PortionSize Study 2 Free-Living Evaluation: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for PortionSize Study 2 Free-Living Evaluation (NCT05166226), a registered phase not stated study. The primary outcome is Energy (Kilocalorie or kcal) Intake.

2. Clinical Question

The registry describes a semi-controlled, free-living evaluation of two apps intended to measure energy and nutrient intake. The primary endpoint is energy intake, expressed as mean daily kilocalories across a three-day free-living period for each app intervention. The planned analysis asks whether the mean from each app is equivalent to the criterion measure, rather than simply asking whether the two app means differ.

A related question compares the measurement errors for PortionSize and MyFitnessPal. For each app, error is defined as the difference between the app’s energy-intake value and the criterion value; weighed foods are given as an example of a criterion variable. The planned comparison uses dependent-samples t-tests, consistent with the registry’s description that participants used both apps in separate periods. However, the record does not report numerical results for either equivalence comparison or for the difference in errors. It therefore does not establish how close either app’s measurements were to the criterion measure or whether their errors differed.

3. Trial Design

This completed study is titled “PortionSize Study 2 Free-Living Evaluation.” Its official title describes a semi-controlled, free-living evaluation of the reliability and validity of the PortionSize and MyFitnessPal apps. The stated objective is to test the accuracy of the apps for measuring energy and nutrient intake in free-living conditions. Enrollment is listed as 45, and the condition is dietary habits. Pennington Biomedical Research Center is the sponsor.

The registry describes each participant as completing two study periods and using both apps. PortionSize and MyFitnessPal were used in separate periods. The primary outcome summary pools data across participants regardless of the order in which the apps were used. This is therefore not described as a comparison between separate groups of participants assigned to one app or the other. The registry’s own analysis description calls for dependent-samples tests of app-specific errors.

The record does not state how the order of app use was determined. Allocation is listed as not stated, and the comparison is not described as a randomized, like-for-like control comparison. Details such as the exact procedures within each period, the criterion-measurement schedule, and any measures taken to manage period or order effects are not reported in the registry record.

The study status is completed. The listed start date is 2022-01-10, and the primary completion date is 2022-11-11. The registry lists one outcome measure but no analysis entries. Consequently, the design and planned statistical approach can be described, but numerical outcome findings cannot be evaluated from 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

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.

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 Mean daily energy intake averaged across 3 day free-living period of each PortionSize app and MyFitnessPal app intervention..

The primary endpoint is energy intake in kilocalories per day. It is defined as mean daily energy intake averaged across a three-day free-living period for each app intervention. The endpoint is continuous, and the registry identifies its summary measure as the mean with standard deviation as the dispersion measure.

Energy intake is used to assess whether measurements made with each app are equivalent to a criterion measure. The registry’s planned approach is equivalence testing: the question is whether the difference between an app’s mean and the criterion mean falls within a prespecified error bound. This differs from asking only whether the means are statistically distinguishable. An ordinary test that fails to detect a difference would not, by itself, establish equivalence.

The criterion measure is also used to define each app’s measurement error. The registry gives energy values from weighed foods as an example of a criterion variable. Comparing the errors from the two apps addresses whether their deviations from the criterion differ. The endpoint therefore concerns measurement performance, not a clinical health outcome or a treatment effect.

The registry does not report the numerical equivalence bounds, endpoint estimates, standard deviations, confidence intervals, or test results. Without those results, the endpoint’s observed precision and the degree of agreement with the criterion measure cannot be assessed.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry describes equivalence testing with the Two One-sided T-test method for comparing each app with the criterion measure. It also describes dependent-samples t-tests for comparing errors between the apps. These are different questions, so their results would need to be interpreted separately. No numerical results from either analysis are reported in the registry record.

What does equivalence testing ask?

In general, an equivalence analysis asks whether a difference is small enough to fall within bounds considered acceptable for the measurement purpose. The null hypothesis is nonequivalence: the confidence interval for the difference extends beyond the prespecified error bounds. The alternative is equivalence: the difference is contained within those bounds. The registry describes this general setup and states that the hypothesized mean difference between an experimental method and the gold standard is zero.

The equivalence bounds are central to interpretation. They determine how much disagreement can be tolerated while still calling the measures equivalent. The registry record does not report the numerical bounds, so their practical meaning cannot be judged. It also provides no confidence interval or test outcome. Thus, the planned method is identifiable, but whether its equivalence criterion was met is not.

Equivalence testing should not be replaced by a conventional test of whether a difference is zero. In general, failing to reject a zero-difference null hypothesis can reflect limited precision and does not establish that differences are sufficiently small. Equivalence requires evidence that the plausible differences lie within the defined bounds. The record’s description of the null and alternative hypotheses reflects this distinction.

What does the paired error comparison ask?

The registry defines error as the difference between a value from an app and the corresponding criterion value. It proposes comparing PortionSize error with MyFitnessPal error using dependent-samples t-tests. In general, a dependent-samples test evaluates paired observations, here reflecting the design in which participants used both apps in separate periods. It focuses on the within-participant contrast between the two error values rather than treating the app measurements as unrelated groups.

The record also says these procedures would indicate whether the error from each method differs from zero. In general, an average error different from zero suggests systematic bias in one direction, while an average near zero can still coexist with substantial variation among individual measurements. A mean-based test does not by itself describe the spread of individual errors or establish agreement for each participant. The registry identifies standard deviation as the endpoint’s dispersion measure but gives no numerical dispersion result.

Why does uncertainty matter?

In general, a confidence interval expresses the uncertainty around an estimated quantity under the model and analysis used. For equivalence, the interval must be considered relative to the equivalence bounds; for a difference in errors, it indicates the range of values compatible with the data and model. A point estimate alone would not communicate the precision of the comparison. The registry record contains no confidence intervals, estimates, or p-values, so uncertainty cannot be quantified for these outcomes.

General statistical caution also applies to continuous measurements: the validity of a t-test depends on the data and model assumptions being suitable for the analysis. The record does not report diagnostic checks, transformations, or alternative methods. No claim about whether those methods were necessary or appropriate can be made from the registry entry alone.

How should planning illustrations be understood?

The stated sample-size assumptions describe a planning illustration computed from the observed effect, rather than the trial’s own design. They specify a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; the effect to detect is expressed as a fraction of the observed effect. The assumptions say that continuous-outcome illustrations give participants per arm. These are planning assumptions, not evidence about the study’s observed power, and they do not establish that the trial was designed using those assumptions. Numerical planning results are not included in the registry record.

In general, such illustrations can show how a hypothetical future study’s required sample size changes under different target effects and assumptions. They should not be read as a post hoc measure of the probability that the completed study detected an effect. Nor do they substitute for the observed estimate and its uncertainty. For this record, no sample-size justification, prespecified design calculation, or numerical outcome analysis is reported.

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(44, 44)
m   <- c(2128.8, 2120)
sp  <- c(762.1, 650.1)
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  44  2128.800  762.100  1897.100  2360.500
CARM  MyFitnessPal  44  2120.000  650.100  1922.352  2317.648

11. Multiplicity, Missing Data and Analysis Populations

The registry lists one outcome measure, the continuous primary measure of daily energy intake. It does not report a multiplicity adjustment or a broader plan for controlling error across multiple analyses. The primary description nevertheless identifies more than one planned statistical question: equivalence of each app to the criterion measure, a comparison of errors between apps, and tests of whether each method’s error differs from zero. The registry does not explain how those questions would be handled together for inferential purposes. The listed outcome count alone does not resolve that issue.

Missing-data procedures are not reported in the registry record. The population description states that all participants completed two study periods using both apps, but the record does not describe missing values at the measurement level, exclusions, imputation, or sensitivity analyses. The adverse-event denominators do not define the analysis population for the primary endpoint.

The registry says outcome measures are summarized by intervention and pooled across participants regardless of app order. It does not give a more detailed analysis-population definition, such as how incomplete app periods or unusable criterion measurements would be handled. The order in which participants used the apps is also not reported, nor is any planned adjustment for order or period effects.

These absences limit assessment of the analysis plan’s implementation and robustness. They do not show that missing data or multiplicity problems occurred; they mean the registry record does not provide the procedures or numerical information needed to evaluate them.

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 reports no serious adverse events and no other adverse events in the reported at-risk groups for either app. For PortionSize, the at-risk count is 44 for serious events and 44 for other events; the affected counts are 0 in both categories. For MyFitnessPal, the at-risk count is 44 for serious events and 44 for other events; the affected counts are also 0. Deaths are reported as 0 for each group. No serious-event terms are listed.

These figures describe the safety entries recorded for the listed groups. They are not measures of energy-intake accuracy and do not answer the primary equivalence question. The enrollment entry is 45, while the adverse-event at-risk counts are 44 for each app group. The registry record does not explain that denominator difference, so it should not be reconciled by assumption or treated as a description of the primary analysis population.

Zero reported events means that no affected participants are recorded in these adverse-event fields for the stated denominators. It does not establish that every possible event was captured or that the apps have no risk. The registry record does not provide further detail on event collection or monitoring procedures.

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

This registry entry establishes that the study was designed to evaluate app-based measurement of energy and nutrient intake in free-living conditions. It identifies a primary energy-intake measure, a criterion-based equivalence question for each app, and a paired comparison of app-specific measurement errors. It also states that participants used both apps in separate periods and that outcomes were pooled across participants regardless of app order.

It does not establish whether PortionSize or MyFitnessPal produced measurements equivalent to the criterion measure. The record contains no numerical endpoint estimates, confidence intervals, standard deviations, or analysis results. It also does not establish whether the error from one app differed from the error from the other. Those conclusions require results that are not reported in the registry record.

Even if numerical results were available, general interpretation would depend on the equivalence bounds, the uncertainty around the estimates, and the suitability of the analysis for the observed data. A finding that average error is near zero would not by itself show that individual app readings closely match the criterion. Conversely, an estimate that differs from zero would need to be considered alongside its precision and the measurement context.

The study addresses a measurement question, not whether either app improves health or changes dietary behavior. Its stated objective is accuracy in measuring intake. No clinical treatment comparison or health-outcome effect is reported in the record. The adverse-event entries are separate from measurement performance and should not be used to infer accuracy.

Thus, the record supports a description of the planned evaluation and its methods, but not a conclusion about comparative accuracy. The distinction is important: a protocol-level statistical question is not an observed result, and completion status alone does not supply the missing outcome estimates.

14. Why This Trial Matters Statistically

Measurement quality is consequential in clinical and behavioral research because conclusions depend on how well recorded values represent the underlying quantity of interest. This study’s question is whether two apps produce energy-intake measurements close enough to a criterion measure under free-living conditions. Such a question calls for methods that assess agreement or equivalence, rather than relying only on a test for a difference between app averages.

The registry’s planned use of equivalence testing makes the acceptable error range central to the scientific interpretation. In general, an equivalence margin should reflect what degree of disagreement is tolerable for the intended measurement use. The registry says bounds were prespecified but does not report their values, so readers cannot evaluate that choice. The absence of outcome estimates also prevents assessment of whether uncertainty was sufficiently narrow.

The paired structure is another important statistical feature. Participants used both apps in separate periods, and the planned error comparison uses dependent-samples tests. In general, paired analysis can account for participant-level differences that would otherwise add variation to a comparison. At the same time, app order and study-period conditions can matter in a repeated-period design. The registry does not report how order was determined or how possible period effects were addressed.

The record illustrates why a clearly described analysis plan and posted numerical results are both needed. The methods identify the intended questions, but without estimates, uncertainty, and the equivalence bounds, statistical readers cannot judge the evidence. The result is a well-defined measurement objective without a reported basis for comparing observed app performance.

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 most direct limitation for interpretation is that numerical outcome analyses are not reported in the registry record. Although the endpoint and planned tests are described, there are no reported means, standard deviations, effect estimates, confidence intervals, or p-values. The record therefore does not permit an assessment of the magnitude or precision of app-to-criterion differences.

The equivalence bounds are described as prespecified but their values are not reported. Without them, readers cannot determine what amount of disagreement was considered acceptable. The registry also does not report numerical results for the planned tests comparing app-specific errors or testing whether each error differs from zero.

The study uses both apps in separate periods, but the order procedure is not stated. The record does not report whether the order was randomized, how period effects were considered, or whether analyses adjusted for order. Allocation is listed as not stated, and the comparison is not a like-for-like control comparison between separate participant groups.

Missing-data handling, analysis exclusions, and sensitivity analyses are not reported. The population description says participants completed both periods, but it does not provide measurement-level completeness details. The record also does not report diagnostic checks or other information needed to assess the assumptions of the planned statistical tests.

These are limitations of what can be assessed from the registry record; they do not establish that the study procedures were inadequate or that the data had particular problems. The record supports a description of the objective and planned methods, but not a detailed evaluation of implementation or observed measurement performance.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: PortionSize Study 2 Free-Living Evaluation — NCT05166226.
• Registry record last updated 2026-05-13; status last verified 2026-04.
• Results first posted 2025-12-15.
• 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

NCT05166226 is a completed study sponsored by Pennington Biomedical Research Center. It evaluates the PortionSize and MyFitnessPal apps for measuring energy and nutrient intake in free-living conditions. The registry lists enrollment of 45 and describes participants as using both apps in separate study periods.

The primary outcome is mean daily energy intake in kcal/day averaged across a three-day free-living period for each app intervention. Planned analyses include equivalence testing against a criterion measure and dependent-samples comparisons of app-specific errors. The record lists one outcome measure and zero analyses.

Numerical outcome results, equivalence bounds, and details of missing-data handling and app-order allocation are not reported in the registry record. Adverse-event entries report zero serious events, zero other events, and zero deaths for the listed at-risk groups of 44 for each app.