Spraying houses in the Peruvian Andes with lambda-cyhalothrin protects residents against cutaneous leishmaniasis

TRANSACTIONS OF THE ROYAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE (2000)94,631-636

Spraying residents

houses in the Peruvian Andes with lambda-cyhalothrin against cutaneous leishmaniasis

protects

‘Department of C. R. Davies’*, E. A. Llanos-Cuentas*, P. Campos2, J. Monge2, E. Leon’ and J. Canales2 Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, Keppel Street, London WClE 7HT, UK; 21nstituto Medicina Tropical ‘Alexander von Humboldt’, Universidad Peruana Cayetano Heredia, Lima, Peru Abstract A household vector control trial was carried out in the Peruvian Andes to measure the effect of spraying inside walls and ceilings with lambda-cyhalothrin on the risk for residents of cutaneous leishmaniasis caused by Leishmania peruviana. The mortality rates of Lutzomyia verrucarum measured with WHO contact bioassay cones set on adobe walls characteristic of the endemic region indicated an LD95 for lambdacyhalothrin of about 20 mg/m’, and no reduction in effectiveness for at least 6 months on indoor adobe walls sprayed with 25 mg/m”. A random selection of 112 houses were sprayed (starting in 1992/93) at 6-monthly intervals with a mean dose of 34 mg/m’, leaving 154 control houses (with closely matched pre-intervention measurements of incidence and sandfly abundance). Comparisons of pre- and post-intervention sandfly indoor abundance, measured at regular intervals for up to 2 years using CDC light traps, in 22 sprayed and 2 1 control houses demonstrated that spraying significantly reduced the indoor abundance of Lu. verrucarum by an average of 78% and of Lu. (Helcocyrtomyia) peruensis by 83%. Spraying was also associated with a significant reduction of 77% in the proportion of bloodfed sandflies collected in light traps. The proportion of susceptible householders acquiring leishmaniasis during the trial was significantly reduced by 54% as a result of spraying. The observed impact of spraying was greatest, 8 1% (95% confidence intervals 20-95%), when the cases detected during the first 6 months after the intervention were excluded from the analysis, suggesting a significant pre-patent period. Keywords: zoonotic cutaneous leishmaniasis, verrucarum, lambda-cyhalothrin, Peru

Leikhmania peruviana, vector control, Lutzomyia pence&

Introduction Leishmaniasis control programmes in most of the 88 endemic countries are limited to patient diagnosis and treatment with pentavalent antimonial drugs (or amphotericin B as a second-line drug). Some countries additionally incorporate prevention strategies such as targeting of reservoir hosts [e.g., culling Leishmania infanturn-infected domestic dogs in Brazil (LACERDA, 1994)], or insecticide spraying of houses and outhouses [e.g., against Phlebotomus (Euphlebotomus) argentzpes in India (KAUL et al., 1994)]. Whilst the effectiveness of dog culling remains controversial (TESH, 1995), few doubts are raised about house-spraying programmes. Vector control for anthroponotic leishmaniases (cutaneous leishmaniasis, CL, caused by L. tropica and visceral leishmaniasis, VL, caused by L. donovam) should certainly be effective, as a reduction in sandfly-human contact will not only protect households but could also protect communities by interrupting the transmission cycle. In contrast, vector control for zoonotic leishmaniases (ZL) is unlikely to eliminate transmission either by a mass-killing effect or by reducing sandfly-reservoir host contact, but it could provide personal protection for householders if the sandfly vectors are sufficiently endophagic. Despite the paucity of trials measuring the effectiveness of spraying, vector-control programmes targeting ZL do operate in some countries where transmission is largely domestic. To our knowledge the only previous trial to quantify satisfactorily the epidemiological impact of house spraying against New World ZL was the comparison of VL incidence in DDT-sprayed and unsprayed villages in Brazil (ALENCAR, 1961). On this basis the Brazilian Ministry of Health continue to spray about 90 000 houses each year with pyrethroids (for residual action) or with organophosphates (for ultra-low-volume spraying) as part of their VL-control strategy (LACERDA, 1994). In the Old World, the only house-spraying trials that have measured the effects on ZL incidence appear to be for CL in Italy (CORRADETTI, 1952) and Iran (NADIM & AMINI, 1970), and there is some doubt whether the former is a zoonosis. None of the more *Author for correspondence; [email protected]

fax +44 (0)2 7323 5687; e-mail

Lutzomyia

recently reported house-spraying trials against ZL provides reliable data on their effectiveness, because (i) they measured only entomological effects (LE PONT et al., 1989; ALEXANDER et al., 1995), (ii) there were no controls (~GHALHAES et al., 1980; LIMA et al., 1988; BENZERROUG et al., 1992), (iii) there were insufflcient reulicates (FALCAO et al., 199 1). or (iv) the housespraying intervention was part of& imebated control programme (XV ZHI-BIAO, 1989). Even the data reported by ALENCAR (196 1) are not wholly reliable, as the control and sprayed villages differed significantly in preintervention incidence. In the Peruvian Andes, CL (uta) is a zoonosis largely caused by L. peruviana transmitted inside houses by Lu. (H..) peruensis, Lu. (H..) avacuchensis or Lu. verrucarum (DAVIES et al.; 199?a).‘T&s was the site in 1944 of the first trial of the effect of insecticides on sandfly populations (HERTIG & FAIRCHILD, 1948). Although the experimental design of these early studies in Peru was flawed (HERRER, 1956), the results indicate that spraying of houses and all surrounding buildings and walls should cause a significant reduction in indoor sandfly abundance. Additional circumstantial evidence that house spraying might be effective comes from the transient re&&tion in uta incidence that coincided with the DDT-snravine camuaigns against malaria and bartonellosis in-the 1560s @iVIEswet al., 1994). There is no national programme for house spraying against uta in Peru; and the effectiveness of local spraying campaigns sporadically organized by regional ministries (often when a sudden increase in the number of cases in a village generates public demand) has not been evaluated. Here we report the entomological and epidemiological impact of a household-based vector-control trial in the Peruvian Andes. Materials and Methods Study site and design The trial was carried out in villages and hamlets endemic for uta in the Departments-of Lima (Co110 and San lose: Canta Province‘l. Ancash (Colcao. San Lucas, cacpara and Chiwip: golognesi I!rovinckj and Piura (El Higueron and La Capia: Huancabamba Province). The habitat, CL epidemiology, clinical symptoms, vector ecology, and possible reservoir hosts in the sites were described previously (DAVIES et al., 1995b,

632

1997a, 1997b; LLANOS-CIJJZNTAS et al., 1999). Comprehensive active search of utu cases by house-to-house visits commenced in March, April and May 199 1 in the study villages in Lima, Ancash and Piura, respectively, and continued at about 3-monthly intervals until early 1996. The first round of house spraying was carried out in September and October 1992 and August 1993 in Lima, Ancash and Piura, respectively. Houses within each village were allocated randomly to the intervention group (sprayed) or the control group (unsprayed), except for (i) houses selected for sandfly sampling which were allocated on the basis of matchedpre-intervention sandflv data, and (ii) houses in the small hamlets which were allocated on the basis of matched pre-intervention epidemiological data. Insecticide spraying and bioassays The inside walls and ceilings (where present) ofhouses were sprayed with Hudson ‘X-pert’ compression spray pumps using lambda-cyhalothrin, supplied as iO% wettable nowder. at an intended deliverv rate of 25 ma/ m2. Coveiage rates were checked by calculating the rat& of insecticide used to the wall and ceiling area in each house. House spraying was repeated at 6-monthly intervals on a total of 4 occasions in Lima and Ancash and on 3 occasions in Piura. The effectiveness of lambda-cyhalothrin (and its persistence) against the local sandfly vectors under field conditions was checked bv WHO contact bioassav cones set against adobe bricks. In one set of bioassays (carried out in the grounds of the Universidad Peruana Cayetano Heredia in Lima with colonized sandflies), the effectiveness of different concentrations of insecticide (5, 10 and 20 mg/m2) was tested on outdoor adobe walls using bricks made in the Ancash field site. The second set of bioassays (carried out with field-caught sandtlies in villages within the studv vallevs of Ancash and Lima) comiared the effectiveness and persistence of 25 mg/mi lambda-cyhalothrin sprayed on indoor and outdoor adobe walls. In each assay, 5 cones containing 20 sandflies (both sexes) were exposed for 1 h against sprayed and unsprayed bricks, after which the recovered sandflies were maintained in uncontaminated plastic tubs for 24 h when mortality was recorded. Dosemortality relationships were calculated after adjusting for control (i.e., background) mortality rates. Collection and analysis of entomological data Sandflies were collected by Centers for Disease Control (CDC) light traps left overnight in the bedrooms of occupied houses, as previously described (DAVIES et al., 1997a). On each visit to a study village, CDC light traps were set up on 2 consecutive nights in each sampled house. All houses in the study were sampled for 2-20 nights pre-intervention (mean 10*6), and then at 1 week, 1 month, 3 month and 6 month time-points for 4 cycles (in Liia and Ancash) and for 3 cycles (in Piura). The 1 month time-point was dropped from the second and third Piura cycles for logistical reasons. Hence, generally each house in Lima and Ancash was sampled on 32 nights post-intervention, and each house in Piura on 20 nights. A total of 22 sprayed houses and 21 control houses were sampled in the project, as 1 sprayed sample house had to be changed mid-way through the survey. All sandflies were counted in the field, and then returned to Lima for identification to species and gender. When specimens were unidentified owing to loss or damage (which was rare), they were assigned species and gender according to the proportions in the village from where they had been collected. Bloodfed status was additionally recorded for sandfhes collected in the Department of Lima both pre- and post-intervention. For each sandflv suecies, the reduction in abundance due to spraying at each time-point was calculated by comparing the average ratio of post-intervention Williams-adjusted geometric mean (GM) abundance:pre-

C. R. DAVIES

ETAL.

intervention GM abundance for the sprayed houses with the equivalent average ratio for the control houses. Houses with zero abundance for a particular species in pre-intervention collections were excluded from the analyses. To demonstrate whether the intervention had a significant impact on sandfly abundance, the effect of spraying on these ‘post:pre ratios’ at each time-point (grid for the average of all time-points) was tested by analvses of deviance carried out in GLIM. suecifvine Poisson errors and scaling for overdispersion whei required (CRAWLEY, 1993). Interaction terms between treatment and either week or spraying cycle were incorporated in the maximal models to test whether the effects of spraying varied significantly with time. Collection and analysis of epidemiological data Durine the first visit to each household in 199 1. clinical aid demographic details were collected for each person by questionnaire and clinical inspection, as described- previously in more detail (DAVIES et al., 1995b. 1997a). and all consentine individuals (86% of the censused population) were giien a leishmahn skin test (LST) using the technique recommended by the World Health Organization. The date of onset of each clinical infection was recorded on the basis of patient (or parental) recall. Every household was revisited during each subsequent field trip to check for changes in clinical status or personnel. All new cases were diagnosed clinically by trained specialists and, with consent, confirmed by parasitological diagnosis and a further LST. An additional 55% of the population with no history of leishmaniasis received a second LST in 1992 nrior to the intervention to check for asymptomatic infections. All patients were provided with free treatment. Comparisons of pre-intervention infection rates were made on the basis of(i) the proportion of people with scars or lesions, (ii) the proportion of people with a positive LST response, and (iii) the number of new cases between January 1990 and the month of intervention. Recall of lesions prior to 1990 (i.e., more than 1 year prior to the first survey) was deemed insufficiently reliable. Comparisons of post-intervention infection rates are based on the number of new cases during the prospective survey amongst the population at risk, i.e., those with no prior history of infection or disease. The effect of spraying on incidence was tested in GLIM (CRAWLEY. 1993) bv an analvsis of deviance, specifying binomial errors, of the log-odds that unin~ fected people in a given household developed CL during the trial. In order to deal with the possibility of significant incubation periods between infection and disease, the analysis was repeated after excluding cases who were detected within specified time periods after the first intervention (O-6 months post-spray). Results Insecticide btiassays The bioassays carried out on colonized Lu. ven-ucarum 1 day after spraying outdoor adobe walls in Lima detected a significant dose response within the concentration range 5-20 mg/m* (Table 1, x2 for trend 24.6, P < O.OOl), with an LD50 of about 3.8 mg/m’ and an ID95 of about 20.1 mg/m2 (estimated by linear regression of the log dose-mortality relationships). There was a suggestion (but not significant) that females were more resistant to insecticide than males. The bioassays carried out on wild-caught sandflies (mostly Lu. venucarum) in endemic villages demonstrated that, with a concentration of 25 mg/m’, the insecticide remained 100% lethal for up to 6 months when sprayed on inside adobe walls; but lethality attributable to spraying dropped significantly with time when sprayed on outside adobe walls (Table 2), from 100% to 88% within 3 months (Fisher’s exact test, P < 0.01) and to 70% within 6 months (Fisher’s exact test, P < 0.001). On the basis of these results an intended indoor coverage

HOUSE SPRAYING FOR LEISHMANIASIS

633

CONTROL

of 25 mg/m’ was chosen for the intervention trial; and the actual mean coverage, calculated from the average data for each house, was 34.2 mg/m2 (median 33.3; quartiles 31.0, 37.2; range 25-50 mg/m”). Entomological impact of intervention The baseline (i.e., pre-intervention) GM sandfly abundances were 5.5 Lu. verrucarum and 0.27 Lu. (H.) peruensisitrap-night in the sprayed houses, and 5.1 Lu. verrucarum and 0.26 Lu. (H.) peruensisltrap-night in the control houses. Both species were collected in each study region, but Lu. (H.) peruensis was not detected in the preintervention collections from 6 ofthe 43 sampled houses. Lu. (H.) ayacuchensis was collected only in Piura, where its pre-intervention GM abundance was 0.69 and 0.871 trap-night in the sprayed and control houses, respectively. Lu. (H.) noguchii was collected only in the departments of Lima and Ancash, with a pre-intervention GM abundance of 0.13 and 0.1 O/trap-night in the sprayed and control houses, respectively (including 10 houses with zero pre-intervention collections, making statistical tests of spraying effects on post-:pre-intervention data impossible). The overall proportion of females was 0.84, 0.53, 0.71 and 0.47 for the 4 species, respectively. The remaining 8 species comprised < 1.7% of the total catch (DAVIES et al., 1997a), and are not referred to again. The relationship between the time since the first intervention and the mean ratio of post-intervention: pre-intervention abundance for Lu. verrucarum and Lu. (H.) peruensis is illustrated in the Figure. For all 4 species, the mean abundance pre- and post-intervention are presented in Table 3, and the mean reduction in abundance attributable to spraying in Table 4. For Lu. verrucarum, a significant effect of spraying on the ratio of post-intervention:pre-intervention abundance was identified at 15 out of the 16 time-points, with the only exception being the final time-point of the first spraying cycle when the reduction observed (44%) was not quite significant (P = 0.06); the overall effect was highly significant (Table 4). As for the other species tested below, no significant interactions were detected between the effect of treatment and either week number or cycle number. So there was no evidence of any diminishing of Table 1. Relationship between concentration of lambda-cyhalothrin sprayed on outdoor adobe walk and 24-h mortality of colonized Lu. verrucururn following l-h contact on walls 1 day postspraying Mortality, Dose (mg/m”) 0 5 1:

Table 2. Relationship adobe walls sprayed spraying

% (number dead/total) Males

Females 0 54.1 62.5 84.2

13.7 74.7 80.6 96.0

(0139) (20137) (20/32) (32/38)

(14/102) (56/75) (79/98) (971101)

64

0

25

75 50 Weeks post-spray

100

(b)

n 0

25

50 75 Weeks post-spray

100

Figure. The relationship between the number ofweeks after the first intervention and the geometric mean (GM) of the ratio of post-intervention sandfly abundance (GM of light-trap collections on 2 consecutive nights) to pre-intervention sandfly abundance (GM of light-trap collections on an average of 11 nights) for control houses (dotted line: n = 21) and sprayed houses (solid line: n = 22). Houses were sprayed at 6-monthly intervals after the first intervention. (a) Lu. vert-ucamm, (b) Lu.

peruensis.

effectiveness with time since the last spray, or any increase in effect with consecutive spraying cycles. For Lu. (H.) peruensis, although a significant effect of spraying was detected at only 4 of the time-points, the effect of spraying on the mean abundance post-intervention was significant (Table 4). Whilst the mean abundance of both Lu. (H.) ayacuchensis and Lu. (H.) noguchii dropped in sprayed houses, no significant effects could be detected. A highly significant reduction in the proportion of sandflies with bloodmeals was observed as a result of spraying in Lima. Before the intervention, the proportion of Lu. vewucarum with blood was not significantly different between ‘sprayed’ houses, 14.3% (241/1688), and control houses, 14.1% (139/984): Yates’ corrected

between 24-h mortality of field-caught Lu. verrucarum, following l-h contact on (either indoors or outdoors) with 25 mg/m* lambda-cyhalothrin, versus time since San Jose (Lima)

Time post-spray

Sprayed I

1 week

100 (91/91)

1 month

100 (82/82)

3 month 6 month

100 (51/51) 100 (83/83)

Values given are percent mortality I, indoors; 0, outdoors.

Yumpe (Ancash)

Control 15.4 52.8 32.0 14.0

(10/65) (38172) (24175) (14/100)

at 24 h (number dead/total).

Sprayed I 100 100 100 100

(90/90) (96/96) (97197) (69169)

Sprayed 0 100 100 90.1 73.1

(80/80) (93/93) (73/81) (49/67)

Control 11.2 8.2 19.4 11.4

(10189) (8/97) (18/93) (8/70)

634

C. R. DAVIES

Table 3. Williams modified geometric mean abundance of four sandfly sprayed and control houses, either before or after the intervention

species (ICDC light trap-night)

Sprayed houses Species LV LP LA LN

After

(25-l 1.2) (0.10-0.37) (0.30-1.19) (0.01-0.26)

1.19 0.095 0.22 0.022

in

Control houses

Before 5.53 0.27 0.69 0.13

ETAL.

Before

(058-2.0) (0.01-0.18) (0.04-0.44) (0~01-0~04)

5.07 0.26 0.87 0.10

After

(2.2-10.6) (0*13-0.40) (0.55-1.25) (0.03-0.18)

3.84 0.31 0.69 0.065

(1.5-8.3) (0.07-0.60) (0.42- 1.00) (0.03-0.10)

Data for LA exclude Ancash and Lima, where this species was absent, and data for L.N exclude Piura where this species was absent. Hence, number of sprayed houses sampled is 22 for Ll’and LP, 10 for LA, 12 for LN; and number of control houses sampled is 2 1 for LVand LP, 10 for LA, 11 for LN. Values given are numbers of sandfliesltrap-night (95% confidence interval). LV, Lu. verrucarum; LP, Lu. peruensis,LA, Lu. ayacuchensis,LN, Lu. noguchii.

Table 4. Reduction in sandfly abundance attributable to house spraying with lambda-cyhalothrin Species” ;r E

% Reduction in abundance (95% CI)

Significanceb

78.2 (34.8 - 92.7) 83.5 (28.9 - 96.1) 71.3 (-16.2 - 92.9) -

z = 2.72, P = 0.006 z = 2.42, I’= 0.016 z = 1.75, I’= 0.08 NS

“Species abbreviations as in Table 3. bDetails given in text.

x2 = 0.00, P = 0.96; but after the intervention the proportion bloodfed in sprayed houses, 2.1% (1 l/53 l), was significantly lower than that in the control houses, 8.9% (54/604): relative risk (95% CI) = 0.23 (0.120.44), Yates’ corrected x2 = 23.4, P < 0.001. Although mortality rates (i.e., the proportion found dead in the traps when collected in the morning) were not recorded in the control houses, a large proportion typically survived overnight in the traps, whereas all sandtlies collected in light traps set up in sprayed houses were dead by the morning. Epidemiological impact of intervention A total of 112 houses were sprayed (Piura 68, Lima 33, Ancash 11) leaving 154 control houses (88, 43 and 23, respectively). Prior to the interventions, the percentage of the population with scars or lesions was the same, 65%, in the sprayed (419/644) and control houses (496/ 765); and the percentage with a positive skin test was marginally higher in the sprayed houses, 72% (413/573), than in the control houses, 69% (543/645). Finally, amongst those people censused at the time of the intervention, the number of people with a history of CL between January 1990 and the date of the intervention was98(i.e., 14*8%)and105(13*7%)inthesprayedand control houses, respectively. Hence, the pre-intervention risk of infection in the control and sprayed houses was well matched. Table 5. Epidemiological

impact

At the start of the interventions, a total of 170 households (75 sprayed and 95 control) included at least 1 previously uninfected (i.e., at risk) household member. The total population at risk in these houses was 437, of which 339 were unscarred and LST negative and 98 were unscarred but with no LST result. Twenty-five new cases were detected during the trial amongst the former (representing 7.4%) and 8 amongst the latter (8.2%). The 61 unscarred but LST-positive people in the 170 houses were excluded Tom the analyses, as they were presumed to have prior subclinical (potentially protective) infections, and only 1 new case was detected in this group during the trial (1.6%). Amongst the 24 1 people at risk in control houses, 24 (10~0%) developed CL, compared to only 9 cases amongst the 196 at risk in the sprayed houses (4.6%) (Table 5). The incidence of new CL cases amongst previously uninfected householder members during the complete trial was significantly reduced by 53.9% (95% CI 4-O-76*6%) in sprayed houses as compared to unsprayed houses (z = 2.063; P = 0.039). However, the effect of spraying was most significant when cases detected during the first 3 months post-spraying were excluded from the analysis (z = 2.331; P = 0.020), the rationale being that some of the first cases might reflect infection events that happened prior to the intervention. After excluding cases from the first 3 months, spraying apparently reduced incidence by 64.8% (95% CI 15.5-84.2%). The reduction in incidence attributable to spraying was even greater, 81.3% (95% CI 19.7-95-O%), when all the data from the first spraying cycle (i.e., the first 6 months) were excluded from tbe analysis (z = 2.249; P = 0.024), suggesting a possible cumulative effect from re-spraying.

Discussion T’he results demonstrate that in the Peruvian Andes spraying the inside walls and ceilings of individual adobe houses with lambda-cyhalothrin can lead to a significant reduction in the indoor abundance of sandfly vectors of leishmaniasis, in the proportion of sandflies caught in indoor light traps with bloodmeals, and in the risk of CL for householders. Anecdotal evidence indicated that

of house spraying Post-intervention

Pre-intervention

Number of incident cases’ Households Control Sprayed

%cL+

%LST+

No. susceptibles (no. households)

65% 65%

69% 72%

TE :;z;

All

>3m

>6m

24 9

21 6

13 2

%CL+, percentage with a cutaneous scar or lesion; %LST+, percentage with a positive skin test response (denominators given in text). “The number of casespost-intervention following a lag period of 0,3 or 6 months.

HOUSESPRAYINGFORLEISHMANIASISCONTROL house spraying also caused significant mortality to those sandflies entering the houses. The intervention was carried out at the household level, rather than the community leve!, so no effect on sandfly population size or adult survwal rate was expected. The apparent effect on sandfly mortality inside sprayed houses is unlikely to impact on the population of sandflies in a village, as only a small proportion of the population would come into contact with sprayed houses. This explains why the indoor abundance in unsprayed houses was relatively unaffected by the spraying of neighbouring houses. Hence, the reduction in the observed abundance inside sprayed houses is presumably because (i) the insecticide had a repellency effect making sandflies less endophagic, (ii) the insecticide affected the behaviour of sandflies after they entered houses, reducing their likelihood of entering a light trap before leaving the house, or (iii) the insecticide killed or inactivated a significant proportion of sandflies before they had a chance to enter a light trap (either through landing on the sprayed walls or possibly through ‘aerial’ insecticidal contact mediated by contaminated dust). Whatever the explanation, the reduction in the size of light-trap collections must reflect a reduction in the biting rate on humans in sprayed houses, as the ratio of human bait:light-trap collections of sandflies in this field site was the same in sprayed and unsprayed houses (DAVIES et al., 1995a). In this trial, lambda-cyhalothrin spraying of inside walls was repeated every 6 months. No significant drop in insecticide effectiveness over this time period was detected either from the bioassayscarried out in the field, or from the sandfly-abundance data collected in the trial. This result contrasts with the reportedly short duration of of the residual effect against Lu. (Lutzomyia) h&a&is cypermethrin sprayed on indoor adobe walls (PASSARAT DE SILANSet al., 1998); whilst reports of the duration of the residual effect of deltamethrin against sandflies are inconsistent (LE PONT et al., 1989; FALCAOet al., 199 1; MARCONDES& NASCIMENTO, 1993; ALEXANDERet al., 1995). Bioassays on outdoor adobe walls in our trial indicated that lambda-cyhalothrin effectiveness also dropped significantly within 3 months of outdoor spraying; a similar rate of reduction in effectiveness was detected by bioassays with colonized Lu. verrucarum on outdoor adobe walli sprayed in Lima (data not shown). The difference in the residual effect on indoor and outdoor walls possibly reflects the impact of direct sunlight or of wind, but cannot be due to rain as the reduction was detected during the dry season. The limited data collected indicate that the dosemortality relationship for the residual effect of lambdacyhalotl&n on Lu. v&-ucamm is consistent with previous findings for sandflies (SCORZAet al.. 1995: KELLY et al.. 1997;-MAZZARRIetal:, 1997). Bioassayswere carried out exclusively on Lu. verrucarum, but the sandfly abundance data from the trial found no evidence that effectiveness differed between the 4 species analysed. Differences in the P values in Table 4 essentially reflect variation in the pre-intervention densities (i.e., it is easier to detect a significant reduction for those speciesthat are the most abundant). Although the number of new casesrecorded during the trial was small, the results demonstrate that house spraying at 6-monthly intervals can reduce the risk of CL by up to 8 1%. This reduction is solely on the basis of household protection, and it is possible that the population effect of mass spraying of villages would enhance the protective effect further. Irrespective of any hypothetical masseffect, the results indicate that householders at risk of uta should experience personal protection by regularly spraying their houses with insecticide. Whether the protection provided by blanket house spraying is sufficient to justify the costs is ultimately an economic decision. Blanket spraying of endemic regions is never likely to be a sustainable strategy, but targeted spraying of

635

villages at greatest risk may be feasible if the decisions are based on accurate knowledge of the spatial distribution of incidence. This information is rarely available in any endemic country for leishmaniasis owing to the paucity of monitoring activity for either sandfly vectors or disease incidence. In practice, where spraying campaigns are operational the decisions where and when to spray tend to be based on the number of cases notified to the Ministry of Health, an unreliable indicator of risk. For example, in the Department of Sfio Paulo, Brazil, spraying against the domestic transmission of zoonotic CL caused by L. bruziliensis is activated by the reporting of 32 casesin a municipality during the sameyear (GOMES & NEWS, 1998). The challenge for the future is to combine empirical measurements of effectiveness, obtained from intervention trials such as reported here, with a comprehensive calculation of the costs entailed by spraying. The derived measurements of cost-effectiveness should then be integrated with a reliable risk map for disease in order to develop an evidence-based strategy for targeting control activities. Comparisons of cost-effectiveness between house spraying and either insecticide-impregnated curtains or bednets should be a priority for those leishmaniases, such as uta, largely transmitted to humans by endophagic sandflies. Acknowledgements

This project receivedfunding from the WHO Special Programme for Research and Training in Tropical Diseases, the International Development Research Centre, and the Wellcome Trust. The insecticide was donated by Zeneca plc, UK. References Alencar, J. E. (1961). Profilaxia do calazar no Cear& Brasil. Rev&a do Institute de Medicina Tropical de Xio Paulo, 3, 175-180. Alexander, B., Jaramillo, C., Usma, M. C., Quesada, B. L., Cadena, H., Roa, W. & Travi, B. L. (1995). An attemut to control ~phl;boto&ne sandfli& (Dipiera: l%ychodidaej by residual spraying with deltamethrin in a Colombian village. Memdrias do Instituw Oswald0 Cruz, 90,42 l-424. Benzerroug, E. H., Benhabylles, N., Izri, M. A. & Belahcene, E. K. (1992). Les pulverisations intra- et peri-domiciliares de DDT dans la lutte contre la leishmaniose cum&e zoonotique en Algkie. Annales de Ia Soci& Beige de Midecine Tropicale, 72, 5-12. Corradetti, A. (1952). The epidemiology and control of oriental sore in Abruzzo, Italy. American Journal of Tropical Medicine and Hygiene, 1,618-622. Crawley, M. J. (1993). GLIM for Ewlogaists. Oxford: Blackwell Scientific Publications. Davies, C. R., Llanos-Cuentas? A., Canales, J., Leon, E., Alvarez, E., Monje, J., Tolentmo, E., Gomero, Q., Pyke, S. & Dye, C. (1994). The fail and rise of Andean cutaneous leishmaniasis: transient impact ofthe DDT campaign in Peru. Transactions of the Royal Society of Tmpical Medicine and Hygiene, S&389-393. Davies, C. R., Lane, R. R., Villaseca, P., Pyke, S. D. M., Campos, P. & Llanos-Cuentas, E. A. (1995a). The relationship between CDC light-trap and human-bait catches of endophagic sandflies (Diptera: Psychodidae) in the Peruvian Andes. Medical and Veterinay Entomology, 9,241-248. Davies, C. R., Llanos-Cuentas, E. A., Pyke, S. D. M. & Dye, C. ( 1995b). Cutaneous leishmaniasis in the Peruvian Andes: an epidemiological study of infection and immunity. Epia’emiologyandInfection, 114,297-318. Davies, C. R., Llanos-Cuentas, E. A., Campos, P., Monje, J., Villaseca, P. & Dye, C. (1997a). Cutaneous leishmaniasis in the Peruvian Andes: risk factors identified &om a village cohort study. American Journal of Tropical Medicine $2 Hygiene, 56,85-95. Davies, C. R., Llanos-Cuentas, E. A., Sharp, S. J., Canales, J.,

Leon, E., Alvarez, E., Roncal, N. & Dye, C. (1997b).

Cutaneous leishmaniasis in the Peruvian Andes: factors associated with variability in clinical symptoms, response to treatment, and parasite isolation rate. Clinical Infectious Diseases, 25, 302-310. FalcBo, A. R., Pinto, C. T. & Ferreira Gontijo, C. M. (1988). Susceptibility of Lutzomyia longtipalpiis to deltamethrin. Memdrias do Instituw Oswald0 Cruz. 83.395-396. FalcBo, A. L., FalcHo, A. R., Pinto, C. T., Ferreira Gontijo, C.

636

C. R. DAVIES ETAL.

M. &Falqueto, A. (1991). Effect of deltamethrin spraying on the sandfly populations in a focus of American cutaneous leeqhmaniasis. Merndrias do Institute Oswaldo Cruz, 86, 399Gomes, A. C. & Neves, V. L. F. C. (1998). Estrategia e perspectivas de controle da leishmaniose tegumentar no Estado de SBo Paula. Revista do Instituw L Medic&a Tropical de LG.7Path, 31,553-558. Herrer, A. (1956). Phlebowmus y DDT en el Peru. Rev&a de Medicina Experimental, 10,99- 137. Hertig, M. & Fairchild, G. B. (1948). The control of Phlebowmus in Peru with DDT. AmericanJournal of TropicalMedicine, 28,207-230. Kaul, S. M., Sharma, R. S., Dey, K. P., Rai, R. N. & Verghese, T. (1994). Impact of DDT indoor residual spraying on PhZebotomusargenttpes in a kala-azar endemic village in eastern Uttar Pradesh. Btdletin of the World Health Organization, 12, 79-81. Kelly, D. W., Mustafa, Z. St Dye, C. (1997). Differential application of lambdacyhalothrin to control the sandfly Lutzomyia longipalpis. Medical and Veterina y Entomology, 11, 13-24. Lacerda, M. M. (1994). The Brazilian leishmaniasis control program. Memdrias do Institute OswaMo Cruz, 89,489-495. Le Pont, F., Mariscal Padilla, J., Desjeux, I’., Richard, A. & Mouchet, J. (1989). Impact de pulverisations de deltamethrine dans un foyer de leishmaniose de Bolivie. Annales de la So&% Beige a%Mtdecine Tropicale, 69,223-232. Lima L. C., de Almeida Marzochi, M. C., Sobroza, P. C. & de Souza, M. A. (1988). Observacdes sobre a leishmaniose tegumentar, cinco anos apes profilaxia. Revista de Saline Btiblica, 22,73-77. Llanos-Cuentas, E. A., Roncal, N., Villaseca, I’., Paz, L., Ogusuku, E., Perez, J. E., Caceres, A. & Davies, C. R. (1999). Natural infections of Leishmaniaperuviana in animals in the Peruvian Andes. Transactions of the Royal Society of Tropical Medicine and Hygiene, 93, 15-20.

family and w-k commitments. E3utit is convenient for me to p!an my study avund my work. I study tw hours on working days and more at the weekends. aq Dr Taysir K. Asi MScInfecriousDiseases Kuwait

MaghalhBes, P. A., Mayrink, W., da Costa, C. A., Melo, M. N., Dias, M., Batista, S. M., Michalick, M. S. M. & Williams, P. (1980). Calazar na zona do Rio Dote-Minais Gerais. Resultados de medidas profilaticas. Revista do Instituw de Medicina Tropical de St50 Paulo, 22, 197-202. Marcondes, C. B. & Nascimento, J. A. (1993). AvaliacHo da eficienca de deltametrina (K-othrine CE) no controle de Lutzomyia Zongipalpi~ (Diptera: Psychodidae), no Municipio de Santa Rita, Paraiba, Brasil. Revista da Sociedade Brasileira de Medicina Tropical, 26, 15- 18. Mazzarri, M. B., Feliciangeli, M. D., Maroli, M., Hemandez, A. & Bravo, A. (1997). Susceptibility of Lutzomyia longspalpis (Diptera: Psychodidae) to selected insecticides in an endemic focus of visceral leishmaniasis in Venezuela. Journal of the American Mosquito ControlAssociation, 13, 335-341. Nadim, A. & Amini, H. (1970). The effect of antimalaria spraying on the transmission of zoonotic cutaneous leishmaniasis. Tropical and Geographical Medicine, 22,479-481. Passarat de Silans, L. N. M., Dedet, J. P. & Arias, J. R. (1998). Field monitoring of cypermethrin residual effect on the mortality rates of the phlebotomine sand fly Lutzomyia Zongipalpis in the state of Paraiba, Brazil. Memorias do Instituto Oswalab Cruz, 93,339-344. Scorza, J. V., Rosario, C. L., Scorza, J. V. Jr & Rojas, E. (1995). Susceptibilidad de hembras silvesnes de Lutzomyia youngi de Trujillo, Venezuela, a insecticidas sinteticos. Boletin de la Direccidn a’e Malariologia y Saneamiento Ambiental, 35, 3 ll326. Tesh, R. B. (1995). Control of zoonotic visceral leishmaniasis: is it time to change strategies? American Journal of Tropical Medicine and Hygiene, S2,287-292. Xu Zhi-Biao (1989). Present situation of visceral leishmaniasis in China. Parasitology Today, 5, 224-228. Received 7 February 2000; reviked 30 May forpublication 1 June 2000

was exited at the possibi/#y to study at a distance from the famous London UniversiYySchool of hygiene and Topical Medicine.. I erjoy very much reading a~/ ardyzhg mimples in my Study Units, because I understand and anticipate implications of hsaith re!ated social and exwmic events. * Tokzhm Akhaeva MScHealth SystemsManagement Kazokkstan

the

2000; accepted

&l have chosen to .stu& by distance lezrnirg because I do not want to leave myjob and twme. t$ virtue of this prcgramme I intend to I.. the uses of epidemiology in a practical seltirg and to become famil&r with essential statist& acd fesearch methcdolog)* Dr Assad Hafeez MScEpidemiology:Principlesand Practice Pakistan

TropicalMedicineacademics.LSHTM For informationaboutLSHTMdistancelearningprogram TheInformationCentre(OO/RSTMflZ) Tel:t44 (0)ZO78628360/8361/8362, E-mail:[email protected] http~/~.lon.ac.uk/extemal