Evaluation of Physician-based Decision Support: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Evaluation of Physician-based Decision Support (NCT03443713), a randomized na study. The primary outcome is Percent of Time With Sensed Glucose 70-180 mg/dl.

2. Clinical Question

The title identifies this as an evaluation of physician-based decision support, and the brief summary describes assessing physician-driven insulin setting changes among people with type 1 diabetes using multiple daily injections and exercising. The arm descriptions, however, specify the exercise type and use of a Dexcom G6 monitor. They do not explain the setting-change procedure, how the physician decisions were made, or how those decisions differed among arms. Those details are not reported in the registry record.

Accordingly, the most directly answerable question from the posted primary results is how the average time in the stated glucose range was described within each assigned exercise group. Comparisons between groups must be interpreted cautiously: arm summaries alone do not quantify uncertainty in a between-group contrast, and no contrast estimate or inferential test is reported in the registry record.

3. Trial Design

This was a randomized, parallel-group study with no masking listed in the registry. The phase is recorded as NA, and the study status is completed. Oregon Health and Science University is listed as the sponsor. The registry reports an enrollment of 25 and gives the study dates as 2018-07-18 for start and 2019-05-08 for primary completion.

The three arms were aerobic exercise, anaerobic exercise, and high intensity interval exercise. The descriptions say subjects in each arm completed their assigned exercise at home and at OHSU and used a Dexcom G6 continuous glucose monitor. The overall study description characterizes it as a short outpatient study with multiple outpatient and home exercise sessions. The number, duration, and scheduling of those sessions are not reported in the registry record.

For the primary endpoint, the registry reports a mean and standard deviation for each arm, with n=8 in the aerobic arm, n=8 in the anaerobic arm, and n=8 in the high intensity interval arm. The primary endpoint description states that 25 subjects were enrolled and that one subject assigned to high intensity interval exercise dropped out before completing the exercise visits. The registry does not explain how the endpoint analysis handled incomplete observations or the relationship between the dropout and the reported endpoint summary.

Random assignment can support a fair comparison when allocation, follow-up, and analysis are appropriately conducted, but the registry record does not report the randomization sequence method, allocation concealment, or an analysis plan. It lists no masking. Whether that was feasible or attempted for any particular study activity is not reported. The posted record does not describe the intended enrollment rationale or a prespecified statistical comparison among the exercise arms.

The study title and brief summary refer to physician-based support and insulin setting changes, while the listed intervention is the Dexcom G6 device and the arm descriptions center on exercise type. The exact role and timing of decision support in the trial procedures are not reported in the registry record. Thus the registry supports a description of a randomized study across three exercise categories, but not a detailed reconstruction of how physician input was implemented.

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.

Start up Visit (2 Hours)

First at Home Period (9-14 Days)

Reasons for not completing this period:

First Exercise Visit (1 Hour)

Dexcom Insertion Visit (30 Minutes)

Analysis population for the primary outcome, as defined in the registry: 25 subjects were enrolled. One subject assigned to High intensity interval exercise dropped out before completing the exercise visits.

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 Percent of Time With Sensed Glucose 70-180 mg/dl, measured over Days 1-7.

The primary endpoint is the percentage of time with sensed glucose between 70–180 mg/dl. It is a continuous measure: each participant’s monitor readings contribute to a proportion of monitored time within the specified range, and the endpoint is averaged across Days 1-7. The registry says the values came from Dexcom G6 sensor data uploaded to Dexcom Clarity.

A time-in-range measure expresses glucose readings over time within a defined interval, rather than reporting a single glucose value. In general, such a summary can describe the distribution of monitored glucose across a period and can be compared descriptively among study groups. It is distinct from a measure of average glucose alone: two groups could have similar mean glucose while differing in how readings are distributed over time. This is general statistical context, not a finding from this study.

The posted arm means are 58.8% for aerobic exercise, 51.2% for anaerobic exercise, and 59.3% for high intensity interval exercise. The corresponding standard deviations are 18.9, 10.8, and 14.6. These values describe the group averages and the variation among observations as summarized in the registry; they do not by themselves show whether an observed difference between groups is compatible with chance variation.

The registry does not report confidence intervals, an adjusted comparison, or a statistical test for the primary endpoint. It also does not give further details about the amount of usable sensor data required for inclusion in the endpoint, or how missing monitor readings were treated. These omissions limit interpretation of precision and completeness.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a mean summarize?

The primary endpoint is continuous, and the registry identifies its summary parameter as a mean. A mean combines the individual endpoint values into an average on the same scale as the underlying measure. Here, the scale is percentage of time with sensed glucose within the stated range during Days 1-7. An arm mean is useful for describing the center of the observed values, but it does not reveal every participant’s value or the shape of the distribution.

The registry reports a standard deviation alongside each mean. In general, the standard deviation describes the typical spread of observations around their mean. A larger standard deviation indicates more variability among the values being summarized; it does not mean that the mean itself has that much uncertainty. The standard deviation is also sensitive to the distribution of values and does not show whether there are clusters, outliers, or skewness. Such distributional details are not reported in the registry record.

What is the distinction between variability and precision?

Variability is variation among observed measurements. Precision concerns uncertainty in an estimate, such as an arm mean or a difference between arm means. A confidence interval is a general method for expressing estimate precision: under its assumptions, it gives a range of values compatible with the data and the procedure used to construct it. A confidence interval for a contrast would help readers assess both the direction and plausible magnitude of a between-arm difference. No confidence interval is reported in the registry record.

A standard deviation cannot substitute for a confidence interval. It describes spread in the observations, while the uncertainty of a mean also depends on the amount and structure of data and on the statistical method. Because the registry does not report confidence intervals or a formal contrast analysis, the precision of comparisons among the three means cannot be assessed from the posted primary summaries alone.

What would an inferential comparison add?

An inferential analysis would estimate a contrast among arms and quantify uncertainty around that contrast, potentially with an interval and a test. The analysis would need to specify which groups are compared, whether the comparison is adjusted for any factors, and how the outcome distribution and incomplete measurements are handled. The registry lists no analysis entries for this outcome. It does not report a p-value, an adjusted effect estimate, or the planned model. Therefore, no inferential conclusion about differences among the exercise arms is available from the registry record.

Randomized assignment is a design feature, not a substitute for a reported analysis. In general, randomization helps balance known and unknown baseline characteristics on average, but the method of generating and concealing assignments can matter. The registry does not describe these methods. It also does not state whether the analysis followed assigned groups or was limited to participants who completed specified procedures. Those analysis-population details are not reported in the registry record.

How should planning calculations be understood?

The registry’s planning description characterizes sample-size calculations as illustrations based on an observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It says planning rows use a stated fraction of the observed effect and that the resulting figures represent participants per arm for non-time-to-event outcomes. These are planning assumptions, not evidence about the probability that this completed study detected an effect. The planning description explicitly distinguishes such illustrations from the trial’s own design.

In general, a sample-size calculation asks how many observations might be needed to detect a specified effect under stated assumptions. Its result depends on the target effect, variability, significance threshold, power target, allocation, and analysis method. A calculation based on an observed effect is particularly dependent on that estimate and should not be interpreted as observed power. No sample-size justification or power target for the trial’s original design is reported in the registry record. The planning illustration cannot repair the absence of a reported inferential analysis.

Why might over-dispersion matter?

In general, over-dispersion means observed variability is greater than a model assumes. When a model underestimates variability, standard errors may be too small and confidence intervals too narrow, which can make evidence appear more precise than it is. Model choice should reflect the outcome’s scale, distribution, and data structure. The posted registry summaries do not identify a fitted model or provide enough distributional information to assess over-dispersion, and no model-based comparison is reported. This is general statistical context rather than a claim that over-dispersion occurred in this study.

The primary endpoint is a percentage derived from repeated sensor readings over time. Repeated readings within a person are not automatically independent observations. General analyses of repeated measurements must account for the way observations are grouped within participants and how participant-level endpoint values are formed. The registry says the percentage was averaged across Days 1-7 but does not describe the underlying calculation beyond that summary, nor does it report a model for repeated data. It is therefore not possible to infer which dependence structure, if any, was used.

What can be learned from the posted summaries?

The means and standard deviations offer a direct descriptive account of the endpoint in each arm. They show the reported average and dispersion, but not the uncertainty around a group contrast, the adjustment strategy, or the robustness of a conclusion to missing observations. A full statistical interpretation would need the prespecified analysis, the number of participants contributing endpoint data, treatment of incomplete sensor data, and estimates with uncertainty for the comparisons of interest. Those details are not reported in the registry record.

10. Confidence Intervals for Each Arm

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

R code and output

arm <- c("Aerobic Exercise", "Anaerobic Exercise", "High Intensity Interval Exercise")
n   <- c(8, 8, 8)
m   <- c(58.8, 51.2, 59.3)
sp  <- c(18.9, 10.8, 14.6)
sd  <- sp      # posted standard deviations
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Aerobic Exercise  8  58.800  18.900  42.999  74.601
CARM  Anaerobic Exercise  8  51.200  10.800  42.171  60.229
CARM  High Intensity Interval Exercise  8  59.300  14.600  47.094  71.506

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 8 outcome measures and reports 4 secondary outcomes. The named secondary measures are percent of time with sensed glucose less than 54 mg/dl and mean sensed glucose, with each name appearing twice in the registry list. Results for these secondary measures are not reported in the available registry record. The record does not explain whether repeated entries refer to different time points, analyses, or other distinctions.

Multiplicity arises when several outcomes or comparisons are examined. In general, the more statistical tests conducted, the greater the chance that at least one result may appear notable by chance if each is interpreted without accounting for the family of tests. Methods for addressing multiplicity can include prespecifying a primary comparison, adjusting thresholds, or treating secondary findings as exploratory. The registry record does not report a multiplicity strategy, the intended hierarchy among outcomes, or inferential results for the secondary outcomes. It is therefore not possible to determine how multiple testing was handled.

The primary endpoint description records 25 enrolled subjects and notes one dropout assigned to high intensity interval exercise before completing exercise visits. The primary outcome summaries report n=8 in each arm. The record does not describe whether these summaries include all randomized participants, only those with complete data, or another analysis population. Nor does it report missing sensor-data counts, the rules for defining usable data, or any imputation method. Missing-data handling is therefore not reported in the registry record.

The registry reports no analysis entries. It does not state whether analyses were based on assigned groups, completed exercise sessions, or available endpoint data. No protocol deviations or sensitivity analyses are described in the registry record. These omissions make it difficult to assess how incomplete participation or measurements could affect the arm summaries and any intended comparisons.

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 non-serious adverse event terms

The registry reports zero serious adverse events in each exercise arm and zero deaths in each arm. It reports other adverse events affecting 2 participants in the aerobic exercise arm, 0 in the anaerobic exercise arm, and 3 in the high intensity interval exercise arm. These are the counts recorded in the registry; the record does not provide descriptions of the individual events.

The adverse-event fields list 8 at risk in the aerobic arm and 8 in the anaerobic arm, and 9 in the high intensity interval arm. These are safety denominators as recorded. The primary endpoint summaries separately report n=8 for each arm, and the primary endpoint description notes one dropout assigned to high intensity interval exercise before completing the exercise visits. The registry does not explain the difference between the safety denominator and the primary endpoint summary for that arm, so they should not be treated as interchangeable.

A zero count in a registry report means no events were recorded in that category for the stated group; it does not establish that the risk is absent. Counts from a small study can provide limited information about uncommon events. The registry lists no serious adverse-event terms and does not provide event narratives, severity details for other adverse events, or further attribution information. These safety counts describe the posted record and are not, by themselves, a comparison of safety among exercise types.

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

The primary results show three arm-specific averages for time with sensed glucose between 70–180 mg/dl during Days 1-7. The mean was 58.8% for aerobic exercise, 51.2% for anaerobic exercise, and 59.3% for high intensity interval exercise. Standard deviations were 18.9, 10.8, and 14.6, respectively. These are descriptive summaries of the endpoint as reported in the registry.

The results do not establish that one exercise type changes time in the stated glucose range more than another. The registry does not report confidence intervals or a statistical test for differences among the arms. It also does not report an analysis plan or explain the handling of incomplete endpoint data. Without those details, the size and precision of any between-arm contrast cannot be judged from the arm means and standard deviations alone.

The study is described as evaluating physician-driven insulin setting changes in people using multiple daily injections and exercising. However, the registry does not give the detailed decision-support procedure or report an analysis that isolates its effect. The posted primary results are grouped by exercise type, so they cannot independently establish the effect of physician input, insulin adjustment, or exercise in general.

The study enrolled 25 subjects, and the primary endpoint description notes a dropout before completion of exercise visits in the high intensity interval arm. The registry does not explain how that event affected endpoint availability or analysis. The endpoint represents Days 1-7, not a long-term outcome, and the record does not report results beyond that stated period.

The appropriate conclusion is limited to the reported summaries and study record: arm-level endpoint means and standard deviations are available, along with limited adverse-event counts. Claims about comparative effectiveness, clinical importance, longer-term outcomes, or a lack of risk are not established by these data. The posted record does not contain the inferential results needed to quantify comparative uncertainty.

14. Why This Trial Matters Statistically

This trial illustrates a common distinction in clinical research between a randomized design and the statistical evidence reported from that design. Random assignment is relevant to how groups are formed, but the interpretation of treatment comparisons also depends on the endpoint definition, analysis population, handling of missing data, and reporting of uncertainty. Here, the registry gives arm-level means and standard deviations for a continuous endpoint but no inferential comparison.

The endpoint is a percentage of time derived from continuous glucose monitoring over Days 1-7. It offers a summary of monitored readings over a defined interval, while also raising practical analytic questions: how missing sensor readings are handled, how repeated observations contribute to each participant’s percentage, and how a group contrast is estimated. The registry reports the averaging period and source of the sensor values but does not give the additional analytic details needed to answer those questions.

The record also highlights the importance of distinguishing a study’s motivating question from its posted outcome summaries. The title and brief summary emphasize physician-based decision support and insulin setting changes. The primary outcome results are reported by exercise arm, and the specific decision-support procedure and its effect are not detailed. Clear reporting of the link between an intervention, its implementation, and the analysis helps readers understand what a trial can answer.

Finally, planning calculations and observed results serve different purposes. A planning illustration can show how assumptions about a target effect and statistical method relate to a future study’s size, but it is not an estimate of the uncertainty in these posted group summaries and is not observed power. With no reported contrast estimate or confidence interval, the central statistical lesson is the need to separate descriptive results from comparative inference.

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 leaves several trial-specific details unavailable. It lists randomized allocation but does not describe sequence generation or allocation concealment. It reports no masking, and the record does not say whether outcome processing or other trial tasks were masked. The phase is listed as NA. The protocol-level rationale for enrollment and the prespecified statistical plan are not reported.

The primary endpoint is described as an average percentage across Days 1-7, but the record does not give the rules for usable sensor data, how missing readings were handled, or whether participants needed a minimum monitoring duration. The dropout before exercise visits is noted for one subject assigned to high intensity interval exercise, but the record does not say whether that subject contributed endpoint data or how the dropout was handled analytically. Although the primary summaries list n=8 per arm, the relationship between those reported counts, enrollment, and endpoint availability is not explained.

No confidence intervals, p-values, model estimates, or analysis entries are reported for the primary outcome. There is no description of adjusted or unadjusted comparisons, assumptions checked, or sensitivity analyses. The registry also does not state how multiple outcome measures or comparisons were to be handled. These gaps limit assessment of uncertainty, robustness, and selective reporting.

The exercise arms are described, but the physician-driven insulin setting changes emphasized in the study summary are not operationally detailed. The record does not report how the settings were selected, when they changed, or how their effects were evaluated separately from exercise assignment. Secondary outcomes are named but their result values are not reported in the registry record. These limitations restrict both the statistical interpretation of the primary summaries and the scope of conclusions that can be drawn about the study’s stated decision-support question.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Evaluation of Physician-based Decision Support — NCT03443713.
• Registry record last updated 2020-03-18; status last verified 2020-03.
• Results first posted 2020-03-18.
• 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

NCT03443713 is a completed randomized, unmasked, parallel-group study titled “Evaluation of Physician-based Decision Support.” Oregon Health and Science University is listed as sponsor. The registry records the study start as 2018-07-18 and primary completion as 2019-05-08, with enrollment of 25. The phase is listed as NA.

Participants were assigned to aerobic exercise, anaerobic exercise, or high intensity interval exercise. The arm descriptions state that participants completed their assigned exercise at home and at OHSU and used a Dexcom G6 monitor. The primary endpoint was percentage of time with sensed glucose between 70–180 mg/dl, averaged across Days 1-7.

The posted primary summaries are means and standard deviations by arm. The registry does not report a statistical comparison, confidence intervals, a statistical analysis plan, or details of missing-data handling. It reports zero serious adverse events and zero deaths in all arms, alongside the listed counts of other adverse events. These results support a descriptive account of the recorded endpoints and events, not a quantified comparative conclusion.