Background
The Roll Back Malaria, United Nations Millennium Development Goals and World Health Assembly universal access and coverage targets for malaria prevention and treatment have been established to reduce disease transmission. To meet these targets malaria control interventions are now being scaled up [
1,
2].
To successfully control malaria, programmes must use current tools efficiently. In moderate to high transmission areas this requires the combination of effective vector interventions, either indoor residual spraying of insecticides (IRS) and/or insecticide-treated nets (ITNs), with effective drug treatment. To achieve this, continuous surveillance, monitoring and evaluation (M & E) need to be integrated into the malaria control programme. There are very few examples of good malaria M & E, which include parasitological, disease transmission and entomological data assessment. These include Bioko island [
3,
4], the cross border Lubombo Spatial Development Initiative (LSDI) [
5], Eritrea [
6] and Zambia [
7].
In Mozambique, the National Malaria Control Programme (NMCP) used DDT for IRS before a change in policy in 1993, when the pyrethroid lambda-cyhalothrin was introduced in the southern part of the country. With the discovery of
Anopheles funestus resistant to pyrethroids in South Africa [
8], the LSDI implemented an insecticide resistance monitoring programme in southern Mozambique that showed in 1999 that both
An. funestus and
Anopheles arabiensis were resistant to pyrethroids [
9,
10] due to elevated p450s [
11]. This resulted in an informed insecticide policy change to the carbamate bendiocarb for IRS [
5]. Bendiocarb was sprayed bi-annually in southern Mozambique until 2005, when, due to the high economic costs associated with this insecticide, an operational change was made back to DDT. No resistance has yet been detected to DDT in Mozambique.
In 2005 the U.S. Government initiated the President's Malaria Initiative (PMI) [
12], which supported the rapid scale-up of malaria prevention and treatment in 15 countries in Africa, including Mozambique. In central Mozambique, an IRS programme was initiated in six districts in Zambézia province. Previously, the NMCP had carried out vector control with sporadic IRS and fogging between 1995 and 2003 in this area. In 2005, the NMCP resumed IRS in Zambézia in three districts, using DDT. Limited expansion of activities occurred in 2006 to cover 5 districts and this effort was strengthened in 2007 by PMI. The IRS was focused on densely populated areas using DDT or lambda-cyhalothrin, the latter being applied on structures not suitable for DDT (i.e. finished or painted walls). In 2009, pyrethroids were the sole class of insecticides purchased for IRS, although all remaining stocks of DDT were sprayed during that year [
13].
Zambézia province was used to pilot and refine a new Malaria Decision Support System (MDSS) [
14], which was implemented in 19 sentinel sites in Zambézia province in 2006 and served to evaluate and monitor the interventions described above. The objective of this work was to determine whether the system, which was continuously improved during the course of the study, could be deployed in a resource poor setting, with low level infrastructure and limited human capacity. The utility to control programme managers of an integrated surveillance system providing real time monitoring through a number of performance and impact indicators, was clearly demonstrated.
Discussion
The success of the IRS-based LSDI programme in southern Mozambique [
3] and the increase in malaria cases in northern Mozambique prompted the MOH and the NMCP to re-initiate a comprehensive vector control programme in Zambézia province. The MDSS project was embedded within the national programme, and demonstrated the feasibility and potential benefit of evaluating vector control interventions even in very resource poor settings. Intervention coverage and its impact on malaria vectors and on community prevalence of malarial infection were monitored as indicators of progress.
Anopheles gambiae s.s and
An. funestus were the dominant malaria vector species in Zambézia province, where
An. gambiae s.s. was predominant. This confirms results from previous studies of
Anopheles distribution in northern Mozambique [
22]. Although
An. arabiensis was detected in this area, it was in low abundance with no sporozoite positive specimens found, indicating low transmission potential.
Mosquito species abundance changes naturally during annual cycles, depending on climatic variables [
15,
23]. Abundance of both
An. gambiae s.s and
An. funestus displayed seasonal variations, but were present throughout the year.
An. gambiae peaked in January to May, at the height of the rainy season, whilst
An. funestus was highest in April to August in 2007, towards the end of the rainy season, thus prolonging the malaria transmission season. However, in 2009,
An. funestus peaked at the same time as
An. gambiae. With this potential pattern of vector abundance the vector control interventions need to remain effective for a duration of eight months each year. This requires either an insecticide with a long residual half-life, such as DDT, or spraying multiple rounds each year.
After three rounds of IRS with DDT a dramatic impact on vector populations was observed. This was similar to that reported for other IRS programmes, including the LSDI programme in southern Mozambique [
5] and Bioko, Equatorial Guinea [
3], where vector densities declined when effective insecticides were applied. In Zambézia, abundance of the vectors fell significantly from 1 to 0.6 and 1.6 to 0.2 per exit trap per 100 nights for
An. gambiae and
An. funestus respectively. This corresponded with a reduction in the sporozoite rates from 4% to 1% in
An. gambiae and 2% to 0% in
An funestus. While remaining low, numbers of
An. arabiensis collected after the interventions doubled, but as none of the samples tested positive for sporozoites, this species appears to have little or no involvement in malaria transmission. However, since
An. arabiensis is known to be an effective malaria vector in other parts of Mozambique [
5,
24,
25] and elsewhere, it is essential that this species continues to be monitored.
The vector control intervention has impacted on disease in the human population, as measured through prevalence surveys over three spray seasons. In the first year, IRS was carried out by the NMCP in Oct/Nov with limited resources. Only 50% of targeted houses were sprayed in each district. There was no measurable impact on disease prevalence in IRS sites after this intervention. In 2007 PMI support for the NMCP in this area was initiated, and IRS coverage increased to >80%, the rate recommend for effective IRS by WHO [
26]. After this intervention there was a significant reduction in
P. falciparum prevalence in children aged between one and 15 years in all IRS districts except Mocuba, where IRS had very little impact.
LLINs provide personnel protection, and with high coverage also offer community protection [
27,
28]. Data from Maganja da Costa (Additional file
1) show that a substantial decrease in parasite prevalence in the human population was associated with LLIN distribution in 2008. In this area, the abundance of
An. gambiae and
An. funestus declined from 52 and 88 per window trap per 100 nights to just 1.6 and 0 respectively. In 2006-2007, the rainfall was exceptionally high in this district leading to higher than usual rice cultivation in 2007. This may have resulted in increased vector abundance in the sentinel sites and increased transmission, which is reflected in the significant increase in prevalence in the 2007 survey. Since 2007, there has been limited rainfall, resulting in no rice cultivation. Both changes in rainfall pattern [
29,
30] and alteration in agricultural practises [
31] will impact on the vector population, in this case to the advantage of malaria control. These severe climatic changes in Maganja da Costa make it difficult to assess the true impact of the LLIN distribution in this area over such a short time. Although not as severe, a climatic change was documented throughout the study area, and needs to be considered when interpreting the impact of interventions.
Prior to the detection of pyrethroid resistance in
An. funestus in 2010 in sentinel sites in Mocuba, no insecticide resistance had been detected in the malaria vectors in this region to any insecticide. A low frequency of resistance may have been masked, as WHO discriminating doses are set at double the insecticide dose that gives 100% mortality of the least susceptible anopheline mosquitoes, making them good indicators of resistance only when resistance levels rise significantly in the mosquito population. Prior to 2010, and the establishment of an insectary in Zambézia, all resistance assays were carried out on wild-caught mosquitoes. This potentially affected resistance testing, as mosquitoes were not standardised for age, physiological state or pre exposure to insecticides. No DDT resistance was detected in the province. The detection of both bendiocarb and pyrethroid resistance at the same time, an unusual pattern of resistance, but one found in 2000 in southern Mozambique, thought initially to be due to cross resistance of elevated monooxygenase [
11] but later data showed both elevated monooxygenase and altered acetylcholineesterase, a major mechanism of carbamate resistance, in
An. funestus populations [
32] suggests that this resistance may have spread.
Monitoring insecticide resistance mechanisms is an integral component of resistance management, allowing for informed decisions on insecticide choice and resistance management [
33], as a number of mechanisms may give rise to different cross resistance patterns [
34,
35]. Within the national programme over this time period it was not possible to monitor for resistance mechanisms, due to the lack of a cold chain to a suitable laboratory. New developments in molecular techniques should make this possible in the future [
36,
37]. However, as the resistance pattern observed in
An. funestus here is very distinct, and similar to that previously detected in southern Mozambique [
11,
32,
38], with resistance to both carbamates and pyrethroids segregating together it is likely that this resistance has not arisen
de novo in central Mozambique. This has been shown to be the result of increased levels of monooxygenases and acetylcholinesterase in the same population [
32,
38].
Currently Mozambique national policy has been to use lambda-cyhalothrin instead of DDT for IRS. The former was used in the 2009 spray round even though pyrethroid resistance has now been detected in Zambézia and previously in southern Mozambique [
32,
38‐
40].
In 1996, South Africa altered its insecticide of choice from DDT to the pyrethroid deltamethrin. After pyrethroid resistance was selected the recorded malaria cases increased more than six fold from 1995 to 1999. During this period there was also an increase in malaria drug resistance [
41], but entomological surveys showed that
An. funestus, which had previously been eliminated by DDT, had re-established in South Africa [
8] from southern Mozambique due to the protection against IRS acquired through pyrethroid resistance. With the re-introduction of DDT [
42] combined with the introduction of an effective drug [
41] malaria control was once again successful in South Africa.
The recently detected low level pyrethroid resistance in
An. funestus in Zambézia province reported here for the first time, and in neighbouring Likoma Island, Malawi, that is close to Zambezia [
43], is likely to become operationally significant and should be closely monitored, as already there are anecdotal reports of
An. funestus resting in pyrethroid sprayed houses and an increase in malaria cases in Zambézia province.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
All authors have read and approved the final manuscript. APA-carried out the entomological investigations and analysis and drafted the manuscript, IK and AR carried out all epidemiological analysis, helped interpret the results and contributed to writing the manuscript, NC-supported entomological work in country, VR-carried out diagnostic tests on mosquitoes from window exit traps, DM-supported field work, SC-carried out all GIS, RM-assisted in design of programme, CW and AS established insecticide resistance in the field, MC-carried out conceptual design and development of MDSS, JH-input into the writing of the manuscript, MC-was responsible for the overall project.