More An. gambiae s.l. were caught by the ITT-B than by the HLC or SRB. The ITT-B and SRB caught between 35% and 15%, respectively, of the number of An. gambiae s.l. caught per night by HLC. Note, however, that by applying the ITT-B for three nights in the same sampling site, its relative sensitivity per week matched the HLC for An. gambiae s.l. and exceeded it for Cx. species. Interestingly, male mosquitoes of almost all the species sampled were found more frequently in the ITT-B than any other sampling method. It should be noted that male mosquitoes are just as useful an indicator of success or failure of a larval control programme even though they do not cause disease. Furthermore, male mosquitoes play an essential role in the life cycle of all mosquitoes and monitoring systems for genetic control strategies such as the release of sterile or genetically modified mosquitoes. The results of this study show that the ITT-B was the most efficient method for collecting An. gambiae s.l. and Cx. species per week because it was possible to conduct more intensive sampling with far less effort. SRB, normally considered as the method of choice for recovering resting mosquito populations in a variety of ecological settings, was insufficiently sensitive in urban Dar es Salaam.
Overall, the ITT-B was by far the most cost-effective sampling method. On the basis that the trap can handle more sampling nights than the HLC at a dramatically reduced cost, more sensitive, extensive, intensive and representative measurements of biting density can be determined over larger sampling areas. The fact that the ITT-B samples the vectors with minimal supervision while the HLC requires intense scrutiny and correspondingly substantial resources, are the primary reason for the difference in their overall costs. Whereas HLC involved daily use of a vehicle to distribute the sampling materials to the respective sampling sites, spot checks as well as picking the vectors the next day, the ITT-B involved none of these. The other major differential cost associated with the HLC was the diagnosis and treatment of the HLC catchers in case of any reported fever. By comparison, the ITT-B requires little or no maintenance so after the initial, expensive outlay of purchasing the traps themselves, it is remarkably affordable because these are very durable and the procedure does not require skillful personnel, intensive supervision or medical expense.
The failure of the ITT-B to reduce the proportion of blood fed mosquitoes suggests that exposure of the occupants does in fact occur, probably during the collection process which necessitates opening of the long zipper that bisects the protective panel. Subsequent follow up discussions with the operators revealed that indeed they do receive bites during the collection process. Therefore, an improved design will be required for routine use. One major disadvantage of the ITT-B often reported by the catchers was the fact that the trap was too heavy to be moved from one sampling station to the next by a single person. This problem was later solved by supplying the operators with bicycles. Also, occasionally the trap was reported to attract other insects but none of these were confirmed to be potential mosquito predators. Otherwise, the protocol was generally well accepted by the trap operators and appears to be easy enough for performance to be maintained with relatively modest incentives.
On the other hand, the SRB proved to be very impractical and on several occasions they were either soaked by rain or stolen. It also often proved difficult to retrieve the mosquitoes from the SRB. These problems, combined with their poor sensitivity and other sundry logistical matters appear to rule out the SRB as a candidate tool for routine mosquito sampling in the city of Dar es Salaam. The SRB have been evaluated previously in terms of efficacy and found to correlate poorly with the HLC  and this study reinforces that view. In earlier studies, it was found out that the proportion of adult mosquitoes resting in man-made shelters depended on the availability of alternative resting sites which varies according to location and changes seasonally  and in a recent study that female mosquitoes prefer larger resting sites over smaller ones . Combining these considerations with the poor sensitivity observed, it appears that SRB are unlikely to provide a useful alternative to the HLC for sampling host-seeking malaria vectors in urban Dar es Salaam, particularly under operational conditions.
The correlation results obtained for ITT-B from this effectiveness trial were slightly different from those of efficacy trials by others [3, 8, 22, 23] (Table S2) (Additional file 2). For example, previous efficacy trials by Govella and others  recorded a much stronger correlation between ITT-B and HLC, than seen in this study. This is most probably because this study was carried out under conditions that involved minimal supervision compared to the intensely controlled efficacy trials. Another likely contributor to this weakened association is the fact that more sampling stations across a very heterogeneous environment were included in this study compared to the relatively few sampling stations for the efficacy trials. Nonetheless, the significant positive correlation between the HLC and the ITT-B (Additional file 2) suggest that this approach may be very useful in programmatic setting and provides a reasonably sensitive and accurate reflection of true mosquito biting densities.
Although the use of window traps installed in existing houses and emptied by resident community-based workers has been described as an effective tool for routine monitoring of indoor-residual spray programmes in southern Africa , no other effectiveness study of this kind has been reported for malaria vector trapping methods. The ITT-B not only represents an option for more accurate and representative measurement of human biting rate over a large sampling area, it is also practical and affordable to use in community-based sampling schemes. Nevertheless, it is recommended that the trap be evaluated in the longer term and on full programmatic scales until the effectiveness of this approach in fully representative conditions of routine practice is established. However, the largest remaining concern is probably the surprisingly high proportion of blood-fed mosquitoes caught, suggesting the design needs to be adapted to avoid human exposure during the empting process before it can be adopted as a routine mosquito-trapping tool.