Mobile Health Intervention for Active Tuberculosis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Mobile Health Intervention for Active Tuberculosis (NCT03544476), a randomized na study. The primary outcome is Number of Participants With Treatment Success.

2. Clinical Question

3. Trial Design

The registry describes a randomized, single-masked, parallel-group study. Participants were assigned to one of two groups: a mobile phone tuberculosis treatment support app plus usual care, or usual care alone. The listed enrollment is 42. The posted outcome groups each contain 21 participants. The person or people who were masked are not reported in the registry record, so the meaning and implementation of single masking cannot be specified further.

The study’s stated purpose was to evaluate and further refine a mobile support tool for people receiving treatment for active tuberculosis. The intervention combined the app with usual care rather than replacing clinical care. App users were asked to report medication administration daily, record side effects when applicable, and complete paper-based adherence testing on randomly selected days during the intensive treatment phase. Usual care included outpatient management, routine clinical and laboratory testing, and clinician-determined follow-up appointments.

The registry lists the study as completed. Its start date is 2019-04-09, and its primary completion date is 2020-03-31. The phase is listed as NA. The registry record names University of Washington as sponsor. Further information about recruitment, sites, allocation concealment, the randomization method, or the masking procedures is not reported in the registry record.

The primary outcome is a count of participants with treatment success at 6 months. The registry also lists engagement, measured as the average number of days participants reported through the app, as a secondary outcome. There are three outcome measures listed, but the additional outcome details are not reported in the registry record. The posted record lists no analyses. The statistical analysis plan, including any adjustment variables or prespecified contrasts, is not reported in the registry record.

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.

Overall Study

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 Number of Participants With Treatment Success, measured over 6 months.

The primary endpoint is the number of participants with treatment success, assessed at 6 months. The registry defines treatment success using World Health Organization standards for tuberculosis treatment: a participant is considered successful if treatment is completed without bacteriological confirmation or if the participant is cured, which the description defines as a negative sputum smear at 6 months and at least once before 6 months. The endpoint is therefore a count of participants meeting a combined outcome definition, not a continuous measure of symptoms, adherence, or time to an event.

A binary outcome is useful when the clinical question is whether participants meet a defined status by a specified point in follow-up. It permits direct comparison of the fraction meeting the definition in each group while retaining an interpretable count. In this trial, the posted counts are 20 participants in the app group and 17 in usual care, with 21 participants in each group. The registry’s definition also distinguishes treatment success from other possible outcomes, including treatment failure, death, default, loss to follow-up, and transfer with unknown outcome.

The combined endpoint depends on the registry’s stated classification rules and on whether outcome status could be established for each participant. The detail of how outcome categories were adjudicated, how missing outcomes were handled, and whether bacteriological confirmation was available consistently is not reported in the registry record. The endpoint measures status at the stated follow-up time; it does not by itself measure the quality of the treatment experience or explain why a participant did or did not meet the definition.

7. Results Posted in the Registry

• Mobile Phone TB Treatment Support App: 95.2% (20/21) versus 81.0% (17/21) on control; risk difference 14.3 percentage points (95% CI -4.8 to 33.4), risk ratio 1.18 (0.94–1.48); unadjusted two-proportion z-test p = 0.153.

8. Statistical Interpretation

9. Statistical Methods Explained

For a binary endpoint, each participant is classified according to whether the outcome occurred. Here, the outcome is treatment success by 6 months. The posted summaries give the number meeting the endpoint and the number in each group: 20 of 21 for the app-plus-usual-care group and 17 of 21 for usual care. The registry also gives the corresponding group proportions as 0.952 and 0.809. These summaries describe the observed participants; the inferential statistics quantify uncertainty when using a sample to learn about a broader population.

What does the risk difference express?

The risk difference is the app-group event proportion minus the usual-care event proportion. The registry reports an estimate of 0.143. On this scale, zero represents equal event proportions. A positive estimate means the event proportion is higher in the app group in these posted results; a negative estimate would mean it is lower. This is a contrast in absolute probability, rather than a ratio of probabilities. The reported 95% confidence interval is -0.048 to 0.334. Because the interval includes zero, the data are compatible with no difference as well as with a range of positive and negative differences.

A confidence interval is a range calculated by a statistical procedure that, under its assumptions and repeated use, would cover the true parameter at the stated frequency. It does not mean that there is a 95% probability that this particular interval contains the true value. General statistical interpretation also depends on the sampling process and model assumptions. The registry record does not report the precise method used to form each interval, so the interval procedure cannot be described more specifically.

How do relative measures differ?

The relative risk compares the event proportion in the app group with the event proportion in usual care. The posted estimate is 1.176, and its confidence interval is 0.936 to 1.478. A relative risk of one corresponds to equal proportions; values above or below one indicate different relative levels in the groups. The odds ratio compares event odds rather than event probabilities. The posted odds ratio is 4.706, with an interval from 0.479 to 46.22. Odds and probabilities are related but are not interchangeable, so the odds ratio should not be read as an absolute percentage-point difference. The broad odds ratio interval indicates that the relative contrast is imprecisely estimated.

With a small number of participants and outcomes, estimates can be sensitive to the outcome count in either group. General statistical reasoning cautions that relative measures may be especially unstable when counts are sparse, and that different measures can appear to tell different stories because they use different scales. The risk difference, relative risk, and odds ratio are complementary descriptions of the same two-group binary outcome; none is a substitute for examining the counts and their uncertainty.

What does the p-value say?

The unadjusted p-value is 0.153. In general, a p-value describes how unusual the observed data, or data more extreme, would be under a specified null model and its assumptions. It is not the probability that the null hypothesis is true, nor does it quantify the probability that the result will replicate. The registry record does not report the test used, the prespecified significance threshold, or whether this was the sole primary comparison. The p-value should therefore be read alongside the effect estimate and confidence interval, not as a standalone verdict.

Adjustment refers to including additional variables in a statistical model, commonly to account for baseline characteristics or improve precision. The posted primary comparison is identified as unadjusted. Whether adjusted analyses were planned, conducted, or intended as the primary analysis is not reported in the registry record. No adjusted estimates should be inferred from the posted summary statistics.

Why can sparse data matter?

General statistical concepts help explain why a small binary-outcome study can yield wide uncertainty. When the total number of outcome observations is limited, moving a small number of participants between outcome categories can materially change estimated group proportions and relative measures. Standard methods may also rely on approximations that work better when data are more abundant. Alternative methods can behave differently in sparse settings, but the registry record does not identify the methods used for the reported estimates. It is not appropriate to assign a particular method to these results without that information.

Over-dispersion is a general issue in count data when observed variability exceeds what a chosen model expects. This trial’s primary outcome is a participant count summarized as a binary status within groups, and the registry does not report a count model or a dispersion estimate. Over-dispersion therefore cannot be assessed from the posted primary summary, and no claim that it affected these results is warranted.

Finally, statistical evidence should be distinguished from clinical importance. A confidence interval describes statistical precision around an estimated contrast; it does not determine whether a difference would matter to patients or clinicians. The registry record does not report a threshold for a clinically important difference. The outcome definition, size of the estimate, uncertainty, and relevance to the intended population all need consideration, but the posted results alone do not settle those questions.

10. Confidence Intervals and Precision

The main precision summary is the 95% confidence interval for the risk difference: -0.048 to 0.334 around a point estimate of 0.143. The interval spans zero, so the reported data do not rule out no difference on the risk-difference scale. It also extends in both directions from the estimate, leaving uncertainty about the direction and size of a population contrast. The estimate is therefore not precise enough to support a narrow claim about how large the difference might be.

The relative measures also have broad intervals. The relative risk is 1.176, with a confidence interval from 0.936 to 1.478. The odds ratio is 4.706, with an interval from 0.479 to 46.22. The intervals include the values associated with no relative difference, and cover a wide range of possible relative effects. The odds ratio estimate should be considered especially cautiously because its interval is very broad.

Confidence intervals express sampling uncertainty under the method’s assumptions; they do not account automatically for every possible source of uncertainty, such as outcome misclassification, missing outcomes, or differences in implementation. The methods used to calculate these intervals are not reported in the registry record. The number of participants is 21 in each group, so the precision is constrained by the modest size of the posted comparison. This is a description of information available in the trial, not a judgment about whether a particular result is clinically important.

Reading the point estimate without its interval would overstate how settled the comparison is. Conversely, an interval that includes zero does not demonstrate that the groups are equivalent. Equivalence requires a prespecified margin and an analysis designed to assess whether the plausible effects lie within that margin; such a margin and analysis are not reported in the registry record. The confidence intervals support an interpretation of substantial uncertainty rather than a definitive conclusion in either direction.

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 planning figures are illustrations based on the observed effect, not the original trial design. The registry record does not report the trial’s own sample-size justification or original design assumptions. The listed 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 set to the stated fraction of the observed effect.

For a binary primary outcome, the entries are participants per arm, not total participants and not total events. The figures are 80 and 106 participants per arm for 80% and 90% power when planning for 100% of the observed effect; 161 and 215 participants per arm when planning for 75% of the observed effect; and 402 and 537 participants per arm when planning for 50% of the observed effect. The 80% power column describes a design that, under the planning assumptions and if the selected effect were the true effect, would have an 80% chance of detecting an effect using the stated test threshold. The 90% column uses the same idea with a higher target chance of detection.

Smaller effects require more participants to distinguish them from random variation under otherwise similar assumptions. Thus, the planning illustrations increase as the effect to detect becomes a smaller fraction of the observed effect. They are not a recommendation or a reconstruction of the design used in this study. Because the inputs are based on the observed estimate, they may be unstable, particularly in a small study. This is not observed power: it is a prospective planning illustration, not a measure of the probability that the completed trial’s result is correct or reproducible. Time-to-event planning displays, if present in a general presentation, would use total event counts under the stated assumptions; the primary endpoint in this trial is binary and its entries are participants per arm.

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, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.810; treatment proportion p₁ = 0.952; average p̄ = 0.881.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.4580 = 0.8976.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.4467 = 0.3760.
4. n = (0.8976 + 0.3760)² / (0.1429)² = 79.5, rounded up to 80 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 <- 0.80952380952381                          # control proportion
p1 <- 0.952380952380952                          # 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(3544476)
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 = 80 (80% power), 106 (90% power)
 75% of observed effect: n per arm = 161 (80% power), 215 (90% power)
 50% of observed effect: n per arm = 402 (80% power), 537 (90% power)
power.prop.test: 79.5 
Simulated power at n = 80 per arm: 0.830 (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.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists three outcome measures and one secondary outcome: engagement, measured as the average number of days participants reported through the app. The primary outcome is treatment success at 6 months. The other outcome details and results are not reported in the registry record. The record lists zero analyses. It does not explain which outcomes were prespecified as primary at each stage, whether there were multiple time points or subgroup comparisons, or whether any outcomes were added or changed after study initiation.

Multiplicity is the general statistical issue that arises when several hypotheses are tested. As the number of tests increases, the chance of seeing at least one apparently unusual result can increase, unless the analysis plan accounts for the testing structure. The registry record does not report a multiplicity strategy, correction, hierarchy among secondary outcomes, or confirmatory testing plan. The posted unadjusted p-value should therefore be interpreted as the reported comparison, without assuming that any family-wise adjustment was or was not conducted outside the record.

Missing data can affect a binary endpoint if some participants do not have a known outcome or are excluded from analysis. The posted group sizes and counts describe 21 participants per group, but the registry record does not specify the analysis population, whether all randomized participants were included, or how incomplete follow-up was handled. It also does not report an imputation method, a missing-outcome category, or sensitivity analyses. The endpoint description includes loss to follow-up as one of the possible treatment outcomes, but that definition does not by itself explain how missing endpoint information was handled in the analysis.

The registry record also does not report whether the primary analysis followed an intention-to-treat principle, whether participants were analyzed according to assigned group, or whether exclusions occurred after randomization. Baseline balance, protocol deviations, and adherence to the planned analysis are not characterized. These omissions limit assessment of how the posted estimates relate to the randomized comparison and how robust they might be to alternative choices about analysis populations or missing outcomes.

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 0% reporting threshold.

The serious adverse-event counts are zero among 21 participants in the mobile phone app group and zero among 21 participants in usual care. The record lists no other adverse events among 21 participants in either group. It lists zero deaths in the app group and one death in usual care. These are posted counts, not estimates demonstrating equal safety or a causal effect of group assignment.

Zero reported events does not establish that the underlying risk is zero. With limited numbers of participants, uncommon events may not appear during the observed follow-up, and the record does not provide detailed information about event ascertainment or duration of safety monitoring. The top serious-event terms are not listed. The cause of the death and whether it was considered related to study participation or treatment are not reported in the registry record.

Counts should be read in the context of group size and the study’s reporting definitions. No comparative safety analysis or confidence interval is reported. The figures are descriptive summaries and do not allow a precise estimate of differences in serious or other adverse events between groups.

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

The posted results show that treatment success at 6 months was recorded for 20 of 21 participants in the mobile phone app plus usual care group and 17 of 21 participants in usual care. The estimated risk difference is 0.143, and the 95% confidence interval is -0.048 to 0.334. These figures describe the enrolled groups under the endpoint definition in the registry record. The interval’s breadth means the estimate is uncertain, and its inclusion of zero means the posted data are compatible with no difference as well as with differences in either direction.

The results do not establish that the app caused the observed group pattern. Randomized assignment is an important design feature, but details needed to assess implementation, baseline comparability, masking, missing outcomes, and analysis populations are not reported in the registry record. The record also does not report the statistical analysis plan or the method used for the confidence interval. Readers should not interpret the point estimate alone as a stable estimate of what would occur in a larger or different population.

The results also do not demonstrate that the groups are equivalent. A nonsignificant or uncertain comparison is not proof that two approaches have the same effect. An equivalence conclusion would require an appropriate design, a prespecified margin defining meaningful similarity, and an analysis that evaluates that margin. Those elements are not reported in the registry record.

The primary outcome captures whether participants met a combined treatment-success definition by 6 months. It does not measure every aspect of treatment, patient experience, app engagement, or adherence behavior. Although engagement is listed as a secondary outcome, its results are not reported in the registry information described. The posted outcome therefore cannot determine whether the app was used as intended or whether any outcome pattern operated through engagement.

The findings are most appropriately read as a small randomized comparison with an imprecise estimate of the primary outcome contrast. They identify the observed counts and the range of uncertainty conveyed by the posted interval, but they do not settle effectiveness, clinical importance, or safety. Further interpretation would depend on fuller methods and outcome reporting, which are not reported in the registry record.

16. Why This Trial Matters Statistically

This trial raises a useful statistical question about evaluating a behavioral and monitoring tool when the primary outcome is a clinical status measured at a defined follow-up. The app intervention was intended to support self-reporting of medication administration and side effects and to prompt adherence testing during intensive treatment. The primary endpoint, however, is treatment success at 6 months. Those are related but distinct constructs: an app-use measure concerns engagement with the tool, while treatment success is a clinical outcome classified using the registry’s stated definition.

The distinction matters because an intervention can affect its intended process without producing a clearly estimated difference in a downstream outcome, and a clinical outcome alone cannot reveal the mechanism behind a group pattern. Engagement is listed as a secondary outcome, but its result is not reported in the registry record. Without it, readers cannot assess how often participants interacted with the app or connect engagement to treatment status.

The study also illustrates why effect estimates and uncertainty should be interpreted together. The risk difference is 0.143, but its confidence interval extends from -0.048 to 0.334. The small number of participants means the point estimate should not be treated as a precise estimate of a broader effect. Reporting risk difference alongside relative measures makes clear that the same comparison can be expressed on different scales, while the wide intervals show that none of those scales eliminates uncertainty.

Finally, this record highlights the value of complete reporting of design and analysis details. Information about masking, randomization, missing outcomes, analysis populations, and multiplicity would help readers understand how the estimates were obtained and what conclusions the design can support. The posted results are informative as a description of the randomized groups, while their statistical interpretation remains limited by the size of the comparison and the methods details not reported in the registry record.

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.

The trial has 42 participants listed in total, with 21 in each posted group. This is a small comparison for estimating a binary outcome, and the confidence intervals are broad. The registry does not report the original sample-size justification, so it is not possible to assess the assumptions or target precision used when the study was planned. The planning illustrations based on the observed effect are not a substitute for the original design rationale.

Important design details are not reported in the registry record. These include eligibility criteria, participant characteristics, study sites, the randomization method, allocation concealment, and which individuals were masked. Without participant characteristics, readers cannot assess group balance or how closely the enrolled population matches other populations receiving treatment for active tuberculosis.

Analysis details are also limited. The record lists no analyses and does not report the statistical analysis plan, the confidence-interval method, missing-data handling, analysis population, or whether participants were analyzed as randomized. It is also not reported whether the outcome was adjudicated or how incomplete outcome information was classified. These gaps make it difficult to assess the robustness of the primary estimate or whether alternative reasonable methods might change its interpretation.

Only the primary outcome counts and comparison statistics are described in detail. Although engagement is listed as a secondary outcome, its results are not reported in the registry record. The additional outcome measures are not characterized in the available information. Safety reporting is also limited to counts; event definitions, attribution, cause of death, and ascertainment procedures are not reported. These limitations restrict interpretation of both the proposed mechanism and the overall balance of outcomes.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Mobile Health Intervention for Active Tuberculosis — NCT03544476.
• Registry record last updated 2025-03-26; status last verified 2025-03.
• Results first posted 2025-03-10.
• 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

The registry describes NCT03544476, titled “Mobile Health Intervention for Active Tuberculosis,” as a randomized, single-masked, parallel-group study sponsored by University of Washington. The study is listed as completed, with enrollment of 42, a start date of 2019-04-09, and a primary completion date of 2020-03-31. Its listed phase is NA.

The comparison was a mobile phone tuberculosis treatment support app plus usual care versus usual care alone. The primary outcome was the number of participants with treatment success at 6 months. The posted counts were 20 of 21 in the app group and 17 of 21 in usual care. The risk difference was 0.143, with a 95% confidence interval from -0.048 to 0.334; the unadjusted p-value was 0.153. The registry also lists engagement as a secondary outcome, but its result is not reported in the registry record.

Serious adverse events were listed for zero participants in either group; one death was listed in usual care. The small group sizes and broad confidence intervals limit precision. Several design, analysis, missing-data, and outcome-reporting details are not reported in the registry record.