Despite high coverage and extensive usage of insecticide treated nets in rural communities of southern Tanzania , partly designed to deter and divert mosquitoes from entering houses , a high number of malaria vectors are still found indoors with an average of 22.22 (CI = 16.93 – 27.51) An. gambiae s.l. and 1.35 (CI = 1.07 – 1.63) An. funestus mosquitoes per trap night per house in Namwawala. In addition, an average of 13.12 (CI = 10.94 – 15.30) An.gambiae s.l. and 2.09 (CI = 1.56 – 2.63) An. funestus were collected in Idete per trap night in a house.
Small houses, constituting the majority of houses in the study area, characterized by relatively low numbers of windows, doors and rooms were associated with relatively high densities of malaria vectors. Although the association of house size and indoor mosquito density remains unknown, it was, however, assumed that smaller houses are likely to concentrate more human odours, which would attract high mosquito numbers. Conversely, houses with more sleeping rooms had a lower density of vectors because they usually have more sleeping spaces, which is likely to encourage consistent use of bed nets by sleepers [31, 32]. Moreover, houses with many rooms are likely to have more nets, which collectively might reduce the number of mosquitoes indoors.
Houses made of mud walls and grass roofs had an increased risk of mosquito bites indoors. Such houses create cooler, darker conditions favoured by resting mosquitoes [33, 34]. Moreover, mud walls as well as grass roofs often have crevices used by mosquitoes to enter the houses unlike cement walls and metal roofs . In addition, lack of or damaged screening over windows as well as open eaves provided entry points and led to increased mosquito abundance inside the houses. These findings are consistent with other studies [16, 35–38] which demonstrated that poorly constructed houses (with mud walls, grass roofs, lack of screening and with eave gaps tend to have increased human-vector exposure), resulting in a higher risk of malaria transmission.
It has been documented that houses with many occupants tend to attract vectors of disease [39, 40]. In this study, the presence of many sleepers in a small house exposed them to a higher risk of An.gambiae s.l. bites but to a lower risk from An. funestus. Large amounts of human emanations from houses with more occupants tend to increase mosquito attractiveness towards that particular house compared to ones with fewer sleepers [41, 42]. The observed inverse relationship between An. funestus and number of occupants inside the house was unexpected; however, it might be due to uneven distribution of An. funestus within the villages. Higher numbers of An. funestus collected during the dry season  were mostly and consistently from a cluster of a few houses located in a particular village hamlet. Therefore, the majority of houses within the sampling area experienced none or low catches.
Furthermore, significant impacts of house risk factors on An. funestus indoor mean catches were not consistent between villages. While this observation remains inconclusive, we postulated the cause to be exceedingly low numbers of An. funestus collected between villages compared to An. gambiae s.l.
Treated nets provided more protective advantages than untreated ones as also observed in previous studies [22, 23, 44, 45]. However, the density of An. gambiae s.l. in Namwawala was higher compared to Idete despite 90% ITN coverage in Namwawala. These results indicate that even at high coverage levels, ITNs still have limitations in reducing the number of malaria vectors entering the houses. Furthermore, recent studies [46, 47] have indicated that poor compliance and usage of bed nets by communities in the tropics is associated with heat discomfort associated with poor airflow caused by bed nets. Although bed nets were procured individually and there was a distribution campaign during the study period, the age of nets as well as usage of ITNs was not systematically investigated in this study, our results illustrate that a risk of transmission remains whenever people are not using treated nets in an optimal way. Improved house designs, and modifications to existing houses could substantially reduce the risk of mosquito-human contact. Although house improvement has been advocated as an efficient intervention for malaria control, the majority of houses in poor rural Africa are temporary and built with minimal material resources. This renders improvements expensive and/or impractical in most rural communities in the short term. Permanent houses (Figure 2b) could be easily and cheaply modified by screening eaves, windows and doors accompanied by community sensitization towards intervention sustainability. Temporary houses (Figure 2a) are less amenable to modifications unless they are rebuilt as more permanent structures. This would have to be addressed through a long-term strategy that sought to build better, inexpensive house models using better construction materials and sustainable financing initiatives, which can be adopted in poor settings. Such an intervention is likely to be beneficial in reducing vector borne diseases and other diseases linked to poor hygiene.