Testing Adaptive Interventions to Improve Physical Activity for Sedentary Women: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Testing Adaptive Interventions to Improve Physical Activity for Sedentary Women (NCT03558828), a randomized na study. The primary outcome is Physical Activity-Steps.

2. Clinical Question

3. Trial Design

The registry lists a randomized, single-masked, sequential study with allocation described as RANDOMIZED and the model described as SEQUENTIAL. The study is marked completed. Its stated aim was to determine an adaptive intervention combining physical activity monitoring, motivational text messages, motivational personal calls, and group meetings. The registry lists Rush University Medical Center as sponsor.

The initial treatment stage assigned participants to an enhanced physical activity monitor alone or the monitor with motivational text messaging. The monitor included goal setting and a physical activity prescription, according to the arm description. At Week 8, response was assessed against a physical activity step goal. The record says that participants who met the goal continued with the same initial component for weeks 9-34. The detailed description of the nonresponder pathway is truncated, so its full assignment rules are not reported in the registry record. The intervention listings include augmented treatments using calls or meetings.

For Weeks 35-50, the registry says all participants returned to an enhanced physical activity monitor-only component. This staged structure matters statistically: the study is not simply a comparison of two fixed treatments maintained unchanged throughout follow-up. Interpretation depends on the treatment sequence and response-based decisions described in the protocol.

The registry gives the study start as 2018-08-24 and primary completion as 2023-08-28. Phase is listed as NA. Details such as the identity of the masked parties, allocation concealment, randomization procedures, and the statistical plan are 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.

Before assignment: Excluded: 453 * Unable to be reached: 163 * Not Interested: 193 * Ineligible: 97 * failed phone screening: 14 * failed health screening: 33 * failed physical activity screening: 50

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Total N=301 participants were analyzed.

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: Baseline characteristics are available for N=301 participants in Step 1 only (Weeks 1-8) as all 301 participants progressed to Steps 2 and 3.

6. Primary Endpoint

The registry lists the primary outcome as Physical Activity-Steps, measured over Baseline plus change is being assessed after 8 weeks, 34 weeks, and 50 weeks for steps..

The primary outcome, Physical Activity-Steps, is a continuous measure reported as the mean number of steps per day. Average daily steps were measured with an ActiGraph worn on the right hip during waking hours for one week. The registry describes assessment at baseline and change after 8 weeks, 34 weeks, and 50 weeks.

Step counts provide a direct numerical measure of ambulatory activity over the measurement period. In general, a continuous endpoint can retain gradations in the measured outcome rather than reducing it to a yes-or-no category. The mean describes the group average, while a confidence interval conveys uncertainty around that estimated mean under the analysis used. A confidence interval does not show the range of individual participants’ step counts.

The posted arm summaries report means and 95% confidence intervals. They do not include a standard deviation in the registry record. The registry also reports a GEE effect size estimate, but does not identify its scale or provide interval bounds or a p-value. Consequently, the endpoint definition and arm-level summaries can be described, but the reported material does not establish a quantified treatment contrast that can be interpreted directly.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a mean describe?

A mean is the arithmetic average of the measured outcome within a group. For this trial, the outcome is average daily steps, expressed in steps/day. In general, a mean gives a useful summary of the group’s central level, but it does not describe how widely individual observations vary. The registry reports group means, but does not report a standard deviation for these summaries.

The arm means are presented with 95% confidence intervals. In general, a confidence interval describes uncertainty in an estimated quantity under the assumptions of the analysis. It is not an interval that is expected to contain a specified proportion of individual participants’ measurements. It also does not, by itself, answer whether two treatment groups differ: that requires a contrast and its uncertainty, or a suitable test.

What does the GEE result add?

The registry lists GEEs and an effect size estimate of .41, with the confidence level marked as 95. Generalized estimating equations are a general approach for estimating average relationships when observations may be correlated, such as repeated measurements from the same participant. In general, accounting for correlation can affect the estimated uncertainty compared with treating every observation as independent. The record does not describe the working correlation structure, model terms, handling of time, or the exact effect-size definition used for this analysis.

The missing interval bounds and p-value limit interpretation of the .41 estimate. Without the scale, it is not clear how to express the estimate in clinical units or what contrast it represents. The estimate cannot be used to reconstruct a confidence interval or a test result. The registry’s stated 95 confidence level does not supply the omitted interval endpoints.

Why does the sequential design matter?

A SMART design studies treatment strategies that can adapt as participants’ responses become known. In general, the effect of a strategy can depend on both the initial assignment and the later decision rule. The registry describes initial treatment assignment, response assessment at Week 8, continued initial treatment for responders through weeks 9-34, and a maintenance phase during Weeks 35-50 in which all participants returned to monitor-only treatment. The full nonresponder assignment description is truncated in the record.

For that reason, the listed intervention groups should not automatically be interpreted as separate, fixed regimens applied uniformly over the whole study. A treatment contrast needs to specify which strategies are being compared and how the adaptive stages are handled. The registry does not provide enough detail in the posted analysis to identify all such contrasts or their estimates.

Why is a two-arm recalculation unavailable?

The registry record states that values are missing or not estimable for a two-arm comparison of this continuous endpoint. A recalculated comparison would require a defined two-arm contrast and suitable summary information for that contrast. The arm means and confidence intervals do not supply those missing elements on their own. Thus, no unadjusted two-arm effect estimate, confidence interval, or p-value is available for interpretation from the posted summaries.

In general, a p-value evaluates how compatible the observed data are with a specified null hypothesis under a statistical model; it is not a measure of the size or importance of an effect. A confidence interval for a contrast would communicate both its direction and precision. Neither can be inferred from the registry’s group means without making additional calculations or assumptions, which are not appropriate for the posted record.

How should the planning illustrations be read?

The sample-size calculations are planning illustrations based on the observed effect, not the trial’s own design. Their 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 a stated fraction of the observed effect. In general, such calculations illustrate how assumptions about a target effect and design relate to a prospective sample size. They are not observed power and do not establish the study’s original sample-size justification. That justification is not reported in the registry record.

10. Analyses Reported by the Sponsor

The registry-reported analysis lists GEEs and gives an effect size estimate of .41 for the intervention groups. Its confidence level is marked as 95, but the interval bounds and p-value are not reported in the registry record. The effect-size scale and the precise contrast represented by the estimate are also not stated. Those omissions prevent a precise interpretation in steps/day or a clear assessment of statistical uncertainty.

No recalculated two-arm comparison is available for this continuous endpoint: the registry record says such values are missing or not estimable. The posted arm means and confidence intervals are descriptive summaries and cannot replace a contrast-specific estimate. The discrepancy flag is false, so there is no stated conflict between the registry-reported analysis and a recalculation. In any case, no two-arm recalculation is available to compare with the reported GEE estimate.

The most defensible reading is therefore limited: the registry reports an overall GEE effect-size estimate, but does not provide enough information to determine its units, precision, or exact comparison. It should not be presented as a quantified difference in steps or as evidence that one adaptive strategy outperformed another.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 3 secondary outcomes described as cardiovascular health, 5 outcome measures, and 2 analyses. It does not report how many hypotheses were tested, which outcomes were included in each analysis, or whether any adjustment for multiple comparisons was planned or applied. The statistical plan is not reported in the registry record. As a result, the record does not support a conclusion about how multiplicity was handled.

In general, testing multiple outcomes, time points, or treatment contrasts can increase the chance of observing an apparently notable result by chance. Methods such as prespecified outcome hierarchies or multiplicity adjustments can address this issue, but whether such methods were used in this trial is not reported in the registry record. The available information is insufficient to infer a testing strategy from the number of posted outcomes or analyses.

The primary outcome timeframe includes baseline and change after 8 weeks, 34 weeks, and 50 weeks. How the repeated measurements were modeled, how missing observations were handled, and whether all time points were analyzed according to a prespecified plan are not reported in the registry record. The GEE method is named, but model specifications and missing-data procedures are not given.

The record states that total N=301 participants were analyzed, and the posted arm summaries include group-specific analysis counts. It does not explain whether this was an intention-to-treat, per-protocol, or other analysis population, how the analysis population was defined, or how many measurements were unavailable at each assessment. The reasons for missing data and any imputation approach are not reported in the registry record. These gaps limit assessment of how well the analysis reflects the randomized assignments and how sensitive the results may be to incomplete follow-up.

12. 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 1% reporting threshold.

Most frequent non-serious adverse event terms

The adverse-event record lists zero serious adverse events in each displayed group and zero deaths in each displayed group. For other adverse events, the monitor-only group has 3 affected participants; the other displayed groups have 0. The numbers at risk are reported separately for each group in the registry record.

These figures are counts in the registry’s adverse-event fields. They do not, by themselves, establish comparative safety. The record does not report event definitions, severity grading beyond the serious-event classification, the collection period, the method of ascertainment, or whether the reported counts represent unique participants or event episodes. It also does not provide statistical comparisons or uncertainty estimates for these counts.

Zero recorded serious events should be read as a report of zero in the registry fields, not as proof that no such events could occur. Likewise, the other-event counts are descriptive and should not be interpreted as evidence of a causal difference between intervention strategies. The relevant exposure and follow-up details are not reported in the registry record, so the counts cannot be converted into rates or used to make a risk comparison.

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

The posted results establish that the registry recorded average daily step counts for the listed intervention groups, with group means and 95% confidence intervals. The outcome was measured by an ActiGraph worn on the right hip during waking hours for one week. Assessment was described at baseline and for changes after 8 weeks, 34 weeks, and 50 weeks. The registry also reports a GEE effect-size estimate of .41, but does not give its scale, interval bounds, or p-value.

The record does not provide an estimable two-arm comparison for the continuous endpoint. Therefore, the posted summaries do not quantify how much one intervention strategy differed from another, nor how precise such a difference would be. The arm means are not substitutes for a direct contrast, and the overall GEE estimate cannot be interpreted as a step-count difference without knowing its scale and model specification.

These results also do not identify which treatment sequence should be selected or establish that a particular component caused a change in physical activity. The study’s adaptive structure makes the definition of the strategy and the response-based treatment decisions important to any causal comparison. Complete details of the nonresponder pathway and the analysis plan are not reported in the registry record.

More broadly, the posted record does not report the missing-data approach, multiplicity strategy, or a detailed account of the analysis population. Those omissions constrain the conclusions that can be drawn from the numerical summaries. The results are best read as descriptive group-level step counts accompanied by an incompletely specified model estimate, rather than as a fully interpretable treatment-effect comparison.

14. Why This Trial Matters Statistically

Statistically, this study is notable because its sequential design is intended to examine adaptive treatment strategies rather than only fixed treatment assignments. A SMART design can address questions about what treatment to begin with and how to adapt treatment after a response is observed. In general, that structure can make the estimand—the precise treatment strategy contrast being estimated—especially important.

The registry record shows why reporting that estimand and its uncertainty matters. It names GEEs and gives an effect-size estimate, but does not specify the scale, contrast, interval bounds, or p-value. It also says that values are missing or not estimable for a two-arm comparison of the continuous endpoint. Readers therefore have descriptive step-count summaries but lack a directly interpretable comparison between adaptive strategies.

The study also illustrates the distinction between descriptive estimates and inferential results. Group means and confidence intervals describe observed summaries and their reported precision; they do not answer every comparative question. A model label alone does not supply the model specification, and an effect-size point estimate alone does not communicate its uncertainty or units.

These reporting details are important for researchers planning or interpreting sequential trials. Clear definitions of treatment strategies, response rules, analysis populations, missing-data procedures, multiplicity handling, and contrast-specific estimates help readers understand what a trial can establish. In this record, several of those elements are not reported, so statistical interpretation must remain focused on the information actually available.

15. 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 specific to interpretation arise from the registry record. The primary endpoint is described and arm-level means with 95% confidence intervals are posted, but the record says values are missing or not estimable for a two-arm comparison of the continuous endpoint. This prevents a direct unadjusted comparison from being read from the posted results.

The reported GEE estimate of .41 lacks its scale, confidence interval bounds, and p-value. The GEE model specification, including its time structure and correlation assumptions, is not reported in the registry record. This makes it difficult to determine what the estimate represents or how precisely it is estimated.

The detailed description of the nonresponder pathway is truncated. Although the record describes the responder pathway and the common maintenance component, it does not give the full decision rules for all adaptive assignments. The exact treatment strategies represented by the reported model therefore cannot be reconstructed from the registry record.

Other important analysis details are not reported: the sample-size justification, interim analyses, missing-data procedures, multiplicity plan, allocation concealment, identity of masked parties, and the definition of the analysis population. Participant characteristics and detailed eligibility criteria are also not reported in the registry record. These omissions limit assessment of the design, risk of bias, precision, and generalizability. They do not change the posted numbers, but they restrict the conclusions those numbers can support.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Testing Adaptive Interventions to Improve Physical Activity for Sedentary Women — NCT03558828.
• Registry record last updated 2025-08-03; status last verified 2025-07.
• Results first posted 2025-05-30.
• 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.

19. Record Summary

The registry describes a completed randomized, single-masked, sequential study sponsored by Rush University Medical Center. It reports total N=301 participants analyzed and identifies average daily steps as the primary outcome, measured with an ActiGraph during waking hours for one week. The primary assessment includes baseline and change after 8 weeks, 34 weeks, and 50 weeks.

Arm-level means and 95% confidence intervals are posted. The registry also reports a GEE effect-size estimate of .41, without interval bounds, a p-value, or a stated scale. Values are missing or not estimable for a two-arm comparison of the continuous endpoint. Details of missing-data handling, multiplicity, and several analysis specifications are not reported in the registry record.