Serologic markers for detecting malaria in areas of low endemicity, Somalia, 2008.

Abstract

Areas in which malaria is not highly endemic are suitable for malaria elimination, but assessing transmission is difficult because of lack of sensitivity of commonly used methods. We evaluated serologic markers for detecting variation in malaria exposure in Somalia. Plasmodium falciparum or P. vivax was not detected by microscopy in cross-sectional surveys of samples from persons during the dry (0/1,178) and wet (0/1,128) seasons. Antibody responses against P. falciparum or P. vivax were detected in 17.9% (179/1,001) and 19.3% (202/1,044) of persons tested. Reactivity against P. falciparum was significantly different between 3 villages (p<0.001); clusters of seroreactivity were present. Distance to the nearest seasonal river was negatively associated with P. falciparum (p = 0.028) and P. vivax seroreactivity (p = 0.016). Serologic markers are a promising tool for detecting spatial variation in malaria exposure and evaluating malaria control efforts in areas where transmission has decreased to levels below the detection limit of microscopy.

Full text

Page 1

Areas in which malaria is not highly endemic are suitable
for malaria elimination, but assessing transmission is diffi -
cult because of lack of sensitivity of commonly used meth-
ods. We evaluated serologic markers for detecting variation
in malaria exposure in Somalia. Plasmodium falciparum or
P. vivax was not detected by microscopy in cross-sectional
surveys of samples from persons during the dry (0/1,178)
and wet (0/1,128) seasons. Antibody responses against P.
falciparum or P. vivax were detected in 17.9% (179/1,001)
and 19.3% (202/1,044) of persons tested. Reactivity against
P. falciparum was signifi cantly different between 3 villages
(p<0.001); clusters of seroreactivity were present. Distance
to the nearest seasonal river was negatively associated with
P. falciparum (p = 0.028) and P. vivax seroreactivity (p =
0.016). Serologic markers are a promising tool for detecting
spatial variation in malaria exposure and evaluating malaria
control efforts in areas where transmission has decreased
to levels below the detection limit of microscopy.
S
where P. falciparum parasite prevalence is >50% in the
general population are located in Africa (1). However, ma-
ub-Saharan Africa has the highest incidence of malaria
caused by Plasmodium falciparum. Almost all areas
laria is not uniformly distributed (1,2) and many parts of
Africa are characterized by low transmission intensity of
malaria (1). These areas are considered suitable for inten-
sive malaria control and disease elimination (3,4).
Assessing malaria transmission intensity and evaluat-
ing interventions are complicated at low levels of malaria
transmission. Assessing transmission intensity directly by
determining the exposure to malaria-infected mosquitoes
(entomologic inoculation rate [EIR]) is diffi cult when mos-
quito numbers are low, sometimes below the detection lim-
its of commonly used trapping methods (5,6), and spatial
and temporal variations in mosquito densities necessitate
long-term intensive sampling (5,7,8). Determination of ma-
laria parasite prevalence in the human population is a com-
monly used alternative (9), but it also becomes less reliable
as an indicator of transmission intensity when endemicity
is low (3,9,10). Therefore, an alternative method is needed
to assess transmission intensity, evaluate interventions, and
obtain information for control programs in areas of low en-
demicity.
Prevalence of antibodies against malaria parasites has
been explored as a means of assessing malaria transmis-
sion intensity (11–13). Antibody seroconversion rates are
less susceptible to seasonal fl uctuations in malaria expo-
sure (11,12), show a tight correlation with EIR (12,13), and
show potential to detect recent changes in malaria transmis-
sion intensity (14). Serologic markers could be particularly
useful in areas of low endemicity, where it may be easier to
detect relatively long-lasting antibody responses than a low
prevalence of malaria infections in the human population or
infected mosquitoes. We used serologic markers of expo-
sure to determine spatial variation in malaria transmission
intensity in an area of low endemicity in Somalia (15).
S erologic Markers for Detec ting
Malaria in Areas of Low Endemic ity,
S omalia, 2008
Teun Bousema, Randa M. Youssef, Jackie Cook, Jonathan Cox, Victor A. Alegana, Jamal Amran,
Abdisalan M. Noor, Robert W. Snow, and Chris Drakeley
RESEARCH
392 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 16, No. 3, March 2010
Author affi liations: London School of Hygiene and Tropical Medi-
cine, London, UK (T. Bousema, J. Cook, J. Cox, C. Drakeley);
University of Alexandria, Alexandria, Egypt (R.M. Youssef); Kenya
Medical Research Institute–Wellcome Trust Research Programme,
Nairobi, Kenya (R.M. Youssef, V.A. Alegana, J. Amran, A.M. Noor,
R.W. Snow); Roll Back Malaria–World Health Organization, Har-
geisa, Somalia (J. Amran); and University of Oxford, Oxford, UK
(A.M. Noor, R.W. Snow)
DOI: 10.3201/eid1603.090732

Page 2

Serologic Markers for Detecting Malaria
Methods
Study Area
This study was conducted in the Gebiley District in
Somaliland in northwestern Somalia. The district has a pre-
dominantly arid landscape with a few seasonal rivers and
patches of irrigated farmlands. It is an area of intense sea-
sonal rainfall with an average annual precipitation of 59.9
mm (2004–2007) and 2 peaks in rainfall in April and Au-
gust. Three moderately sized communities were randomly
selected from census maps by using spatial random sampling
techniques in Arcview version 3.2 (Environmental Systems
Research Institute, Redlands, CA, USA) (16). These com-
munities were the villages of Xuunshaley (9.72140°N,
43.42416°E), Badahabo (9.68497°N, 43.65616°E), and
Ceel-Bardaale (9.81777°N, 43.47455°E). The research
protocol was reviewed and approved by the Research Eth-
ics Review Committee of the World Health Organization
(RPC246-EMRO) and the Ethical Committee of the Minis-
try of Health and Labor, Republic of Somaliland.
Data Collection
Two cross-sectional surveys were conducted. The fi rst
survey was conducted in March 2008 to determine parasite
carriage at the end of the dry season (16). The purpose of
the survey and the procedures were fi rst discussed with the
clan elders; thereafter, each household was visited, and in-
formed consent was sought from each head of household.
Households that agreed to participate were geolocated by
using a global positioning system (Garmin eTrex; Garmin
International, Inc., Olathe, KS, USA), and information was
collected on demographic characteristics, bed net use, and
travel history of the participants. Distance to seasonal rivers
or other water bodies and distance to the nearest livestock
enclosure was determined by using the global positioning
system.
Individual written consent was obtained from all liter-
ate adults; illiterate adults provided consent by a thumbprint
in the presence of an independent literate adult witness.
For children <18 years of age, consent was obtained from
parents or guardians, and children 12–18 years of age who
could not write also provided consent by a thumb print.
One fi ngerprick blood sample was obtained from each
respondent for the preparation of a P. falciparum antigen–
specifi c rapid diagnostic test (RDT) (Paracheck-Pf; Orchid
Biomedical Systems, Goa, India) sample and thick and thin
blood smears. One-hundred high-power microscopic fi elds
were examined and an additional 100 fi elds were exam-
ined if the fi rst 100 fi elds were negative. RDT results were
used for treatment with sulfadoxine-pyrimethamine and 3
doses of artesunate according to national guidelines. A sec-
ond cross-sectional survey was conducted at the end of the
wet season (August–September 2008) by using procedures
identical to those described above, except that part of the
fi ngerprick blood sample was placed on fi lter paper (3 MM;
Whatman, Maidstone, UK) as described by Corran et al.
(17).
Entomologic Surveys
Presence of Anopheles spp. mosquitoes in the area was
determined by larvae collections in all permanent water
bodies (artifi cial rain water reservoirs, wells, boreholes,
stagnant storage pits, and riverbeds) in the 3 villages at the
end of the wet season. Locally produced 250-mL dippers
with a white surface were used. Five to 10 dips were made
in the large water bodies and the presence of Anopheles
spp. larvae was visually assessed and recorded.
Elution of Serum
Filter paper samples were stored at 4°C with desiccant
until processed. A 3.5-mm blood spot, equivalent to ≈3 μL
of blood (17), was punched from the fi lter paper and placed
in a labeled well of a low-binding 96-well titer plate. A to-
tal of 300 μL of reconstitution buffer (phosphate-buffered
saline [PBS], 0.05%Tween, and 0.1% [wt/vol] sodium az-
ide) was added, and plates were sealed and rocked gently
at room temperature overnight and subsequently stored at
4°C. The reconstituted blood spot solution was equivalent
to a 1:100 dilution of whole blood or a 1:200 dilution of
serum.
ELISAs
All reconstituted fi lter paper spots were tested at a
fi nal serum dilution equivalent of 1:1,000 for human im-
munoglobulin G antibodies against P. falciparum mero-
zoite surface protein 119 (MSP-119) and 1:2,000 for anti-
bodies against apical membrane antigen 1 (AMA-1) by
using described ELISA methods (12,17). Briefl y, recom-
binant MSP-119 (Wellcome genotype) and AMA-1 (3D7)
were coated overnight at 4°C at a concentration of 0.5 μg/
mL. Plates were washed by using PBS, 0.05% Tween 20
(PBS/T) and blocked for 3 h with 1% (wt/vol) skim milk
powder in PBS/T. Positive controls (a pool of hyperim-
mune serum) and negative controls (European malaria-
negative volunteers) were added in duplicate to each plate.
The plates were washed and horseradish peroxidase–conju-
gated rabbit anti-human immunoglobulin G (Dako, Rosk-
ilde, Denmark) (1:5,000 dilution in PBS/T) was added to
all wells. Plates were developed for 20 min by using an
o-phenylenediamine dihydrochloride substrate solution.
Reactions were stopped with 2 mol/L H2SO4. Plates were
read immediately at 492 nm and optical density (OD) val-
ues recorded. For P. vivax, an identical protocol was used
with MSP-119 (0. 5 μg/mL) (18) and AMA-1 (0. 5 μg/mL).
Serum in this protocol was used at 1:1,000 dilutions for
both antigens.
Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 16, No. 3, March 2010 393

Page 3

RESEARCH
Data Management and Statistical Analyses
Data were double-entered and imported into STATA
version 10 (StataCorp LP, College Station, TX, USA). Du-
plicate OD results were averaged and normalized against
the positive control sample on each plate. A cutoff value
above which samples were considered antibody positive
was defi ned by using a mixture model as described (17).
Distribution of normalized OD values was fi tted as the sum
of 2 Gaussian distributions by using maximum-likelihood
methods. The mean OD of the Gaussian distribution cor-
responding to the seronegative population plus 3 SD val-
ues was used as the cutoff value for seropositivity (J. Cook
et al., unpub. data). A separate cutoff value was generated
for each antigen (MSP-119 and AMA-1) for each species
(P. vivax and P. falciparum). The seroconversion rate was
estimated by fi tting a simple reversible catalytic model to
the measured seroprevalence by age in years by using max-
imum-likelihood methods. The serologic-derived annual
EIR was then estimated by using the MSP-119 seroconver-
sion rate and a calibration curve derived from determined
values (11).The titer of antibody responses was estimated
by using the formula dilution/[maximum OD/(OD test se-
rum – minimum OD) – 1]; the median titer and interquartile
range (IQR) are given. Because of low overall antibody
prevalence, antibody responses were combined by species
to determine the presence of any reactivity against P. falci-
parum or P. vivax. As a quantitative measure of reactivity
to either malaria species, the highest titer in the MSP-119
and AMA-1 ELISAs was used.
Factors associated with P. falciparum or P. vivax se-
roreactivity were determined for each village separately by
using generalized estimating equations adjusting for corre-
lation between observations from the same household. The
following factors were tested in the models: age in years,
distance to the nearest seasonal river (in 100 m), distance
to the nearest enclosure of livestock (in 100 m), number of
household members, number of houses in a 100-m radius,
roofi ng material, wall material, fl oor material, travel histo-
ry, recent or regular bed net use, and an indicator of house-
hold wealth. The household wealth index was calculated on
the basis of principal component analysis on characteristics
such as ownership of a television, radio, telephone, bicycle,
motorbike, cattle, and access to electricity (19). Variables
that were signifi cant at p = 0.10 in univariate analyses were
added to the multivariate model and retained in the fi nal
multivariate model if their association with immune re-
sponses was statistically signifi cant at p<0.05.
For detection of spatial clusters in immune responses,
age-adjusted log10-transformed ODs were calculated as de-
scribed by Wilson et al (20). First, Loess lines were fi tted to
scatter plots of age against log-transformed ODs for each
antigen separately. For P. falciparum MSP-119 and P. vivax
AMA-1, the linear regression was split at 49 and 46 years of
age. Log-transformed ODs were adjusted for age by linear
regression. SaTScan software (21) was used for detection
of spatial clustering in log-transformed age-adjusted OD
values by using the normal probability model. A circular-
shaped window was used to systematically scan the area of
each village separately; statistical signifi cance of the clus-
ters was explored by using 999 Monte Carlo replications to
ensure adequate power for defi ning clusters. The upper limit
was specifi ed as 50% of the village population. Signifi cant
increases in ODs were detected by calculation of the likeli-
hood ratio for each window. Only clusters were reported
that appeared for MSP-119, AMA-1, and their combined
age-adjusted ODs. Maps were made by using ArcGIS ver-
sion 9.1 (Environmental Systems Research Institute).
Results
The 2 cross-sectional surveys were completed in
March (dry season, n = 1,178) and August–September (wet
season, n = 1,128) 2008. These surveys were characterized
by a clear seasonality with no rainfall detected during No-
vember 2007–March 2008 and a median monthly rainfall
of 114.5 mm in April–August 2008. None of the survey
participants were positive by rapid diagnostic test, and P.
falciparum or P. vivax parasites were not detected on any
of the examined blood slides (Table 1). Available hospital
records indicated 2/283 slide-confi rmed, RDT-confi rmed
malaria cases in the study area in July and August 2008
(T. Bousema, unpub. data). Travel history was not avail-
able for these persons. During August–September 2008, a
total of 464 potential breeding sites were examined in Xu-
unshaley (n = 40), Badahabo (n = 42), and Ceel-Bardaale (n
= 382). In Ceel-Bardaale, 158 Anopheles mosquito larvae
were found at 81 of 382 examined sites. In the 2 other vil-
lages, no Anopheles larvae were observed.
Malaria Exposure Assessed by Immunologic Methods
In August–September 2008, serum samples were col-
lected from 1,128 persons in Xuunshaley (n = 271), Ba-
dahabo (n = 160), and Ceel-Bardaale (n = 697) (Table 2).
In the 3 months before the survey, 19 persons reported
having traveled to areas that are known to have higher
malaria endemicity for a median of 4 (IQR 2–20) days.
Persons who reported traveling to areas highly endemic
for malaria were more likely to have a positive response
to P. falciparum (odds ratio [OR] 2.62, 95% confi dence
interval [CI] 0.98–7.01, p = 0.054) but not to P. vivax
(OR 1.18, 95% CI 0.42–3.32, p = 0.75), after adjustment
for age and village of residence. These 19 persons were
excluded from further analyses.
All antigens tested showed a clear increase in sero-
prevalence with a person’s age (Figure 1). The data did not
suggest a recent reduction in malaria transmission inten-
sity (14). The EIR for P. falciparum based on seroconver-
394 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 16, No. 3, March 2010

Page 4

Serologic Markers for Detecting Malaria
sion rates for MSP-119 and AMA-1 (11) was <0.1 infec-
tious bites/person/year. When MSP-119 and AMA-1 data
were combined, 17.9% (179/1,001) of the persons tested
showed reactivity against P. falciparum (i.e., had antibod-
ies against P. falciparum MSP-119, AMA-1, or both) and
19.3% (202/1,044) against P. vivax. There was a signifi cant
positive association between reactivity against P. falcipar-
um and P. vivax (p<0.001). However, only 39.8% (66/166)
of persons with antibodies against P. falciparum also re-
sponded against P. vivax antigens, and there was no appar-
ent correlation between antibody titers against antigens of
the 2 malaria species (p>0.58).
Spatial Patterns in Seroreactivity
P. falciparum antibody prevalence was 9.4% (23/244)
in Xuunshaley, 21.7% (30/138) in Badahabo (p = 0.001),
and 20.4% (126/619) in Ceel-Bardaale (p<0.001) (Table
2). P. vivax antibody prevalence was 16.1% (40/248) in
Xuunshaley, 21.0% (31/148) in Badahabo (p = 0.11), and
20.2% (131/648) in Ceel-Bardaale (p = 0.13) (Table 2).
Age-adjusted P. falciparum seroreactivity was signifi -
cantly increased in a cluster of 18 households (108 persons)
in Ceel-Bardaale (p = 0.002) (Figure 2). In Xuunshaley,
there was a small cluster of 6 households (27 persons) with
a higher age-adjusted P. vivax seroreactivity (p = 0.005).
Factors Associated with Seroreactivity
Seroreactivity data were analyzed for villages sepa-
rately because villages were >7 km apart and were therefore
likely to have their own transmission characteristics. In all 3
villages, P. falciparum antibody prevalence increased with
age (Table 3). For Ceel-Bardaale, an independent negative
association was found between P. falciparum antibody re-
sponses and distance to the nearest seasonal river (OR 0.94,
95% CI 0.88–0.99, p = 0.03) after adjustment for age and
correlation between observations from the same household.
Within the group of persons who had a positive antibody
response against P. falciparum, the titer increased with age
in Xuunshaley (β = 1.74, SE = 0.81, p = 0.031) and Ceel-
Bardaale (β = 11.48, SE = 3.49, p = 0.001).
Similar to P. falciparum, P. vivax antibody prevalence
increased with age in all 3 villages (Table 3). For Ceel-
Bardaale, distance to the nearest seasonal river was nega-
tively associated with P. vivax immune response (OR 0.93,
Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 16, No. 3, March 2010 395
Table 1. Characteristics of persons included in cross-sectional survey for Plasmodium falciparum and P. vivax infection, Somalia,
2008*
Characteristic
No. persons
Age, y, median (IQR)
Female
Reported regular bed net use
Reported fever in 14 d preceding survey
Temperature >37.5°C at time of survey
Positive rapid diagnostic test result
Plasmodium falciparum parasite prevalence†
P. vivax parasite prevalence†
*IQR, interquartile range (25th?75th percentile). Values are % (no. positive/no. tested) unless otherwise indicated.
†Determined by screening 200 high-power microscopic fields.
Dry season, Mar–Apr
1,178
17 (6–36)
48.6 (573/1,178)
1.9 (22/1,158)
4.8 (57/1,179)
0.8 (10/1,177)
0 (0/1,173)
0 (0/1,173)
0 (0/1,173)
End of wet season, Aug–Sep
1,128
15 (6–37)
50.8 (573/1,128)
2.2 (25/1,128)
0.6 (7/1,128)
1.1 (12/1,124)
0 (0/1,106)
0 (0/1,106)
0 (0/1,106)
Table 2. Immune responses against Plasmodium falciparum and P. vivax in study participants, by village, Somalia, 2008*
Village†
Badahabo
160
17.5 (5–35)
Characteristic
No. persons
Median age, y (IQR)
P. falciparum immune response
Combined
MSP-1
Xuunshaley
271
20 (7–40)
Ceel-Bardaale
697
13 (6–35)
p value‡
0.04
9.4 (23/244)
5.1 (13/254)
251.2 (155.0–285.3)
4.8 (12/252)
169.1 (137.0–190.9)
21.7 (30/138)
13.4 (19/142)
214.8 (169.8–275.5)
9.2 (13/141)
189.3 (158.4–225.5)
20.4 (126/619)
15.0 (95/634)
248.5 (190.5–397.2)
8.9 (58/653)
233.7 (170.7–487.0)
<0.001
<0.001
0.12
0.02
0.12
AMA-1
P. vivax immune response
Combined
MSP-1
16.1 (40/248)
11.9 (30/252)
333.9 (271.4–463.9)
5.2 (13/252)
151.5 (146.0–202.1)
21.0 (31/148)
13.9 (21/151)
342.4 (280.1–374.8)
7.4 (11/149)
183.5 (155.7–275.7)
20.2 (131/648)
10.4 (58/648)
291.5 (248.7–393.3)
12.8 (85/665)
227.1 (184.6–390.8)
0.33
0.33
0.39
0.001
0.10
AMA-1
*IQR, interquartile range (25th–75th percentile); MSP-1, merozoite surface protein 1; AMA-1, apical membrane antigen 1.
†Values are % prevalence (no. positive/no. tested) or approximate median titer (IQR) only for seropositive persons unless otherwise indicated.
‡Adjusted for age and correlations between observations from the same household, when applicable.

Page 5

RESEARCH
95% CI 0.87–0.99, p = 0.02) after adjustment for age and
correlation between observations from the same household.
P. vivax antibody titer did not increase with age or any oth-
er factor in those persons who were seropositive. House-
hold factors, socioeconomic factors, distance to the nearest
livestock enclosure, and use of mosquito netting were not
independently associated with immune responses against
P. falciparum or P. vivax.
Although seroprevalence and antibody titers were
higher in older age groups, seroreactivity was also observed
in young children. P. falciparum antibodies were detected
in 22 children <5 years of age (median titer 216.5, IQR
173.2–248.5); 10 had antibodies against P. vivax (median
titer 220.1, IQR 190.4–262.4), and 2 of these children had
antibodies against P. falciparum and P. vivax. Thirty chil-
dren <5 years of age who responded to malaria antigens
were from all 3 villages (3 from Xuunshaley, 7 from Bada-
habo, and 20 from Ceel-Bardaale). Travel to areas in which
malaria was highly endemic in the past 3 months was not
reported for any of these children with antibodies against P.
vivax, P. falciparum, or both. In children <5 years of age, a
response against P. falciparum antigens was not related to
a response against P. vivax antigens (p = 0.30).
Discussion
We showed that serologic markers can be used to de-
tect heterogeneity in malaria transmission in the Gebiley
District of Somalia where malaria transmission occurs at
levels too low to be detected by microscopy. None of the
slides or rapid diagnostic tests showed parasite carriage in
the population, and MSP-119 and AMA-1 seroprevalence
data showed a clear increase in seroreactivity with age and
evidence for variation in exposure to malaria between and
within villages.
Malaria is perceived as a public health problem in the
study area (22), and the 2 slide-confi rmed malaria cases con-
fi rm local clinical malaria episodes. Malaria transmission
in the Gebiley District could not be confi rmed by micros-
copy or RDT in 2 large cross-sectional surveys in the gen-
eral population. However, our serologic fi ndings confi rmed
the occurrence of malaria transmission in the area. Using a
validated model to relate age-specifi c seroconversion rates
to EIR (11), we estimated that P. falciparum transmission
intensity in this area in Somalia was low (EIR <0.1 infec-
tious bites/person/year). Because of the longevity of anti-
body responses, this estimate should be interpreted as an
average EIR experienced over several years. The low EIR
appeared to be supported by examination of breeding sites
at the end of the wet season, which confi rmed the presence
of malaria vectors at a low density. We did not directly de-
termine the EIR by sampling adult mosquitoes because the
low density of mosquitoes would have required intensive
sampling over different seasons (6,23).
Serologic data showed a clear age-dependency in
malaria-specifi c immune responses, which suggested ex-
posure-driven age acquisition of antibody response. Once
acquired, antibody responses to MSP-119 and AMA-1 will
persist for several years (L.C. Okell, unpub. data) (12), and
the rate of acquisition in younger age groups is therefore
critical for determining current malaria transmission in-
tensity. The maximum seroprevalence for individual ma-
laria antigens did not exceed 25% in the oldest age groups,
which is comparable to areas of low malaria endemicity in
northeastern Tanzania (12).
Because of the longevity of antibody responses, sero-
reactivity may not necessarily be the result of recent ex-
posure or exposure in the study area (11,24–26). Consid-
erable changes in transmission intensity in the study area
would have been detected by the model (14). However,
especially in adults, exposure to parasites earlier in life and
a history of traveling to malaria-endemic areas can obscure
immune responses resulting from recent local transmission
(25). Our data indicate that although antibodies may have
been acquired outside the study area, ongoing local ma-
laria transmission at a low intensity is likely. Elimination
of false-positive results to reliably detect low-level local
malaria transmission is necessary.
Cross-reactivity between immune responses to malaria
and other parasites have been reported (27,28) but are ex-
pected to be more pronounced when whole parasite extract
is used instead of recombinant proteins representing single
antigens. The chance of cross-reactive antibody responses
may be minimized by using sera at a minimum dilution of
1:80 (27). Our serum samples were tested at considerably
higher dilutions and we observed no relation in antibody
titers between the homologous antigens of P. falciparum
and P. vivax. Moreover, our method for calculating sero-
positivity derives its seronegative population from within
396 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 16, No. 3, March 2010
Table 3. Factors associated with Plasmodium falciparum or P. vivax seroprevalence in 3 villages, Somalia, 2008*
P. falciparumP. vivax
Village
Xuunshaley
Badahabo
Ceel-Bardaale
Factor
Age
Age
Age
OR (95% CI)
1.02 (1.00–1.04)
1.03 (1.01–1.05)
1.03 (1.02–1.04)
0.94 (0.88–0.99)
p value
0.029
0.002
<0.001
0.028
OR (95% CI)
1.04 (1.02–1.06)
1.03 (1.01–1.05)
1.03 (1.02–1.04)
0.93 (0.88–0.99)
p value
<0.001
0.006
<0.001
0.016 Distance to river†
*OR, odds ratio; CI, confidence interval. Estimates are adjusted for correlation between observations from the same household.
†Nearest seasonal river.