Study of Azithromycin to Prevent Recurrent Trichiasis Following Surgery in Ethiopia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of Azithromycin to Prevent Recurrent Trichiasis Following Surgery in Ethiopia (NCT00347776), a randomized phase 4 study. The primary outcome is Recurrent Trichiasis in Tetracycline and Azithromycin Groups.

2. Clinical Question

The trial addresses recurrence after surgery, rather than the initial correction of trichiasis alone. Its registry description says that recurrence was identified by clinical assessment for one or more eyelashes touching the globe or evidence of epilation. The comparison pooled participants from the two azithromycin arms, so the posted primary result does not distinguish treatment of the participant alone from treatment of the participant and immediate family members.

The registry describes the objective as determining the impact of post-surgical treatment on recurrence at one and two years. However, the posted primary outcome timeframe lists assessments through 12 months, and the reported primary analysis is a time-to-event analysis. The registry record does not report a separate posted primary result at two years. The result should therefore be read in the context of the outcome and follow-up information actually posted, rather than as an answer to every time point mentioned in the study objective.

3. Trial Design

This was a randomized, single-masked, parallel-group Phase 4 trial sponsored by Johns Hopkins University. The registry lists the condition as trichiasis, an enrollment of 1452, and a completed status. The record lists the study start as 2001-08 and primary completion as 2003-12.

Participants undergoing trichiasis surgery were assigned to a control arm receiving topical tetracycline or to azithromycin treatment. The registry lists two azithromycin arms: a single 1g oral dose to the participant, and a single oral dose to the participant and immediate family members. For the posted primary comparison, those two intervention arms were combined, giving a comparison of the combined azithromycin groups with the tetracycline control group.

“Randomized” indicates that treatment assignment was by chance, while “parallel” means participants were assigned to a treatment group for the comparison rather than crossing between treatments. “Single-masked” indicates that the registry records masking of one party, but it does not identify which party was masked. Further details of the masking procedure are not reported in the registry record.

The primary outcome was recurrence of trichiasis, assessed clinically at 2 weeks, 1.5 months, 6 months and 12 months after surgery. The registry also lists surgical failure and adverse events as secondary outcomes. The record does not report the randomization procedure, allocation concealment, the identity of the masked party, the original sample-size justification, or the detailed protocol. Those details should not be inferred from the trial’s completed status or enrollment.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

1.5 Month Follow-up

Reasons for not completing this period:

6 Month Follow-up

Reasons for not completing this period:

12 Month Follow-up

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Recurrence rates (expressed as person-years) in the tetracycline arm was compared with the 2 azithromycin arms combined (483+ 485= 968 participants)

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Recurrent Trichiasis in Tetracycline and Azithromycin Groups, measured over Primary outcome assessed at 2 weeks,1.5 months, 6 months and 12 months post-surgery.

The primary endpoint concerns recurrent trichiasis after surgery. The registry defines recurrence through clinical assessment: one or more eyelashes touching the globe, or evidence of epilation. This definition links the endpoint to signs observed in the eye, rather than to a laboratory measure or a participant-reported symptom alone.

Recurrence is relevant to the trial question because the study compares post-surgical antibiotic strategies in people who have had surgery for trichiasis. An endpoint that records whether recurrence occurs can be used to compare the groups’ observed outcomes. The registry lists assessments at 2 weeks, 1.5 months, 6 months and 12 months after surgery, and its reported primary analysis uses a log-rank method and a Cox proportional-hazards parameter. These methods account for the timing of events, rather than treating every participant’s follow-up as if it were identical.

The posted arm results are counts of participants with an event: 45 of 484 in the tetracycline control and 62 of 968 in the combined azithromycin group. The registry’s outcome description also refers to recurrence rates expressed as person-years. The posted record does not provide the event-time data, person-time denominators, or a full account of how the counts relate to the time-to-event analysis. The counts and the time-to-event estimate therefore describe related aspects of the outcome, but they are not interchangeable summaries.

The registry objective mentions one- and two-year recurrence, while the listed primary outcome timeframe ends at 12 months. A separate two-year primary endpoint result is not reported in the registry record.

7. Results Posted in the Registry

• Intervention: 6.4% (62/968) versus 9.3% (45/484) on control; risk difference -2.9 percentage points (95% CI -5.9 to 0.1), risk ratio 0.69 (0.48–1.00); unadjusted two-proportion z-test p = 0.047.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Surgical Failure

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Intervention: 2.1% (20/968) versus 2.5% (12/484) on control; risk difference -0.4 percentage points (95% CI -2.1 to 1.2), risk ratio 0.83 (0.41–1.69); unadjusted two-proportion z-test p = 0.613.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Intervention and the control. The interval gives the range of effects that remain compatible with the data.

The registry lists surgical failure and adverse events as secondary outcomes, and it posts a result for surgical failure at 6 weeks. Across the posted secondary information, there is no broad, clearly consistent pattern that independently confirms or overturns the primary comparison. These outcomes concern additional aspects of the post-surgical experience and should be interpreted as supporting context, not as substitutes for the primary endpoint.

The surgical-failure analysis is reported as logistic regression with an odds ratio. The adverse-event record lists event counts but does not provide a corresponding comparative statistical analysis. The registry also lists four outcome measures and three analyses overall, while only two secondary outcomes are posted in the secondary outcome summary. The scope and relationship of every registered measure to every reported analysis are not fully explained in the registry record.

General explanation: when multiple outcomes are examined, the chance of finding an apparently notable result by chance can increase if each is assessed separately. A multiplicity adjustment or a prespecified hierarchy can address this issue, but neither a multiplicity procedure nor an outcome hierarchy is reported in the registry record. Unadjusted secondary results therefore carry less weight than a clearly prespecified primary analysis, especially when the number of planned comparisons and the analysis plan are not fully documented.

The secondary outcomes can help describe what was measured beyond recurrence, but the posted information does not justify treating any one secondary finding as definitive evidence about the primary question. The record does not report a complete analysis population for each secondary outcome, or enough detail to establish how missing observations were handled. Those omissions limit how confidently the secondary results can be integrated with the primary analysis.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Surgical Failure
• Adverse Events

9. Statistical Interpretation

10. Statistical Methods Explained

The primary registry analysis is described as log-rank, with a Cox proportional-hazards parameter. The posted estimate is 0.67, its 95% confidence interval is 0.45 to 0.98, and the reported p-value is .047. A separate unadjusted recalculation uses the posted participant counts and reports a risk difference of -0.029, a 95% confidence interval from -0.059 to 0.001, and p = 0.047. These are distinct methods and effect measures, so their estimates and intervals need not have the same numerical form.

What does a time-to-event method compare?

General explanation: A time-to-event method considers both whether an event occurred and when it occurred during follow-up. A log-rank test compares the event patterns over time between groups. A Cox proportional-hazards model expresses that comparison as a hazard ratio. The hazard is the instantaneous event rate among participants still at risk at a given time; it is not the same as the proportion of all participants who experienced an event. The registry record does not post participant-level event times, censoring details, or diagnostics for the model.

The registry lists multiple assessment times and describes recurrence rates in person-years. The reported log-rank and Cox analysis is consistent with an analysis of event timing, but the record does not describe the detailed time origin, censoring rules, or handling of participants without an observed event. Those procedures are not reported in the registry record.

What does the unadjusted risk difference mean?

General explanation: A risk difference compares the observed event proportions in two groups by subtracting one group’s proportion from the other, using a stated comparison direction. It is expressed on the absolute scale of event proportions. The recalculation uses the posted counts: 45 events among 484 control participants and 62 among 968 participants in the combined intervention group. Its risk difference is -0.029. This is a simple unadjusted summary of the posted counts; it does not incorporate event timing or covariate adjustment.

A confidence interval gives a range of values compatible with the data and the method’s assumptions. The recalculated 95% interval, from -0.059 to 0.001, crosses zero. General explanation: zero is the null value for a risk difference, representing equal event proportions. The p-value of 0.047 is also from the unadjusted recalculation. A p-value describes how unusual data at least as incompatible with the null hypothesis would be under that hypothesis and the analysis assumptions; it is not the probability that the null hypothesis is true.

Why do the estimates differ in form?

The sponsor reports a hazard ratio from a time-to-event analysis, whereas the recalculation reports a risk difference from event counts. General explanation: a hazard ratio compares event rates over follow-up among participants still at risk, while a risk difference compares event proportions over the summarized observation period. Different measures answer related but non-identical questions. A difference between their confidence intervals is not, on its own, evidence of an error.

The sponsor’s hazard ratio confidence interval, 0.45 to 0.98, does not include 1, the null value for a hazard ratio. The recalculated risk-difference interval crosses zero. Their p-values are similar as posted, but this does not make the analyses equivalent. The registry does not report enough detail to reproduce the sponsor’s time-to-event analysis from the arm-level counts.

Why can event timing and over-dispersion matter?

General explanation: If participants have different lengths of follow-up, a count of events alone may not capture the full time-at-risk pattern. Time-to-event methods can account for different observed event times and censored follow-up, provided those data and the analysis rules are available. Over-dispersion is a concern in count models when the observed variation exceeds what the assumed distribution allows; it can make uncertainty estimates too small if ignored. The primary endpoint is posted as participant event counts and analyzed by a time-to-event method, not as an over-dispersed count outcome. No over-dispersion parameter is reported in the registry record, and it is not appropriate to infer one.

The registry’s primary result should therefore be read as a reported time-to-event comparison, alongside a simpler recalculation based on the posted counts. The latter offers a transparent absolute-scale summary, but it cannot reconstruct or supersede the sponsor’s analysis. The trial’s original statistical plan, including any covariate adjustments, model checks, or planned interim analyses, is not reported in the registry record.

11. Confidence Intervals and Precision

The sponsor’s primary time-to-event estimate is 0.67, with a 95% confidence interval from 0.45 to 0.98. General explanation: a confidence interval describes the range of effect values reasonably compatible with the observed data under the method’s assumptions. For a hazard ratio, 1 represents no relative difference in hazards. The reported interval lies below 1, but its upper limit is close to that null value. This indicates that the estimate is not highly separated from the null on the hazard-ratio scale.

The unadjusted recalculation reports a risk difference of -0.029, with a 95% confidence interval from -0.059 to 0.001. For a risk difference, zero is the null value. Because this interval includes zero, the posted counts are compatible with a range that includes no difference in event proportions. Its endpoints also permit effects in opposite directions. The interval’s width is important: the point estimate alone does not convey the uncertainty in the comparison.

These intervals apply to different effect measures and analysis methods. The sponsor’s interval is for a hazard ratio derived from a time-to-event analysis; the recalculated interval is for an unadjusted difference in event proportions. They should not be compared as if they were intervals for the same quantity. The registry does not post the event-time and person-time information that would permit an independent reconstruction of the sponsor’s interval.

Confidence intervals quantify statistical uncertainty under an analysis; they do not account for every possible source of uncertainty, such as limitations in measurement, follow-up, or applicability to other settings. The record does not report details needed to assess all such sources. The posted estimates therefore offer a bounded but incomplete picture of precision.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size figures are planning illustrations based on the observed effect, not the trial’s original design calculations. The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each row considers a stated fraction of the observed effect: 100%, 75%, or 50%. The original trial’s own sample-size assumptions and justification are not reported in the registry record.

For the binary participant-event outcomes, the figures are participants per arm. The planning note specifies that time-to-event planning tables, if used, give total numbers of events instead; the primary outcome’s planning figures are participants per arm. The 80% power column gives the number of participants per arm the planning calculation would require to have an 80% chance of detecting the specified effect under its assumptions. The 90% power column gives the corresponding number for a higher chance of detection. These are planning targets, not observed trial outcomes.

Smaller effects are harder to distinguish from random variation, so planning for a smaller fraction of the observed effect requires more participants per arm. The row for 75% of the observed effect represents a more modest target effect than the row for the full observed effect; the 50% row represents a still smaller target. The increasing sample-size requirements across those rows illustrate this general relationship.

The curve should be read as a visual representation of the same planning relationship between a target effect and required sample size, under the stated assumptions. It does not show how much power the completed trial actually achieved. This is not observed power. The observed enrollment and outcome data are separate from these hypothetical planning calculations, and the figures do not establish that the original trial was designed using the observed effect.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.093; treatment proportion p₁ = 0.064; average p̄ = 0.079.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.3804 = 0.7456.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.3798 = 0.3197.
4. n = (0.7456 + 0.3197)² / (-0.0289)² = 1,356.2, rounded up to 1,357 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.0929752066115702                          # control proportion
p1 <- 0.0640495867768595                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(347776)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 1,357 (80% power), 1,816 (90% power)
 75% of observed effect: n per arm = 2,514 (80% power), 3,364 (90% power)
 50% of observed effect: n per arm = 5,882 (80% power), 7,874 (90% power)
power.prop.test: 1356.2 
Simulated power at n = 1,357 per arm: 0.805 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The sponsor’s primary analysis is reported as a log-rank comparison with a Cox proportional-hazards parameter. It gives a hazard ratio of 0.67, a 95% confidence interval from 0.45 to 0.98, and p = .047. The separate recalculation is unadjusted and uses the posted participant counts; it gives a risk difference of -0.029, a 95% confidence interval from -0.059 to 0.001, and p = 0.047.

The recalculation is not a reproduction of the sponsor’s method. It does not use the event timing underlying the time-to-event analysis, and it does not estimate a hazard ratio. The sponsor’s own result should be treated as the trial’s reported analysis; the registry’s discrepancy indicator is false. The difference between the confidence intervals reflects, in part, that the analyses concern different effect measures and use different information.

The registry record does not provide enough information to independently reproduce the sponsor’s analysis. In particular, it does not post participant-level event times, person-time, or full analysis procedures. The unadjusted result is therefore useful as a count-based companion summary, not as a replacement for the sponsor’s reported time-to-event result.

15. Multiplicity, Missing Data and Analysis Populations

The registry identifies recurrence as the primary outcome and lists surgical failure and adverse events as secondary outcomes. It also records four outcome measures and three analyses overall, while the secondary outcome summary posts two outcomes. The registry record does not report a complete statistical analysis plan, a prespecified hierarchy among outcomes, or a correction for multiple comparisons.

General explanation: examining several outcomes can increase the likelihood that at least one appears statistically notable by chance, even when each test is considered separately. A multiplicity strategy can control this issue, and a prespecified primary outcome can help distinguish the main confirmatory question from secondary or exploratory questions. Without the plan, it is not possible to determine whether the posted secondary analyses were planned in advance, whether additional analyses were conducted, or how the investigators intended to interpret multiple results.

The record does not describe a participant flow by analysis, the number of participants with available outcome data at each assessment, or the handling of missing observations and loss to follow-up. It also does not report an intention-to-treat definition, a per-protocol analysis population, or how participants were handled if treatment assignment or follow-up differed from plan. These matters are relevant to understanding who contributed to the reported outcome counts and time-to-event analysis.

The registry lists the primary outcome assessment times as 2 weeks, 1.5 months, 6 months and 12 months after surgery, but does not give the number at risk at each time or explain the censoring rules used for the Cox analysis. The objective mentions one- and two-year recurrence, while a separate two-year primary result is not posted. Missing follow-up details should not be filled in by assumptions about standard practice.

Interim analyses, stopping rules, subgroup analyses, covariate adjustment, and procedures for missing data are not reported in the registry record. Their absence from the record is a reporting limitation; it does not establish that such procedures were or were not used.

16. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent non-serious adverse event terms

The adverse-event figures list 0 serious adverse events among 484 control participants and 0 among 968 participants in the combined intervention group. The record also lists 13 other adverse events in the control group and 28 in the combined intervention group. Deaths are listed as 2 in the control group and 0 in the intervention group.

These are counts, not rates adjusted for follow-up time, and the group sizes differ. The registry does not provide definitions for “other” adverse events, event-level descriptions, timing, severity grading beyond the serious classification, or a comparative statistical analysis of these counts. It also does not explain how many participants were assessed for adverse events over the full follow-up period.

Zero recorded serious adverse events in these groups means none are listed in the posted record; it does not establish that the risk of serious events is absent. The counts alone do not establish that treatment caused or prevented an adverse event or death. They should be read as the safety information posted in the registry, with the reporting and follow-up details that remain unavailable.

17. What the Result Does, and Does Not, Mean

The posted results describe a comparison of recurrent trichiasis after surgery between topical tetracycline and two azithromycin groups combined. The primary registry analysis uses a log-rank method and reports a Cox proportional-hazards estimate of 0.67, with a 95% confidence interval from 0.45 to 0.98 and p = .047. The unadjusted count-based recalculation reports a risk difference of -0.029, with a 95% confidence interval from -0.059 to 0.001 and p = 0.047.

These findings do not mean that every person in one group had the same outcome, nor do they establish what would happen for an individual. Statistical estimates describe comparisons between groups under their analysis assumptions. The time-to-event estimate and the count-based recalculation use different measures and should be interpreted separately. The uncertainty intervals indicate that the magnitude of the comparison is not known precisely.

The trial record does not report a separate two-year primary result, although the study objective mentions recurrence at one and two years. The posted primary outcome timeframe lists assessments through 12 months. The registry also does not provide detailed information about missing data, analysis populations, or all procedures used in the primary model. Those gaps constrain interpretation of the reported estimates.

The results do not establish that one treatment works or does not work, and they are not medical advice. They describe the comparison and outcomes posted for this trial. Applying the findings beyond the study population, procedures, treatments, or follow-up requires evidence and context not contained in the registry record.

18. Why This Trial Matters Statistically

This trial illustrates why an effect estimate should be read together with its endpoint, analysis method, and uncertainty. The same primary comparison is presented through a time-to-event analysis and an unadjusted comparison of participant counts. The former uses a hazard-ratio scale and the latter a risk-difference scale. These methods answer related but distinct statistical questions, so a reader should not expect their confidence intervals to match or treat one as a direct replication of the other.

The registry also makes the endpoint definition important. Recurrence is based on clinical signs, while the analysis is described as time-to-event and the outcome text refers to person-years. Counts alone do not reveal when recurrence occurred or how much follow-up each participant contributed. This distinction matters when interpreting a result from a trial with repeated assessments and a time-to-event analysis.

The trial’s two azithromycin strategies are combined for the posted primary comparison. That choice means the primary estimate speaks to the pooled azithromycin arms versus tetracycline, rather than providing a separate estimate for each azithromycin strategy. It is a useful reminder to check what groups an estimate actually represents.

Finally, planning sample sizes based on an observed effect are illustrations, not evidence of the completed trial’s power or of the assumptions used when it was designed. The absence of the original sample-size justification, multiplicity plan, and detailed missing-data procedures limits reconstruction of the full statistical plan. Together, these features show why registry results should be read with attention to both the calculations that are posted and the design details that are not.

19. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry provides limited detail for independently evaluating the primary analysis. It reports a log-rank method and a Cox proportional-hazards parameter, but does not post event times, censoring rules, the number at risk over time, person-time denominators, model diagnostics, or covariate-adjustment details. The primary outcome description refers to person-years, while the posted arm summaries are participant event counts; the relationship between those summaries is not fully documented.

The primary comparison combines two azithromycin arms. The posted estimate therefore does not distinguish the single participant dose strategy from the strategy involving the participant and immediate family members. The registry does not give separate primary results for those intervention arms in the summarized comparison.

The study objective mentions outcomes at one and two years, whereas the listed primary outcome timeframe runs through 12 months. A separate two-year primary result is not reported in the registry record. The record also does not describe participant flow, missing outcome data, analysis populations, allocation concealment, or which party was masked. It does not report the original sample-size justification, multiplicity procedures, interim analyses, or the complete analysis plan.

The safety summaries contain event counts but limited definitions and no posted comparative analysis. The secondary outcome information is also incomplete relative to the recorded total number of outcome measures and analyses. These reporting limitations do not show that procedures were absent; they mean the registry record does not permit their evaluation.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Study of Azithromycin to Prevent Recurrent Trichiasis Following Surgery in Ethiopia — NCT00347776.
• Registry record last updated 2017-08-07; status last verified 2017-06.
• Results first posted 2017-08-07.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

23. Record Summary

NCT00347776 is a completed, randomized, single-masked, parallel-group Phase 4 trial sponsored by Johns Hopkins University. The registry lists 1452 participants and trichiasis as the condition. It compares topical tetracycline with oral azithromycin after trichiasis surgery, with two azithromycin strategies combined in the posted primary comparison.

The primary outcome is recurrent trichiasis, assessed clinically by evidence of one or more eyelashes touching the globe or epilation. The registry lists assessments at 2 weeks, 1.5 months, 6 months and 12 months after surgery. The sponsor reports a log-rank and Cox time-to-event analysis; an unadjusted recalculation summarizes the posted event counts as a risk difference. The record also lists surgical failure and adverse events as secondary outcomes.

Important details, including the original sample-size justification, missing-data procedures, analysis populations, multiplicity plan, and a separate two-year primary result, are not reported in the registry record.