SuPRA: Using Wearable Activity Trackers With a New Application to Improve Physical Activity in Knee Osteoarthritis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for SuPRA: Using Wearable Activity Trackers With a New Application to Improve Physical Activity in Knee Osteoarthritis (NCT02585323), a randomized na study. The primary outcome is Daily Number of Minutes in Moderate/Vigorous Physical Activity.

2. Clinical Question

The study is relevant to a behavioral question: whether support combining self-monitoring technology, education, and counselling can change measured activity in a population for whom physical activity is an important part of care. The outcome is device-measured rather than solely self-reported, but it captures a particular definition of bouted moderate/vigorous activity. It does not, by itself, describe every dimension of activity, sedentary behavior, symptoms, or daily functioning.

3. Trial Design

The registry identifies SuPRA as a randomized trial with no masking and a crossover model. It records enrollment of 51 and a completed status. The sponsor is the University of British Columbia. The recorded study start was 2017-04-01 and the primary completion date was 2019-08-21. The registry lists phase as not applicable.

The behavioral intervention paired the Fitbit Flex with FitViz and a brief education session and physiotherapist counselling. In the immediate group, the intervention was delivered in Months 1-3. The delayed group was assigned the same intervention with a 3 month delay, starting in Month 4. The record says that in Months 6-9 participants continued to use Fitbit/FitViz without the physiotherapist phone call. The available description does not fully specify every period, sequence, or measurement occasion, so those details should not be inferred from the crossover label alone.

There is an important design-description ambiguity. The registry lists the model as crossover, while its note about analysis states that the trial is cluster-randomized and warns that a simple individual-level re-analysis would be misleading. The registry record does not explain how those descriptions fit together or identify the unit of randomization in further detail. The sponsor's own analysis is therefore the reported statistical result; an independent individual-level re-analysis is not presented.

The primary outcome was measured at baseline, 13 weeks, 26 weeks, and 39 weeks. The record also lists secondary measures covering sedentary minutes, five KOOS domains, PHQ-9, and Partners in Health Scale. It reports 16 outcome measures and 16 analyses. The specific statistical plan, allocation procedures, and analysis population are not reported in the registry record. The study description states the intervention rationale, but a sample-size justification 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.

Baseline

Week 13 (RCT Period)

Reasons for not completing this period:

Week 26 (All Received Intervention)

Reasons for not completing this period:

Week 39 (All Received Intervention)

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: 13 weeks: Immediate (I) Group: 1 withdrew due to Fitbit wear issues; 1 withdrew, reason unknown. Delayed (D) Group: 1 refused assessment 24 weeks: I Group: 1 withdrew due to family issues, 1 reason unknown 29 weeks: I Group: 2 withdrew - reason unknown

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 Daily Number of Minutes in Moderate/Vigorous Physical Activity, measured over Baseline, 13 weeks, 26 weeks, 39 weeks.

The primary endpoint was the daily number of minutes spent in moderate/vigorous physical activity (MVPA), expressed as a mean in minutes per day. The registry says the measure came from a SenseWear Mini sensor worn over a 7-day period. It used bouted MVPA: a bout was defined as at least 10 consecutive minutes at a level of at least 3 METs, with allowance for an interruption of up to two minutes below the threshold.

This definition focuses on sustained activity rather than counting every isolated minute that reaches the intensity threshold. A daily mean gives an interpretable summary of the measured activity over the monitoring period. Device measurement can provide an outcome distinct from participants' recollection or perceptions, although the registry record does not report further information about device wear adherence, valid-day rules, or measurement quality procedures.

MVPA was selected because the trial aimed to improve physical activity and reduce sedentary time. It represents one particular activity target: the time spent in qualifying bouts at the stated intensity. It is not a complete measure of all movement, sedentary behavior, fitness, or health. The record lists daily sedentary minutes as a secondary outcome and includes KOOS symptom, pain, sport and recreation, activity of daily living, and quality of life measures, alongside PHQ-9 and Partners in Health Scale. Those measures address other aspects of participants' experience and are not interchangeable with the primary activity endpoint.

The registry's primary analysis reports an adjusted difference in mean change, rather than simply reporting a group mean at a single occasion. The exact occasion contrast and the direction of group subtraction are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

ANCOVA, or analysis of covariance, is a regression-based method that compares an outcome between groups while adjusting for one or more other variables. In a trial, adjustment can account for measured baseline values or other prespecified characteristics, depending on the model. General explanation: adjustment may improve precision when the covariates are informative and appropriately included, but it does not remove uncertainty about unmeasured factors or compensate automatically for an unclear design.

For this study, the registry labels the reported method ANCOVA and the parameter an adjusted difference in mean change. It reports 13.1 minutes per day, a 95% confidence interval from 1.6 to 24.5, and p = 0.05. The record does not describe the covariates, the exact outcome occasion or occasions used in the contrast, or the coding that determines the order of the group subtraction. Because the contrast direction is not stated, the positive estimate should not be translated into a claim that the immediate or delayed group had the greater change.

What does the confidence interval convey?

A confidence interval expresses sampling uncertainty around an estimate under the assumptions of the model and analysis. General explanation: a wider interval indicates less precision than a narrower interval on the same scale. The interval runs from 1.6 to 24.5 minutes per day. It provides more context than the point estimate alone, but does not identify the probability that a particular treatment effect is true, nor does it establish practical importance. The registry does not report a threshold for meaningful change in this endpoint.

How should the p-value be read?

The reported p-value is 0.05. In general, a p-value describes how incompatible the observed data are with a specified null model, conditional on the assumptions used for the analysis. It is not the probability that the null hypothesis is true, and it does not measure the magnitude of an effect. The record does not state whether the p-value was adjusted for the number of outcomes or analyses. The result should therefore be considered alongside the interval, analysis details, and the broader set of measured outcomes.

Why does the design description matter?

The registry gives the trial model as crossover but its analysis note describes a cluster-randomized design. These labels have different implications for how measurements relate to one another and how uncertainty should be estimated. General explanation: observations from participants in the same randomized cluster may be correlated, so an analysis that treats every participant as independent can misrepresent precision. The registry itself states that a simple individual-level re-analysis would be misleading and that only the sponsor's own analysis is shown. The record does not clarify the relationship between the crossover and cluster-randomized descriptions. Consequently, the sponsor's reported estimate should not be replaced with a simple calculation from group summaries.

The record's reported parameter is an adjusted difference in mean change, while the endpoint itself is a continuous measure summarized by a mean and standard deviation. General explanation: a mean describes the average outcome, and a standard deviation describes variability among observations around that mean. The posted result does not give the underlying group means or standard deviations, so no separate unadjusted estimate can be reconstructed in the registry record in this account. No recalculated estimate is reported.

What about planning illustrations?

Sample-size illustrations based on an observed effect are planning scenarios, not estimates of what the completed study was powered to detect. General explanation: a planning calculation asks how many participants per arm might be needed under stated assumptions, whereas observed power is not a substitute for interpreting the confidence interval from completed data. The registry's stated illustration assumptions are a two-sided significance level of 0.05, equal allocation between arms, a normal approximation, and effects defined as fractions of the observed effect. Those illustrations do not establish the original design's sample-size justification, which is not reported in the registry record. The registry also notes that time-to-event illustrations use total events, while other illustrations use participants per arm; this is a continuous endpoint, not a time-to-event outcome.

In general, over-dispersion means that variability is greater than a model assumes. A model that fails to account for relevant dependence or extra variability may give confidence intervals that are too narrow and p-values that are too small. This is a general statistical concern, not evidence that a particular problem occurred in this study. The registry's warning about individual-level re-analysis is a reason to respect its stated analysis structure, but the full sponsor model is not reported in the registry record.

Finally, multiple endpoints and analyses raise a general multiplicity question: when many statistical tests are performed, the chance of at least one apparently notable result can increase. The record lists 16 outcome measures and 16 analyses, but does not report a multiplicity adjustment or identify the full testing hierarchy. The primary outcome is identified, yet the statistical plan governing interpretation across outcomes is not reported in the registry record. This limits what can be inferred from a single p-value considered in isolation.

10. Analyses Reported by the Sponsor

The registry reports the sponsor's own ANCOVA analysis as an adjusted difference in mean change of 13.1 minutes per day, with a 95% confidence interval from 1.6 to 24.5 and p = 0.05. The record's analysis note says the trial is cluster-randomized and that a simple individual-level re-analysis would be misleading; it therefore shows only the sponsor's analysis. No recalculated estimate is presented alongside it.

The registry also identifies the overall model as crossover, without explaining how that description relates to the cluster-randomized note. In the absence of the sponsor's full model specification, group contrast coding, and analysis population, the estimate should be read as the sponsor-reported result rather than reconstructed from a simplified comparison. The exact named-group direction of the estimate is not reported in the registry record.

The posted confidence interval and p-value convey uncertainty under the sponsor's analysis, but the record does not give enough detail to evaluate all model choices independently. In particular, the covariates and handling of missing observations are not reported. The result can be described accurately as reported, while recognizing that the registry record does not fully expose the analysis needed for a complete methodological audit.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 16 outcome measures and 16 analyses. The primary endpoint is daily minutes in bouted MVPA. Listed secondary outcomes include daily sedentary minutes; KOOS symptom, pain, sport and recreation, activity of daily living, and quality of life; PHQ-9; and Partners in Health Scale. The record does not report whether statistical testing across these outcomes was adjusted for multiplicity, whether a testing hierarchy was specified, or which results were considered confirmatory. It is therefore not possible to assess from the registry record whether the reported p-value was interpreted within a prespecified family of tests.

Information on analysis populations is also limited. The record reports enrollment and some reasons for withdrawal around outcome assessments, but does not state whether the primary analysis followed an intention-to-treat principle, used a per-protocol population, or applied another definition. The number of participants contributing to the primary ANCOVA is not reported in the registry record. Nor are the approach to missing outcomes, assumptions about missingness, or any sensitivity analyses reported.

The outcome description records withdrawals at 13 weeks: in the immediate group, one participant withdrew because of Fitbit wear issues and one withdrew for an unknown reason; in the delayed group, one refused assessment. It also lists immediate-group withdrawals at 24 weeks, one due to family issues and one for an unknown reason, and at 29 weeks, two for unknown reasons. These timings do not align directly with all listed primary measurement occasions, which are baseline, 13 weeks, 26 weeks, and 39 weeks. The record does not explain the relationship between the withdrawal timings and the analysis data.

For a repeated-measure or crossover design, missing observations can affect which comparisons remain available and may influence estimates if missingness is related to outcome or intervention exposure. General explanation: statistical methods can use incomplete repeated observations under assumptions, but the validity of those methods depends on the design and missing-data process. No method for addressing missingness is reported in the registry record. The design note also cautions against simple individual-level re-analysis, while the relation between the stated crossover model and cluster-randomized description remains unclear.

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 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry reports serious adverse events for the immediate group and delay group separately. In the immediate group, 26 participants were listed at risk; there were 0 serious adverse events, 7 other adverse events, and 0 deaths. In the delay group, 25 participants were listed at risk; there were 0 serious adverse events, 3 other adverse events, and 0 deaths. No top serious adverse event terms are listed.

These figures describe the adverse-event counts recorded in the registry. They do not establish whether an event was caused by the intervention, whether events were anticipated, or how events were collected and reviewed; those details are not reported in the registry record. The category “other adverse events” is not further specified in the registry entries, so the counts should not be assigned a particular type or severity beyond the registry's label.

The absence of reported serious adverse events in these group summaries is not evidence that no adverse events of any kind occurred: the registry separately lists other adverse events. Nor do these counts establish a comparative safety effect. The record does not report an analysis of adverse-event differences or information about exposure-adjusted event rates. These small-group counts are descriptive registry entries, not a substitute for a detailed safety analysis.

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

The posted result describes an adjusted difference in mean change in daily bouted MVPA: 13.1 minutes per day, with a 95% confidence interval from 1.6 to 24.5 and p = 0.05. It is a statistical estimate for the comparison specified in the sponsor's analysis. The interval indicates uncertainty around that estimate, and the p-value should not be interpreted as the size or importance of the difference. Because the registry does not report group contrast coding, the estimate cannot be assigned confidently to a direction between the immediate and delayed groups.

The endpoint is a device-measured activity outcome with a defined bout threshold and monitoring period. The finding does not describe all physical activity, sedentary time, symptoms, function, quality of life, mental health, or self-management. Although those other outcomes appear in the registry, their results are not reported in the registry record. The primary result should not be generalized to those measures.

The study evaluated a bundled behavioral intervention: education, Fitbit/FitViz use, and physiotherapist counselling. The registry's description does not separate the contribution of each component. The delayed intervention schedule also means that intervention exposure varied over the study period; the record does not provide enough detail to reconstruct every period-specific comparison.

The result does not establish a clinically meaningful benefit, because a meaningful-change threshold is not reported. It also does not establish effects beyond the measured study periods or in populations not described by the record. The record lists enrollment of 51, but does not report the primary analysis population or the handling of missing observations. Those omissions matter when judging precision and generalizability.

Finally, the trial model is labeled crossover, while the statistical note calls it cluster-randomized and cautions against a simple individual-level re-analysis. The sponsor's adjusted analysis is therefore the result to describe, but the registry record does not provide enough methodological detail to independently evaluate every aspect of that analysis. The evidence is best read as one reported estimate with an interval and limitations, not as a universal conclusion about the intervention.

14. Why This Trial Matters Statistically

This trial raises several useful statistical questions for researchers studying behavioral interventions. First, the intervention is a package: wearable tracking, an application, education, goal setting, and physiotherapist counselling. The primary comparison concerns the package as described, rather than the isolated effect of any one component. Recognizing that distinction helps keep interpretation aligned with the trial question.

Second, activity was defined operationally. The primary endpoint was not a broad, undefined measure of being active; it was daily bouted MVPA measured by a SenseWear Mini over 7-day periods, using a stated intensity threshold and allowance for brief interruption. Such definitions determine what a treatment contrast means. A change in this outcome need not imply a corresponding change in every movement pattern or patient-reported outcome.

Third, the record illustrates why design labels and analysis details matter. It identifies a crossover model but also includes an analysis note describing cluster randomization and warning against individual-level re-analysis. These features influence the dependence structure that an analysis may need to account for. Because the registry does not reconcile the descriptions, a reader should avoid over-interpreting a simplified comparison.

Finally, the reported adjusted estimate is accompanied by a confidence interval and p-value, but the registry does not fully describe the contrast, missing-data method, or multiplicity plan. These omissions are common reasons to distinguish a numerical result from a complete statistical account. The result remains interpretable as the sponsor's reported analysis, while its scope and precision must be considered alongside the unavailable methodological details.

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 this record constrain interpretation. The most immediate is ambiguity in the design description: the trial model is listed as crossover, while the statistical note describes cluster randomization. The record does not explain the relationship between these descriptions, identify the clusters, or provide enough detail to reconstruct the allocation and analysis structure. It explicitly cautions that a simple individual-level re-analysis would be misleading.

The primary contrast is incompletely described. The registry reports an adjusted difference in mean change but does not state the group subtraction order, the precise time-point comparison, or the covariates in the ANCOVA. The estimate therefore cannot be assigned to the immediate or delayed group direction with confidence, and the adjustment cannot be independently assessed from the registry record.

The record lists withdrawals at 13 weeks, 24 weeks, and 29 weeks, while primary outcome time points are baseline, 13 weeks, 26 weeks, and 39 weeks. It does not explain how these withdrawal entries relate to each analysis occasion, how many observations entered the primary analysis, or how missingness was handled. No analysis-population definition or missing-data sensitivity analysis is reported in the registry record.

The registry lists 16 outcomes and 16 analyses but does not report a multiplicity strategy or complete statistical plan. It also does not report the original sample-size justification, a threshold for a clinically meaningful difference, or enough information about device adherence and valid monitoring periods to evaluate measurement completeness. The sample-size illustrations described in the record are planning exercises based on the observed effect, not evidence about the original design or observed power.

These limitations do not negate the sponsor-reported result, but they restrict independent verification and the strength of conclusions that can be drawn from the registry entry alone.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: SuPRA: Using Wearable Activity Trackers With a New Application to Improve Physical Activity in Knee Osteoarthritis — NCT02585323.
• Registry record last updated 2025-10-06; status last verified 2025-09.
• Results first posted 2025-10-06.
• 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

SuPRA was a randomized behavioral trial of Fitbit Flex and FitViz use, education, and physiotherapist counselling for people with knee osteoarthritis. The registry lists a crossover model, no masking, enrollment of 51, and completed status. The immediate group received the intervention in Months 1-3; the delayed group began in Month 4. The primary endpoint was daily minutes of bouted MVPA measured with a SenseWear Mini over a 7-day period, with assessments listed at baseline, 13 weeks, 26 weeks, and 39 weeks.

The sponsor-reported ANCOVA result was an adjusted difference in mean change of 13.1 minutes per day, with a 95% confidence interval from 1.6 to 24.5 and p = 0.05. Group contrast direction is not reported. The record notes withdrawals, but does not describe the primary analysis population or missing-data handling. It also describes the trial as cluster-randomized in an analysis note, despite listing a crossover model, and says a simple individual-level re-analysis would be misleading. Those design and reporting gaps limit independent interpretation.