Children's Healthy Living Community Randomized Trial: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Children's Healthy Living Community Randomized Trial (NCT01881373), a randomized na study. The primary outcome is Change in Prevalence of Overweight and Obesity Using Body Mass Index Percentile for Age and Sex.

2. Clinical Question

3. Trial Design

This was a completed, randomized, parallel, unmasked community trial sponsored by the University of Hawaii. The registry lists enrollment of 9840. The program focused on environmental changes intended to support active play and healthy food intake in remote, underserved native populations in the Pacific Region. Community engagement and capacity building were part of the described approach.

The record describes CHL program communities, a comparison community receiving a delayed optimized program, and Temporal communities assessed for anthropometric trends. The stated assessment plan included children in six communities selected in each jurisdiction. The intervention is described as multiple-component and environmentally focused, rather than as a treatment delivered solely to individual children.

The primary outcome was assessed at baseline and 24 months. Children recruited at each time were not the same children, and the number of communities was consistent across time. This distinction matters: the record describes repeated community-level assessment of recruited children, not follow-up of the same children from baseline to the final assessment. It does not establish that individual participants were randomized or followed longitudinally.

The registry lists the trial as randomized but also identifies its design as cluster-randomized. That structure affects how comparisons should be analysed because children within a community may share features and exposures. The registry's caution is that a simple individual-level re-analysis would be misleading, so only the sponsor's own analysis is shown. Details of the randomization sequence, allocation concealment, community assignment, and any stratification are not reported in the registry record.

The registry lists 23 outcome measures and 28 analyses. It names eight secondary outcomes covering beverage and water intake, physical activity, sedentary behavior, screen time, sleep disturbance, and fruit and vegetable intake. The available primary results concern overweight and obesity prevalence; results for the listed secondary outcomes are not detailed in the registry record for this analysis.

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.

Before assignment: Eighteen selected communities were matched for size and randomized within jurisdiction to intervention or control. Nine additional communities were monitored with no randomization for temporal trends of obesity. 2-8 year old children and their households and environments were measured and evaluated in each of these communities.

Baseline

24 Months

78 Months

Analysis population for the primary outcome, as defined in the registry: Children 2-8y in selected communities of the Pacific Region; the children from each community recruited each time were NOT the same children. The number of communities is consistent across time.

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: 2 to 8 year old children

6. Primary Endpoint

The registry lists the primary outcome as Change in Prevalence of Overweight and Obesity Using Body Mass Index Percentile for Age and Sex, measured over Baseline, 24 months.

The primary endpoint is the change in prevalence of overweight plus obesity from baseline to 24 months, expressed as a percentage of children 2 to 8 years old. Weight and height are used to derive body mass index (BMI), defined as weight in kilograms divided by height in meters squared. Children are classified using BMI percentiles for age and sex according to CDC 2000 criteria.

Overweight is defined as BMI at or above the 85th percentile and below the 95th percentile; obesity is at or above the 95th percentile. The endpoint combines these categories. The record defines the comparison category as healthy weight, from the 5th percentile up to but not including the 85th percentile, and excludes children below the 5th percentile.

Prevalence is a population-level measure: it describes the proportion of assessed children meeting the combined overweight or obesity definition at a given assessment. It can be used to compare the distribution of weight status across communities and assessment times. It does not directly measure an individual child's change in BMI or weight status.

The registry describes the endpoint as a change over time, but the reported analyses are labelled mean differences in final values. The registry record does not report the underlying group-specific endpoint values or enough group-code detail to reconcile those reported comparisons with the stated change endpoint.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is continuous in the registry's analysis classification, and its dispersion is identified as a standard error. The endpoint itself is a prevalence measure based on BMI categories. The registry reports logistic regression and labels the resulting parameter a mean difference in final values. Since model specification and outcome coding are not reported in detail, the relationship between the logistic model and the mean-difference label cannot be reconstructed from the registry record alone.

What does a confidence interval express?

A confidence interval describes uncertainty around an estimated comparison under the assumptions of the analysis. Its width reflects how precisely the comparison is estimated: a narrower interval indicates greater precision than a wider interval, all else being equal. For the first reported comparison, the 95% interval runs from -7.29 to -0.47. For the second, it runs from -8.58 to 3.32. The first interval is entirely on one side of zero, while the second includes zero. This describes the estimates' statistical uncertainty; it does not by itself establish practical importance, explain the cause of a difference, or identify which named group had the higher value.

How should the p-values be read?

A p-value summarizes how unusual data at least as incompatible with a specified null model would be if that model and the analysis assumptions held. The reported values are 0.02 and 0.28. These values do not measure the probability that the null hypothesis is true, the probability that an intervention is effective, or the size of an effect. They also do not answer whether a difference is important in practice. Interpretation depends on the estimate, its interval, the design, and the analysis assumptions.

The registry labels both comparisons as superiority analyses. That label identifies the stated comparison type; it is not a substitute for understanding group order, outcome coding, or the uncertainty around each estimate. Since the group-code mapping and subtraction order are not reported, the negative estimates cannot be given a named-group direction from the record alone.

Why does cluster randomization matter?

In a cluster-randomized design, assignment occurs at the group or community level rather than independently for every person. People in the same community may have more similar outcomes than people in different communities. General statistical reasoning indicates that treating all individuals as independent can understate uncertainty when outcomes are correlated within clusters. The registry explicitly warns that a simple individual-level re-analysis would be misleading and shows only the sponsor's analysis. The specific model adjustments used to account for clustering are not reported in the registry record.

The number of communities is consistent across time, but the children assessed at each time are not the same. This is a repeated set of community assessments with different recruited children, not an individual-level longitudinal analysis of the same children. A comparison based on final values does not necessarily convey the same information as a comparison of within-child change, and the record does not report participant-level follow-up or a specific model for changes in community prevalence.

What can be inferred from the reported comparisons?

The reported estimates and intervals quantify the comparisons the sponsor analysed. They do not provide the separate prevalences for each group, and the registry does not state the mapping of OG000, OG001, and OG002 to named groups. Consequently, it is not possible from the registry record to verify which community groups form each comparison or translate the direction of the estimates into named arms. The endpoint description also refers to change, whereas the analysis parameter is labelled as a difference in final values. Clarification of the analysis scale and group ordering is not reported.

The registry lists 23 outcome measures and 28 analyses. When many outcomes or comparisons are analysed, the chance of at least one apparently unusual result can increase. General statistical practice may address this issue through prespecified testing strategies or multiplicity adjustments, but whether either was used is not reported in the registry record. Thus, the p-values should not be assumed to have been adjusted for the broader set of analyses.

The sample-size assumptions described for planning illustrations are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation, with the effect to detect expressed as a fraction of the observed effect. These illustrations are planning exercises, not observed power and not the trial's own design justification. The trial-specific sample-size justification is not reported in the registry record. Because the trial is cluster-randomized, individual-level planning assumptions alone would not describe all relevant design features; the registry does not report whether the planning illustration incorporates clustering.

Other details needed to fully assess the statistical analysis—including the exact model formula, covariates, handling of missing outcomes, analysis population, protocol deviations, and any interim analyses—are not reported in the registry record. These gaps do not invalidate the posted estimates, but they limit independent assessment of how the analyses were conducted and how broadly they can be interpreted.

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("CHL Program", "Control Community", "Temporal")
n   <- c(2859, 2786, 2726)
m   <- c(-2.8, 1.4, -0.4)
sp  <- c(1.16, 1.49, 2)
sd  <- sp * sqrt(n)      # standard errors converted to SDs
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  CHL Program  2859  -2.800  62.025  -5.075  -0.525
CARM  Control Community  2786  1.400  78.646  -1.522  4.322
CARM  Temporal  2726  -0.400  104.422  -4.322  3.522

11. Analyses Reported by the Sponsor

The sponsor's own analysis is the relevant analysis reported for this cluster-randomized trial. The registry explicitly cautions that a simple individual-level re-analysis would be misleading, because it would fail to respect the community-level design. Accordingly, the posted comparisons are described as logistic-regression results rather than replaced by an unadjusted calculation treating children as independent observations.

The first sponsor-reported comparison has an estimate of -3.88, a 95% confidence interval from -7.29 to -0.47, and p = 0.02. The second has an estimate of -2.63, a 95% confidence interval from -8.58 to 3.32, and p = 0.28. These are the registry's own reported values. A separate recalculated estimate is not reported in the registry record, and the registry does not provide enough group-code mapping or model detail to reconstruct a like-for-like comparison independently.

The registry's caution means that an individual-level recalculation should not be treated as a competing estimate or as a check of the sponsor's clustered analysis. The sponsor's own analysis takes precedence for interpreting these posted results. The available record does not establish the exact clustering correction, covariates, or other model choices, so the sponsor estimates can be reported but not independently reproduced from the registry entries.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists 23 outcome measures and 28 analyses. It does not report whether the primary outcome or comparisons were designated in a statistical analysis plan, whether a multiplicity strategy was used, or whether any adjustment was made for testing across outcomes and analyses. The reported p-values of 0.02 and 0.28 should therefore be read as the values attached to the posted comparisons, not assumed to be multiplicity-adjusted.

Missing-data methods are not reported in the registry record. The record does not state how missing height, weight, or other outcome information was handled, whether any imputation was used, or whether participants were excluded from analyses for missing observations. It also does not report the analysis population definition, such as the handling of communities or children who did not contribute a measurement at the final assessment.

The primary endpoint description says that recruited children were not the same at baseline and 24 months. This means the assessment is not a paired comparison of the same children's outcomes. The registry does not report how changes in the composition of recruited children were assessed or accounted for in the model.

Interim analyses, stopping rules, and any prespecified monitoring plan are not reported in the registry record. The registry also does not provide a detailed statistical analysis plan, including model covariates, cluster-level variance assumptions, or a plan for secondary outcome analyses. Those omissions make it difficult to assess prespecification, missingness, and the full context of the reported p-values. They do not justify assuming that particular methods were or were not used; the relevant details are simply not reported in the registry record.

13. 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 figures list 3359 participants at risk in the CHL Program group, 3265 in the Delayed Optimized CHL Program group, and 3216 in the Temporal group. For each group, the registry reports zero serious adverse events, zero other adverse events, and zero deaths.

These figures describe what was recorded in the registry for the listed at-risk groups. They should not be interpreted as proof that no adverse events occurred outside the recorded reporting framework, nor do they establish that the interventions are free of risk. The registry does not provide serious adverse-event terms, since no top serious terms are listed, or a fuller account of event ascertainment and reporting procedures.

The safety figures are distinct from the primary anthropometric outcome. They do not provide information about changes in overweight or obesity prevalence, and they do not resolve the limitations in group-code mapping or statistical-analysis detail. The record's reported zeros are best read as registry-reported counts, not as a broader safety conclusion.

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

The posted results describe sponsor-reported comparisons of final values for the primary outcome, prevalence of overweight plus obesity based on BMI percentile for age and sex. One comparison has an estimate of -3.88 with a 95% confidence interval from -7.29 to -0.47 and p = 0.02. The other has an estimate of -2.63 with a 95% confidence interval from -8.58 to 3.32 and p = 0.28. The first interval excludes zero; the second includes zero and permits values on either side of it.

These estimates do not, by themselves, show which named group had a higher prevalence. The registry reports coded comparisons but does not map those codes to the named CHL program, delayed optimized program, and Temporal groups, or state the subtraction order. Nor does it provide group-specific prevalence values that would make the comparison's direction explicit. The parameter is labelled a mean difference in final values, while the endpoint is described as change from baseline to 24 months; the registry record does not explain this distinction.

The result is not an estimate of individual children's change. Different children were recruited at the assessment times, so the outcome reflects community samples rather than the same children's trajectories. The community-level randomized design also makes the sponsor's model important; a simple analysis treating children as independent would be misleading according to the registry.

The confidence intervals describe statistical precision, not clinical importance. The registry does not state a threshold for a meaningful change, provide enough detail to independently reproduce the analysis, or report whether multiplicity was addressed. The p-values should not be interpreted as probabilities that a result is true or as direct measures of treatment benefit.

Accordingly, the findings are limited to the reported model comparisons and their uncertainty. They do not establish that the program caused a particular change, that two named groups were equivalent, or that the observed differences are clinically important. The missing group mapping, analysis details, and information about missing data constrain stronger interpretation.

15. Why This Trial Matters Statistically

This trial is statistically notable because its intervention and comparison structure is community-based, while its primary outcome is measured in children recruited from those communities. In a cluster-randomized study, the unit of assignment and the unit of measurement are not necessarily the same. Statistical methods need to respect the grouping of children within communities; treating every child's observation as independent can give misleading uncertainty. The registry directly cautions against that kind of individual-level re-analysis.

The assessment schedule also creates an important distinction between population change and individual change. Baseline and 24-month assessments recruited different children, although the number of communities remained consistent. Thus, the primary outcome tracks prevalence in community samples over time; it does not follow each child's progression. This distinction affects what a change in prevalence can mean and what kinds of conclusions are supported.

The registry's reporting illustrates why estimates should be considered together with confidence intervals and design context. The first comparison's interval excludes zero, while the second comparison's interval includes zero. Those statements are about the posted estimates and their statistical precision. The group-code mapping, however, is not reported, so the direction cannot be assigned to named arms from the record alone.

Finally, the record lists many outcomes and analyses but does not describe a multiplicity plan, missing-data methods, or a detailed model specification. These are not minor labels: they shape how confidently a reader can interpret nominal p-values, understand uncertainty, and assess reproducibility. The trial therefore offers a useful case in interpreting community-level randomized evidence while distinguishing design details, reported estimates, and unreported analysis details.

16. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry does not map the codes used in the reported comparisons to the named CHL program, delayed optimized CHL program, and Temporal groups. It also does not state the subtraction order. As a result, the negative estimates cannot be attributed to a particular named group as higher or lower. This is a central limitation for a plain-language interpretation of direction.

The endpoint is described as change from baseline to 24 months, but the reported parameter is labelled a mean difference in final values. Group-specific prevalence values and the detailed computation of the reported difference are not reported in the registry record. The relation between the endpoint description, logistic-regression method, and mean-difference estimate is therefore not fully transparent.

The trial assessed different children at baseline and 24 months. The results cannot describe within-child change, and differences in the composition of recruited children across assessment times could affect the observed community prevalence. The registry does not report details of recruitment at each time or methods addressing changes in sample composition.

The record identifies a cluster-randomized design but does not report the exact model specification or how the analysis accounted for clustering. It also does not provide covariate details, missing-data methods, analysis-population definitions, or an account of interim analyses. The listed 23 outcome measures and 28 analyses make the absence of a reported multiplicity strategy relevant to interpreting the p-values.

Finally, the record gives no trial-specific sample-size justification. Planning assumptions describe an illustration rather than the trial's own design rationale, and should not be read as observed power. These limitations restrict independent reproduction and causal interpretation of the reported comparisons without changing the values the registry records.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Children's Healthy Living Community Randomized Trial — NCT01881373.
• Registry record last updated 2024-03-04; status last verified 2023-06.
• Results first posted 2020-05-01.
• 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.

20. Record Summary

The Children's Healthy Living Community Randomized Trial (NCT01881373) was a completed, randomized, parallel, unmasked community trial sponsored by the University of Hawaii. The registry lists enrollment of 9840 and describes an environmentally focused CHL program, a delayed optimized comparison program, and Temporal communities.

The primary outcome was change in prevalence of overweight plus obesity by BMI percentile for age and sex, assessed at baseline and 24 months among children 2-8y in selected Pacific Region communities. Different children were recruited at each assessment. The registry reports two logistic-regression comparisons as mean differences in final values: -3.88 (95% confidence interval -7.29 to -0.47; p = 0.02) and -2.63 (95% confidence interval -8.58 to 3.32; p = 0.28).

The group-code mapping, analysis details, missing-data methods, and multiplicity plan are not reported in the registry record. The record cautions that a simple individual-level re-analysis would be misleading for this cluster-randomized design.