Infection with Chlamydia trachomatis after mass treatment of a trachoma hyperendemic community in Tanzania: a longitudinal study.
ABSTRACT Data from studies done in communities where trachoma is mesoendemic suggest that ocular infection with Chlamydia trachomatis can be eliminated after one mass treatment with antibiotics. However, there are no comparable long-term data from trachoma hyperendemic communities. Our aim, therefore, was two-fold: first, to ascertain the disease pattern of trachoma and ocular infection with C trachomatis in a trachoma hyperendemic community after mass treatment; and, second, to ascertain the risk factors for incident infection.
We did a longitudinal study of a trachoma hyperendemic community (n=1017) in Tanzania. We did surveys, including ocular swabs, at baseline, 2, 6, 12, and 18 months to identify the presence, and quantity, of C trachomatis after single mass treatment of all individuals aged 6 months or older with azithromycin 20 mg per kg; pregnant women without clinical disease received topical tetracycline.
Mass treatment (coverage 86%) significantly reduced the prevalence of infection from 57% (495 of 871) to 12% (85 of 705) at 2 months. Infection remained fairly constant to 12 months, with evidence of increasing numbers and load of infection by 18 months post-treatment. Incident infection at 6 months was 3.5-times more likely if another member of the household had more than 19 organisms per swab at 2 months. Travel outside the village, and visitors to the household, did not increase the risk of infection within households up to 12 months.
In this trachoma hyperendemic community, infection levels after high antibiotic coverage persisted at a low level to 18 months, with evidence for re-emergence after 1 year. Fairly light loads of infection were associated with household transmission. Yearly mass treatment over a few years could be sufficient to eliminate infection.
www.thelancet.com Vol 366 October 8, 2005
In 1998, WHO1set the target of worldwide elimination
of trachoma-induced blindness by the year 2020.
The SAFE strategy (surgery for trichiasis, antibiotics
for infection, facial cleanliness, and environmental
improvements to reduce transmission of Chlamydia
trachomatis) has since been implemented to control the
disease. Antibiotics are used to reduce the prevalence of
disease and the pool of infection with C trachomatis.
WHO recommends mass treatment with antibiotics
when the prevalence of active trachoma in children
exceeds 20%,2and a working group in 2003 proposed
mass treatment when the prevalence exceeds 10%.
Results of studies indicate that mass treatment with
systemic azithromycin in mesoendemic and hypo-
endemic communities greatly reduces infection to the
point of elimination after 2 years.3–5However, data from
hyperendemic communities in Ethiopia suggest re-
emergence of disease after a single dose.6Such re-
emergence might arise because of inadequate antibiotic
coverage, and a WHO workshop proposed that
community coverage of at least 80% be set as a treatment
goal. The long-term effect on infection of such degrees
of coverage in hyperendemic communities is, however,
unknown and empirical data are needed.
If untreated individuals were the source of re-
emergent infection, clustering of emergent infection
within households of untreated cases would be
expected. If contact with individuals from villages not
treated with antibiotics were the source of new
infections, then re-emergent infection would be more
likely in families that received more visitors or did more
travelling than in those who interacted less with others.
Every potential source of re-infection should be
identified, since every source would affect a trachoma
Our aim was two-fold: first, to ascertain the disease
pattern of trachoma and ocular infection with
hyperendemic community after mass treatment with
azithromycin; and, second, to ascertain the risk factors
for incident infection at 6 months after mass treatment.
months in a trachoma
Between 2000 and 2002, we did a longitudinal study of
the population of Maindi, a village in Kongwa,
Tanzania. We did a census of every household at the
start of the study, recording the age, sex, and
relationship to head of household of every individual.
Lancet2005; 366: 1296–300
Dana Center for Preventive
Ophthalmology, Johns Hopkins
University, Baltimore, MD, USA
Kongwa Trachoma Project,
London School of Hygiene and
Tropical Medicine, London, UK
Medical Research Council
Laboratories, The Gambia,
West Africa (RLBaileyFRCP)
ProfSheilaWest, Wilmer Eye
Institute, Room129, Johns
Hopkins School of Medicine,
600North Wolfe Street,
Baltimore, MD21287, USA
Infection with Chlamydia trachomatisafter mass treatment
of a trachoma hyperendemic community in Tanzania:
Sheila K West, Beatriz Munoz, Harran Mkocha, Martin J Holland, AuraAguirre, Anthony W Solomon, Allen Foster, Robin L Bailey,
Background Data from studies done in communities where trachoma is mesoendemic suggest that ocular infection
with Chlamydia trachomatis can be eliminated after one mass treatment with antibiotics. However, there are no
comparable long-term data from trachoma hyperendemic communities. Our aim, therefore, was two-fold: first, to
ascertain the disease pattern of trachoma and ocular infection with C trachomatis in a trachoma hyperendemic
community after mass treatment; and, second, to ascertain the risk factors for incident infection.
Methods We did a longitudinal study of a trachoma hyperendemic community (n=1017) in Tanzania. We did
surveys, including ocular swabs, at baseline, 2, 6, 12, and 18 months to identify the presence, and quantity, of
C trachomatis after single mass treatment of all individuals aged 6 months or older with azithromycin 20 mg per kg;
pregnant women without clinical disease received topical tetracycline.
Findings Mass treatment (coverage 86%) significantly reduced the prevalence of infection from 57% (495 of 871) to
12% (85 of 705) at 2 months. Infection remained fairly constant to 12 months, with evidence of increasing numbers
and load of infection by 18 months post-treatment. Incident infection at 6 months was 3·5-times more likely if
another member of the household had more than 19 organisms per swab at 2 months. Travel outside the village,
and visitors to the household, did not increase the risk of infection within households up to 12 months.
Interpretation In this trachoma hyperendemic community, infection levels after high antibiotic coverage persisted
at a low level to 18 months, with evidence for re-emergence after 1 year. Fairly light loads of infection were
associated with household transmission. Yearly mass treatment over a few years could be sufficient to eliminate
www.thelancet.com Vol 366 October 8, 2005 1297
For these analyses, households could be part of a larger
compound of adjacent households called a kaya.
The study was approved by the Johns Hopkins Institute
Review Board and the Tanzanian National Institute for
Medical Research. All participants provided oral informed
After the initial census, we did a baseline survey of
everyone in the village. We identified individuals with
trachoma, using the WHO simplified grading scheme,
and magnifying loupes (?2·5 magnification).7Active
trachoma was defined as having had at least one eye with
follicular trachoma or trachoma intense (severe
trachoma), or both. To detect and quantitate infection
with C trachomatis, we wiped a dacron swab across the
tarsal plate of the right eye, according to a standard
protocol.8We repeated these assessments, including
everyone in the village at the time of each follow-up, at
2, 6, 12, and 18 months after treatment with antibiotics.
We placed swabs in dry vials, kept them cold on frozen
freezer blocks in the field for the day, and then stored
them at –20°C until they could be shipped on freezer
blocks to the London School of Hygiene and Tropical
Medicine, UK, for processing. We used the Amplicor
C trachomatis qualitative PCR assay (Roche Molecular
Systems, Branchburg, NJ, USA) to identify positive
samples; the Amplicor test targets DNA of the common
cryptic plasmid of C trachomatis. Details about how we
processed samples have already been published.8Briefly,
PCR was done on eluted samples, according to the
manufacturer’s instructions. If PCR inhibition was
evident, we retested samples after diluting an aliquot one
in five with a mixture of Amplicor CT/NG lysis buffer and
specimen diluent in equal amounts. We assigned positive
and negative results, and treated ambiguous results as
outlined by the manufacturer. For this report, as in
previous published work,8we defined positive samples as
infection. We processed all positive samples for quantity
of omp1, using the quantitative PCR assay developed and
described previously8and outlined below.
We did real-time quantitative PCR with the LightCycle
(Roche Molecular Systems). The target for quantitative
PCR amplification was a 123 bp fragment within constant
domain 3 of omp1. Before assaying samples in the
LightCycler, we further processed them to purify and
concentrate the DNA, since the Amplicor buffer was not
compatible with the quantitative PCR assay. We
quantified every sample twice, on two replicate aliquots of
the re-suspended DNA. Every assay included six
concentrations of the standard (ten-fold dilutions of the
stock solution from 105to one copy per capillary) and a
We began treatment with antibiotics immediately after
completion of the baseline survey. Everyone in the village
older than age 6 months was offered a single dose of oral
azithromycin 20 mg per kg up to a maximum dose of 1 g.
We weighed children to identify the correct dose. We
offered women who told us they were pregnant topical
tetracycline if they had no signs of trachoma, otherwise
they were treated with azithromycin.
We did census updates for every household every
2weeks, and recorded any changes—ie, births, departures
and arrivals, and deaths. Anyone who spent at least 1 day
outside of the village was recorded as having travelled. We
recorded by neighbourhood any new households that
were set up in, or that moved out of, the community. If
individuals joined a household from outside the village
and stayed for at least two census updates, ocular swabs
were taken. We did not take swabs from visitors who
stayed for less than two census updates.
We report quantitation of infection as number of
organisms (copies of omp1) per swab, as previously
described.8We recorded as positive those samples that
were Amplicor positive but LightCycler negative on two
runs, for reasons described elsewhere. We used
maximum likelihood estimations to calculate the number
of organisms per swab as 4·1 in these instances.8We
modelled predictors of incident infection at 6 months
Number in censusTreatment compliance; number (%) Baseline trachoma status
Number examined Active trachoma; number (%) Severe trachoma; number (%)
Table 1: Prevalence of active trachoma at baseline and compliance with azithromycin treatment, by age
Reason absent from village Present in village
DeathMoved out* Temporary absence† Total numberNumber (%) examined
* According to surveillance. †Individuals returned for at least one subsequent follow-up.
Table 2:Follow-up rates of 1017 individuals included in first census
www.thelancet.com Vol 366 October 8, 2005
post-treatment, where incident infection was defined as
individuals who were Amplicor negative at 2 months and
positive at 6 months. We chose an interval of 6 months to
allow enough time for C trachomatis to re-emerge after
treatment. Logistic regression analyses were used to
control for confounding and to investigate interactions.
We used general estimating equations to adjust the
standard errors of the estimates for clustering of other,
unmeasured, factors at the compound level. We used
similar approaches for measurement of incident infection
at 12 and 18 months.
Role of the funding source
The sponsor of the study had no role in study design,
data collection, data analysis, data interpretation, or
writing of the report. The corresponding author had full
access to all the data in the study and had final
responsibility for the decision to submit for publication.
Maindi consists of 212 households in 101 compounds
dispersed over 2·2 square miles. Subsistence farming is
the main occupation. At baseline, the community
(n=1017) was hyperendemic for trachoma; 77% of
children aged younger than 8 years had active trachoma
and 24% had severe trachoma (table 1). 82% (85 of 104)
of individuals with severe trachoma and 72% (167 of 230)
of those with trachoma follicular only were infected with
C trachomatis. Those infected with C trachomatis and who
had severe trachoma had the highest load of infection
(median 1292 omp1 copies per conjunctival swab; range
4–2 000 874). The overall population median was
19 copies per swab. Of those with infectious loads greater
than the overall median, 71% (157 of 221) were younger
than age 15 years.
Overall treatment coverage was 86% (table 1). Table 2
shows the follow-up rates over time for the village. The
most common reason for absence of treatment or follow
up at 2 months was travel outside the village. Typically,
the farms of these families are at least a 2-day walk from
Mass treatment at this level of coverage had a great and
sustained effect on prevalence of C
(figure 1). Rates of trachoma also fell, but began to re-
emerge after 1 year (figure 1). Unlike for the prevalence
of active disease, there was no evidence for a further
decline in infection after 2 months. The median (range)
load of those infected declined from 19 (4–2 000 874) to
10 (4–415 351) omp1 copies at 2 months after treatment,
then rose to 19 (4–2 292 274), 128 (4–16 823), and
225 (4–3 460 754) omp1 copies per positive swab at 6, 12,
and 18 months, suggesting that infection load increased
with time since treatment (figure 2).
The prevalence of infection in the cohort at 6 months
post-treatment included individuals who were infected
at 2 months and remained so at 6 months and those
newly infected. For the analyses of factors associated
Time since treatment (months)
Infection with C trachomatis
2468 10 12141618
C trachomatis load (omp1 copies/swab in log units)
Time since treatment (months)
Figure 1: Prevalence of infection and active trachoma over time
Figure 2: Infection load (omp1 copies per swab) in those infected over time
Number not infected
at 2 months
Number (%) with incident
Odds ratio (95% CI)*
Infection in another household member at 2 months
Infection load greater than overall population median in another household member at 2 months
No354 26 (7%)
Yes117 26 (22%)
Treated at baseline
Visitors or new additions to household
No 361 38 (11%)
Yes110 14 (13%)
Travelled outside village
No 309 39 (11%)
Yes, 1–2 trips86
Yes, ?3 trips37
*Odds ratios from model that included all variables listed, with infection or infection load greater than median as infection
variable, depending on model. 95% CIs adjusted for clustering of other, unmeasured, factors within compound.
Table 3: Predictors of incident infection at 6 months post-treatment in participants not infected at
www.thelancet.com Vol 366 October 8, 2005 1299
with infection at 6 months, we therefore divided the
population into two groups: those who had had infection
at 2 months, and those who had not. 544 people had
infection data at both timepoints.
66 people were infected at 2 months post-treatment and
also had infection data at 6 months post-treatment. Of
this group, 56% were also infected at 6 months. Because
the sample is small, detailed analyses are inconclusive.
Nevertheless, a strong predictor of infection at 6 months
was having an infection load greater than median at
2 months (OR 42·8, 95% CI 4·5–407).
Table 3 shows predictors of incident infection at
6 months post-treatment in individuals not infected at
2 months—ie, by comparison with those who were not
infected at either timepoint. Incident infection was more
common in children aged younger than 8 years than in
the other age groups, and the risk was 3·5 times greater
in households where another household member had an
infectious load greater than the overall median at
2 months post-treatment. Visitors and new additions to
the household were not related to infection at 6 months,
nor was travel outside the village. Incident infection was
not related to the presence of low-level infection in a
household member at 2 months (table 3). Risk factors
were the same for incident infection at 12 months (data
We assessed the association between clinical
indicators of trachoma and infection over time (table 4).
At baseline, the prevalence of infection in those with
either trachoma follicular or severe trachoma was high.
After treatment, the prevalence of infection in those with
trachoma follicular changed; only a third of affected
individuals were infected, and less than 20% had an
infection load greater than the overall median. By
contrast, close to two thirds of those with severe
trachoma after treatment had infection, and almost half
had high infection loads. Moreover, the prevalence of
severe trachoma, though lower than the prevalence of
infection over time (figure 1), tracked the changes in
infection more closely than did trachoma follicular,
suggesting that in hyperendemic areas severe trachoma
could be a marker for infection post-treatment, or a
target for post-treatment surveillance.
Even though azithromycin coverage in Maindi
exceeded WHO recommended levels, ocular infection
with C trachomatis remained in the community for up
to 18 months after treatment, albeit at a low level.
Findings based on a model from data6in another
hyperendemic area indicated that re-emergence would
arise to almost pretreatment levels by 1 year after a
single mass treatment. We noted, however, that
infection and trachoma did not increase consistently
until 12 months after treatment. Our empirical data
suggest that yearly treatment for a few years might be
sufficient to reduce the prevalence of infection notably.
However, further longitudinal research is warranted,
since differences in the rate of re-emergence in
different communities is large.6
The sources of infection at 6 months post-treatment
were previous infection at 2 months and infection from
people living in the same household who had high
infection loads at 2 months. Results of other studies
done in Tanzania9,10indicate that some individuals have
constant infection and disease. As such, that previous
infection is a risk factor for later infection at 4 months
with no treatment intervention, is not surprising.
Previous treatment did not significantly affect rate of
incident infection at 6 months. The presence in the
household at 2 months of infected individuals was not
predictive of incident infection at 6 months—ie, light
infectious loads did not lead to transmission. This
finding is important, since it suggests that a threshold
of infectious load is required for transmission. Since
the threshold level in our study was fairly low, we
propose that the environment for transmission must
have been favourable. If the environment for
transmission is less facile, then higher loads might be
needed. The median load among the few infected
people in a hypoendemic area of The Gambia, for
example, was 108 copies per swab,11yet the proportion
of the population with infection and disease in that
study was very low. Thus, although the median value in
Tanzania was much lower, the threshold apparently
existed under favourable conditions for continued
transmission. If the environment for transmission is
more hostile, then high loads may be needed. It is
noteworthy that high antibiotic coverage could result in
loads sufficiently low as to interrupt re-emergence, as
has been seen in mesoendmeic and hypoendemic
Another explanation for the differences in load
between infected people in The Gambia and in our
population could be that constant transmission
stimulates an immune response that results in a lower
Number (%) with trachoma Number (%) with trachoma follicular
follicular and any infection and infection load ?19 copies
Number (%) with severe
trachoma and any infection
Number (%) with severe
trachoma and infection load ?19 copies
Table 4: Infection in individuals with trachoma over time
www.thelancet.com Vol 366 October 8, 2005
load with infection in hyperendemic communities.
Once transmission falls, and immune stimulation is
less intense, infection results in higher loads. Our
findings lend support to this notion, in that they show a
steady increase in average load of infection post-
treatment. However, the prevalence of infection in our
study did not increase greatly.
In our population, infection at 6 months was not
related to either the presence of visitors in the
household or to travel outside the village. Almost half
of our cohort travelled in the 4 months before the 6-
month survey. However, much of the travel was related
to farm work, where contact with other infected
persons from outside the village was unlikely.
Although we attempted to keep track of travel by
visiting the households every 2 weeks, we might have
missed some trips, and some individuals might have
mistakenly reported the same travel event on more
than one occasion. Finally, in other cultures, travel
outside the village might entail more mingling than is
generally the case in the individuals we studied, and
might, therefore, be a greater source of re-emergent
The findings of our community-based study show
that antibiotic coverage as high as 86% decreases, but
does not eliminate, ocular infection with C trachomatis
over 18 months. Trachoma and infectious load began to
rise 12 months after treatment, suggesting that yearly
mass treatments for this community would be effec-
tive. By 6 months post-treatment, people who were not
infected at 2 months were more likely to become
infected if another household member had an infec-
tious load greater than the median of 19 copies at
2 months. In this area of Tanzania, the environment
for transmission of C trachomatis, at fairly low loads,
seems favourable and argues for careful study of the
longer-term effect on infection of a yearly mass
S K West designed the study, oversaw data collection, and wrote the
report. B Munoz did all data analyses and, together with the other
authors listed, helped to write and review the report. H Mkocha obtained
data. M J Holland and A Aguirre did laboratory analyses. A W Solomon,
A Foster, and R L Bailey assisted in the design of the study.
D C W Mabey was the main investigator and helped to write and review
Conflict of interest statement
We declare that we have no conflict of interest.
This work was funded by a grant from the Wellcome Trust, Burrough
1WHO. Global elimination of blinding trachoma. Geneva World
Health Organization 51stWorld Health Assembly Resolution 51.11.
Geneva: World Health Organization, 1998.
2WHO. Prevention of blindness: primary healthy care management
of trachoma. Geneva: World Health Organization, 1993: 16.
3Schachter J, West SK, Mabey D, et al. Azithromycin in control of
trachoma. Lancet 1999; 354: 630–35.
4 Solomon AW, Holland MJ, Alexander ND, et al. Mass treatment
with single-dose azithromycin for trachoma. N Engl J Med 2004;
5 Burton MJ, Holland MJ, Makalo P, et al. Re-emergence of
Chlamydia trachomatis infection after mass antibiotic treatment
of a trachoma endemic Gambian community. Lancet 2005; 365:
6 Melese M, Chidambaram JD, Alemayehu W, et al. Feasibility of
eliminating ocular Chlamydia trachomatis with repeat mass
antibiotic treatments. JAMA 2004; 292: 721–25.
7 Thylefors B, Dawson CR, Jones BR, West SK, Taylor HR. A simple
system for the assessment of trachoma and its complications.
Bull World Health Organ 1987; 65: 477–83.
8Solomon AW, Holland MJ, Burton MJ, et al. Strategies for control
of trachoma: observational study with quantitative PCR. Lancet
2003; 362: 198–204.
9 Smith A, Munoz B, Hsieh YH, et al. OmpA genotypic evidence for
persistent ocular Chlamydia trachomatis infection in Tanzanian
village women. Ophthalmic Epidemiol 2001; 8: 127–35.
10West SK, Munoz B, Lynch M, et al. Risk factors for constant, severe
trachoma among preschool children in Kongwa, Tanzania.
Am J Epidemiol 1996; 143: 73–78.
11 Burton MJ, Holland MJ, Faal N, et al. Which members of a
community need antibiotics to control trachoma? Conjunctival
Chlamydia trachomatis infection load in Gambian villages.
Invest Ophthalmol Vis Sci 2003; 44: 4215–22.