This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Reducing Emergency Diabetes Care for Older African Americans (NCT03466866), a randomized phase 3 study. The primary outcome is Number of Incident Diabetes-related ED Visits and/or Hospitalizations.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03466866 |
| Phase | Phase 3 |
| Condition | Diabetes Mellitus |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Enrollment | 156 |
| Arms in the primary outcome | 2 |
| Lead sponsor | Thomas Jefferson University |
| Results posted | Yes |
2. Clinical Question
Population
The trial concerns African Americans aged 50 years and older with diabetes after an emergency department visit. The primary analysis is described as including all randomized participants. The registry reports enrollment of 156, while the brief summary describes a planned study population of 230; those figures describe different aspects of the record and should not be treated as interchangeable.
Intervention
PREVENT combines in-home diabetes education with diabetes-specific Behavioral Activation and facilitated telehealth visits involving primary care clinicians and a diabetes nurse educator. Community Health Workers deliver the intervention, and the arm description states that they are race-concordant with participants.
Comparator
Enhanced Usual Care (EUC) is intensive home-based diabetes education. The brief summary says it matches PREVENT in the intensity of home visits and diabetes self-care education, while the arm description specifies that it does not include goal setting or telehealth visits.
Primary question
Does assignment to PREVENT, compared with EUC, change the rate of diabetes-related emergency department visits and/or hospitalizations during the 12 months after the index emergency department visit? Each qualifying visit or hospitalization is counted as an event; an emergency visit that leads to hospitalization is counted once.
3. Trial Design
The registry describes a Phase 3, randomized, double-masked, parallel-group trial comparing PREVENT with Enhanced Usual Care. Both interventions are behavioral. The record lists 156 participants enrolled, with 78 in the PREVENT arm and 78 in the EUC arm. The brief summary describes a planned population of 230, but the registry enrollment figure and the brief-summary target are not the same measure. The registry record does not explain the difference.
The trial began on 2019-01-10, had a primary completion date of 2023-08-31, and is listed as completed. These dates describe the trial record; they do not establish when every participant completed follow-up. The primary outcome timeframe is 12 months after the index emergency department visit.
PREVENT extends from the emergency department into the community. It combines home diabetes education, Behavioral Activation intended to reinforce self-care, and telehealth visits facilitated with primary care clinicians and a diabetes nurse educator. EUC provides home-based diabetes education. The brief summary describes EUC as matching PREVENT in the number and duration of home visits and in diabetes self-care education, while omitting PREVENT's other active elements. The arm description specifically notes no goal setting or telehealth visits in EUC.
The primary comparison is reported as a Poisson regression analysis of the event rate, expressed as an incidence rate ratio. The registry identifies all randomized participants as the primary-analysis population. The detailed randomization procedures, implementation of masking, allocation concealment, and any deviations from assigned treatment are not reported in the registry record. The record also does not describe how the analysis accounted for treatment adherence or crossover, if either occurred.
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
| Stage | PREVENT (Preventing and Reducing Emergency Visits in Diabetes Through Education and Telehealth) | EUC (Enhanced Usual Care) |
|---|---|---|
| STARTED | 78 | 78 |
| COMPLETED | 62 | 62 |
| NOT COMPLETED | 16 | 16 |
Analysis population for the primary outcome, as defined in the registry: All randomized participants
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.
| Characteristic | Unit | PREVENT (Preventing and Reducing Emergency Visits in Diabetes Through Education and Telehealth) | EUC (Enhanced Usual Care) | Total |
|---|---|---|---|---|
| Participants analysed | 78 | 78 | 156 | |
| Age, Continuous | years | 56.4 (SD 9.5) | 57.2 (SD 9.5) | 56.8 (SD 9.5) |
| Sex: Female, Male: Female | Participants | 48 | 50 | 98 |
| Sex: Female, Male: Male | Participants | 30 | 28 | 58 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 1 | 2 | 3 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 77 | 76 | 153 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Asian | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Black or African American | Participants | 78 | 78 | 156 |
| Race (NIH/OMB): White | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): More than one race | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 0 | 0 | 0 |
| Region of Enrollment: United States | participants | 78 | 78 | 156 |
| Stratification variable: Baseline HbA1c < 9.0% vs. > or equal to 9.0%: Baseline HbA1c < 9.0% | Participants | 40 | 42 | 82 |
| Stratification variable: Baseline HbA1c < 9.0% vs. > or equal to 9.0%: Baseline HbA1c > or equal to 9.0% | Participants | 38 | 36 | 74 |
| Stratification variable: Primary care physician at Jefferson: Jefferson primary care physician | Participants | 40 | 40 | 80 |
| Stratification variable: Primary care physician at Jefferson: Non-Jefferson primary care physician | Participants | 38 | 38 | 76 |
| Stratification variable: Discharge status: Discharged home from index visit | Participants | 38 | 39 | 77 |
| Stratification variable: Discharge status: Admitted from index visit | Participants | 40 | 39 | 79 |
6. Primary Endpoint
The registry lists the primary outcome as Number of Incident Diabetes-related ED Visits and/or Hospitalizations, measured over 12 months.
| Element | Registry entry |
|---|---|
| Result type | Mean |
| Measure of spread | 95% Confidence Interval |
| Unit | Incidence rate ratio |
| Endpoint classification | Continuous Endpoint |
The primary endpoint is the number of incident diabetes-related emergency department visits and/or hospitalizations during the 12 months after the index emergency department visit. The registry treats each qualifying emergency visit or hospitalization as an event. If an emergency department visit results in a hospitalization, that episode is counted once rather than as two events. Events are ascertained through chart reviews and participant self-report.
This count-based endpoint captures repeated acute-care events, rather than only whether a participant had any event. An event-rate analysis can therefore use information about the frequency of events across the follow-up period. The registry reports the comparison as an incidence rate ratio, which expresses the event rate in the PREVENT arm relative to the rate in EUC. It is a relative measure of rates, not a direct statement about the number of people who experienced an event.
Such an endpoint is relevant to the trial's stated aim of reducing diabetes-related emergency care. It also has important interpretive limits: the count does not, by itself, distinguish between participants with no events and those with repeated events, nor does it describe the severity or circumstances of each event. The record does not provide event-count distributions by participant or detail the specific rules used to classify an event as diabetes-related.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry reports an incidence rate ratio of .67 for PREVENT compared with EUC from a Poisson regression analysis. The 95% confidence interval is .42 to 1.07, and the reported p-value is .12. The point estimate is below the no-difference value for a rate ratio, so its direction favors a lower event rate in PREVENT. However, the confidence interval includes values on both sides of the no-difference point. The estimate is therefore imprecise enough to be compatible with a range of effects, including little or no difference and a rate higher in PREVENT.
The p-value of .12 does not meet the conventional .05 threshold. It should not be read as proof that the interventions are equivalent, or as proof that there is no difference. A non-significant result can arise when the estimated effect is uncertain; the interval is important because it shows that uncertainty. Conversely, the point estimate alone does not establish that PREVENT reduces events. The registry result does not support a definitive claim about the direction or size of the underlying effect.
Descriptive arm results are also reported: the PREVENT arm has a value of 0.88 with a 95% confidence interval of 0.62 to 1.27, and the EUC arm has a value of 1.29 with a 95% confidence interval of 0.93 to 1.79. These arm-level estimates provide context for the comparative model, but they do not replace the regression estimate. The registry identifies the arm values in the context of an incidence rate ratio outcome; it does not provide further detail about adjustments that may distinguish the model-based comparison from the arm-level summaries.
The primary outcome is based on event counts, so repeated events can contribute to the result. The rate ratio does not reveal whether the observed pattern was broadly shared across participants or driven by a smaller number with repeated acute-care episodes. Participant-level event distributions are not reported in the registry record.
The registry's brief summary describes the study as testing whether PREVENT can reduce emergency visits and hospitalizations, but the posted primary result is an estimate with appreciable uncertainty. Statistical interpretation should stay with that estimate and its interval rather than turn the trial's aim into a conclusion. The result does not establish that the two approaches have the same effect, nor does it establish a clinically important benefit for either approach. The registry record does not report a threshold for a clinically important difference.
9. Statistical Methods Explained
The primary outcome is an event count, and the registry reports Poisson regression as the comparative method. In general, Poisson regression models counts in relation to exposure time or another specified scale and can estimate a relative event rate. The reported incidence rate ratio compares the modeled rate for PREVENT with that for EUC. A rate ratio below the no-difference point indicates a lower estimated rate in the first arm; a rate ratio above it indicates a higher estimated rate. This general interpretation does not, on its own, establish causation or clinical importance.
What does the confidence interval communicate?
A confidence interval describes the precision of an estimate under the assumptions of the analysis. A narrower interval indicates greater precision, while a wider interval indicates more uncertainty about the effect size. The registry's 95% confidence interval for the comparison is .42 to 1.07. It spans the no-difference point, so the data do not sharply distinguish a lower rate from no difference or a higher rate. The interval should be read alongside the point estimate, not as a range of effects that are equally likely or as a guarantee about the true effect.
How should the p-value be read?
The reported p-value is .12. In general, a p-value measures how unusual data at least as incompatible with a specified null model would be if that model and the test assumptions held. It is not the probability that the null hypothesis is true, and it does not quantify the probability that the result occurred by chance. The record does not report the full testing procedure, including whether the test was one-sided or two-sided, beyond the reported result. The p-value should therefore be understood as part of the sponsor-reported analysis, not as a complete description of the modeling process.
Why can event-count models need special attention?
Poisson models commonly rely on a relationship between the expected count and its variability. In general, clinical event counts may be more variable than a basic Poisson model assumes, a feature called over-dispersion. If present and not appropriately handled, it can affect uncertainty estimates and tests. The registry record identifies Poisson regression but does not report whether over-dispersion was assessed or whether the analysis used a correction, a robust variance method, or another approach. No conclusion about the adequacy of that aspect of the model can be drawn from the method label alone.
Repeated events introduce another general consideration: events from the same participant may be related rather than statistically independent. Methods need to account appropriately for the data structure if that dependence affects uncertainty. The record describes an event-count endpoint but does not report the detailed model specification or how within-participant dependence was handled. It also does not give the participant-level distribution of event counts.
What does the analysis population mean?
The registry describes the primary efficacy analysis as including all randomized participants. This identifies the stated population for the primary analysis, but the record does not explain how participants with incomplete follow-up, missing outcome information, or other post-randomization circumstances were handled. The phrase alone does not establish whether every randomized participant contributed complete outcome data or how missingness was addressed.
General statistical principles help explain why these details matter. A model's estimate depends on how the outcome, follow-up, and exposure are defined, and its uncertainty depends on the model assumptions and variance calculation. The registry gives the endpoint definition, the follow-up timeframe, and the named regression method, but not a full statistical analysis plan. Details such as covariate adjustment, offsets, model diagnostics, treatment of incomplete observation time, and sensitivity analyses are not reported in the registry record. They should not be inferred from the reported incidence rate ratio.
The registry also lists four outcome measures and four analyses, and names three secondary outcomes: perceived access to health care, actual access to care, and change from baseline to 12 months on Diabetes Self-Care Inventory scores. The record does not describe how the secondary outcomes were analyzed or how the primary and secondary testing was organized. The analysis labels and outcome list do not by themselves show whether any adjustment for multiple testing was planned or used.
10. Analyses Reported by the Sponsor
| Method | Estimate | 95% CI | P-value |
|---|---|---|---|
| Poisson regression | Incidence rate ratio .67 | .42 to 1.07 | 0.120 |
The registry reports the primary comparison from a Poisson regression analysis: an incidence rate ratio of .67, a 95% confidence interval of .42 to 1.07, and p = .12. This is the reported inferential result to use when describing the trial's primary comparison. The arm-level values are descriptive summaries and should not be treated as a substitute for that regression analysis.
No separate recalculated comparison is reported in the registry record, so there is no independent recalculated estimate to set against the regression result. The reported record does not identify a discrepancy between an analysis and a recalculation. The sponsor's model result and its uncertainty should therefore be presented as reported, without implying that an unreported re-analysis confirms or contradicts it.
The numerical result is consistent with a point estimate below the no-difference value, but the confidence interval includes values on either side of that point. The p-value is .12. Together, these features call for a cautious description: the registry reports an estimated lower event rate with uncertainty that does not establish a clear difference. Neither the point estimate nor the p-value alone supports a definitive statement about efficacy.
The record does not provide the full model specification, diagnostics, or a detailed account of adjustments. As a result, the sponsor-reported result can be described accurately, but the analysis cannot be independently reproduced from the registry information alone.
11. Multiplicity, Missing Data and Analysis Populations
The registry lists the primary outcome and three secondary outcomes: perceived access to health care, actual access to care, and change from baseline to 12 months on Diabetes Self-Care Inventory scores. It also reports four outcome measures and four analyses. The record does not specify the complete set of statistical tests, their order, or whether the secondary outcomes were confirmatory or exploratory.
When several outcomes or hypotheses are tested, the chance of at least one apparently notable result can increase. General approaches to multiplicity include adjusting the testing threshold, arranging tests in a prespecified hierarchy, or treating secondary analyses as exploratory. Which approach was planned or used for this trial is not reported in the registry record. The primary p-value of .12 should be interpreted as posted; there is not enough information to reconstruct a multiplicity procedure or to determine whether that value was adjusted.
The registry identifies all randomized participants as the primary-analysis population, but does not describe missing outcome data or incomplete follow-up. The missing-data strategy, the amount and pattern of missingness, and any sensitivity analyses are not reported in the registry record. This matters because an event count over a follow-up period can be affected by how time without observed data is handled. No assumptions about those procedures should be made from the stated population alone.
The record also does not report whether the analysis was conducted according to randomized assignment in every case, how treatment discontinuation or crossover was handled, or whether participants were excluded after randomization. Those details are relevant to interpreting randomized comparisons. The stated analysis population is informative, but is not a substitute for a description of participant flow and analysis rules.
12. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| PREVENT (Preventing and Reducing Emergency Visits in Diabetes Through Education and Telehealth) | 36 | 78 | 46.2% |
| EUC (Enhanced Usual Care) | 39 | 78 | 50.0% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| PREVENT (Preventing and Reducing Emergency Visits in Diabetes Through Education and Telehealth) | 0 of 78 |
| EUC (Enhanced Usual Care) | 1 of 78 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| PREVENT (Preventing and Reducing Emergency Visits in Diabetes Through Education and Telehealth) | 0 of 78 |
| EUC (Enhanced Usual Care) | 0 of 78 |
Most frequent serious adverse event terms
| Event term | Organ system | PREVENT (Preventing and Reducing Emergency Visits in Diabetes Through Education and Telehealth) | EUC (Enhanced Usual Care) |
|---|---|---|---|
| Cardiac | Cardiac disorders | 9 of 78 | 16 of 78 |
| Infection | Infections and infestations | 10 of 78 | 9 of 78 |
| Gastrointestinal | Gastrointestinal disorders | 5 of 78 | 10 of 78 |
| musculoskeletal | Musculoskeletal and connective tissue disorders | 9 of 78 | 5 of 78 |
| Neurological (excluding stroke) | Nervous system disorders | 6 of 78 | 6 of 78 |
| Diabetes-related | Endocrine disorders | 5 of 78 | 6 of 78 |
| Pulmonary | Respiratory, thoracic and mediastinal disorders | 2 of 78 | 6 of 78 |
| shortness of breath | Respiratory, thoracic and mediastinal disorders | 4 of 78 | 3 of 78 |
| Blood disorders | Blood and lymphatic system disorders | 2 of 78 | 3 of 78 |
| renal | Renal and urinary disorders | 2 of 78 | 3 of 78 |
Serious events with the highest arm-level rate: cardiac (20.5%), gastrointestinal (12.8%), infection (12.8%), musculoskeletal (11.5%), diabetes-related (7.7%).
The registry reports serious adverse events for 36 of 78 participants in PREVENT and 39 of 78 in EUC. It reports zero other adverse events affected in each arm and deaths of 0 in PREVENT and 1 in EUC. These are safety counts as recorded in the registry; they are not estimates of the causal effect of either intervention.
The record lists the most common serious-event terms as Cardiac (20.5%), Gastrointestinal (12.8%), Infection (12.8%), musculoskeletal (11.5%), Diabetes-related (7.7%), and Neurological (excluding stroke) (7.7%). These categories describe reported serious-event terms. The registry information does not establish that an event was caused by an intervention, and the percentages should not be interpreted as comparative risks without knowing their exact denominators and coding rules.
Serious adverse event counts are not directly comparable to the primary endpoint's count of diabetes-related emergency visits and hospitalizations. The endpoints have different definitions, and the registry does not explain the overlap between reported serious events and primary-outcome events. Details about seriousness criteria, event timing, attribution, and how multiple events for the same participant were counted are not reported in the registry record. The safety figures therefore provide a limited description rather than a complete comparative safety assessment.
13. What the Result Does, and Does Not, Mean
The primary result is an estimated incidence rate ratio of .67 for PREVENT compared with EUC, with a 95% confidence interval of .42 to 1.07 and p = .12. The estimate points toward a lower event rate in PREVENT, but the interval includes the no-difference value and also permits an estimate in the opposite direction. The result is uncertain; it does not establish a clear between-arm difference.
This finding does not prove that PREVENT and EUC are equivalent. Equivalence requires a defined margin and an analysis designed to assess whether the difference lies within that margin. The registry record does not report an equivalence margin or an equivalence analysis. Likewise, a non-significant p-value does not demonstrate that an intervention has no effect.
The result also does not establish a clinically meaningful benefit. The registry does not report a threshold for clinical importance, nor does the rate ratio alone express the absolute number of events avoided or the number of participants experiencing any event. The arm-level values and the model-based comparison describe event rates, not a complete account of individual experience.
The study was randomized, which is an important design feature for estimating treatment comparisons, but interpretation still depends on implementation, outcome ascertainment, missing-data handling, and analysis assumptions. The registry describes chart review and participant self-report for event ascertainment, but does not report further validation or adjudication details. It also does not provide the full statistical plan or participant-level event distribution.
The appropriate reading is therefore bounded: the posted analysis estimates a lower rate for PREVENT, but the precision and reported test do not establish that the interventions differ. The result neither confirms the trial's stated hypothesis nor shows that the approaches have the same effect. It should be interpreted for the specified population, interventions, outcome, and follow-up period rather than generalized beyond those details.
14. Why This Trial Matters Statistically
This trial poses a statistically useful question because its primary outcome counts repeated acute-care events rather than recording only whether any event occurred. That structure can preserve information about recurrence, but it also means that interpretation depends on the distribution of events across participants and on appropriate modeling of repeated counts. The registry reports an incidence rate ratio from Poisson regression, making the model's assumptions and handling of within-participant dependence important to understand.
The active comparator is also relevant to interpretation. EUC is described as intensive home-based diabetes education, with PREVENT adding Behavioral Activation and telehealth elements. The comparison is therefore framed as the added contribution of those elements beyond education, rather than a comparison with no intervention. The registry's brief summary says the arms were matched in home-visit intensity and diabetes self-care education.
The estimated rate ratio of .67 is potentially compatible with a lower rate in PREVENT, but its 95% confidence interval of .42 to 1.07 includes the no-difference point. The reported p-value of .12 underscores that the result is not statistically conclusive under conventional testing. This combination illustrates why estimates and intervals are more informative than a binary reading of a p-value.
Finally, the record's description of a target population of 230 alongside enrollment of 156 makes clear why planned and enrolled numbers should be distinguished. The registry does not explain the difference, and no inference about the reason is warranted. Overall, the trial is informative as a randomized comparison with a clearly defined event-based primary question, while incomplete reporting of analysis details limits independent assessment of precision and assumptions.
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 trial-specific details needed to assess the result are not reported in the registry record. The brief summary describes a planned population of 230, whereas the enrollment field reports 156 and the arms each list 78. The record does not explain this difference or provide a sample-size justification. It is therefore not possible to determine from the registry entry whether the planned target changed or why enrollment ended at the reported figure.
The primary method is identified as Poisson regression, but the model specification is not given. Covariate adjustment, exposure-time handling, over-dispersion assessment, variance estimation, and treatment of within-participant dependence are not reported. These details matter for an event-count outcome and for the confidence interval and p-value. No independent assessment of model adequacy can be made from the method name alone.
The registry says events were ascertained through chart reviews and participant self-report, but does not report the operational criteria for classifying events as diabetes-related, the completeness of records, or how disagreements were resolved. The record also does not provide participant-level event counts, follow-up completeness, missing-data patterns, or sensitivity analyses. These omissions limit assessment of how robust the event-rate comparison is to ascertainment and follow-up.
Although the trial is described as double-masked, the registry record does not explain who was masked or how masking was maintained for behavioral interventions. The allocation procedure, adherence, treatment receipt, and any crossover or discontinuation are also not described in detail. The record does not report adjustment for multiple outcomes or the prespecified role of secondary analyses. These limitations concern reporting available in the registry entry; they do not establish that the procedures were absent from the trial.
16. Related Tutorials
17. Related Statistical Calculators
18. Sources
- ClinicalTrials.gov: Reducing Emergency Diabetes Care for Older African Americans — NCT03466866.
- Registry record last updated 2024-05-14; status last verified 2024-04.
- Results first posted 2024-05-14.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
19. Record Summary
NCT03466866 is a completed Phase 3, randomized, double-masked, parallel-group trial sponsored by Thomas Jefferson University. It compares PREVENT, a behavioral intervention combining home diabetes education, Behavioral Activation, and facilitated telehealth, with Enhanced Usual Care, described as home-based diabetes education.
The registry reports enrollment of 156, with 78 participants in each arm. The primary endpoint is the number of diabetes-related emergency department visits and/or hospitalizations over 12 months after the index emergency department visit. The primary comparison, analyzed with Poisson regression, is an incidence rate ratio of .67 (95% confidence interval .42 to 1.07; p = .12).
The estimate favors a lower event rate for PREVENT, but the interval includes the no-difference point and the result is not statistically conclusive. The registry lists three secondary outcomes. Detailed information about multiplicity, missing data, model specification, and participant-level event patterns is not reported in the registry record.