- Email: [email protected]
Do untreated bednets protect against malaria?
TRANSACTIONS
OF THE ROYAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE (200 1) 95457-462
Do untreated bednets protect against malaria? Margaret PinderZ, Gijs E. L. Walraven’ and Sign E. Clarke’*, Claus Begh’, Richard C. Brown’, Steve W. Lindsay’p3 ‘Danish Bilharziasis Laboratory, Jegersborg Alli 10, DK-2920 Charlottenlund, Denmark; ‘Medical Research Council Laboratories, I? 0. Box 273, Banjul, The Gambia; 3School of Biological and Biomedical Sciences, University of Durham, South Road, Durham DHl3LE, UK Abstract Bednets are thought to offer little, if any, protection against malaria, unless treated with insecticide. There is also concern that the use of untreated nets will cause people sleeping without nets to receive more mosquito bites, and thus increase the malaria risk for other community members. Regular retreatment of nets is therefore viewed as critical for malaria control. However, despite good uptake of nets, many control programmes in Africa have reported low re-treatment rates. We investigated whether untreated bednets had any protective benefit (in October and November 1996) in The Gambia where nets, although widely used, are mostly untreated. Cross-sectional prevalence surveys were carried out in 48 villages and the risk of malaria parasitaemia was compared in young children sleeping with or without nets. Use of an untreated bednet in good condition was associated with a significantly lower prevalence of Plasmodium falciparum infection (51% protection [95% CI 34-64%], P < 0.001). This finding was only partly explained by differences in wealth between households, and children in the poorest households benefited most from sleeping under an untreated net (62% protection [14-83%], P = 0.018). There was no evidence that mosquitoes were diverted to feed on children sleeping without nets. These findings suggest that an untreated net, provided it is in relatively good condition, can protect against malaria. Control programmes should target the poorest households as they may have the most to gain from using nets. Keywords: malaria,
Plasmodium
fulciparum, disease control, vector control, bednets, children, poverty, The Gambia
Introduction Treating mosquito nets and other fabrics with insecticides makes them an effective method of reducing malaria attacks and deaths (&ONSO et al., 1993; LENGELER, 1998). Whilst malaria control programmes in Africa have reported good uptake or purchase of nets, very few people, for a variety of reasons, re-treat their nets (WINCH et al. 1997; SNOW et al., 1999). For example, despite a long history of net use in The Gambia, and a national impregnation programme (NIBP) since 1992, fewer than 15% of nets are regularly retreated (CHAM et al., 1997). In practice, this means that large numbers of people are sleeping under untreated nets. It is thus important to know whether these people are protected against malaria. An intact net, if used correctly, can provide an effective barrier to mosquitoes (PORT & BOREHAM, 1982; LINES et al., 1987; LINDSAY et al., 1989; BURKOT et al, 1990), but whether the reduction in biting is sufficient to reduce malaria infection and morbidity rates in the human population is not clear. No study has convincingly demonstrated a protective effect for bednets that have not been impregnated with insecticide. Only 3 randomized intervention studies have been carried out. In community-based trials, children in villages using untreated nets had 38% fewer clinical attacks in The Gambia (SNOW et al., 1988), and 33% fewer in Tanzania (-WELL et al., 1999). Neither result, however, was statistically significant. In only one trial has a sirrnificant difference in the incidence of malaria attack; been seen (NEVILL et aZ., 1988), but the possibility that this difference was due to increased infection among the control children, who slept in the same room as those using nets, cannot be ruled out. Now that the enhanced impact of impregnated nets has been demonstrated, further randomized trials of untreated nets are unlikely, and the question of whether untreated nets can protect against malaria may never be definitivelv resolved. Evidence for a orotective effect of untreated nets will remain poor, be&g based largely on anecdotal accounts and cross-sectional data, in which users of untreated nets were self-selected (BRADLEY et al., 1986; CAMPBELL et al., 1987; LINDSAY & for correspondence; fax +45 77 32 77 33, e-mail [email protected]
*Author
1988; GENTON et aZ., 1994; D’ALESSANDRO et aZ., 1995). In these studies, confounding by other factors, such as wealth, could have been responsible for the lower prevalence observed among those that used nets compared to those that did not. It is also possible that the risk of transmission in non-net users was exacerbated because mosquitoes were diverted from users of untreated nets to feed on non-net users in the same house or village. In other words, the observed difference could have resulted from an increase of malaria among non-users rather than from a decrease among net users. Previous analyses lacked data on confounders, such as socio-economic status, and other measures used to reduce mosquito biting. Data from a cross-sectional study carried out in The Gambia included this information, and thus provided an opportunity to take account of the most likely alternative explanations when we compared the risk of malaria infection in children sleeping with or without nets, and thereby re-evaluate whether untreated nets can protect those who sleep under them from malaria. GIBSON,
Materials and Methods Study area and population Data were available from 48 villages in The Gambia, in a rural area with low-to-moderate malaria transmission concentrated in a single seasonal peak each year (LINDSAY et al., 1993). The study area was located about 200 km inland, and covered about 2500 km2 (between UTM 410 OOOE and 480 OOOE) around the towns of Farafenni, Soma and Kaur. The population, described in more detail by AIKINS et al. (1993), comprises predominantly subsistence farmers of the ethnic groups Mandinka, Wollof and Fula. Villages are discrete units, typically consisting of a cluster of family compounds, composed of closely related households. Health services within the study area include: 4 government health centres and private pharmacies located in towns; government disp&sarie< in 10 villages; and urimarv health care PHC) workers in the larger viliages. Outreach clinics also’operate once a mon& in 26 of the PHC villages, and are equally well attended by villagers from both PHC and non-PHC villages (HILL et al., 2000). A geographically stratified random sample (not proportional to size) was drawn from all villages
458 within the study area. The selected villages comprised 11 PHC and 37 non-PHC villages. Parasitological and entomological surveys Cross-sectional prevalence surveys of children aged 6 months-5 years were carried out during the peak transmission period in October and November 1996. In each village, a cluster sample of up to 30 children was selected, from compounds located on the periphery of the village closest to the alluvial floodplains where mosquitoes breed (B@GH et al., 1999a; THOMAS & LINDSAY, 2000). Children were examined, standing, for spleen enlargement. Fingerprick blood samples were obtained for blood films, and haemoglobin was measured using a Hemocue”’ haemoglobinometer. Thick films were examined at X 1000 magnification, and the number of parasites per high-powered field (HPF) was recorded. Parasite density per & was calculated assuming a blood volume per HPF of 0.002 ~.I.L(GREENWOOD & ARMSTRONG,-199 1). Children with 25000 uarasites/uL were classified as having high-density para&taemia.‘A film was considered negative if no parasites had been seen after 100 HPF had been examined. Plasmodium species were identified from thin films. Axillary temperature was measured with an electronic thermometer. To take account of diurnal variation in temperature (AMSTRONG SCHELLENBERG et al., 1994), surveys were usually carried out between 08:OO and lO:OO, and a fractional polynomial function used to fit a flexible curve to the iarcation in observed temneratures. usine STATATM. Fever was defined as a temperature grea?er than the 95th percentile value for the time of measurement (equivalent to 237~5°C at 09:15 and >38*O”C at 13:OO). Parasitepositive children with a raised temperature, but with symptoms suggesting another cause for fever, were not regarded as malaria cases. Mosquitoes were collected every second week throughout the transmission season (June-December), using light-traps. In each village, collections were made in 2 of the compounds selected for parasitological survey (B~GH et al., 1999b). For each village, the total number of infective bites per person in the transmission season (EIR) was estimated from the sum of the monthly EIRs, calculated as the geometric mean number of mosquitoes multiplied by the mean proportion of infective mosquitoes and number of days in the month. Household surveys and interviews Information on bednet usage and insecticide treatment was collected. The presence of a bednet was validated by observation, and the condition of the net recorded. A net was considered to be in good condition if it was long enough to be tucked under the mattress, was not torn or otherwise badly damaged, and had no more than 5 small holes (finger-width, approximate diameter ~2 cm). All other nets were classified as being in poor condition. All nets were tested for permethrin (M~~LLER et al., 1994), the insecticide used by the NIBP in The Gambia. Women responsible for the care and upbringing of at least one child under 5 years were interviewed, using a predominantly open-ended questionnaire, to obtain detailed socio-economic data, and information on knowledge and practices in relation to malaria. Up to 21 interviews were carried out in each village. The lottery method was used to initially select 21 compounds at random, followed by random selection of one woman per compound. A household was defined as the family sub-unit within a compound, comprising the informant, her husband and co-wives, and their children. Families were categorized into socio-economic groups based on 9 household features, which included ownership of household commodities, livestock, and house structure (S. E. Clarke et al., in preparation). Families possessing <4 of these features were categorized as
SIANE.CLARKE
ETAL.
‘poorer’ (24% of women interviewed), whilst families which possessed 4-5 of these features were regarded as ‘average’ (46%), and those with 26 as ‘wealthier’ (30%). In some cases the women interviewed cared for children who had been selected for inclusion in the parasitological survey, and we were able to relate the household practices and socio-economic differences in relation to bednet use to the parasitological findings. Statistical methods Data were analysed in EpiInfo (version 6.01b) and SPSS (version 9.0). Percent protection [with 95% confidence intervals (CI)] was obtained from 1 - odds ratio, and its confidence intervals. Categorical data were tested using the x2 test, and correlation and linear regression analysis used to test continuous data. Ethical considerations Ethical approval was obtained from the joint Gambian Government-Medical Research Council Ethics Committee. Informed consent was obtained from village leaders, and parents or guardians of the children, and from the women interviewed. All inhabitants of the study villages who requested medical attention were examined, and given treatment. Results A total of 1196 children were examined, of whom 468 (39%) were infected with I? falciparum parasites, including 7 children with dual infections with I? malariae. Children in whom only I? malariae (9) or l? ovale (3) were detected were excluded from analysis. A further 18 children were also excluded, owing to conflicting or insufficient information on bednet use. Among the 1166 children included in the analysis, 914 (78%) slept under a bednet, although many nets (388) were in a poor condition. Insecticide was detected on few nets (6.3%) by the end of the rains, when transmission was at its peak. Less than half of the children surveyed (48%) slept under a net in good condition or one treated with insecticide. Bednet use did not vary with age (x” = 2.43, d.f.4, P = 0.66). Neither did the condition of the net change with age, amongst those children sleeping under untreated nets (x’ = 7.28, d.f.4, P = 0.12). Net use, and condition, did not differ between PHC and non-PHC villages (x2 = 4.50, d.f.3, P = 0.21). Individual effect of bednets Less malaria was seen among children sleeping under an untreated net in good condition compared to those without a net (Table 1). Significant differences were observed in the prevalence gf I? falciparum parasitaemia (x2 = 20.78, d.f.1, P < O.OOl), high-density parasitaemia (x” = 14.51, d.f.1, P < 0.001) and severe-tomoderate anaemia (x2 = 9.84, d.f.1, P = 0.002). A similar pattern was seen with respect to clinical attacks, although few clinical cases of malaria were recorded. Use of an insecticide-impregnated net, irrespective of its condition, was associaiedwith the lowest prevalence of parasitaemia and 69% protection [95% CI 42-84%] (data not shown, x2 = 16.21, d.f.1, P < 0.001). Effect of bednet coverage in villages Both the mosquito density and the total estimated malaria transmission risk during the 5-month transmission season differed considerably between the 48 study villages, from
UNTREATED
Table
459
BEDNETS AND MALARIA
1. Effect of untreated
bednets
on malaria
in children
aged 6 months-5
1996)
Untreated net in good condition” (n = 481)
No net (n = 252)
Number of cases (%)
Protection (95% CI)
128 (50.8%) 65 (25.8%)
161 (335%) 69 (14.3%)
5 1% (34-64%) 52% (30-67%)
15 (6.0%) 11 (4.4%)
16 (3.3%) 14 (2.9%)
46% (-ll-74%) 35% (-46-71%)
0.09 0.29
80 (31.7%) 105 (43.0%)
131 (27.3%) 149 (31.2%)
19% (-13-42%) 40% (17-56%)
0.21 0.002
Number of cases (%) parasite prevalence All parasitaemias High-density (>500O/~L) parasitaemia
years (The Gambia,
P
l? falciparum
Malaria cases Fever and parasitaemia Fever and high-density parasitaemia Enlarged spleen (Hackett’s grade l-5) Severe-moderate anaemia (Hb < 8 g/dL)
1
“A net which is long enough to tuck under the mattress, that was either untorn or had <5 small finger-sized holes. Hb, haemoglobin concentration. Coverage of good nets: 0--- O-49% 100,
. 50-92%
c; 0, 0 ’
Transmission
‘.
147 infective bites intensity: In (EIR+ 1)
Fig. 1. Prevalence of P. falciparum parasitaemia in young children in 48 Gambian villages experiencing different intensities of transmission, measured as total number of infective bites/ person/transmission season. Villages with a high percentage of children sleeping under untreated nets in good condition (0) are compared to those with a lower coverage of nets (3). EIR, entomological inoculation rate.
The prevalence of parasitaemia in villages decreased as the percentage of children sleeping under a good net increased (r = -0.386, P = 0.007). This was most clearly seen amongst villages where the total number of infective bites per person during the transmission season was a15 (r = -0.725, P < 0.001: Fig. 2). When differences in transmission risk between villages were controlled for in the analysis, this association with bednet coverage became more marked (overall: r = -0.577, d.f.2,45; P < 0,001 and, in villages with more than 15 infective bites: r = -0.746, d.f.2,33; P < 0.001). Untreated
nets and diversion
of mosquito biting
In order to see whether the use of untreated nets increased the risk of malaria infection for other unprotected children in the village, we looked at infection among children who did not use a good or treated net. The prevalence of parasitaemia in unprotected children was compared between villages with similar transmission intensity but different levels of net use. To minimize the influence of insecticide-treated nets, data for 6 villages where treated nets were more common were excluded. In most of the 42 villages analysed, none of the children slept under a treated net (28 villages) and in 14 villages the coverage ranged between 3% and
I
.
I
1
20 40 60 80 Coverage of good nets (%)
100
I
I
Fig. 2. Effect of coverage of untreated nets in good condition on the prevalence of P. julciparum in Gambian villages with 315 infective bites per person per transmission season (n = 36). Villages for which the estimates of transmission may be inaccurate, owing to low mosquito numbers, are excluded.
11%. The prevalence of parasitaemia among children sleeping unprotected in villages where many children slept under a good/treated net was similar or lower than in villages where good nets were uncommon (Table 2). Household practices and socio-economic net use
status in relation to
Typically the poorest households lived in houses with mud-brick walls and thatched roof. They generally did not possess a radio or manufactured bed, and had no means of transport. Livestock ownership in rural Gambia is common, but 15% of the households categorized as ‘poorer’ had no animals. In contrast, over 70% of households in the other 2 socio-economic groups owned a radio, manufactured bed and some livestock. House construction improved markedly with wealth, and in the ‘wealthier’ group more families had homes with concrete walls (32%) and almost all (94%) had a metal roof (S. E. Clarke et al., in preparation). The prevalence of malaria parasitaemia declined markedly with increasing wealth, from 5 1% in the children in the poorer families to 42% in the middle group, and 33% in the wealthier households (x2 .,. = 10.93, d.f.2, P = 0.004). Both parasitological and interview data were available for 618 children (53% of those surveved). The distribution of key variables (age, Z? falcipa&n ‘prevalence and use of nets) within the data subset did not differ from the main data set. Children who slept under impregnated nets (31) or for whom there were no data on net use (4) were excluded from analysis. Use of untreated bednets, and their condition, was strongly
460
E. CLARKE
SIti
Table 2. Prevalence of Plasmodium falciparum who did not sleep under a good net (n = 551)
among
Gambian
ETAL.
children
Coverage of good or treated nets in village” Low (O-49% children) Transmission intensityb
Number
l? falciparum prevalence
212 136 58
o-19 20-69 70-165
High (50-93%
I? falciparum prevalence
Number
43% 54% 57%
children)
40 :i
P
35% 48% 26%
0.38 0.44 0.00 1
“Data from 6 villages where > 12% of children slept under insecticide-treated nets were excluded. bEstimated total number of infective bites per person during the rainy season.
associated with socio-economic status of the household (x” = 32.43, d.f.4, P < 0.001). Children’s net use increased from 64% in the poorest households to 73% in the middle group and 86% in the wealthiest households. The proportion of nets which were in good condition increased from 50% to 52% and 69%, respectively. In contrast, there were no significant differences in other household practices that would have explained the lower prevalence in children sleeping under an untreated net in good condition. Few households had used insecticide coils (17%) or aerosol sprays (3%) in the previous 7 nights, and insecticide use did not differ between households with nets in good condition and households without nets. Although more households used local repellents (traditional plant products), their use was greatest in households without a net (67% vs 47%, x2 = 14.62, d.f.1, P < 0.001). In general, few families kept antimalarial drugs in the home in case of illness (8%). Although a greater proportion of households with good nets kept chloroquine (12% vs 4% without nets, x2 = 6.90, d.f.1, P = 0.009), recent use of the drug to treat malaria in a child under 5 years at home did not differ significantly (10% vs 7%, x2 = 1.44, d.f.1, P = 0.23). Health facility attendance in connection with a recalled episode of malaria in a child (59% of cases overall) did not differ between groups. When the data were stratified by socio-economic group, and children sleeping under an untreated net in good condition were compared with those not using a net (Table 3), a lower prevalence of parasitaemia was seen only amongst the poorest households (62% protection [95% CI 14-83%], x2 = 5.56, d.f.1, P = 0.018). Among other household groups, children sleeping under a good net had only a marginally lower prevalence than children who did not use a net. Discussion The prevalence of I? falciparum infection was significantly lower in children who used untreated mosquito nets, provided the nets were in good condition, than in children not using a net. This trend was consistently
Table 3. Effect of untreated economic groups
bednets
on prevalence
observed across all malaria parameters, including clinical attacks. That children sleeping under untreated nets also had significantly lower prevalences of highdensity parasitaemia and severe-to-moderate anaemia suggests that nets may also be associated with reduced malaria morbidity (DELACOLLETTE & VAN DER STUYFT, 1993). The magnitude of the apparent protection was similar to that seen in the randomized trials of untreated nets (SNOW et al., 1988; hflAXWEI.L et al., 1999). As expected, the prevalence of infection was lowest among children who slept under insecticidetreated nets, irrespective of the condition of the net (LINES et al., 1987; CARNEVALE et al., 1992). The differences observed occurred in an area where unprotected individuals receive up to 165 infective bites per transmission season. Whether untreated nets provide any protection at higher transmission intensities remains uncertain. These findings replicate those reported previously (BRADLEY et al., 1986; CAMPBELL et al., 1987; LINDSAY & GIBSON, 1988; GENTON et al., 1994; D’ALEsSANDRO et al., 1995). However, as in these other crosssectional studies, net users in our data set were selfselected, and further analysis is required. Interpretation of data from cross-sectional prevalence studies is fraught with inherent difficulties, most notably the possibility that other confounding factors could explain any observed association. Even when data on potential confounding factors are collected, a larger sample is required in order to simultaneously control for each additional confounder in the analysis. Almost inevitably, therefore, analysis will be limited to those factors likely to have the strongest confounding effect. Furthermore, one can never exclude the possibility of residual confounding or that another important confounder has been overlooked (DAYEY SMITH et al., 1992). Only a randomized intervention study can avoid confounding, but it may not always be ethical to do so. This is the case with respect to untreated nets, now that numerous studies have established that an insecticide-treated net provides excellent protection from clinical attacks of malaria and deaths (ALONSO et al., 1993; LENGELER,
of P. falcipamcm
group
Number
Poorer Average Wealthier 95% CI, 95% confidence interval.
54 ;z
in different
household
socio-
Net in good condition
No net Socio-economic
malaria
P. falciparum prevalence
Number
P. falciparum prevalence
Protection (95% CI)
P
63% 44% 35%
48 101 95
40% 41% 31%
62% (14-83%) -
0.018 0.669 0.693
UNTREATED
461
BEDNETS AND MALARIA
1998). Unfortunately the results of the few randomized trials of untreated nets that have been carried out have been inconclusive. Whilst untreated nets have been consistently shown to provide protection against clinical attacks, this protection has not been statistically significant (SNOW et al., 1988; MAXWELL et al., 1999). Whether untreated nets provide significant protection from malaria will therefore never be fully resolved. Our data, which include information on several potentially important confounders, provided an opportunity to assess some of the alternative explanations for the apparent protective effect seen by earlier descriptive studies, and thus shed some fresh light on untreated nets. Even if untreated nets were to provide individual protection, it is feared that their use may increase the risk for others, since unprotected individuals will receive more mosquito bites than usual if they share a room with someone sleeping under an untreated net (LINES et al, 1987). It has also been suggested that this could, through increasing exposure to bites amongst the control group, explain the apparent protection seen in some studies. Had the effect we observed been due to diversion of mosquito feeding, unprotected children from villages where many children slept under nets should have had higher prevalences than those seen among unprotected children in villages where net use was low. This was not the case. Indeed the data suggest that children sleeping without nets may even receive fewer infective bites in villages where bednet coverage is high. Diversion may be important only for people sleeping unprotected in the same house as net users. Thus use of untreated nets is unlikely to increase the risk for others sleeping in the same village. Although net users in our data set were self-selected, and it is thus possible that the apparent protection was due to other differences between those that use nets and those that do not, we are able to show that the protective effect of good nets was not due to differences in exposure (mosquito density), other methods of personal protection, or socio-economic status. Both use and condition of bednets improved with household socio-economic status and, as expected, partially explained the apparent protection. Increasing socio-economic status was associated with lower infection rates, such that amongst wealthier households the use of an untreated net provided no additional benefit. The protection afforded by wealth appeared to be primarily indirect. Treatment practices, use of insecticides or mosquito repellents did not differ between households that used nets and those that did not. Amongst children not using nets, the prevalence of infection was much higher in those from the poorest families, perhaps indicating that exposure is greatest among the poor and they may have most to gain from the use of a net. This may explain why use of an untreated bednet in good condition by children in poor households was associated with significantly lower prevalence of infection. There is no doubt as to the enormous protective benefit gained by sleeping under an impregnated net, and that re-treatment of nets with insecticide should be vigorously promoted. However, unless re-impregnation rates dramatically improve, it is useful to know whether untreated nets can confer any individual protection, and whether they generate an increased risk to others. Our results suggest that untreated nets do not pose a threat to others sleeping in the village. Furthermore, net use should be particularly encouraged amongst the poorest families, who may derive a protective benefit even if their nets are not treated. Unfortunately these people, who may have most to gain from a net, can least afford to buy one. Bednet programmes that rely on social marketing will effectively exclude this group, unless they incorporate mechanisms, such as credit systems or vouchers, which facilitate access for the poorest.
Acknowledgements
The oroiect was funded bv the Danish Bilharziasis Laboratory, Denmark, and carried out in collaboration with the Medical Research Council (MRC) Laboratories in The Gambia. We thank Professor K. I’. W. J. McAdam and the MRC staff at Farafenni and Fajara for their support and assistance with the fieldwork. We are grateful to Sister M’boge, Chief Nursing Officer, Ministry of Health, Social Services and Women’s Affairs, for her assistance in identifying a nurse, and to nurse Lamin Fatty for his work during the parasitological surveys. A special acknowledgement is due to the field workers who carried out the surveys and interviews, for their hard work and enthusiasm, in particular Yorro Bah, Samba Baldeh, Malick John and Hawa Manneh, and also to Kebba Jammeh who read the blood slides. Thanks are also due to Paul Milligan for his assistance in generating the time-adjusted temoerature function, and to Derek Charlwood and Jo Lines for their comments -on the manuscript. We are especially grateful to the people in the study villages who took part in these investigations.
References
Aikins, M. K., Pickering, H., Alonso, I’. L., D’Alessandro, U., Lindsay, S. W., Todd, J. & Greenwood, B. M. (1993). A malaria control trial using insecticide-treated bednets and targeted chemoprophylaxis in a rural area of The Gambia, West Africa. 4. Perceptions of the causes of malaria and of its treatment and prevention in the study area. Transactions of the Royal Society of Tropical Medicine and Hygiene, 87,
supplement 2, S/25-S/30. Alonso, I’. L., Lindsay, S. W., Armstrong Schellenberg, J. R. M., Keita, K., Gomez, I?., Shenton, F. C., Hill, A. G., David, P. H., Fegan, G., Cham, K. & Greenwood, B. M. (1993). A malaria control trial usine insecticide-treated bednets and targeted chemoprophylax& in a rural area of The Gambia, West Africa. 6. The impact of the interventions on mortality and morbidity from malaria. Transactions of the Royal Society of Tropical Medicine and Hygiene, 87, supplement 2, Sl37-Sl44. Armstrong Schellenberg, J. R. M., Greenwood, B. M., Gomez, I’., Menendez, C. & Alonso, P. L. (1994). Diurnal variation in body temperature of Gambian children. Transactions of the Royal Society of Tropical Medicine and Hygiene, 88.429-431.
Bradley, A. K., Greenwood, B. M., Greenwood, A. M., Marsh, K.. Bvass, P.. Tulloch. S. & Haves. R. (1986). Bednets (&s&to-nets) and morbidity from’ malaria. Lancer, ii, 204-207.
Burkot, T. R., Garner, P., Paru, R., Dagoro, H., Barnes, A., McDougall, S., Wirtz, R. A., Campbell, G. & Spark, R. (1990). Effects of untreated bednets on the transmission of Plasmodium falciparum, P. vivax and Wuchereria bancrofti in Papua New Guinea. Transactions of the Royal Society of Tropical Medicine and Hygiene, 84, 773-779.
Bsgh, C., Thomas, C. J., Clarke, S. E. & Lindsay. S. W. (1999a). Localised breeding of Anopheles gambiae complex mosquitoes along the River Gambia, West Africa. In: Mariation in malaria transmission in rural Gambia, PhD thesis, C. Bmgh, University of Copenhagen, Denmark, pp. 20-37. Bngh, C., Clarke, S. E., Thomas, C. J., & Lindsay, S. W. (1999b). Sharp decline in malaria transmission away from river systems in The Gambia, West Africa. In: Vuriation in malaria transmission in rural Gambia, PhD thesis, C. Begb, University of Copenhagen, Denmark, pp. 40-54. Campbell, H., Byass, P. & Greenwood, B. M. (1987). Bednets and malaria suppression. Lancet, i, 859-860. Carnevale, P., Bitsindou, P., Diomande, L. & Robert, V. (1992). Insecticide impregnation can restore the efficiency of torn bednets and reduce man-vector contact in malaria endemic areas. Transactions of the Royal Society of Tropical Medicine and Hygiene, 86,362-364.
Cham, M. K., Olaleye, B., D’Alessandro, U., Aikins, M., Cham, B., Maine, N., Williams, L. A., Mills, A. & Greenwood, B. M. (1997). The impact of charging for insecticide on the Gambian National Impregnated Bednet Programme. Health Policy and Planning, 12, 240-247.
Clarke, S. E., Bsgh, C. & Lindsay, S. W. (1999). Personal protection measures against biting in response to mosquito densini in The Gambia. In: Variation in malaria transmission in n&l Gambia, PhD thesis, C. B0gl-1,University of Copenhagen, Denmark, pp. 104-l 18. D’Alessandro, U., Olaleye, B. O., McGuire, W., Thomson,
462
SIbiN E. CLARKE
M. C., Langerock, I’., Bennett, S. & Greenwood, B. M. (1995). A comparison of the efficacy of insecticide-treated and untreated bednets in preventing malaria in Gambian children. Transactions of the Royal Society of Tropical Medicine and Hygiene, 89,596-598. Davey Smith, G., Phillips, A. N. & Neaton, J. D. (1992). Smoking as an ‘independent’ risk factor for suicide: illustration of an artefact from observational epidemiology? Lancer, 340,709-712. Delacollette, C. & van der Stuyft, I’. (1993). High parasitaemia incidence rates can be used to estimate malaria morbiditv rates. Annals of Tronical Medicine and Parasitolopv, I_ 87,_ 537-539.
”
-
Genton, B., Hii, J., Al-Yaman, F., Paru, R., Beck, H. I’., Gimry, M., Dagoro, H., Lewis, D. & Alpers, M. P. (1994). The use of untreated bednets and malaria infection, morbidity, and immunity. Annals of Tropical Medicine and Parasitology, 88, 263-270. Greenwood, B. M. & Armstrong, J. R. M. (1991). Comparison of two simple methods fo;hetermining malaria parasite density. Transactions of the Royal Society of Tropical Medicine and Hygiene, 85, 186-188. Hill, A. G., MacLeod, W. B., Joof, D., Gomez, I’., Ratcliffe, A. A. & Walraven, G. (2000). Decline of mortality in children in rural Gambia: the influence of village-level Primary Health Care. Tropical Medicine and International Health, 5, 107-118. Lengeler, C. (1998). Insecticide treated nets and curtains for malaria control (Cochrane Review). In: The Cochrane Libray, Issue 3, Oxford: Update Software. Lindsay, S. W. & Gibson, M. E. (1988). Bednets revisitedold idea, new angle. Parasitology Today, 4, 270-272. Lindsay, S. W., Shenton, F. C., Snow, R. W. & Greenwood, B. M. (1989). Responses of Anopheles gambiae complex mosquitoes to the use of untreated bednets in The Gambia. Medicaland Veterinary Entomology, 3, 253-262. Lindsay, S. W., Alonso, I’. L., Armstrong Schellenberg, J. R. M., Hemingway, J., Thomas, P. J., Shenton, F. C. & Greenwood, B. M. (1993). A malaria control trial using insecticide-treated bed nets and targeted chemoprophylaxis in a rural area of The Gambia, West Africa. 3. Entomological characteristics of the study area. Transactions of the Royal Society of Tropical Medicine and Hygiene, 87, supplement 2, 19-23. Lines, J. D., Myamba, J. & Curtis, C. F. (1987). Experimental
TRANSACTIONS
I
1 CD-ROM Review
1
I
I
Malaria (2nd edition). A Topics in International Health CD-ROM. Wallingford: CAB1 Publishing, 2000. Price: Institutional Rate E120/US$195; Student/ Individual Rate L3O/LJS$55; Developing Country Rate E4XJSf80; Reduced Developing Country Rate E201 US$35. ISBN O-85199-494-6. The Wellcome Trust has produced a second edition of its CD-ROM on malaria, part of their Topics in International Health series. It has been compiled with guidance from distinguished malaria research workers. The software is easy to load and the programme is very user friendly, much more so than the first edition. The content consists of 13 interactive tutorials covering most aspects of malaria and a collection of images, 866 in all. Each tutorial has between 40 and 60 pages. A great deal of information
can be obtained
by simply
working through a tutorial on, say, Malaria in Z+egnancy or Zmmunity, page by page. In addition, however, there are pop-up windows on nearly every page to add more detail as required. The design of the tutorials is excellent with some very high quality images and drawings, They are a pleasure to use. Those who have used the first edition will find this one much more compreSpecial
hut trials of permethrin-impregnated mosquito nets and eave curtains against malaria vectors in Tanzania. Medical and Veterinary Entomology, 1, 37-51. Maxwell, C. A., Mvamba, 1.. Niunwa, K. 1.. Greenwood. B. M.. & Curtis, -C. F.- (1999). Comparison of bednets impregnated with different pyrethroids for their impact on mosquitoes and on re-infection with malaria after clearance of pre-existing infections with chlorproguanil-dapsone. Transactions of the Roval Society of Troaical Medicine and Hy.&ne,93,4111. ” ” * Z Miiller, O., Quidones, M., Cham, K., Aikins, M. & Greenwood, B. (1994). Detecting permethrin on treated bednets. Lancet, 344,1699-1700. Nevill, C. G., Watkins, W. M., Carter, J. Y. & Munafu, C. G. (1988). Comparison of mosquito nets, proguanil hydrochloride, and placebo to prevent malaria. British Medical Journal, 297,401-403. Port, G. R. & Boreham, P. F. L. (1982). The effect of bednets on feeding by Anopheles gambiae Giles (Diptera: Culicidae). Bulletin of Entomological Research, 72, 483-488. Snow, R. W., Rowan, K. M., Lindsay, S. W. & Greenwood, B. M. (1988). A trial of bednets (mosquito nets) as a malaria control strategy in a rural area of The Gambia, West Africa. Transactions of the Royal Society of Tropical Medicine and Hygiene, 82, 212-215. Snow, R. W., McCabe, E., Mbogo, C. N. N., Molyneaux, C. S., Some, E. S., Mung’ala, V. 0. & Nevill, C. G. (1999). The effect of delivery mechanisms on the uptake of bed net re-impregnation in Kilifi District, Kenya. Health Policy and Planning, 14, 18-25. Thomas, C. I. & Lindsav, S. W. (2000). Local-scale variation in malaria-infection amonest rural Gambian children estimated by satellite remote sensing. Transactions of the Royal Sociey of Tropical Medicine and Hygiene, 94, 159- 163. Winch, P. T., Makemba, A. M., Makame, V. R.. Mfaume, M. S., Lynch, M. C., Premji, i., Minjas, J. N. &Shiff, C. J: (1997). Social and cultural factors affecting rates of regular retreatment of mosquito nets with insecticide in Bagamoyo District. Tanzania. Tropical Medicine and International Health, 2, 760-770. 1
Received 22 December 2000; revised 29 March acceptedfor publication 9 May 2001
2001;
OF THE ROYAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE (2001) 95,462
I
hensive.
ETAL.
mention
should
be made
of the new
tutorials on Pathogenesis, Vector Biology and Roll Back Malaria, the last replacing the WHO Global Malaria Control Strategy. The now renamed tutorial on Parasite Biology contains a section on molecular approaches, which the previous edition did not include. There is a cross-referencing amongst tutorials and a search feature that will bring up all the information within the whole programme on a topic that you select. A comprehensive glossary helps the non-medical student with the 5 medically orientated tutorials, and the medical student with some scientific terms that may be unfamiliar. I was not able to display the pages on a full screen. Also, it is not possible to copy the images and text; but this is hardly surprising. I did find a few points of detail that were either incorrect or less clear-cut than stated. However, it would be inappropriate to highlight these since, overall, this CD-ROM is excellent for all who require training in malaria. It is particularly recommended for students who need to acquire an all-round knowledge of malaria but do not have the necessary expertise around them. Geoffrey Department
of Infectious and Tropical Diseases London School of Hyniene and Tropical Medicine
Keppel Street - -London WClE 7HT, UK
Targett
OF THE ROYAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE (200 1) 95457-462
Do untreated bednets protect against malaria? Margaret PinderZ, Gijs E. L. Walraven’ and Sign E. Clarke’*, Claus Begh’, Richard C. Brown’, Steve W. Lindsay’p3 ‘Danish Bilharziasis Laboratory, Jegersborg Alli 10, DK-2920 Charlottenlund, Denmark; ‘Medical Research Council Laboratories, I? 0. Box 273, Banjul, The Gambia; 3School of Biological and Biomedical Sciences, University of Durham, South Road, Durham DHl3LE, UK Abstract Bednets are thought to offer little, if any, protection against malaria, unless treated with insecticide. There is also concern that the use of untreated nets will cause people sleeping without nets to receive more mosquito bites, and thus increase the malaria risk for other community members. Regular retreatment of nets is therefore viewed as critical for malaria control. However, despite good uptake of nets, many control programmes in Africa have reported low re-treatment rates. We investigated whether untreated bednets had any protective benefit (in October and November 1996) in The Gambia where nets, although widely used, are mostly untreated. Cross-sectional prevalence surveys were carried out in 48 villages and the risk of malaria parasitaemia was compared in young children sleeping with or without nets. Use of an untreated bednet in good condition was associated with a significantly lower prevalence of Plasmodium falciparum infection (51% protection [95% CI 34-64%], P < 0.001). This finding was only partly explained by differences in wealth between households, and children in the poorest households benefited most from sleeping under an untreated net (62% protection [14-83%], P = 0.018). There was no evidence that mosquitoes were diverted to feed on children sleeping without nets. These findings suggest that an untreated net, provided it is in relatively good condition, can protect against malaria. Control programmes should target the poorest households as they may have the most to gain from using nets. Keywords: malaria,
Plasmodium
fulciparum, disease control, vector control, bednets, children, poverty, The Gambia
Introduction Treating mosquito nets and other fabrics with insecticides makes them an effective method of reducing malaria attacks and deaths (&ONSO et al., 1993; LENGELER, 1998). Whilst malaria control programmes in Africa have reported good uptake or purchase of nets, very few people, for a variety of reasons, re-treat their nets (WINCH et al. 1997; SNOW et al., 1999). For example, despite a long history of net use in The Gambia, and a national impregnation programme (NIBP) since 1992, fewer than 15% of nets are regularly retreated (CHAM et al., 1997). In practice, this means that large numbers of people are sleeping under untreated nets. It is thus important to know whether these people are protected against malaria. An intact net, if used correctly, can provide an effective barrier to mosquitoes (PORT & BOREHAM, 1982; LINES et al., 1987; LINDSAY et al., 1989; BURKOT et al, 1990), but whether the reduction in biting is sufficient to reduce malaria infection and morbidity rates in the human population is not clear. No study has convincingly demonstrated a protective effect for bednets that have not been impregnated with insecticide. Only 3 randomized intervention studies have been carried out. In community-based trials, children in villages using untreated nets had 38% fewer clinical attacks in The Gambia (SNOW et al., 1988), and 33% fewer in Tanzania (-WELL et al., 1999). Neither result, however, was statistically significant. In only one trial has a sirrnificant difference in the incidence of malaria attack; been seen (NEVILL et aZ., 1988), but the possibility that this difference was due to increased infection among the control children, who slept in the same room as those using nets, cannot be ruled out. Now that the enhanced impact of impregnated nets has been demonstrated, further randomized trials of untreated nets are unlikely, and the question of whether untreated nets can protect against malaria may never be definitivelv resolved. Evidence for a orotective effect of untreated nets will remain poor, be&g based largely on anecdotal accounts and cross-sectional data, in which users of untreated nets were self-selected (BRADLEY et al., 1986; CAMPBELL et al., 1987; LINDSAY & for correspondence; fax +45 77 32 77 33, e-mail [email protected]
*Author
1988; GENTON et aZ., 1994; D’ALESSANDRO et aZ., 1995). In these studies, confounding by other factors, such as wealth, could have been responsible for the lower prevalence observed among those that used nets compared to those that did not. It is also possible that the risk of transmission in non-net users was exacerbated because mosquitoes were diverted from users of untreated nets to feed on non-net users in the same house or village. In other words, the observed difference could have resulted from an increase of malaria among non-users rather than from a decrease among net users. Previous analyses lacked data on confounders, such as socio-economic status, and other measures used to reduce mosquito biting. Data from a cross-sectional study carried out in The Gambia included this information, and thus provided an opportunity to take account of the most likely alternative explanations when we compared the risk of malaria infection in children sleeping with or without nets, and thereby re-evaluate whether untreated nets can protect those who sleep under them from malaria. GIBSON,
Materials and Methods Study area and population Data were available from 48 villages in The Gambia, in a rural area with low-to-moderate malaria transmission concentrated in a single seasonal peak each year (LINDSAY et al., 1993). The study area was located about 200 km inland, and covered about 2500 km2 (between UTM 410 OOOE and 480 OOOE) around the towns of Farafenni, Soma and Kaur. The population, described in more detail by AIKINS et al. (1993), comprises predominantly subsistence farmers of the ethnic groups Mandinka, Wollof and Fula. Villages are discrete units, typically consisting of a cluster of family compounds, composed of closely related households. Health services within the study area include: 4 government health centres and private pharmacies located in towns; government disp&sarie< in 10 villages; and urimarv health care PHC) workers in the larger viliages. Outreach clinics also’operate once a mon& in 26 of the PHC villages, and are equally well attended by villagers from both PHC and non-PHC villages (HILL et al., 2000). A geographically stratified random sample (not proportional to size) was drawn from all villages
458 within the study area. The selected villages comprised 11 PHC and 37 non-PHC villages. Parasitological and entomological surveys Cross-sectional prevalence surveys of children aged 6 months-5 years were carried out during the peak transmission period in October and November 1996. In each village, a cluster sample of up to 30 children was selected, from compounds located on the periphery of the village closest to the alluvial floodplains where mosquitoes breed (B@GH et al., 1999a; THOMAS & LINDSAY, 2000). Children were examined, standing, for spleen enlargement. Fingerprick blood samples were obtained for blood films, and haemoglobin was measured using a Hemocue”’ haemoglobinometer. Thick films were examined at X 1000 magnification, and the number of parasites per high-powered field (HPF) was recorded. Parasite density per & was calculated assuming a blood volume per HPF of 0.002 ~.I.L(GREENWOOD & ARMSTRONG,-199 1). Children with 25000 uarasites/uL were classified as having high-density para&taemia.‘A film was considered negative if no parasites had been seen after 100 HPF had been examined. Plasmodium species were identified from thin films. Axillary temperature was measured with an electronic thermometer. To take account of diurnal variation in temperature (AMSTRONG SCHELLENBERG et al., 1994), surveys were usually carried out between 08:OO and lO:OO, and a fractional polynomial function used to fit a flexible curve to the iarcation in observed temneratures. usine STATATM. Fever was defined as a temperature grea?er than the 95th percentile value for the time of measurement (equivalent to 237~5°C at 09:15 and >38*O”C at 13:OO). Parasitepositive children with a raised temperature, but with symptoms suggesting another cause for fever, were not regarded as malaria cases. Mosquitoes were collected every second week throughout the transmission season (June-December), using light-traps. In each village, collections were made in 2 of the compounds selected for parasitological survey (B~GH et al., 1999b). For each village, the total number of infective bites per person in the transmission season (EIR) was estimated from the sum of the monthly EIRs, calculated as the geometric mean number of mosquitoes multiplied by the mean proportion of infective mosquitoes and number of days in the month. Household surveys and interviews Information on bednet usage and insecticide treatment was collected. The presence of a bednet was validated by observation, and the condition of the net recorded. A net was considered to be in good condition if it was long enough to be tucked under the mattress, was not torn or otherwise badly damaged, and had no more than 5 small holes (finger-width, approximate diameter ~2 cm). All other nets were classified as being in poor condition. All nets were tested for permethrin (M~~LLER et al., 1994), the insecticide used by the NIBP in The Gambia. Women responsible for the care and upbringing of at least one child under 5 years were interviewed, using a predominantly open-ended questionnaire, to obtain detailed socio-economic data, and information on knowledge and practices in relation to malaria. Up to 21 interviews were carried out in each village. The lottery method was used to initially select 21 compounds at random, followed by random selection of one woman per compound. A household was defined as the family sub-unit within a compound, comprising the informant, her husband and co-wives, and their children. Families were categorized into socio-economic groups based on 9 household features, which included ownership of household commodities, livestock, and house structure (S. E. Clarke et al., in preparation). Families possessing <4 of these features were categorized as
SIANE.CLARKE
ETAL.
‘poorer’ (24% of women interviewed), whilst families which possessed 4-5 of these features were regarded as ‘average’ (46%), and those with 26 as ‘wealthier’ (30%). In some cases the women interviewed cared for children who had been selected for inclusion in the parasitological survey, and we were able to relate the household practices and socio-economic differences in relation to bednet use to the parasitological findings. Statistical methods Data were analysed in EpiInfo (version 6.01b) and SPSS (version 9.0). Percent protection [with 95% confidence intervals (CI)] was obtained from 1 - odds ratio, and its confidence intervals. Categorical data were tested using the x2 test, and correlation and linear regression analysis used to test continuous data. Ethical considerations Ethical approval was obtained from the joint Gambian Government-Medical Research Council Ethics Committee. Informed consent was obtained from village leaders, and parents or guardians of the children, and from the women interviewed. All inhabitants of the study villages who requested medical attention were examined, and given treatment. Results A total of 1196 children were examined, of whom 468 (39%) were infected with I? falciparum parasites, including 7 children with dual infections with I? malariae. Children in whom only I? malariae (9) or l? ovale (3) were detected were excluded from analysis. A further 18 children were also excluded, owing to conflicting or insufficient information on bednet use. Among the 1166 children included in the analysis, 914 (78%) slept under a bednet, although many nets (388) were in a poor condition. Insecticide was detected on few nets (6.3%) by the end of the rains, when transmission was at its peak. Less than half of the children surveyed (48%) slept under a net in good condition or one treated with insecticide. Bednet use did not vary with age (x” = 2.43, d.f.4, P = 0.66). Neither did the condition of the net change with age, amongst those children sleeping under untreated nets (x’ = 7.28, d.f.4, P = 0.12). Net use, and condition, did not differ between PHC and non-PHC villages (x2 = 4.50, d.f.3, P = 0.21). Individual effect of bednets Less malaria was seen among children sleeping under an untreated net in good condition compared to those without a net (Table 1). Significant differences were observed in the prevalence gf I? falciparum parasitaemia (x2 = 20.78, d.f.1, P < O.OOl), high-density parasitaemia (x” = 14.51, d.f.1, P < 0.001) and severe-tomoderate anaemia (x2 = 9.84, d.f.1, P = 0.002). A similar pattern was seen with respect to clinical attacks, although few clinical cases of malaria were recorded. Use of an insecticide-impregnated net, irrespective of its condition, was associaiedwith the lowest prevalence of parasitaemia and 69% protection [95% CI 42-84%] (data not shown, x2 = 16.21, d.f.1, P < 0.001). Effect of bednet coverage in villages Both the mosquito density and the total estimated malaria transmission risk during the 5-month transmission season differed considerably between the 48 study villages, from
UNTREATED
Table
459
BEDNETS AND MALARIA
1. Effect of untreated
bednets
on malaria
in children
aged 6 months-5
1996)
Untreated net in good condition” (n = 481)
No net (n = 252)
Number of cases (%)
Protection (95% CI)
128 (50.8%) 65 (25.8%)
161 (335%) 69 (14.3%)
5 1% (34-64%) 52% (30-67%)
15 (6.0%) 11 (4.4%)
16 (3.3%) 14 (2.9%)
46% (-ll-74%) 35% (-46-71%)
0.09 0.29
80 (31.7%) 105 (43.0%)
131 (27.3%) 149 (31.2%)
19% (-13-42%) 40% (17-56%)
0.21 0.002
Number of cases (%) parasite prevalence All parasitaemias High-density (>500O/~L) parasitaemia
years (The Gambia,
P
l? falciparum
Malaria cases Fever and parasitaemia Fever and high-density parasitaemia Enlarged spleen (Hackett’s grade l-5) Severe-moderate anaemia (Hb < 8 g/dL)
1
“A net which is long enough to tuck under the mattress, that was either untorn or had <5 small finger-sized holes. Hb, haemoglobin concentration. Coverage of good nets: 0--- O-49% 100,
. 50-92%
c; 0, 0 ’
Transmission
‘.
147 infective bites intensity: In (EIR+ 1)
Fig. 1. Prevalence of P. falciparum parasitaemia in young children in 48 Gambian villages experiencing different intensities of transmission, measured as total number of infective bites/ person/transmission season. Villages with a high percentage of children sleeping under untreated nets in good condition (0) are compared to those with a lower coverage of nets (3). EIR, entomological inoculation rate.
The prevalence of parasitaemia in villages decreased as the percentage of children sleeping under a good net increased (r = -0.386, P = 0.007). This was most clearly seen amongst villages where the total number of infective bites per person during the transmission season was a15 (r = -0.725, P < 0.001: Fig. 2). When differences in transmission risk between villages were controlled for in the analysis, this association with bednet coverage became more marked (overall: r = -0.577, d.f.2,45; P < 0,001 and, in villages with more than 15 infective bites: r = -0.746, d.f.2,33; P < 0.001). Untreated
nets and diversion
of mosquito biting
In order to see whether the use of untreated nets increased the risk of malaria infection for other unprotected children in the village, we looked at infection among children who did not use a good or treated net. The prevalence of parasitaemia in unprotected children was compared between villages with similar transmission intensity but different levels of net use. To minimize the influence of insecticide-treated nets, data for 6 villages where treated nets were more common were excluded. In most of the 42 villages analysed, none of the children slept under a treated net (28 villages) and in 14 villages the coverage ranged between 3% and
I
.
I
1
20 40 60 80 Coverage of good nets (%)
100
I
I
Fig. 2. Effect of coverage of untreated nets in good condition on the prevalence of P. julciparum in Gambian villages with 315 infective bites per person per transmission season (n = 36). Villages for which the estimates of transmission may be inaccurate, owing to low mosquito numbers, are excluded.
11%. The prevalence of parasitaemia among children sleeping unprotected in villages where many children slept under a good/treated net was similar or lower than in villages where good nets were uncommon (Table 2). Household practices and socio-economic net use
status in relation to
Typically the poorest households lived in houses with mud-brick walls and thatched roof. They generally did not possess a radio or manufactured bed, and had no means of transport. Livestock ownership in rural Gambia is common, but 15% of the households categorized as ‘poorer’ had no animals. In contrast, over 70% of households in the other 2 socio-economic groups owned a radio, manufactured bed and some livestock. House construction improved markedly with wealth, and in the ‘wealthier’ group more families had homes with concrete walls (32%) and almost all (94%) had a metal roof (S. E. Clarke et al., in preparation). The prevalence of malaria parasitaemia declined markedly with increasing wealth, from 5 1% in the children in the poorer families to 42% in the middle group, and 33% in the wealthier households (x2 .,. = 10.93, d.f.2, P = 0.004). Both parasitological and interview data were available for 618 children (53% of those surveved). The distribution of key variables (age, Z? falcipa&n ‘prevalence and use of nets) within the data subset did not differ from the main data set. Children who slept under impregnated nets (31) or for whom there were no data on net use (4) were excluded from analysis. Use of untreated bednets, and their condition, was strongly
460
E. CLARKE
SIti
Table 2. Prevalence of Plasmodium falciparum who did not sleep under a good net (n = 551)
among
Gambian
ETAL.
children
Coverage of good or treated nets in village” Low (O-49% children) Transmission intensityb
Number
l? falciparum prevalence
212 136 58
o-19 20-69 70-165
High (50-93%
I? falciparum prevalence
Number
43% 54% 57%
children)
40 :i
P
35% 48% 26%
0.38 0.44 0.00 1
“Data from 6 villages where > 12% of children slept under insecticide-treated nets were excluded. bEstimated total number of infective bites per person during the rainy season.
associated with socio-economic status of the household (x” = 32.43, d.f.4, P < 0.001). Children’s net use increased from 64% in the poorest households to 73% in the middle group and 86% in the wealthiest households. The proportion of nets which were in good condition increased from 50% to 52% and 69%, respectively. In contrast, there were no significant differences in other household practices that would have explained the lower prevalence in children sleeping under an untreated net in good condition. Few households had used insecticide coils (17%) or aerosol sprays (3%) in the previous 7 nights, and insecticide use did not differ between households with nets in good condition and households without nets. Although more households used local repellents (traditional plant products), their use was greatest in households without a net (67% vs 47%, x2 = 14.62, d.f.1, P < 0.001). In general, few families kept antimalarial drugs in the home in case of illness (8%). Although a greater proportion of households with good nets kept chloroquine (12% vs 4% without nets, x2 = 6.90, d.f.1, P = 0.009), recent use of the drug to treat malaria in a child under 5 years at home did not differ significantly (10% vs 7%, x2 = 1.44, d.f.1, P = 0.23). Health facility attendance in connection with a recalled episode of malaria in a child (59% of cases overall) did not differ between groups. When the data were stratified by socio-economic group, and children sleeping under an untreated net in good condition were compared with those not using a net (Table 3), a lower prevalence of parasitaemia was seen only amongst the poorest households (62% protection [95% CI 14-83%], x2 = 5.56, d.f.1, P = 0.018). Among other household groups, children sleeping under a good net had only a marginally lower prevalence than children who did not use a net. Discussion The prevalence of I? falciparum infection was significantly lower in children who used untreated mosquito nets, provided the nets were in good condition, than in children not using a net. This trend was consistently
Table 3. Effect of untreated economic groups
bednets
on prevalence
observed across all malaria parameters, including clinical attacks. That children sleeping under untreated nets also had significantly lower prevalences of highdensity parasitaemia and severe-to-moderate anaemia suggests that nets may also be associated with reduced malaria morbidity (DELACOLLETTE & VAN DER STUYFT, 1993). The magnitude of the apparent protection was similar to that seen in the randomized trials of untreated nets (SNOW et al., 1988; hflAXWEI.L et al., 1999). As expected, the prevalence of infection was lowest among children who slept under insecticidetreated nets, irrespective of the condition of the net (LINES et al., 1987; CARNEVALE et al., 1992). The differences observed occurred in an area where unprotected individuals receive up to 165 infective bites per transmission season. Whether untreated nets provide any protection at higher transmission intensities remains uncertain. These findings replicate those reported previously (BRADLEY et al., 1986; CAMPBELL et al., 1987; LINDSAY & GIBSON, 1988; GENTON et al., 1994; D’ALEsSANDRO et al., 1995). However, as in these other crosssectional studies, net users in our data set were selfselected, and further analysis is required. Interpretation of data from cross-sectional prevalence studies is fraught with inherent difficulties, most notably the possibility that other confounding factors could explain any observed association. Even when data on potential confounding factors are collected, a larger sample is required in order to simultaneously control for each additional confounder in the analysis. Almost inevitably, therefore, analysis will be limited to those factors likely to have the strongest confounding effect. Furthermore, one can never exclude the possibility of residual confounding or that another important confounder has been overlooked (DAYEY SMITH et al., 1992). Only a randomized intervention study can avoid confounding, but it may not always be ethical to do so. This is the case with respect to untreated nets, now that numerous studies have established that an insecticide-treated net provides excellent protection from clinical attacks of malaria and deaths (ALONSO et al., 1993; LENGELER,
of P. falcipamcm
group
Number
Poorer Average Wealthier 95% CI, 95% confidence interval.
54 ;z
in different
household
socio-
Net in good condition
No net Socio-economic
malaria
P. falciparum prevalence
Number
P. falciparum prevalence
Protection (95% CI)
P
63% 44% 35%
48 101 95
40% 41% 31%
62% (14-83%) -
0.018 0.669 0.693
UNTREATED
461
BEDNETS AND MALARIA
1998). Unfortunately the results of the few randomized trials of untreated nets that have been carried out have been inconclusive. Whilst untreated nets have been consistently shown to provide protection against clinical attacks, this protection has not been statistically significant (SNOW et al., 1988; MAXWELL et al., 1999). Whether untreated nets provide significant protection from malaria will therefore never be fully resolved. Our data, which include information on several potentially important confounders, provided an opportunity to assess some of the alternative explanations for the apparent protective effect seen by earlier descriptive studies, and thus shed some fresh light on untreated nets. Even if untreated nets were to provide individual protection, it is feared that their use may increase the risk for others, since unprotected individuals will receive more mosquito bites than usual if they share a room with someone sleeping under an untreated net (LINES et al, 1987). It has also been suggested that this could, through increasing exposure to bites amongst the control group, explain the apparent protection seen in some studies. Had the effect we observed been due to diversion of mosquito feeding, unprotected children from villages where many children slept under nets should have had higher prevalences than those seen among unprotected children in villages where net use was low. This was not the case. Indeed the data suggest that children sleeping without nets may even receive fewer infective bites in villages where bednet coverage is high. Diversion may be important only for people sleeping unprotected in the same house as net users. Thus use of untreated nets is unlikely to increase the risk for others sleeping in the same village. Although net users in our data set were self-selected, and it is thus possible that the apparent protection was due to other differences between those that use nets and those that do not, we are able to show that the protective effect of good nets was not due to differences in exposure (mosquito density), other methods of personal protection, or socio-economic status. Both use and condition of bednets improved with household socio-economic status and, as expected, partially explained the apparent protection. Increasing socio-economic status was associated with lower infection rates, such that amongst wealthier households the use of an untreated net provided no additional benefit. The protection afforded by wealth appeared to be primarily indirect. Treatment practices, use of insecticides or mosquito repellents did not differ between households that used nets and those that did not. Amongst children not using nets, the prevalence of infection was much higher in those from the poorest families, perhaps indicating that exposure is greatest among the poor and they may have most to gain from the use of a net. This may explain why use of an untreated bednet in good condition by children in poor households was associated with significantly lower prevalence of infection. There is no doubt as to the enormous protective benefit gained by sleeping under an impregnated net, and that re-treatment of nets with insecticide should be vigorously promoted. However, unless re-impregnation rates dramatically improve, it is useful to know whether untreated nets can confer any individual protection, and whether they generate an increased risk to others. Our results suggest that untreated nets do not pose a threat to others sleeping in the village. Furthermore, net use should be particularly encouraged amongst the poorest families, who may derive a protective benefit even if their nets are not treated. Unfortunately these people, who may have most to gain from a net, can least afford to buy one. Bednet programmes that rely on social marketing will effectively exclude this group, unless they incorporate mechanisms, such as credit systems or vouchers, which facilitate access for the poorest.
Acknowledgements
The oroiect was funded bv the Danish Bilharziasis Laboratory, Denmark, and carried out in collaboration with the Medical Research Council (MRC) Laboratories in The Gambia. We thank Professor K. I’. W. J. McAdam and the MRC staff at Farafenni and Fajara for their support and assistance with the fieldwork. We are grateful to Sister M’boge, Chief Nursing Officer, Ministry of Health, Social Services and Women’s Affairs, for her assistance in identifying a nurse, and to nurse Lamin Fatty for his work during the parasitological surveys. A special acknowledgement is due to the field workers who carried out the surveys and interviews, for their hard work and enthusiasm, in particular Yorro Bah, Samba Baldeh, Malick John and Hawa Manneh, and also to Kebba Jammeh who read the blood slides. Thanks are also due to Paul Milligan for his assistance in generating the time-adjusted temoerature function, and to Derek Charlwood and Jo Lines for their comments -on the manuscript. We are especially grateful to the people in the study villages who took part in these investigations.
References
Aikins, M. K., Pickering, H., Alonso, I’. L., D’Alessandro, U., Lindsay, S. W., Todd, J. & Greenwood, B. M. (1993). A malaria control trial using insecticide-treated bednets and targeted chemoprophylaxis in a rural area of The Gambia, West Africa. 4. Perceptions of the causes of malaria and of its treatment and prevention in the study area. Transactions of the Royal Society of Tropical Medicine and Hygiene, 87,
supplement 2, S/25-S/30. Alonso, I’. L., Lindsay, S. W., Armstrong Schellenberg, J. R. M., Keita, K., Gomez, I?., Shenton, F. C., Hill, A. G., David, P. H., Fegan, G., Cham, K. & Greenwood, B. M. (1993). A malaria control trial usine insecticide-treated bednets and targeted chemoprophylax& in a rural area of The Gambia, West Africa. 6. The impact of the interventions on mortality and morbidity from malaria. Transactions of the Royal Society of Tropical Medicine and Hygiene, 87, supplement 2, Sl37-Sl44. Armstrong Schellenberg, J. R. M., Greenwood, B. M., Gomez, I’., Menendez, C. & Alonso, P. L. (1994). Diurnal variation in body temperature of Gambian children. Transactions of the Royal Society of Tropical Medicine and Hygiene, 88.429-431.
Bradley, A. K., Greenwood, B. M., Greenwood, A. M., Marsh, K.. Bvass, P.. Tulloch. S. & Haves. R. (1986). Bednets (&s&to-nets) and morbidity from’ malaria. Lancer, ii, 204-207.
Burkot, T. R., Garner, P., Paru, R., Dagoro, H., Barnes, A., McDougall, S., Wirtz, R. A., Campbell, G. & Spark, R. (1990). Effects of untreated bednets on the transmission of Plasmodium falciparum, P. vivax and Wuchereria bancrofti in Papua New Guinea. Transactions of the Royal Society of Tropical Medicine and Hygiene, 84, 773-779.
Bsgh, C., Thomas, C. J., Clarke, S. E. & Lindsay. S. W. (1999a). Localised breeding of Anopheles gambiae complex mosquitoes along the River Gambia, West Africa. In: Mariation in malaria transmission in rural Gambia, PhD thesis, C. Bmgh, University of Copenhagen, Denmark, pp. 20-37. Bngh, C., Clarke, S. E., Thomas, C. J., & Lindsay, S. W. (1999b). Sharp decline in malaria transmission away from river systems in The Gambia, West Africa. In: Vuriation in malaria transmission in rural Gambia, PhD thesis, C. Begb, University of Copenhagen, Denmark, pp. 40-54. Campbell, H., Byass, P. & Greenwood, B. M. (1987). Bednets and malaria suppression. Lancet, i, 859-860. Carnevale, P., Bitsindou, P., Diomande, L. & Robert, V. (1992). Insecticide impregnation can restore the efficiency of torn bednets and reduce man-vector contact in malaria endemic areas. Transactions of the Royal Society of Tropical Medicine and Hygiene, 86,362-364.
Cham, M. K., Olaleye, B., D’Alessandro, U., Aikins, M., Cham, B., Maine, N., Williams, L. A., Mills, A. & Greenwood, B. M. (1997). The impact of charging for insecticide on the Gambian National Impregnated Bednet Programme. Health Policy and Planning, 12, 240-247.
Clarke, S. E., Bsgh, C. & Lindsay, S. W. (1999). Personal protection measures against biting in response to mosquito densini in The Gambia. In: Variation in malaria transmission in n&l Gambia, PhD thesis, C. B0gl-1,University of Copenhagen, Denmark, pp. 104-l 18. D’Alessandro, U., Olaleye, B. O., McGuire, W., Thomson,
462
SIbiN E. CLARKE
M. C., Langerock, I’., Bennett, S. & Greenwood, B. M. (1995). A comparison of the efficacy of insecticide-treated and untreated bednets in preventing malaria in Gambian children. Transactions of the Royal Society of Tropical Medicine and Hygiene, 89,596-598. Davey Smith, G., Phillips, A. N. & Neaton, J. D. (1992). Smoking as an ‘independent’ risk factor for suicide: illustration of an artefact from observational epidemiology? Lancer, 340,709-712. Delacollette, C. & van der Stuyft, I’. (1993). High parasitaemia incidence rates can be used to estimate malaria morbiditv rates. Annals of Tronical Medicine and Parasitolopv, I_ 87,_ 537-539.
”
-
Genton, B., Hii, J., Al-Yaman, F., Paru, R., Beck, H. I’., Gimry, M., Dagoro, H., Lewis, D. & Alpers, M. P. (1994). The use of untreated bednets and malaria infection, morbidity, and immunity. Annals of Tropical Medicine and Parasitology, 88, 263-270. Greenwood, B. M. & Armstrong, J. R. M. (1991). Comparison of two simple methods fo;hetermining malaria parasite density. Transactions of the Royal Society of Tropical Medicine and Hygiene, 85, 186-188. Hill, A. G., MacLeod, W. B., Joof, D., Gomez, I’., Ratcliffe, A. A. & Walraven, G. (2000). Decline of mortality in children in rural Gambia: the influence of village-level Primary Health Care. Tropical Medicine and International Health, 5, 107-118. Lengeler, C. (1998). Insecticide treated nets and curtains for malaria control (Cochrane Review). In: The Cochrane Libray, Issue 3, Oxford: Update Software. Lindsay, S. W. & Gibson, M. E. (1988). Bednets revisitedold idea, new angle. Parasitology Today, 4, 270-272. Lindsay, S. W., Shenton, F. C., Snow, R. W. & Greenwood, B. M. (1989). Responses of Anopheles gambiae complex mosquitoes to the use of untreated bednets in The Gambia. Medicaland Veterinary Entomology, 3, 253-262. Lindsay, S. W., Alonso, I’. L., Armstrong Schellenberg, J. R. M., Hemingway, J., Thomas, P. J., Shenton, F. C. & Greenwood, B. M. (1993). A malaria control trial using insecticide-treated bed nets and targeted chemoprophylaxis in a rural area of The Gambia, West Africa. 3. Entomological characteristics of the study area. Transactions of the Royal Society of Tropical Medicine and Hygiene, 87, supplement 2, 19-23. Lines, J. D., Myamba, J. & Curtis, C. F. (1987). Experimental
TRANSACTIONS
I
1 CD-ROM Review
1
I
I
Malaria (2nd edition). A Topics in International Health CD-ROM. Wallingford: CAB1 Publishing, 2000. Price: Institutional Rate E120/US$195; Student/ Individual Rate L3O/LJS$55; Developing Country Rate E4XJSf80; Reduced Developing Country Rate E201 US$35. ISBN O-85199-494-6. The Wellcome Trust has produced a second edition of its CD-ROM on malaria, part of their Topics in International Health series. It has been compiled with guidance from distinguished malaria research workers. The software is easy to load and the programme is very user friendly, much more so than the first edition. The content consists of 13 interactive tutorials covering most aspects of malaria and a collection of images, 866 in all. Each tutorial has between 40 and 60 pages. A great deal of information
can be obtained
by simply
working through a tutorial on, say, Malaria in Z+egnancy or Zmmunity, page by page. In addition, however, there are pop-up windows on nearly every page to add more detail as required. The design of the tutorials is excellent with some very high quality images and drawings, They are a pleasure to use. Those who have used the first edition will find this one much more compreSpecial
hut trials of permethrin-impregnated mosquito nets and eave curtains against malaria vectors in Tanzania. Medical and Veterinary Entomology, 1, 37-51. Maxwell, C. A., Mvamba, 1.. Niunwa, K. 1.. Greenwood. B. M.. & Curtis, -C. F.- (1999). Comparison of bednets impregnated with different pyrethroids for their impact on mosquitoes and on re-infection with malaria after clearance of pre-existing infections with chlorproguanil-dapsone. Transactions of the Roval Society of Troaical Medicine and Hy.&ne,93,4111. ” ” * Z Miiller, O., Quidones, M., Cham, K., Aikins, M. & Greenwood, B. (1994). Detecting permethrin on treated bednets. Lancet, 344,1699-1700. Nevill, C. G., Watkins, W. M., Carter, J. Y. & Munafu, C. G. (1988). Comparison of mosquito nets, proguanil hydrochloride, and placebo to prevent malaria. British Medical Journal, 297,401-403. Port, G. R. & Boreham, P. F. L. (1982). The effect of bednets on feeding by Anopheles gambiae Giles (Diptera: Culicidae). Bulletin of Entomological Research, 72, 483-488. Snow, R. W., Rowan, K. M., Lindsay, S. W. & Greenwood, B. M. (1988). A trial of bednets (mosquito nets) as a malaria control strategy in a rural area of The Gambia, West Africa. Transactions of the Royal Society of Tropical Medicine and Hygiene, 82, 212-215. Snow, R. W., McCabe, E., Mbogo, C. N. N., Molyneaux, C. S., Some, E. S., Mung’ala, V. 0. & Nevill, C. G. (1999). The effect of delivery mechanisms on the uptake of bed net re-impregnation in Kilifi District, Kenya. Health Policy and Planning, 14, 18-25. Thomas, C. I. & Lindsav, S. W. (2000). Local-scale variation in malaria-infection amonest rural Gambian children estimated by satellite remote sensing. Transactions of the Royal Sociey of Tropical Medicine and Hygiene, 94, 159- 163. Winch, P. T., Makemba, A. M., Makame, V. R.. Mfaume, M. S., Lynch, M. C., Premji, i., Minjas, J. N. &Shiff, C. J: (1997). Social and cultural factors affecting rates of regular retreatment of mosquito nets with insecticide in Bagamoyo District. Tanzania. Tropical Medicine and International Health, 2, 760-770. 1
Received 22 December 2000; revised 29 March acceptedfor publication 9 May 2001
2001;
OF THE ROYAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE (2001) 95,462
I
hensive.
ETAL.
mention
should
be made
of the new
tutorials on Pathogenesis, Vector Biology and Roll Back Malaria, the last replacing the WHO Global Malaria Control Strategy. The now renamed tutorial on Parasite Biology contains a section on molecular approaches, which the previous edition did not include. There is a cross-referencing amongst tutorials and a search feature that will bring up all the information within the whole programme on a topic that you select. A comprehensive glossary helps the non-medical student with the 5 medically orientated tutorials, and the medical student with some scientific terms that may be unfamiliar. I was not able to display the pages on a full screen. Also, it is not possible to copy the images and text; but this is hardly surprising. I did find a few points of detail that were either incorrect or less clear-cut than stated. However, it would be inappropriate to highlight these since, overall, this CD-ROM is excellent for all who require training in malaria. It is particularly recommended for students who need to acquire an all-round knowledge of malaria but do not have the necessary expertise around them. Geoffrey Department
of Infectious and Tropical Diseases London School of Hyniene and Tropical Medicine
Keppel Street - -London WClE 7HT, UK
Targett