Phone Coaching for Weight Loss Maintenance: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Phone Coaching for Weight Loss Maintenance (NCT04293055), a randomized na study. The primary outcome is Percent Change in Body Weight From Baseline to Month 12.

2. Clinical Question

3. Trial Design

This was a randomized, parallel-group behavioral study with no masking. The registry lists enrollment of 77 and describes the study as completed. The listed study start was 2020-03-04, and the primary completion date was 2023-08-23. The sponsor was The Miriam Hospital. The registry record lists the phase as not applicable.

The maintenance strategy was offered to people who had achieved and maintained at least 5% weight loss in a 1-year behavioral weight-loss program. Both randomized groups received monthly newsletters and encouragement to weigh daily using a smart scale that electronically transmitted weight data to research staff. At baseline, participants were randomized to the possibility-of-coaching group or the maintenance-only group. The registry summary describes the coaching condition as activated when a participant regained more than 1.5% of weight. At that point, the coaching group received 4 brief weekly calls, while the control group received no additional intervention contact.

The primary analysis population was narrower than the overall enrollment: only people who regained 1.5% of weight and triggered the randomized condition were included. The posted primary endpoint reports 35.0 participants in the possibility-of-coaching group and 31.0 in the maintenance-only group. Multiple imputation was used to correct for missing data, according to the endpoint description. The registry record does not report the number of participants who triggered the condition before missing-data handling or explain the imputation procedure.

The parallel design makes the central comparison the difference between randomized groups, while the trigger-based population defines who contributes to the primary analysis. The registry record does not report further details about allocation procedures, adherence, masking beyond the listing of none, or the original statistical plan.

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: Only individuals who regained 1.5% weight and triggered randomization condition were included in the analysis. Multiple imputation used to correct for missing data.

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 Percent Change in Body Weight From Baseline to Month 12, measured over Baseline and Month 12.

The primary endpoint is percent change in body weight from baseline to Month 12. It summarizes the change in body weight over the maintenance follow-up as a continuous outcome, with a group mean reported for each randomized condition. A mean provides a compact description of the group’s average result, while the comparison of means estimates how far apart the groups are on that scale.

The endpoint description specifies that the analysis included only participants who regained 1.5% of weight and triggered the randomized condition. That makes the endpoint relevant to the group that reached the trigger, rather than to every person enrolled at baseline. The registry also states that multiple imputation was used to correct for missing data, but does not report the imputation method or the amount and pattern of missingness.

Using a continuous weight-change measure preserves more information than reducing outcomes to categories such as regained or maintained. It can describe the average direction and magnitude of change in each group. The posted results give means of 5.18 and 5.71, with standard errors of 0.72 and 0.76, respectively. The endpoint is useful for comparing the groups’ average weight-change results at the stated follow-up, but it does not by itself describe individual variation, the course of weight change at every point, or outcomes beyond the registry’s stated timeframe.

7. Results Posted in the Registry

• Maintenance Program Only: mean 5.71 versus 5.18 on control; difference 0.53 (95% CI -1.52 to 2.58), standardized effect d = 0.12; z-test p = 0.613.

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.

Percent Weight Change Over the Coaching Period

• 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:
  • Maintenance Program Only: mean 0.37 versus -1.05 on control; difference 1.42 (95% CI 0.57 to 2.27), standardized effect d = 0.81; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Maintenance Program Only than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The registry lists one secondary outcome: percent weight change over the coaching period, from the trigger date to 40 days post-trigger date. Its posted comparison differs in direction from the primary result: the secondary endpoint reports a difference between groups with an unadjusted p-value of <0.001, while the primary Month 12 comparison has an unadjusted p-value of 0.613 and a confidence interval that includes zero. This contrast suggests that the short-term and longer-term summaries do not tell the same statistical story.

The secondary endpoint complements the primary endpoint by addressing a shorter period immediately following the trigger, when coaching calls were offered in the coaching condition. It does not answer the same question as the Month 12 endpoint. A difference across the coaching period cannot by itself establish that a difference persisted through Month 12, and the primary result cannot be replaced by a secondary result selected for its stronger statistical signal.

Unadjusted secondary results generally carry less evidential weight when several outcomes or analyses could be examined, because the chance of observing a seemingly unusual result can increase across multiple comparisons. The registry lists one secondary outcome, but it does not report a multiplicity plan or any adjustment for multiple testing. The record also lists no reported analysis entries. Without those details, the secondary p-value should be treated as an additional finding that merits cautious interpretation rather than as confirmatory evidence on its own.

The registry does not report whether the secondary outcome was designated in advance as confirmatory, how missing secondary measurements were handled, or whether there were other analyses of the same timeframe. These gaps limit assessment of how much weight to place on the secondary result. The overall pattern is therefore a notable short-term contrast alongside an imprecise primary longer-term comparison, with the latter remaining central to the trial’s stated primary question.

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

• Percent Weight Change Over the Coaching Period

9. Statistical Interpretation

10. Statistical Methods Explained

The registry reports a continuous primary endpoint, group means, standard errors, a mean difference, a confidence interval, and an unadjusted p-value. These quantities answer related but distinct questions. The means describe the average in each group; the mean difference summarizes the contrast; the confidence interval expresses uncertainty around that contrast; and the p-value measures compatibility with a specified null model under the analysis assumptions.

What does a mean difference express?

A mean difference is the subtraction of one group mean from the other, following a stated comparison direction. In the posted primary comparison, the direction is maintenance-only minus possibility of coaching. The reported estimate is 0.53. Keeping the direction clear is important: reversing the order of the groups would reverse the sign of the difference while leaving the underlying separation unchanged. The estimate is in the endpoint’s percent-change scale, not a count of participants or a percentage-point probability.

A group mean can make a continuous outcome easier to compare, but it does not show how individual values are distributed. People in a group may have results that vary widely even when the average is close to another group’s average. The registry provides standard errors for each group mean, not standard deviations in the endpoint results. A standard error describes the estimated uncertainty in a mean; it is not a direct summary of how much individuals’ outcomes vary.

What does the confidence interval add?

A confidence interval gives a range of values reasonably compatible with the data and the procedure used to calculate it. Its purpose is to show the precision of the estimated group difference, rather than to provide a second estimate or a guarantee that the true value lies within the interval. For the primary comparison, the interval runs from -1.522 to 2.582 and includes zero. That means the posted result does not rule out no difference, and it also allows values on either side of zero. The width signals that uncertainty around the group contrast remains substantial.

The registry does not provide confidence limits for each group mean. The reported interval applies to the difference between groups, which is the direct contrast relevant to the primary question. A reader should not infer an interval for either group mean from the posted standard errors without additional assumptions and calculations.

How should the p-value be read?

The primary unadjusted p-value is 0.613. In general, a p-value describes how unusual a result at least as discrepant as the observed one would be under a specified null model, assuming the analysis model and its conditions are appropriate. It is not the probability that the null hypothesis is true, nor a measure of the size or practical importance of a difference. A larger p-value does not demonstrate equivalence, and a smaller one does not establish that an effect is important or free from bias.

The primary comparison is labelled unadjusted. The registry record does not report whether covariates were included, whether alternative models were examined, or whether the displayed p-value corresponds to any additional analyses. The record lists no reported analysis entries. Therefore, the posted comparison should be read as the result available in the registry, not as a complete account of every analysis that may have been considered.

Why does the analysis population matter?

Although the registry lists enrollment of 77, the primary endpoint description restricts analysis to individuals who regained 1.5% of weight and triggered the randomized condition. The endpoint group counts are 35.0 and 31.0. The trial’s question therefore concerns the group meeting that trigger condition, not automatically all people who entered the study. Because the coaching offer was activated after regain, the treatment contrast also reflects a strategy of possible coaching after a trigger, compared with no additional contact after the trigger.

The endpoint description says multiple imputation was used to correct for missing data. In general, multiple imputation uses a model to replace missing values with plausible values across multiple completed datasets, then combines the results to reflect uncertainty from the imputation process. Its credibility depends on the imputation model and assumptions about why data are missing. Those details, the amount of missingness, and any sensitivity analyses are not reported in the registry record, so the imputation cannot be evaluated in depth from the posted description.

Why can over-dispersion matter?

General statistical explanation: some outcomes vary more than a simple model expects. This extra variability, often called over-dispersion, can make standard errors too small if the model assumes less variation than the data contain. That can make confidence intervals appear narrower and p-values appear more decisive than warranted. The primary endpoint is a continuous measure, and the registry reports standard errors, but it does not report evidence of over-dispersion or describe a model that would require an over-dispersion assessment. This general issue should not be mistaken for a reported flaw in the study.

Overall, the posted statistics support a cautious reading: the primary mean difference is modest in relation to its uncertainty, and the interval includes zero. The registry record does not report the full statistical plan, so it is not possible to establish whether the displayed method was the only planned analysis or how other analytic choices were handled.

11. Confidence Intervals and Precision

The primary mean difference is 0.53, with a 95% confidence interval from -1.522 to 2.582. This interval is the key precision measure for the group contrast. It includes zero, so the primary comparison is compatible with no average difference, but it also includes values indicating a difference in either direction. The interval’s breadth means the posted result does not determine the size or direction of the underlying contrast with much precision.

The endpoint group means are 5.18 and 5.71, with standard errors of 0.72 and 0.76. Standard errors describe uncertainty in estimated means, not the spread of participant-level outcomes. The registry does not post standard deviations for the primary result, nor does it give confidence limits for each group mean. The direct interval for the mean difference is therefore more informative for judging the precision of the comparison than comparing the two standard errors by themselves.

The secondary mean difference is reported as 1.42, with an unadjusted p-value of <0.001. Its timeframe is from the trigger date to 40 days post-trigger date. The registry provides a confidence interval for this comparison, but the primary conclusion should not be inferred from the secondary endpoint: it addresses a different time window, and its result is unadjusted. A relatively small p-value does not remove uncertainty about the longer-term primary comparison.

Precision is not the same as importance. A confidence interval can describe uncertainty around an estimate without deciding whether values within it matter to participants or researchers. The registry does not report a threshold for a clinically meaningful difference, so the interval cannot be judged against such a threshold from the record alone.

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 sample-size planning figures are participants per arm, not total event counts. The posted planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set as the stated fraction of the observed effect. These are planning illustrations calculated from the observed effect, not the trial’s original design assumptions; those original assumptions are not reported in the registry record.

The 80% power column describes the number of participants per arm that the planning calculation indicates would be needed for an 80% chance of detecting the specified effect under its assumptions. The 90% power column gives the corresponding planning number for a higher chance of detection. These are prospective design quantities: they describe hypothetical sample sizes for a future study under the stated assumptions. This is not observed power.

The planning display includes scenarios for planning for 100% of the observed effect, planning for 75% of the observed effect, and planning for 50% of the observed effect. For the primary endpoint, the listed participants per arm for 80% power are 1,008, 1,792, and 4,031 across those scenarios, respectively; the corresponding figures for 90% power are 1,350, 2,399, and 5,397. The planning values grow as the assumed effect becomes smaller, because a smaller group difference is harder to distinguish from random variation under the same general assumptions.

The planning curve should be read as a visual representation of how the required per-arm sample size changes with the effect used for planning. It does not show the actual precision of the completed trial or replace the primary confidence interval. The required sample sizes also depend on the assumptions used, and the registry record does not report whether the trial’s original design used these same assumptions.

13. How the Sample Size Is Derived

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

Formula

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

Worked calculation for this trial

1. Observed difference Δ = 0.530; pooled standard deviation σ = 4.246.
2. n = 2 × (1.96 + 0.842)² × 4.246² / 0.530² = 1,007.7, rounded up to 1,008 participants per arm for 80% power.

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

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 0.53                      # observed difference in means
sigma <- 4.24643968519512                     # 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(4293055)
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 = 1,008 (80% power), 1,350 (90% power)
 75% of observed effect: n per arm = 1,792 (80% power), 2,399 (90% power)
 50% of observed effect: n per arm = 4,031 (80% power), 5,397 (90% power)
power.t.test: 1008.7 
Simulated power at n = 1,008 per arm: 0.801 (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 record does not report a statistical plan describing how multiple outcomes or analyses were to be handled. It lists two outcome measures, including the primary Month 12 endpoint and one secondary endpoint, but provides no multiplicity adjustment or prespecified testing hierarchy. When more than one comparison is considered, unadjusted p-values do not account for the increased opportunity to observe a small p-value by chance. This is why an unadjusted secondary result generally deserves less weight than a result supported by a prespecified multiplicity strategy.

The record lists no reported analysis entries. It does not report whether there were interim analyses, additional exploratory comparisons, model adjustments, or subgroup analyses. These details are not reported in the registry record, so the displayed statistics cannot be placed within a fuller account of the analysis process. The absence of such reporting does not establish that those analyses occurred; it means the registry does not document them.

The primary endpoint description states that multiple imputation was used to correct for missing data. In general, multiple imputation fills in missing values repeatedly using a model based on observed information and combines estimates across completed datasets. This approach can account for uncertainty caused by missing values when its assumptions and model are suitable. The registry does not report the imputation variables, number of imputations, missing-data amount, reasons for missingness, or sensitivity analyses. The assumptions underlying the treatment of missing data therefore cannot be assessed from the record.

The primary analysis population includes only people who regained 1.5% of weight and triggered the randomized condition. The registry does not describe analyses of participants who did not reach the trigger or say whether the primary analysis follows an intention-to-treat approach in a broader randomized population. It also does not report how the trigger-defined population was handled if outcome data were missing. These limits matter because the analysis population defines the group to which the posted primary comparison directly applies.

Together, the unreported multiplicity details, absence of listed analysis entries, and limited missing-data description call for restraint in treating any one p-value as a complete summary of evidence. The posted estimate and interval remain interpretable as reported, but the registry does not support a comprehensive assessment of every analytic choice.

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 registry reports serious adverse events, other adverse events, and deaths by group. For the possibility-of-coaching group, 41 were at risk for serious adverse events and 41 were at risk for other adverse events; the reported affected counts were 0 for each category, and deaths were 0. For the maintenance-only group, 36 were at risk for each adverse-event category; the reported affected counts were 0 for each category, and deaths were 0.

These are registry-reported counts for the listed groups. The adverse-event denominators differ from the endpoint analysis counts, and the record does not explain the relationship between those populations. The results should therefore be read using the adverse-event figures as reported rather than assumed to represent the same analysis population as the primary endpoint.

No serious adverse-event terms are listed. Zero reported events in these categories does not establish that no adverse events occurred outside the registry’s reporting definitions or observation and reporting process. The record does not provide further detail about event collection, monitoring procedures, or the duration of safety follow-up.

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

The primary result estimates the average Month 12 percent weight change difference between the trigger-defined randomized groups. The posted estimate is 0.53, and the confidence interval ranges from -1.522 to 2.582. Because the interval includes zero and spans values on either side, the registry result does not provide a precise estimate of the direction or size of the average difference. The unadjusted p-value of 0.613 is consistent with that uncertainty, but it should not be interpreted as proof that the groups are identical.

The finding applies to participants who regained 1.5% of weight and triggered the randomized condition. It does not directly describe people who did not reach that trigger, nor does it automatically generalize to people who did not first achieve and maintain at least 5% weight loss in a 1-year behavioral program. The strategy being compared also includes the shared maintenance program in both groups. The contrast is therefore the possibility of coaching after a trigger versus maintenance support without additional contact after the trigger.

The secondary result concerns weight change over the coaching period and has an unadjusted p-value of <0.001. It provides information about a shorter timeframe, but it does not establish a lasting difference at Month 12. The two outcomes have different timeframes and should not be collapsed into one conclusion. The primary estimate remains imprecise, while the secondary result should be interpreted cautiously in the absence of reported multiplicity procedures.

The registry record does not report a threshold for a clinically meaningful difference, a full statistical analysis plan, details of the multiple-imputation procedure, or sensitivity analyses. Those omissions prevent a complete evaluation of practical importance and robustness. The results describe the posted data and their statistical uncertainty; they do not establish what would happen for every individual or under different intervention conditions.

Accordingly, the central message is not a claim that phone coaching does or does not work. It is that the posted primary comparison in the trigger-defined population has a confidence interval that includes no difference and remains broad, while the secondary short-term comparison is statistically distinct in its reported unadjusted analysis. Further interpretation depends on study details that are not reported in the registry record.

17. Why This Trial Matters Statistically

This trial is statistically interesting because it evaluates a conditional intervention strategy: phone coaching was available after a weight-regain trigger, rather than being delivered continuously to everyone in the intervention group. That design makes the population and timing of the analysis central to interpretation. The primary result concerns participants who reached the trigger, so it answers a narrower question than whether the maintenance strategy affects all participants enrolled at baseline.

The results also illustrate why a point estimate should be read together with its confidence interval. The primary mean difference is 0.53, but its 95% confidence interval extends from -1.522 to 2.582. The estimate alone might suggest a small separation; the interval shows that the data are compatible with a wider range of possible differences, including zero. The primary p-value of 0.613 is not evidence of equivalence, and the registry provides no equivalence margin.

A second statistical lesson is that outcomes at different times can give different signals. The short-term secondary comparison has an unadjusted p-value of <0.001, whereas the primary Month 12 comparison is less precise and has an interval including zero. Distinct timeframes answer distinct questions. Statistical evidence for a shorter interval cannot be assumed to persist at the primary follow-up.

Finally, the record shows how missing-data and multiplicity reporting affect interpretation. Multiple imputation is noted, but its implementation and assumptions are not described. The registry also does not report a multiplicity plan. These limitations do not invalidate the posted results, but they constrain how confidently the estimates can be evaluated and how much weight should be assigned to the secondary finding.

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.

The primary analysis is limited to participants who regained 1.5% of weight and triggered the randomized condition. Its results therefore do not directly address participants who did not reach the trigger. The registry record does not report the number who triggered before the analysis or describe outcomes for those who did not trigger, limiting assessment of the overall randomized strategy.

The primary mean-difference confidence interval, -1.522 to 2.582, is broad and includes zero. This indicates limited precision for the longer-term group contrast. The registry does not provide a clinically meaningful difference threshold, so the possible values in the interval cannot be classified as important or unimportant using a threshold from the record.

Multiple imputation is reported for the primary endpoint, but the registry does not describe the imputation model, missingness, or sensitivity analyses. The robustness of the primary result to missing-data assumptions therefore cannot be determined. The registry also does not report the full statistical plan, multiplicity handling, or whether further analyses were conducted. The listed analysis count is zero, leaving no additional analysis details in the record.

The secondary outcome is measured over a shorter period than the primary endpoint and is reported with an unadjusted p-value. No multiplicity strategy is reported, which limits the confirmatory weight of that result. The record does not provide detailed adherence information or explain how much coaching contact participants received beyond the description of 4 brief weekly calls for those assigned to coaching after the trigger.

Finally, the safety record reports no affected participants in the listed event categories, but it does not describe event ascertainment procedures or the relation between safety denominators and the endpoint analysis population. These reporting limits make it inappropriate to infer more about safety or the broader experience of participants than the registry counts document.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Phone Coaching for Weight Loss Maintenance — NCT04293055.
• Registry record last updated 2024-12-31; status last verified 2024-12.
• Results first posted 2024-12-31.
• 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

NCT04293055 is a completed randomized, parallel-group study of phone coaching as a weight-loss-maintenance strategy. Both groups received a low-intensity maintenance program; the intervention group could receive 4 brief weekly coaching calls after weight regain triggered the condition. The primary endpoint was percent change in body weight from baseline to Month 12 among participants who triggered randomization.

The posted primary mean difference was 0.53, with a 95% confidence interval from -1.522 to 2.582 and an unadjusted p-value of 0.613. The interval includes zero and leaves substantial uncertainty about the group difference. The single posted secondary outcome covered the coaching period and had an unadjusted p-value of <0.001. The registry reports multiple imputation for the primary endpoint but does not describe its details, the multiplicity plan, or the original statistical design assumptions.