Optimizing Function and Independence Through STRIDE: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Optimizing Function and Independence Through STRIDE (NCT03300336), a randomized na study. The primary outcome is Proportion of Patients Discharged to Skilled Nursing Facility.

2. Clinical Question

STRIDE was designed to implement an inpatient mobility program and evaluate patient outcomes before and after implementation. Its primary outcome concerns discharge destination: discharge to a skilled nursing facility rather than home. This is an administrative and chart-based outcome, not a patient-reported measure. The posted analysis reports an odds ratio of 0.77, with a 95% confidence interval from 0.64 to 0.93 and p = 0.01. Those statistics describe the model’s comparison, but the registry does not connect the analysis group codes to the intervention and usual-care labels. The direction of the odds ratio therefore cannot be assigned confidently to a particular study arm from the reported information alone.

3. Trial Design

The registry describes STRIDE as a randomized, unmasked, crossover study using a stepped-wedge implementation design. Its summary says the program was implemented at 8 VAMC sites, with patient outcomes evaluated before and after implementation. The study also aimed to evaluate usual versus enhanced implementation, although the arm descriptions list implementation of STRIDE and usual care before implementation; further details about an enhanced implementation comparison are not reported in the registry record.

The listed enrollment is 13857. This is the registry’s enrollment figure, not necessarily the number included in the primary outcome analysis. The registry record does not report the analyzed sample size for that outcome in the information. It gives the study start as 2017-10-05, primary completion as 2020-03-29, and status as completed. The condition listed is Early Ambulation, and the intervention is categorized as behavioral.

In a stepped-wedge design, implementation is introduced across sites over time rather than being applied to all sites from the outset. In general, this structure means that comparisons can depend on both intervention status and the time period in which outcomes are observed. Because the record describes a cluster-randomized design, the assignment structure is at the site level, not simply at the level of individual patients. That distinction matters for analysis: observations within a site may be related, and a model should account for the design rather than treating every patient observation as independent.

The record says patients could be in both intervention and usual-care groups if they were hospitalized in the respective periods. For the primary outcome, it states that the first hospitalization for each patient during the study period was analyzed, but also describes possible presence in both groups. The exact handling of that feature, along with the full outcome population definition, is not reported in the registry record. Masking was listed as none; the record does not describe further steps to limit awareness of allocation.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Analysis population for the primary outcome, as defined in the registry: Discharge to skilled nursing facility was analyzed for the first hospitalization for each patient during the study period. Patients can be in both intervention and usual care groups if hospitalized during the respective intervention and usual care time periods; however, for this outcome patients will be analyzed in only one group, based upon the earliest of their hospitalizations if had more than one.

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: Patient-hospitalizations

6. Primary Endpoint

The registry lists the primary outcome as Proportion of Patients Discharged to Skilled Nursing Facility, measured over Assessed at hospital discharge, an average of 7 days.

The primary endpoint is the proportion of patients discharged to a skilled nursing facility rather than home. It is assessed at hospital discharge, described as an average of 7 days. The registry says the outcome is determined using administrative data pulls and chart review, not patient report.

Discharge destination is a consequential measure of where a patient receives care after leaving the hospital. In general, it can serve as an indicator of the level of support needed at discharge and may reflect both patient circumstances and decisions made during care. It is not, by itself, a direct measure of mobility, recovery, or quality of life. A difference in this endpoint would not establish why discharge destinations differed, nor would it describe outcomes beyond the discharge assessment.

The reported statistical result is an odds ratio from a generalized linear model. The registry gives the estimate as 0.77, with a 95% confidence interval from 0.64 to 0.93. However, the displayed analysis identifies groups by codes, and their mapping to intervention and usual care is not reported in the registry record. Accordingly, the odds ratio can be described as the posted model estimate, but its direction should not be attributed to a named arm without that mapping.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary result is reported as an odds ratio from a generalized linear model. In general, a generalized linear model relates an outcome to explanatory variables using a specified link function and outcome distribution. The registry record does not report the specific link function, distributional details, covariates, or other model specification. It is therefore possible to explain the reported measure, but not to reconstruct the full analysis from the registry description.

An odds ratio compares the odds of an outcome between two groups. Odds are not the same as probabilities or proportions. In general, an odds ratio below the null value means the numerator group has lower odds than the denominator group; an odds ratio above the null value means higher odds. The reported value is 0.77. Since the analysis lists group codes rather than a mapping to the arm labels, the registry record does not establish which group is the numerator. It is important not to turn the reported estimate into a claim that STRIDE had lower or higher odds than usual care without that information.

The reported 95% confidence interval is 0.64 to 0.93. In general, a confidence interval describes the uncertainty in an estimate under the assumptions of the analysis. It is not a guarantee that the true value lies in that interval, and it does not account automatically for sources of bias or model misspecification. The interval is reported alongside p = 0.01. A p-value describes how unusual data at least as incompatible with a specified null model would be if that null model and the analysis assumptions held. It is not the probability that the null hypothesis is true, nor does it express the probability that the result is due to chance.

The trial’s design makes the analysis structure particularly important. The registry characterizes the study as cluster-randomized and stepped-wedge. In cluster randomization, groups such as sites, rather than individual patients, are assigned to intervention conditions. In general, patient outcomes within the same site may be correlated because patients share care settings and practices. An analysis that assumes all patient observations are independent can understate uncertainty when that correlation is present. The stepped-wedge structure also means that intervention status changes across time, so time and implementation status can be entangled unless the analysis accounts for the design.

The record says the program was implemented at 8 VAMC sites and that outcomes were compared before and after implementation. It does not describe the sequence of implementation, the timing of each site’s transition, or how calendar time was represented in the model. Those details are not reported in the registry record. The registry explicitly notes that a simple individual-level re-analysis would be misleading for this cluster-randomized design; consequently, no individual-level recalculation is presented and the sponsor’s model result is the available analysis.

In general, a stepped-wedge design can use information from periods before and after implementation, while accounting for differences between sites and changes over time. The design alone does not guarantee that all time-related influences are controlled. The registry record does not report the model’s site effects, time terms, clustering adjustment, or any sensitivity analyses. Without those specifications, readers cannot determine exactly how the model addressed within-site dependence or period effects.

The outcome is a binary discharge destination: skilled nursing facility versus home. The registry reports the result as an odds ratio, not a difference in proportions or a risk ratio. In general, odds ratios can differ numerically from risk ratios, particularly when outcomes are common. The observed group proportions and analysis denominators are not reported in the result information, so an absolute difference cannot be inferred from the odds ratio alone. No conversion to a probability scale is justified from the reported result without additional information.

The record describes the primary outcome population as the first hospitalization for each patient during the study period and notes that a patient could be represented in both intervention and usual-care groups if hospitalized during the respective periods. The population description ends before its full detail is available. The registry record does not explain how repeated hospitalization information was handled, whether the stated first-hospitalization rule was applied separately within periods, or how patients present in both groups contributed to the model. These details matter because the unit of analysis and repeated observations affect the interpretation of uncertainty.

The registry also includes planning illustrations based on the observed effect, with a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation. The illustrations describe the effect to detect as a stated fraction of the observed effect; for time-to-event outcomes, their totals concern events, and for other outcomes they concern participants per arm. These are planning scenarios, not observed trial results and not observed power. The registry record does not report that these assumptions were the basis of the original trial design or sample-size justification.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Intervention", "Usual Care")
n   <- c(6141, 6722)
m   <- c(0.16, 0.2)
sp  <- c(0.02, 0.03)
sd  <- sp * sqrt(n)      # standard errors converted to SDs
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Intervention  6141  0.160  1.567  0.121  0.199
CARM  Usual Care  6722  0.200  2.460  0.141  0.259

11. Analyses Reported by the Sponsor

The primary result reported by the sponsor is the generalized linear model odds ratio of 0.77, with a 95% confidence interval from 0.64 to 0.93 and p = 0.01. The registry identifies the comparison groups only by codes in the analysis result; the mapping of those codes to intervention and usual care is not reported in the registry record. The estimate can therefore be described as the sponsor-reported comparison without assigning its direction to a named arm.

No simple individual-level recalculation is presented. The registry identifies the study as cluster-randomized and states that an individual-level re-analysis would be misleading. In general, such a recalculation could fail to account for clustering and the stepped-wedge timing, producing an inappropriate estimate of uncertainty. The sponsor’s model result is therefore the posted analysis for interpretation, rather than a competing unadjusted calculation.

The record does not include enough model detail to independently reproduce the sponsor’s analysis. The link function, adjustment variables, treatment of time, clustering method, analyzed denominators, and group-code mapping are not reported in the registry record. This lack of detail limits independent verification; it does not justify replacing the sponsor result with a calculation that ignores the design.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists 13 outcome measures and 13 analyses. It identifies the skilled-nursing-facility discharge proportion as the primary outcome and lists secondary outcomes concerning physical function, disability, limitations, satisfaction with care, health utility, health-related quality of life, and mobility in the environment. The registry record does not report whether the analyses were organized into a prespecified hierarchy or whether any adjustment for multiple comparisons was planned or applied.

In general, testing multiple outcomes can increase the chance of finding at least one apparently notable result even when no true effects exist. Multiplicity procedures can address that concern, but the registry does not describe such a procedure. The reported p-value for the primary outcome should therefore be read as the result of the reported analysis, not as evidence that multiplicity was or was not controlled.

The registry gives an enrollment figure of 13857, but the analyzed denominator for the primary outcome is not reported in the result information. The primary population description says the first hospitalization for each patient during the study period was analyzed and notes that patients could be in both groups in the respective periods. The full population description is incomplete. The record does not state how missing discharge information was handled, how many observations were excluded, or whether all enrolled participants contributed to the primary analysis.

In general, missing-data methods matter when outcome availability is related to patient characteristics, care setting, or outcome status. The registry record does not report the amount or pattern of missing outcome data, the assumptions used for any missingness, or any imputation or sensitivity analysis. It also does not give a full analysis-population definition, such as whether the analysis followed an intention-to-treat principle at the site level. These omissions limit assessment of whether the analyzed population reflects the randomized design and how robust the result may be to missing data.

13. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The registry lists serious adverse events as affecting 0 participants in the Intervention group and 0 in the Usual Care group. It also lists 0 other affected participants and 0 deaths for each group. The listed at-risk counts are 7134 for Intervention and 6723 for Usual Care. No serious adverse event terms are listed.

These are registry-recorded counts for the reported categories. They do not, by themselves, establish that no adverse events occurred in any broader sense. The record’s presentation does not provide definitions, ascertainment procedures, follow-up details, or a fuller account of how events were collected and classified. The meaning of zero affected participants depends on those processes and on the completeness of reporting.

The at-risk counts should not be treated as the primary outcome denominators: the registry does not state that they are the analyzed populations for discharge destination. Nor do these safety figures explain the primary outcome result. They are a separate part of the record and should be interpreted within the limits of the adverse-event reporting information available.

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

The posted result establishes that the registry reports a model-based odds ratio of 0.77 for discharge to a skilled nursing facility rather than home, with a 95% confidence interval from 0.64 to 0.93 and p = 0.01. It describes a comparison made by the sponsor’s generalized linear model in a cluster-randomized stepped-wedge study. The interval and p-value characterize statistical uncertainty under the model and its assumptions.

The result does not establish which named arm had lower odds, because the registry’s group codes are not mapped to intervention and usual care in the reported analysis. It also does not quantify an absolute difference in discharge proportions. An odds ratio is a relative measure of odds, and the registry does not provide the group-specific proportions or analyzed denominators needed to translate the estimate into an absolute scale.

The endpoint is discharge destination, not a direct measure of mobility, function, or quality of life. A difference in discharge destination would not, on its own, show that the intervention caused a change in mobility or explain the pathway to that destination. The registry record does not report enough detail about model adjustment, missing data, or the handling of time and site clustering to assess every assumption behind the estimate.

The result should not be extended to outcomes beyond the primary endpoint or to settings beyond the study described. The listed secondary outcomes cover other patient-reported and functional domains, but their detailed results are not included in the primary result information summarized in the registry. The study’s completed status and enrollment figure do not substitute for the analyzed sample size or complete outcome reporting.

In general, statistical evidence for a modeled comparison and evidence about its practical importance are distinct. The odds ratio and its confidence interval convey the reported statistical result, while clinical interpretation also depends on the group mapping, underlying outcome proportions, study context, and analysis details. Several of those details are not reported in the registry record, so conclusions should remain limited to the posted model result and its stated endpoint.

15. Why This Trial Matters Statistically

This trial is statistically informative because its design links randomization to sites and implementation timing rather than assigning every patient independently. That structure is central to interpreting the primary result. When patients share sites, their outcomes may be related; when implementation occurs over time, calendar periods may also shape comparisons. These are general features of cluster-randomized stepped-wedge studies, and they explain why an unadjusted patient-level recalculation would not be an adequate substitute for the sponsor’s analysis.

The trial also illustrates why the effect measure and outcome definition matter. Discharge to a skilled nursing facility is an administrative and chart-based endpoint, whereas STRIDE is a mobility program and the registry lists additional functional and patient-reported outcomes. The primary odds ratio cannot be treated as a direct estimate of mobility change or as an absolute difference in discharge probabilities.

Finally, statistical interpretation depends on reporting completeness. The registry provides an estimate, confidence interval, p-value, design description, and enrollment figure, but not the group-code mapping, full model specification, analyzed denominator, or missing-data approach. Those details would help readers understand the direction and robustness of the result. The case underscores that a numerical estimate can be precisely reported while important elements needed for independent interpretation remain unavailable.

16. 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 trial-specific limitations follow from the information in the registry record. The analysis groups are identified by codes, but their mapping to intervention and usual care is not reported. This prevents a confident arm-specific interpretation of the odds ratio, even though the estimate, confidence interval, and p-value are stated.

The study is described as cluster-randomized, crossover, and stepped-wedge, but the record does not give the sequence or timing of implementation across sites, or the full statistical model. It does not specify how clustering, site differences, calendar time, or any repeated patient observations were handled. These details are especially relevant because the primary population description says a patient could appear in both intervention and usual-care groups in different periods.

The record’s primary population description is incomplete. Although it states that the first hospitalization for each patient during the study period was analyzed, it then notes possible inclusion in both groups and ends before fully describing how the outcome population was defined. The analyzed denominator and the handling of missing discharge information are not reported in the primary result information.

The enrollment figure of 13857 should not be mistaken for the primary outcome analysis size. The registry also lists secondary outcomes and analyses but does not report whether multiplicity was addressed. Lastly, while serious adverse event counts are listed as zero affected participants in both groups, the event definitions and ascertainment details are not reported. These reporting gaps limit independent evaluation and the scope of conclusions that can be drawn from the posted result.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Optimizing Function and Independence Through STRIDE — NCT03300336.
• Registry record last updated 2023-09-28; status last verified 2023-09.
• Results first posted 2021-07-19.
• 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.

20. Record Summary

NCT03300336, titled “Optimizing Function and Independence Through STRIDE,” is listed as a completed randomized, unmasked crossover study. The registry describes a stepped-wedge implementation of the STRIDE inpatient mobility program at 8 VAMC sites. Usual care is described as pre-implementation care, and the listed enrollment is 13857.

The primary outcome is discharge to a skilled nursing facility rather than home, assessed at hospital discharge using administrative data and chart review. The sponsor-reported generalized linear model result is an odds ratio of 0.77, with a 95% confidence interval from 0.64 to 0.93 and p = 0.01. The mapping of analysis group codes to the named arms is not reported in the registry record, so the estimate’s direction cannot be assigned confidently to STRIDE or usual care.

The record also lists secondary functional and patient-reported outcomes. Important analysis details, including the full model specification, analyzed denominator, missing-data handling, and group-code mapping, are not reported in the registry record.