deet mosquito repellent sold through social marketing provides personal protection against malaria...

DEET mosquito repellent sold through social marketing

provides personal protection against malaria in an area of

all-night mosquito biting and partial coverage of

insecticide-treated nets: a case–control study of effectiveness

Mark Rowland1,2, Tim Freeman1, Gerald Downey2, Abdul Hadi1 and Mohammed Saeed1

1 HealthNet International, University Town, Peshawar, Pakistan2 London School of Hygiene and Tropical Medicine, London, UK

Summary DEET (Diethyl-3-methylbenzamide), the widely used mosquito repellent, has the potential to prevent

malarial infection but hitherto there has been no study demonstrating this possibility during normal

everyday use. MosbarTM, a repellent soap containing DEET, was promoted through social marketing in

villages in eastern Afghanistan. This was followed up with a case–control study of effectiveness against

malarial infection conducted through local clinics. Mosbar was purchased by 43% of households.

Reported use of insecticide-treated nets (ITNs) was 65% among the control group. There was a strong

association between Mosbar use and ITN use, as 81% of Mosbar users also possessed ITN. The use of

Mosbar was associated with a 45% reduction in the odds of malaria (95% CI: )11% to 72%, P ¼ 0.08)

after adjusting for ITN and other unmatched factors. Ownership of ITNs was associated with a 46%

reduction in the odds of malaria (95% CI: 12% to 67%, P ¼ 0.013) after adjusting for Mosbar and

other unmatched factors. The greatest reduction in the odds of malaria was associated with

combined use of Mosbar and ITN (69% reduction, 95% CI: 28% to 87%, P ¼ 0.007). The association

between recalled use of Mosbar 10 days ago (nearer the time of infection) and reduction in malarial

infections (adjusted odds ratio 0.08, 95% CI: 0.01–0.61, P ¼ 0.001) was significantly stronger than that

shown by current use of Mosbar. Most purchasers of Mosbar were satisfied with the product (74%),

although a minority said they preferred to use only ITN (8%). The local mosquito vectors,

Anopheles stephensi and A. nigerrimus, started biting shortly after dusk and continued biting until

early morning. It was shown that Mosbar prevented biting throughout this period. In regions where

mosquito vectors bite during evening and night, repellents could have a useful supplementary role to

ITN and their use should be more widely encouraged.

keywords repellents, DEET, insecticide-treated nets, malaria, mosquito, social marketing, Afghanistan

Introduction

The world market for synthetic insect repellents is

enormous. It is estimated that 31% of the USA popu-

lation uses a DEET repellent every year, and worldwide

use exceeds 200 million applications annually (Barnard

2000). While these figures testify to the utility and

popularity of skin repellents for providing protection

against mosquitoes and other biting insects, there have

been no studies showing that everyday use of skin

repellents can provide demonstrable or sustained pro-

tection against malaria. Until very recently (Rowland

et al. 2004) no controlled trial has shown a clear

effect upon malarial infection (e.g. Kroeger et al. 1997;

McGready et al. 2001), and hence the potential role of

repellents in public health has been relatively neglected.

Over the last decade the main emphasis of international

malaria control organizations has been to improve

coverage of insecticide-treated nets (ITN) (Nabarro &

Taylor 1998). While ITN are unquestionably a simple,

cheap and effective means of protection, it would seem

that in regions of the world where vectors bite in the

evening before people go to sleep, the level of protection

offered by ITN can be disappointingly low (Dolan et al.

1993; Luxemburger et al. 1994). For this reason there

has been renewed interest in finding out whether

mosquito repellents can provide additional protection or

improve upon the protection shown by sole use of ITN

Tropical Medicine and International Health

volume 9 no 3 pp 343–350 march 2004

ª 2004 Blackwell Publishing Ltd 343

(Lindsay et al. 1998; McGready et al. 2001; Pates et al.

2002).

DEET-based products of variable quality are sold

through general stores in most developing countries.

Mosbar is a repellent ‘soap’ (Yap 1986), cheaper than the

majority of repellent products, and was preferred over

other brands of DEET in a user acceptability study among

Afghans (M. Rowland, unpublished data). Mosbar is not

normally available commercially in Pakistan and Afghan-

istan. We adopted a strategy typical of any new household

product against mosquitoes: we promoted the product to

householders who could choose whether to buy it or not.

Our aims were to find out which group of people would be

encouraged to buy, whether this group already used other

forms of personal protection, and whether the repellent

had any protective effect against malaria. The protective

effect was measured by a clinic-based case–control study.

Materials and methods

Mosbar social marketing

Mosbar is a commercially available insect repellent made

from DEET (20%) and permethrin (0.5%) (Yap 1986). It

is a solid formulation which, when wetted and applied as a

lather to the skin, dries to a repellent film. In preparation

for the case–control study a social marketing project

centred on door-to-door sales of Mosbar was implemented

in several villages of Behsud district, Nangahar province,

eastern Afghanistan, in autumn 1999 and early summer

2000. The insect repellency effect of Mosbar was explained

to each household, and an opportunity to buy the product

at the standard retail price of US$0.50 was offered.

Promotion was aided by giving the product a suitable local

name, making the packaging attractive to local people, and

providing information in the local languages of Dari and

Pashtu. The salesmen were Ministry of Health (MoH)

staff, known to the community but having no previous

sales experience.

Case–control study

Cases and controls were identified by a process of passive

case detection in which outpatients with febrile illness were

examined for malaria at one of two participating clinics in

Behsud. This recording exercise began in July and continued

until December 2000. A variety of details were collected

from each patient including age and gender, whether they

were using Mosbar currently or had been using it 10 days

previously, whether they were using ITN currently or

10 days previously. The information on ITN use was

collected because ITNs were known to be a protective factor

for malaria and also a potential confounder of the relation-

ship between Mosbar use and outcome (malarial infection).

The case definition was a resident of Dobella with clinical

symptoms of malaria (temperature >37.5 �C, history of

fever, head or body ache) who attended one of the clinics

and tested positive for malaria parasites by microscopy.

Individuals from Dobella attending the clinic with febrile

illness but shown by microscopy to be malaria negative

were identified as controls. The two exposures of interest

were current use of Mosbar and use of Mosbar 10 days ago

(10 days being an approximate incubation time for malarial

infection). The use of ITN and other potential confounders

was recorded at the same time as exposure and outcome

status. This process of data collection was the same for cases

and controls. Only about 15% of outpatients with fever

actually had malaria, therefore the number of controls

exceeded the number of cases several fold. The selection of

cases and controls was made independently of Mosbar use,

and individual matching was not undertaken.

The unmatched case–control study was analysed using

logistic regression to estimate the odds of malaria in

Mosbar users vs. non-users of Mosbar and in ITN users vs.

non-users of ITN (Stata Release 5; Stata Corporation,

College Station, TX, USA). Univariate models were

initially fitted to the data in order to obtain crude odds

ratios. Multiple logistic regression models were then fitted,

with interpretation of odds ratios adjusted for the effect of

potential confounders and other explanatory variables.

Tests of interaction were also performed.

User acceptability

A survey of user experiences and preferences was conduc-

ted in Dobella and Narmasi villages in November 1999,

several months after the door-to-door sales campaign.

Ninety-nine householders who had purchased Mosbar

were selected at random and interviewed (one refused) by a

local anthropologist using a structured questionnaire.

Information was sought about Mosbar use among family

members, side-effects, popularity and willingness to

continue buying.

Entomology

We conducted a small efficacy study against mosquitoes in

June and July 1998 on the outskirts of Jalalabad in an area

of rice fields near the Kabul River. Four volunteers slept

outside on beds. Each was accompanied by a collector

whose task was to catch mosquitoes from the exposed skin

of the sleeping volunteer. Two sleepers were treated with

Mosbar and two were left untreated. All exposed skin was

treated including face and limbs. Throughout the night

Tropical Medicine and International Health volume 9 no 3 pp 343–350 march 2004

M. Rowland et al. Effectiveness of DEET

344 ª 2004 Blackwell Publishing Ltd

mosquitoes were caught, and every hour the collections

were stored and ambient temperature recorded. The

experiment ran for four nights. Individuals treated with

Mosbar served as untreated controls on alternate nights.

All volunteers were provided with chemoprophylaxis.

Ethical clearance

Ethical clearance for clinical, entomological and social

studies on Mosbar repellent was provided by the LSHTM

Ethics Committee. The Eastern Afghanistan Ministry of

Public Health granted permission for the project.

Results

Study characteristics

Of the 709 individuals who took part in the case–control

study, there were 96 cases of malaria. There were only

three cases of Plasmodium falciparum and one case of

mixed infection, and these were grouped with the 92

P. vivax cases for purposes of analysis. The first malaria

episode was recorded on the 1 July and the final episode on

the 20 December 2000.

The proportion of malaria episodes reported differed

greatly between the two clinics, with only 3.2% (9/284) of

individuals examined at the MoH clinic receiving a positive

malaria diagnosis. This contrasted with the 20.5% of

individuals (87/424) who received a positive malaria

diagnosis at the Austrian Relief Committee clinic (ARC).

Examination procedure differed between the two clinics.

The medical officer at the ARC clinic screened all patients

before referring them to the microscopist for diagnosis. At

the MoH clinic a slide was taken from anyone with

symptoms of malaria, as there was no medical officer

resident.

There was a greater proportion of children aged 9 and

younger among cases (71/96, 74%) than controls (349/613,

57%). There was a higher proportion of females represented

in the control group than in the case group. Overall, more

females (435/706, 61.6%) took part in the study.

Mosbar use recorded among the control groups (20.2%,

124/613) is one possible indicator of the uptake of Mosbar in

Dobella village. This percentage use among controls was

2.6-fold higher than recalled use of 10 days previously

(7.8%, 48/613). Mosbar use was greater among the controls

than cases both on the day of presentation and from their

recollection of 10 days previously, as shown in Table 1.

Table 1 Descriptive data for cases (N ¼ 96) and controls (N ¼ 613) and logistic regression models for explanatory variables

Cases n (%) Controls n (%)

Unadjusted odds

ratio (95% CI) P-value

Adjusted odds


Clinic

Austrian Relief Committee 87 (90.6) 337 (55.1) 1 1<0.001Ministry of Health 9 (9.4) 275 (44.9) 0.36 (0.25, 0.51) <0.001 0.39 (0.27, 0.56)

Age (years)

<1 24 (25.0) 134 (21.9) 1 1

1–4 33 (34.4) 162 (26.4) 1.14 (0.64, 2.02) 1.18 (0.65, 2.17)5–9 14 (14.6) 53 (8.7) 1.47 (0.71, 3.07) 2.15 (0.98, 4.74)

10–14 15 (15.6) 140 (22.8) 0.60 (0.30, 1.19) 0.89 (0.43, 1.85)

>15 10 (10.4) 124 (20.2) 0.45 (0.21, 0.98) 0.026 0.69 (0.30, 1.55) 0.146

Gender

Male 45 (47.4) 226 (37.0) 1 10.123Female 50 (52.6) 385 (63.0) 0.65 (0.42, 1.00) 0.054 0.69 (0.43, 1.11)

Mosbar use

No 85 (88.5) 489 (79.8) 1 10.082Yes 11 (11.5) 124 (20.2) 0.51 (0.26, 0.99) 0.045 0.55 (0.28, 1.11)

ITN ownership

No 61 (63.5) 213 (34.7) 1 10.013Yes 35 (36.5) 400 (65.3) 0.30 (0.20, 0.48) <0.001 0.54 (0.33, 0.88)

Mosbar use 10 days ago

No 95 (99.0) 565 (92.2) 1 1<0.001Yes 1 (1.0) 48 (7.8) 0.12 (0.02, 0.91) 0.040 0.08 (0.01, 0.61)

ITN use 10 days ago

No 18 (18.7) 334 (54.5) 1 10.003Yes 78 (81.3) 279 (45.5) 0.19 (0.11, 0.33) <0.001 0.42 (0.23, 0.77)




The area around Dobella has one of the highest

concentrations of ITNs in Afghanistan, estimated at 66%

coverage from HealthNet sales data. This is consistent with

the estimated use of ITN (65.3%, 400/613) obtained from

clinic controls data (Table 1). ITN use was three times

higher than the estimated Mosbar use. Only 16.5%

(66/400) of individuals in the control groups who stated

they were using ITN on the day of presentation recalled

they were not using nets 10 days previously. By contrast, a

much higher proportion, 61.3% (76/124), of individuals

from the control group who stated they were using Mosbar

on the day of presentation recalled they were not using

Mosbar 10 days previously. There was a strong association

between Mosbar use and ITN use: among the control

groups 25.2% (101/400) of ITN users were Mosbar users

but only 10.8% (23/213) of non-ITN users were Mosbar

users (v2 ¼ 18.0, d.f. ¼ 1, P < 0.0001). ITN use was

higher among the controls group than among the cases

group both on the day individuals presented at the clinic

and from recollection of 10 days previously.

Effectiveness

The univariate logistic regression model revealed that the

5–9-year age group was associated with greater odds of

malarial infections. There was an association between clinic

type and odds of malaria due to the screening of patients

prior to microscopy in the ARC clinic but not in the MoH

clinic (Table 1). Univariate analysis revealed that the use of

Mosbar was associated with a 49% reduction in the odds of

malaria (95% CI: 1% to 74%). This beneficial effect was

also apparent in the ownership of ITNs (70% reported

reduction, 95% CI: 52% to 80%). The recalled use of

Mosbar or ITNs 10 days previously was also significantly

associated with reduction in the odds of malaria (Table 1).

Multiple logistic regression models included Mosbar use

as the exposure variable, and ITN use, gender, age and

clinic type as explanatory variables. To avoid including a

combination of variables that were essentially measuring

the same quantity, current use of Mosbar and ITN were

not included in the same model as recalled use of Mosbar

and ITN of 10 days ago. The effects of Mosbar and ITN

use 10 days ago, controlling for the effect of other

variables, were obtained by fitting a separate logistic

regression model. There was no evidence that the effect of

Mosbar varied by clinic, age or gender (likelihood ratio

testing P ¼ 0.78, 0.19, 0.46, respectively).

The use of Mosbar was associated with a 45% reduction

in the odds of malaria (95% CI: )11% to 72%) after

adjusting for other unmatched factors. This effect was not

significant, unlike the effect shown in the univariate model,

but the magnitude of the reduction was large and had the

sample size been larger, it might have reached statistical

significance. The ownership of ITNs was associated with a

significant 46% reduction in the odds of malaria (95% CI:

12% to 67%) after adjusting for other unmatched factors.

This result was also smaller than that shown in the

univariate model (70%). Confounding appears to be

present in the relationship between ITN ownership and

malarial infections, with 24% of the observed reduction in

the univariate analysis being explained by other variables

in the final model.

The recalled use of Mosbar 10 days ago was associ-

ated with a 92% reduction in the odds of malaria (95%

CI: 39% to 99%) after adjusting for other unmatched

factors. Use of ITN 10 days ago was associated with a 58%

reduction in the odds of malaria (95% CI: 23% to 77%).

Both these results were statistically significant (Table 1).

The combined use of Mosbar and ITN was associated with

a 69% reduction in the odds of malaria (95% CI: 28% to

87%) after adjusting for other unmatched factors, whereas

use of Mosbar or ITN as unitary protection was associated

with only a 50% or 48% reduction, respectively (Table 2).

However, the added benefit of combining the use of Mosbar

Table 2 Descriptive data for cases (N ¼ 96) and controls (N ¼ 613) and logistic regression models for unitary and combined use of

Mosbar and ITN, adjusting for effect of other explanatory variables (age, gender and clinic)

Cases n (%) Controls n (%)

Unadjusted odds


Adjusted odds


Current use

Neither Mosbar nor ITN 57 (23.1) 190 (76.9) 1 1

Mosbar only 4 (14.8) 23 (85.2) 0.58 (0.20, 1.67) 0.327 0.50 (0.16, 1.52) 0.221ITN only 28 (8.6) 299 (91.4) 0.31 (0.19, 0.51) <0.001 0.52 (0.31, 0.89) 0.017

Both Mosbar and ITN 7 (6.5) 101 (93.5) 0.23 (0.10, 0.52) <0.001 0.31 (0.13, 0.72) 0.007

Use 10 days ago

Neither Mosbar nor ITN 77 (23.2) 255 (76.8) 1 1

Mosbar only 1 (4.0) 24 (96.0) 0.14 (0, 0.82) 0.025 0.11 (0.02, 0.87) 0.036

ITN only 18 (5.5) 310 (94.5) 0.19 (0.11, 0.33) <0.001 0.43 (0.24, 0.79) 0.007Both Mosbar and ITN 0 (0) 24 (100) 0 (0, 0.53) 0.008 0




and ITN over ITN alone (adjusted odds ratio 0.74; CI: 0.18

to 3.07; P ¼ 0.684) or Mosbar alone (adjusted odds ratio

0.54; CI: 0.22 to 1.33; P ¼ 0.178) was not significant. This

was also the case for combined use of Mosbar and ITN over

unitary use of Mosbar or ITN 10 days previously.

Model adequacy was tested using the Hosmer-

Lemeshow goodness-of-fit statistic. There was no evidence

for lack of fit (v28 ¼ 6.96, P ¼ 0.541).

Acceptability of Mosbar

Of the 613 controls participating in the study, 124 were

using Mosbar (20.4%). An alternative measure of its uptake

were sales records: 43% of households in Dobella had

bought it. The survey of 99 Mosbar-purchasing households

revealed that most were satisfied with the repellent (74%)

and willing to purchase more (64%). In 51% of the 99

households everyone used Mosbar, in 25% only adults did,

in 5% only children did, and in 18% no one was using

Mosbar despite having bought it. A small minority consid-

ered it was too much effort to apply Mosbar on every family

member every night (6%), and some preferred to use only

ITN (8%). Two complained they contracted malaria despite

using Mosbar, one did not like the smell, and one described a

burning sensation during use. Most could recite correctly

health messages included in the packaging. Most indicated

that the effect of Mosbar lasted all night (76%) and that one

bar would last an individual 30 days or more (66%).

Entomology

All species of mosquito started biting shortly after dusk.

Landing rates peaked between the hours of 21:00 and

23:00 and waned between midnight and 04:00 (Figure 1).

About 70% of culicine biting and 80% of anopheline

biting occurred before midnight. Temperature ranged

between 35–38 �C at 19:00, falling to 29–33 �C by 05:00

hours. During all four nights not one mosquito, either

culicine or anopheline, landed on the skin of the sleepers

treated with Mosbar. Although a number of mosquitoes

were observed to fly close to the treated individuals, none

actually landed.

Discussion

There was borderline evidence of an association between the

current use of Mosbar repellent and a reduction in malarial

infections. There was stronger evidence of an association

between the recalled use of Mosbar 10 days previously and a

reduction in malarial infections. An interval of just 10 days

seems unlikely to be a cause of recall bias, and yet there was a

61.3% difference in stated use over this interval. It is

conceivable that some individuals, on being probed by clinic

staff, might have felt tempted to say they were using the

repellent currently even if they were not but would be more

truthful about whether they were using Mosbar 10 days ago.

There was indication that individuals were less disciplined in

applying repellent each night than in using ITN each night.

The discrepancy between current use of ITN and stated use

10 days ago (30.3%) was much less than that for Mosbar

(61.3%). Thus the greater reduction in the odds of malarial

infection associated with recalled use of Mosbar 10 days

previously might be a more accurate measure of the level of

protection to be gained from regular use of DEET repellent.

It was certainly the case that users of DEET repellent stood to

gain considerable protection from mosquito biting, much of

which occurs in the early part of the night in this region of

Asia, as demonstrated in the entomological study. It follows

19:0

0

20:0

0

21:0

0

22:0

0

23:0

0

24:0

0

01:0

0

02:0

0

03:0

0

04:0

0

0

1

2

3

4

5

6

7

Mean mosquitolanding catch per

person

Culicines – control

Culicines – mosbar

A. nigerrimus – control

A. nigerrimus – mosbar

A. stephensi – control

A. stephensi – mosbar

A. pulcherrimus – control

A. pulcherrimus – mosbar

Time

Figure 1 Hourly mosquito biting ratesand the effect of Mosbar in a rice-growing

area in eastern Afghanistan.




that the recalled use of Mosbar 10 days ago among the

control group (7.8%) might be a more accurate measure of

the actual level of use in the community than the declared

current use among the controls (20.4%). Both measures

were substantially less than the number of families initially

buying Mosbar (43%), and we know from the user survey

that not all family members were regular users.

It has been demonstrated on previous occasions that ITN

provide protection against malaria in this region (Rowland

et al. 1996, 1997, 2002b), and the present case–control

study provides further evidence of this. The level of

protection associated with current ITN use was similar to

that of recalled use of ITN 10 days previously.

Malaria in the case–control study was not analysed

separately for P. falciparum or P. vivax infections due to the

small number of P. falciparum infections. The proportion of

P. falciparum to P. vivax infections tends to be less in eastern

Afghanistan than in Pakistan (Shah et al. 1997; Rowland

et al. 2002a). In the year prior to the case–control study a

cluster randomized trial of Mosbar had been undertaken in

an Afghan refugee camp just over the border in Pakistan

(Rowland et al. 2004). This showed that Mosbar was

effective in reducing the odds of P. falciparum malaria, with

a protective efficacy of 56% (95% CI: 24% to 75%). No

effect was shown against P. vivax malaria (protective

efficacy of )29%, 95% CI: )94% to 14%).

The conclusion reached in the case–control study

regarding the effect of Mosbar on P. vivax differs from that

of the cluster randomized trial. This is likely to be due to

the case–control study having started 2 months earlier in

July, a period when there is more active transmission of

P. vivax (Rowland et al. 2002a). The cluster randomized

trial ran from September to January, the period when the

frequency of P. falciparum infections increases and

P. vivax infections wanes.

Ordinarily case–control studies are more likely to suffer

from bias error than controlled trials, where treatments are

randomly allocated to units. The selection of clinic-based

controls (‘hospital controls’) can prove to be an incorrect

method of sampling controls from the study population if

admission to the clinic is related to the exposure of interest.

In the present study, where the exposure is Mosbar use, such

an association is unlikely. Selection biases still may result

because individuals who are taken ill with other febrile

conditions (i.e. the controls) may not necessarily be repre-

sentative of the population of individuals free of the disease

under study (malaria). The possibility of observer bias was

minimized in this study by ensuring that the recording of

exposure and outcome variables (and possible confounding

variables) was the same for cases and controls.

While the use of clinic/hospital controls has its advan-

tages, being convenient and cheap, with data of a

comparable quality, and with recall comparable with that

of cases, there are disadvantages with using hospital

controls. The question arises whether hospital controls are

representative of the general population. Villagers visiting

the clinic with what they regard as malaria-like symptoms

might simply be more health-conscious than other villag-

ers. This would lead to a greater representation of Mosbar

users in the control group than in the reference population,

and might overestimate the effect of Mosbar.

Information was collected on ITN use, age and sex, and

this was controlled for in the analysis. Other individual

patient characteristics that could be considered potential

confounders include socio-economic status, with the affluent

members of Dobella being possibly more proactive in

obtaining protection against malaria (as they could better

afford such interventions). As a result, there would be a

greater representation of more affluent villagers among

Mosbar users and the effect (upon malarial infections) of

Mosbar use would then be confounded with the effect of

socio-economic status. No information on socio-economic

status was collected during the present study. However, a

previous study did collect such information in relation to

ITN ownership, and demonstrated that ITN were more

likely to be purchased by the more affluent 25% of

households in the community (Howard et al. 2003). Since

ITN ownership and Mosbar use are correlated, there is a

high likelihood that socio-economic status is also a con-

founder of Mosbar use. This raises the key question: is there

an association between socio-economic status and malarial

risk that is confounding the recorded protective effect of ITN

and Mosbar interventions? In an earlier study of ITN

effectiveness in the same district of Afghanistan there was an

association between ITN ownership and socio-economic

status, but no association between socio-economic status

and malarial risk (Webster et al. 2003). In another study in

the same district there appeared to be no difference in

malaria prevalence, at the time of buying, between house-

holds that bought ITN and households that did not

(Rowland et al. 2002b). There is, therefore, convincing

evidence that malarial risk is similar across households of

differing socio-economic status (within the same village) and

a high likelihood that the effect of the Mosbar intervention –

as was the case with the earlier ITN intervention – is real.

The household survey appeared to confirm the accepta-

bility of Mosbar. The number of families purchasing

Mosbar was satisfactory, especially considering this was

the first year Mosbar had been marketed in Dobella.

However, it must be noted that Mosbar was marketed

door-to-door. The more usual form of selling such prod-

ucts, through shopkeepers in the bazaar, brought only

limited sales of Mosbar albeit after little or no product

promotion (T. Freeman, unpublished data). It seems that




both health information and good product promotion are

required for optimal selling of Mosbar as household

protection against malaria.

The group least at risk of contracting malaria were users

of Mosbar and ITN together. ITN do not, of course,

provide protection in the hours before people retire to bed,

and in regions of the world where vector mosquitoes start

biting in early evening, there would appear to be consid-

erable advantage to be gained from supplementing ITN

with use of repellents.

It is not clear whether the use of repellents in the target

villages would continue at the present level in years to

come. The door-to-door sales campaigns should be con-

tinued, alongside sales of repellent through shops over a

wider area, and long-term use and behaviour monitored.

The estimates of effectiveness described here might be

improved upon with a larger study. Effectiveness studies

should now be considered in other endemic countries.

Because the study was not powerful enough to show a

significant improvement of combined intervention over

unitary interventions, and because there might be an

element of residual confounding due to socio-economic

status, there is a case for confirming the extra benefit of

combined interventions in a factorial design, cluster rand-

omized controlled trial comparing ITN, Mosbar, and both

interventions together. Nevertheless, the evidence presen-

ted here, together with the evidence from the earlier cluster

randomized trial (Rowland et al. 2004) is sufficient to

justify marketing of repellents as a public health measure.

To reduce operational costs the social marketing of

repellents might be carried out alongside that of ITN. This

raises questions of equity and affordability and a strategy

that also addresses the needs of poor should be considered.

Acknowledgements

We thank the staff of the MoH and ARC clinics in Behsud,

entomologists Hamid Rehman and Mohammad Kamal,

ITN manager Noorullah, and two anonymous referees for

their helpful comments. HealthNet International’s malaria

control and research programme is supported by the

European Commission, and WHO/UNDP/World Bank

Special Programme for Research and Training in Tropical

Diseases. MR is supported by the UK Department for

International Development and the Gates Malaria Partner-

ship. However, none of these donors can accept respon-

sibility for any information provided or views expressed.

References

Barnard DR (2000) Repellents and Toxicants for Personal

Protection. WHO/CDS/WHOPES/GCDPP/2000.5. Global

Collaboration for Development of Pesticides for Public

Health, WHO, Geneva.

Dolan G, Ter Kuile FO, Jacoutot V et al. (1993) Bed nets for the

prevention of malaria and anaemia in pregnancy. Transactions

of the Royal Society of Tropical Medicine and Hygiene 87,

620–626.

Howard N, Rowland M, Chandramohan D et al. (2003) Socio-

economic factors associated with the purchasing of insecticide

treated nets in Afghanistan, and their implications for social

marketing. Tropical Medicine and International Health 8,

1043–1050.

Kroeger A, Gerhardus A, Kruger G, Mancheno M & Pesse K

(1997) The contribution of repellent soap to malaria control.

American Journal of Tropical Medicine and Hygiene 56,

580–584.

Lindsay SW, Ewald JA, Samung Y, Apiwathnasorn C & Nosten F

(1998) Thanaka and DEET mixture as a mosquito repellent for

use by Karen women. Medical and Veterinary Entomology 12,

295–301.

Luxemburger C, Perea WA, Delmas G, Pruja C & Moren A (1994)

Permethrin impregnated bed nets for the prevention of malaria

in schoolchildren on the Thai-Burmese border. Transactions

of the Royal Society of Tropical Medicine and Hygiene 88,

155–159.

McGready R, Simpson JA, Htway M, White NJ, Nosten F &

Lindsay SW (2001) A double-blind randomised therapeutic trial

of insect repellents for the prevention of malaria in pregnancy.

Transactions of the Royal Society of Tropical Medicine and

Hygiene 95, 137–138.

Nabarro DN & Taylor EM (1998) The Roll Back Malaria

campaign. Science 280, 2067–2068.

Pates HV, Lines JD, Keto AJ & Miller JE (2002) Personal

protection against mosquitoes in Dar es Salaam, Tanzania, by

using a kerosene oil lamp to vaporise transfluthrin. Medical and

Veterinary Entomology 16, 277–284.

Rowland M, Bouma M, Ducornez D et al. (1996) Pyrethroid

impregnated bed nets for self protection from malaria for

Afghan refugees. Transactions of the Royal Society of Tropical

Medicine and Hygiene 90, 357–361.

Rowland M, Hewitt S, Durrani N, Saleh P, Bouma M & Sondorp

E (1997) Sustainability of pyrethroid-impregnated bednets for

malaria control in Afghan communities. Bulletin of the World

Health Organization 75, 23–29.

Rowland M, Mohammed N, Rehman H et al. (2002a) Mosquito

vectors and transmission of malaria in eastern Afghanistan.

Transactions of the Royal Society of Tropical Medicine and

Hygiene 96, 620–626.

Rowland M, Webster J, Saleh P et al. (2002b) Prevention of

malaria in Afghanistan through social marketing of insecticide

treated nets: evaluation of coverage and effectiveness by

cross-sectional surveys and passive surveillance. Tropical

Medicine and International Health 7, 813–822.

Rowland M, Downey G, Rab A et al. (2004) DEET mosquito

repellent provides personal protection against malaria: a house-

hold randomized trial in an Afghan refugee camp in Pakistan.

Tropical Medicine and International Health 9, 335–342.




Shah I, Rowland M, Mehmood P et al. (1997) Chloroquine

resistance in Pakistan and the upsurge of falciparum malaria in

Pakistani and Afghan refugee populations. Annals of Tropical

Medicine and Parasitology 91, 591–602.

Webster J, Chandramohan D, Freeman T et al. (2003) A health

facility based case-control study of effectiveness of insecticide

treated nets: potential for selection bias due to pre-treatment

with chloroquine. Tropical Medicine and International Health

8, 196–201.

Yap HH (1986) Effectiveness of soap formulations containing

DEET and permethrin as personal protection against outdoor

mosquitoes in Malaysia. Journal of the American Mosquito

Control Association 2, 63–67.

Authors

Abdul Hadi, Tim Freeman and Mohammed Saeed, HealthNet International, 11A Circular Lane, University Town, Peshawar, Pakistan.

E-mail: [email protected] (Hadi and Saeed), [email protected] (Freeman).

Mark Rowland (corresponding author) and Gerald Downey, London School of Hygiene and Tropical Medicine, Keppel Street,

London WC1E 7HT, UK. Fax: 00 44 20 7299 4720; E-mail: [email protected], [email protected]




deet mosquito repellent sold through social marketing provides personal protection against malaria...

Documents