Pragmatic Trial Comparing Telehealth Care and Optimized Clinic-Based Care for Uncontrolled High Blood Pressure: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Pragmatic Trial Comparing Telehealth Care and Optimized Clinic-Based Care for Uncontrolled High Blood Pressure (NCT02996565), a randomized na study. The primary outcome is Change in Systolic Blood Pressure Between Baseline and 12 Months (mm Hg).

2. Clinical Question

The comparison is between care models assigned at the clinic level, rather than between separate treatments assigned within the same clinic. That distinction matters when interpreting the results: participants in a clinic share a care setting, and outcomes may be more alike within clinics than across clinics. The registry record identifies clinic assignment but does not report the details of how that dependence was handled in the primary analysis.

The question is about comparative change over time, not the absolute blood pressure level at a single visit. A negative mean change represents a reduction from baseline on the outcome’s stated scale. The estimate of the difference between arms is therefore central to deciding whether the posted results distinguish the models on the primary endpoint. The registry record does not report a primary result for patient preference.

3. Trial Design

This was a completed, randomized, parallel-group pragmatic trial. Clinics were randomized to offer either Best Practice Clinic-Based Care or Telehealth Care, and eligible participants received the model assigned to their primary-care clinic. The registry identifies the study as unmasked. The phase is listed as NA. HealthPartners Institute is named as the sponsor.

The trial enrolled 3071 participants. The registry reports a start date of 2017-11-15 and a primary completion date of 2020-04-17. Its description places the study in a large care system in Minnesota and frames the comparison as one between team-based approaches to managing hypertension. The intervention descriptions characterize the clinic-based and telehealth arms by the care model offered at the assigned clinic; the telehealth description specifically notes pharmacist management.

Because clinics, rather than individual participants, were randomized, the design is cluster-randomized in practical terms. General statistical context: outcomes from people treated in the same clinic may be correlated because they share staff, workflows, and care environments. An analysis that ignores this correlation can give uncertainty estimates that are too narrow. The registry identifies a planned contrast for the primary outcome, but the specific model details and handling of clinic-level clustering are not reported in the registry record.

The trial is described as pragmatic, indicating a comparison conducted in a care-system setting rather than a tightly controlled efficacy setting. That label alone does not specify the eligibility rules, implementation fidelity, or how much clinical care varied within each model; those details are not reported in the registry record. The record also does not report a sample-size justification, an interim-analysis plan, or further details of the randomization procedure.

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.

Before assignment: Among 57 primary care clinics screened for eligibility, 36 were excluded (16 no MTM; 20 not using automated BP monitor) and 21 were randomized (17 individual clinics plus 4 clinics as 2 co-located pairs). Among 69,480 patients screened in primary care encounters, 3,071 met all inclusion criteria and were enrolled in the study and received care according to the clinic-randomized treatment arm.

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 Change in Systolic Blood Pressure Between Baseline and 12 Months (mm Hg), measured over Trajectory over 12 months.

The primary endpoint is mean change in systolic blood pressure between baseline and 12 months, measured in mm Hg and collected from medical records. It describes how the recorded systolic pressure changed over the follow-up period, rather than reporting only a single measurement. A negative value indicates a decrease relative to baseline on the endpoint’s scale.

Systolic blood pressure is the pressure measurement used in the primary question about managing uncontrolled hypertension. Comparing its change between care models gives a common quantitative outcome for assessing whether the assigned approaches were associated with different trajectories over the stated follow-up. General statistical context: a continuous measure preserves information about the magnitude of change, rather than reducing the outcome to a yes-or-no category. The registry’s primary result is expressed as arm means with 95% confidence intervals and as a mean difference between arms with its own 95% confidence interval.

The arm-level estimates are -17.97 mm Hg for Best Practice Clinic-Based Care, with a 95% confidence interval from -19.43 to -16.5, and -18.72 mm Hg for Telehealth Care, with a 95% confidence interval from -20.2 to -17.24. The reported net mean difference is -0.76 mm Hg, with a 95% confidence interval from -2.84 to 1.33. The interval for the contrast includes zero, so the posted primary result does not establish a clear difference between the care models. The registry record does not provide further detail on the timing or number of blood-pressure measurements contributing to the endpoint.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is continuous, so its arm-level summaries are means and its comparative estimate is a mean difference. A mean describes the center of the observed values, while a confidence interval conveys uncertainty in an estimate. General statistical context: a 95% confidence interval is generated by a procedure designed to cover the underlying parameter in 95% of repeated samples under its assumptions. It is not a statement that there is a 95% probability that this particular interval contains the parameter.

What does the primary contrast represent?

The reported planned contrast is a mean difference between the arms, with a 95% confidence interval from -2.84 to 1.33 and a p-value of 0.45. A contrast compares estimated outcomes across groups; it is not the same as either group’s change considered alone. Zero is the no-difference value for a mean difference. Since the reported interval includes zero, the data do not clearly distinguish the groups on this endpoint. The interval also communicates the range of contrasts compatible with the analysis, subject to its assumptions.

The trial is designated as a superiority comparison. The null hypothesis for a conventional superiority test is that the comparative effect is zero; a p-value describes how unusual the data, or more extreme data, would be under that null model and its assumptions. It is not the probability that the null hypothesis is true. A p-value of 0.45 is not evidence that the treatment models are equivalent. Demonstrating equivalence would require an equivalence margin and a suitable design and analysis, neither of which is reported in the registry record.

Why does clinic assignment matter?

Clinics were randomized to a care model. General statistical context: when assignment occurs in clusters, participants within a cluster can have correlated outcomes. This dependence reduces the amount of independent information compared with a design in which every person is independently randomized. Appropriate analysis may need to account for the cluster structure. The registry names a planned contrast for the primary comparison, but the model specification, cluster adjustment, and any small-sample corrections are not reported in the registry record. Those details cannot be inferred from the estimate alone.

The record identifies an unmasked trial. General statistical context: lack of masking can matter through differences in behavior, measurement, care delivery, or outcome reporting. Here the primary endpoint is described as collected from medical records, but the record does not specify measurement procedures, whether assessors knew the assigned model, or how the outcome data were checked. Unmasked status is a design feature to consider, not proof that bias occurred.

How should secondary comparisons be read?

Three secondary outcomes have arm-level values and comparative analyses in the listed results. The registry reports mixed-model risk ratios for these outcomes. A risk ratio compares the outcome probability in one group with that in another; a value of one is the no-relative-difference value. The listed estimates for monitoring blood pressure and satisfaction favor Telehealth Care, while the smoking comparison does not clearly distinguish the groups. These analyses are secondary and should not replace the primary comparison.

Simple recalculations based on posted summary counts are unadjusted comparisons. They are not the same as the registry’s reported mixed-model analyses, which may account for features of the design or analysis that a calculation from aggregate counts cannot reproduce. Where estimates from these approaches differ, their interpretation should retain that distinction. The primary endpoint has a sponsor-reported planned contrast; a separate simple unadjusted two-arm recalculation is not available for that continuous outcome in the posted record.

What do the planning illustrations mean?

The sample-size illustrations are calculations based on observed secondary-outcome effects, not the trial’s own design justification. The stated assumptions are 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. General statistical context: smaller target effects usually require more participants to detect, all else equal. These planning counts are hypothetical design illustrations. They are not observed power, do not describe the trial’s actual enrollment rationale, and do not establish the probability that the trial detected an effect.

The registry lists 30 outcome measures and 30 analyses, while 29 secondary outcomes are posted. The analysis plan, outcome hierarchy beyond the primary endpoint, multiplicity strategy, analysis populations, and missing-data methods are not reported in the registry record. These omissions constrain how fully the reported estimates and their uncertainty can be evaluated.

10. Analyses Reported by the Sponsor

The primary result is a reported planned contrast: a net mean difference of -0.76 mm Hg, with a 95% confidence interval from -2.84 to 1.33 and p = 0.45. This is the central comparative estimate for the primary endpoint. The registry record also reports the mean change and confidence interval within each arm, but those arm-specific estimates are not substitutes for the between-arm contrast.

The simple unadjusted recalculation available for certain binary secondary outcomes cannot be used to recreate the primary continuous-outcome contrast. The registry record indicates that values are missing or not estimable for a two-arm recalculation of this continuous endpoint. Accordingly, the sponsor-reported planned contrast is the primary comparison to interpret; there is no competing recalculated primary estimate to set beside it.

The reported primary analysis does not clearly separate the care models: its confidence interval includes zero, and its p-value is 0.45. The registry record does not give enough model detail to independently reconstruct the analysis, including how clinic-level assignment was handled. The reported contrast can be interpreted as posted, while recognizing that this methodological detail and other aspects of the statistical plan are not reported in the registry record.

11. Multiplicity, Missing Data and Analysis Populations

The registry identifies one primary outcome and lists 30 outcome measures and 30 analyses. It also reports 29 secondary outcomes. Multiple outcomes and analyses create a multiplicity issue: general statistical context, the more tests considered, the greater the opportunity for some results to appear notable by chance when each is assessed separately. A prespecified hierarchy or adjustment can help control how such findings are interpreted. The registry record does not report whether multiplicity was addressed, whether secondary analyses were exploratory, or whether any hierarchy among secondary outcomes was prespecified.

The reported primary comparison is therefore the clearest basis for the main conclusion. Secondary findings can add context, but absent a reported multiplicity plan they should not be treated as though each were an isolated confirmatory test. The record does not report the statistical analysis plan or additional criteria defining confirmatory versus exploratory analyses.

Outcome denominators vary across the primary and secondary results. This makes it important to distinguish the enrollment count from the number contributing to each analysis. The registry record does not report why outcome counts differ, the amount or pattern of missing data, or the methods used to handle missing observations. It also does not report whether sensitivity analyses assessed the effects of missing data. Consequently, the direction and possible impact of missingness cannot be determined from the posted results.

The record does not identify the analysis population, such as whether the primary result followed an intention-to-treat approach, or how participants with incomplete follow-up were handled. It does not report a per-protocol analysis or the criteria for including participants in any analysis population. These details matter for understanding whether the reported comparison reflects assignment to the clinic model, receipt of the model, or a subset with available outcome data.

Finally, because clinics were randomized, the handling of within-clinic correlation is relevant to both standard errors and confidence intervals. The registry reports the planned contrast but not the full model specification or cluster-adjustment details. These omissions do not invalidate the posted estimate, but they limit an independent assessment of its precision and assumptions.

12. Safety Results

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

Deaths

Other (non-serious) adverse events

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

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hypotension (1.9%), hyponatremia (0.5%), hyperkalemia (0.1%), hypokalemia (0.1%).

The registry reports serious adverse events among 1646 participants at risk in Best Practice Clinic-Based Care, with 29 affected, and among 1422 participants at risk in Telehealth Care, with 34 affected. These are reported counts, not evidence by themselves that an event was caused by the assigned care model. The record also lists 42 deaths in the clinic-based arm and 43 in the telehealth arm; it does not state the causes or attribute them to an intervention.

The registry lists hypotension at 1.9%, hyponatremia at 0.5%, hyperkalemia at 0.1%, and hypokalemia at 0.1% among the top serious-event terms. The record does not provide arm-specific figures for these terms in the information, so they should not be used to compare event rates between arms.

Safety counts should be read alongside the denominators, event definitions, follow-up period, and completeness of reporting. The registry record does not report further detail on event ascertainment, event timing, severity criteria, or causal assessment. Counts alone cannot establish comparative safety or explain why an event occurred.

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

The primary result addresses whether the two assigned care models differed in mean systolic blood-pressure change from baseline to 12 months. The reported estimate is close to zero relative to its uncertainty, and the 95% confidence interval includes zero. The analysis therefore does not establish a clear difference between the models on the primary outcome. This is not a demonstration that the models are equivalent, nor is it proof that their effects are identical.

The arm-level results show negative mean changes in both groups. Those values describe the observed average changes in this trial; they do not establish what would have happened to the same participants under a different care model, and they are not comparisons against a no-care group. Randomization supports the between-arm comparison, subject to the design and analysis assumptions, rather than a causal interpretation of each arm’s change on its own.

The secondary results offer information about selected outcomes, including self-monitoring, satisfaction, and smoking. Their pattern is mixed and they do not replace the primary endpoint. The record does not report the multiplicity plan, missing-data approach, or full analysis model details, which limits the strength of conclusions that can be drawn from secondary findings or the precision of the primary estimate.

The findings apply to the trial setting and the participants and clinics included in this pragmatic comparison. The registry describes a large care system in Minnesota, but does not provide enough detail to judge how closely other care settings or populations resemble it. The results do not imply that all clinics would implement the models in the same way or achieve the same outcomes.

Overall, the posted statistics describe a randomized comparison with no clear primary-outcome separation and some secondary differences. They do not establish that one model is universally preferable, that the models are interchangeable, or that any particular participant will experience a given outcome. The registry record does not report a primary result for the trial’s stated aim concerning patient preference.

14. Why This Trial Matters Statistically

This trial is statistically informative because it compares care models under clinic-level randomization in a pragmatic setting. That design makes the analysis more than a simple comparison of two independent sets of individual participants: outcomes may share features of the clinic to which participants were assigned. Clear reporting of how such clustering is handled is important for interpreting the uncertainty around the estimated difference. The registry identifies the clinic-level assignment but does not report the full primary model specification.

The primary result also illustrates why within-arm changes and between-arm effects answer different questions. Both arms have negative mean changes, but the randomized question is whether the change differs between the assigned models. The reported contrast and its confidence interval address that question directly. Its interval includes zero, so the primary outcome does not clearly distinguish the models, despite the negative mean changes in each arm.

The record includes several secondary outcomes, which can illuminate other aspects of care, but these are not interchangeable with the primary outcome. Considering multiple secondary comparisons without a reported multiplicity plan requires restraint: an isolated finding may be less persuasive when many outcomes are being examined. The secondary pattern can be useful for generating or refining questions, while the primary estimate remains the main comparative result.

Finally, the registry record exposes important limits on statistical interpretation: it does not report the sample-size justification, analysis populations, missing-data methods, or a complete description of the analysis models. Those details are essential for evaluating how robustly the reported estimates reflect the randomized design. Recognizing these limits allows the results to be read as evidence with measurable uncertainty, rather than as a simple verdict about either care model.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

Several limitations specific to this trial affect interpretation. Clinics were randomized, but the registry does not report the number of clinics, their distribution between arms, or the details of how clinic-level correlation was handled in the primary analysis. These design details matter because the precision of a cluster-randomized comparison depends on the cluster structure, not only on the total participant count.

The trial was unmasked. Although the primary endpoint was collected from medical records, the registry does not describe measurement procedures, whether assessors knew the assigned care model, or whether blood-pressure measurements were standardized across clinics. The record therefore does not establish the direction or size of any possible measurement or implementation effects.

The primary outcome is based on change over 12 months, but the registry does not report the number or timing of measurements used to form each participant’s outcome, how baseline values were selected, or how repeated measurements contributed to the estimated trajectory. The listed primary analysis gives a mean contrast, but the registry does not report enough model detail to assess all assumptions behind that estimate.

The denominators differ across outcomes, and reasons for those differences are not reported. The registry record does not describe missing-data patterns, how incomplete follow-up was handled, or whether sensitivity analyses were performed. It also does not identify the primary analysis population. The sample-size justification and any interim analyses are not reported in the registry record.

The trial’s setting is described as a large care system in Minnesota. Further details needed to assess generalizability, including participant characteristics and clinic characteristics, are not reported in the record used for this analysis. The trial description names patient preference as an aim, but a preference result is not reported among the posted outcomes. These gaps limit conclusions about the reach of the findings and about aims beyond the primary blood-pressure comparison.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Pragmatic Trial Comparing Telehealth Care and Optimized Clinic-Based Care for Uncontrolled High Blood Pressure — NCT02996565.
• Registry record last updated 2024-09-23; status last verified 2020-12.
• Results first posted 2024-09-23.
• 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

The registry record describes a completed, randomized, parallel-group pragmatic trial comparing Best Practice Clinic-Based Care with Telehealth Care for participants with uncontrolled hypertension. Clinics were randomized to offer one of the two models, and the trial enrolled 3071 participants in a large care system in Minnesota. The study was unmasked and sponsored by HealthPartners Institute.

The primary outcome was mean change in systolic blood pressure between baseline and 12 months, collected from medical records. The reported mean difference was -0.76 mm Hg, with a 95% confidence interval from -2.84 to 1.33 and p = 0.45. The interval includes zero and does not establish a clear between-model difference on the primary outcome.

The record lists 30 outcome measures and 30 analyses, with 29 secondary outcomes posted. Selected secondary results show a mixed pattern. The registry does not report the full analysis plan, multiplicity strategy, missing-data methods, or detailed handling of clinic-level correlation. Serious adverse event counts and deaths are also reported, but the listed figures do not establish causation.