Does the Diagnosis of Trachoma Adequately Identify Ocular Chlamydial Infection in Trachoma-Endemic Areas?

By the year 2020, the World Health Organization (WHO) hopes to have successfully eliminated blinding trachoma, an ocular disease caused by Chlamydia trachomatis [1]. With progression of the WHO's program, it will become increasingly important to determine which villages and which individuals within these villages are affected and therefore will benefit from treatment with antibiotics. Although the DNA amplification tests now available to diagnose C. trachomatis are highly sensitive and specific, they are expensive and rarely performed in trachoma-endemic areas. Thus, current trachoma programs use the prevalence of clinically “active trachoma” as a guide for antibiotic treatment of chlamydial infection. Using the ligase chain reaction (LCR), we evaluated the validity of clinically determined “active trachoma” as a surrogate for chlamydial eye infection among children in a hyperendemic Egyptian community

Subjects, materials, and methods The sample population consisted of 1059 subjects aged 1–15 years from a clinical trial (the Egyptian arm of the Azithromycin in the Control of Trachoma [ACT] study), which compared mass treatment with oral azithromycin with that of topical tetracycline in a trachoma-endemic area of lower Egypt [2]. Children were studied because they are most likely to be infected and because scarring later in life can make clinical diagnosis difficult [3, 4]. In the ACT study, one village was assigned community-wide oral azithromycin treatment (3 doses with intervals of 1 week: 1 g of capsules for adults and 20 mg/kg of oral suspension for children) and the other village received community-wide treatment with 1% topical tetracycline (once daily for 6 weeks). The final examination occurred, on average, 14 months after the initiation of treatment. Data reported for the pretreatment visit from both villages were combined. In the present study, data reported for the posttreatment visit were from the azithromycin-treated village only, because we did not want to combine posttreatment results from villages that had received different treatments. The baseline population in the 2 villages consisted of 731 children aged 1–10 years and 328 children aged 11–15 years. At 14 months of follow-up, there were 354 children aged 1–10 years in the azithromycin-treated village. In the present study, we attempted to treat every individual present, although there were a small number of individuals who refused treatment in each village, and some individuals or families could not be reached at scheduled treatment times. Ninety-five percent received at least 1 dose of azithromycin. Clinical and laboratory results were reported for all individuals in each village, regardless of compliance to medication [2]

Ophthalmologists experienced in the WHO's detailed grading scale performed the examination. The upper tarsal conjunctiva of the right eye was examined by use of a binocular loupe (magnification, ×2.5) and a hand light. The detailed WHO grading was later converted into the WHO's simplified grading [5]. The WHO’s simplified grading definition for follicular trachomatous inflammation (TF) is the presence of ⩾5 follicles in the upper tarsal conjunctiva [5]. The definition for intense trachomatous inflammation (TI) is the presence of inflammatory thickening of the upper tarsal conjunctiva that obscures more than half the normal deep tarsal vessels [5]. Participants were considered to be clinically active if they were graded as either TF or TI in the right eye

At each examination, swabs from the right upper conjunctiva were collected for testing by use of LCR (Abbott Laboratories) to detect C. trachomatis DNA. The swab was placed in tubes with LCR transport medium and were kept on ice until they were refrigerated within 5 h of collection. Samples were transported to the United States at ambient temperature and were stored at −80°C. LCR testing was done over a period of 2 years in a laboratory at the University of California, San Francisco. Laboratory staff were unaware of the clinical status of study participants. Chlamydial infection in this study was defined as a positive LCR test for C. trachomatis using a sample-to-cut-off ratio of 1.00, as recommended by Abbott Laboratories. Although the LCx test (Abbott Laboratories) is not approved by the US Food and Drug Administration for ocular chlamydial specimens, the test detects the plasmid common both to the trachoma serovars (A, B, and C) and the genital serovars (D–K)

Results Before treatment, the prevalence of clinical activity (TF and/or TI) in children aged 1–10 years in the 2 study villages was 48% (353/731 children; 95% confidence interval [CI], 45%–52%) and the prevalence of infection, as determined by LCR, was 49% (355/731 children; 95% CI, 45%–52%) (table 1). Both clinical activity and infection decreased with age (figure 1). The peak of both clinical activity (120 [63%] of 190 children; 95% CI, 56%–70%) and infection (110 [59%] of 186 children; 95% CI, 52%–66%) occurred in children aged 3–4 years (figure 1). The prevalence of infection as determined by LCR decreased more gradually with age than did the prevalence of clinical activity (figure 1). By age 12, the prevalence of clinically active trachoma decreased to 7% (7/96 children; 95% CI, 3%–14%), whereas the prevalence of infection decreased only to 24% (23/95 children; 95% CI 16%–34%) (figure 1)

Before treatment, 69% of those children classified as clinically active had chlamydial infection (table 1). This is the positive predictive value (PPV) of the clinical exam for detecting chlamydial infection. Overall, the clinical exam detected 69% of those children identified as infected, as determined by LCR, at pretreatment (table 1). This is the sensitivity of the clinical exam for detecting infection

Ninety-seven percent of uninfected children did not have TI (specificity of TI in detecting infection; table 2). Therefore, only 3% of uninfected children were graded TI. Eighty-eight percent of children classified as TI were identified as infected (PPV) (table 2). However, using TI alone as a clinical sign detected only 24% of infected children (sensitivity; table 2). Only 1 case of TI also was not TF; this case was LCR positive

The clinical examination proved to be less reliable in detecting infection in older children. Among children aged 1–5 years, 78% of clinically active cases were LCR positive (PPV, 78%), versus only 17% in children aged 11–15 years (PPV, 17%) (tables 3 and 4; P<.001, χ² test)

Before treatment, the clinical exam detected a greater proportion of those identified as infected (table 1), compared with that at 14 months after treatment with azithromycin (table 5). That is, in the azithromycin-treated village, 33% of infections in children were subclinical before treatment, compared with 50% after treatment (P=.35, χ² test). Although the sensitivity of the clinical examination for detecting infection in children decreased modestly after azithromycin treatment, the proportion of clinically active children who were infected (PPV) declined more drastically, from 67% before treatment (table 4) to 10% 14 months after treatment (table 5; P<.001, χ² test)

Discussion The validity of active trachoma as a marker for ocular chlamydial infection is not clear. It is well known that not all cases of clinically active trachoma also have concurrent infection, particularly in low prevalence areas [2, 6–9]. A report from 1 village in Australia found only 17% of clinically active cases had evidence of chlamydia by DNA amplification [8]. A recent study in a community in Surkhet Valley, Nepal, with 6% clinical activity, revealed no evidence of chlamydia by DNA amplification tests in any of the cases of clinically active trachoma [7]. Our present study found that almost one-third of children with clinically active trachoma were not infected, as determined by LCR

Conversely, it is well known that chlamydial infections without clinically active disease also are quite common [6, 10–12]. Our study found that, in a hyperendemic area, the clinical examination detected only 69% of infected children aged 1–10 years, which means that targeting treatment only to clinically active children still may miss almost one-third of infected cases, which is problematic since these individuals can remain a potential source of infection within the community. In older children (aged 11–15 years), the clinical exam detected only 17% of infection. Thus, current WHO recommendations for mass treatment of all individuals in a community if prevalence is ⩾20% seem reasonable [13]

Many trachoma programs target antibiotic treatment to entire households if one child with clinically active trachoma is found [13]. If treatment had been targeted to the entire household of any child with clinically active trachoma aged 1–10 years, then 332 (94%) of the 355 children identified as infected (95% CI, 90%–96%) would have been treated, and 154 (77%) of 199 households (95% CI, 71%–83%) would have been treated. The average number of individuals per household in this study was 10.5, and the average number of children aged 1–10 years per household was 3.8. Although targeted treatment of households would have detected the vast majority of infected children aged 1–10 years, this treatment strategy may have little actual savings over mass treatment, since ∼4 of 5 households would have been treated with targeted household treatment. Thus, in highly endemic populations, it is reasonable for trachoma control programs use mass treatment as opposed to targeting treatment to households with clinically active children

Using TI alone as a clinical marker missed the majority (76%) of infected children. Furthermore, only a single infection was TI and not also TF. Thus the TI grade does not contribute much more than using the TF grade alone. Other factors which may limit the utility of TI in children include the following: (1) TI is a rare event in some communities, so a large sample size would be necessary for a statistically meaningful result [13]; (2) anecdotally, some graders tend to have difficulty with the TI grading and there is some indication that there is more variation among graders in the TI grading [5]; and (3) an epidemic of adenoviral conjunctivitis would mimic the appearance of TI due to chlamydial infection [7]. Although there is some evidence that the TI grade predicts future scarring [3, 4], some investigators are now advocating not using TI at all in determining the prevalence of chlamydial infection [13]

For those programs that must continue to use the clinical exam to detect infection, the results of the study support the use of children as a sentinel group, particularly those aged 1–5 years. First, preschool children had the highest prevalence of clinical activity as found previously in other settings [3]. Furthermore, the clinical exam detected infection most reliably in this age group. That is, the clinical exam detected a much greater proportion of infection in children aged 1–5 years (78%), compared with those aged 6–10 years (58%), and an even smaller proportion of those aged 11–15 years (17%)

After treatment the clinical exam greatly overestimated the prevalence of infection in children. The proportion of clinically active children who were infected (PPV) declined from 67% before treatment to 10% 14 months after treatment. Trachomatous follicles are known to persist for months after chlamydial infection has been cleared [5, 14]. Because of this, the clinical exam may underestimate the efficacy of the antibiotic program

The clinical diagnosis of active trachoma is not a reliable marker of infection in children, at least in the short term. The clinical exam is less reliable in teenagers and less reliable after treatment. Because the antibiotic component of a trachoma control program would ideally be based on the prevalence of C. trachomatis infection, an appropriate strategy could be to sample populations by use of one of the DNA amplification tests to detect infection, perhaps pooling specimens to increase the cost-effectiveness [15]. When laboratory testing is not available, clinical examination will be more useful if it is limited to a sentinel group of children. Our results imply that monitoring preschool children provides the most information. In any event, clinical information is clearly useful for identifying hyperendemic areas before treatment and for monitoring areas with a prevalence of blinding trachoma

• Received August 27, 2002.
• Accepted December 24, 2002.

• © 2003 by the Infectious Diseases Society of America