Sweet Preference and Alcohol Craving: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Sweet Preference and Alcohol Craving (NCT01296646), a randomized phase 2 study. The primary outcome is Percent Heavy Drinking Days.

2. Clinical Question

The study focused on whether two participant characteristics might help distinguish treatment response: hedonic response to sweet taste and initial craving for alcohol. Sweet preference was classified as Sweet Liker or Sweet Disliker; craving was classified as high or low. These categories defined four participant profiles. The registry describes the study’s principal aim as testing whether the combination of sweet-preference status and initial craving predicts drinking outcomes during naltrexone treatment.

The primary outcome was percent heavy drinking days over the 12 weeks of medication. A second stated aim concerned percent abstinent days, also described in the study summary. The registry record does not report results for either aim, so it cannot establish whether any of the participant characteristics predicted either outcome, or whether a prediction was specific to naltrexone rather than placebo. Nor does it report an estimated treatment effect for the primary outcome.

This is therefore a question about treatment-response prediction, not simply whether participants with different taste preferences or craving levels had different outcomes. To answer the treatment-response question, an analysis would need to distinguish prognostic differences that might occur in either group from differences in outcome associated with receiving naltrexone rather than placebo. The registry does not report the analysis used to make that distinction.

3. Trial Design

The registry describes a randomized, quadruple-masked, Phase 2 study with a factorial model. Enrollment was 80, and the study status is completed. The recorded start was 2010-01 and the primary completion was 2012-06. The sponsor was University of North Carolina, Chapel Hill. Alcohol dependence and alcohol abuse are listed as conditions.

The study organized participants by sweet-taste response and initial alcohol craving: Sweet Liker, High Craver; Sweet Liker, Low Craver; Sweet Disliker, High Craver; and Sweet Disliker, Low Craver. Within each described profile, half were assigned to naltrexone and half to placebo. All participants were described as receiving Brenda Therapy Sessions. The registry does not provide participant counts for each profile or report the allocation method beyond stating that the study was randomized.

Quadruple masking indicates that the registry records masking at four roles or levels, but it does not identify those roles or describe how masking was maintained. The registry also does not specify whether the factorial label refers to the statistical model, the assignment structure, or both. Sweet preference and craving are participant characteristics in the arm descriptions; the record does not say they were randomized.

The primary completion date is recorded, but details about recruitment, follow-up completion, deviations from assigned treatment, and the number of participants included in each outcome analysis are not reported in the registry record. The listed enrollment should not be treated as the number analyzed for the primary endpoint. The record documents the intended treatment comparison and subgroup structure, but leaves important operational details and the outcome analysis unavailable.

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: Subjects were screened and, if met criteria, randomized to placebo or naltrexone generally within 5-10 days.

Overall Study

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Percent Heavy Drinking Days, measured over 12 weeks.

The primary endpoint is Percent Heavy Drinking Days, measured as a mean percentage of days over the entire medication period, with a timeframe of 12 weeks. The registry says it was derived from the Timeline Follow-back (TLFB). A heavy drinking day is defined as 5 or more standard drinks for a man and 4 or more standard drinks for a woman. A standard drink is described as 12-14 grams of ethanol or the amount contained in a 12 oz beer, 5 oz of wine or 1 1/2 oz of hard liquor.

Expressing the outcome as a percentage relates heavy drinking days to the medication period rather than reporting only a count of such days. The measure can therefore describe the proportion of observed days meeting the heavy-drinking definition. It is a continuous outcome in the registry’s outcome classification, and the specified summary is a mean with standard deviation as the dispersion measure.

The endpoint was selected to address the study’s question about whether sweet preference and initial craving predict drinking patterns during treatment. The registry also identifies Percent Days Abstinent as a secondary outcome. It does not report the primary endpoint’s observed mean, standard deviation, group-specific results, or an estimated comparison. As a result, the measurement definition is available, but the magnitude and precision of the endpoint results are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry classifies the primary outcome as continuous and identifies the mean as its parameter of interest, with standard deviation as the dispersion measure. The outcome is a percentage of days, derived from TLFB reports across the medication period. A mean describes the average outcome in a group. A standard deviation describes how spread out individual observations are around that average. They provide different information: the mean concerns the group’s central level, while the standard deviation concerns participant-to-participant variability.

What would a treatment comparison estimate?

A randomized comparison of naltrexone with placebo could estimate the difference in mean percent heavy drinking days between assigned groups, provided the analysis and outcome data were available. Random assignment is intended to make the groups comparable on average at baseline, although it does not ensure that every participant characteristic is balanced in a particular trial. The registry does not report an estimated comparison or its uncertainty, so no conclusion about the observed treatment contrast can be drawn from the posted record.

Because the trial addresses prediction, a treatment contrast alone would not fully answer its central question. A predictive factor is informative when the relationship between treatment assignment and outcome varies according to that factor. In statistical terms, an analysis may examine interaction between treatment and participant characteristics. This differs from a prognostic association, in which a characteristic is related to the outcome regardless of assigned treatment. The registry does not report whether interaction terms were fitted, how the four profiles were compared, or which model was used.

Why does the factorial label need detail?

The record labels the model factorial and describes four combinations of sweet-preference and craving categories, with naltrexone or placebo within each profile. That description identifies the factors relevant to the research question, but it does not specify the fitted model or the contrasts used. The sweet-preference and craving categories are participant characteristics in the arm descriptions. The registry does not say that those characteristics were assigned experimentally. Their observed associations can therefore be affected by differences among participants, even in a randomized medication comparison.

For example, an analysis could represent sweet preference and craving as separate predictors, include a combined profile, or test whether the association of one characteristic depends on the other. Those choices answer different questions and can use the available observations differently. This is a general statistical distinction; the registry record does not report which approach was used. It also does not provide coefficients or other model results that would make the distinction empirically assessable.

How should uncertainty be assessed?

A confidence interval communicates uncertainty around an estimate. A narrower interval generally indicates greater precision under the model’s assumptions, while a wider interval indicates less precision. A point estimate without an interval can conceal substantial uncertainty, and a p-value alone does not communicate the size of an effect. No confidence interval, p-value, or primary-outcome estimate is reported in the registry record. Therefore, neither the magnitude nor the precision of any primary comparison can be evaluated.

General statistical considerations also matter for a percentage outcome. Percentages are bounded, and their distributions can be asymmetric or concentrated near a boundary. Whether those features are important depends on the observed data. Some models assume a particular residual distribution; if the data vary more than the model allows, standard errors and inferential results may be unreliable. Over-dispersion is one name for variability greater than a model’s assumptions permit. These are general considerations, not reported properties of this trial’s data. The registry does not identify a distributional model, diagnostic checks, or any adjustment for non-constant variance.

What analysis details are available?

The record identifies the primary outcome, its mean parameter, standard deviation, and 12-week timeframe. It describes two aims in the study summary: one involving percent abstinent days and one involving percent heavy drinking days. It does not report a statistical analysis plan, model specification, covariate adjustment, handling of repeated measurements, or methods for missing outcome values. It also does not describe interim analyses. These details should not be inferred from the randomized or factorial design labels.

The registry’s planning assumptions describe a planning illustration based on the observed effect, using a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; the effect to detect is described as a stated fraction of the observed effect. Such assumptions, by themselves, do not supply an observed treatment result or demonstrate that a trial had adequate power. Planning illustrations are not observed power, and the registry does not report a trial-specific sample-size justification in the statistical plan.

For a complete assessment, readers would need the primary-outcome summaries, analysis population, model and contrast definitions, uncertainty estimates, and information on missing observations. The registry record does not report these items. The available design details explain the intended question, but they are not a substitute for the analysis needed to evaluate it.

10. Multiplicity, Missing Data and Analysis Populations

The study summary names two principal aims: prediction of Percent Days Abstinent and prediction of Percent Heavy Drinking Days. The record also identifies Percent Days Abstinent as a secondary outcome and Percent Heavy Drinking Days as the primary outcome. Testing more than one outcome or several predictors can create multiplicity: the more statistical questions evaluated, the greater the chance that at least one result appears notable by chance if no adjustment or hierarchy is used. This is a general statistical concern, not a claim about what occurred in this trial.

The registry does not report a multiplicity strategy, such as a prespecified testing hierarchy, adjusted significance thresholds, or a correction across outcomes and subgroup contrasts. It also does not report which comparisons were treated as confirmatory or exploratory. Because no analyses are listed, it is not possible to determine how the outcome aims or the four participant profiles were handled in the analysis.

Missing-data methods and analysis populations are not reported in the registry record. There is no account of outcome completion, the amount or pattern of missing TLFB information, or how missing days were treated in the percentage calculation. The record does not specify whether analyses followed assignment regardless of treatment received, included only participants with complete outcomes, or used another population definition. Enrollment of 80 does not establish the number contributing to any particular outcome analysis.

These omissions limit interpretation in distinct ways. Multiplicity methods affect how evidence across several questions should be judged; missing-data methods affect which observations contribute and what assumptions support the estimate; and population definitions clarify to whom an estimate applies. Without these details and the outcome results themselves, the registry cannot establish the robustness of any finding or the representativeness of the analyzed participants.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: heart attack (2.5%).

The registry reports serious adverse event counts for Placebo and Naltrexone, with 40 participants at risk in each group. The serious adverse event count is 1 for Placebo and 0 for Naltrexone. The record lists Heart Attack as a top serious term at 2.5%. It does not provide further detail about event timing, individual cases, relatedness to treatment, or how the term-level percentage was calculated.

Other adverse event counts are listed as 2 for Placebo and 3 for Naltrexone, again with 40 at risk in each group. The registry does not give a breakdown of these events by type or severity in the information reported. The death entries are null, so deaths are not reported in the registry record; null should not be interpreted as confirmation that no deaths occurred.

These are event counts, not estimates of the primary drinking outcome. Small differences in counts should be interpreted cautiously, especially without event definitions, follow-up detail, or an analysis of uncertainty. The safety information is descriptive in the registry record and does not establish a causal difference between treatment groups. It also cannot be used to draw conclusions about the efficacy or predictive aims of the trial.

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

The registry record supports a description of the study’s design and intended question: participants were randomized, masking was recorded as quadruple, and treatment groups received naltrexone or placebo alongside Brenda Therapy Sessions. The study was organized around sweet-preference and craving profiles and specified Percent Heavy Drinking Days over 12 weeks as the primary endpoint. Those details clarify what the trial set out to examine.

The record does not report the primary outcome results or any analyses. It therefore does not show whether mean percent heavy drinking days differed between treatment groups, whether outcomes differed across sweet-preference or craving profiles, or whether either characteristic predicted a different response to naltrexone. It cannot establish the direction, magnitude, or precision of any such association. Lack of a reported result is not evidence that the result was null.

The stated aims should also not be mistaken for demonstrated prediction. A characteristic may be associated with an outcome in both treatment groups without identifying who responds differently to a treatment. Assessing treatment-specific prediction requires an analysis that compares the treatment-outcome relationship across characteristic levels. The registry does not report that analysis or its findings.

Similarly, the randomized design does not make every comparison in the study causal. Randomization applies to treatment assignment; the record describes sweet preference and craving as participant profiles, not randomized interventions. Associations involving those characteristics may reflect other differences among participants. The available record does not include enough model or participant information to assess that possibility.

The safety counts provide limited descriptive information, but they do not fill the gap in primary outcome reporting. Overall, the registry documents a completed trial and its planned research question, yet its posted results are insufficient to judge whether the hypotheses were supported. Conclusions about treatment benefit, lack of benefit, or useful prediction require results that are not reported in the registry record.

13. Why This Trial Matters Statistically

This trial addresses a statistically important distinction in treatment research: identifying an overall average treatment contrast is not the same as identifying characteristics that predict variation in treatment response. The study’s question combines sweet-taste preference and initial alcohol craving, with an interest in whether those participant profiles are informative for outcomes under naltrexone. That question requires attention to treatment-by-characteristic relationships, not merely separate descriptions of the treatment groups or the profiles.

The design also illustrates why a factorial label needs to be interpreted alongside the actual assignment and analysis details. The registry describes four combinations of sweet preference and craving, and a naltrexone or placebo assignment within each. Those categories can organize comparisons, but each additional contrast can affect precision and complicate interpretation. A prespecified analysis is important for making clear which interactions and subgroup comparisons are central.

For continuous percentage outcomes, the choice of summary and model matters. Means and standard deviations offer understandable group descriptions, but analysis should also account for the outcome’s distribution and the way observations were collected. These are general methodological considerations; the registry does not report how they were addressed in this trial.

The record’s missing outcome estimates highlight a practical issue in trial interpretation. Registration details can define the question and help readers understand a design, but evaluating evidence requires the observed outcomes, denominators, uncertainty estimates, and analysis methods. Without them, the distinction between a carefully stated aim and a demonstrated finding remains essential. The trial is statistically noteworthy for its predictive question; the available registry record does not provide the results needed to resolve it.

14. 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 consequential limitation for interpretation is that the registry record does not report results for the primary endpoint. It gives no group means, standard deviations, treatment contrast, confidence interval, or p-value for Percent Heavy Drinking Days. Although the study status is completed, completion alone does not make the trial’s outcome findings available. The registry lists no analyses.

The record also does not report outcome denominators, participant flow, or counts within the four sweet-preference and craving profiles. Enrollment of 80 does not establish how many participants were randomized, completed the medication period, contributed TLFB data, or entered a particular analysis. Missing-data frequency and handling are not reported, making it impossible to assess whether the observed outcomes, if analyzed, would represent all randomized participants.

Important analytical details are absent, including the model specification, treatment-by-profile contrasts, covariate adjustment, multiplicity procedures, and any diagnostic checks for the continuous percentage outcome. The registry does not say how the factorial label was operationalized or whether the profile categories were analyzed as separate factors or combined categories. It also does not identify the roles covered by quadruple masking.

The primary outcome description defines heavy drinking days by sex-specific thresholds and gives the TLFB measurement approach, but the record does not provide further detail on outcome ascertainment, adherence, or changes during follow-up. The safety entries include counts and a serious event term, but not event timing or relatedness. Taken together, these limitations mean the posted record can describe the trial’s intent and some safety information, but cannot support an evaluation of the primary result or the robustness of the proposed prediction analyses.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Sweet Preference and Alcohol Craving — NCT01296646.
• Registry record last updated 2013-11-25; status last verified 2013-10.
• Results first posted 2013-11-25.
• 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.

18. Record Summary

NCT01296646, titled Sweet Preference and Alcohol Craving, is a completed randomized, quadruple-masked Phase 2 study sponsored by University of North Carolina, Chapel Hill. Enrollment was 80. The registry records a start of 2010-01 and primary completion in 2012-06.

Participants were grouped by sweet-taste preference and initial alcohol craving, with naltrexone or placebo assigned within each profile; all were described as receiving Brenda Therapy Sessions. The primary endpoint was mean Percent Heavy Drinking Days over 12 weeks, derived from TLFB. Percent Days Abstinent was listed as a secondary outcome.

The registry reports limited adverse event counts but no primary outcome estimates or analyses. The results needed to assess treatment differences or whether sweet preference and craving predict response are not reported in the registry record.