Entomological aspects and the role of human behaviour in malaria transmission in a highland region of the Republic of Yemen

Abstract

Background: The Republic of Yemen has the highest incidence of malaria in the Arabian Peninsula, yet little is known of its vectors or transmission dynamics.

Methods: A 24-month study of the vectors and related epidemiological aspects of malaria transmission was conducted in two villages in the Taiz region in 2004-2005.

Results: Cross-sectional blood film surveys recorded an overall malaria infection rate of 15.3 % (250/1638), with highest rates exceeding 30 % in one village in May and December 2005. With one exception, Plasmodium malariae, all infections were P. falciparum. Seven Anopheles species were identified among 3407 anophelines collected indoors using light traps (LT) and pyrethrum knockdown catches (PKD): Anopheles arabiensis (86.9 %), An. sergentii (9 %), An. azaniae, An. dthali, An. pretoriensis, An. coustani and An. algeriensis. Sequences for the standard barcode region of the mitochondrial COI gene confirmed the presence of two morphological forms of An. azaniae, the typical form and a previously unrecognized form not immediately identifiable as An. azaniae. ELISA detected Plasmodium sporozoites in 0.9 % of 2921 An. arabiensis (23 P. falciparum, two P. vivax) confirming this species as the primary malaria vector in Yemen. Plasmodium falciparum sporozoites were detected in An. sergentii (2/295) and a single female of An. algeriensis, incriminating both species as malaria vectors for the first time in Yemen. A vector in both wet and dry seasons, An. arabiensis was predominantly anthropophilic (human blood index = 0.86) with an entomological inoculation rate of 1.58 infective bites/person/year. Anopheles sergentii fed on cattle (67.3 %) and humans (48.3; 20.7 % mixed both species), but only 14.7 % were found in PKDs, indicating predominantly exophilic behaviour. A GIS analysis of geographic and socio-economic parameters revealed that An. arabiensis were significantly higher (P < 0.001) in houses with televisions, most likely due to the popular evening habit of viewing television collectively in houses with open doors and windows.

Conclusions: The predominantly indoor human biting vectors recorded in this study could be targeted effectively with LLINs, indoor residual spraying and/or insecticide-treated window/door curtains reinforced by education to instil a perception that effective and affordable malaria prevention is achievable.

Fig. 1

The study sites in Taiz Governorate, Yemen, during the wet season in 2004 and 2005: Ukaysh, showing the large oasis (a) and its main perennial stream (b), and a cluster of houses situated on the hill overlooking the oasis (c); Al-Sa’dah showing the Wadi Mae’an’ wetland (d, e) and flooded land at Wadi Al-Ahmar (f)

Fig. 2

Examples of houses at the study sites: a Cluster of older traditional type house structures, each with a single room and no windows; b–d most common types of houses in this region, with one or two rooms and with windows; e two-storey house with cement block upper floor over traditional ground floor; f modern cement block houses with two or three rooms, known locally as ‘Helal’ houses

Fig. 3

Estimated precipitation and mean temperature data (a) and specific humidity (b) at Ukaysh, Taiz governorate, Yemen from January 2004 to March 2005 (http://iridl.ldeo.columbia.edu); c Monthly abundance of Anopheles arabiensis (geometric mean number of mosquitoes per house collected by pyrethroid knockdown catch [PKD], with 95 % confidence interval) at Ukaysh (March 2004–March 2005). The red arrow indicates the date (1 Aug 2004) when The National Malaria Control Programme carried out larviciding at all the known larval habitats of Anopheles at this site

Fig. 4

Numbers of An. arabiensis (log mean numbers/house) collected in houses in Ukaysh (a) between March 2004 and March 2005, with number of occupants (b) and presence or absence of televisions (c) in the same houses. Numbers of An. arabiensis were significantly higher (P < 0.001) in houses with televisions (5.7; 95 % CI 2.8–10.8) than those without (1.4; 95 % CI 0.8–2.1), and were positively correlated with the number of human occupants in the house (Pearson’s Bivariate correlation; r = 0.361; P = 0.005)

Fig. 5

A television set within a house in Ukaysh, Taiz governorate, Yemen. Typically, up to 18 people would gather within this room every night between 19:00 and 23:00 with windows and doors left open for ventilation