CV Wizard: Does a Clinical Decision Support Tool Improve CVD Risk Factor Control in Safety Net Clinics?: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for CV Wizard: Does a Clinical Decision Support Tool Improve CVD Risk Factor Control in Safety Net Clinics? (NCT03001713), a randomized na study. The primary outcome is 1-year Change in 10-year CVD Risk.

2. Clinical Question

The trial asks whether introducing CV Wizard, a prioritized, point-of-care clinical decision support system, is associated with a change in estimated 10-year cardiovascular disease risk in safety net community health centers. The project’s broader aims include improving CVD risk factor control and addressing disparities in care among low-income, racially and ethnically diverse patients receiving primary care at these centers. The primary registry outcome focuses on estimated CVD risk rather than directly measuring heart attacks or strokes.

3. Trial Design

The registry describes a randomized, parallel-group trial with no masking. The intervention is behavioral: implementation of the CV Wizard clinical decision support system. Centers, rather than individual patients, are described as the units randomized. The immediate-implementation arm included 30 safety net community health centers, with implementation planned at the start of study year 2. The delayed-implementation arm also included 30 centers, with implementation planned 18 months later. Both groups therefore had access to the intervention under different timing arrangements; the comparator is delayed implementation, not a permanently untreated group.

The registry lists enrollment as 70. The outcome summaries, however, report 10980.0 for the immediate-implementation group and 7247.0 for the delayed-implementation group. The record does not explain how these figures relate to the enrollment entry, what each outcome-group count represents, or whether the counts refer to patients, records, or another analytic unit. That discrepancy in scale should be kept in view when interpreting the summaries, without assuming that one entry is erroneous.

The trial is marked completed. Its listed start date is 2018-02-22, and primary completion is recorded as 2020-03-31. The study description says the project intended to evaluate CVD care, disparities in care and outcomes, and patient engagement in treatment choices. Those aims provide context, but the registry posts only the primary outcome summary and indicates no posted secondary outcomes. The protocol’s detailed allocation procedures, implementation fidelity measures, and analytic handling of the center-level design are not reported in the registry record.

4. Participant Flow and Accounting

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

Overall Study

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 1-year Change in 10-year CVD Risk, measured over 12 months.

The primary endpoint is titled “1-year Change in 10-year CVD Risk,” with a timeframe of 12 months. It is a continuous measure reported as a mean, with a 95% confidence interval. The risk estimate is calculated using the American College of Cardiology/American Heart Association pooled risk equations. The listed inputs are age, race and ethnicity, sex, systolic blood pressure, total cholesterol, high-density lipoprotein cholesterol, and diabetes, smoking, and antihypertensive medication status.

A risk score combines information about several cardiovascular risk factors into an estimate of future CVD risk. In general, using a continuous score can retain more gradation than reducing an outcome to whether a participant did or did not experience an event. It can therefore describe changes in a modeled risk estimate over the specified follow-up period. It is not itself a count of heart attacks or strokes, and the registry outcome summary does not report those clinical events as the primary endpoint.

The posted group means are 0.4 for immediate implementation and 0.2 for delayed implementation. The record gives the unit as percent risk of CVD, but does not explain in the results description how to interpret the scale of the reported change, including its direction. The estimates should therefore be read as the registry’s reported group summaries, not as a direct estimate of the intervention’s causal effect.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous and summarized as a mean with a 95% confidence interval. In general, a mean describes the arithmetic average of measurements in an analyzed group. A confidence interval expresses sampling uncertainty around an estimate, conditional on the statistical model and data structure. Neither the model used to construct these intervals nor the method used to account for assignment by center is reported in the registry record.

What does a group confidence interval tell us?

In general, a confidence interval gives a range of values compatible with the observed data under a specified procedure and its assumptions. Its width reflects precision: a narrower interval indicates less sampling uncertainty under that procedure, while a wider interval indicates more. The intervals posted for this trial apply to the group means, not directly to the difference between the immediate- and delayed-implementation groups. A comparative question requires an estimate of that difference and its uncertainty. Neither is posted in the registry record, so the group intervals cannot substitute for a contrast-specific analysis.

Confidence intervals also depend on the unit of analysis and the sampling model. When participants are grouped within clinics, measurements from the same clinic may be more alike than measurements from different clinics. In general, this dependence is called clustering or within-cluster correlation. Ignoring it can make estimated uncertainty too small because observations within a clinic do not necessarily provide as much independent information as observations across separate clinics. The trial description says centers were randomized, but the registry does not report whether the outcome analysis incorporated center-level clustering, how many centers contributed outcome data, or what variance estimator was used.

Why does the center-level design matter?

In general, a cluster-randomized study assigns groups such as clinics to study conditions while collecting measurements on people treated at those sites. The allocation unit and outcome unit are then different. Analysis should reflect the allocation structure so that uncertainty in the treatment comparison is not calculated as though each person had been assigned independently. Possible methods include models or variance adjustments that account for clustering, but the specific method for this trial is not reported in the registry record. It is therefore not possible to assess from the posted summaries whether the reported intervals reflect the randomized design.

The outcome is a modeled risk estimate based on several patient characteristics and clinical factors. In general, a risk equation transforms measured predictors into an estimated probability or risk over a specified horizon. Such a score can serve as a surrogate or intermediate measure, but it is not identical to observing the clinical events that the model seeks to predict. The registry names the pooled risk equations and their listed inputs; further details about score calculation, handling of unavailable inputs, and the definition of change are not reported in the registry record.

What can be inferred from the posted summaries?

The registry gives a mean and interval for each arm, but no direct contrast, test statistic, or p-value. In general, a treatment-effect analysis would compare outcomes across assigned groups using a prespecified estimand and a method suited to the design. The estimand clarifies what effect is being targeted, such as a difference in mean change between assignment groups and the population to which that contrast applies. The registry does not report the estimand, covariate adjustment, model form, or inferential procedure. A direct comparative conclusion cannot be reconstructed from the arm summaries alone.

The enrollment entry and group counts also raise a reporting question. The record lists enrollment as 70 but gives outcome counts of 10980.0 and 7247.0. There is no explanation of whether these counts describe people, encounters, records, or a different unit, and no information on exclusions or data aggregation. Without that information, the analytic denominator and the relationship between sample size and precision cannot be evaluated. The entries should be reported as they appear rather than reconciled by assumption.

Sample-size calculations based on an observed effect are planning illustrations, not observed evidence about the trial’s power. In general, a planning calculation asks how many participants per arm might be needed under specified assumptions about the effect, significance level, allocation, and statistical model. Such an illustration does not establish that the trial had that power, and using an observed estimate for planning can yield an optimistic or unstable target. The registry describes an illustration based on the observed effect, two-sided significance level 0.05, equal allocation, and a normal approximation, with the detectable effect set as a stated fraction of the observed effect. Numerical results for these planning scenarios are not reported in the registry record. The trial’s own sample-size justification and assumptions are not reported in the registry record.

In general, over-dispersion means that observed variation is greater than expected under a model’s assumed distribution. It matters when the variance model is too restrictive, because standard errors and intervals may then fail to represent the actual uncertainty. The registry does not report an over-dispersion assessment or describe a distributional model beyond the primary outcome being continuous. It would not be appropriate to assert that over-dispersion occurred or that a particular correction was used.

Finally, the registry lists zero posted analyses. That entry does not reveal whether analyses were conducted outside the registry, whether results were submitted elsewhere, or whether no analysis was performed. The analysis plan, interim analysis procedures, multiplicity adjustments, and handling of missing outcome data are not reported in the registry record. These absences limit evaluation of how the displayed summaries were selected and how robustly they represent the randomized comparison.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists two outcome measures overall but posts no secondary outcome results. The primary endpoint is 1-year change in 10-year CVD risk. The broader study description mentions CVD care, disparities in care and outcomes, and patient engagement in treatment choices, but the registry does not provide corresponding secondary estimates. Whether these aims were formal endpoints, how they were ordered, and whether any multiplicity procedure was prespecified are not reported in the registry record.

Multiplicity arises in general when investigators test or estimate multiple outcomes, time points, subgroups, or contrasts. If many tests are conducted, the chance of at least one apparently notable result can increase unless the analysis plan addresses that issue. The registry does not report the number of planned comparisons, a hierarchy of outcomes, or any adjustment strategy. No conclusion about multiplicity control can be drawn from the posted primary outcome summary alone.

Missing-data handling is also not described. The record does not report the amount or pattern of missing endpoint data, reasons for missingness, imputation methods, or sensitivity analyses. In general, missingness can affect estimates when the availability of an outcome is related to its value or to treatment assignment. Without these details, the completeness of the outcome data and the sensitivity of the means to missing observations cannot be assessed.

The analysis populations are not reported in the registry record. It is not stated whether results are based on all randomized centers or participants, on those with available outcome measurements, or on another defined population. Nor is it clear how the outcome-group counts relate to the enrollment entry. These reporting gaps prevent a full assessment of whether the analysis preserved the randomized comparison and how observations were included.

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

The posted results establish that the registry reports a mean primary-outcome value of 0.4 in the immediate-implementation group and 0.2 in the delayed-implementation group, each with a 95% confidence interval. They describe the outcome as 1-year change in estimated 10-year CVD risk. These summaries are descriptive arm-level results, not a reported estimate of the effect of immediate implementation relative to delayed implementation.

They do not establish whether the groups differed statistically. A comparison requires a contrast and its uncertainty, and neither a comparative confidence interval nor a p-value is reported. The arm-level intervals should not be read as a test of the treatment difference. In addition, the record does not state how change is coded, so the values alone cannot be labeled as increases, decreases, or improvement.

The endpoint is a calculated risk estimate based on the pooled risk equations, not a count of observed heart attacks, strokes, or other cardiovascular events. The trial’s stated aims include reducing disparities and improving patient engagement, but no secondary results for those aims are posted. The primary summaries therefore do not establish effects on clinical events, disparities, or engagement.

The registry’s enrollment entry of 70 and its outcome-group counts of 10980.0 and 7247.0 are not reconciled. The meaning of the outcome counts and the analysis population are not reported. The method for accounting for randomization by center is also not reported, so the precision of the arm summaries cannot be evaluated against the design from the record alone.

In general, a randomized design can support causal comparisons when the analysis follows the randomized assignment and appropriately handles the design and missing data. The registry information does not include enough detail to assess those conditions for this endpoint. The results should therefore be described narrowly: they are the group means and confidence intervals posted for the primary outcome, with important details about the direct comparison and analysis not reported in the registry record.

12. Why This Trial Matters Statistically

This trial is statistically notable for assigning the intervention at the community health center level while reporting a continuous patient-level risk measure. That design makes the unit of randomization important to interpretation. In general, clustered allocation affects the amount of independent information in the data and the calculation of uncertainty. The registry’s lack of detail on center-level analysis means that readers cannot verify how the design is reflected in the intervals.

The comparison is between immediate and delayed implementation, not between implementation and a group that never receives the system. This timing structure matters when defining the effect of interest and the follow-up period. The posted primary endpoint uses a modeled 10-year risk estimate and is measured at 12 months, so it captures a risk-score change rather than long-term observed cardiovascular events.

The record also illustrates why outcome summaries need a clear denominator and analysis population. Enrollment is listed as 70, while the two outcome groups have counts of 10980.0 and 7247.0. Without an explanation of those entries, readers cannot determine the analytic unit or connect the counts to the randomized centers and participants. That ambiguity affects assessment of precision and reproducibility.

Finally, the registry reports group means but not the comparative effect that answers the randomized question. A careful statistical report would make the contrast, its uncertainty, the analytic population, and the approach to clustering explicit. These items are not reported in the registry record, so the posted information permits a limited descriptive reading rather than a complete evaluation of the intervention comparison.

13. 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 interpretation arise from the registry record. The primary endpoint is described as 1-year change in estimated 10-year CVD risk, but the summary does not specify the direction convention for change or explain how the baseline value was incorporated. That prevents assigning a favorable or unfavorable meaning to the two reported means solely from their signs and magnitudes.

The record gives arm-specific means and 95% confidence intervals but does not post a between-group estimate or statistical test. It is therefore not possible to characterize the comparative effect or its precision from the registry summaries alone. The analysis method for a continuous outcome, any covariate adjustment, and whether center-level clustering was addressed are not reported in the registry record.

The registry lists enrollment as 70, but the outcome summaries report 10980.0 and 7247.0. The record does not explain what those group counts represent, how they were formed, or why they differ in scale from the enrollment entry. The outcome analysis population and participant flow are not reported. This limits assessment of the denominators and of how the result relates to randomized assignment.

Although the study description identifies broader goals concerning care, disparities, and patient engagement, secondary outcome results are not posted. Missing-data procedures, subgroup analyses, multiplicity handling, and interim analyses are not reported in the registry record. The adverse-event fields list zero values for both groups, but the record does not clarify whether this means no events occurred or whether event reporting was absent. These gaps counsel against extending the primary endpoint summaries to outcomes or analyses that are not posted.

14. Related Tutorials

15. Related Statistical Calculators

16. Sources

• ClinicalTrials.gov: CV Wizard: Does a Clinical Decision Support Tool Improve CVD Risk Factor Control in Safety Net Clinics? — NCT03001713.
• Registry record last updated 2025-02-11; status last verified 2025-01.
• Results first posted 2025-02-11.
• 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.

17. Record Summary

The registry describes a completed randomized, parallel-group trial with no masking, comparing immediate and delayed implementation of the CV Wizard clinical decision support system in safety net community health centers. The primary endpoint is 1-year change in estimated 10-year CVD risk, reported as a mean with a 95% confidence interval. The immediate-implementation group has a reported mean of 0.4 (95% confidence interval 0.3 to 0.5); the delayed-implementation group has a mean of 0.2 (95% confidence interval 0.1 to 0.3).

No direct between-group estimate, test, or secondary outcome results are posted. The registry lists enrollment as 70, while the outcome groups report counts of 10980.0 and 7247.0; their relationship is not explained. The analysis population, missing-data procedures, and method for accounting for center-level randomization are not reported in the registry record. The summaries support a description of the posted group values, but not a complete assessment of the randomized treatment contrast.