Using Peer Support to Aid in Prevention and Treatment in Prediabetes: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Using Peer Support to Aid in Prevention and Treatment in Prediabetes (NCT03689530), a randomized na study. The primary outcome is Change in Glycosylated Hemoglobin A1c (HbA1c) at 6 Months.

2. Clinical Question

3. Trial Design

UPSTART, formally titled “Using Peer Support to Aid in Prevention and Treatment in Prediabetes,” was a randomized, parallel-group behavioral trial sponsored by the University of Michigan. The registry records an enrollment of 450, a single masking designation, and a completed status. The recorded start date is 2018-10-02, and the primary completion date is 2021-10-25. The phase is listed as NA.

Participants were assigned to peer support or enhanced usual care. In the peer-support arm, participants were matched with a peer supporter. The enhanced usual-care arm received brief education and a folder of information and resources. The registry does not report the specific allocation procedure, what the single masking designation means in practice, or whether participants, staff, or outcome assessors were masked.

The primary outcome was change in HbA1c from baseline to 6 months. The registry also lists secondary outcomes covering HbA1c and body weight at 12 months, enrollment in and attendance at formal diabetes-prevention programs, physical activity, sugary soda consumption, patient activation, and perceived confidence in taking steps to prevent diabetes. The listed measures span laboratory, physical, behavioral, and patient-reported domains.

For the primary outcome, results are reported for 135 participants in the peer-support arm and 132 in enhanced usual care. The registry notes that primary-outcome data were unavailable for 23 participants in the intervention arm and 37 in the control arm because of scheduling issues and COVID. The relationship between those missing observations and the analysis population, and the trial’s original sample-size justification, 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

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Data was unavailable for 23 participants in the Intervention arm and 37 participants in the Control arm because of scheduling issues and COVID.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Includes 95 peer supporters who delivered the intervention.

6. Primary Endpoint

The registry lists the primary outcome as Change in Glycosylated Hemoglobin A1c (HbA1c) at 6 Months, measured over baseline to 6 months.

The primary endpoint is change in glycosylated hemoglobin A1c (HbA1c) from baseline to 6 months, measured as a percentage. HbA1c is the percentage of red blood cells with glucose-coated hemoglobin and indicates average blood sugar levels over the preceding 2-3 months. The registry notes that 5.7% to 6.4% is considered the prediabetes range.

Change is useful because it focuses on how the measure moved over the follow-up period, rather than only on a single measurement at the end. The registry explains that negative change values indicate a reduction from baseline. For this reason, a negative mean change is interpreted as a drop in HbA1c, not as a negative HbA1c level.

The comparison of interest is the difference in mean change between the randomized groups. A difference near zero indicates that the average changes were similar in the observed groups; uncertainty around that difference matters when judging whether the result is compatible with a range of effects. HbA1c is a continuous measure, so the registry reports group means and standard deviations. Those summaries describe the average change and the spread of individual changes, respectively.

HbA1c is relevant to a prevention trial because it measures a marker of blood sugar over time. The primary endpoint alone, however, does not capture every aspect of program engagement or behavior change. The registry therefore also lists outcomes related to formal program participation, physical activity, body weight, and patient-reported measures.

7. Results Posted in the Registry

• Peer Support Arm (Intervention): mean -0.07 versus -0.06 on control; difference -0.01 (95% CI -0.06 to 0.04), standardized effect d = -0.05; z-test p = 0.698.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Change in Glycosylated Hemoglobin A1c (HbA1c) at 12 Months

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • Peer Support Arm (Intervention): mean -0.06 versus -0.06 on control; difference 0.00 (95% CI -0.06 to 0.06), standardized effect d = 0.00; z-test p = 1.000.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Peer Support Arm (Intervention) and the control. The interval gives the range of effects that remain compatible with the data.

Change in Body Weight at 12 Months

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • Peer Support Arm (Intervention): mean -0.79 versus 0.22 on control; difference -1.01 (95% CI -2.25 to 0.23), standardized effect d = -0.19; z-test p = 0.111.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Peer Support Arm (Intervention) and the control. The interval gives the range of effects that remain compatible with the data.

Change in Whether Participant Enrolled in a Formal Program to Prevent Diabetes

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Peer Support Arm (Intervention): 25.4% (34/134) versus 14.3% (20/140) on control; risk difference 11.1 percentage points (95% CI 1.7 to 20.5), risk ratio 1.78 (1.08–2.93); unadjusted two-proportion z-test p = 0.021.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Peer Support Arm (Intervention) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes extend beyond HbA1c to body weight, participation in formal prevention programs, session attendance, physical activity, sugary soda consumption, patient activation, and perceived confidence. The results span different types of measures, so they do not form one single summary of benefit or harm. The secondary pattern is not uniformly aligned with a change in HbA1c: the available results include a signal in formal program enrollment, while the continuous outcomes do not show a clear difference in the posted unadjusted comparisons.

These outcomes can help describe whether the intervention may have affected engagement or behavior even when the primary biomarker comparison is inconclusive. They do not overturn the primary result, which remains the central outcome for the trial’s main question. A secondary change in program enrollment, for example, would address uptake of a formal program rather than directly demonstrate a difference in HbA1c.

Unadjusted secondary results carry less weight because several outcomes are examined. Without a reported multiplicity plan, the registry does not establish how the analysis addressed the increased chance of false-positive findings across multiple comparisons. Secondary outcomes may also differ in completeness, measurement, and clinical meaning. Their results are best regarded as context for the primary finding and as potential evidence to evaluate in further research, rather than as independent confirmation of an overall effect.

The registry lists 22 posted secondary outcomes, but the detailed numerical results considered in this analysis cover only a subset of the listed outcome domains. The record does not report a single combined secondary-outcome analysis. Conclusions should therefore remain outcome-specific and should avoid treating the broad list of measures as though it were a single validated endpoint.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Change in Glycosylated Hemoglobin A1c (HbA1c) at 12 Months
• Change in Body Weight at 12 Months
• Change in Whether Participant Enrolled in a Formal Program to Prevent Diabetes
• Change in Number of Sessions Participant Attended in a Formal Program to Prevent Diabetes
• Change in Frequency and Duration Participant Engages in Moderate to Vigorous Physical Activity as Shown by Minutes Per Week
• Change in the Number of Times Drinking Regular Soda or Pop That Contains Sugar
• Change in Patient Activation as Measured by the Patient Activation Measure 13 (PAM-13)
• Change in Participant's Perceived Confidence in Their Ability to Take Steps to Prevent Diabetes Using the Williams Perceived Competence Scale

9. Statistical Interpretation

10. Statistical Methods Explained

What does the mean change represent?

HbA1c is continuous, and the registry reports mean change and standard deviation for each arm. The mean is the arithmetic average of the observed changes, while the standard deviation describes how dispersed those changes are around the average. The negative group means indicate decreases from baseline, according to the registry’s endpoint description. They do not mean that participants had negative HbA1c values.

The difference in means compares the average change in peer support with the average change in enhanced usual care. Its sign depends on the order of subtraction. The posted estimate of -0.01 is reported with peer support as the intervention comparison; it indicates a slightly more negative mean change in that group. The estimate is not an adjusted model coefficient and should not be confused with the sponsor’s mixed-model result.

Why report a confidence interval?

A confidence interval communicates uncertainty around an estimated contrast. It is more informative than the point estimate alone because it shows which values remain compatible with the observed data under the method used. For the primary difference, the interval crosses zero. Thus, the summaries do not distinguish a small difference in either direction from no difference with sufficient precision. The interval does not assign probabilities to individual possible effects, and it does not establish that the true effect must lie within its endpoints.

What does the unadjusted comparison test?

The recalculated p-value is based on a simple, unadjusted comparison of the posted group means and standard deviations. A p-value describes how compatible the observed test statistic is with a null model of no difference, subject to the assumptions of the test. It is not the probability that the null hypothesis is true, and it does not measure the size or practical importance of an effect. The recalculated p-value of 0.698 is not evidence that the groups are exactly equivalent.

The registry separately reports a mixed-model analysis for the primary outcome. Mixed models can use a model-based approach to repeated or otherwise structured observations, depending on how the analysis is specified. The registry record does not report the model formula, covariance structure, adjustment variables, or detailed handling of missing data. It is therefore not possible to reproduce that model from the posted summaries or to identify precisely why its p-value differs from the unadjusted recalculation.

Why can variability affect precision?

When individual outcomes vary widely relative to the difference between group means, the estimated contrast is less precise. The standard deviations describe that variability. A confidence interval combines information about the group summaries and sample sizes with assumptions about the comparison method. Larger samples can generally improve precision, while greater variability can reduce it. These are general statistical principles; they do not establish the trial’s original design rationale.

Over-dispersion is a concern mainly for outcomes whose variance is greater than a chosen model expects, especially count outcomes. If a model underestimates that variance, standard errors may be too small and uncertainty may be understated. The primary endpoint is continuous, and the registry does not report an over-dispersion assessment. For the behavioral and participation outcomes, the appropriate variance model depends on the outcome type and analysis specification; those details are not reported in the registry record.

How should the secondary comparisons be treated?

The registry includes outcomes with different measurement scales, including continuous measures and a binary indicator of enrollment in a formal program. A mean difference is suitable for comparing continuous outcomes, while a risk difference compares the proportion with an event or status between groups. These effect measures answer different questions and should not be compared as though they were on a common scale.

Some secondary comparisons have unadjusted p-values and intervals. Unadjusted results do not account for the fact that several outcomes are examined. As the number of comparisons grows, the chance of at least one apparently notable finding arising by chance can increase. The registry does not report a multiplicity strategy, so secondary findings should be treated as supportive or exploratory rather than as standalone confirmation.

All recalculations are simple, unadjusted analyses based on posted summary statistics. They do not reconstruct the trial’s full analysis plan, which is not reported in the registry record. The sponsor’s reported mixed-model analysis is the trial’s own analysis and takes precedence when describing the sponsor’s inferential result.

11. Confidence Intervals and Precision

The primary estimate is -0.01 for the difference in mean HbA1c change between peer support and enhanced usual care. Its confidence interval extends across zero. That pattern means the posted results do not identify a clear direction of difference with precision: the data remain compatible with a modest advantage for either group and with no difference. The interval should be considered alongside the point estimate, rather than replaced by a binary reading of a significance threshold.

The observed group means, -0.07 and -0.06, are close together, while the standard deviations are 0.2 and 0.22. The spread of individual changes is therefore relevant to interpreting the small difference in averages. A mean can conceal variation: participants within the same arm may have had changes that differed substantially from the group average. The registry does not provide the participant-level distribution, so the summaries cannot describe its shape or identify subgroups.

Precision is also affected by how much outcome information is available. The registry reports 135 participants with primary outcome data in the intervention arm and 132 in the control arm, and notes unavailable data for 23 and 37 participants, respectively. The reported confidence interval reflects the analysis used to produce it, but the registry does not explain in detail how missing observations were handled. If missingness is related to outcomes, estimates based only on available data may not represent the full randomized groups.

There is no basis for interpreting the result as evidence of equivalence. Equivalence requires a prespecified margin and an analysis designed to assess whether differences fall within that margin; neither is reported in the registry record. The appropriate conclusion is narrower: the estimated contrast is small, its interval includes zero, and the available result does not establish a between-group difference.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are participants per arm for 80% power and participants per arm for 90% power. They are planning illustrations based on the observed effect, not the trial’s own original sample-size calculation. The registry record does not report the original design assumptions or sample-size justification.

The illustration assumes a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It considers detecting the stated fraction of the observed effect. The rows for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect represent progressively smaller target effects. Smaller effects need more participants to detect with the same power, all else being equal.

The 80% power column describes a planning target in which a study designed under the stated assumptions would have an 80% chance of detecting the specified effect using the chosen test, if that effect were present and the assumptions held. The 90% column uses a higher chance of detection and consequently requires more participants for the same target effect. These are design concepts, not guarantees about the result of a particular study.

For the primary comparison, the illustration gives 6,931 participants per arm for planning around 100% of the observed effect at 80% power and 9,279 participants per arm at 90% power. For 75% of the observed effect, the figures are 12,322 and 16,496 participants per arm; for 50%, they are 27,724 and 37,115 participants per arm. These large illustrations reflect the small observed primary difference under the stated assumptions. They should not be interpreted as a judgment about the trial’s original design. This is not observed power.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -0.010; pooled standard deviation σ = 0.210.
2. n = 2 × (1.96 + 0.842)² × 0.210² / -0.010² = 6,930.9, rounded up to 6,931 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- -0.01                      # observed difference in means
sigma <- 0.210124850577433                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(3689530)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 6,931 (80% power), 9,279 (90% power)
 75% of observed effect: n per arm = 12,322 (80% power), 16,496 (90% power)
 50% of observed effect: n per arm = 27,724 (80% power), 37,115 (90% power)
power.t.test: 6931.9 
Simulated power at n = 6,931 per arm: 0.798 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The registry reports the sponsor’s primary comparison as a mixed-model analysis with a p-value of 0.744. The simple recalculation from the posted group summaries gives an unadjusted p-value of 0.698. These values differ because the sponsor’s model-based analysis and the summary-statistic recalculation are not the same analysis. The model’s specification and the details needed to reproduce it are not reported in the registry record.

Both analyses support the same broad reading: neither indicates clear statistical evidence of a difference in the primary HbA1c outcome at 6 months. The recalculation is useful for interpreting the posted means, variability, and contrast, but it should not replace the trial’s own analysis when describing the sponsor’s result. It is an unadjusted calculation from summary statistics, not a reconstruction of the full mixed model.

The registry does not report an adjusted effect estimate or confidence interval for the mixed-model result, nor does it provide enough model detail to compare the two methods point by point. The sponsor’s reported analysis therefore takes precedence for the trial’s inferential result; the recalculation provides a transparent, limited comparison based on the group summaries.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary HbA1c outcome and many secondary outcomes. Testing multiple outcomes creates a multiplicity issue: even if there were no true differences, a larger set of comparisons can increase the chance that at least one result appears statistically notable by chance. The registry does not report whether the trial used a formal adjustment, a hierarchy of outcomes, or another strategy to address multiplicity. The unadjusted secondary comparisons should therefore be interpreted cautiously.

The record notes unavailable primary-outcome data for 23 participants in the intervention arm and 37 in the control arm, attributing this to scheduling issues and COVID. It does not report whether those observations were missing at random, whether other missing-data reasons occurred, or how the analysis handled missing measurements. Missingness can affect both the estimate and its uncertainty if participants without measurements differ systematically from those with measurements.

The registry reports group-specific numbers contributing to the posted primary summaries, but it does not provide a full account of the analysis population, exclusions after randomization, or whether the primary analysis followed an intention-to-treat principle. It also does not report sensitivity analyses that test how conclusions might change under different assumptions about missing data. These omissions limit assessment of robustness; they do not by themselves show that the analysis was biased.

Likewise, detailed information about the original statistical analysis plan, any interim analyses, and the prespecified multiplicity approach is not reported in the registry record. The reported mixed-model analysis is identified, but its specification and missing-data assumptions are not described. The available evidence supports a cautious reading of the posted comparisons, with particular care around secondary p-values and incomplete outcome data.

16. 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 5% reporting threshold.

The registry reports serious adverse events and other adverse events by arm. In enhanced usual care, 0 serious adverse events and 0 other adverse events were recorded among 178 participants at risk. In peer support, 0 serious adverse events and 0 other adverse events were recorded among 177 participants at risk. The registry also records 0 deaths in enhanced usual care and 1 death in peer support.

These figures describe events recorded in the registry and should not be interpreted as proof that an intervention caused or prevented an event. The record does not provide a serious-event term or further detail about the reported death, including its cause or relationship to the intervention. The “at risk” counts are the registry’s denominators for these safety summaries and should not be assumed to match the numbers contributing to the primary outcome.

Zero recorded adverse events in a category does not establish that the risk is absent. Safety interpretation depends on event definitions, duration and completeness of follow-up, and how events were collected, details that are not reported. The figures are descriptive and do not support a causal conclusion about comparative safety.

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

The primary result describes the average change in HbA1c among participants with reported outcome data and compares that change between peer support and enhanced usual care. The group means decreased by similar amounts, and the estimated difference was small. The interval includes zero, so the posted evidence does not establish a clear difference in HbA1c change at 6 months.

This is not the same as showing that the interventions are equivalent or that peer support has no possible effect. A nonsignificant comparison can arise when the true difference is small, when estimates are imprecise, or both. The confidence interval indicates uncertainty, while the sample-size illustrations show how challenging it would be to plan for small effects under the stated assumptions. Neither converts the observed result into proof of no effect.

The findings also do not answer every question about the intervention. HbA1c is one health measure, while the intervention was intended to support engagement with prevention activities and health-care navigation. The registry lists behavioral and patient-reported secondary outcomes, but those outcomes do not replace the primary endpoint and should be interpreted with attention to multiplicity and missing information.

The result applies to this randomized comparison as recorded, with the available details about the intervention, comparator, outcome, and follow-up. The registry does not provide a full eligibility description or enough demographic detail to determine how well the results generalize to every group described in the study rationale. It also does not report the model specifications or complete missing-data strategy. Those limitations constrain broader claims.

In short, the posted primary data show similar average HbA1c changes in the two arms, with uncertainty that does not establish a between-group difference. They neither demonstrate a definitive advantage nor rule out all potentially relevant effects. The secondary outcomes and safety counts add context, but they should not be combined into a stronger conclusion than the record supports.

18. Why This Trial Matters Statistically

This trial addresses an important statistical challenge in behavioral research: an intervention can aim to change engagement, access, and sustained behavior, while the primary endpoint is a clinical marker measured over time. A sound interpretation keeps those constructs distinct. HbA1c, formal program enrollment, and patient-reported confidence do not measure the same thing, even when they are part of a common prevention strategy.

The randomized comparison provides a framework for evaluating peer support against enhanced usual care, but the primary estimate is small relative to the observed variation and is not precise enough to establish a difference. This illustrates why a point estimate alone is inadequate. The interval and the amount of missing outcome information are essential to understanding how much the observed comparison can support.

The trial also illustrates the limits of interpreting a broad set of secondary outcomes. A behavioral signal may be relevant to the intervention’s proposed pathway, yet it does not automatically imply a change in a biomarker. Multiple comparisons further mean that isolated secondary findings need cautious interpretation unless the analysis plan clearly addresses them.

Finally, the planning illustrations connect observed effects to future study design without claiming that the trial had a particular power or that a result is more likely because of its sample size. Small effects require larger studies to estimate or detect reliably under conventional planning assumptions. That lesson is useful for designing subsequent research, while the original design rationale remains unreported in the registry record.

19. 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 primary-outcome data were unavailable for 23 intervention participants and 37 control participants because of scheduling issues and COVID. The registry does not describe the characteristics or outcomes of those participants, the missing-data method, or sensitivity analyses. The possibility that missingness affects the observed comparison therefore cannot be assessed fully.

Although the registry identifies a mixed-model analysis, it does not report the model formula, covariates, repeated-measures structure, or assumptions. This limits independent reproduction and makes it difficult to determine exactly how the sponsor’s result relates to the unadjusted comparison. The original sample-size justification and any interim analyses are also not reported in the registry record.

The registry’s single masking designation is not explained, and it does not specify which participants or study personnel were masked. For a behavioral intervention, knowledge of group assignment may be relevant to engagement and self-reported outcomes, but the record does not provide enough detail to evaluate this issue.

The registry’s study rationale emphasizes low-income Latino and African American adults, but the detailed participant characteristics and full eligibility criteria are not reported in the registry record. Generalizability to populations beyond those enrolled cannot be assessed from the outcome summaries alone. The listed secondary outcomes are numerous, and no multiplicity strategy is reported. Finally, the adverse-event record is concise and does not provide event descriptions or attribution details. These limitations affect interpretation; they do not establish that the trial conduct or analysis was flawed.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Using Peer Support to Aid in Prevention and Treatment in Prediabetes — NCT03689530.
• Registry record last updated 2023-03-03; status last verified 2023-02.
• Results first posted 2023-03-03.
• 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.

23. Record Summary

UPSTART was a completed, randomized, parallel-group behavioral trial sponsored by the University of Michigan. It compared peer support, in which participants were matched with a peer supporter, with enhanced usual care consisting of brief education and information and resources. The registry records enrollment of 450 and a single masking designation.

The primary outcome was mean change in HbA1c from baseline to 6 months. The posted means were -0.07 for peer support and -0.06 for enhanced usual care; the unadjusted difference was -0.01, with an interval spanning zero. The sponsor reported a mixed-model analysis that likewise did not establish clear evidence of superiority.

Primary-outcome data were unavailable for 23 intervention participants and 37 control participants because of scheduling issues and COVID. The registry lists secondary outcomes across clinical, behavioral, and patient-reported domains. Safety summaries record no serious or other adverse events in either arm and 1 death in peer support, with no death in enhanced usual care. Details about missing-data handling, multiplicity, and the full analysis plan are not reported in the registry record.