CTSI Pilot: Improving Adherence to Diabetic Diet: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for CTSI Pilot: Improving Adherence to Diabetic Diet (NCT04051008), a randomized na study. The primary outcome is Calories Purchased.

2. Clinical Question

3. Trial Design

The registry describes a completed, randomized, single-masked, parallel-group study with an enrollment of 65. Participants were assigned to Control, Online, or Default. The record describes a behavioral intervention rather than a medication or device trial. The study's stated goal was to test approaches informed by behavioral economics to encourage healthier grocery shopping among people with diabetes or at risk of diabetes.

Control participants shopped in person as usual and received recipe cards based on the DASH diet and diabetes dietary recommendations. Online participants shopped at a local grocery store using an online platform and also received the recipe cards. Default participants used the online platform with a pre-filled cart intended to match diabetic-diet goals and support recipes from the provided materials; they could change its contents. The registry's arm description for the Default group ends before fully describing the materials, so further details are not reported in the registry record.

The primary outcome was calories purchased at post-test, identified as week 5. Receipt data from target grocery stores were entered into Nutritionist Pro software to calculate total calories purchased. The primary endpoint counts were 19 in Control, 19 in Online, and 22 in Default. These counts reflect participants who posted week-5 receipt data, as described in the outcome definition. The registry does not report why these counts differ from the overall enrollment or whether all randomized participants were included in any particular analysis population.

The meaning of single masking is not described in detail. Who was masked, how masking was maintained, and the original design's statistical assumptions 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: The sample sizes reflect the number of participants in each group who provided receipt data during Week 5 of the 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.

Population: The sample size for the baseline analysis population is different from the Participant Flow information because 1 participant did not have usable baseline survey data. Therefore there are n=64 baseline participants included in this section and n=65 randomized participants overall.

6. Primary Endpoint

The registry lists the primary outcome as Calories Purchased, measured over At post-test (week 5).

The primary endpoint is total calories purchased from target grocery stores at post-test, identified as week 5. Receipts were entered into Nutritionist Pro software and used to calculate calories. The posted group summaries are means with standard deviations, so the endpoint is treated as a continuous measure: each participant's receipt information contributes to a numerical total, and the group mean describes the average among participants with recorded week-5 receipt data.

Calories purchased were selected to reflect shopping behavior, which is the direct focus of these interventions. The registry's rationale is that online shopping and default choices may shape grocery selection, while receipts can quantify what was purchased. This is a measure of purchases, not a direct measure of what participants ate. The registry lists separate secondary outcomes concerning dollars spent and nutritional quality of foods consumed, but the primary endpoint itself should not be interpreted as a complete assessment of diet or health.

The registry does not describe how receipt data were handled when information was incomplete, how products were matched to the software database, or how purchases outside target grocery stores were addressed. These details are not reported in the registry record, so the endpoint's scope is limited to the stated receipt-based calculation.

7. Results Posted in the Registry

• Group 2 - Online: mean 25,339 versus 24,443 on control; difference 895.90 (95% CI -9,991 to 11,783), standardized effect d = 0.05; z-test p = 0.872.
• Group 3 - Default: mean 18,527 versus 24,443 on control; difference -5,916 (95% CI -15,104 to 3,272), standardized effect d = -0.40; z-test p = 0.207.

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.

Dollars Spent

• 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:
  • Group 2 - Online: mean 88.79 versus 84.71 on control; difference 4.08 (95% CI -30.74 to 38.90), standardized effect d = 0.07; z-test p = 0.818.
  • Group 3 - Default: mean 72.59 versus 84.71 on control; difference -12.12 (95% CI -39.66 to 15.42), standardized effect d = -0.27; z-test p = 0.388.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Group 2 - Online and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Group 3 - Default and the control. The interval gives the range of effects that remain compatible with the data.

Change in Dollars Spent

• 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 errors.
• Result:
  • Group 2 - Online: mean -8.65 versus -7.20 on control; difference -1.45 (95% CI -34.07 to 31.17), standardized effect d = -0.03; z-test p = 0.931.
  • Group 3 - Default: mean -24.72 versus -7.20 on control; difference -17.52 (95% CI -49.23 to 14.19), standardized effect d = -0.34; z-test p = 0.279.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Group 2 - Online and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Group 3 - Default and the control. The interval gives the range of effects that remain compatible with the data.

Nutritional Quality of Foods Consumed

• 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 errors.
• Result:
  • Group 2 - Online: mean 21.13 versus 20.99 on control; difference 0.14 (95% CI -3.56 to 3.84), standardized effect d = 0.02; z-test p = 0.941.
  • Group 3 - Default: mean 22.00 versus 20.99 on control; difference 1.01 (95% CI -2.66 to 4.68), standardized effect d = 0.18; z-test p = 0.590.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Group 2 - Online and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Group 3 - Default and the control. The interval gives the range of effects that remain compatible with the data.

Across the posted secondary outcomes, the point estimates do not all move in the same way. The Online group is close to Control on spending and nutritional quality, while the Default group has lower spending estimates and a higher nutritional-quality estimate. The change-in-spending results also point toward a larger decrease for Default than for the other groups. Taken together, these directions are broadly consistent with the possibility that the default-cart approach influenced shopping patterns, but they are not a uniform or conclusive pattern.

That overall picture is relevant to the primary result because the outcomes describe related aspects of grocery purchasing and food quality. It does not resolve the uncertainty in calories purchased: each endpoint measures something different, and none should be treated as a substitute for the primary endpoint. The secondary findings may help identify outcomes and design questions for future studies, but the small groups and uncertain estimates limit how much can be inferred from their pattern.

The registry gives unadjusted comparisons for the secondary outcomes and does not report secondary analyses in its analysis records. When many outcomes and comparisons are examined, some estimates may appear notable by chance even if there is no underlying effect. Without a reported multiplicity strategy, the secondary p-values and intervals do not provide the same evidential footing as a prespecified, multiplicity-controlled confirmatory analysis. They are best regarded as exploratory signals that require further evaluation, not independent confirmation of the primary result.

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

• Dollars Spent
• Dollars Spent
• Change in Dollars Spent
• Nutritional Quality of Foods Consumed

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is continuous, and the registry reports a mean and standard deviation for each group. A mean describes the average recorded calorie total among participants with receipt data. A standard deviation describes how spread out individual observations are around that average; it is not the same as the uncertainty in the estimated mean. The reported group standard deviations are substantial, which helps explain why the comparisons are imprecise despite the point estimates being easy to state.

What does a mean difference represent?

Each posted comparison expresses the intervention group's mean relative to the Control mean. A positive difference means the intervention group's observed mean is higher; a negative difference means it is lower. For calories purchased, the Online estimate is positive and the Default estimate is negative. These are directions in the observed data, not proof of a causal effect. Random allocation can support a causal interpretation when the design and analysis are appropriate, but the registry record does not give enough detail about the analysis population, missing data, or full statistical plan to assess those conditions.

How should the confidence interval be read?

A confidence interval expresses uncertainty around an estimated difference. In general, a wider interval indicates less precision. An interval that includes zero is compatible with no difference under the calculation used, while also allowing values away from zero. For both primary comparisons, the posted intervals include zero and cover differences in both directions. The ranges should therefore be considered alongside the point estimates rather than replaced by a simple significant-or-not label. The intervals are posted as 95% confidence intervals; the registry does not report whether they were adjusted for multiple comparisons.

What does an unadjusted p-value say?

A p-value measures how compatible the observed data are with a specified no-difference model, under the assumptions of the calculation. It is not the probability that the null hypothesis is true, nor does it measure the size or importance of a difference. The posted p-values are unadjusted. Because the registry lists multiple outcomes and comparisons and does not report a multiplicity procedure, these p-values should be treated as exploratory rather than as a familywise confirmatory analysis.

Why do group sizes and variation matter?

Uncertainty generally increases when sample sizes are small or observations vary widely. Here, the primary endpoint uses receipt data from 19 Control participants, 19 Online participants, and 22 Default participants, with sizeable within-group standard deviations. Those features are consistent with broad confidence intervals. The registry does not report a statistical model that accounts for any additional structure in the observations, and the posted primary comparisons are not described as adjusted model estimates.

What are standardized effects and planning illustrations?

The comparisons also include standardized effect-size values. A standardized effect expresses a mean difference relative to a pooled estimate of variability, putting the difference on a scale that is less dependent on the endpoint's measurement units. It can help describe magnitude, but it does not remove uncertainty or establish that the observed effect will recur. The planning illustrations use fractions of the observed effects to show how participant requirements change under stated assumptions. Such figures depend strongly on the effect chosen and on assumptions about allocation, significance level, and the calculation method; they are not the trial's original sample-size justification.

The registry does not report whether the original trial had a formal sample-size calculation, its design assumptions, an interim analysis plan, or a prespecified model for the primary outcome. Those parts of the statistical plan are not reported in the registry record. They should not be inferred from the planning illustrations or from the posted p-values.

11. Confidence Intervals and Precision

The primary comparisons are imprecise. For Online versus Control, the estimated difference is 895.9 Kcal, with a 95% confidence interval from -9991.0 to 11780.0. For Default versus Control, the estimate is -5916.0 Kcal, with a 95% confidence interval from -15100.0 to 3272.0. Each interval covers zero and spans values in both directions, so the results do not pin down whether mean purchases were higher or lower in either intervention group.

A confidence interval is not a range containing most individual participants' outcomes. It represents uncertainty in the estimated group difference under the calculation used. The observed standard deviations describe participant-to-participant variation, while the confidence intervals describe uncertainty about the mean difference. Both sources matter: the participant counts are modest, and the outcome varies substantially across participants.

The interval widths make clear that the point estimates alone are not a complete description. In particular, a point estimate near zero does not mean the true difference must be near zero when the interval is broad. Conversely, an estimate favoring fewer calories in the Default group does not establish that this difference is reliable. The registry record does not report an adjusted confidence interval or a method accounting for multiple comparisons, so the posted intervals should be interpreted as unadjusted comparisons.

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, not total events. The calculation uses a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It explores detecting 100%, 75%, or 50% of the observed effect. These assumptions describe the planning illustration, not the original trial's design; the trial's own sample-size assumptions are not reported in the registry record.

The 80% power column indicates the number of participants per arm that the stated planning calculation estimates would be needed to have an 80% chance of detecting the specified effect under its assumptions. The 90% power column gives the corresponding requirement for a 90% chance. These are conditional planning quantities, not guarantees: their interpretation depends on the assumed effect, variability, allocation, significance level, and analysis approximation.

Read each row as planning for a particular fraction of the observed effect. Planning for 75% or 50% of the observed effect uses a smaller target difference than planning for the full observed effect. Smaller target effects are harder to distinguish from random variation, so they require more participants under the same assumptions. The figures can be very sensitive to the observed effect, particularly when that effect is estimated from a small pilot sample.

These planning calculations are not observed power. They do not tell whether the completed study had adequate power after results were known, and they are not evidence that the original design used these assumptions. The registry record does not report the original sample-size justification.

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 = 895.900; pooled standard deviation σ = 17120.206.
2. n = 2 × (1.96 + 0.842)² × 17120.206² / 895.900² = 5,732.4, rounded up to 5,733 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 <- 895.900000000001                      # observed difference in means
sigma <- 17120.2063674536                     # 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(4051008)
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 = 5,733 (80% power), 7,675 (90% power)
 75% of observed effect: n per arm = 10,191 (80% power), 13,643 (90% power)
 50% of observed effect: n per arm = 22,930 (80% power), 30,697 (90% power)
power.t.test: 5733.4 
Simulated power at n = 5,733 per arm: 0.800 (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. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary outcome, multiple secondary outcomes, and comparisons of each intervention group with Control. It does not report a multiplicity plan describing how the chance of false-positive findings across those outcomes and comparisons was addressed. The posted p-values are unadjusted, and the confidence intervals are not identified as multiplicity-adjusted. When many hypotheses are considered, interpreting each result as though it were the only test can overstate the strength of evidence. This matters especially for secondary outcomes, which are useful for exploration but should not be treated as confirmatory without a prespecified analysis strategy.

The registry record reports enrollment of 65, while the primary endpoint's week-5 receipt-data counts are 19 for Control, 19 for Online, and 22 for Default. It does not report reasons why a participant may lack receipt data, the amount or pattern of missing data, or whether missingness differed by group. It also does not report imputation methods, sensitivity analyses for missing outcomes, or whether the primary analysis followed intention-to-treat principles. These details are not reported in the registry record, so the potential effect of missing data cannot be evaluated from the posted results.

The record does not specify a detailed analysis population for the primary outcome or explain whether all randomized participants were analyzed according to their assigned groups. It also does not report an interim analysis plan, adjustment covariates, or a prespecified statistical model. The absence of those details in the registry does not show that the study lacked such procedures; it means they cannot be established from this record. The posted results should therefore be read with these reporting limitations in mind.

15. 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.

The adverse-event record lists 21 participants at risk in Control, 21 in Online, and 23 in Default. It records zero serious adverse events and zero other adverse events in each group, and zero deaths in each group. These counts describe what is recorded for the listed groups; they do not establish that no events occurred outside the registry's reporting process or that the interventions have no risks.

Safety counts should be interpreted in light of the study size and the event information recorded. The registry does not list serious adverse-event terms, and the posted serious-event counts are zero. It does not provide narrative detail about how events were collected, how long participants were monitored for adverse events, or how events were defined. Those details are not reported in the registry record. The figures therefore describe the recorded event counts but cannot support broad conclusions about safety.

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

The results describe differences in average receipt-based calorie purchases at post-test among participants who provided week-5 receipt data. The Default point estimate is lower than Control, while the Online point estimate is slightly higher. However, both unadjusted confidence intervals include zero and are broad. The data therefore leave substantial uncertainty about the direction and size of the underlying group differences.

These findings do not show that the groups were equivalent. A failure to establish a clear difference is not proof that the true difference is zero, particularly with small endpoint groups and broad intervals. Nor does a point estimate in the anticipated direction establish that an intervention caused a change. Interpretation depends on the design, the completeness of outcome data, and the analysis plan, details that are not fully reported in the registry record.

The measured primary outcome is calories purchased from target grocery stores, not calories consumed, dietary adherence in all settings, blood glucose, weight, or another clinical outcome. The registry also reports secondary measures, but those do not transform the primary result into evidence of a health benefit. The stated behavioral rationale gives a reason to study the shopping environment; the posted results address purchasing measures at the specified study time.

The trial is described as a pilot, and its results can inform questions for further research: whether receipt collection is feasible, how the groups' shopping measures vary, and what outcomes merit more precise evaluation. The registry record does not report a definitive confirmatory analysis or a basis for making clinical recommendations. The findings should be understood as an uncertain estimate of shopping behavior in this study, rather than a conclusion about treatment effectiveness or individual care.

17. Why This Trial Matters Statistically

This study is statistically instructive because its intervention targets a behavior in a real-world purchasing environment, while its primary outcome is a quantitative measure derived from receipts. The design makes clear that outcome definition matters: calories purchased is not the same as calories consumed, and a shopping intervention's effect on one measure may not translate directly to another.

The three-group structure also distinguishes the online-shopping component from the added default-cart component. The Control group shopped in person as usual, while both intervention groups used online shopping and the Default group additionally encountered a pre-filled cart. That structure motivates comparisons that ask whether the two approaches show different patterns, but the posted results compare each intervention group with Control and do not report a direct Online-versus-Default comparison.

The pilot-scale endpoint counts and wide confidence intervals show why effect estimates must be read together with precision. Point estimates are informative descriptions of this sample, but uncertainty is substantial. Planning illustrations can help readers understand how assumptions about a target effect affect participant requirements; they do not replace the original design justification, which is not reported in the registry record.

Finally, the multiple secondary measures illustrate the importance of prespecifying outcomes and addressing multiplicity. Related outcomes can provide context, but unadjusted exploratory results should not be mistaken for several independent confirmations. The study raises useful statistical questions about outcome measurement, missing receipt data, and future design, while leaving important analysis details unreported.

18. 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 are specific to interpreting this record. The primary endpoint is based on receipts from target grocery stores, so it captures only the purchases described by that measure. The registry does not report how shopping outside those stores was handled, how receipt records were validated, or whether the nutritional software's product matching was checked. Purchases also do not directly establish what participants consumed.

The week-5 endpoint counts are smaller than the overall enrollment, and reasons for the difference are not reported in the registry record. Without information about missing receipts, it is not possible to assess whether participants with available data differed from those without them. The record also does not report whether the analyses followed assigned groups or included all randomized participants.

The primary group standard deviations are large, and the observed groups are small, limiting precision. The record does not describe the original sample-size justification or statistical plan, including model specification, planned covariate adjustment, or any interim analyses. The planning illustrations are explicitly based on observed effects and stated calculation assumptions; they cannot establish what the original design anticipated.

Finally, the record does not report a multiplicity strategy despite multiple outcomes and comparisons. Secondary findings are therefore exploratory in interpretation. Details of masking, intervention fidelity, and the completeness of the Default arm description are also limited in the record. These reporting gaps constrain assessment of bias and generalizability but do not, by themselves, demonstrate that a particular bias occurred.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: CTSI Pilot: Improving Adherence to Diabetic Diet — NCT04051008.
• Registry record last updated 2023-12-13; status last verified 2023-11.
• Results first posted 2023-12-13.
• 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.

22. Record Summary

The registry describes a completed randomized, single-masked, parallel pilot study with enrollment of 65. Adults with type 2 diabetes or at risk of developing it were assigned to usual in-person shopping with recipe cards, online grocery shopping with recipe cards, or online shopping with a pre-filled default cart.

The primary endpoint was total calories purchased from target grocery stores at post-test in week 5, calculated from receipt data. The posted comparisons show uncertain differences between each intervention group and Control: both 95% confidence intervals include zero. Secondary outcomes provide additional exploratory measures of spending and nutritional quality, but the record does not report a multiplicity plan or detailed analysis population. The posted results are useful as pilot estimates; important design and analysis details are not reported in the registry record.