Melese M, Chidambaram JD, Alemayehu W, Lee DC, Yi EH, Cevallos V, Zhou Z, Donnellan C, Saidel M, Whitcher JP, Gaynor BD, Lietman TM. Feasibility of Eliminating Ocular Chlamydia trachomatis With Repeat Mass Antibiotic Treatments. JAMA. 2004;292(6):721–725. doi:10.1001/jama.292.6.721
Author Affiliations: ORBIS International, Addis Ababa, Ethiopia (Drs Melese and Alemayehu); WHO Collaborating Center, F.I. Proctor Foundation (Drs Chidambaram, Saidel, Whitcher, Gaynor, and Lietman, Mr Lee, and Mss Yi, Cevallos, Zhou, and Donnellan), Department of Ophthalmology (Drs Gaynor, Whitcher, and Lietman), Department of Epidemiology and Biostatistics (Drs Whitcher and Lietman), Institute for Global Health (Drs Whitcher and Lietman), University of California, San Francisco.
Context Mass antibiotic administrations for ocular chlamydial infection play
a key role in the World Health Organization's trachoma control program. Mathematical
models suggest that it is possible to eliminate trachoma locally with repeat
mass treatment, depending on the coverage level of the population, frequency
of mass treatments, and rate that infection returns into a community after
each mass treatment. Precise estimates of this latter parameter have never
Objective To determine the rate at which chlamydial infection returns to a population
after mass treatment and to estimate the treatment frequency required for
elimination of ocular chlamydia from a community.
Design, Setting, and Participants Longitudinal cohort study of 24 randomly selected villages from the
Gurage Zone in Ethiopia conducted February 2003 to October 2003. A total of
1332 children aged 1 to 5 years were monitored for prevalence of ocular chlamydial
infection pretreatment and 2 and 6 months posttreatment.
Interventions All individuals older than 1 year were eligible for single-dose oral
azithromycin treatment. Pregnant women were offered tetracycline eye ointment.
Main Outcome Measures Prevalence of ocular chlamydial infection, measured by polymerase chain
reaction, in children aged 1 to 5 years, in each of 24 villages at each time
point was used to estimate the rate of return of infection and the treatment
frequency necessary for elimination.
Results The prevalence of infection was 56.3% pretreatment (95% confidence interval
[CI], 47.5%-65.1%), 6.7% 2 months posttreatment (95% CI, 4.2%-9.2%), and 11.0%
6 months posttreatment (95% CI, 7.3%-14.7%). Infection returned after treatment
at an exponential rate of 12.3% per month (95% CI, 4.6%-19.9% per month).
The minimum treatment frequency necessary for elimination was calculated to
be once every 11.6 months (95% CI, 7.2-30.9 months), given a coverage level
of 80%. Thus, biannual treatment, already being performed in some areas, was
estimated to be more than frequent enough to eventually eliminate infection.
Conclusion The rate at which ocular chlamydial infection returns to a community
after mass treatment suggests that elimination of infection in a hyperendemic
area is feasible with biannual mass antibiotic administrations and attainable
Mass antimicrobial administrations have been used in several control
programs and have been contemplated for many others. They have proven to be
effective against some parasitic diseases (eg, onchocerciasis and filariasis),
but at times have not lived up to expectations (eg, malaria).1- 3 Various
forms of mass treatment have been used for bacterial diseases, including sexually
transmitted chlamydia and syphilis.4,5 The
World Health Organization (WHO)6 and its partners
are now using repeated mass azithromycin administrations to control the ocular
strains of chlamydia that cause trachoma, the world's leading cause of infectious
blindness.7 Trachoma meets the critical criteria
for eradicability: there is an effective treatment for the ocular strains
of Chlamydia trachomatis, and there is no known animal
reservoir. Currently, there is little evidence of emerging chlamydial resistance
to macrolides;8 however, susceptibility testing
in chlamydia is difficult to measure and rarely performed, and further surveillance
may be needed.9 Trachoma has already disappeared
from most developed countries—the last documented case of indigenous
active trachoma in the United States appears to have been in the 1970s.10 Nevertheless, the general consensus among public
health workers is that the incidence of ocular chlamydial infection cannot
be reduced to zero in the most hyperendemic areas with antibiotics alone.
Can trachoma infection be eliminated from the most hyperendemic areas
with repeated mass antibiotic administrations? Mathematical models reveal
that it is theoretically possible to eliminate infection locally even without
complete antibiotic coverage by progressively reducing the prevalence of infection
with each treatment.11 Elimination is dependent
on the efficacy of the antibiotic in an individual, the coverage and frequency
of treatment, and the initial rate at which infection returns to a community
after mass treatment,11 but precise estimates
of the important latter parameter have never been reported. Here we determine
the rate at which chlamydial infection returns to a hyperendemic population
in Ethiopia, and from this we estimate the treatment coverage and frequency
(ie, biannual, annual) required to eliminate infection. That is, we determine
whether elimination of ocular chlamydia from severely affected areas is a
A geographical area was selected from the Gurage Zone of Ethiopia that
included 3 subdistricts and about 112 000 people (Figure 1). A stratified sample of 24 villages was randomly chosen
from a complete list of all villages (8 from each of the 3 subdistricts).
A census was conducted (February and March 2003), and all village residents
aged 1 year and older were eligible to participate in the study. A single
oral dose of azithromycin (1 g to adults, 20 mg/kg to children) was offered
within 2 weeks of the baseline examination to all members of the community
except children younger than 1 year. Children younger than 1 year were excluded
because azithromycin was approved for use in Ethiopia only for children 1
year and older. Adherence to therapy was essentially 100% of those treated,
since administration of the single-dose antibiotic was directly observed.
Pregnant women were offered topical tetracycline ointment. Guardians were
asked to bring all children aged 1 to 5 years, the ages most likely to harbor
infection, to a central location in their village for examinations at baseline
and 2 and 6 months after treatment (±1 week, from March 2003 to October
2003). Verbal consent was obtained from the parent or guardian of each child.
The right upper tarsal conjunctiva of each child was everted and swabbed.
Swabs were placed immediately at 4°C and at −20°C within 6 hours,
and transported at 4° C to the University of California, San Francisco
for processing with the Amplicor polymerase chain reaction (PCR) test (Roche
Molecular Systems, Branchburg, NJ) according to protocol.
Posttreatment samples from the same village were pooled by random selection
into groups of 5, and 200 µL of each of the 5 samples was pooled into
a single tube for processing.12- 14 The
prevalence in each village was then estimated from the proportion of positive
pools, using maximum likelihood estimation as previously described.15
Laboratory controls were included according to the Roche Amplicor protocol.
In addition, negative field controls were obtained from at least 5 random
children from each village. Immediately after the study swab and before changing
gloves for the next patient, a second swab was passed within 1 inch of the
conjunctiva without touching. These control swabs were processed in a manner
identical to the study swabs; if a pooled control was found to be positive,
then all samples in that pool were individually retested. All specimens were
processed in a masked manner.
The rate of return of infection after treatment was determined from
the observed increase in prevalence from 2 to 6 months after treatment. Using
this rate, the treatment frequency necessary to achieve elimination was obtained
from the following inequality11:
where efficacy is the efficacy of the antibiotic
in an individual, and period is the duration between
treatments. The left-hand side of the inequality represents how much a single
mass treatment with a given antibiotic efficacy and coverage reduces the level
of infection at each treatment. The right-hand side represents the exponential
increase of infection during the period between treatments. For eventual elimination,
the fraction of infection reduced by each treatment must be greater than the
increase of infection on its return between treatments.
To model the future prevalence, the level of infection was extrapolated
from the observed prevalence data. The baseline prevalence and the rate of
return of infection defined the parameters for a logistic growth model. Mass
treatments are incorporated into the model by periodically lowering the prevalence
assuming 80% coverage and that the antibiotic will eliminate infection in
95% of individuals treated. By varying the frequency of these mass treatments,
the resulting projections predict the feasibility of eliminating infection.
Results from 2 months and 6 months were compared using a t test paired by village. Intervillage variance was estimated, and
confidence intervals (CIs) were used to express uncertainty due to sampling
error. All statistical calculations were performed in STATA 7.0 (Stata Corp,
College Station, Tex) using the village as the unit of observation. A local
sensitivity analysis was performed by standard techniques16:
differentiating the necessary treatment frequency function with respect to
either coverage or initial rate of return, around the observed coverage and
rate of return. P<.05 was considered statistically
All research was conducted in accordance with the Declaration of Helsinki.
We obtained ethical approval from the institutional review board of the University
of California, San Francisco, and the National Ethical Clearance Committee
of the Ethiopian Science and Technology Commission (registered with the Office
for Human Research Protections), prior to commencing the study.
For the mass antibiotic distribution, 92% of the total 10 169 individuals
aged 1 year and older were covered by treatment, and 93% of the total 1478
children aged 1 to 5 years were covered, relative to the census (N = 24 villages).
The 3 most common reasons for not receiving treatment were temporary absence
from the village at the time of treatment, migration, and death. Refusal of
treatment was rare. Village PCR participation rates were comparable at each
time point, as depicted in Table 1.
We found that 99.1% of negative field controls were PCR negative (449/453).
Prior to treatment, the mean village prevalence of infection based on
PCR positivity was 56.3% (N = 24 villages; 95% CI, 47.5%-65.1%). After treatment,
the mean prevalence dropped to 6.7% (95% CI, 4.2%-9.2%) at 2 months; by 6
months, it had risen to 11.0% (95% CI, 7.3%-14.7%) (P =
.005 for 2 vs 6 months). Village-level prevalences at each time point are
categorized into various strata in Table
1. After mass treatment, the exponential rate of return of infection
was calculated to be 12.3% per month (95% CI, 4.6%-19.9% per month) (Figure 2). Treatment every 6 months is more
than enough to eventually eliminate infection.
Using these empirical data and the inequality above, we estimated the
minimum treatment frequency necessary for elimination to be once every 11.6
months (95% CI, 7.2-30.9 months), given 80% treatment coverage of the population
(Figure 3). This coverage level
of 80% was chosen for the projections because it has been achieved by other
trachoma programs17 and is the target recommended
by the WHO. The estimation of the necessary treatment frequency is locally
sensitive both to the initial rate of return after treatment (changing by
0.12 months for every 1% relative change in rate) and to the coverage level
(changing by 0.32 months for every 1% absolute change in coverage level).
The dependence of the necessary treatment frequency on the coverage level
is displayed in Figure 3. Also,
the effect that the uncertainty in the estimation of the rate of return has
on the necessary treatment frequency is depicted in Figure 3 by the 95% CIs.
These results imply that elimination of ocular chlamydia in this area
of Ethiopia is feasible. Biannual treatment with 80% coverage should be more
than sufficient to eventually reduce the local incidence of infection to zero
(Figure 2). This coverage level
is realistic and within the range of previous trachoma programs.17 Greater
coverage would allow less frequent treatment (Figure 3). These calculations require estimation of the rate of
return of infection into a community after mass treatment, which has not been
available previously for several reasons. First, most trachoma programs monitor
disease by following clinical activity, which does not correlate well with
infection after antibiotic treatment.18,19 Also,
intervillage variance is sufficiently high that multiple villages need to
be monitored to make a reasonable estimate.20,21 Finally,
in some areas that have been studied, trachoma appears to be disappearing
even in the absence of an organized control program, in which case infection
may never return after treatment.22- 24
Currently, the WHO's goal for trachoma programs is neither eradication
(global reduction of infection to zero) nor true elimination (local reduction
of infection to zero), but the more conservative target of "elimination of
trachoma as a public health concern".4 This
is defined as less than 5% clinical activity in children. The rationale of
the program is that infection can be reduced with several mass antibiotic
treatments, and that other sustainable, nonantibiotic measures such as face-washing
and fly control can prevent infection from returning to a community. So far,
it has been difficult to prove that any particular nonantibiotic measure has
a significant effect on chlamydial infection, although there are reasons to
be optimistic.25 If true local elimination
of ocular strains of chlamydia is feasible with antibiotics alone, then this
would provide a rationale for the trachoma program even without adjunctive
measures. If other measures prove to be as effective as hoped, then antibiotics
could be given less frequently and to a lower percentage of the population.
The prevalence of ocular chlamydia in a community before treatment may
be a key factor determining the rate that infection returns after treatment—the
greater the baseline prevalence, the more rapid the return.11,26 Trachoma
in this area of Ethiopia is as severe as anywhere else in the world, so if
biannual treatments can eliminate infection in this region, success should
be possible elsewhere, perhaps requiring even less frequent treatments. Longer
term empirical studies will be necessary to determine whether infection can
indeed be eliminated locally and to determine appropriate dosing frequencies
for less endemic areas. Although studies have found only minimal pneumococcal
resistance after mass treatment,27 the potential
for emerging chlamydial resistance should be monitored, particularly when
multiple rounds of treatment are used.
If mass periodic treatments with incomplete coverage of the population
can eliminate trachoma as these results suggest, then researchers can concentrate
on whether similar results can be obtained by targeting only a core group
most likely to be infectious. Mathematical models imply that this too is possible,
although treatment may need to be given more than twice per year.11 For many bacterial diseases, treatment targeted to
the entire population may not be appropriate. However, if only a core group
needs to be treated, then mass repeat antibiotic administration may prove
to be a valuable tool for a variety of bacterial scourges.