This document summarizes a study on the serological profile and validity of ELISA testing for Schistosoma haematobium infection in Basrah, Iraq. Key findings include:
- Antibody levels increased with age up to 15 years then plateaued, with no significant differences between males and females except in the 15-24 age group.
- Those positive for current infection by urine examination had significantly higher antibody levels and seropositivity compared to those negative by urine examination. ELISA detected infection in 83% of urine-positive individuals.
- For age groups 6-14 and 15-24, antibody levels and seropositivity were significantly higher in those reporting swimming compared to those who did not
The epidemiology of schistosomiasis the Basrah study 2 serological profile.pdf
1. 460
TRANSACTIONS
OFTHEROYAI.
SOCIETY
OFTROPICAL
MEDICINE
ANDHYGIENE
(1987) 81, 460467
The epidemiology of schistosomiasis in the later stages of a
control programme based on chemotherapy: the Basrah study.
2. The serological profile and the validity of the ELISA in
seroepidemiological studies
ALIM YACOUB,B. A. SOUTHGATE
AND JANE E. LILLYWHITE
Department of Tropical Hygiene, London School of Hygiene and Tropical Medicine, Keppel Street, London,
WClE 7HT
Abstract
be
An epidemiological study ofSchistosoma haematobium infection in Al-maadan locality, consideredto
the main remaining endemic focus in Basrah, southern Iraq was carried out. The association
between the serological profile of the population, measuredby the enzyme linked immunosorbent
assay(ELISA), and various factors including current infection by urine examination, water contact
pattern, past history of, and treatment for, schistosomiasis,and cercarial dermatitis wasinvestigated.
Further study of the serological data by the relative operating characteristic (ROC) analysis to assess
the validity of the ELISA in detecting schistosomalinfection, showed that current infection, past
history of infection, age,cercarial dermatitis and categoryof household were significantly associated
with the serological profile of the population. The analysisallowed quantification of the effectsof past
history of infection and cercarial dermatitis on the validity of the ELISA in detecting schistosomiasis.
Introduction
The application of serological techniques in
epidemiological surveys has contributed significantly
to our understanding of the natural historv of many
parasitic diseasesand their patterns of dis&bution in
the communitv (LOBEL & KAGAN. 1978: LUCAS.
1976). Until aGo& 10 years ago, ser&pidemiological
studies had a limited use in schistosomiasisbecause
the serological testsemployed lacked sufficient sensi-
tivity or specificity or were technically difficult to
apply in the field (for critical reviews of the subjectsee
KAGAN& PELLEGRINO,
1961;KAGAN,1982;HOUBA,
19801.
-- --,.
The development of the enzyme linked immuno-
sorbent assav (ELISA) and its anolication for the
serodiagnosig df schisiosomiasis (ENGVAL & PERL-
MANN, 1972;HULDT et al., 1975),with the ability to
processlarge numbers of sampleswith small amounts
of reagents.,has stimulated researchto assess
the test
asa screenmgtool in epidemiological studies (FARAG
& BARAKAT, 1978; SCHINSKIet al., 1978; JANITS-
CHKEet al:, 1981;MCLARENet al., 1978, 1979).The
use of purfied eggantigens hasenhanced further the
validity of the ELISA in the detection of Schistosoma
munsoni infection (MCLAREN et al., 1981).
HILLYER et al. (19791 and LONG et al. (1981)
carried out populati& based studies to evaluaie thi
ELISA. in comoarison with other serodiaenostic
tests,in detect&S. mansoniinfection in PuertāoRico
and St Lucia respectively. YOGORE
et al. (1983, 1984)
and LEWERTet al. (1984) reported the results of an
elegant seroepidemiological study of S. japonicum in
the Philippines; they concluded that the ELISA could
provide a more accurate measurement than stool
Correspondence
to be addressed
to: Dr B. A. Southgate,
Department of Tropical Hygiene, London School of
Hygieneand Tropical Medicine, Keppel Street,London,
WClE 7HT.
examination of the prevalence and incidence of S.
japonicum infection and recommended its use for
monitoring the impact of the control programme
already in operation.
However, there are still the following important
gaps in our knowledge of the seroepidemiology of
schistosomiasis.(1) Studies should not be limited to
obtaining serological and parasitological data but
should collect information on other factorsrelevant to
exposure to the infection (LOBEL & KAGAN, 1978).
For example, although pasthistory of schistosomiasis
and exposure to non-human schistosomeshave long
been known to influence the interpretation of serolo-
gical data, the extent of their effect has not been
quantified mainly becauseof lack of epidemiological
information on their distribution.
(2) Evaluation of the validity of any serological test
hasbeenbasedmainly on its sensitivity and specificity
with respect to a reference parasitological test (the J
index used in some studies is derived from these 2
measures). One of the main limitations of such an
approach is the need to selecta cut-off point. Several
methods have been suggested,none entirely satisfac-
tory (DE SAVIGNY& VOLLER, 1980), and there is
always an element of arbitrariness in such selection;
there is a need to apply a statistical or mathematical
technique which allows evaluation of the validity of
the test across different cut-off points.
(3) Seroepideniiological studies of S. haematobium
infection in areas where this parasite is the only
prevalent human schistosomehavenever beencarried
out. The lack of a standardized high quality homolo-
gous antigen which can be produced in sufficient
amounts to be used in a field study has been an
imnortant deterrent, significant crossreactionshavine:
be& reported with het&ologous S. mansonicrude egg
or worm antigens (ISMAIL, 1980; MCLAREN et al.,
1978; TATICHEFF & MELAMED, 1983).
2. A. YACOUB et al. 461
To explore someof these issues, we carried out a
seroepidemiological study of S. haematobium infection
in Basrah, southern Iraq, with the following objec-
tives.
(1) Description of the serological profile of S.
huemutobium infection in an endemic community in
Basrah in relation to age and sex, water contact
activities, current infection as detected by urine
examination, cercarial dermatitis, category of the
household, and evidence of previous schktosomiasis.
(2) Assessment of the validitv of the ELISA in
detecting S. haetnatobium infe&on in the field by
using relative operating characteristic (ROC) analysis.
ROC analysis allows assessmentto be carried out
across different cut-off points.
Materials and Methods
Study population
S. haemutobium is the only prevalent human schistosome
in southern Iraq; a control programme based on screening
primary schoolchildren by urine examination and treating
those found infected, and the application of molluscicides,
was launched in 1953. Reports of the Endemic Diseases
Centre of Basrah indicated that a very low prevalence of
infection had been attained among schoolchildren in the
endemic areas. A more recent survey of schoolchildren in
different areas of Basrah (YACOUB, 1985) showed that
infection existed mainly in Al-maadan locality, 30 km north
of Basrahcitv. Detailed descriotions of the studv nooulation
and the house-to-housesurve);sare given in pa& i of this
series(YACOUB
& SOUTHGATE,
1987).After interviewing all
members of a household, blood sampleswere collected by
finger prick on Whatman No. 3 filter paper, dried, and kept
in plastic bagsat -20Ā°C during the period of the field work.
The households were divided into Categories 1 and 2,
describedby YACOUB & SOUTHGATE (1987).In this paper we
report the effect of this categorization on the serological
profile of the community.
Laboratory procedures
Urine sampleswere examined asdescribed by YACOUB &
SOUTHGATE (1987).
The ELISA test, employing S. munsoni soluble egg
antigen, wasbasedon the method described by MCLARENet
al. (1978) with some modifications. Bound antibody was
measuredusing a peroxidase-labelled polyvalent anti-human
IgG conjugate (Dako Ltd) and orthophenylene diamine/
HzOz assubstrate. The reaction wasstopped and read on a
Multiskan photometer (Flow Laboratories) at 492 mn when
the absorbanceof the reference positive reached a value of
l%lO.A cut-off value of 0.40 wasselectedasgiving the best
discrimination between positives and negatives based on
testing random blood samplesfrom individuals with known
narasitoloaical results.
Statistical analyses
Geometric mean absorbance values (A 492nm) and the
proportions of seropositiveswere related to eachvariable in
the study using the &i-squared test with continuity correc-
tion, the calculation of standardized normal deviates(SND),
or analysis of variance (F test).
To evaluate the relative contribution of each variable
taking into account the effect of others, multiple regression
analysis was performed using the ELISA readings (log,,
transformed) as the dependent variable while age, sex,
current infection, water contact activities, history of haema-
turia, treatment for schistosomiasis,cercarial dermatitis, and
the category of the household were introduced as the
independent variables. Age was introduced as a dummy
variable; a computer program based on the statistical
package for the social sciences(SPSS), using the stepwise
method, was employed for the analysis.
ROC analysis
ROC analysis, originally developed in radar studiesto
separate observer variability from innate detectability of a
signal, wasapplied in the evaluation of screeningtestsfor the
first ume by ERDREICH & LEE(1981).It allows assessment
of
the discriminating ability of a test across different cut-off
points; we report its first application for the evaluation of the
validity of a serodiagnostic test for parasitic infections.
ROC analysis consists essentially of 2 components, the
construction of the ROC curve and the calculation of various
detectability measures. The ROC curve is constructed by
plotting the proportion of true positives detected (i.e.,
sensitivity) on the Y axis of a graph, and the proportion of
falsepositives (i.e., 1-specificity) on the X axis, for different
cut-off points. The 45ā slope diagonal, or āchance lineā,
results if the test is unable to discriminate between true and
false positives. The higher the actual curve lies above the
chanceline, the better is the testās discriminant ability. The
point on the curve at the maximum perpendicular distance
from the chanceline representsthe optimum cut-off point.
The areaunder the curve provides a quantitative measureof
the discriminant ability of the test and can be estimated by
the maximum likelihood method. A computer program
developedby DORFMAN & ALF (1969)wasusedfor our data.
The area under the curve ranges from 0.5 (for zero
detectability) to 1.0 (for perfect detectability).
ROC analysis was applied to evaluate the ELISA in
detecting infection in the following groups of individuals.
Group 1: thosewith current infection by urine examination;
group 2: individuals with current infection and past history
of schistosomiasis,or both; and group 3: individuals with
current infection, past history of schistosomiasis,history of
cercarial dermatitis, or any 2 or 3 of these features.
Household visits
Results
261 individuals were surveyed and have been
described by YACOUB & SOUTHGATE (1987); 246
useable blood samples were collected.
Serological profile and study variables
The results of the detailed analysescarried out to
explore the associationbetweenthe absorbancevalues
and eachof the study variablesaregiven in Tables 1to
10. They are briefly summarized below.
Serology, age and sex
There was in general an increase in the level of
antibodies up to 15yearsold; afterwards aplateauwas
reached. There was no sirmificant different between
malesand femalesexceptii the 15-24year agegroup.
The samepattern was observedwhether the geomet-
ric mean (Table 1) or the proportion of seropositives
(Table 2) was used.
Serology and current infection
Table 3 shows that the geometric means of the
absorbancevalues and the proportion of seropositives
among those positive by urine examination were
significantly higher than those found negative by
urine examination. About 83% of those wsitive bv
urine examination had values above the selected
cut-off noint (i.e.. the sensitivitv of the ELISA) while
46% of those negative by urine examination had
values below O-4(specificity of the test). The preva-
lence ratesby both serology and parasitology for each
age group are shown in Table 4. About 15% of the
population were positive by parasitology comparedto
57% positive by serology. The greatest discrepancy
wasobservedamongthe 25yearsand aboveagegroup
of the population.
3. 462 EPIDEMIOLOGY OF SCHISTOSOMIASIS IN BASRAH. 2.
Table l-Geometric means and 95% confidence intervals of the
absorbance values with respect to age and sex
Age
(years)
O-5
6-10
11-14
15-24
25-44
45+
95%
Geometric confidence
Sex Number mean interval
males 22 0.19 0.15-0.24
females 15 0.24 0.17-0.34
total 37 0.21 0.17-0.25
males 31 0.36 0.28-0.46
females 21 0.46 0.36-059
total 52 0.40 0.33-0.48
males 28 0.59 0.50-0.70
females 25 0.50 0.41-0.61
total 53 0.55 0.48-0.63
males 21 0.68 0.62-0.75
females 27 0.34 0.27-0.43
total 48 046 0.39-0.54
males 7 0.42 0.25-0.69
females 23 0.41 0.33-0.50
total 30 0.41 0.33-0.50
males 7 0.55 0.43-0.70
females 19 0.42 0.33-0.53
total 26 0.45 0.37-0.54
F=13.38, P<O.OOl
Serology and water contact pattern
Earlier analysis (not presented in this paper)
showedthat swimming wasclosely associatedwith age
and that it wasthe 6-14year agegroup which reported
the maximum proportion of those who swam during
the summer prior to the survey. The geometric mean
of the absorbance values and the proportion of
seropositives with respect to swimming for each age
group areshown in Tables 5 and 6 respectively. It can
be seenthat there were significant differences in both
quantities betweenthosewho reported swimming and
those who did not, for the 6-14 and 15-24year age
groups.
Since bathing, fetching water and washing clothes
were reported mainly by individuals above 5 years
old.,and sincethere wasno variation in reporting such
activities within the remaining age groups, the
analysiswith respectto serologywascarried out for all
agegroups excluding the first 5years.Using either the
geometric mean or the proportion of seropositives,
there wasno significant difference betweenthosewho
reported any of theseactivities and thosewho did not
(Table 7).
Serology and history of haematuria, treannent fm schisto-
somiasis and cexarial dermatitis
In general, those who reported a history of haema-
turia or treatment for schistosomiasishad significantly
Table 2-Distribution of seropositives with respect to age and sex
Age
(years)
o-5
6-10
11-14
15-24
25-44
45+
Total
Males
Number Number
tested positive (%)
:: 3 (13.6)
16 (51.6)
2 tzl g:;i
7
;t I;:::i
11: 72 (62.1)
x*=39.7, P<O.OOl
Females Total
NUllher Number Number Number
tested positive(%) tested positive (%)
15 3 (20.0) 6 (16.2)
2 :; ~$i:ij
;:
:; t:;:ii
I: :i i:::ij
:i
:: Ig:i
1::
12 (63.0)
::
18 (69.2)
69 (53.1) 246 141
x*=16.09,
(57.3)
P<O.Ol x*=38.25, P<O.OOl
Table 3-Geometric means (GM) of absorbance values and 95%
confidence intervals, and the proportion of seropositives, with
respect to current infection
Urine
examination
Positive
Negative
Total
Number GM
36 0.59
198 0.40
234 0.42
95%
Confidence
interval
0.50-0.70
0.37-044
0.39-0.45
% Sero-
positive
83.3
54.5
59.0
Table 4-Degree of discrepancy between parasitological and
serological results for different age groups of Al-maadan popula-
tion
Age % Positive by * Positive by Index of
(y=rs) parasitology serology discrepancy*
o-5 2.6 16.2 83.9%
6-14 24.8 67.6 63.3%
15-24 16.7 62.5 73.3%
25+ 3.7 60.7 93.9%
Total 15.3 57.3 73.3%
*Calculated from the formula (proportion positive by serology -
proportion positive by parasitology) x lOO/proportion
positive by serology
higher levels of antibodies than those who did not
(Table 8). The geometric mean of the absorbance
values of thosewho reported shara (cercarial dermati-
tis) was significantly higher than the geometric mean
of those who did not (Table 9). However, among
thosepositive by parasitology, there wasno significant
difference between thosewho reported this condition
and those who did not, while a significant difference
occurred when the analysiswasconfined to thosewho
were negative by urine examination.
Serology and the category of the household
The geometric mean of the absorbance values of
membersof householdsof category 1wassignificantly
lower than the geometric mean of members of
households of category 2 (SND = 2.02, 0.01
<P<O*O5) (see Table 10). However, the difference
between the two categories with respect to the
proportion of seropositives was not significant
(SND = 1.611, 0.10 <P<O*25).
Results of the multiple regression analysis
Sincemany of the study variables were interrelated,
it was necessary to disentangle the association be-
4. A. YACOUB et cd. 463
Table S-Geometric means (GM) and 95%confidence intervals (CI) of the absorbance values with respect to
swimming for each age group
Age Swimming
(years) No. GM (95% CI) No.
Not swimming
GM (95% CI)
o-5 20 0.25 (O-19-0.33) 17 0.17
6-14* 0.51
(0*14-0.21)
80 (0.45-0.58) 0.34
E-24*
(0.27-0.43)
19 0.70 (0.62-0.78) zz 0.35 (0.28-0.43)
25+
Total 12:
0.35 (0.21-0.60) 0.44 (O-38-0.51)
0.47 (0.42-0.52) 15: 0.35
*Significant difference at 0.05 level
(0.31-0.39)
Table bPropmtion of seropositives with respect to age among
those who reported swimming and those who did not
Age Swimming Not Swimming
(years) No % Positives No. % Positives x2 P
O-5 20 20.0 17 5.9 * 0.45
6-14 80 73.8 25 48.9 4.7 0.05 >P>O.O2
15-24 19 94.7 29 41.4 * >O,OOl
25+ 4 50.0 52 61.5 * 0.99
Total 123 68.3 123 46.3 11.2 >O.OOl
*Fisherās exact test (two-tailed)
tween each of them and the serological data by
controlling for the others. The significant results of
the multiple regression analysis are shown in Table
11; history of haematuria, age, current infection,
category of the household, and cercarial dermatitis
were significantly associated with the serological
profile of the Al-maadan population. These factors
contributed 27% of the variance and the multiple
correlation coefficient was 0.52. The negative sign
attachedto the regressioncoefficient of the categoryof
the household denoted that individuals in the second
Table 7-Geometric means (GM) and 95% confidence intervals (CI) of absorbance values and proportions of
seropositives with respect to washing clothes, bathing and fetching water
Activity No. GM (95% CI) % Positive x2 P
Bathing
Yes
No
Total
WY:? clothes
No
Total
Fetching water
Yes
No
Total
33 0.51 (0.41-O-63)
176 044 (0.41-0.48)
209 0.46 (O-42-0.49)
1:;
209
1:;
209
0.43 (0.38-O-50) 61.3 0.24 0~75>P>o*50
0.46 (0.42-0.51) 66.0
0.46 (0.42-O-49) 64.6
0.44 (0.37-0.52)
0.46 (0*42-0.50)
0.46 (O-43-0.49)
75.8 1.60 0.25>2'>0* 10
62.5
64.6
61.7 0.09 0.9O>P>O.75
65.4
64.6
Table g-Geometric means (GM) and 95% conhdence intervals (CI) and proportions of seropositives with
respect to history of haematuria and treatment for schistosomiasis
Category
History of haematuria
Yes
No
Total
History of treatment
Yes
No
Total
No. GM
65 0.67
181 0.34
246 0.41
50 0.69
194 0.35
244 o-41
(95% CI)
(O-61-0.73)
(0.31-0.37)
(0.38-0.44)
(0.64-0.76)
(0.32-0.39)
(0.38-044)
% Positive x2 P
93.8 46.2 <O*OOl
44.2
57.3
96.0 37.1 <o.oo 1
46.9
57.0
5. 464 EPIDEMIOLOGY OF SCHISTOSOMIASIS IN BASRAH. 2.
Table 9-Serology and cetcarial dermatitis: geometric means (GM) of absorbance values of individuals with
respect to current infection and cercarial dermatitis
Category Number
With dermatitis
1:;
Without dermatitis
Total 209
SNDt = 4.78, P<O*OOl
For those infected with S. haematobium
With dermatitis
Without dermatitis :i
Total
SND = 1.61, P =3?l-11
For those not infected with S. haematobium
With dermatitis
;;
Without dermatitis
Total 170
SND = 3.88, P<O*OOl
Results of urine examination were not available from 4 individuals
GM (95% CI)*
0.56 (O-SO-0.62)
0.39 (0-35-0-43)
0.46 (O-42-0.50)
0.68 (055-0.83)
0.49 (O-38-0.63)
0.59 (0~50-0~70)
O-53 (0.47-0.60)
0.38 (0.34-0.43)
0.44 (0.40-0.48)
*CI = confidence intervals
j-SND = standardized normal deviate
Table lO-Geome.tric means (GM) of absorbance values and the
proportions
of sempositives with respect to the category of the
household
significantly associatedwith the level of antibodies in
the multiple regression analysis.
Category NUdXr GM (95% Cl)* % seropositive
The results of the ROC analysis
The Figure showsthe ROC curves for the 3 groups
1 80 0.36 (0.31-0.42) 50.0
2 166 0.43 (0.39-0.47) 60.8
of individuals; closed circles represent the 0.4 cut-off
Total 246 0.41 (0.38-044) 57.3
point, other cut-off points arelabelled. It can be seen
that all 3 curves were above the chance line, i.e. the
l CI = confidence intervals ELISA can discriminate between those with current
Table 11-Significant results of multiple regression analysis using logre ELBA as the dependent variable and
history of haematuria, treatment for schistosomiasis, age, sex, water contact activities, cercarial dermatitis,
and the category of household as the independent variables
Variable
History of
haematuria
b* SE(b)*
O-18 0.04
P* t P
0.33 4.8 <O*OOl
Cercarial
dermatitis
0.11 0.04 o-22 3.2 LO.01
Category of
household
-0.08 0.03 -0.15 -2.4 co.05
Current
infection
0.11 0.04 0.16 2.5 <o-o5
Age 0.05 o-02 0.15
Constant -0.56 0.07
Multiple r = 0.52, SE(r) = 0.22, ?=0*27
*b=regression coefficient, SE = standard error, p = standardized b
2.2 <0*05
-8.1 <o-o01
category contributed positively to the antibody level
relative to the first category. The standardized coef-
ficient provides an estimate of the degree of the
contribution of each independent variable to the
explained variance. Sex, water contact activity, and
history of treatment for schistosomiasis were not
infection and those negative by urine examination;
between thosewith current infection or apast history
of haematuria, or both, and those who were negative
by urine examination and did not report such history;
and finally between those with current infection or a
past history of either haematuria or cercarial dermati-
6. A. YACOUB t?Cal. 465
Table U-The area under the curve (Fig.) and its standard
deviation for each category of individual tested by ELISA for S.
hamurtobium
infection
Grout
Standard
Area deviation
Individuals with current infection
Individualswith current infection
or pastexposure
0.70 00478
0.82 0.0302
Individualswith current infectionor
pastexposureand cercarialdermatitis
0.88 0.0283
0.2 0.4 0.6 0.8 1
PROPORTION OF FALSE POSITIVES
- Individuals with current lnfectlon
- - - - lndlvldualr wlth current lnfectlon and past
history of schlstosomlarls
+e.*. Indlvlduals wlth current infectlon, past history
and cercarlal dermatitis
Fig. The applicationof the relativeoperatingcharacteristic
(ROC)
analysisto serologicaldata of Al-maadanpopulation.
Theclosed
circlesindicatetheselected
cut-offpointusedin thestudy
(W4).Opencirclesindicatetheothercut-offpointsinvestigated
inthe
study;valuesfrom abovedownwards
followthe sameorderfor all
three groupsof individuals.
tis, or both, and those negative by urine examination
who did not report a history of either condition. The
heights of the 3 curves show that the ELISA has a
better discriminating ability when those who gave a
history of haematuria were considered positive in
addition to thosewho were positive by urine examina-
tion. Quantitatively, this observation is expressedby
the values of the areaunder each of the 3 curves, as
shown in Table 12. The Table alsoshowsestimatesof
the standard deviations for the performanceof testsof
signiiicance. There was a siani&ant difference be-
t&en the areas under the &rves for the first and
secondgroups (SND=2*12, O*Ol<P<O*OS).Howev-
er, the diffecence between the second and the third
groups was not significant (SND = 1.45, 0.10~
P<O*20).
In summary, although the test discriminated be-
tween those with current infection and those who
were negative by urine examination (area=0*7), its
discriminating ability wasfurther enhancedby giving
consideration to the variable pasthistory of haematur-
ia (area=0.82). Although the inclusion of those with
cercarial dermatitis increased the area to O-88, this
increase was not significant.
Discussion
In the present study S. mansoniegg antigen was
used to screenfor S. haematobium. Since the latter is
the only endemic human schistosome in southern
Iraq, crossreactionsdue to other human schistosomes
can be excluded. The sensitivity of the ELISA in
detecting S. hematobium usink S. matsoni egg
antigen was 83%. This firmre is comnarable to that
rep&ted by MCLAREN e?al. (19783 using similar
antigen to test Egyptian children and adults infected
with S. haemutobium (83% and 76%). JANITSCHKE et
al. (1981) reported a sensitivity of 86% when S.
mansoni worm antigen was used to test for S.
haematobium cases.ISMAIL(1980) tested sera from
Egyptian casesof S. haemutobium and S. man.&. He
showed that egg antigens of both parasites were
equally reactive againstthe samesetof sera.Very few
crossreactionswere reported from thesestudieswhen
serafrom individuals with intestinal nematodeswere
tested. Cross reactions were reported, however, by
HILLYER & GOMEZ DE RIO (1979) with trichinosis.
cysticercosisand fascioliasis. ākāhes; conditions could
not affect our results since the first 2 do not exist in
southern Iraq and human fascioliasis is very rare.
The ROC analvsis revealed some interesting
findings; it showed<thatthe ELISA was of adequati
validity when used to discriminate between indi-
viduals with respectto infection status asdetectedby
urine examination. The discriminant ability of the test
wasquantified by measuring the areaunder the ROC
curve (O-7).This is a summary index of the perform-
ance of the test across various cut-off points. The
ROC curve also showed that the optimum cut-off
point with respect to current infection was 0.4 (the
point of the highest perpendicular distance from the
chanceline). This wasin fact the cut-off point usedin
this study to calculate the sensitivity iu;d specificity
firmres. However. ROC analvsis allows flexibilitv in
&e choiceof cut-hff points depending on the extent to
which all positive casesneed to be identified and the
implicatitins of such choicefor the health caresystem.
ROC analysis can thus be considered as a powerful
tool in seroepidemiologicalstudies of schistosomiasis.
An attempt wasmadeto quantify the effect of apast
history of infection and cercarial dermatitis, using
ROC analysis, on the validity of the ELISA in
detecting schlstosomiasis. By including those indi-
viduals with a past history of infection as positive
cases, the area under the curve was significantly
increased to 0.82; when cercarial dermatitis was
included asavariable, the increasein the discriminant
ability of the test was not significant.
The specificity of the ELISA in detecting current
infection was 46%, calculated from those negative by
parasitology in the Al-maadan population. This figure
is low compared to those reported by LONG et al.
(1981)and HILLYER etal. (1979)using the ELISA to
detect S. man.& infection in St Lucia and Puerto
Rico, who recorded specificity figures of 75% and
92%. MOTT & DIXON (1982)reported figures of 97%
7. 466 EPIDEMIOLOGY OF SCHISTOSOMIASISIN BASRAH. 2.
and 90% using crude egg antigen to test European and
Amazon Indian sera from persons who had never been
to endemic areas. However, they reported a 47%
specificity figure using the same antigen to test
post-treatment sera from residents in S. mansoni
areas.
The variation in the reported specificity figures
from different studies (including the present one) can
be attributed to the following factors.
(1) Variation in the choice of subjects to determine the
specificity of the test. Higher figures are likely to
be obtained when testing sera from individuals
living in non-endemic areas.
(2) The presence and nature of schistosomiasis control
programmes in the endemic areas. Individuals
who had had the infection and who had been
treated were usually positive serologically even if
they were parasitologically negative (see Table 8).
Since a control programme based on selective
chemotherapy has been in operation in Basrah for
many years, a significant proportion of positive
reactions could be attributed to the persistence of
antibodies after successful cure.
(3) The sensitivity of the parasitological test used as
the reference test for calculating specificity. For
example, YOGORE et al. (1983) showed that a
single stool examination reduced the underesti-
mate to 29%. This factor might have operated in
the present study, and is discussed by YACOUB &
SOUTHGATE (1987).
(4) Variation in the frequency of cercarial dermatitis
(whether caused by human or non-human schisto-
somes), and of unisexual and unmated infections.
Since serological data reflect the accumulated
experience of a parasitic infection by individuals in a
community, the value of such data can be enhanced if
they are age related (DRAPER et al., 1972; LOBEL &
KAGAN, 1978). Our study showed that age was one of
the significant factors associated with the serological
profile of schistosomiasis in the Al-maadan commun-
ity (Table 11). However, it has also been shown that
factors apart from current infection and age might be
equally or more important when attempting to
interpret the antibody profile of this community; past
history of infection and cercarial dermatitis were
significantly associated with the profile. In endemic
areas of Basrah the causative organism of cercarial
dermatitis (shara) is not known, and it may be caused
by cercariae of S. haematobium. The only investiga-
tion in these areas was made by WATSON & NAJIM
(1956), who concluded that true cercarial dermatitis
from non-human schistosomes was rare in southern
Iraq and attributed the cases they examined to S.
haematobium.
The finding that category of household was a
significant factor associated with the serological pro-
file indicates that the distribution of antibodies is
non-random and that there is a degree of aggregation
in the 2 categories of households studied. The degree
of clusteringāor non-random distribution can alsoā be
investigated using such parameters as the distance of
the household from the source of water or the
distributian of infection at individual household level.
Studies carried out by KLUMPP (1983), using para-
sitoloeical data. and NOSENAS et al. (1975). using
both -serological and parasitological data, āprovidi
examples.
Swimming was the only water contact activity
associated with positive serology (Tables 5, 6, 7);
however, none of tiiese activities contributed signi-
ficantly to the antibody level when other factors were
considered in the multiple regression analysis. This
might be explained by the fact that serology measures
the life-time experience of an individual, while
infection detected by parasitology reflects variation in
recent exposure through water contact.
In conclusion, the results of this seroepidemiologic-
al survey demonstrate the relative importance of
variables, in addition to parasitological data, when
attempting to interpret the serological profile of a
community exposed to schistosomal infection.
Acknowledgements
We thank all staff at the Endemic DiseasesCentre. Basrah
for their help during the field work. We are very grateful to
all schoolchildren and inhabitants of Al-maadan locality, for
without their cooperation this study would not have been
possible. Thanks also go to Dr C. C. Draper for allowing us
io process the blood samples at the -Seroepidemioibgy
Laboratory of the London School of Hygiene and Tropical
Medicine and for his advice on certain technical issues.The
computer program for the ROC analysiswaskindly provided
by the Institute of Psychiatry, London.
Part of the material in this paper wasincluded in a thesis
by A. Yacoub acceptedfor the award of aPhD degreeby the
University of London.
References
De Savigny, D. & Voller, A. (1980).The communication of
ELISA data from laboratory to clinician. Journal of
Immunoassay, 1, 105-128.
Dorfman. D. D. & Alf. E. (1969X Maximum-likelihood
e&n&ion of parameters oi sign&detection theory and
determination of confidence intervals*-*rating [sic]
method data. Journal of Mathematical Psychology, 6,
487-496.
Draper, C. C., Voller, A. & Carpenter, R. G. (1972). The
epidemiologic interpretation of serologic datain malaria.
~4~47;~ Journal of Tropical Medicine and Hygiene, 21,
El-Gindv. k. (1956). Monthlv orevalence rates of natural
infei&on wiih S. āhaematobiki cercariaein B. truncatus in
central Iraq. Bulletin of Endemic Diseases, 7 (l/2), 1l-3 1.
El-Gindv, M. & Radhawv, I. (1965). Viabilitv, incubation
period and hatchability-of 3. trukatus eggsfrom central
Iraq under outdoor conditions in different seasons.
Bulletin of Endemic Diseases, 7 (3/4), 13-25.
Engvall, E. & Perlmann, P. (1972).Enzyme linked immuno-
sorbent assay (ELISA): III. Quantitation of specific
antibodies by enzyme labelled anti-immunoglobulin in
antigen coated plates. Journal of Immunology, 109,
129-135.
Erdreich, L. & Lee, E. T. (1981). Use of relative operating
characteristic analysisin epidemiology. Ammican3mrd
of Epidemiology, 114, 649-662.
Farag, H. F. & Barakat, R. M. R. (1978). The enzyme
linked immunoassay in the diagnosis of bilharziasis.
Tropenmedizin und Parasitologic, 29, 12-14.
Hillyer, G. V. & Gomez de Rio, I. (1979). The enzyme
linked immunosorbent assay(ELISA) for the immuno-
diagnosisof schistosomiasis.AmericanJournal of Tropical
Medicine and Hygiene, 28, 237-241.
Hillyer, G. V., Ruiz-Tiben, E., Knight, W. B., Gomez de
Rio, I. & Pelley, R. P. (1979). Immunodiagnosis of
infection with S. mansoni: comparison of ELISA,
radioimmunoassay and precipitation tests performed
with antigens from eggs. American Journal of Tropical
Medicine and Hygiene, 28, 661-669.
Houba, V. (1980).Schistosomiasis.In: Immunological Znves-
tigation of Tropical Parasitic Diseases,
Houba, V. (editor).
8. A. YACOUB et al. 467
Edinburgh, London and New York: Churchill Living- of the Royal Societyof Tropical Medicine and Hygiene, 73,
stone, pp. 130-147. 636-639.
Huldt, G., Lagerquist, B., Phillips, T., Draper, C. C. &
Voller, A. (1975). Detection of antibodies in schistoso-
miasis by enzymelinked immunosorbent assay(ELISA).
Annals of Tropical Medicine and Parasitology, 69, 483-
488.
Ismail. M. M. (1980).Observations on someseroloeical testsin
Schistosoma haematobium infections in n&t and ex-
perimental animals. Ph.D. thesis, University of London,
pp. 105122.
McLaren, M. L., Lillywhite, J. E., Dunne, D. W. &
Doenhoff, M. J. (1981). Serodiagnosisof human Schisto-
soma mansoni infections: enhanced sensitivity and spe-
cificity in ELISA using a fraction containing S. mansoni
antigens w, and LX,. Transactions of the Royal Societyof
Tropical Medicine and Hygiene, 75, 72-79.
Mott, K. E. & Dixon, H. (1982). Collaborative study on
antigens for immunodiagnosis of schistosomiasis.Bulle-
tin of the World Health Organization, 60, 729-753.
Nosenas, J. S., Matsuda, H., Blas, B. L., Tanaka, H. &
Santos, A. T., Jr (1975). Evaluation of the circumoval
precipitation test using dried blood on filter paper as a
diagnostic tool in epidemiological survey for schistoso-
miasis. Japanese 3oumal of Experimental Medicine, 45,
367-375.
Janitschke, K., El-Kalouby, A. H., Braun-Munzinger, R.
A., El-Baz, H. & Mahmoud, M. (1981). Evaluation of
the ELISA test as an epidemiological tool in schistoso-
miasis. Journal of Tropical Medicine and Hygiene, 84,
147-154.
Kagan, I. G. (1982). Serodiagnosisof schistosomiasis.In:
Immunoparasitologv: Principles and Methods in Malaria
and Schistosomiasis Research. Strickland, G. T. & Hun-
fiā~ri-l,&, Jr (editors). USA: Praeger Publications, pp.
Kagan, I. G. & Pellegrino, J. (1961). A critical review of
immunological methods for the diagnosisof bilharziasis.
Bulletin of the World Health Organization, 25, 61l-674.
Klumpp, R. K. (1983). A study of the transmission of
Schistosomahaematobium in Volta Lake, Ghana. Ph.D.
thesis, University of London, pp. 238-247and 302-306.
Lewert, R. M., Yogore, M. G., Jr & Bias, B. L. (1984).
Seroepidemiology of schistosomiasisjaponica by ELISA
in the Philinuines. II. Unreliabilitv of stool examination
in the meaāsurementof incidence: American Journal of
Tropical Medicine and Hygiene, 33, 872-881.
Lobe& H. 0. & Kagan, I. G. (1978). Seroepidemiology of
parasitic diseases.Annual Reviews of Microbiology, 32,
329-347.
Long, E. G., McLaren, M., Goddard, M. J., Bartholomew,
R. K., Peters, P. & Goodgame, R. (1981). Comparison
of ELISA, radioimmunoassayand stool examination for
Schistosoma mansoni infection. Transactions of the Royal
Society of Tropical Medicine and Hygiene, 75, 365-371.
Lucas, A. 0. (1976). Surveillance of communicable diseases
in tropical Africa. International Journal of Epidemiology,
5, 39-43.
McLaren, M., Draper, C. C., Roberts, J. M., Minter-
Goedbloed, E., Ligthart, G. S., Teesdale, G. H., Amin,
M. A., Omer, A. H. S., Bartlett, A. & Voller, A. (1978).
Studies on the enzyme linked immunosorbent assay
(ELISA) test for Schistosoma mansoni infections. Annals
of Tropical Medicine and Parasitology, 72, 243-253.
McLaren, M., Long, E. G., Goodgame, R. W. & Lilly-
white, J. E. (1979). Application of the enzyme linked
immunosorbent assay(ELISA) for the serodiagnosisof
Schistosoma mansoniinfections in St Lucia. Transactions
Peters, P. A., Warren, K. S. & Mahmoud, A. A. F. (1976).
Rapid accurate quantification of schistosome eggs via
Nuclepore filters. Journal of Parasitology, 62, 154-155.
Schinski, V. D., Clutter, W. C. & Murrell, K. D. (1976).
Enzyme and ā*sI-labelled anti-immunoglobulin assaysin
the immunodiagnosis of schistosomiasis.American Jour-
nal of Tropical Medicine and Hygiene, 25, 824-831.
Taticheff, S. StMelamed, M. (1983).Evaluation of ELISA -
with S. mansoni egg antigen - in the serodiagnosisof
schistosomiasis.Ethiopian Medical 3ourna1, 21, 27-33.
Watson, J. M. & Najim, A. T. (1956). Studies on
bilharziasis in Iraq. Part 12.Observationson schistosome
dermatitis. Journal of the Iraq Medical Professions, 4,
4-10.
Yacoub, A. (1985). The epidemiology of Schistosomahaema-
tobium infection in Basrah. southern Iraa: seroloeical.
parasitological and behavioural studies. Ph.D. thesis;
University of London.
Yacoub, A. & Southgate,B. A. (1987).The euidemiologv of
schistosomiasisin the later stagesof acontrol programme
basedon chemotheraov: the Basrahstudv. 1.Descriotive
epidemiology and pa%itological results. Transact& of
the Royal Societyof Tropical Medicine and Hygiene, 81,
449~A<9
. ._ ,_..
Yogore, M. G., Jr, Lewert, R. M. & Blas, B. L. (1983).
Seroepidemiology of schistosomiasisjaponica by ELISA
in the Philippines. 1. Underestimation by stool examina-
tion of the prevalence of infection in schoolchildren.
American Journal of Tropical Medicine and Hygiene, 32,
1322-1334.
Yogore, M. G.,, Jr, Lewert! R. M. & Bias, B. L. (1984).
Seroepidermologyof schistosomiasisjaponica by ELISA
in the Philippines. III. Selectivemasschemotherapy with
praziquantel in a control programme. American Journal
of Tropical Medicine and Hygiene, 33, 882-890.
Accepted for publication 9 June 1986