Hispaniola is the only Caribbean island where malaria is endemic, and the greatest burden is borne by Haiti, located on the western half of the island [1]. Anopheles albimanus is the predominant vector transmitting malaria on Hispaniola, and the causative agent is Plasmodium falciparum [2]. The national policy of the Haitian Ministry of Health [Ministère de la Santé Publique et de la Population (MSPP)] is to screen individuals with symptoms consistent with malaria using one of the WHO recommended rapid diagnostic tests (RDT); if positive, two slides each carrying thick and thin blood smear are obtained; one is read by local hospital technicians, while the second is sent to the national laboratory in Delmas [Laboratoire National de la Santé Publique (LNSP)] for confirmation. Although microscopy remains globally the gold standard for diagnosing malaria [3, 4], the accuracy of slide reading and their quality control remains questionable in Haiti [5]. Malaria remains high on the differential list of non-respiratory acute febrile illnesses in Haiti, despite the fact that in recent studies only about 3–17 % of these cases are confirmed as malaria by either microscopy or RDT [6, 7]. Official reporting of malaria cases to the LNSP is recognized as under-representing the true burden of malaria [2].

Since the vast majority of malaria cases on Hispaniola occur in Haiti, it should remain the focus of malaria control and future eradication efforts [2, 8, 9]. Eradication of malaria from Hispaniola is viewed as a reasonable goal, considering that it is an island isolated geographically from other malarious lands. Further, the disease is transmitted by a comparatively poor, exophilic vector [10], and chloroquine-sensitive P. falciparum appears to be the single agent of malaria [2, 9, 11]. However, the effort to eliminate malaria on Hispaniola will face frequent natural disasters, poor infrastructure, lack of skilled laboratory technicians, and geographic barriers presented by mountainous terrain, combined with limited resources available to people seeking medical help and their negative perception of the value of the healthcare provided to them [12]. All of these factors have hampered past efforts at control [2].

Rapid diagnostic tests have been identified as a key component of the current control strategy to prepare for eradication, thus keeping microscopy as only a confirmatory secondary test [13]. RDTs have a ≥75 % sensitivity at 200 parasites/µL of blood [14] with a false positive rate ≤10 % and an invalid rate ≤5 %, according to WHO published guidelines. In July 2010 Haiti MSPP approved the use of three RDTs in the country: First Response Malaria Ag HRP2 (Premier Medical Corporation Ltd., Watchung, NJ, USA), CareStart Malaria HRP2 (Pf) (Access Bio, Inc, Monmouth Junction, NJ, USA), and SD Bioline Malaria Ag Pf (Standard Diagnostics, Inc., Yongin, Korea) [15].

Moving beyond the malaria control phase to the elimination phase, it is suspected that in comparison to the whole population carrying malaria parasites, the febrile cases represent only the tip of the iceberg. The sub-population harbouring asymptomatic, sub-microscopic infections remains obscure. A recent review has described the potential for P. falciparum to persist in an asymptomatic host for more than a decade, making these infections potentially very important in elimination efforts [16]. Further, the importance of these asymptomatic infections rises when transmission levels are low such as in Haiti, where the overall official prevalence seems to range between 0.1–1 % [2, 17]. Despite the low prevalence of febrile cases, multiple studies have pointed out the high prevalence of asymptomatic carriers using only microscopy for diagnosis [7, 18]. Additional sub-microscopic infections remain unidentified, and will require implementing new survey tools capable of identifying them.

For the past five years malaria transmission in Haiti has been fairly stable, with a promising low level of overall transmission. However, in Africa and other areas around the world  where the transmission levels drop significantly, substantial numbers of sub-microscopic carriers remain, enabling the parasite to restart transmission when conditions are right [19–22]. In Tanzania, the prevalence of asymptomatic malaria infections as determined by microscopy was 1.9 %, whereas by real-time quantitative nucleic acid sequence-based amplification with a lower limit of detection, it was 32.5 % [22]. In this study a scheme is presented for testing field-collected samples from remote and undeveloped areas in Haiti to identify sub-microscopic infections using qRT-PCR in an attempt to further understand disease transmission and associated risk factors.

Using the synchronized ring stage cultured parasites in a series of blood dilutions, the limit of detection of the qRT-PCR assay for P. falciparum was 0.0003 parasitized red cells per µL of whole blood (0.3 per mL) (Fig. 2a). In testing serial dilutions of RNA extracted from a blood sample with a higher parasitaemia, the results were highly correlated with RNA extracted from the blood dilution series, however the limit of detection for the diluted RNA was even lower, 0.0001 parasitized red cells per µL (0.1 per mL) (Fig. 2b). Assays to determine the limit of detection of this method were conducted three times independently with parasites from in vitro culture to demonstrate reproducibility before using it with field-collected samples. The standard curve obtained using RNA extracted from the blood dilution series (Fig. 2a) was used to estimate the parasitaemia of all field samples (Fig. 2c), since extraction of RNA from the blood samples used to prepare this curve most closely corresponds to the preparation of RNA from the field samples. Samples were collected from rural sites in six departments across Haiti, and all samples were tested for malaria parasites using RDT, microscopy and qRT-PCR. Among the 563 samples tested, ten and 16 were found positive for malaria by RDT and microscopy, respectively. Using the qRT-PCR test to assess the infection status of these subjects, an additional 92 were identified for a total of 108. The majority (55.6 %) of these infections hosted a parasitaemia of less than one parasite per ml of blood (Fig. 2c). Results by department for all tests are shown in Table 1. All samples testing positive by the qRT-PCR were re-extracted for total nucleic acid using a modified protocol of the same extraction kit as per the manufacturer’s protocol. The assay was then repeated using the same reaction conditions with and without the reverse transcription step to evaluate the difference in detecting rRNA and genomic DNA, respectively. By qPCR only 20 of the 108 qRT-PCR-positive samples were identified as positive.

Location	RDT			Slide			RT-PCR
Department	(n)	No. Pos.	% Pos.	(n)	No. Pos.	% Pos.	(n)	No. Pos.	% Pos.
Artibonite	108	0	0	108	0	0	108	3	2.8
Central Plateau	122	0	0	122	0	0	122	6	4.9
Grand Anse	194	10	5.1	194	14	7.1	194	79	40.7
Nord	39	0	0	39	1	2.6	39	10	25.6
Ouest	63	0	0	63	0	0	63	2	3.2
Sud-Est	37	1	2.7	37	2	5.4	37	8	21.6
Total	563	14	2.4	563	17	3.0	563	108	19.1

Overall 3.0 % of samples tested were positive for malaria by microscopic analysis, 2.4 % by RDT and 19.1 % by qRT-PCR. However, a large variation in prevalence of infection was detected across departments. Departments having a low prevalence of malaria infections (i.e., fewer than five sub-microscopic carriers per 100 in the population) are Artibonite, Central Plateau and Ouest. High prevalence (i.e., more than 20 sub-microscopic carriers per 100 in the tested population) was observed in the Nord, Sud-Est and Grand-Anse Departments. Thus, a large reservoir of asymptomatic, sub-microscopic carriers is present in various locations across the peninsula interspersed with departments with very low prevalence.

A survey method capable of identifying virtually every individual infected with the parasite will be crucial in the malaria elimination phase. It will allow making a reliable estimate of the proportion of sub-microscopic infections which must be factored in when developing intervention plans (e.g., mass drug administration or test and treat), and, secondly to test the efficacy of intervention measures after implementation to evaluate success [22]. PCR has been extensively used as the most sensitive survey tool with a detection limit of 0.01–0.2 parasites/µL of blood [20]. PCR depends on detecting genomic DNA, commonly a segment of the small sub-unit ribosomal RNA gene due to its multiple copy numbers and availability of relatively conserved regions. The power of detecting the ribosomal RNA itself has been recognized by others [22, 28–32], and by qRT-PCR the limit of detection is as low as 0.0003 parasites/µL of blood, as reported here. The limited use of qRT-PCR rather than conventional one step or nested PCR can be attributed to factors such as the potential challenge of maintaining RNA quality and the increased cost of using a probe-based assay. The value of the extremely low limit of detection offered by the qRT-PCR test employed here is seen in the large number of asymptomatic carriers identified which were undetected by any other method, including qPCR.

The qRT-PCR assay described here has the potential to detect blood stage infections from the first emergence of the parasite into the blood from the liver throughout the course of blood stage infection, even during what appears to be an asymptomatic, chronic phase. A single exo-erythrocytic (EE) stage schizont releases merosomes into the blood stream, which ultimately rupture releasing tens of thousands of merozoites. Assuming an average blood volume of 5 L and a single exoerythrocytic stage schizont releasing merozoites, then two or more infected red cells may be expected per ml of blood to initiate the asexual growth phase. To detect this parasitaemia, the test must have a limit of detection equal to or lower than 0.004 parasites/µL of blood. The qRT-PCR targeting the small ribosomal RNA of P. falciparum with a limit of detection of 0.0003 parasites/µL of blood exceeds this threshold by an order of magnitude. Comparing the qRT-PCR assay to qPCR on total nucleic acids using the same primer sets and reaction conditions, excluding the reverse transcription step, C_t values for qRT-PCR were an average of 14 lower than for qPCR for the same sample. The large number of rRNA molecules in each parasite (>10⁴) [33], their presence in free merozoites, and the fact that they are packaged separately in ribosomes which may be released as separate particles upon cell lysis, all likely contribute to the ability to detect as few as three parasitized red blood cells per 10 ml of blood, even when sampling only 0.4 ml of blood and ultimately assaying RNA from only 5 % of the blood sample.

In order for Haiti to reach the pre-elimination phase of malaria, the number of active cases must be less than one per 1000 population at risk in preparation for a reduction to zero locally acquired cases. Haiti would have to maintain zero locally acquired cases for at least 3 years prior to becoming certified malaria-free country by WHO [34]. The current status seems ambiguous, since official records indicate a reduced overall number of clinical cases from ~80,000 in 2010 to ~17,000 in 2014 [6] and hyperendemic foci of malaria remain despite current control efforts [7, 18]. All official published data acquired from hospitals and dispensaries across Haiti represent only data from febrile cases, yet in an area experiencing high transmission only 17 % of all acute febrile illnesses were attributed to malaria [7].

Comparatively high malaria prevalence has been reported previously in the Grande Anse, Nord and Sud-Est Departments [7, 18], however, none of the reported data accurately captured the sub-microscopic levels of infection. There is no national policy on active cross-sectional surveillance to monitor number of asymptomatic or sub-microscopic carriers of parasite in the country. Needless to say Haiti currently has no tools available to identify the sub-microscopic population, but counting the sub-microscopic infections found in these departments raises the already high prevalence to a remarkably high level. Over the past few years there has been an elevated level of malaria transmission in Grand-Anse. A community-based survey there found that gametocytes were rare in the asymptomatic population (0.9 %) [7], however the survey methods used to detect gametocytes have the same higher limit of detection as microscopy for malaria diagnosis as described here. Thus, it is expected that a higher percentage of gametocytaemic individuals would have been found in this population, if assessed by PCR-based methods. In the present study 40 % of the asymptomatic population in the same area were infected with the parasite. In spite of the low levels of parasitaemia in this population, it is likely over time that they produce gametocytes in sufficient quantities to contribute in disease transmission [20]. Given the large percentage of individuals having such a low parasitaemia (Fig. 2c), it is highly likely that they have chronic infections, with only a small fraction sampled at the very onset of a new malaria infection. Despite this logic, whether the parasites were persistently present in the blood stream from a previous infection or recurrent re-infections is yet to be studied. Further, the methodology used does not necessarily detect only viable parasites, rather stable nucleic acids derived therefrom, thus further analysis is required to be able to accurately estimate the parasite burden in individuals positive by the qRT-PCR test. The study is also limited having screened only females of child-bearing age. Whether or not the same level of high prevalence is seen in males, elders and children in the population requires further investigation.

In spite of the availability of bed nets to 64 % of the population, no statistical significance for their proposed role in protecting individuals from infection was found. This finding is consistent with the fact that the main vector for malaria transmission in Haiti is An. albimanus, an outdoor-biting mosquito with a preference for blood meals at dawn and dusk [35]. The survey showed that those with history of previous infection or with a household member that is or was infected are twice as likely to be positive for sub-microscopic infection. The results were statistically significant and come in agreement with the observation, that those who are closely associated with others infected with malaria are more likely to be infected. Finally, travel outside the screening site was a protective factor against infection. Further information relative to the pattern of travel, the length of absence, and travel destination(s) is needed to understand more clearly the basis of the protective effect, however it is consistent with the interpretation that transmission is focal within sites and that travel to other locations in Haiti where transmission is lower is protective.

There is a large variation in malaria prevalence among study sites, and these results suggest that using this highly sensitive qRT-PCR based survey tool would facilitate accurately mapping the reservoirs of malaria across the country. Given that samples obtained in a field setting are stable for transportation to established laboratories for rapid completion of the assay, the results can be obtained in a timely fashion for mapping purposes and for assessing effectiveness of elimination efforts. Obtaining such data will be crucial to effectively guiding the utilization of resources available for elimination and evaluating success.

The findings of this study demonstrate the superior efficacy of the qRT-PCR as a surveillance tool in detecting malaria infections, including submicroscopic infections, as compared to other available tools frequently employed including PCR. DRS provides a highly useful preservative for parasite nucleic acids in field conditions, making this assay feasible for use under field conditions in underdeveloped countries by allowing successful transportation of stabilized samples to established laboratories for analysis. The study identified a large number of sub-microscopic infections in Haiti that were not previously described. Plans for malaria elimination on Hispaniola must account for the large number of asymptomatic, sub-microscopic infections present in Haitians in regions where malaria transmission is found.