correspondence Mark Rowland, London School of Hygiene and Tropical Medicine, Keppel Street, London WCIE 7HT, UK
Summary We conducted a community-randomized controlled trial in Pakistan to determine the efficacy of indoor residual spraying with alphacypermethrin ('Fendona', Cyanamid, NJ, USA), applied at 25 mg/m2, to prevent falciparum and vivax malaria. Wettable powder (WP) and suspension concentrate (SC) formulations were tested against an unsprayed control in an irrigated rice-growing area of Sheikhupura district, Punjab province. The study area of 180 km2 was divided into nine sectors, which were assigned at random to control, WP, or SC treatments in replicates of 3. Sentinel villages were selected from each sector for entomological and disease monitoring. Malaria was monitored by fortnightly active case detection (ACD) and by cross- sectional parasite surveys on schoolchildren. Mosquito populations were monitored by space spraying of rooms and by cattle-landing catches. The spray campaign took place in June 1997 and covered 96% of compounds. During the 7 months after spraying, the incidence of falciparum malaria was 95% lower and that of vivax malaria 80% lower in WP-sprayed than unsprayed sectors. Similar results were obtained for sectors sprayed with the SC formulation. Cross-sectional surveys gave estimates of efficacy comparable to those obtained by ACD. Anopheles culicifacies was 80% less abundant and A.stephensi, the predominant anopheline, was up to 68% less abundant in sprayed areas over the 7-month period. Reductions in anopheline parous rates indicated that the single-spray treatment was effective for the entire transmission season. Sprayed surfaces lacked odour, which contributed to the popularity of the campaign. Alphacypermethrin is a promising insecticide for the control of malaria in Pakistan and South Asia generally.
Falciparum and vivax malaria are major health problems in Pakistan. In the last decade there has been a sixfold increase in falciparum malaria, which now comprises 42% of all malaria cases recorded by the National Malaria Control Programme ( Shah et al. 1997 ). Factors associated with the upsurge include the spread of chloroquine resistance across the country ( Rowland et al. 1997a ; Shah et al. 1997 ), warmer autumns favouring prolonged transmission ( Bouma et al. 1996 ), and a chronic decline in vector control activities. Indoor spraying of residual insecticide is the primary method of malaria control in Pakistan and South Asia, and has been shown to prevent morbidity and economic losses ( Dezulueta et al. 1980 ; Mills 1991; Rowland et al. 1994 ; 1997b). The organophosphorus insecticide malathion has been the mainstay of malaria control in Pakistan over the past 25 years and still gives adequate control if campaigns are properly targeted and managed ( Rowland et al. 1997b ), despite malathion resistance in several species of anopheline ( Rathor et al. 1985 ; Suleman 1993). The only synthetic pyrethroid to be evaluated in Pakistan is lambdacyhalothrin, which performed better than malathion in Afghan refugee camps in North-west Frontier Province ( Rowland et al. 1994 ). Further trials on pyrethroids as house-spraying agents are warranted.
The National Institute of Malaria Research and Training (NIMRT) has conducted several efficacy and safety trials of new candidate insecticides under the WHO Pesticide Evaluation Scheme. Alphacypermethrin ('Fendona', Cyanamid), an alphacyano pyrethroid, is being proposed as a new agent for indoor residual spraying and treatment of bednets ( WHO 1998). We describe a community-randomized controlled trial involving two formulations of alphacypermethrin for indoor residual spraying undertaken in Sheikhupura district, Punjab province.
Materials and methods
The wettable powder (5% WP) and suspension concentrate (10% SC) formulations of ‘Fendona’ were tested against a no treatment control. The study area incorporated three Union Councils and covered 180 km2 of Sheikhupura district, about 60 km west of Lahore, the provincial capital. The area was mapped by NIMRT, divided into nine sectors of approximately equal population size and area, and then each sector was assigned at random to untreated, WP, or SC spraying in replicates of three. All 60 villages within the study area were included in the trial. The study area was a flat, irrigated, agricultural plain, with rice as the main crop during the malaria season (June–November). The area appeared environmentally homogeneous. At night, cattle, water buffalo, and other domestic animals were kept in the villages in owners' compounds, and hence almost all sources of blood meals and indoor resting sites were situated in the villages. House compounds were constructed from mud and brick and separated by high perimeter walls.
Census surveys and spray operations
Village elders attended meetings to learn about the study and to give their consent to the project. Two or three sentinel villages totalling 2000 population were selected from each sector for disease and entomological monitoring and enumerated. The population and number of houses in sentinel villages were, respectively, 6017 and 1206 in WP sectors, 5677 and 1124 in SC sectors, and 6567 and 1328 in control sectors; hence sentinel villages were consistent between treatment groups. Sentinel villages were chosen carefully so that the distance to villages in a different treatment group was at least 1.5 km.
Supervised spray operations were completed within 24 days in June 1997. The target dose was 25 mg alphacypermethrin per m2. Spraymen used Hudson X-pert spray pumps (Hudson, IL, USA) to spray living quarters, storage rooms and animal shelters. Each sprayman sprayed 9–10 compounds per day. Spray coverage was 96% of compounds (2665 compounds sprayed) in WP sectors and 97% (1991 sprayed) in SC sectors.
Active case detection
To measure a reduction in malaria incidence of 50%, given an incidence of 50 episodes per 1000 person years, we estimated that 2000 people in each of the nine sectors should be followed by ACD. This required the survey teams to visit 400 houses in each sector every fortnight. Any member of a household reporting fever during the previous 3 days had a blood smear taken and was treated presumptively with chloroquine. Confirmed cases were given a full course of 25 mg chloroquine per kg body weight over 3 days. ACD surveys were started in April, 2 months before the spray campaign, and continued until January.
Cross-sectional surveys were undertaken in April–May and September, i.e. before and after spraying. One or two schools were selected from sentinel villages in each sector and thick and thin smears were taken from 200 to 300 children aged 5–15 present on the day. Smears were stained with Giemsa, and 200 fields were examined before any smear was classified as negative. Children positive for parasites were given a full course of chloroquine.
Space spray catches were undertaken fortnightly in two sleeping rooms, two storage rooms, and two animal shelters in each sector. The monitoring rooms were deliberately left unsprayed, and the same rooms were used throughout the study. Female anophelines were identified to species level and scored. To determine parous rates, up to 100 unfed and freshly fed specimens were dissected per night; if the number caught exceeded 100, the dissections were done according to the proportion of each species in the overall catch.
One night of buffalo bait collection was carried out from dusk to dawn, inside and outside an animal shelter, in each sector each month. Buffalo were preferred to humans as buffalo landing catches were easier to do, presented no health hazard, and yielded more mosquitoes. The same animals were used throughout. Catching stations were sprayed at the same time as the rest of the village.
Residual activity on mud, wood, thatch, brick, and cement was monitored every 2 weeks using standard WHO bioassay cone techniques. Forty-five mosquitoes were exposed for 30 min in each test. Susceptibility to alphacypermethrin was monitored using standard WHO resistance test kits. Exposure was to 25 mg/m2 for 1 h, and mortality was scored after 24 h recovery. Climatic variables were recorded daily at the Lahore Meteorological Station.
Overall incidence rates were calculated for each treatment group for the periods before and after spraying, and nonparametric tests (Kruskal–Wallis and Wilcoxon rank sum tests) were used to test for differences between treatment groups. The point-estimates of treatment effects were calculated using the overall data for each treatment arm, but confidence intervals are not reported as they assume independence of observations and are therefore incorrect. As there were only three sampling units (i.e. sectors) in each treatment arm, it was not possible to obtain precise estimates of effect adjusting for the nonindependence (intracluster correlation) of individuals within the same sector ( Donnar & Klar 1994). The same procedure was used to analyse the cross-sectional prevalence data. Mosquito densities were analysed using analysis of variance and multiple range tests (least significant difference method) on log-transformed data. Parous rates were examined using χ2 tests.
The mean monthly temperature peaked (36°) in June and fell to 27° by November. The first significant rains fell in July, at the onset of the monsoon. Humidity rose to 65%, conducive to transmission of falciparum malaria, and remained at that level until November.
Active malaria case detection
In the preintervention period from May to June there was no significant difference in incidence rate between any of the 3 intervention groups for falciparum (Kruskal–Wallis, χ2 = 0.4, d.f. = 2, P = 0.81) or vivax malaria (χ2 = 2.5, d.f. = 2, P = 0.29) ( Table 1). Overall falciparum incidence was around 5 episodes per 1000 person years (ppy) while vivax incidence was 10 times higher. In the postintervention period from July to January, incidence rates in control and insecticide treatment groups diverged significantly ( Table 1). The overall incidence of falciparum malaria rose to 29 ppy in the control group but remained < 3 ppy in the insecticide treatment groups (Kruskal–Wallis, χ2 = 7.2, d.f. = 2, P = 0.03). Similarly, the overall incidence of vivax malaria was 19 ppy in the control but fell to 4 ppy in the SC and WP groups. The result was not significant using the Kruskal–Wallis test (χ2 = 4.3, d.f. = 2, P = 0.11). As there was no evidence that one formulation was better than the other against falciparum (signed rank, P = 0.51) or vivax malaria (signed rank, P = 0.83), the data from both insecticide treatment arms were grouped and compared with the control, giving a significant difference (signed rank test, P = 0.02). Estimates of incidence rate ratios are shown in Table 1. The protective efficacy, 100*(1-IRR)%, of both insecticide treatments was 90–95% against falciparum malaria and around 80% against vivax malaria. Malaria was fairly rare and seemed to affect all age groups equally (P = 0.28).
Table 1. Incidence of malaria per 1000 person years before and after indoor residual spraying
Figure 1 shows the changes in monthly incidence rate. Vivax malaria incidence peaked in late spring and declined after the monsoon. Falciparum malaria increased significantly in August and peaked in October. Transmission of both species was much lower in the sprayed sectors. Figure 2 presents the data in terms of monthly slide positivity rate (ratio of microscopy-confirmed malaria cases to all febrile illness detected); the rates were notably lower in the two insecticide treatment groups after spraying.
Cross-sectional malaria surveys
The prevalence of falciparum malaria was only around 1% in the pre-intervention survey, while that of vivax malaria stood around 5% ( Table 2). There was no significant difference between any treatment group pre-intervention, either for falciparum (Kruskal–Wallis, χ2 = 0.4, d.f. = 2, P = 0.81) or vivax malaria (χ2 = 2.5, d.f. = 2, P = 0.29). In the postintervention survey 3 months after spraying, falciparum malaria prevalence increased to 3.9% in the control group but remained < 1% in the SC and WP groups. Similarly, vivax malaria prevalence remained at 7.5% in the control group but fell to < 3% in the SC and WP groups. Both results were of borderline significance using Kruskal–Wallis (P = 0.06), but if the data are regrouped to compare both insecticide treatments (n = 6) against the control (n = 3), the result is significant for both falciparum (signed rank, P = 0.02) and vivax malaria (signed rank, P = 0.04). Odds ratios for treatment effects are shown in Table 2. The protective efficacy, 100*(1-OR)%, of the spray intervention indicates that falciparum malaria prevalence was reduced by 100% in the SC and by 86% in the WP groups, while vivax malaria prevalence was reduced by 77% in the SC and by 63% in the WP group. The protective efficacies for vivax malaria were lower than for falciparum malaria.
Table 2. Results of cross-sectional parasite surveys in schoolchildren before and after indoor residual spraying
Entomological indicators: mosquito densities
Six anopheline species were collected from the study area. Anopheles stephensi was by far the most abundant species; geometric mean density in space spray catches was 697 specimens per sector per month. Next was A. subpictus with 288 per sector per month, while A. culicifacies was only 137.
Seasonal abundance is shown in Figure 3. Few anophelines were caught in May and June (the rice crop was only planted in June). Culicines, however, were more abundant in May than in any other month and were presumably breeding in sites other than paddy fields. The July monsoon marked an increase in abundance of A. culicifacies, A. stephensi, A. annularis, and A. pulcherrimus. A. subpictus first appeared in October and A. nigerrimus in November. A. stephensi showed two peaks, first in July–August, then in November. The densities of all anophelines except A. pulcherrimus and A. nigerrimus were significantly lower in the sprayed sectors ( Table 3).
Table 3. Geometric mean density of mosquitoes per sector from space spray collections inside sentinel living rooms and animal sheds over the postintervention period from July to January. Six rooms were space sprayed in each sector every two weeks. Means followed by s were significantly different from untreated sectors
The parous rate of A. stephensi in untreated sectors increased from 62% during May–June to 80% during July–November, while in WP sectors the rate decreased from 81% to 24% and in SC sectors from 91% to 26% over the same period. The difference in parous rate between untreated and insecticide-treated sectors was highly significant (χ2 = 433, d.f. = 2, P < 0.0001). Similar trends were apparent in A. culicifacies, A. subpictus, A. annularis and A. pulcherrimus, indicating that the insecticide treatments were reducing longevity.
The trend in parous rate with time is shown in Figure 4. Parous rates in A. stephensi dropped to < 20% after spraying but then gradually increased over the next 5 months (χ2trend= 22, P < 0.0001), while in the unsprayed sectors the rate remained fairly steady at around 80% (χ2trend= 0.3, P = 0.59). These results indicate that insecticide residual activity was gradually deteriorating between July and December. The pattern was not so clear for other species because fewer specimens were caught.
Landing catches on water buffalo
The species caught during buffalo landing catches were the same as those caught in indoor resting (space-spraying) catches ( Table 4). The ratio of landing to indoor resting catches was 100 times greater for A. nigerrimus than for most other species, indicating marked exophilic behaviour. A. pulcherrimus was the most abundant species in landing catches both indoors and outdoors, and this may indicate a greater degree of exophily and zoophily than in other species.
Table 4. Results of indoor and outdoor cattle landing catches in untreated and alphacypermethrin-sprayed villages. The ratio of the number of mosquitoes caught from buffalo bait and from space spraying was calculated for each sector and mean ratios calculated.
More mosquitoes were caught in outdoor landing catches and in unsprayed sectors than in indoor landing catches and in sprayed sectors, with the exception of A. nigerrimus and A. pulcherrimus, the exophilic species ( Table 4). However, fewer A. pulcherrimus were found in space-spray collections in alphacypermethrin-treated sectors than in untreated sectors. These findings indicate a mix of indoor and outdoor resting for A. pulcherrimus and only partial control of this species by indoor residual spraying.
All surfaces treated with 25 mg/m2 alphacypermethrin (mud, brick, cement, wood, and thatch) killed 100% of A. culicifacies and A. stephensi for at least 112 days after treatment ( Figure 5). Even after 184 days, when cone bioassay tests were finished, all types of treated surface were still killing around 50%. There was no evidence that the decay rate on cement or mud was any faster than on wood or thatch.
Insecticide resistance tests
Resistance tests showed 100% mortality of laboratory-reared and wild-caught A. stephensi and A. culicifacies (N = 80). Hence there was no evidence for resistance in the study area.
Few side-effects were reported during or after spraying. In common with other alphacyano pyrethroid insecticides, transitory skin irritation and headache were reported by some spraymen and a few householders shortly after spraying. The effect wore off after a few hours. Since no persistent odour or residue was evident after spraying, and because both nuisance and vector mosquitoes were controlled, many people expressed appreciation for the spray campaign.
This is the first evaluation of indoor residual spraying with alphacypermethrin in a randomized controlled trial against malaria. Both WP and SC formulations were very effective against indoor resting anophelines, and reduced the incidence of falciparum malaria by 90% or more and of vivax malaria by 80%. It is conceivable that the incidence rates derived from ACD surveys would be subject to error or bias due to staff or householders' fatigue during repeated surveys. However, these estimates of protective efficacy were consistent with estimates obtained from cross-sectional surveys during the peak malaria season (occasional cross-sectional surveys on schoolchildren would seem less prone to staff error).
Alphacypermethrin appeared to have less effect against vivax malaria. This result is misleading because relapses of vivax malaria from pre-trial infections would occur at equal frequency in insecticide-treated and untreated sectors and could not be distinguished from new infections. The relapse rate is high during the spring in Pakistan, just before the monsoon, and coincides with a growth in anopheline populations ( Fox & Strickland 1989; Rowland et al. 1997b ). The phenomenon of relapse makes it more difficult to eliminate vivax malaria from subtropical areas by vector control compared to falciparum malaria, which does not relapse. Several years of indoor spraying campaigns in Afghan refugee camps of North-west Frontier Province have brought to an end the public health problem of falciparum malaria, but vivax malaria still persists ( Rowland 1999).
For the last 3 decades, malathion has been the insecticide of choice for malaria control in Pakistan ( Dezulueta et al. 1980 ; Shah et al. 1997 ). Recent field trials in North-western Pakistan have demonstrated that malathion can still reduce malaria incidence by up to 55% if campaigns are properly supervised ( Rowland et al. 1997b ). Pyrethroids are now being regularly used for malaria control in Pakistan ( Rowland et al. 1994 ; Shah et al. 1997 ), although there have been few evaluations and none against a control group. A comparative study of lambdacyhalothrin and malathion in Afghan refugee camps in Frontier Province showed that lambdacyhalothrin was twice as effective in controlling falciparum malaria but only equal to malathion in controlling vivax malaria ( Rowland et al. 1994 ). The results presented here indicate that alphacypermethrin is as good as lambdacyhalothrin in controlling falciparum and superior to lambdacyhalothrin in controlling vivax malaria. However, caution must be exercised when drawing conclusions from trials conducted in different provinces, as conditions might not have been the same. Where alphacypermethrin and lambdacyhalothrin have been compared side-by-side (in experimental hut studies in West Africa), they showed equal toxicity against free-living anophelines ( Darriet 1991). By contrast, bioassay tests indicated that alphacypermethrin was more persistent than lambdacyhalothrin on mud, wood, cement and thatch surfaces ( Dorta et al. 1993 ; Sulaiman et al. 1996 ). Alphacypermethrin's notable residual life was confirmed in bioassay tests in Pakistan: mortality began to decline only after 3 months. Under natural conditions, where free flying anophelines may remain in sprayed rooms for 2–3 days ( Mahmood & Reisen 1981), residues may be lethal for much longer. In India resting densities of Culex quinquefasciatus and Aedes subpictus were reduced for at least 27 weeks after alphacypermethrin-spraying of huts ( Amalraj et al. 1987 ).
The gradual increase in parous rate in the sprayed areas indicated a gradual deterioration of residual activity. But even by the end of the year the parous rates between sprayed and unsprayed sectors had still not fully converged. That sprayed surfaces remained toxic several months after spraying was indicated by the lower abundance of A. subpictus in sprayed areas. This species did not make a significant appearance until November, 5 months after the spray campaign, and yet was controlled even then. The only anopheline whose density was similar in sprayed and untreated areas was A. nigerrimus, an exophilic species ( Reisen & Milby 1986) that was much less likely to come into contact with treated surfaces.
A. culicifacies, the purported primary vector in the Punjab ( Reisen & Boreham 1982), had more or less disappeared by September, and yet falciparum malaria did not reach its peak until October. A. stephensi was five times more prevalent in the study area and still abundant in November. In North-west Frontier Province, A. stephensi is also more common than A. culicifacies, and circumsporozoite antigen positivity rates were found to be about the same in the two species ( Rowland et al. 1997b ). Sheikhupura may be an exceptional area of the Punjab, but circumstantial evidence points to A. stephensi being the more significant vector there.
Alphacypermethrin appears highly suited for the control of malaria in Pakistan and South Asia. It is highly toxic to vectors, requires only one treatment per year, and was appreciated by the community. The absence of odour means it is more acceptable to householders than standard organophosphorate insecticides. In international tenders it is price-competitive against its main rivals deltamethrin, lambdacy- halothrin and cyfluthin.
We thank the staff of the National Institute of Malaria Research and Training for their hard work and productivity, particularly Mr M. Sarwar and Mr M. Yasin. Mr Alamdar Shirazi of Cyanamid provided logistic support whenever required. The field study was funded by Cyanamid, with technical support from WHOPES. HealthNet International's Malaria Control Programme is supported by the European Commission (DG1) and the United Nations High Commissioner for Refugees. Research costs were partly covered by the Department for International Development of the United Kingdom and by HealthNet International.