Study design
This is a descriptive study using routinely collected malaria surveillance data. First, the temporal evolution of national malaria cases from 2000 to 2016 was evaluated. Secondly, a space–time cluster-analysis using a Bernoulli model was conducted for the sub-set of information gathered during the 2007–2016 period, which is when both high resolution geographical and temporal data were available.
Study setting
Suriname is a malaria-endemic country along the northern coast of South America. The coastal area has been free of malaria since 1968; however, the interior has recorded a high malaria incidence and prevalence in the early years of this Millennium (around 160 malaria cases/1000 persons at risk per year). Since then, malaria incidence decreased sharply and steadily to elimination level from 2004 to 2009 [1, 2], with sporadic outbreaks in gold-mining areas.
The tropical rainforests in the interior provide excellent habitat to the main malaria vector Anopheles darlingi [15]. Secondary (potential) vectors include Anopheles nuneztovari and Anopheles oswaldoi [16]. The mass distribution of long-lasting impregnated bed nets between 2006 and 2009 and repeated large river flooding in 2006 in high-transmission risk areas are thought to have negatively impacted the An. darlingi populations in these areas during that time [17]. The interior, however, continues to be an important risk area, since this vector has proven efficient in malaria transmission even in low densities [15].
Road infrastructure in the interior of the country is very limited. Priority modes of transportation are boats (dug-out canoes) and small airplanes, as well as all-terrain vehicles in remote mining areas, resulting in challenging logistics for the provision of health services.
Malaria parasite species identified in Suriname include P. falciparum, Plasmodium vivax and Plasmodium malariae. Mixed infections have been reported.
Study population
The study population consisted of all subjects who were tested for malaria, and which are recorded in the Surinamese national surveillance database. The population at risk in Suriname is composed of stable and mobile populations in the interior of the country. The stable populations are Maroon (descendants of African slaves) and Amerindian (native) populations living in tribal villages along rivers in the forests of the interior. Both the Maroon and Amerindian populations consist of several tribes, each with its own language. Being of African descent, many of the Maroons have a Duffy-negative phenotype which prevents them from becoming infected by P. vivax.
Since 2007, the population at risk was extended to include the mobile gold-mining communities in remote areas in the forest. These are mostly migrant miners of Brazilian origin (Portuguese speaking), but also include a small portion of Surinamese Maroons [18], Chinese and nationals of regional (Latin-American) countries. The total number of population at risk varied from 47,372 in 2000 to 84,700 in 2016. This increase was due to both stable population growth and the inclusion of mobile migrants. The number of Maroons and Amerindians are based on health registration data, since most people in the villages are registered at the Medical Mission Primary Health Care organization (Medical Mission) since birth. The Medical Mission is a government-funded, non-governmental organization providing primary health care to the stable communities in the interior. The number of mobile migrants is unknown and varies depending, among other things, on gold availability, gold price and military counter-intervention in neighbouring countries (especially in French Guiana). It is estimated at 20,000 people.
Malaria interventions
The most important malaria interventions during the study period included: (i) passive screening, and health education at central level and in the villages of the interior; (ii) bed net production and distribution in cooperation with local women organizations until 2006; (iii) introduction of artemisinin-combination therapy (ACT) for P. falciparum infections at the end of 2004; (iv) mass distribution of free long-lasting insecticide-treated nets ((LLINs) and retreatment-tablets) in combination with a large awareness campaign between 2006 and 2009, followed by regular distribution of free LLINs at screening points and during active case detection surveys (ACDs) in high-risk areas since then; (v) indoor residual spraying (IRS) in high-risk areas in 2006 (discontinued following a steep decrease in number of cases); (vi) national introduction of rapid diagnostic tests (RDTs) in 2006; (vii) introduction of single-dose primaquine in addition to ACT for P. falciparum infections in 2007; and, (viii) passive and active case detection surveys (ACDs) in remote risk areas (mining areas) since 2009. Implementation of (changes of) nationwide interventions was approved by the National Malaria Board, a multi-sectorial advisory board within the Ministry of Health.
Data sources and collection
The source of this study data is the national malaria surveillance database. It includes all subjects screened for malaria within the Surinamese health system. Access to diagnosis and treatment is free for all. This means that the surveillance data include subjects from the stable and mobile communities, including documented and undocumented migrants. Malaria diagnosis was done by microscopy screening of blood smears (parasite detection in 200 (routine screening) or 500 (non-routine screening) fields of a thick smear) or by RDTs. All RDT results were cross-checked with blood smears. As much as possible people with a positive diagnosis were provided with treatment on the spot.
A system for internal and external quality control of microscopy was in place. Slides (all positives, 10% of negatives) were sent to the national reference laboratory at the Bureau of Public Health (Ministry of Health) for re-check. The Bureau of Public Health took part in the regional External Quality Assurance Programme (EQAP). Microscopy refreshment training was organized on an annual basis (at least during the last decade).
The Surinamese national surveillance system includes the following components: (i) the laboratory of the Bureau of Public Health reports on malaria data from the medical centre of the Bureau of Public Health, from hospitals, from Regional Health Services, from the blood bank and from private laboratories; (ii) the Medical Mission which has 56 clinics in the Interior (8); and, (iii) the Ministry of Health Malaria Programme (Malaria Programme), which targets its interventions at high-risk populations, especially mobile migrant gold miners. The programme has been supported throughout the years by the Global Fund to Fight AIDS, Tuberculosis and Malaria, the Inter-American Development Bank (IDB), the US Agency for International Development (USAID), the Pan American Health Organization and private companies (i.e., Newmont Mining Company). It provides malaria services in a migrant clinic in Paramaribo, operates a small number of border malaria-screening posts along the border with French Guiana, maintains a malaria service deliverer (MSD) network in gold-mining areas and performs ACD surveys in remote mining areas (Fig. 1).
The MSD network consists of lay people from the high-risk (mining) areas and communities (including migrants) who are trained and supervised by the Malaria Programme to provide health information, bed nets and diagnosis and treatment to their peers. Where possible a relationship has been established with local companies (mining and logging) to train MSDs among their personnel.
ACDs were almost exclusively done by the Malaria Programme MSDs and MSD-supervisors. Incidental ACDs were executed by the Medical Mission in the villages. During ACDs, mass screening for malaria was done with RDTs. Cross-checking of RDT results with blood smear took place during the ACDs in the field or after the ACDs at the central level.
National aggregation of malaria data in the surveillance system was done by the Malaria Programme. For the time-trend analysis, the data on subjects with a positive malaria test, diagnosed within the national health structures or identified during ACDs between 2000 and 2016, were used. Data on malaria-risk population, mortality and hospitalization as a result of malaria infection were maintained by the epidemiology unit of the Bureau of Public Health. These data were also used for the time-trend analysis.
To identify significant aggregation of cases in notification points over time or space, a spatiotemporal cluster-analysis was conducted. For this the screened subjects, both tested positive (cases) and negative (controls), registered between 2007 and 2016 by the Malaria Programme were used. This included passive and active screening.
For both analyses, malaria cases were defined as people in whom the, regardless of the presence or absence of clinical symptoms, presence of malaria parasites in the blood was confirmed by microscopic examination. In addition, people who in the absence of a blood smear result, had a positive RDT result were included as cases. Similarly, malaria-negative persons (controls) were defined as people with a negative blood smear result, or people who in the absence of a blood smear result, had a negative RDT result.
Analysis and statistics
Morbidity and mortality trends
For the temporal trend analysis, graphs and trend lines were created using Epi Info™ version 7.2.1.0 (Centers for Disease Control and Prevention (CDC), Atlanta, GA, USA) and Tableau Software, version 9.2. The variables assessed over time were population at risk, number of malaria cases, malaria hospitalizations, and malaria deaths. For the persons screened, the variables gender and age (mean and standard deviation), country of origin, nationality, diagnosis (blood smear or RDT), and health service provider (organization) were assessed. For the positive cases, locality of infection (based on travel history; considering import versus autochthonous cases), and malaria parasite species were included.
Calculations were made for the Annual Parasite Index (API = total confirmed cases in a year × 1000/total population); the proportion of non-indigenous (imported) malaria cases (the total confirmed imported cases × 100/total confirmed cases); the proportional contribution of each parasite species to the total number of cases (total number of confirmed cases for one species × 100/total confirmed number of cases for all species); the proportion of cases reported/notified for each surveillance system (the number of cases per surveillance system × 100/total number of cases).
Trend lines on number of autochthonous malaria cases were evaluated with a linear trend model to determine significant changes over time (significance at p < 0.05). For the period 2014–2016 the autochthonous cases were mapped using Tableau software, version 9.2, in order to assess the presence of recent areas of transmission or so-called ‘hot spots’. For the purpose of mapping all national cases without known locality of infection were mapped in the capital of the country; Paramaribo and the Coastal Area of Suriname however have been malaria-free since 1968.
Spatiotemporal clusters of malaria cases by notification points
The space–time cluster analysis was conducted using SaTScan software (v.8.0) [4]. Specifically, a Bernoulli model was used to evaluate the distribution of positive malaria cases in notification points relative to the control group (which were defined as the suspected cases that were tested but were negative for malaria). The spatial and temporal unit of analysis was locality of notification and month, respectively. Notification points included the Malaria Programme clinic in Paramaribo, the border screening points (Albina, Tumatu, Antonio do Brinco) along the border with French Guiana and the MSD service points in the mining areas in the interior of Suriname. These notification points almost all exclusively provide malaria health services. For ACDs, the area of ACD was recorded as a notification point, with one coordinate for the survey.
A maximum spatial and temporal window of 50% of the study area and study period, respectively, was used as recommended by Kulldorf [19]. Significant clusters of high incidence (Bernoulli model, p < 0.05) were mapped using Google Earth Pro Version 7.3.0.3832 (32-bit) and cluster characteristics were described in a Table.
Ethics approval
Ethics approval was received from the Committee for Human-centred Scientific Research (Ministry of Health) in Suriname (VG-15-17). Exception of ethical review was obtained from the Pan American Health Organization (PAHO) Ethical Review Committee considering that the study is based on routine programme information (PAHO-2017_07_066). The confidentiality of the study subjects was protected and individual data were not shared.