High-Dose Isoniazid Among Adult Patients With Different Genetic Variants of INH-Resistant Tuberculosis (TB): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for High-Dose Isoniazid Among Adult Patients With Different Genetic Variants of INH-Resistant Tuberculosis (TB) (NCT01936831), a randomized phase 2 study. The primary outcome is Daily Change in log10 Colony-forming Unit (CFU).

2. Clinical Question

3. Trial Design

This was a completed, randomized, open-label, parallel phase 2 study of isoniazid in adults with tuberculosis. The registry lists total enrollment as 282. Participants were organized into dose cohorts according to the mutation group assigned to their tuberculosis strain: an inhA mutation only, neither inhA nor katG mutations, or a katG mutation with or without an inhA mutation.

The dose cohorts were Group 1 at 5 mg, 10 mg, or 15 mg; Group 2 at 5 mg; and Group 3 at 15 mg or 20 mg. Each cohort received isoniazid daily with vitamin B6 for 7 days. The trial was unmasked. The registry record does not report the randomization procedure, how allocation was implemented, or whether allocation was randomized within each mutation group or across cohorts. It also does not report a detailed protocol rationale for the cohort structure.

The primary endpoint was daily change in sputum log10 CFU per mL over 7 days. CFU results were not collected for Group 3, so Group 3 is absent from the posted primary endpoint comparisons. The primary outcome analysis population was further restricted to Step 2 participants in Groups 1 and 2 who took at least one dose and had both baseline and Day 7 CFU counts. The reported sample sizes for the four primary outcome cohorts were 12, 11, 12, and 11 participants, respectively.

The registry lists 12 outcome measures and eight secondary outcome measures, including pharmacokinetic, microbiological, and additional early bactericidal activity measures. The record’s analysis listing contains no posted analyses for those other measures. The trial was completed, with primary completion recorded as 2021-09-22.

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.

Only the first 6 of 9 groups are shown.

Step 1: Group Identification

Step 2: Arm Randomization

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Participants who took at least one dose of study treatment in Step 2 in Groups 1 and 2. CFU was not collected for Group 3. Participants who had missing Baseline or Day 7 CFU counts were not included.

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.

Population: All participants

Only the first 6 of 10 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Daily Change in log10 Colony-forming Unit (CFU), measured over Measured at baseline and Day 7.

The primary endpoint is the mean daily change in log10 colony-forming units per milliliter of sputum (log10 CFU per mL sputum per day), measured from baseline to Day 7. CFU is a measure of viable bacteria capable of forming colonies in culture. The endpoint is intended to describe the short-term change in sputum bacterial burden during the treatment period.

The registry defines the measure as the Day 7 log10 CFU per mL minus baseline log10 CFU per mL, divided by 7. Baseline is the mean of the pre-entry visit and entry visit sputum colony counts. Under this definition, a negative daily change indicates a decline in measured bacterial burden over the 7-day period; a value closer to zero indicates a smaller average decline.

Early bactericidal activity measures are used to characterize short-term changes in bacterial counts after treatment begins. As a general statistical point, expressing the change per day facilitates comparison of average change across a specified observation period, but it does not describe each participant’s trajectory or establish what happens beyond the observed period. The registry reports the outcome as a mean with standard deviation. It does not report CFU collection for Group 3, and participants with missing baseline or Day 7 CFU counts were excluded from the primary endpoint population.

7. Results Posted in the Registry

• Group 1: 10mg Cohort: mean -0.18 versus -0.06 on control; difference -0.12 (95% CI -0.23 to -0.01), standardized effect d = -0.92; z-test p = 0.027.
• Group 1: 15mg Cohort: mean -0.21 versus -0.06 on control; difference -0.15 (95% CI -0.27 to -0.03), standardized effect d = -1.03; z-test p = 0.011.
• Group 2: 5mg Cohort: mean -0.15 versus -0.06 on control; difference -0.09 (95% CI -0.19 to 0.01), standardized effect d = -0.71; z-test p = 0.085.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is a continuous measure: each participant’s CFU change is expressed on a numeric scale, and the registry reports the mean for each cohort. A mean describes the arithmetic center of a set of measurements. The accompanying standard deviation describes how dispersed measurements are around that mean. It is not the uncertainty of the mean itself; uncertainty about a mean is reflected more directly in a confidence interval.

What does the mean difference represent?

The posted mean differences compare each listed cohort’s mean with the mean for Group 1 at 5 mg. A negative difference means the compared cohort’s estimated daily change is more negative than the reference cohort’s estimate. For example, the Group 1 10 mg comparison has a mean difference of -0.12. This is a difference in mean daily change on the endpoint scale, not a percentage change and not a measure of an individual participant’s response.

The reported means and differences are unadjusted. In general, an unadjusted comparison contrasts the observed group summaries without accounting for other covariates through a statistical model. The registry record does not report a covariate-adjusted model or enough detail to establish whether additional modeling was used for the primary result. Comparisons across mutation groups can also reflect differences in the groups beyond dose; Group 2’s 5 mg cohort is not genetically defined in the same way as Group 1’s reference cohort.

How should the p-values be read?

A p-value describes how unusual data at least as discrepant as the observed data would be under a specified null model. The posted unadjusted p-values are 0.027 for Group 1 at 10 mg, 0.011 for Group 1 at 15 mg, and 0.085 for Group 2 at 5 mg. A p-value does not measure the probability that the null hypothesis is true, the probability that the result occurred by chance, or the size and importance of a difference. Its interpretation depends on the statistical model and analysis plan. The registry record does not report the original prespecified testing strategy or whether these comparisons were primary, exploratory, or selected after examining the data.

Multiple comparisons matter because assessing several contrasts can raise the chance of at least one apparently notable result under repeated testing. The record lists multiple outcomes and three primary endpoint comparisons, but does not report a multiplicity procedure. The p-values should therefore be read as the posted unadjusted values rather than as values corrected for the number of tests.

Why do small cohorts matter?

Each primary outcome cohort has 11 or 12 participants. Small samples can produce uncertain estimates, and a few participant measurements can have substantial influence on a group mean. Standard deviations give a sense of within-cohort variation, while confidence intervals convey uncertainty around between-cohort differences. Neither removes uncertainty caused by limited sample size or by exclusions from the analysis population.

The endpoint calculation uses the difference between Day 7 and baseline values divided by 7. It is a summarized daily change over the interval. It does not necessarily mean that each participant’s CFU count changed at a constant rate on every day. The registry also lists other EBA measures based on early and late phases and on model-based estimates, but numerical results for those analyses are not included in the posted analysis listing.

What limits the analysis population?

The primary outcome population includes participants in Groups 1 and 2 who took at least one dose of study treatment in Step 2 and had both baseline and Day 7 CFU counts. People with missing baseline or Day 7 counts were excluded, and CFU was not collected for Group 3. The record does not report the number excluded for missing measurements, the reasons for missingness, or methods such as imputation to address missing values. The reported cohort sample sizes should thus be understood as the analyzed participants for this outcome, not as the total enrollment.

In general, randomization can reduce systematic differences between assigned groups, but the registry record does not describe the allocation process or confirm which comparisons were protected by random assignment. The study is identified as randomized, yet the mutation-defined groups are not interchangeable by design. Interpretation should distinguish the within-Group 1 dose comparisons from the comparison of Group 2 with Group 1. No further claims about the statistical design or analysis plan are supported by the registry record.

10. Confidence Intervals and Precision

A confidence interval expresses uncertainty around an estimated comparison. In general, it gives a range of values reasonably compatible with the data and the statistical method used, under the assumptions of that method. Wider intervals indicate less precision; intervals that include zero are compatible with no mean difference as well as with differences in either direction. An interval that excludes zero indicates that the corresponding unadjusted estimate is not centered on zero at the stated interval level, but it does not establish clinical importance or remove the need to consider design and multiplicity.

For Group 1 at 10 mg, the posted mean difference is -0.12 and its confidence interval excludes zero. The Group 1 at 15 mg difference is -0.15, and its interval also excludes zero. The Group 2 at 5 mg difference is -0.09, and its interval includes zero. These intervals are based on small cohorts: the reference group has 12 participants, and the comparison groups have 11, 12, and 11 participants, respectively. The limited cohort sizes contribute to uncertainty, and each cohort’s standard deviation describes appreciable variation in individual values relative to the reported means.

The registry’s comparison intervals are reported with more than three decimal places, so exact endpoints are not reproduced in this explanation. Their inclusion or exclusion of zero can be described without implying greater precision than is appropriate. The record does not report whether confidence intervals were adjusted for the multiple comparisons or whether they derive from a prespecified analysis plan. These are unadjusted comparison intervals and should be considered alongside the unadjusted p-values, the short observation period, the analysis exclusions, and the mutation-defined cohort structure.

Precision is not the same as validity or relevance. A narrow interval would not address potential differences between groups, missing outcome data, or the limits of a 7-day endpoint. Conversely, a wide interval does not prove that there is no difference; it indicates that the estimate is less precise. The posted intervals quantify sampling uncertainty according to the reported comparison, not uncertainty about generalizing to other populations or treatment settings.

11. 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 effects in the posted comparisons, not the trial’s original design assumptions. Each row considers a stated fraction of the observed effect: the full observed effect, 75% of the observed effect, or 50% of the observed effect. The columns give the estimated participants per arm for 80% power and participants per arm for 90% power. These values are participants per arm, not total participants or total events.

The planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect. These figures are not observed power; they are illustrations of the sample size that would be needed under those assumptions to detect an effect of the specified size with the stated power.

In plain terms, 80% power describes a design that would have an 80% chance of detecting an effect of the planned size under the assumptions if that effect were present. The 90% power column describes the corresponding design with a higher chance of detection, and it generally requires more participants. Smaller effects need more participants to detect than larger effects. The planning values are calculated separately for each posted comparison, so they vary with the estimated effect and variability in that comparison.

These calculations are not a reconstruction of the trial’s recruitment plan. The registry record does not report the original sample-size justification, assumptions, or target power for the study. The illustration also relies on observed effects from small cohorts, which may be imprecise; planned sample sizes based on those estimates should not be treated as definitive design recommendations.

12. 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, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -0.120; pooled standard deviation σ = 0.131.
2. n = 2 × (1.96 + 0.842)² × 0.131² / -0.120² = 18.7, rounded up to 19 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
delta <- -0.12                      # observed difference in means
sigma <- 0.130857974628257                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^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 (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(1936831)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 19 (80% power), 25 (90% power)
 75% of observed effect: n per arm = 34 (80% power), 45 (90% power)
 50% of observed effect: n per arm = 75 (80% power), 100 (90% power)
power.t.test: 19.7 
Simulated power at n = 19 per arm: 0.784 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-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.

13. Multiplicity, Missing Data and Analysis Populations

The registry record lists 12 outcome measures, including the primary CFU endpoint and secondary pharmacokinetic, microbiological, and EBA outcomes. Eight secondary outcome measures are described. The analysis listing contains no posted analyses for those outcomes. For the primary endpoint, three cohort comparisons are reported, all using Group 1 at 5 mg as the reference. The record does not report whether a hierarchy of hypotheses, a multiplicity adjustment, or another strategy for controlling false-positive findings was specified.

Multiplicity is relevant when several outcomes or comparisons are tested. As a general statistical principle, the more hypotheses assessed, the greater the possibility of obtaining a small p-value by chance somewhere among the tests, unless the analysis accounts for the testing structure. The p-values posted for the primary comparisons are explicitly unadjusted in the comparison results. They should not be interpreted as family-wise adjusted evidence, and the available record does not establish how the three contrasts relate to a prespecified decision rule.

The record defines the primary endpoint population as participants who took at least one study-treatment dose in Step 2 in Groups 1 and 2 and had baseline and Day 7 CFU counts. Participants missing either CFU measurement were excluded. CFU was not collected for Group 3. The analyzed cohort sizes are 12, 11, 12, and 11. The registry record does not report how many otherwise eligible participants had missing measurements, the reasons for missingness, whether missingness differed by cohort, or any imputation or sensitivity analysis.

Because the primary endpoint is based on participants with observed baseline and Day 7 measurements, the posted estimates describe that analyzed population. If missingness is related to bacterial measurements or other participant characteristics, exclusion could affect the estimates; the record does not provide enough information to assess that possibility. The record also does not report a detailed statistical analysis plan, whether the analysis was intention-to-treat, or a protocol-defined treatment of deviations. The study is identified as randomized, but details of random sequence generation, concealment, and allocation within the mutation-defined cohorts are not reported in the registry record.

14. 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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: pneumothorax spontaneous (7.1%), pneumothorax (6.2%).

Serious adverse event figures are reported by dose cohort as the number affected among the number at risk. The registry reports 0 of 13 in Group 1 at 5 mg, 1 of 14 in Group 1 at 10 mg, 0 of 16 in Group 1 at 15 mg, 1 of 16 in Group 2 at 5 mg, 0 of 11 in Group 3 at 15 mg, and 0 of 10 in Group 3 at 20 mg. The death count is 0 in each of the six cohorts.

The registry also lists “Pneumothorax spontaneous” at 7.1% and “Pneumothorax” at 6.2% among the top serious adverse event terms. These terms and percentages are reported as registry safety data; they should not be assigned to a particular cohort unless the record makes that attribution. Serious adverse events and other adverse events are distinct summaries. The other adverse event figures show 12 of 13, 14 of 14, 15 of 16, 16 of 16, 11 of 11, and 10 of 10 affected across the respective cohorts.

These counts are descriptive. The cohorts are small, and the registry record does not provide comparative safety analyses or details that would support attributing an event to isoniazid or comparing risks across doses. They also do not capture, by themselves, the severity or duration of every event.

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

The posted primary outcome describes average short-term changes in sputum bacterial counts in three dose cohorts within the inhA-only mutation group and one cohort with neither inhA nor katG mutations. All four reported means are negative, consistent with decreases in measured CFU over the 7-day endpoint period under the registry’s definition. The mean declines in the Group 1 10 mg and 15 mg cohorts were more negative than the Group 1 5 mg reference mean, and their posted unadjusted confidence intervals exclude zero. The Group 2 5 mg comparison is also more negative than the reference mean, but its interval includes zero.

These results do not establish that any dose is effective for treating tuberculosis or that a dose is preferable. The endpoint is an early bactericidal activity measure, not a reported long-term clinical outcome. It covers 7 days, and the registry record does not provide results for Group 3 CFU because CFU was not collected for that group. It also does not provide numerical results for the listed secondary outcomes in the analysis listing.

The cohort comparisons must be interpreted in light of the mutation groups. Comparing two Group 1 cohorts addresses different doses within the same stated mutation category, while comparing Group 2 with Group 1 also changes the mutation category. The record does not report enough information about randomization and allocation to determine precisely which contrasts can be interpreted as randomized treatment comparisons.

The small analyzed samples, exclusions for missing baseline or Day 7 measurements, multiple outcomes, and unadjusted comparisons all limit certainty. The reported means and intervals quantify only some aspects of uncertainty and do not resolve the design and analysis limitations. The registry record does not report a sample-size justification or a detailed statistical analysis plan. The findings should therefore be read as posted short-term cohort estimates rather than as a definitive assessment of longer-term treatment outcomes or a complete analysis of every enrolled participant.

16. Why This Trial Matters Statistically

This study poses a statistical challenge common in early clinical research: the outcome is measured over a short interval, the cohorts are small, and participants are grouped by biological characteristics as well as dose. The endpoint’s negative values represent decreases in sputum CFU, while the comparison estimates describe differences in average daily change. Keeping those two quantities distinct helps prevent a mean difference from being mistaken for an absolute bacterial count or a percentage change.

The mutation-defined structure also matters. A contrast between Group 1 doses is not equivalent to a contrast between Group 1 and Group 2, because the latter compares different mutation categories as well as cohort assignment. A statistical comparison can quantify the observed difference, but interpretation depends on whether the groups are comparable and how treatment allocation was conducted. The registry identifies the study as randomized but does not report the allocation details needed to evaluate those questions.

Precision is another important issue. Each primary outcome cohort contains only 11 or 12 participants, so means and differences are estimated with limited data. The posted intervals provide a useful complement to p-values: two intervals exclude zero and one includes zero, while all remain subject to the small sample sizes and unadjusted testing. The record’s omission of multiplicity details and a formal analysis plan makes it especially important not to treat individual p-values as a complete inferential account.

Finally, the registry separates the overall enrollment from the analyzed CFU population. The primary endpoint excludes participants with missing baseline or Day 7 counts and does not include Group 3. Readers therefore need to distinguish enrollment, outcome availability, and the size of each analyzed comparison. Those distinctions are central to judging what an early bactericidal activity estimate can and cannot establish.

17. 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.

Several trial-specific limitations constrain interpretation. The primary outcome covers only baseline to Day 7, so it describes an early response interval rather than later treatment outcomes. CFU was not collected for Group 3, leaving the two katG-containing dose cohorts without primary endpoint results. The record lists other pharmacokinetic and EBA outcomes, but numerical analyses for those measures are not posted in the analysis listing.

The primary endpoint analysis includes only participants who took at least one Step 2 dose in Groups 1 and 2 and had both baseline and Day 7 counts. The number and characteristics of participants excluded for missing CFU results are not reported. The cohort sizes for the posted comparisons are small, and the registry does not report sensitivity analyses to assess the effect of missing observations.

Although the design is described as randomized, the registry record does not report the randomization procedure, allocation concealment, or how assignment related to mutation-defined cohorts. This limits assessment of whether each comparison can be interpreted as a randomized contrast. The comparison of Group 2 with Group 1 is especially distinct from the dose contrasts within Group 1 because the mutation categories differ.

The record does not report the original sample-size justification, a detailed statistical analysis plan, prespecified multiplicity handling, or adjusted models. The three posted p-values are unadjusted. Safety counts are also descriptive, with no comparative analysis reported. Together, these limitations mean the posted estimates should be interpreted narrowly as short-term cohort summaries and comparisons, not as a comprehensive analysis of all outcomes or a definitive measure of longer-term clinical benefit.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: High-Dose Isoniazid Among Adult Patients With Different Genetic Variants of INH-Resistant Tuberculosis (TB) — NCT01936831.
• Registry record last updated 2023-02-17; status last verified 2023-02.
• Results first posted 2023-02-17.
• 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.

21. Record Summary

NCT01936831 is a completed randomized, open-label, parallel phase 2 study of isoniazid dose cohorts among adults with tuberculosis categorized by resistance-associated mutations. Total enrollment was 282. Cohorts received daily isoniazid at doses from 5 mg/kg to 20 mg/kg with vitamin B6 for 7 days.

The primary outcome was mean daily change in sputum log10 CFU per mL from baseline to Day 7. The posted primary analyses include four cohorts in Groups 1 and 2, with 11 or 12 participants per cohort; Group 3 CFU was not collected. Two comparisons with the Group 1 5 mg reference have unadjusted confidence intervals that exclude zero, while the Group 2 comparison interval includes zero. The registry record lists secondary outcomes but no posted analyses for them. The original sample-size justification and detailed statistical analysis plan are not reported in the registry record.