Healthy Aging Resources to Thrive (HART): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Healthy Aging Resources to Thrive (HART) (NCT03739762), a randomized na study. The primary outcome is Change in Time Spent Sitting or Lying Down at 6 Months.

2. Clinical Question

3. Trial Design

HART, formally titled Healthy Aging Resources to Thrive, is recorded as a randomized, unmasked, parallel-design trial sponsored by Kaiser Permanente. Its official title describes a randomized trial to reduce sitting time and improve cardiometabolic health in obese older adults. The registry lists the phase as not applicable. Enrollment is recorded as 283, and the status is completed. The recorded start date is 2019-02-05 and the primary completion date is 2022-11-07.

The brief description characterizes the study as a two-stage randomized control design. At the first stage, participants were to be assigned to either an intervention aimed at decreasing sitting time or a healthy living control that did not focus on sitting time. After six months, the intervention group was to be randomized again: one part would continue the intervention, and the other would receive no further intervention. The summary states that all participants would be followed for 12 months. The registry excerpt does not provide outcome results for that later follow-up or explain how the second-stage assignments were analyzed.

The primary endpoint concerns change in sitting or lying time at six months, so the posted primary comparison addresses the initial intervention and control groups at that time point. The record describes the i-STAND program as including vibrating wristbands, standing desks, a workbook, and coaching calls. The control offered health coaching on topics unrelated to reducing sitting time and did not offer prompting devices or desks. Because masking is recorded as none, participants and trial personnel were not masked according to the registry entry. Further details about allocation procedures, concealment, and implementation fidelity 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: We had 11 participants consent\* to the study but who were never randomized thus \*not enrolled\*. Of these, 3 completed a baseline visit but before randomization, 2 never responded to our attempts to reach them (5 calls and one letter) and 1 said they were too busy to continue. The remaining 8 were non responsive to our attempts to reach them after initial consent. All of these participants were remote and thus their randomization took place up to 2 weeks after the baseline visit.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Number of participants who had a 6-month measure for change in sitting time

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 Change in Time Spent Sitting or Lying Down at 6 Months, measured over Change from Baseline at 6 months.

The primary endpoint is mean change from baseline in minutes spent sitting or lying down during waking hours at six months. The measure is described as objective: participants wore an activPAL for one week following each measurement visit. The registry identifies the population for this endpoint as participants with a six-month measure for change in sitting time. The question is not simply how much time participants spent sitting at one visit; it is how that measure changed from baseline by the six-month assessment.

Sitting and lying time is a direct behavioral target for an intervention designed to reduce sitting. Measuring it objectively can avoid relying solely on participants’ recollection or estimates of their own behavior. That general rationale does not mean an activity monitor captures every aspect of movement or explains why a participant’s measured time changed. The registry identifies minutes as the unit and reports the endpoint as a mean, with 95% confidence intervals for the arm-level changes.

The endpoint is distinct from the listed secondary outcomes, which include change in weight, BMI, and waist circumference at six months. It is also distinct from the trial’s broader title, which refers to cardiometabolic health. The posted primary result therefore addresses the measured sitting-or-lying outcome at the specified follow-up, rather than every behavior or health outcome of interest.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it is measured in minutes, and the registry identifies the reported quantity as a mean. For a continuous endpoint, a mean change summarizes the average change in the measured group; it does not describe how widely individual participant changes vary. The registry reports 95% confidence intervals around the arm-level means and around the sponsor’s comparative estimate. These intervals express statistical uncertainty under the analysis assumptions. A narrower interval would indicate greater precision, while a wider interval would indicate less precision; neither interval guarantees that the true value lies within it.

What do the arm-level changes describe?

The i-STAND mean change is -40.36 minutes, and the Healthy Living Control mean change is -8.51 minutes. Each is a within-arm change from baseline at six months, as described for the primary endpoint. The negative signs indicate that the measured average was lower at follow-up than at baseline. The confidence intervals describe uncertainty around each arm’s mean change, not the range of individual changes. Overlap or non-overlap between separate arm-level intervals is not, by itself, the registry’s formal test of the between-arm comparison.

What does the regression comparison describe?

The registry reports a linear regression analysis with a mean difference in final values of -31.85 minutes, a 95% confidence interval from -52.91 to -10.79, and p = 0.003. A regression model estimates a relationship between an outcome and one or more predictors. Here, the record identifies a linear regression and the comparison groups, but does not report the full model specification. It is therefore not possible to state whether the model adjusted for baseline values or other variables, or to describe the exact assumptions and diagnostics used.

The confidence interval for the regression contrast excludes zero. In general, when zero represents no difference in a mean contrast, an interval that excludes zero is evidence against that null value at the corresponding confidence level, subject to the model and data assumptions. The reported p-value supplies another measure of compatibility with a null comparison under those assumptions. It should not be read as an effect size, the probability that the result occurred by chance, or a measure of clinical importance. The registry’s estimate and its uncertainty are the relevant reported quantities.

Why does the outcome distribution matter?

Linear methods typically summarize a continuous outcome with a mean contrast. Their usefulness depends on whether the model gives a reasonable description of the outcome and its uncertainty. General statistical considerations include the distribution of residuals, influential observations, dependence among observations, and unequal variability across groups. These are general concepts, not findings about this trial. The registry record does not report diagnostic plots or checks, so no conclusion about those assumptions can be drawn from the available entry.

Over-dispersion is most often discussed for count outcomes, where observed variability exceeds what a simple count model expects. It is not the primary issue implied by this continuous minutes endpoint. More generally, a mismatch between a model and the data can make standard errors and confidence intervals unreliable. Robust methods or a different model may sometimes address such problems, but the registry does not report whether any such methods were used. The arm-level spread values are not reported in the record.

What does the superiority label mean?

The registry labels the analysis as superiority. In general, a superiority test evaluates whether the evidence supports a difference in a specified direction or supports a nonzero contrast, according to the trial’s test and analysis plan. The reported contrast is negative, and the registry gives p = 0.003. The record does not report the detailed hypothesis statement, sidedness of the test, or statistical plan, so those particulars should not be inferred beyond the posted label and results.

The study’s randomized design is relevant to interpretation because random assignment is intended to support a fair comparison between groups. Randomization does not eliminate every source of uncertainty, and the registry does not provide details on allocation concealment, adherence, crossover, or departures from assigned treatment. The analysis population and handling of missing observations are also not described. Those details matter because the estimate reflects the analyzed data and the methods used to include participants.

How should the confidence interval and p-value be read together?

They provide complementary information about the estimated contrast. The interval gives a set of values compatible with the observed data and model at the stated confidence level, while the p-value quantifies the evidence against the tested null under the analysis assumptions. Neither substitutes for the estimated magnitude or the context of the endpoint. The interval ranges from -52.91 to -10.79 minutes, which conveys uncertainty in the reported contrast while retaining its negative direction. Statistical evidence does not, on its own, establish the practical importance of the difference.

The registry reports six outcome measures and six analyses, while three secondary outcomes are named in the entry. It does not report whether or how the primary and secondary analyses were adjusted for multiplicity. When many outcomes or comparisons are examined, the chance of at least one apparently notable result can rise; this is a general statistical concern, not a claim that the primary result is invalid. The absence of a reported multiplicity procedure limits what can be said about the family of analyses as a whole.

10. Analyses Reported by the Sponsor

The posted sponsor analysis is a linear regression comparison labeled as a mean difference in final values. It reports -31.85 minutes, with a 95% confidence interval from -52.91 to -10.79 and p = 0.003. The record labels the comparison as a superiority analysis. No discrepancy is flagged between the posted result and a recalculated result, and the sponsor’s own analysis is the relevant reported comparative result.

The arm-level values separately report mean change from baseline: -40.36 minutes for i-STAND and -8.51 minutes for Healthy Living Control. Those summaries are not a substitute for the sponsor’s regression estimate. The registry’s wording distinguishes the comparison as a difference in final values, while the endpoint description and arm summaries concern change from baseline. Without a full model specification, it is not possible to establish precisely how the regression result relates to the arm-level change summaries or whether any adjustment was made.

The confidence interval and p-value describe statistical uncertainty and evidence under the sponsor’s analysis. Further details on covariates, missing-data handling, analysis population, and model diagnostics are not reported in the registry record. Accordingly, the sponsor result should be read with the stated endpoint and method, without treating unavailable analytical details as known.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists three secondary outcomes: change in weight, BMI, and waist circumference at six months. It also records six outcome measures and six analyses. This indicates that the primary endpoint is part of a broader set of recorded outcomes and analyses, but the entry does not provide a complete account of every comparison or the timing of all analyses. The methods used to control the chance of false-positive findings across multiple outcomes are not reported in the registry record.

Multiplicity is a general statistical issue: when researchers test several hypotheses, interpreting each p-value in isolation can overstate the strength of evidence across the full set. Adjustment methods, hierarchical testing, or a prespecified primary-outcome strategy can address different aspects of this concern. The record does not say whether any such approach was planned or used. The reported primary p-value should therefore be read as the result for its stated comparison, not as a complete multiplicity-adjusted assessment of every outcome.

The endpoint population is described as participants with a six-month measure for change in sitting time. The posted results list 124 participants for i-STAND and 122 for Healthy Living Control. The registry does not explain why other enrolled participants are not represented in these endpoint counts, whether all were randomized, or whether missing six-month measurements arose from withdrawal, missed visits, unusable monitor data, or other reasons. Such explanations are not reported in the registry record.

Missing-data methods can affect an estimate if participants with observed outcomes differ systematically from those without them. The entry does not report imputation, weighting, sensitivity analyses, or assumptions about missingness. It also does not identify an intention-to-treat or per-protocol analysis population. No conclusion about the direction or size of any missing-data effect is justified without those details. The reported result should be understood as the analysis presented for the stated endpoint population.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: unrelated sae (1.4%).

The safety entry reports serious adverse events for i-STAND and Healthy Living Control. The at-risk counts are 140 and 143, respectively; six serious adverse events are recorded in each group. Deaths are recorded as zero in both groups. These are descriptive registry counts. They do not by themselves establish whether an event was caused by the intervention, nor do they summarize the severity, duration, or clinical consequences of every event.

The record also lists other adverse events, with counts of 72 for i-STAND and 47 for Healthy Living Control among the stated at-risk groups. These counts should not be interpreted as numbers of participants affected unless the record makes that distinction; an adverse-event count can reflect events rather than unique people. The registry identifies Unrelated SAE as a leading serious-event term and gives 1.4%. It does not provide enough detail in this entry to interpret that percentage’s precise denominator or determine how events were adjudicated.

Comparisons of safety counts require caution because the registry does not report event-level descriptions, exposure time, statistical comparisons, or an attribution analysis. A count alone is not a rate adjusted for time under observation, and no such rate should be inferred. The figures describe what the safety entry records; they are not a complete assessment of safety or a basis for medical advice.

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

The primary finding is a difference in the posted six-month sitting-or-lying time results between the randomized groups. The intervention arm’s mean change was negative, and the control arm’s mean change was also negative but closer to zero. The sponsor’s linear regression comparison is negative, with a 95% confidence interval that does not include zero and a reported p-value of 0.003. In the context of the endpoint, this pattern is consistent with a larger average reduction in measured sitting or lying time in i-STAND than in Healthy Living Control at six months.

This result does not mean that every participant reduced sitting time, or that the average change describes each person’s response. A group mean can coexist with substantial variation in individual outcomes, and arm-level spread estimates are not reported in the registry record. The confidence interval gives uncertainty around the estimated group contrast under the analysis assumptions; it is not a prediction interval for individuals.

The result is limited to the measured behavior and follow-up described. It does not directly establish effects on weight, BMI, waist circumference, or all aspects of cardiometabolic health. Those measures are listed as secondary outcomes, but their results are not included in this entry. Although the trial summary describes follow-up for 12 months and a second randomization after six months, the primary result concerns the six-month comparison and does not establish later effects.

Statistical evidence and clinical importance are different questions. The reported confidence interval and p-value address uncertainty in a statistical comparison, while judging practical importance would require context about meaningful changes and participant experience that is not reported in the registry record. The entry also lacks detail on missing data, model specification, and multiplicity procedures. These limits do not erase the reported result, but they define how narrowly it should be interpreted.

14. Why This Trial Matters Statistically

HART is statistically interesting because it evaluates a behavioral target with an objective measure and a randomized comparison. The primary endpoint focuses on minutes spent sitting or lying during waking hours, assessed with an activPAL worn for one week after measurement visits. That approach gives the study a direct measure aligned with the intervention’s stated goal, rather than relying only on a participant’s general impression of sitting behavior.

The design also includes a second-stage randomization after six months for the original intervention group. Such a design can, in principle, address questions about continuing versus stopping an intervention after an initial period. The registry summary describes that planned structure, although the posted primary result concerns the initial comparison at six months and does not report the later-stage results. The distinction between the initial randomized comparison and later treatment paths matters when defining the question each analysis can answer.

The record makes visible both an arm-level summary and a regression-based comparison. These are useful reminders that within-group change and between-group contrast are different quantities. A group may change over time for many reasons, while the comparative estimate addresses how the groups differ under the reported analysis. Confidence intervals help communicate precision, and the reported p-value conveys evidence under the model, but neither alone describes individual variation or practical importance.

The trial also illustrates why analysis details matter. The registry does not report the full regression specification, missing-data methods, analysis population, or multiplicity strategy. These omissions limit assessment of robustness and of the broader outcome family. Statistical interpretation is strongest when the endpoint, estimand, model, and population are clear; the available record supports a focused reading of the six-month primary result, not a broader claim about every outcome or follow-up period.

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.

The registry entry does not provide the full statistical analysis plan or enough detail to reconstruct the sponsor’s linear regression. In particular, covariate adjustment, baseline handling, model diagnostics, and the exact test specification are not reported in the registry record. The comparison is labeled a mean difference in final values, while the endpoint is described as change from baseline; the information available does not resolve all details of that distinction.

The endpoint counts are smaller than the recorded enrollment, but the entry does not explain participant flow into the six-month analysis. The reasons for unavailable measurements, missing-data assumptions, imputation methods, and sensitivity analyses are not reported. The analysis population is also not named as intention-to-treat, per-protocol, or another defined population. These gaps limit assessment of how representative the posted endpoint analysis is of the randomized groups.

The record lists six outcome measures and six analyses, while naming three secondary outcomes. It does not describe the prespecified hierarchy of outcomes or any multiplicity correction. Results for the listed secondary outcomes are not included in this entry. The brief description mentions a second randomization after six months and follow-up for 12 months, but the primary result reported is at six months; the later results are not reported in this record.

The trial is recorded as unmasked. The registry does not provide details about adherence, intervention exposure, contamination between groups, or fidelity to the coaching program. The intervention and control also differ in their stated content and offered devices, so the posted contrast concerns those assigned programs as described, not an isolated effect of one component. These limitations should be considered alongside the reported estimate without assuming that any one of them changed the result in a particular direction.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Healthy Aging Resources to Thrive (HART) — NCT03739762.
• Registry record last updated 2024-06-25; status last verified 2024-05.
• Results first posted 2023-12-05.
• 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

HART is a completed, randomized, unmasked, parallel-design trial sponsored by Kaiser Permanente. It enrolled 283 and studied an intervention intended to reduce sitting time in older adults with obesity and a sedentary lifestyle. The two arms were i-STAND and Healthy Living Control. The trial summary describes a second-stage randomization after six months and follow-up for 12 months.

The primary endpoint was mean change from baseline in waking-hour sitting or lying time at six months, measured objectively with an activPAL. The posted arm means were -40.36 minutes for i-STAND and -8.51 minutes for control. The sponsor’s regression comparison reported a mean difference in final values of -31.85 minutes, with a 95% confidence interval from -52.91 to -10.79 and p = 0.003.

The registry does not report full model details, missing-data methods, or multiplicity procedures. The reported result should be interpreted as a six-month comparison for the specified endpoint, not as a result for every listed outcome or the later follow-up.