Mobile Strategies for Women's and Children's Health: Optimizing Adherence and Efficacy of PMTCT/ART: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Mobile Strategies for Women's and Children's Health: Optimizing Adherence and Efficacy of PMTCT/ART (NCT02400671), a randomized na study. The primary outcome is Maternal Virologic Failure.

2. Clinical Question

The study is intended to examine whether communication by SMS, and the option for a participant to respond to a nurse, are associated with maternal ART adherence and retention in care. Those aims are described in the registry summary; the posted primary outcome is maternal virologic failure. These concepts are related but are not interchangeable: a viral-load measure is not a direct measure of adherence or retention, and the registry’s posted primary result is expressed as counts of VL measurements rather than as a reported arm-level prevalence comparison.

The registry lists maternal ART adherence, maternal ART resistance, and maternal perceptions of the intervention and care received among the secondary outcomes. Results available for maternal ART resistance can be considered alongside the primary outcome, but they answer a different question. They do not establish whether the SMS strategies changed adherence, retention, perceptions, or virologic failure. The registry record does not report posted results for those other secondary outcomes in the material described.

3. Trial Design

NCT02400671 is a completed, randomized, parallel-group trial titled Mobile Strategies for Women’s and Children’s Health: Optimizing Adherence and Efficacy of PMTCT/ART. The official title describes an evaluation of mHealth strategies to optimize adherence and efficacy of PMTCT/ART. The University of Washington is listed as sponsor. The registry records enrollment of 825 and lists the phase as NA.

The study has three arms: Two-way SMS, One-Way SMS, and Control. Two-way SMS participants were to receive weekly push messages with questions and could text back to the study nurse. One-Way SMS participants were to receive weekly push messages. The control arm is described as receiving standard of care with no intervention. Allocation is listed as randomized, the model as parallel, and masking as none. The registry record does not report the method used to generate the random sequence, allocation concealment procedures, or reasons for the absence of masking.

The brief summary describes comparisons of unidirectional messaging, bidirectional dialogue between participant and provider, and control, with ART adherence and retention in care as outcomes of interest. The primary outcome entry instead specifies maternal virologic failure, assessed using VL measurements after the first 4 months post-ART. Its stated timeframe is 2 years postpartum. The reported primary arm entries are 90.0 for Two-way SMS, 119.0 for One-Way SMS, and 99.0 for Control; accompanying arm counts are 249.0, 235.0, and 243.0, respectively. Because the outcome is designated as a count of units, these entries should not be recast as participant-level event rates.

The design description does not specify the allocation ratio or the full analysis population for every outcome. Those details, as well as a sample-size justification and fuller protocol information, are not reported in the registry record. The separate planning illustrations based on observed ART-resistance effects should not be mistaken for the trial’s original design calculations.

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.

Before assignment: 825 participants were enrolled but it was later identified that one participant was enrolled twice. Only one of these records is included in analysis for a total participant number of 824

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Participants with VL data available \>=4 months post-ART initiation

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 Maternal Virologic Failure, measured over 2 years postpartum.

Maternal virologic failure is the primary endpoint. The registry defines failure as HIV RNA at least 1000 c/ml after the first 4 months post-ART. The stated timeframe is 2 years postpartum, and the population is participants with VL data available at or beyond 4 months after ART initiation.

Viral-load measurements provide an objective clinical measure of virologic status. Comparing the prevalence of measurements meeting the stated threshold can therefore address whether the arms differ in virologic failure, rather than relying only on reported behavior or perceptions. The registry specifies Generalized Estimating Equations (GEE) with a log-binomial link for that comparison.

The posted primary result, however, is identified as a count of units, with VL measurements as the unit. It lists arm values of 90.0, 119.0, and 99.0 and accompanying counts of 249.0, 235.0, and 243.0. No arm-level prevalence estimates, confidence intervals, or GEE comparison results are reported. The counts alone do not show how many measurements met the failure threshold, how repeated measurements were handled in the posted result, or the precision of an arm contrast. Thus the registry’s endpoint definition and planned comparison are clear, but the available primary result does not provide the specified comparative estimate.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry describes the primary comparison as a GEE analysis with a log-binomial link. This is a model-based approach for comparing outcomes across study arms when the data structure may include correlated observations. General statistical explanation: GEE estimates population-average relationships while accounting for within-cluster dependence through a working correlation structure. In a longitudinal setting, repeated measurements from the same participant may be correlated, so treating each measurement as fully independent can understate uncertainty. The registry does not report the working correlation structure, the specific variance estimator, or other implementation details.

What does a log-binomial link do?

General statistical explanation: a link function connects the expected outcome to the predictors in a regression model. A log link models the logarithm of the expected outcome. For a binary outcome, this form can express relative prevalence or risk contrasts directly, subject to the model assumptions. The registry says it will compare prevalence of virologic failure using a log-binomial link, but it does not post the fitted estimates, confidence intervals, or p-values for that analysis. The link description alone is not a result.

How should the primary count entries be read?

The registry identifies the primary result as COUNT OF UNITS and names VL measurements as the unit, while its endpoint description refers to prevalence of virologic failure. Those descriptions do not provide enough information to turn the posted arm values into prevalence estimates. In particular, the entries do not state how many of the counted measurements met the HIV RNA threshold, nor do they give an arm-specific denominator in terms of eligible measurements. The accompanying arm counts should not automatically be treated as denominators for a rate, because the outcome unit is measurements and the registry does not explain the relationship between these quantities. A valid comparison requires the analysis definition and model output, not arithmetic applied to the posted counts.

What do the secondary risk differences express?

General statistical explanation: a risk difference expresses the absolute contrast between two proportions. A value below zero indicates a lower observed proportion in the named intervention arm than in control, while a value above zero indicates a higher observed proportion. A confidence interval describes a range of values compatible with the data and model assumptions; an interval that spans zero includes no difference as a plausible value. For the ART-resistance comparisons, both intervals span zero. The unadjusted p-values are 0.178 for Two-way SMS versus Control and 0.609 for One-Way SMS versus Control. These are unadjusted results and are not proof of no effect. The registry does not post adjusted comparisons.

What are the planning illustrations for?

The registry includes planning illustrations calculated from the observed ART-resistance effects, not from the trial’s original design. The stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The effect to detect is described as a stated fraction of the observed effect. For planning for 100% of the observed effect, the figures are 1,075 participants per arm for 80% power and 1,438 participants per arm for 90% power for the Two-way SMS comparison; for One-Way SMS, they are 7,553 and 10,111. For planning for 75% of the observed effect, the corresponding figures are 2,083 and 2,788 for Two-way SMS, and 13,068 and 17,494 for One-Way SMS. For planning for 50% of the observed effect, they are 5,073 and 6,790 for Two-way SMS, and 28,591 and 38,275 for One-Way SMS. These are hypothetical planning illustrations, not observed power and not evidence that the trial was designed to detect those effects.

Planning values are sensitive to the observed effect used as their starting point. An effect estimated with uncertainty can yield unstable planning figures, particularly when the observed contrast is small. The registry’s assumption of equal allocation applies to those illustrations; it should not be assumed to describe actual trial allocation. The registry record does not report the trial’s original sample-size justification, design effect assumptions, or planned power calculation.

The endpoint is described as involving measurements over time, but no dispersion estimate or details for handling repeated observations are posted with the primary result. General statistical explanation: variation and correlation influence precision, and over-dispersion can make a simple model’s uncertainty estimates too narrow if the model does not account for it. The registry’s choice of GEE is compatible with accounting for correlated data, but details needed to assess the fitted analysis are not reported. No recalculation is warranted from the posted primary counts.

10. Multiplicity, Missing Data and Analysis Populations

The trial has three arms, a primary endpoint, and multiple secondary outcomes. The registry lists maternal ART adherence, maternal ART resistance, and maternal perceptions of intervention and care received as secondary outcomes; it also lists 7 outcome measures overall. The posted ART-resistance results include two intervention-versus-control comparisons. Testing several outcomes or contrasts increases the chance that at least one unadjusted result may appear notable even when there is no underlying difference. General statistical explanation: multiplicity procedures, such as a prespecified testing hierarchy or adjusted significance criteria, can address this issue, but the appropriate method depends on the analysis plan and scientific priorities.

A multiplicity procedure or decision rule is not reported in the registry record. No reported analyses are listed for the secondary outcome record, and adjusted estimates are not posted. The ART-resistance p-values of 0.178 and 0.609 are explicitly unadjusted; they should not be treated as confirmatory evidence in the absence of a reported plan for interpreting multiple comparisons. The registry does not report interim analyses or stopping rules.

Missing-data handling is also not reported in the registry record. The primary population is described as participants with VL data available at or beyond 4 months post-ART initiation, which indicates that availability of VL data is relevant to inclusion in that outcome analysis. The record does not state how missing measurements were handled, whether participants without eligible VL data were excluded or otherwise accounted for, or whether sensitivity analyses were conducted. It also does not provide reasons for missing data or describe imputation.

The reported arm counts differ by outcome, but the registry does not explain the relationship between each outcome’s analysis population and randomized enrollment. No conclusions about attrition, exclusions, or the number of participants contributing repeated measurements should be inferred from these counts alone. The analysis population for the secondary ART-resistance outcome and the approach to missing secondary outcome data are not reported in the registry record. More complete information on prespecified analyses, missingness, and the population included in each model would be needed to assess robustness.

11. 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 serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: infant death (5.8%), infant hospitalization (2.2%), maternal death (1.8%), maternal hospitalization (1.8%).

The registry lists serious adverse events and other adverse events by arm, together with counts at risk. Serious adverse event counts are 32 among 276 at risk in Two-way SMS, 18 among 271 at risk in One-Way SMS, and 24 among 277 at risk in Control. Other adverse event counts are 15 among 276 at risk, 17 among 271 at risk, and 14 among 277 at risk, respectively. Death counts are 5, 3, and 4 in the same arm order.

These are descriptive counts, not adjusted comparisons. No confidence intervals, p-values, or comparative safety analyses are reported in the registry record. The counts alone do not establish that an intervention caused or prevented an event, and differences in counts should not be interpreted as differences in risk without a clearly defined outcome denominator and analysis. The registry also lists infant death, infant hospitalization, maternal death, and maternal hospitalization among the most frequent serious-event terms, with percentages of 5.8%, 2.2%, 1.8%, and 1.8%, respectively. The record does not provide enough detail to interpret those percentages as comparative risks or to determine the events’ relationship to the intervention. These safety data should therefore be read as reported event summaries, not as causal evidence.

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

The randomized design is intended to support comparisons of the two SMS strategies with standard care. However, the registry’s primary endpoint entry does not include the comparative estimates that would be needed to judge how precisely the arms differ in maternal virologic failure. It posts counts of VL measurement units, while describing a prevalence comparison using GEE with a log-binomial link. Without that model output and a clear account of the eligible measurement denominators, the primary question cannot be answered from the posted counts alone.

The secondary ART-resistance comparisons are more fully quantified, but both are unadjusted and their uncertainty intervals span zero. They do not demonstrate a clear difference, and they do not rule out effects in either direction. Because they are secondary outcomes among multiple outcomes and comparisons, they should not carry the same weight as a reported, prespecified primary analysis. They also do not directly answer questions about adherence, retention, perceptions, or virologic failure.

These limitations do not mean that the trial had no effect, nor do the reported entries establish that an effect occurred. Statistical uncertainty and incomplete reporting constrain what can be concluded. The registry record does not report a sample-size justification, the full missing-data approach, a multiplicity plan, or detailed analysis populations. Those omissions make it difficult to assess whether the available analyses align with the intended design and how sensitive the results may be to missing data or modeling choices.

The findings should be understood as a record of the stated trial design and the results posted for selected outcomes, not as a complete account of all study analyses. The record supports describing the randomized comparison and the observed entries, but stronger conclusions about effectiveness, equivalence, or safety would go beyond the reported statistical evidence. No treatment recommendation follows from these registry results.

13. Why This Trial Matters Statistically

This trial illustrates why outcome definitions, analysis populations, and reported model estimates need to align. The primary question concerns prevalence of maternal virologic failure, but the posted primary result is designated as a count of VL measurement units. A count may be useful descriptively, yet it does not itself express an arm-specific prevalence or a comparative effect. The registry’s stated GEE method indicates an intended model-based comparison, but the fitted estimates and their uncertainty are not reported.

The distinction is especially relevant when measurements may be correlated within participants. General statistical explanation: repeated observations from the same person are not necessarily independent. Methods that account for such dependence can affect both estimated uncertainty and the interpretation of the result. GEE is one approach for population-average comparisons, but an analysis label alone cannot substitute for its results or implementation details.

The secondary ART-resistance results also demonstrate the importance of uncertainty and multiplicity. The unadjusted estimates have intervals that include no difference, and multiple outcomes and contrasts are part of the broader study. A p-value should be read alongside the effect estimate, interval, and analysis context rather than as a stand-alone verdict. Planning illustrations based on observed effects likewise do not reveal the original study’s design adequacy or observed power.

For clinical researchers and biostatisticians, the central statistical lesson is that a transparent result should connect the endpoint definition, eligible data, model, estimate, and uncertainty. When any of these links is missing, restraint is necessary. The registry record provides enough information to identify the intended primary comparison and describe selected secondary and safety counts, while leaving important elements of the inferential analysis unavailable.

14. 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 limitations are specific to the information recorded for this trial. The primary outcome is defined as prevalence of virologic failure, with a threshold of HIV RNA at least 1000 c/ml, but the posted result is a count of VL measurements. The record does not provide the number of qualifying failure measurements, the eligible measurement denominators, or the GEE estimates. It is therefore not possible to assess the primary arm contrasts or their precision from the posted entries.

The registry does not report whether repeated VL measurements contributed for individual participants, how within-participant correlation was modeled, or which variance estimator was used. The stated GEE log-binomial method is not accompanied by a full specification. The primary analysis population is described in terms of VL data availability from 4 months post-ART initiation, but exclusions, missingness, and sensitivity analyses are not reported.

For ART resistance, the posted estimates are unadjusted, and the registry does not report a multiplicity strategy, a prespecified hierarchy, or adjusted comparisons. The remaining listed secondary outcomes do not have comparative results described in the registry record. The adverse-event figures are descriptive counts without comparative estimates or uncertainty intervals, and the record does not establish event attribution.

The trial’s sample-size justification and original statistical plan are not reported in the registry record. The planning illustrations use observed secondary effects and should not be read as design documentation. Additional limitations include the open-label design, with no masking, and the absence of registry entries on allocation concealment. These are limitations of the available record; they do not establish how unreported procedures were handled.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Mobile Strategies for Women's and Children's Health: Optimizing Adherence and Efficacy of PMTCT/ART — NCT02400671.
• Registry record last updated 2022-11-02; status last verified 2022-10.
• Results first posted 2022-11-02.
• 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.

18. Record Summary

NCT02400671 is a completed, randomized, parallel-group, unmasked trial sponsored by the University of Washington, with enrollment of 825. It compared weekly Two-way SMS and One-Way SMS strategies with standard of care among HIV-infected Kenyan mothers in a PMTCT-ART setting.

The primary outcome is maternal virologic failure after the first 4 months post-ART, defined as HIV RNA at least 1000 c/ml, with a stated timeframe of 2 years postpartum. The registry describes a GEE analysis with a log-binomial link, but the posted primary result consists of counts of VL measurement units rather than comparative model estimates.

Posted maternal ART-resistance comparisons are unadjusted, and both confidence intervals span zero. Details on multiplicity, missing-data handling, the original sample-size justification, and several analysis populations are not reported in the registry record. The safety entries are descriptive counts without comparative analyses.