Antibiotics for trachoma

 

Summary of findings    [Explanations]

 

Summary of findings for the main comparison. Antibiotic versus control for trachoma: individuals

---

Antibiotic versus control for trachoma
Patient or population: patients with trachoma Settings: individuals Intervention: antibiotic Comparison: control
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	control	antibiotic
Active trachoma Follow-up: 3 months	Medium risk population^1		RR 0.78 (0.69 to 0.89)	1961 (9 studies)	⊕⊕⊝⊝ low^2,3
	800 per 1000	624 per 1000 (552 to 712)
Ocular chlamydia trachomatis infection Follow-up: 3 months	Medium risk population^1		RR 0.81 (0.63 to 1.04)	297 (4 studies)	⊕⊕⊝⊝ low^4,5
	600 per 1000	486 per 1000 (378 to 624)
Active trachoma Follow-up: 12 months	Medium risk population^1		RR 0.74 (0.55 to 1)	1035 (4 studies)	⊕⊕⊝⊝ low^6,7,8
	750 per 1000	555 per 1000 (413 to 750)
Ocular chlamydial trachomatis infection Follow-up: 12 months	Medium risk population		RR 0.25 (0.08 to 0.78)	129 (1 study)	⊕⊕⊝⊝ low^9
	190 per 1000	48 per 1000 (15 to 148)
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

---

 ^1 Median risk in control groups in included studies (rounded to nearest 10 per 1000)
^2 Serious limitations in design: None of the trials reported methods to conceal the allocation. Two trials only attempted to mask the assessment of active trachoma.
^3 Serious inconsistency: Risk ratios ranged from 0.40 to 1.02.
^4 Serious limitations in study design: None of the trials reported adequate allocation concealment. Three out of four trials masked outcome assessment.
^5 Serious imprecision: 95% confidence intervals include 1 (0.63 to 1.04)
^6 Serious limitations in design: None of the trials reported allocation concealment or masking of outcome assessment.
^7 Serious inconsistency: Risk ratios ranged from 0.50 to 1.05.
^8 No downgrading for imprecision: The pooled estimate of effect is imprecise (confidence intervals 0.55 to 1). However, we felt that this inconsistency probably arises due to limitation in study design and inconsistency and therefore did not additionally downgrade the quality of evidence on the basis of this criteria.
^9 Very serious limitations in study design: Only one small trial which did not report adequate methods of allocation concealment and did not mask outcome assessment.

 Summary of findings 2 Oral versus topical antibiotic for trachoma: individuals

 Summary of findings 3 Oral azithromycin compared to control for trachoma: communities

 Summary of findings 4 Oral azithromycin compared to topical tetracycline for trachoma: communities

 

Background

 

Description of the condition

Trachoma is the world's leading infectious cause of blindness (Resnikoff 2004). Active trachoma affects an estimated 41 million people, the majority of them children (Mariotti 2009). About 1.3 million people are blind as a consequence of trachoma (Resnikoff 2004). It is a disease of poverty and is associated with poor water supplies and sanitation.

There are two phases of trachoma. In the first phase, most frequently seen in infancy and childhood, there are repeated attacks of conjunctivitis caused by the organism Chlamydia trachomatis (C. trachomatis). The conjunctivitis is characterised by the presence of follicles on the under surface of the upper eyelid and by vascular changes in active trachoma. Active trachoma is associated with discharge from the eyes and nose that is particularly noticeable on the faces of children, but the active stage may also be asymptomatic in children and adults. When symptomatic, symptoms may persist for months after the infection is cleared. C. trachomatis is transmitted from child to child and from child to mother and back to child through eye-finger-eye contacts, fomites and via eye-seeking flies.

Repeated conjunctival infections over a number of years lead to the second phase of disease, characterised by scarring and shortening of the upper eyelid. Ultimately, the lashes turn inwards to rub on the cornea, causing pain, corneal abrasions and secondary infection. Blindness results from corneal opacification. The blinding phase affects women more commonly than men and typically starts in adult life (Burton 2009). The treatment at this stage is surgery to reposition the eyelid margin.

 

Description of the intervention

Active trachoma has been treated with antibiotics since the 1950s and a variety of regimens have been used. The antibiotic can be applied directly to the conjunctiva (topical) or taken orally (systemic). Antibiotics applied topically are usually in the form of an ointment and a variable amount is squeezed onto the inner surface of the lower eyelid. This route gives a high concentration of the antibiotic to the conjunctiva but a low dose to the nasopharynx, which is also a reservoir for the organism. Ointments may cause stinging eyes and temporary blurred vision, and they are difficult to apply to small children.

Oral treatment gives a higher dose of antibiotic to sites of infection outside of the eye, but systemic antibiotics can have various adverse effects in the person taking them. Bacteria anywhere in the body may also develop antibiotic resistance. A full course of oral treatment has a higher compliance rate than a course of topical antibiotic.

Efforts in trachoma control have used various antibiotic treatment regimens and have also been aimed at different subgroups within a trachoma endemic area. Examples of subgroups are only those individuals with clinical signs of disease (detected actively or passively), and active cases together with family contacts or high-risk groups including school children. Because many individuals harbour infection without demonstrating physical signs, it has been suggested that trachoma control cannot be achieved by antibiotic treatment given only to subgroups of a trachoma endemic community (Bailey 1993; Kamiya 1956; Sutter 1983). This led to the concept of community-based interventions, where all residents of a community should receive treatment irrespective of disease status.

The desired primary endpoint of any intervention against the active disease is reduction of blindness but this could only be demonstrated 20 to 30 years after the start of the intervention. The usual surrogate outcome measure in trachoma intervention trials is clinically active disease. In some trials a secondary endpoint is laboratory evidence of ocular C. trachomatis infection.

 

Why it is important to do this review

International interest in trachoma was given a boost in 1997 when the WHO launched a new initiative for trachoma control, based on the 'SAFE' strategy. The components giving their name to the acronym are surgery, antibiotics, facial cleanliness and environmental improvement. Cochrane systematic reviews on surgery for trichiasis (Yorston 2006), face washing (Ejere 2004) and environmental sanitary interventions (Rabiu 2007) have also been completed.

The World Health Organization (WHO) recommended topical treatment is 1% tetracycline ointment to both eyes, either twice daily for six weeks or on five consecutive days each month for six months. Compliance with this treatment is poor due to the side effects of the ointment and the length of the treatment programme. The WHO recommended oral antibiotic is azithromycin, given as a single dose of one gram in adults and 20 mg/kg of body weight in children. Azithromycin has low plasma levels but high intracellular concentrations and a long half-life. It has been shown to be an effective treatment of genital chlamydial infections.

It is important to do this review to systematically evaluate the safety and effectiveness of these recommended treatment regimens.

 

Objectives

The aim of this review is to assess the evidence in relation to the antibiotic arm of the SAFE strategy by assessing the effects of antibiotics on both active trachoma (primary objective) and on C. trachomatis infection (secondary objective). In particular, when this review was first published in 2002, the aim was to investigate the strength of evidence that antibiotics were more effective than placebo in reducing disease and to compare the effects of oral azithromycin with topical tetracycline.

These objectives were modified when the review was updated in 2010. It was decided to consider individually randomised and cluster-randomised trials separately as we felt that they were addressing different questions and were likely to be measuring different effects. The following two objectives were identified.

1. What is the effect of antibiotic treatment for individuals on active trachoma and ocular C. trachomatis infection?

2. What is the effect of community treatment with antibiotics on the prevalence of active trachoma and ocular C. trachomatis infection?

 

Methods

 

Criteria for considering studies for this review

 

Types of studies

This review includes only randomised controlled trials (RCTs) of antibiotic treatment for active trachoma. Clinical and community-based trials were included in this review. In clinical trials the unit of randomisation was the individual with active trachoma and outcomes were reported at an individual level. In community-based trials the unit of randomisation was a community, in which some individuals had active trachoma, and outcomes may have been reported at an individual or a community level.

 

Types of participants

Participants in the trials were people who were usually resident in a trachoma endemic area.

 

Types of interventions

We included trials in which the interventions were:

1. topical or oral administration of an antibiotic at any dose or frequency compared to placebo or no treatment;
2. topical administration of an antibiotic at any dose or frequency compared to oral administration of an antibiotic at any dose or frequency.

We excluded studies if the antibiotic was combined with an environmental or educational intervention unless this component was used uniformly across the trial and only the antibiotic treatment varied in the different groups.

 

Types of outcome measures

We measured outcomes at three, 12 and 24 months after the start of treatment. Three months was the time at which the maximum effect on active trachoma was expected given that clinical signs take several months to resolve after the clearance of infection. We selected 12 months to represent the period during which recurrence of infection or relapse would be most likely to occur, and we selected 24 months to reflect the expected long-term result of one course of treatment. A course of treatment may be a single or multiple doses of an oral antibiotic or interrupted applications of a topical antibiotic applied over six weeks to several months.

In order to take into account the fact that studies may not have collected outcomes at those exact times, we defined the following ranges for each:

• three months, i.e. outcomes measured before six months;
  
• 12 months, i.e. outcomes measured between six months and 18 months;
  
• 24 months, i.e. outcomes measured after 18 months.
  

If more than one outcome measurement was available, then we selected the nearest measurement to three, 12 or 24 months.

 

Primary outcomes

The primary outcome for this review was active trachoma. There are five main trachoma grading scales (Dawson 1975b; Dawson 1981b; MacCallan 1936; Thylefors 1987; WHO 1962). All these scales, except for MacCallan, quantify the number of follicles and the degree of vascular engorgement of the under surface of the upper eyelid as seen with low magnification (usually x 2.5). The Dawson scales subdivide the follicular and papillary activity as F 0 to 3 and P 0 to 3. The Thylefors scale is a simplified version defining active trachoma by the grades TF (mild-moderate) and TI (intense). The MacCallan scale is not directly comparable with the other scales as scarring is included as an indicator of active disease. The four more recent scales are broadly comparable. A minor inconsistency between them is that Dawson's F1 is defined as five or fewer follicles in zones two and three, and F2 as "more than 5 follicles in zones 2 and 3 together, but less than 5 in zone 3"; whereas TF is five or more follicles in zones two and three. This means that the divisions between F1 and F2 and 'not TF' and TF do not quite coincide.

In this review we defined the absence of active trachoma as:

• not TF and not TI (Thylefors scale);
  
• (P0 or P1 or P2) AND (F0 or F1) (WHO and Dawson scales).
  

We defined active trachoma as TF, TI, or both, in the Thylefors scale; or any other grade for P or F in the WHO or Dawson scales.

 

Secondary outcomes

The secondary outcome was a positive test for C. trachomatis infection. A variety of tests have been used to demonstrate presence of the pathogen. Historically, staining of conjunctival cells to show inclusion bodies was the first method of identifying infection. This was followed by culture of the organism, which was time consuming and lacking in sensitivity. The demonstration of antigen by various antibody staining methods followed, and finally identification of chlamydial DNA by various amplification methods. The tests, in order of increasing sensitivity, are:

1. culture by C. trachomatis isolation in eggs or tissue culture;
2. staining of conjunctival smears with giemsa or iodine;
3. direct fluorescent antibody cytology;
4. indirect enzyme immunoassay;
5. DNA hybridisation;
6. DNA amplification with the ligase chain reaction;
7. DNA amplification with the polymerase chain reaction.

We recorded adverse effects, if reported.

 

Search methods for identification of studies

 

Electronic searches

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) 2010, Issue 11, part of The Cochrane Library. www.thecochranelibrary.com (accessed 12 December 2010), MEDLINE (January 1950 to December 2010), EMBASE (January 1980 to December 2010), the metaRegister of Controlled Trials (mRCT) (www.controlled-trials.com) (December 2010) and ClinicalTrials.gov (www.clinicaltrials.gov) (December 2010). There were no language or date restrictions in the search for trials. The electronic databases were last searched on 12 December 2010.

See: Appendices for details of search strategies for CENTRAL (Appendix 1), MEDLINE (Appendix 2), EMBASE (Appendix 3), mRCT (Appendix 4) and ClinicalTrials.gov (Appendix 5).

 

Searching other resources

We used the Science Citation Index to search for articles that cited the included studies. We searched the reference lists of identified articles for any other potentially relevant studies. We also contacted experts in the field, either directly or through the membership of the WHO workshops, requesting information on unpublished trials.

 

Data collection and analysis

 

Selection of studies

For the first publication of this review, one author assessed the titles identified from the initial searches and selected all titles that made reference to treatment for trachoma. When the review was updated in 2005, and again in 2011, two authors screened the search results. The searches also found references to genital C. trachomatis infections and to laboratory tests on C. trachomatis. We excluded titles that clearly referred to either of these groups at the first viewing. Two authors independently obtained the full copies of all possibly relevant papers and assessed them according to the 'Criteria for considering studies for this review'. We assessed the trials meeting these criteria for quality.

 

Data extraction and management

Two authors independently extracted data. Discrepancies were resolved before entry into Review Manager 5 (RevMan 2008).

For the update of the review in 2011, JE checked the original data collection and entry. Appendix 6 summarises the changes that were made. For the new trials that were identified, two authors (JE, AWS) extracted data independently and resolved discrepancies by discussion. Data were entered by both authors onto two spreadsheets and cross-checked. Data were cut and pasted into RevMan from the spreadsheet (JE).

 

Assessment of risk of bias in included studies

This was a new feature for the update in 2011. We assessed the risk of bias using the Cochrane Collaboration's tool for assessing the risk of bias (Higgins 2008a).

We assessed the extent to which bias could have been introduced in the following aspects of study design and execution: sequence generation, allocation concealment, blinding (masking), incomplete outcome data and selective outcome reporting.

We did not assess sequence generation and allocation concealment for cluster-randomised studies but considered two additional criteria: recruitment bias and baseline imbalances (Higgins 2008b).

Two authors (JE, AWS) independently assessed risk of bias, compared results and resolved discrepancies by discussion.

 

Measures of treatment effect

The primary outcome for the review was active trachoma and the secondary outcome was ocular C. trachomatis infection. Both of these are dichotomous (adverse) outcomes and our preferred effect measure was the risk ratio.

 

Unit of analysis issues

This review includes trials in which individuals were randomly allocated to treatment and trials in which communities were the unit of allocation (cluster-randomised trials). A correct analysis of cluster-randomised trials includes an adjustment for the fact that people within a cluster tend to be more similar to each other than to people from other clusters i.e. the observations are not independent. The effect of cluster-randomisation is to increase the size of standard errors and hence widen the confidence intervals compared with a study of the same size using individual participant randomisation (Donner 1982).

For the update in 2011, our preferred method of analysis of cluster-randomised studies was as follows: for those studies that reported the effect measure using an analysis that properly accounted for the cluster design, we planned to enter and pool data from different studies using the generic-inverse variance method in RevMan. However, we were aware that cluster-randomised trials are not always analysed and reported appropriately. For those studies that did not report such an effect measure we planned to perform an approximate analysis (Higgins 2008b) as follows:

• calculate a 'design effect' of 1 + (M - 1) ICC (where ICC = intra-cluster correlation coefficient and M = average cluster size);
  
• multiply the standard error of an analysis at the individual level by the square root of the design effect.
  

Estimates from the literature suggest that the ICC can vary from 0.05 to 0.2 (Katz 1988; West 1991). We planned sensitivity analyses using iCC estimates of 0.05. 0.1 and 0.2.

 

Dealing with missing data

The clinical need to change or discontinue antibiotic therapy (for an individual undergoing treatment for a single episode of infection of disease, or a community undergoing a single round of mass treatment) is likely to be rare. This reduces the potential problems associated with performing the analysis on an intention-to-treat basis. More serious problems may arise from losses to follow up and non-compliance. Some of the trials have been done in largely transient populations in which losses to follow up rapidly accumulate as people move on. Such losses were assumed to be independent of the outcome measures, therefore we did not exclude studies on this basis.

 

Assessment of heterogeneity

We assessed heterogeneity by considering clinical and study design differences between trials and by examining the forest plots. We also considered statistical measures of heterogeneity such as the χ² test and I² statistic.

 

Assessment of reporting biases

As less than 10 trials were included in the meta-analyses in this version of the review we did not assess publication bias. In future updates that include more trials, we will assess the possibility of small study effects, including publication bias, using a funnel plot (plotting the risk ratio along the x axis versus standard error along the y axis).

We did not judge the possibility of selective reporting of outcomes to be a problem in most of the included trials because the two main outcomes of this review, active trachoma and C. trachomatis infection, were usually reported.

We included all trials irrespective of the language of publication, however we cannot exclude the possibility that negative trials have been published in less accessible journals (see publication bias above).

We did not find any evidence of multiple (duplicate) reporting publication bias. Data from one of our included trials (ACT 1999 The Gambia) were published twice, with ocular C. trachomatis infection being the focus of one publication and active trachoma the focus of the other, but the relationship of the data was clear from the publications.

 

Data synthesis

In the original review, the review authors pooled outcomes from community-based trials in which non-affected and affected cases were treated with outcomes from individual-based trials in which only affected cases were treated. The original protocol planned but did not carry out a sensitivity analysis to determine the effect of using only data from cases that were active at baseline.

In the update, we considered these community-based and individually randomised trials separately as they are asking different questions and are likely to be estimating different treatment effects.

Where appropriate, data were pooled using a random-effects model. If there were three trials or less we used a fixed-effect model. In cases where there was substantial heterogeneity or inconsistency, that is the individual study estimates were different sides of the null line and/or confidence intervals did not overlap, with corresponding high levels of I² we did not pool the results.

 

Subgroup analysis and investigation of heterogeneity

We considered type of antibiotic (oral or topical) to be a potential source of clinical heterogeneity. This subgroup analysis was not specified explicitly but was implied in the objectives of the original protocol which were to consider oral and topical antibiotics separately, in particular oral azithromycin and topical tetracycline. A further subgroup analysis considered just those trials in which communities were randomised to oral azithromycin, topical tetracycline, or both, where the antibiotic was administered using regimens consistent with WHO guidelines current in 2010, compared either to each other, placebo or no treatment.

 

Sensitivity analysis

As set out above under 'unit of analysis issues', we considered the possible effect of assumptions about the size of the intra-class correlation coefficient (ICC) on the results.

 

Results

 

Description of studies

See: Characteristics of included studies; Characteristics of excluded studies.

The characteristics of the included studies and reasons for exclusion of studies are detailed in the 'Characteristics of included studies' and 'Characteristics of excluded studies' tables.

 

Results of the search

Details of the original searches for previous versions of this review are in Appendix 7. The electronic searches were updated in December 2010. After de-duplication, the search identified a total of 341 references. The Trials Search Co-ordinator scanned the search results and removed 251 references which were not relevant to the scope of the review. Two authors independently reviewed the remaining 90 references and three new studies were identified (Atik 2006; Lee 2007; TANA 2009). Three studies previously classified as excluded studies were considered to be eligible for inclusion (Bailey 1993; Chidambaram 2006; Resnikoff 1995). One new report with relevant data for ACT 1999 The Gambia was identified.

Fourteen individually randomised studies were included in this review (Attiah 1973; Bailey 1993; Bowman 2000; Cochereau 2007; Darougar 1980; Dawson 1969 Sherman; Dawson 1969 Stewart; Dawson 1997; Foster 1966; Hoshiwara 1973; Peach 1986; Shukla 1966; Tabbara 1996; Woolridge 1967). The citations Dawson 1969 Sherman and Dawson 1969 Stewart refer to two arms of the same trial, which were conducted in different schools. As the results were reported separately in the paper, they have been treated as separate studies.

Eight community-based studies were included in this review (ACT 1999 Egypt; ACT 1999 Tanzania; ACT 1999 The Gambia; Atik 2006; Chidambaram 2006; Lee 2007; Resnikoff 1995; TANA 2009). The three ACT citations used the same protocol, which was applied in different countries and reported in the same article. Chidambaram 2006, Lee 2007 and TANA 2009 used communities with 'delayed treatment' as a comparator group. Although this had the disadvantage that baseline data were not available, it is an ethical solution to community randomisation.

 

Included studies

 

Types of participants

In individually randomised studies, all participants had active trachoma in at least one eye. The groups described were children aged six to 12 years in Egypt (Attiah 1973), people aged nine months and older in The Gambia (Bailey 1993), children aged six months to 10 years in The Gambia (Bowman 2000), children age one to 10 years in Pakistan and Guinea-Conakry (Cochereau 2007), pre-school children in Iran (Darougar 1980), boarding school residents aged 12 to 21 years in USA (Dawson 1969 Sherman; Dawson 1969 Stewart), children aged two to 10 years in Egypt (Dawson 1997), boarding school residents aged eight to 20 years in USA (Foster 1966), boarding school residents aged seven to 13 years in USA (Hoshiwara 1973), children under 21 years in northern Australia (Peach 1986), school children aged five to 13 years in India (Shukla 1966), children aged seven to 14 years in Saudi Arabia (Tabbara 1996), primary school children in Taiwan (Woolridge 1967).

In the community-based studies, three studies (ACT 1999 Egypt; ACT 1999 Tanzania; ACT 1999 The Gambia) included all residents of the study villages irrespective of age or trachoma status. Chidambaram 2006 and TANA 2009 offered treatment to everyone over the age of one year but data were reported for children aged one to five and one to 10 years old respectively. The settings were Egypt, Ethiopia, Saudi Arabia, The Gambia and Tanzania. One study (Atik 2006) included all residents of study villages who were over the age of six months; the setting was Vietnam. In three studies, carried out in Ethiopia (Chidambaram 2006; Lee 2007) and Mali (Resnikoff 1995), only children aged one to five years were examined but all residents of treated villages were offered treatment.

 

Types of intervention

There were various treatment strategies applied over periods of three weeks to 12 months.  Table 1 summarises the different treatment schedules for the individually randomised studies.

Eleven trials investigated oral antibiotics (Bowman 2000; Bailey 1993; Cochereau 2007; Darougar 1980; Dawson 1969 Sherman; Dawson 1969 Stewart; Dawson 1997; Foster 1966; Hoshiwara 1973; Shukla 1966; Tabbara 1996). Five trials used azithromycin 20 mg/kg (Bailey 1993; Bowman 2000; Cochereau 2007; Dawson 1997; Tabbara 1996). Other oral antibiotics included doxycycline (Darougar 1980; Hoshiwara 1973), trisulphapyrimidines (Dawson 1969 Sherman; Dawson 1969 Stewart), sulphamethoxypyridazine (Foster 1966) and sulphadimethoxine (Shukla 1966).

Almost all of the included individually-randomised trials had a treatment arm with topical antibiotics (the exception was Hoshiwara 1973.) Almost all of these trials of topical antibiotic used tetracycline or oxytetracycline. Most used 1% formulations. One trial used 0.25% (Attiah 1973) and in one trial people with 'severe disease' were given erythromycin 250 mg four times daily for two weeks in addition to topical tetracycline (Bailey 1993). A couple of trials did not report the dose (Bowman 2000; Peach 1986). Dawson 1997 used tetracycline 1% with polymyxin 10,000 units/g. Shukla 1966 used sulphafurazole 15%. There was considerable variation in the treatment schedules used. Topical treatment was applied one to four times daily, for one to seven days over a six week to 12 month period. Cochereau 2007 was the only trial to use topical azithromycin 1.5% on a two-day and three-day treatment schedule.

Some trials compared oral and topical treatments (Bailey 1993; Bowman 2000; Cochereau 2007; Darougar 1980; Dawson 1969 Sherman; Dawson 1969 Stewart; Dawson 1997; Foster 1966; Shukla 1966; Tabbara 1996). In three of these trials there was also an untreated control group (Darougar 1980; Foster 1966; Shukla 1966).

 Table 2 summarises the treatments used in the cluster-randomised studies. The majority of the cluster-randomised trials investigated azithromycin 20 mg/kg and this was compared to tetracycline or oxytetracycline 1% (ACT 1999 Egypt; ACT 1999 Tanzania; ACT 1999 The Gambia; Atik 2006). One trial (Atik 2006) assessed the effect of targeted treatment (with azithromycin 20 mg/kg) on schoolchildren with active trachoma, aged five to 15 years, plus their family members within the selected clusters rather than mass treatment of the entire cluster. One trial compared oxytetracycline 1% against no treatment (Resnikoff 1995). Three trials compared azithromycin 20 mg/kg against no (delayed) treatment (Chidambaram 2006; Lee 2007; TANA 2009). Specific exclusion criteria were usually given for pregnant women who were given either oral erythromycin or topical tetracycline instead of azithromycin.

 

Types of outcome measures

Most trials used active trachoma as the main outcome measure, the exceptions being Lee 2007, Chidambaram 2006 and TANA 2009 which focused on ocular chlamydial infection. The trachoma grading scales used after 1962 do not have scarring as a feature of active trachoma and so the underlying principles in the grades are more or less equivalent in all of the studies, using only the presence of follicles and papillae for diagnosis of active disease.

The secondary outcome measure was presence of C. trachomatis. The laboratory tests used were culture in McCoy cells, one to five identifiable inclusions per 100 to 1000 cells, elementary bodies ≤ 200 or ≥ 200 on conjunctival smear immunofluorescence, polymerase chain reaction (PCR) and ligase chain reaction (LCR).

 

Follow up

Most studies reported outcomes at three months. Fewer trials reported outcomes at 12 months and only one trial reported outcomes at 24 months.

 

Excluded studies

See the 'Characteristics of excluded studies' table for information on excluded studies, including reasons for exclusion.

 

Risk of bias in included studies

See Figure 1 and Figure 2. There were 14 individually randomised trials included in this review and eight cluster-randomised trials. We did not assess sequence generation and allocation concealment for cluster-randomised studies but considered two additional criteria: recruitment bias and baseline imbalances (see other potential sources of bias).

Figure 1. Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.

Figure 2. Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

 

Allocation

Sequence generation and allocation concealment were poorly described, with only one trial (Bowman 2000) reporting adequate methods for both of these criteria. Dawson 1997 and Woolridge 1967 reported adequate sequence generation but not allocation concealment, which was likely to be a more important source of bias.

 

Blinding

Assessment of ocular C. trachomatis infection is relatively easy to mask as it is straightforward to anonymise laboratory samples. Eleven out of 14 studies that reported ocular chlamydial infection also reported masking the assessment of the laboratory samples. Clinical assessments of trachoma are more difficult to mask, especially in the cluster-randomised studies where one community received treatment and another did not, or in cases where the treatments differed for example oral versus topical antibiotic. Only five studies reported efforts to mask the assessment of active trachoma (Bailey 1993; Bowman 2000; Cochereau 2007; Dawson 1969 Sherman; Dawson 1969 Stewart).

 

Incomplete outcome data

Only three studies provided data suggesting that incomplete outcome data were unlikely to bias the results, that is they reported high follow-up rates which were equal between intervention groups (Bailey 1993; Dawson 1997; TANA 2009).

 

Selective reporting

There was little suggestion of selective outcome reporting.  Table 3 and  Table 4 show the outcome reporting grid. In most cases, where an outcome was not reported it was because the study follow up was not conducted at that time point or "Not mentioned but clinical judgement says unlikely to have been measured", which is unlikely to introduce bias. Three trials did not report active trachoma. For Chidambaram 2006 and TANA 2009, data on active trachoma were collected and not reported; the authors have supplied unpublished data on active trachoma for TANA 2009. For Lee 2007 the focus of the study was the presence of C. trachomatis DNA on flies and in the eyes of children, so it was plausible that a clinical assessment was not done. Nine trials (out of a total of 21 included studies) did not report ocular C. trachomatis infection but there was nothing in the reports of these studies to suggest that the data had been collected and not reported. In one trial Cochereau 2007 it was clear that data on ocular infection had been collected but not reported.

All studies (with the exception of Lee 2007) reported outcomes at three months. Thirteen (out of 21) trials reported outcomes at 12 months. Only one trial reported 24 months outcomes for both treatment and control groups (Atik 2006), although Chidambaram 2006 reported outcomes for the treated communities only at 24 months. Again, there was no suggestion from the published reports that the non-publication of outcomes at 12 or 24 months was related to study results ( Table 3,  Table 4).

 

Other potential sources of bias

Recruitment bias can occur when individuals are recruited to the trial after the clusters have been randomised as the knowledge of whether each cluster is an ‘intervention’ or ‘control’ cluster could affect the types of participants recruited (Higgins 2008b). None of the included studies discussed this issue.  

When small numbers of clusters are randomised, there is a possibility of chance baseline imbalance between the randomised groups in terms of either the clusters or the individuals (Higgins 2008b). This was a particular problem with the cluster-randomised trials included in this review. Four of the trials randomised only two communities to treatment or control (ACT 1999 Egypt; ACT 1999 Tanzania; Atik 2006; Resnikoff 1995) and one trial randomised only six communities (three in each group) to treatment or control (Lee 2007). Only two studies were of a reasonable size (Chidambaram 2006: 15 communities; TANA 2009: 24 communities). In ACT 1999 The Gambia eight communities were pair matched.

Reporting of the baseline comparability of clusters or statistical adjustment for baseline characteristics (ACT 1999 Egypt; ACT 1999 Tanzania; Lee 2007) can help reduce concern about the effects of baseline imbalances, however it is difficult to interpret differences in treatment effect between only two communities because there may be some other unknown confounding factor that explains the difference in effect.

 

Effects of interventions

See:  Summary of findings for the main comparison Antibiotic versus control for trachoma: individuals;  Summary of findings 2 Oral versus topical antibiotic for trachoma: individuals;  Summary of findings 3 Oral azithromycin compared to control for trachoma: communities;  Summary of findings 4 Oral azithromycin compared to topical tetracycline for trachoma: communities

 

What is the effect of antibiotic treatment of the individual on active trachoma and ocular C. trachomatis infection?

 

Antibiotic versus control (placebo)

 Analysis 1.1 shows the effect of any antibiotic treatment on active trachoma at three months. Nine trials randomising 1961 people contributed to this analysis. There was considerable heterogeneity between trials (I² = 73%). The treatment effects observed in the different trials ranged from a risk ratio of 0.40 (95% CI 0.20 to 0.79) (Dawson 1969 Stewart) to a risk ratio of 1.02 (95% CI 0.83 to 1.25) (Darougar 1980). However, most of the trials suggested an apparent beneficial effect of treatment on active trachoma measured at three months follow up. The pooled risk ratio was 0.78 (95% CI 0.69 to 0.89).

 Analysis 1.2 shows the effect of any antibiotic treatment on ocular C. trachomatis infection at three months. Fewer trials contributed to this analysis (four trials, n = 297). However, in contrast to the effect on active trachoma there was no evidence of any heterogeneity in treatment effect between trials (I² = 0). The treatment effect appeared to be of a similar order of effect as for active trachoma but did not achieve conventional levels of statistical significance (pooled risk ratio of 0.81, 95% CI 0.63 to 1.04).

 Analysis 1.3 shows the effect of any antibiotic treatment on active trachoma at 12 months. Four trials randomising 1035 people contributed to this analysis. Again there was evidence of considerable heterogeneity between trials (I² = 90%). The treatment effects observed in the different trials ranged from a risk ratio of 0.50 (95% CI 0.41 to 0.62) (Shukla 1966) to a risk ratio of 1.05 (95% CI 0.88 to 1.24) (Foster 1966). However, three of the four trials showed a statistically significant beneficial effect of treatment on active trachoma measured at 12 months follow up. The pooled risk ratio was 0.74 (95% CI 0.55 to 1.00).

 Analysis 1.4 shows the effect of any antibiotic treatment on C. trachomatis infection at 12 months. Only one trial provided data on ocular chlamydial infection at 12 months (Darougar 1980). The effect was strong with a risk ratio of 0.25. Although this was statistically significant the estimate of treatment effect was imprecise with a wide confidence interval (0.08 to 0.78), reflecting the small sample size of the trial.

One source of clinical heterogeneity in these trials was whether oral or topical antibiotic was used. One of the objectives of the review was to compare oral and topical treatment, in particular oral azithromycin and topical tetracycline.

 Analysis 2.1 shows the results separately for the trials that considered oral antibiotic versus control and the trials that considered topical antibiotic versus control on active trachoma at three months. Although statistical heterogeneity was reduced by considering these trials separately, there remained substantial heterogeneity (I² of 60% and 68%). The pooled estimate of treatment effect for oral antibiotics on active trachoma at three months was 0.81 (95% CI 0.67 to 0.97) and for topical antibiotics was 0.82 (0.72 to 0.92). A similar picture was seen for active trachoma at 12 months ( Analysis 2.3). There were not enough data to make a reliable comparison of the effects of oral and topical antibiotics versus control on C. trachomatis infection ( Analysis 2.2;  Analysis 2.4).

Subgroup analyses such as these can be misleading because there may be other reasons for differences between trials apart from the type of antibiotic used. Direct comparison of oral versus topical antibiotic within trials is a more reliable estimate of relative effect.

 

Oral versus topical antibiotic

 Analysis 3.1 shows the effect of oral versus topical antibiotic on active trachoma at three months from within-trial comparisons (six trials, n = 953). There was considerable statistical heterogeneity (I² = 63%). The estimates of effect were spread across the null line with three trials reporting a beneficial effect of oral antibiotics and three trials reporting a beneficial effect of topical antibiotics. Three of the six trials (Darougar 1980; Dawson 1997; Foster 1966) had findings consistent with a hypothesis of no difference in effect. Similarly for active trachoma at 12 months ( Analysis 3.3), C. trachomatis infection at three ( Analysis 3.2) and 12 months ( Analysis 3.4), there was no consistent evidence as to whether oral or topical antibiotics were more effective.

Examining the trials for clinical heterogeneity suggested that the interventions used in Bowman 2000 were different. In particular, this study focused on "practical operational conditions" which meant that the topical treatments were unsupervised. Excluding this trial from the analyses substantially reduced the observed inconsistency (I² = 0) with a pooled risk ratio for the remaining five included trials of 1.04 (95% CI 0.94 to 1.16). Similar improvements in consistency were seen when Bowman 2000 was excluded from the 12 months analyses (i² changed from 56% to 29%, pooled risk ratio 1.01 (95% CI 0.85 to 1.20).

 Analysis 4.1,  Analysis 4.2,  Analysis 4.3 and  Analysis 4.4 show the specific comparison between oral azithromycin and topical tetracycline for active trachoma and C. trachomatis infection at three and 12 months. There was considerable heterogeneity in the results of these studies for active trachoma ( Analysis 4.1). As before, excluding Bowman 2000 from the analyses substantially reduced the inconsistency (I² = 0) and the pooled risk ratio of the two remaining trials was 1.01 (95% CI 0.80 to 1.28). Only two trials reported data at 12 months. Bowman 2000 reported a beneficial effect of azithromycin compared to tetracycline (risk ratio 0.66, 95% CI 0.45 to 0.98). Dawson 1997 reported a smaller effect which was not statistically significant (risk ratio 0.90, 95% CI 0.65 to 1.23).

Data from Bailey 1993 have not been included in the graphical analyses because they compared oral antibiotic (single-dose azithromycin) with a combination of topical/oral antibiotic (topical tetracycline with oral erythromycin for severe cases). A total of 194 people with active disease were randomly allocated to treatment, 97 in each group. Approximately 60% of these people were antigen positive at baseline. At 26 weeks 21/97 had active disease in the azithromycin group and 27/97 in the tetracycline/erythromycin group (risk ratio 0.78 95% CI 0.47 to 1.28). Approximately 42% of each group was antigen positive. Data from Cochereau 2007 also have not been included in the meta-analyses because they compared oral azithromycin with two regimens of topical azithromycin and treated people accompanying the children to the treatment centre. They found that trachoma resolved in 93.0%, 96.3% and 9.6.6% of the two-day group, three-day group and oral treatment group 60 days after treatment.

 

What is the effect of community treatment with antibiotics on active trachoma and ocular C. trachomatis infection?

All the cluster-randomised community-based trials discussed in this section compared oral azithromycin to control or oral azithromycin to topical tetracycline.

 

Antibiotic versus control (placebo)

Four community-based trials compared azithromycin versus no treatment (Atik 2006; Chidambaram 2006; Lee 2007; TANA 2009). None of these trials reported outcomes at three months. Atik 2006 published data on active trachoma at 12 months and unpublished data were supplied for TANA 2009 ( Analysis 5.1). The two trials reported very different results. Atik 2006 reported a non-significant increased risk of active trachoma in the community treated with azithromycin (risk ratio 1.14, 95% CI 0.67 to 1.94) whereas in TANA 2009, communities receiving mass treatment with azithromycin had a reduced prevalence of active trachoma in children 12 months after treatment (risk ratio 0.58, 95% CI 0.52, 0.65). It is difficult to explain the differences between the two studies, however, Atik 2006 only compared two communities compared to the 24 randomised in TANA 2009. Of the two studies TANA 2009 was judged to be at lower risk of bias (Figure 2).

These analyses do not take into account the cluster design of the studies. Data from TANA 2009 suggested an intra-cluster correlation coefficient (ICC) of approximately 0.06. Adjusting the results of TANA 2009 for this ICC gave a 95% CI 0.47 to 0.72. In fact the results of this study were reasonably robust to assumptions about the ICC; adjusting for an ICC of 0.2 gave a 95% CI of 0.41 to 0.83.

Atik 2006, Chidambaram 2006, Lee 2007 and TANA 2009 reported C. trachomatis infection at 12 months ( Analysis 5.2). In all four studies communities treated with azithromycin were less likely to have C. trachomatis infection at 12 months compared to untreated communities. These studies gave different estimates of effect (0.61 in Atik 2006, 0.49 in Chidambaram 2006, 0.04 in Lee 2007 and 0.32 in TANA 2009). The pooled risk ratio was 0.35 (95% CI 0.21 to 0.60). Although it is likely that the size of the pooled effect estimate is unreliable, given the differences between the studies, all of the studies indicated a statistically significant beneficial effect of antibiotic treatment on C. trachomatis infection.

The conclusions did not change as a result of adjusting for the extra variation introduced by the cluster design of the studies. Adjusting for an ICC of 0.2 gave a confidence interval for the pooled risk ratio of 0.20 to 0.63.

Data from Resnikoff 1995 are not included in the forest plots as it was difficult to extract data in a suitable format. The study randomly allocated four villages in factorial fashion to treatment with 1% oxytetracycline or health education. Individuals treated with tetracycline experienced a higher cure rate than people who were not and communities treated with tetracycline experienced a lower incidence and prevalence of the disease.

 

Oral versus topical antibiotic

Only one study compared oral and topical community-based treatment for trachoma, the 'Azithromycin in Control of Trachoma' study (ACT). As this study took place in three different countries in Africa (Egypt, The Gambia and Tanzania), it is included in the analyses as three studies.

Even though all three studies had the same interventions and the one study protocol there was still considerable heterogeneity of effect. However, it must be remembered that in two locations only two communities were randomised to oral versus topical treatment (ACT 1999 Egypt; ACT 1999 Tanzania).

Analyses 6.1 to 6.4 show the effect of community-based treatment with azithromycin versus topical tetracycline ( Analysis 6.1;  Analysis 6.2;  Analysis 6.3;  Analysis 6.4). In ACT 1999 Egypt and ACT 1999 The Gambia there was some evidence that azithromycin was more effective than topical tetracycline in reducing the risk of active trachoma and C. trachomatis infection at three and 12 months. However, these results were not very robust to assumptions about the ICC. Adjusting for an ICC of 0.05 resulted in confidence intervals including one for all the results. In ACT 1999 Tanzania, the findings were less consistent, with a risk ratio greater than one (favouring topical treatment) for active trachoma and C. trachomatis infection at 12 months and risk ratio less than one (favouring oral treatment) for C. trachomatis infection at three months.

 

Adverse effects

 Table 5 summarises the information on adverse effects reported in the included studies. Data on adverse effects were sparsely reported. In 12 of the 22 included studies there was no mention of adverse effects in the study report. In TANA 2009 data on adverse effects were collected systematically: 96/671 individuals treated with azithromycin reported an adverse effect of treatment (14.3%, 95% CI 11.7% to 17.2%); 72 of these (10.7% 95% CI 8.5% to 13.3) were gastrointestinal (abdominal pain, vomiting, nausea, diarrhoea, constipation and related issues); no serious adverse events were recorded in this study. A specific analysis of childhood mortality suggested that azithromycin treatment reduced the rate of childhood mortality in these communities. The mortality rate for children aged one to nine years was 4.1 per 1000 person-years (95% CI 3.0 to 5.7) in the treated communities compared to 8.3 per 1000 person-years (95% CI 5.3 to 13.1) in the untreated communities.

 

Discussion

 

Summary of main results

Trials included in this review provide evidence that individuals with trachoma benefit from antibiotic treatment ( Summary of findings for the main comparison). Antibiotic treatment reduces the risk of active trachoma and ocular C. trachomatis infection up to 12 months after treatment. The trials included in this review were clinically and statistically heterogeneous and most had serious limitations in their design. This makes it difficult to estimate the size of the effect; the current best guess would be an approximate 20% risk reduction. Overall the quality of the evidence is graded as low. Oral and topical treatments appeared to have similar effects if used as prescribed ( Summary of findings 2). One trial (Bowman 2000) compared oral antibiotic and unsupervised topical treatment and found the oral antibiotic to be more effective "under practical operational conditions", which may have been due to poor compliance with the more complex topical treatment regimen.

Only three of the more recent trials in individuals used azithromycin, which is the currently recommended oral antibiotic treatment. None of these trials had a no treatment group. However, in the individually-randomised trials there was no evidence that azithromycin was less effective than topical tetracycline.

We identified four community-based trials comparing azithromycin versus no treatment. These trials were of variable quality and size, however there was one large, good quality trial conducted in Ethiopia (TANA 2009) that provided good evidence that community-based treatment with a single dose of azithromycin reduces the prevalence of active trachoma and ocular chlamydial infection in children up to 12 months after treatment ( Summary of findings 3).

Only one trial compared oral versus topical community-based treatment ( Summary of findings 4). This study was conducted in three countries in Africa and therefore is included as three separate studies in this review. Data from this study were inconsistent. In The Gambia and Egypt there was some evidence that oral azithromycin was more effective than topical tetracycline, particularly with regards to ocular infection. However, after adjustment for the cluster design of the study these findings were not statistically significant and were not replicated consistently in the Tanzanian arm of the study.

None of the included trials reported any serious adverse events associated with either of the currently used antibiotics, azithromycin and topical tetracycline. However, for many of the trials it was not clear whether data on adverse effects had been collected systematically. In the one trial that did report these data, between 10% and 15% of people experienced symptoms such as nausea and vomiting with azithromycin treatment.

 

Overall completeness and applicability of evidence

A strength of the evidence is that the included trials come from many different countries and populations. However, it is unfortunate that heterogeneity between trial results means that we cannot estimate with any confidence the size of the effect for treatment of trachoma with oral or topical antibiotics, although it is likely that both oral and topical treatments have a beneficial effect.

Almost all the trials in individuals were done in children and the generaliseability of these findings to adults is uncertain. In the community-based trials, data were reported for adults and children. With the small number of trials it was not possible to determine whether the effects are different in these groups but one study (TANA 2009) provided data on ocular infection after mass treatment in both children and adults. The observed risk ratio in children was 0.32 (95% CI 0.26 to 0.40) and in adults was 0.49 (0.33 to 0.71).

Evidence for community-based or mass treatment campaigns is sparse. It was not possible to determine who should be treated on the basis of available data, what the important factors are in planning treatment strategies and whether it is the whole community, all children under 10 years of age, all women and children or families of all children with active trachoma. There is some evidence that frequent treatment of children may be an effective strategy to reduce the community prevalence of infection (TANA 2009).

Azithromycin was given in the trials as a single, double or triple dose but it was not possible to determine whether there was any difference in effect. Where azithromycin is not donated, there is a major cost difference between topical tetracycline and oral azithromycin, but it was not possible to determine which is the more cost-effective strategy per extra case cured. Some populations in which trachoma is endemic are subject to migration and that may account in part for the low follow-up rates in the community trials; it may also have implications in determining the most effective treatment in those populations where new infected cases migrate into the community.

Our review does not directly address the evidence for the WHO guidelines. In part this is because we did not identify any trial data that directly tested the efficacy of the mass antibiotic administration schedules currently recommended. These are that where the baseline district-level prevalence of TF in one to nine year-old children is 10% or greater, mass antibiotic treatment should be undertaken annually for three years before a repeat district-level survey (Solomon 2006).

 

Quality of the evidence

The included trials were published from 1966 onwards and their quality was variable. The quality of evidence for most outcomes was low, particularly for the comparison antibiotics versus no treatment ( Summary of findings for the main comparison). Reporting of sequence generation and allocation concealment was not good and it was often difficult to assess the effect of incomplete data due to inadequate reporting. There was considerable heterogeneity of results. However, masking of outcome assessment was reported for laboratory analyses (less so for clinical assessments of active trachoma) and there was little evidence of selective outcome reporting. There was moderate quality evidence for the comparison of oral and topical antibiotics for the outcome active trachoma ( Summary of findings 2).

The community-based trials were also of poor methodological quality with the exception of one study (TANA 2009) ( Summary of findings 3;  Summary of findings 4). The main problem with the included studies was that in most of the studies only two communities were randomly allocated to treatment (only two trials randomised sufficient number of clusters). Although adjustment for baseline characteristics can alleviate this problem to some extent, the interpretation of these studies is always problematic as it is difficult to exclude the alternative explanation that there is some characteristic that is different between the communities (apart from treatment of trachoma) and which may be the real cause of any observed differences in outcome. There was also little information on other potential sources of bias in cluster-randomised trials such as recruitment bias.

Three community-based trials (Chidambaram 2006; Lee 2007; TANA 2009) had a 'delayed treatment' design that involved randomly selecting clusters for treatment and comparing the prevalence of trachoma 12 months after treatment with a random selection of untreated clusters that are then enrolled in the treatment programme. This study design overcomes the ethical dilemma of surveying communities for trachoma and then withholding treatment for 12 months but has the disadvantage that baseline data on trachoma are not available in the control group.

 

Potential biases in the review process

This review has been substantially revised for the update. New methods, such as assessment of risk of bias and subgroup and sensitivity analyses, have been incorporated. A new protocol was not written. It is possible that the update could have been influenced by knowledge of the trial results.

 

Agreements and disagreements with other studies or reviews

This update, like previous versions of the Cochrane review of antibiotics for trachoma (Mabey 2002; Mabey 2005), found some evidence of benefit of treatment of individuals with clinical signs of active trachoma, but only limited evidence to support the use of oral azithromycin in preference to topical tetracycline. In contrast to those previous Cochrane reviews, we found some good quality evidence demonstrating the effectiveness of community-based treatment, with some limited evidence of greater benefit of mass treatment with azithromycin over mass treatment with topical tetracycline.

 

Authors' conclusions

 

Implications for practice

Whilst the data are not of sufficient quality to make firm conclusions, there is some evidence that people treated with either oral or topical antibiotics may experience a reduction in the risk of active trachoma (perhaps of the order of 20% relative risk reduction). It is likely that oral azithromycin and topical tetracycline have similar effects if used as prescribed.

Community-based trials show that mass administration of antibiotics reduces the prevalence of active trachoma and ocular infection with C.trachomatis. There were no trial data that directly tested the mass antibiotic administration schedules currently recommended by WHO.

 

Implications for research

The Alliance for the Global Elimination of Trachoma has endorsed the donation of azithromycin for the treatment of trachoma. This would be an ideal setting in which to conduct community-randomised trials, under operational conditions,  comparing the effect of mass distribution of azithromycin to that of placebo or no treatment. Opportunities for ethically conducting such trials occur in countries and districts newly enrolling in trachoma control programmes. Inequities are bound to exist in some settings at start-up, when antibiotics and resources for their distribution are generally in limited supply. Allocating interventions randomly in these circumstances is reasonable, with roll-out of the intervention to areas initially randomised to 'control' in later treatment rounds. Such an approach has been used in several of the trials included in this review. Trials to determine optimal dosages and dosage intervals of azithromycin at various levels of endemicity, test the most appropriate thresholds for starting and stopping mass treatment, and to determine which subgroups will need to be treated at various stages of the pathway towards elimination are also required; where potential strategies to evaluate could be selected on the basis of recent mathematical modelling work. Cost effectiveness per extra case cured should be one of the outcome measures. The adverse effects of azithromycin and emergence of resistance are also areas that should be addressed.

 

Acknowledgements

The authors would like to thank:
The Systematic Review Training Unit at the Institute of Child Health and Neal Alexander for statistical input;
Hugh Taylor and David Mabey for peer review comments on earlier versions of this review;
The Cochrane Eyes and Vision Group for developing and executing the electronic searches;
Tom Lietman for responding to queries about two included trials (Chidambaram 2006; TANA 2009) and providing unpublished data on active trachoma for TANA 2009; and
Sharon Haymes for comments on the 2011 update.

 

Data and analyses

Download statistical data

 

Comparison 1. Any antibiotic versus control (individuals)

---

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Active trachoma at three months	9	1961	Risk Ratio (M-H, Random, 95% CI)	0.78 [0.69, 0.89]
2 Ocular C.trachomatis infection at three months	4	297	Risk Ratio (M-H, Random, 95% CI)	0.81 [0.63, 1.04]
3 Active trachoma at 12 months	4	1035	Risk Ratio (M-H, Random, 95% CI)	0.74 [0.55, 1.00]
4 Ocular C. trachomatis infection at 12 months	1		Risk Ratio (M-H, Fixed, 95% CI)	Subtotals only

---

 

Comparison 2. Subgroup analysis: oral and topical antibiotics versus control (individuals)

---

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

---

 1 Active trachoma at three months

9

Risk Ratio (M-H, Random, 95% CI)

Subtotals only

---

1.1 Oral antibiotic

6

599

Risk Ratio (M-H, Random, 95% CI)

0.81 [0.67, 0.97]

---

1.2 Topical antibiotic

6

1478

Risk Ratio (M-H, Random, 95% CI)

0.82 [0.72, 0.92]

---

 2 Ocular C. trachomatis infection at three months

4

Risk Ratio (M-H, Fixed, 95% CI)

Subtotals only

---

2.1 Oral antibiotic

4

259

Risk Ratio (M-H, Fixed, 95% CI)

0.85 [0.66, 1.11]

---

2.2 Topical antibiotic

1

85

Risk Ratio (M-H, Fixed, 95% CI)

0.18 [0.02, 1.37]

---

 3 Active trachoma at 12 months

4

Risk Ratio (M-H, Fixed, 95% CI)

Subtotals only

---

3.1 Oral antibiotic

3

429

Risk Ratio (M-H, Fixed, 95% CI)

0.87 [0.76, 1.00]

---

3.2 Topical antibiotic

4

724

Risk Ratio (M-H, Fixed, 95% CI)

0.79 [0.71, 0.88]

---

 4 Ocular C. trachomatis infection at 12 months

1

Risk Ratio (M-H, Fixed, 95% CI)

Subtotals only

---

4.1 Oral antibiotic

1

91

Risk Ratio (M-H, Fixed, 95% CI)

0.36 [0.10, 1.23]

---

4.2 Topical antibiotic

1

85

Risk Ratio (M-H, Fixed, 95% CI)

0.14 [0.02, 1.04]

---

 

Comparison 3. Oral versus topical antibiotics (individuals)

---

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Active trachoma at three months	6	953	Risk Ratio (M-H, Random, 95% CI)	0.98 [0.82, 1.18]
2 Ocular C. trachomatis infection at three months	3		Risk Ratio (M-H, Fixed, 95% CI)	Totals not selected
3 Active trachoma at 12 months	5	886	Risk Ratio (M-H, Random, 95% CI)	0.93 [0.75, 1.15]
4 Ocular C. trachomatis infection at 12 months	2		Risk Ratio (M-H, Fixed, 95% CI)	Totals not selected

---

 

Comparison 4. Oral azithromycin versus topical tetracycline (individuals)

---

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Active trachoma at three months	3		Risk Ratio (M-H, Fixed, 95% CI)	Totals not selected
2 Ocular C. trachomatis infection at three months	2		Risk Ratio (M-H, Random, 95% CI)	Totals not selected
3 Active trachoma at 12 months	2	447	Risk Ratio (M-H, Fixed, 95% CI)	0.76 [0.59, 0.99]
4 Ocular C. trachomatis infection at 12 months	1		Risk Ratio (M-H, Fixed, 95% CI)	Totals not selected

---

 

Comparison 5. Oral azithromycin versus control (communities)

---

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Active trachoma at 12 months	2		Risk Ratio (M-H, Random, 95% CI)	Totals not selected
2 Ocular C. trachomatis infection at 12 months	4	4345	Risk Ratio (M-H, Random, 95% CI)	0.35 [0.21, 0.60]

---

 

Comparison 6. Oral azithromycin versus topical tetracycline (communities)

---

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Active trachoma at 3 months	3		Risk Ratio (M-H, Random, 95% CI)	Totals not selected
2 Ocular C. trachomatis infection at 3 months	3		Risk Ratio (M-H, Random, 95% CI)	Totals not selected
3 Active trachoma at 12 months	3		Risk Ratio (M-H, Random, 95% CI)	Totals not selected
4 Ocular C. trachomatis infection at 12 months	3		Risk Ratio (M-H, Random, 95% CI)	Totals not selected

---

 

Appendices

 

Appendix 1. CENTRAL search strategy

#1 MeSH descriptor Trachoma
#2 MeSH descriptor Chlamydia trachomatis
#3 trachoma* or tracoma*
#4 (#1 OR #2 OR #3)
#5 MeSH descriptor Anti-Bacterial Agents
#6 antibiotic*
#7 MeSH descriptor Azithromycin
#8 azithrom*cin*
#9 MeSH descriptor Tetracycline
#10 tetracycline*
#11 MeSH descriptor Chlortetracycline
#12 chlortetracycline*
#13 MeSH descriptor Macrolides
#14 macrolide*
#15 (#5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14)
#16 (#4 AND #15)

 

Appendix 2. MEDLINE search strategy

1 randomized controlled trial.pt.
2 (randomized or randomised).ab,ti.
3 placebo.ab,ti.
4 dt.fs.
5 randomly.ab,ti.
6 trial.ab,ti.
7 groups.ab,ti.
8 or/1-7
9 exp animals/
10 exp humans/
11 9 not (9 and 10)
12 8 not 11
13 exp trachoma/
14 trac?oma$.tw.
15 exp chlamydia trachomatis/
16 or/13-15 (14120)
17 exp antibacterial agents/
18 antibiotic$.tw.
19 exp azithromycin/
20 azithrom?cin$.tw.
21 exp tetracycline/
22 tetracycline$.tw.
23 exp chlortetracycline/
24 chlortetracycline$.tw.
25 exp macrolides/
26 macrolide$.tw.
27 or/17-26
28 16 and 27
29 12 and 28

The search filter for trials at the beginning of the MEDLINE strategy is from the published paper by Glanville (Glanville 2006).

 

Appendix 3. EMBASE search strategy

1 exp randomized controlled trial/
2 exp randomization/
3 exp double blind procedure/
4 exp single blind procedure/
5 random$.tw.
6 or/1-5
7 (animal or animal experiment).sh.
8 human.sh.
9 7 and 8
10 7 not 9
11 6 not 10
12 exp clinical trial/
13 (clin$ adj3 trial$).tw.
14 ((singl$ or doubl$ or trebl$ or tripl$) adj3 (blind$ or mask$)).tw.
15 exp placebo/
16 placebo$.tw.
17 random$.tw.
18 exp experimental design/
19 exp crossover procedure/
20 exp control group/
21 exp latin square design/
22 or/12-21
23 22 not 10
24 23 not 11
25 exp comparative study/
26 exp evaluation/
27 exp prospective study/
28 (control$ or prospectiv$ or volunteer$).tw.
29 or/25-28
30 29 not 10
31 30 not (11 or 23)
32 11 or 24 or 31
33 exp trachoma/
34 trac?oma$.tw.
35 exp chlamydia trachomatis/
36 or/33-35
37 exp antibiotic agent/
38 antibiotic$.tw.
39 exp azithromycin/
40 azithrom?cin$.tw.
41 exp tetracycline/
42 tetracycline$.tw.
43 exp chlortetracycline/
44 chlortetracycline$.tw.
45 exp macrolide/
46 macrolide$.tw.
47 or/37-46
48 36 and 47
49 32 and 48

 

Appendix 4. metaRegister of Controlled Trials search strategy

trachoma and antibiotics

 

Appendix 5. ClinicalTrials.gov search strategy

Trachoma AND Antibiotics

 

Appendix 6. Changes made to data in the 2011 update of the review

---

			Current review				Original review
			Treatment		Control		Treatment		Control		Comments
Comparison		Study	n	N	n	N	n	N	n	N
1.1	Active trachoma at 3 months	Peach 1986	216	340	138	189	284	408	182	233	Missing data counted twice in original review
1.3	Active trachoma at 12 months	Woolridge 1967	121	202	95	120	149	202	100	120	Error in data extraction in original review
2.1	Active trachoma at 3 months	Peach 1986	216	340	138	189	284	408	182	233	Missing data counted twice in original review
2.3	Active trachoma at 12 months	Woolridge 1967	121	202	95	120	149	202	100	120	Error in data extraction in original review
3.1	Active trachoma at 3 months	Bowman 2000	48	152	68	139	56	158	83	156	People with missing data counted as having trachoma in original review
3.1	Active trachoma at 3 months	Shukla 1966	53	83	29	62	53	83	34	42	Error in data extraction in original review
3.1	Active trachoma at 3 months	Tabbara 1996	15	31	12	29	15	32	12	32	In the original review, people who were not followed up were included in the denominator. This makes the assumption that people who were not followed up had inactive trachoma.
3.2	Chlamydia trachomatis infection at 3 months	Tabbara 1996	6	30	4	26					Not included in previous review
3.3	Active trachoma at 12 months	Bowman 2000	32	149	45	139					Not included in previous review
3.4	Chlamydia trachomatis infection at 12 months	Dawson 1997	8	105	5	33	7	105	5	33	Error in data extraction in original review
5.1	Active trachoma at 12 months	Atik 2006	21	523	35	994					Not included in previous review
5.2	Chlamydia trachomatis infection at 12 months	Atik 2006	23	659	68	1192					Not included in previous review
5.2	Chlamydia trachomatis infection at 12 months	Lee 2007	2	170	56	185					Not included in previous review
		ACT study	Data for the ACT trial in the original review was not exactly the same as the published data and included unpublished outcomes. The original review authors had access to individual patient data which was not available to the current authors. In the absence of access to the original data, we felt it was unwise to make any changes to the data included in the review.

---

 

Appendix 7. Results of searches for previous versions of the review

The original electronic searches identified 566 reports of studies, of which 51 reported antibiotic treatment trials for trachoma. A total of 15 studies (8678 participants) met the inclusion criteria. Ten trials compared antibiotic to placebo or no treatment (Attiah 1973; Darougar 1980b; Dawson 1969i; Dawson 1969ii; Foster 1966; Hoshiwara 1973; Peach 1986; Shukla 1966; Tabbara 1988; Woolridge 1967). One trial (Tabbara 1988) was later excluded as it was not possible to identify patient outcomes as both eyes of the same patient were in some instances used in the randomisation and the results were reported as eyes not patients. The citations Dawson 1969i and Dawson 1969ii refer to two arms of the same trial, which were conducted in different schools; as the results are reported separately in the paper they have been treated as separate studies. Some of the above studies reported the comparison of topical against oral antibiotics. A further six trials compared topical tetracycline to oral azithromycin (Bowman 2000; Dawson 1997; Schachter 1999i; Schachter 1999ii; Schachter 1999iii; Tabbara 1996). The three Schachter 1999 citations used the same protocol but applied in different countries and reported in the same article. Schachter 1999i refers to results from Egypt, Schachter 1999ii from The Gambia and Schachter 1999iii from Tanzania. One further trial was excluded as an oral antibiotic, erythromycin, was used in conjunction with topical tetracycline in severe cases of trachoma, the comparison being oral azithromycin (Bailey 1993b).
The electronic searches were updated in 2005 and 206 new reports of studies were identified from the electronic searches. Hard copies of two reports were obtained for further scrutiny. One study was excluded as the trial did not take place in a trachoma endemic region (Isenberg 2002) and the other study by Humet 1989 was excluded as it did not assess the treatment of ocular Chlamydia trachomatis.

 

What's new

Last assessed as up-to-date: 11 December 2010.

---

Date	Event	Description
18 February 2011	New search has been performed	Issue 3 2011: Electronic searches updated and 6 new trials included. Risk of bias assessed for all studies and summary of findings tables added.
9 February 2011	New citation required and conclusions have changed	Issue 3 2011: Review substantively updated. New authorship.

---

 

History

Protocol first published: Issue 3, 1999
Review first published: Issue 1, 2002

---

Date	Event	Description
31 March 2008	Amended	Converted to new review format.
1 February 2005	New citation required and conclusions have changed	Substantive amendment

---

 

Contributions of authors

Original version of the review: Denise Mabey (DM) screened the search results, graded selected trials, extracted some data and wrote the review. DM was the guarantor for the review. Nicole Fraser-Hurt (NF) graded selected trials, extracted the data and contributed to writing the review.Chistine Powell screened the search results and worked on the update of the review.

Major update of review and change of authors 2011: JE and AWS screened search results, assessed the risk of bias of all included studies, extracted data from new included trials and substantially rewrote the text of the review.

The Cochrane Eyes and Vision Group editorial team developed the search strategies and undertook the electronic searches.

 

Declarations of interest

The Edna McConnell Clark Foundation supported DM and NF for one half day a week over a 10 month period to undertake the original review. SightSavers International part funded JE's salary to update the review. AWS is a member of the International Trachoma Initiative (ITI)'s Trachoma Expert Committee and has received research support from both ITI and Pfizer, the manufacturers of azithromycin.

 

Sources of support

 

Internal sources

• No sources of support supplied
  

 

External sources

• The Edna McConnell Clark Foundation, USA.
  
• Christian Blind Mission, Germany.
  
• Sightsavers International, UK.
  

 

Differences between protocol and review

In 2011 this review was substantially revised. The major change was that we considered separately the individually-randomised and cluster-randomised trials, and the Cochrane Collaboration's tool for assessing risk of bias was implemented. The greater detail required in Revman 5 format has meant that we have completed some aspects of the methods - such as assessment of heterogeneity - for this update which were not discussed in detail in the original protocol.

 

Notes

This review was first published as Mabey D, Fraser-Hurt N. Antibiotics for trachoma. Cochrane Database of Systematic Reviews 2002, Issue 1. Art. No.: CD001860. DOI: 10.1002/14651858.CD001860.pub2. The 2011 updated version has been written by a new review team.

 

Index terms

Medical Subject Headings (MeSH)

*Chlamydia trachomatis; Administration, Oral; Administration, Topical; Anti-Bacterial Agents [administration & dosage; *therapeutic use]; Azithromycin [administration & dosage]; Randomized Controlled Trials as Topic; Tetracycline [administration & dosage]; Trachoma [*drug therapy]

MeSH check words

Humans

* Indicates the major publication for the study