PortionSize Study 3: Assessment in Free-living Conditions: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for PortionSize Study 3: Assessment in Free-living Conditions (NCT05906160), a registered phase not stated study. The primary outcome is Energy (Kilocalorie or kcal) Intake.

2. Clinical Question

3. Trial Design

PortionSize Study 3 is registered as a completed study of app-based energy-intake assessment in free-living conditions. The registry lists an enrollment of 63, a start date of 2023-06-22, and a primary completion date of 2024-11-25. Pennington Biomedical Research Center is named as the sponsor, and Dietary Habits is the listed condition.

The study description says participants used both PortionSize and MyFitnessPal in separate periods. It describes the goal as testing the apps’ accuracy against doubly labeled water, characterized as the gold-standard criterion measure. The recorded endpoint timeframe is 4 days for each app and an overlapping 12 days for DLW. The registry says outcome measures are summarized by app, pooling participants regardless of the order in which they used the apps.

This is not described as a randomized, like-for-like comparison between separate treatment and control groups. Allocation is not stated, and the registry record’s comparison note cautions that the arms are not a like-for-like control comparison. The app summaries should therefore be read as assessments of each method against the criterion measure, not as an estimate of the effect of assigning one app instead of the other.

The record lists 1 outcome measure and 0 analyses. Although the study status is COMPLETED, completion status does not itself indicate that results were posted. The record does not report participant characteristics, details of the app-use sequence, or a fuller account of allocation.

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.

Period 1

Period 2

Analysis population for the primary outcome, as defined in the registry: All participants completed two study periods using both the PortionSize and MyFitnessPal apps. Outcome measures are summarized by intervention (PortionSize vs. MyFitnessPal), pooling data across all participants regardless of the order in which apps were used. One participant allocated to the PortionSize arm was removed from analysis due to an unusable DLW sample.

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 Energy (Kilocalorie or kcal) Intake, measured over 4 days for each app (PortionSize and MyFitnessPal) and an overlapping 12 days for DLW.

The primary endpoint is mean energy intake, expressed in kcal/day. For each app, the registry describes a mean difference between energy intake measured by the app and the estimate from doubly labeled water (DLW). The stated subtraction is DLW mean daily energy intake subtracted from the mean measured by the app. Under this convention, a negative difference indicates that the app estimate is lower than the DLW estimate; a positive difference indicates that it is higher.

Comparing an app estimate with a criterion measure addresses agreement: whether the app’s mean estimate is sufficiently close to the DLW estimate under the planned definition of equivalence. This is different from asking whether the apps differ from each other, or whether app use changes energy intake. The registry describes the endpoint as a continuous measure summarized by a mean and standard deviation.

The timeframe is 4 days for each app, with DLW measured over an overlapping 12 days. The record does not report endpoint values, standard deviations, confidence intervals, or the pre-specified error bounds used to judge equivalence. Without those results and bounds, the degree of agreement cannot be assessed from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does equivalence testing ask?

General explanation: An equivalence test asks whether the difference between two measurement methods is small enough to fall within an acceptable range set in advance. The question is not simply whether a difference can be detected. Instead, the analysis assesses whether the data support a difference bounded closely enough around zero to satisfy the chosen criterion.

The registry describes TOST, the Two One-side T-test method. In general, this approach evaluates the difference against both sides of the acceptable range: whether it is greater than the lower bound and whether it is less than the upper bound. Equivalence is supported only when both one-sided tests support being within the bounds. The registry states the null hypothesis as nonequivalence, with the alternative hypothesis being equivalence. It does not report the pre-specified error bounds or the test results.

How should the mean difference be read?

The outcome is a continuous measure: mean energy intake in kcal/day. The registry defines the app-minus-criterion direction by subtracting the DLW mean daily estimate from the app mean. Under that definition, a negative value means the app estimate is lower than the DLW estimate, while a positive value means it is higher. A difference near zero would indicate little average directional difference, but whether it qualifies as equivalent depends on the pre-specified bounds and the uncertainty around the estimate.

General explanation: A mean describes the average level in the analyzed observations. A standard deviation describes how spread out individual observations are around that mean. The standard deviation is not the same as the uncertainty in the estimated mean difference. Assessing uncertainty in a mean difference generally requires information about the analysis and the variability of the relevant observations. The record identifies the dispersion measure as standard deviation but supplies no endpoint summary statistics.

What role does a confidence interval play?

General explanation: A confidence interval expresses uncertainty around an estimated quantity under the assumptions of the statistical method. For equivalence, the interval is assessed against the acceptable lower and upper bounds. An interval that extends outside a bound does not establish that the difference is sufficiently small, even if the point estimate is close to zero. The registry record does not report a confidence interval or its confidence level, so this assessment cannot be made.

How do the measurement periods affect interpretation?

The registry gives a timeframe of 4 days for each app and an overlapping 12 days for DLW. These periods are not described as identical in length. General explanation: When measurements from methods cover different periods, the analysis needs to define how the measurements correspond and how the difference is estimated. Such choices affect what the resulting mean difference represents. The registry record does not describe those details, so no particular alignment or averaging procedure should be assumed.

The outcome description says all participants completed two study periods using both apps, and that summaries are pooled regardless of app order. This supports reading each app summary as an assessment across participants rather than as a comparison of separate randomized groups. It does not reveal whether the order was balanced, whether order was accounted for analytically, or whether period effects were assessed; those details are not reported in the registry record.

What do the planning illustrations represent?

General explanation: A sample-size calculation asks how many observations a future study might need under stated assumptions. It is not a description of the observed precision or a substitute for the actual results. The registry’s planning assumptions describe a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The target effect is described as a stated fraction of the observed effect. The record does not report the resulting sample-size values or the fractions used.

The registry explicitly characterizes these calculations as planning illustrations based on the observed effect, not as the trial’s own design. Therefore, they should not be read as evidence of observed power or as the basis for concluding whether the study was adequately designed. The stated illustration conventions distinguish participants per arm for non-time-to-event outcomes from total events for time-to-event outcomes. The primary endpoint described for this study is continuous, and the registry reports 0 analyses, so the record does not permit interpretation of an observed-effect-based calculation.

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("PortionSize", "MyFitnessPal")
n   <- c(62, 62)
m   <- c(-1055.7, -734.8)
sp  <- c(802.7, 628.5)
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  PortionSize  62  -1055.700  802.700  -1259.548  -851.852
CARM  MyFitnessPal  62  -734.800  628.500  -894.409  -575.191

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 1 outcome measure and 0 analyses. It identifies energy intake as the primary endpoint, assessed for PortionSize and MyFitnessPal against DLW. The record does not report a multiplicity plan, such as how testing across the app-specific assessments would be handled or whether any additional analyses were planned. The presence of a single listed outcome measure does not establish that only one statistical test was conducted; the record contains no analyses with which to assess that question.

Missing-data methods are not reported in the registry record. The endpoint description says all participants completed two study periods using both apps, but it does not specify whether all intended measurements were available, how incomplete daily records would be handled, or how missing DLW or app measurements would affect analysis. Completion of study periods and completeness of analyzable endpoint data are not interchangeable statements.

The registry says outcomes are summarized by app, pooling participants regardless of the order in which the apps were used. It does not describe a statistical analysis population beyond this outcome-summary statement, nor does it report whether exclusions or deviations were considered. The population description also does not provide enough detail to determine how every participant’s measurements contributed to each app-specific estimate.

There is no reported analysis plan for confidence intervals, adjustment, sensitivity analyses, or treatment of potential period or order effects. These are not necessarily absent from the study conduct; they are not reported in the registry record. The listed completion status does not resolve these reporting gaps.

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.

The registry’s adverse-event record lists 32 participants at risk in the PortionSize group and 31 in the MyFitnessPal group. It reports 0 serious adverse events, 0 other adverse events, and 0 deaths in each group. No serious-event terms are listed.

These figures describe the adverse-event entries in the registry; they do not provide an estimate of comparative risk between the apps. The study description says participants used both apps in separate periods, and the record cautions that the arms are not a like-for-like control comparison. The group labels in the safety record should therefore not be treated as randomized parallel groups.

General explanation: A zero count in a registry field means no events were recorded for that field in the posted entry. It does not, on its own, describe the completeness of event surveillance, the duration of observation for safety, or the precision of a risk estimate. The registry record does not provide further safety-monitoring details or event definitions. These counts should be read narrowly as the reported serious-event, other-event, and death figures, not as a broad claim about safety.

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

The study addresses measurement agreement: how closely energy intake estimated with each app matches the DLW criterion estimate in free-living conditions. The registry specifies a continuous endpoint, a mean-difference direction, and an equivalence-testing approach. It also states that the app periods were separate and that app-specific summaries pool participants regardless of the order of use.

Those design descriptions do not establish the observed level of agreement. The record lists 0 analyses and reports no mean differences, standard deviations, confidence intervals, p-values, equivalence bounds, or test conclusions. It is therefore not possible from the registry record to say whether either app’s estimates were equivalent to DLW, whether either tended to overestimate or underestimate, or how precise an estimate was.

General explanation: Equivalence is a defined statistical conclusion, not a synonym for a small-looking difference or a non-significant test. It depends on an acceptable range specified before analysis and on uncertainty being sufficiently limited relative to that range. The registry describes this framework but does not include the information needed to judge its application.

The study also does not provide a randomized comparison of the apps as competing interventions. Both apps were used in separate periods, and the comparison note says the arms are not a like-for-like control comparison. The study’s stated question is about app measurement relative to DLW, not whether one app is superior to the other or whether app use affects behavior.

Accordingly, the registry record documents the study objective and planned endpoint but does not report a result that supports a conclusion about measurement accuracy. It also does not establish how results would generalize beyond the participants and conditions studied. Such conclusions would require posted outcome analyses and additional information not reported in the registry record.

14. Why This Trial Matters Statistically

This study is statistically notable for framing app performance as an equivalence question rather than relying only on a test for a difference from the criterion measure. General explanation: Failure to detect a difference is not the same as demonstrating that two measurement methods are close enough for a defined purpose. Equivalence testing makes the acceptable range central to interpretation.

The registry identifies doubly labeled water as the criterion measure and defines the sign of the mean difference, which makes the intended direction of any future estimate clear. Its focus on free-living conditions also situates the measurement question outside a description limited to app use alone. The specified app and DLW timeframes are different, however, and the record does not explain how those periods were aligned analytically.

The design description also matters for comparisons. Participants used both apps in separate periods, and summaries are pooled regardless of app order. This is not a randomized parallel-group comparison, so app-specific summaries should be interpreted as separate assessments against the criterion rather than as a direct causal contrast between app assignments.

Finally, the record illustrates a distinction between study completion and statistical evidence. The status is COMPLETED, yet the registry lists 0 analyses and provides no endpoint results. Without estimates, uncertainty, and equivalence bounds, readers cannot evaluate the central statistical question. The planned method is informative about what the researchers intended to test, but it cannot stand in for the results of that test.

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 central limitation for interpretation is that the registry record lists 0 analyses and no endpoint results. It provides no app-specific mean differences, standard deviations, confidence intervals, p-values, or equivalence conclusions. The acceptable error bounds needed to interpret equivalence are also not reported. Consequently, neither the direction nor the precision of the observed app-to-DLW differences can be assessed.

The measurement periods are described as 4 days for each app and an overlapping 12 days for DLW, but the record does not explain how estimates from these periods were matched or analyzed. The description says the app summaries pool across participants regardless of order, but does not report how order or period effects were assessed. The study is not described as a randomized, like-for-like comparison, and allocation is not stated.

Participant characteristics, missing-data handling, the statistical analysis population, and a multiplicity plan are not reported in the registry record. Although the description says all participants completed two study periods, it does not establish the completeness of every measurement used for the endpoint. The record’s adverse-event counts likewise lack further details about monitoring or event definitions.

The planning assumptions are explicitly described as illustrations based on the observed effect, not the trial’s own design. The record does not give the resulting sample-size values or all inputs needed to evaluate those illustrations. These gaps limit what can be concluded from the registry record; they do not establish that the study conduct lacked procedures that were not recorded.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: PortionSize Study 3: Assessment in Free-living Conditions — NCT05906160.
• Registry record last updated 2026-10-19; status last verified 2026-07.
• Results first posted 2026-10-19.
• 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

NCT05906160, titled “PortionSize Study 3: Assessment in Free-living Conditions,” is listed as COMPLETED. The registry reports enrollment of 63, sponsorship by Pennington Biomedical Research Center, a start date of 2023-06-22, and a primary completion date of 2024-11-25.

The study objective is to assess energy-intake measurement by PortionSize and MyFitnessPal against doubly labeled water. Participants used both apps in separate periods; the outcome description says app summaries pool across participants regardless of order. The primary endpoint is mean energy intake in kcal/day, with a planned equivalence analysis of the app-to-DLW mean difference.

The record lists 1 outcome measure and 0 analyses. It does not report endpoint estimates, uncertainty, equivalence bounds, or conclusions. The adverse-event entries report 0 serious events, 0 other events, and 0 deaths for each app group.