the receiver operator curve, and Youlden's Index calculated as sensitivity plus specificity minus 1.

In order to estimate the annual population incidence of malaria, data from the fever treatment seeking behaviour study were used to estimate the proportion of patients with malaria attending clinics participating in the surveillance network. In case the treatment seeking behaviour of patients with non-malarial fever differed significantly from those with malaria, we used an algorithm to assign each respondent a malaria diagnosis. This was done according to whether malaria was confirmed by a blood test, the treatment people received, and the status of the blood film at the time of the household survey. A sensitivity analysis was performed by assuming best-case and worse case scenarios where there was uncertainty in the algorithm.

The study was approved by the Ethics committee of the National Institute of Health Research and Development, Indonesian Ministry of Health (Jakarta, Indonesia) and the Ethics committee of Menzies School of Health Research (Darwin, Australia). Written informed consent was obtained from adult patients and parents of enrolled children.

Between January 2004 and December 2005, 99,158 laboratory-confirmed episodes of malaria were reported from patients seeking treatment at health facilities in the surveillance network. Of these, 68,802 (69%) were reported from community outpatients, 24,334 (25%) hospital outpatients and 6,022 (6.1%) from hospital admissions. Laboratory confirmation of malaria was made in 27% (68,802/253,987) of patients who had a malaria smear at community clinics. At the hospital, 15% (24,334/168,217) of the total outpatient workload and 34% (6,022/17,823) of the total inpatient workload was attributable to patients with malaria (Figure 2). Although treatment guidelines recommend that all patients have microscopic confirmation of malaria prior to treatment, a further 2,148 outpatients were reportedly diagnosed and treated following clinical diagnosis, ie 8.1% (2,148/26,482) of all outpatients treated for malaria.

Children under five years of age, accounted for 22% (14,855/68,802) of community outpatients with malaria, 29% (7,029/24,334) of hospital outpatients and 34% (2,050/6,022) of patients admitted to hospital (overall p < 0.001).

In total 58% (57,938) of cases had infection with P. falciparum and 37% (36,471) P. vivax (Table 1). The proportion of malaria due to P. vivax was greatest in community outpatients (42% 29,144/68,802), but fell to 25% (6,181/24,334) in hospital outpatients and 19% (1,146/6,022) in hospital inpatients (p < 0.001). In children under one year of age malaria was most likely to be attributable to infection with pure P. vivax (overall 55% 2,684/4,889) and this was apparent in all types of facility; overall p < 0.001 (Figure 2). Malaria attributable to P. vivax fell to 43% (8,112/19,045) in children aged 1–4 years, to 33% (4,029/12,096) in children 5 to 9 years old and 34% (21,618/63,077) thereafter.

Demographic data was gathered for 5,255 people from 825 households. Overall 46% (2,409) were female with 6% (92/1,441) of women of childbearing age pregnant. Of the 5,255 respondents questioned 28% (1,494) were highland Papuans, 26% (1,371) lowland Papuans, with the remaining 45% (2,390) of non-Papuan origin. The median age was 21 years [Interquartile Range: 8–32], with 17% (870) children aged less than 5 years (Figure 3). Of the 3,284 adults (age > 15 years), 26% (866) reported being housewives, 14% (456) were private employees, 9% (299) were farmers and 20% (668) were unemployed. At the time of interview, 3,896 household members were present, with microscopy results available in 3,890, representing 74% of all household members. Absent members were most likely to be adults (76%, 1,032/1,359) of whom 78% were male (805/1,032).

The overall prevalence of asexual parasitaemia was 16.3% (634/3,890) with P. falciparum infection present in 290 (46%), P. vivax infection in 248 (39%), P. malariae in 24 (4%) and mixed infections in 72 (11%) (Figure 4a). The prevalence of malaria varied with age. The median age of respondents with P. falciparum parasitaemia was 19 years, with prevalence peaking at 10% (69/707) in young adults aged 15–24 years. In contrast, the median age of those with P. vivax infection was significantly lower (13.5 years, p = 0.01) with a peak of 9.5% (61/642) in children aged one to four years (Figure 4a). The proportion of patients with parasitaemia due to P. vivax infection was greatest in infancy (67% 12/18) falling to 47% (61/130) in children aged one to four years, with no change thereafter (overall 36% 175/486); p = 0.004 (figure 4b). The overall risk of parasitaemia was 19% (402/2067) in Papuans, significantly higher than in non-Papuans (12.7% 232/1,823); p = 0.001. The relative proportion of parasitaemia attributable to non-Papuans was 22% (66/230) in children under 15 years, but rose to 49% (166/338) in adults; p < 0.001 (Figure 4b).

In total 68% (434/634) of respondents with asexual parasitaemia were asymptomatic (afebrile with no history of fever in preceding 24 hours). Although this proportion did not differ for P. falciparum, P. vivax, or mixed infections, asymptomatic carriage was significantly more likely in people with pure P malariae infections (91.7% (22/24); Odds Ratio = 5.3 [95%CI 1.2–33]), p = 0.023. Patients with P. vivax parasitaemia were significantly less likely to be afebrile if they were less than 1 year of age (33% 4/12) compared to 75% (85/113) of older children and 70% (86/123) of adults (p = 0.004 and 0.02 respectively), although this age difference was not significant for the other species (Figure 4a). The risk of fever did not vary with sex or ethnicity.

The density of asexual parasitaemia of any species was significantly higher if the patients were febrile (Table 2). A P. falciparum parasitaemia above 1,734 μl^-1 had a 57% (57/100) sensitivity for predicting fever and a positive predictive value of 61% (57/93). The corresponding threshold for P. vivax was 310 μl^-1, with parasitaemia above this level having a sensitivity of 77% (56/73) and associated with fever in 46% (56/122) of cases.

In 2005, a review of 13,766 laboratory records revealed that 96.6% of patients with malaria had one blood film for each episode of malaria, 3.25% had two blood films and 0.13% had three films. The total number of cases treated as community outpatients were therefore multiplied by 0.967 to account for multiple sampling. Hence the total number of patients with confirmed malaria reported by the surveillance network during the two-year period was 95,886.

In the household survey, a history of fever within the last month was reported in 35% (1813/5255) of household members, of whom treatment seeking behaviour could be assessed in 1015 (56%). Of these respondents, 60% (607) were estimated to have had malaria (based upon their treatment and subsequent microscopy at the time of the survey) of whom 39% (238/607) would have been diagnosed and thereby picked up by the surveillance network. The sensitivity analysis defined the lower and upper estimates as 37.7% (261/693) and 48% (206/426) respectively. After correcting for multiple sampling at the community clinics, the surveillance network identified a total 47,943 reports of malaria per annum for a population of 140,4000 people, representing 39% of the total burden of malaria. The overall incidence of malaria, therefore, was estimated as 876 per 1,000 per year (Range: 711–906). The corresponding breakdown was 512 (Range: 416–529) for P. falciparum, 322 (Range: 262–333) for P. vivax, 15.7 (Range: 12.7–16.2) for P. malariae, and 26.3 (Range: 21.3–27.2) for mixed infections.