Optimizing Function and Independence Through iHI-FIVES: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Optimizing Function and Independence Through iHI-FIVES (NCT03474380), a randomized na study. The primary outcome is Patient Number of Days Not at Home.

2. Clinical Question

This is an evaluation of a program implemented at sites, not simply a comparison of two individually assigned treatments. The record describes a stepped-wedge design in its brief summary, labels the model as crossover, and identifies the design as cluster-randomized in its analysis note. Those descriptions indicate that timing and site context matter when interpreting the comparison. The exact sequence of implementation, the assignment procedure, and the mapping of analysis groups to the intervention and usual-care periods are not reported in the registry record.

3. Trial Design

The registry record describes Optimizing Function and Independence Through iHI-FIVES as a randomized study, with no masking and a crossover model. Its brief summary says the program was implemented at 8 VAMC sites in a stepped-wedge design. The record also flags the trial as cluster-randomized for purposes of interpreting the analysis. These descriptions point to a design in which implementation timing at the site level is important; they do not describe a straightforward individually randomized, parallel-group comparison.

The intervention is behavioral: implementation of iHI-FIVES. The comparator is the pre-implementation period, described as usual care. The primary outcome is assessed for participants according to whether they were in usual care or intervention based on the 6-month time interval in which they were identified and/or enrolled at a site. This makes the time period and site part of the context for interpreting the contrast. The detailed schedule, number and order of rollout periods, and any rules for assigning participants to periods are not reported in the registry record.

The registered enrollment is 1406. That is the enrollment figure in the record; it should not be treated as the number contributing to the primary endpoint analysis, because endpoint-specific analysis counts are not reported in the registry record. The start date is 2018-04-23, the primary completion date is 2020-10-20, and the study status is completed. The record lists the phase as NA, which is consistent with the registry’s entry rather than a drug-development phase label.

The posted primary analysis uses a generalized linear mixed model and reports a rate ratio. Further information about the model’s link function, distribution, fixed and random effects, covariate adjustment, and treatment of time is not reported in the registry record. Nor does the record state how the usual-versus-enhanced implementation evaluation relates to the primary outcome analysis. Those omissions limit reconstruction of the design and analysis from the registry entry alone.

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.

Month 1-6

Month 7-12

Month 13-18

Month 19-24

Analysis population for the primary outcome, as defined in the registry: Veterans referred to home based care and determined to have a caregiver

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: 571 Veterans and 265 Caregivers in Intervention; 327 Veterans and 170 Caregivers in Usual Care Baseline Measures are described at the participant level by usual care and intervention, pre-specified in the study protocol. Participants were either in usual care or intervention based on the 6-month time interval that were identified and/or enrolled at a site. 73 Providers were enrolled for interviews. Baseline characteristics were not captured for these providers.

6. Primary Endpoint

The registry lists the primary outcome as Patient Number of Days Not at Home, measured over 180 days.

The primary endpoint is the patient number of days not at home during 180 days after study determination of eligibility. The description includes days in settings such as an emergency department, inpatient hospital, or post-acute facility. It is a count of days, assessed through administrative data rather than reported by the patient.

This measure captures time spent outside the home in the specified care settings over a defined follow-up period. In general, an endpoint based on days can reflect both whether such a stay occurs and how long it lasts. It may therefore provide a broader summary of use of these settings than an outcome that records only whether a person had any stay. That general interpretation does not establish which settings contributed most to the result in this trial; setting-specific results are not reported in the registry record.

The endpoint is described as a mean of a continuous measure, while the analysis reports a rate ratio from a generalized linear mixed model. The record does not explain the precise relationship between the summarized mean and the modeled rate, or provide the group-specific outcome values. It also does not report endpoint-specific denominators or the distribution of days not at home. These details would help readers understand the scale and variability of the observed data.

The outcome was defined at the participant level according to usual care or intervention based on the site’s 6-month interval. This timing is important because the program was rolled out in a site-based design. The endpoint description therefore identifies both the 180-day follow-up window and the implementation period used to classify participants.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a rate ratio represent?

A rate ratio compares modeled rates between two groups. A value of 1 corresponds to equal rates under the numerator and denominator ordering used in the analysis. Values below 1 indicate a lower rate in the numerator group, and values above 1 indicate a higher rate. This is a general explanation of the measure. For this result, the record names the analysis groups as OG000 and OG001 but does not state which is the numerator or denominator, so the estimate cannot be translated into an intervention-versus-usual-care direction from the registry entry alone.

The endpoint is days not at home over 180 days, and the analysis reports a rate ratio. Rate ratios are often used for outcomes expressed as counts or rates, but the registry does not describe the specific model distribution, link function, or offset, if any. It also does not explain how the model’s rate relates to the endpoint’s mean number of days. Those details should not be inferred from the label alone.

How should the confidence interval be read?

A confidence interval describes uncertainty around an estimate under the statistical model and its assumptions. In general, a 95% confidence interval is generated by a procedure that, under repeated applications in settings where the model assumptions hold, covers the underlying parameter at the stated frequency. It is not a range containing 95% of individual participants’ outcomes, and it is not a guarantee that the true effect lies within the particular interval.

Here, the estimate is 0.58 and the reported 95% confidence interval is 0.31 to 1.09. For a rate ratio, the value 1 is the reference for equal rates. Because the interval extends across 1, the result does not rule out equality at the level represented by this interval. The interval also allows estimates in either direction relative to equality. Its interpretation remains relative, and it does not provide an absolute difference in days.

Why use a generalized linear mixed model?

A generalized linear mixed model combines a generalized linear model framework with terms that can represent grouping or correlation in the data. In general, mixed models can be useful when observations are organized within clusters or repeated settings, because outcomes from the same cluster may be more alike than outcomes from different clusters. Accounting for that structure can affect uncertainty estimates and the estimated contrast.

The record says the trial is cluster-randomized and reports a generalized linear mixed model. That makes the warning against a simple individual-level re-analysis especially relevant: treating every participant as if independently assigned can ignore the design structure. However, the registry does not report which random effects or fixed effects were included, how site and time were represented, what distribution and link were used, or whether covariates were adjusted for. It is therefore not possible to confirm from the registry record how the model handled each design feature.

Over-dispersion is a general concern for count outcomes when observed variation exceeds what a simple count distribution assumes. If present and not adequately handled, it can make uncertainty estimates too small or otherwise misrepresent the data. The registry does not report a dispersion assessment or the model distribution, so over-dispersion cannot be evaluated from this entry. This is a general methodological consideration, not a claim that over-dispersion occurred in this study.

What does the p-value add?

A p-value measures how incompatible the observed data are with a specified null hypothesis, given the model and its assumptions. It is not the probability that the null hypothesis is true, nor is it a measure of the size or practical importance of an effect. The reported value is 0.091. The registry’s planning assumptions mention a two-sided significance level of 0.05, but the entry does not state whether that threshold was prespecified for the primary analysis or how multiple outcomes were handled.

The p-value and confidence interval give related but not identical ways to describe statistical uncertainty. Neither determines whether the observed rate ratio is important in practical terms. Absolute outcome summaries and a prespecified threshold for a meaningful difference are not reported in the registry record.

How do the planning illustrations relate to the trial result?

The registry describes sample-size illustrations as calculations based on the observed effect, with 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. These are planning illustrations, not evidence of the trial’s observed power and not a description of the trial’s original design justification. The registry does not report a sample-size justification for the trial itself.

In general, a planning calculation estimates the sample size needed under specified assumptions to detect a chosen effect with a target power. Such an illustration depends on its assumptions and should not be read as showing that the completed study had that power. The available record does not report the detailed values for the illustrations in the registry record, so no numerical planning result can be stated.

10. Analyses Reported by the Sponsor

The reported primary result is the sponsor’s generalized linear mixed model analysis: rate ratio 0.58, 95% confidence interval 0.31 to 1.09, and p = 0.091. The registry analysis note says that a simple individual-level re-analysis would be misleading for this cluster-randomized design and that only the sponsor’s own analysis is shown. That warning means an unadjusted calculation from posted summary statistics should not be treated as a competing estimate or as a correction to the sponsor’s result.

No recalculated primary effect is reported in the registry record. Therefore, there is no independent recalculation to compare numerically with the sponsor’s analysis. The sponsor’s own analysis takes precedence for describing the posted result because the design involves clustering and the record specifically cautions against a simple individual-level re-analysis.

Precedence does not mean that every modeling choice can be checked from the registry entry. The record does not report the model specification, the ordering of the analysis groups in the rate ratio, or a detailed analysis plan. Readers can describe the reported estimate and its interval, while recognizing that those missing details limit independent appraisal and interpretation of direction.

11. Multiplicity, Missing Data and Analysis Populations

The record lists four outcome measures and four analyses. The primary outcome is patient number of days not at home. Three secondary outcomes are caregiver satisfaction with the veteran’s VA care, caregiver subjective burden, and caregiver depressive symptoms. The results for those secondary outcomes are not included in the registry record described for the posted primary analysis.

When a study examines multiple outcomes or analyses, the chance of observing at least one apparently unusual result can increase. Statistical approaches can address multiplicity, for example by adjusting inference or by defining a hierarchy of outcomes. This is a general explanation; the registry record does not report whether any multiplicity procedure was used, whether the primary outcome was protected from other comparisons, or how the analyses were prioritized. The four analyses should not be assumed to represent four independent tests, and no adjustment should be inferred.

Missing-data methods are also not described. The primary outcome is assessed through administrative data, but the record does not say whether data were complete for all eligible participants, how unavailable records were handled, or whether any imputation or sensitivity analysis was conducted. Administrative ascertainment describes the data source; it does not establish that every outcome value was observed.

The analysis population is not reported in sufficient detail. The registered enrollment is 1406, but endpoint-specific numbers analyzed, exclusions, and the handling of participants with incomplete follow-up are not reported in the registry record. The record does not state whether analyses followed an intention-to-treat principle or another population definition. These omissions make it difficult to connect enrollment to the primary estimate or assess the potential effect of missing outcomes.

The record includes sample-size planning assumptions, but it does not report the trial’s own sample-size justification. The assumptions used for planning illustrations should not be taken as evidence of a prespecified multiplicity strategy, a missing-data plan, or the exact analysis 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 0% reporting threshold.

The adverse-event record lists an intervention group with 265 at risk and 0 serious events, 0 other events, and 0 deaths. It lists a usual-care group with 170 at risk and 0 serious events, 0 other events, and 0 deaths. These are the adverse-event figures recorded for those groups.

Zero recorded events means that no events of the listed categories were reported in the stated at-risk groups in this record. It does not establish that no health events occurred outside the reported categories, periods, or collection process. The registry does not describe the adverse-event collection methods, the relevant observation window, event definitions, or whether the groups’ at-risk figures represent the same analysis population as the primary endpoint.

These figures should also be read in the context of a behavioral program and a site-based implementation design. They are not evidence about the primary endpoint of days not at home, and they do not replace the separate efficacy analysis. The record lists no top serious-event terms. That absence should not be interpreted as a comprehensive statement about all possible harms or the completeness of event ascertainment.

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

The posted result gives an estimate of the relative rate for the primary endpoint under a generalized linear mixed model. The rate ratio is 0.58, with a 95% confidence interval from 0.31 to 1.09 and p = 0.091. The interval includes the reference value of 1, and the record does not identify which analysis group is the numerator. Consequently, the result does not establish from the registry entry alone whether the intervention group had a lower or higher modeled rate than usual care.

The result does not show an absolute difference in days not at home. No group-specific mean values or absolute rates are reported for the primary outcome, so the rate ratio cannot be translated into a number of days from this record. The interval communicates substantial uncertainty about the relative rate, and the reported p-value should not be treated as a measure of clinical importance or as the probability that the program has a particular effect.

The endpoint is based on administrative data and refers to time in several types of care settings over 180 days. It is not a direct measure of caregiver satisfaction, burden, or depressive symptoms. Those are listed as secondary outcomes, but their results are not reported in the registry record. Conclusions about those outcomes cannot be drawn from the primary endpoint result.

The trial’s site-based, stepped-wedge description and cluster-randomized analysis note matter. An analysis that ignores clustering could misrepresent the uncertainty, which is why the registry cautions against a simple individual-level re-analysis. At the same time, details needed to inspect the sponsor’s model and connect its analysis groups to intervention status are not reported in the registry record.

Overall, the posted statistics describe one modeled comparison with uncertainty, not a definitive statement that the program does or does not change days away from home. They also do not establish the result’s relevance to other populations or implementation settings. The registry result should be read within the defined population, endpoint, follow-up period, and design limitations recorded for this study.

14. Why This Trial Matters Statistically

This trial is statistically notable because the intervention was implemented in a site-based design rather than as a simple individually randomized comparison. The brief summary describes a stepped-wedge rollout across 8 VAMC sites, while the record also labels the model as crossover and flags the trial as cluster-randomized. Such designs require attention to clustering and timing. The registry’s caution against a simple individual-level re-analysis makes that point explicit.

The endpoint also illustrates the distinction between an outcome’s measurement scale and the model’s reported effect measure. Days not at home is a count over a fixed period, while the primary result is a rate ratio. A relative estimate can convey direction and scale only when its group ordering is known, and it does not automatically tell readers the absolute number of days associated with the comparison. The record does not supply that group ordering or absolute outcome summaries.

The confidence interval is important alongside the point estimate. The estimate is below 1, but its 95% confidence interval crosses 1, and it spans a range of relative rates. This is a useful example of why a single point estimate should not be read as a precise result and why statistical evidence should not be reduced to a yes-or-no reading of a p-value.

Finally, the record’s missing model details, analysis populations, and multiplicity information show how registry results can be informative while still leaving important questions open. Readers can identify the outcome, design context, and reported uncertainty, but cannot fully reconstruct the analysis or judge how secondary results were handled from the listed information alone.

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 record describes the study design in more than one way: the brief summary says stepped-wedge, the model is labeled crossover, and an analysis note calls the design cluster-randomized. These labels may describe related features, but the record does not explain how they fit together. The site rollout sequence and the timing structure are not reported in detail, limiting assessment of the design from the registry entry.

The reported analysis identifies groups as OG000 and OG001 but does not state which group forms the numerator or denominator of the rate ratio. This prevents a reliable assignment of the estimate’s direction to intervention versus usual care. The model specification is also not reported: the distribution, link function, random-effects structure, adjustment variables, and treatment of site and time are unavailable.

Outcome-specific sample sizes, group-specific days not at home, and absolute rates are not reported. The registered enrollment of 1406 does not establish how many participants contributed to the primary analysis. The record does not describe missing-data handling, analysis-population definitions, or sensitivity analyses. It also does not report whether the four listed analyses were adjusted for multiplicity.

The primary endpoint is described as days in several types of care settings, but setting-specific results are not reported. The record does not explain whether all types of days were weighted or counted in the same way beyond describing the count of days not at home. Details of eligibility, participant characteristics, and exposure to the program are not reported in the registry record.

Finally, the registry warns that simple individual-level re-analysis would be misleading, and no recalculated result is reported. This limits independent numerical verification. The posted model result can be described, but the available record does not support a complete assessment of its robustness or broader generalizability.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Optimizing Function and Independence Through iHI-FIVES — NCT03474380.
• Registry record last updated 2024-06-10; status last verified 2023-12.
• Results first posted 2024-06-10.
• 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

Optimizing Function and Independence Through iHI-FIVES is a completed, randomized, unmasked study of a behavioral caregiver skills training program. The record describes implementation at 8 VAMC sites in a stepped-wedge design and labels the model as crossover. The registered enrollment is 1406.

The primary endpoint is patient days not at home over 180 days among veterans referred to home based care and determined to have a caregiver. The outcome uses administrative data. The posted generalized linear mixed model result is a rate ratio of 0.58, with a 95% confidence interval of 0.31 to 1.09 and p = 0.091.

The registry does not report the ordering of the analysis groups in the rate ratio, group-specific endpoint values, model details, analysis populations, missing-data methods, or multiplicity procedures. Its analysis note cautions that a simple individual-level re-analysis would be misleading for the cluster-randomized design.