Starting Treatment With Agonist Replacement Therapies (START): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Starting Treatment With Agonist Replacement Therapies (START) (NCT00315341), a randomized phase 4 study. The primary outcome is Hepatic Safety.

2. Clinical Question

The registry describes a randomized, parallel-group comparison of buprenorphine/naloxone and methadone. Its stated context is a Food and Drug Administration request to compare the treatments on indices of hepatic safety. The primary outcome classifies participants according to whether liver transaminase measurements stayed within, rose above, or moved below a threshold based on the upper limit of normal.

This is a question about patterns of laboratory measurements during treatment, rather than a single measurement at one visit. The endpoint description refers to baseline and any and all subsequent samples. The population specified for the endpoint is restricted to evaluable subjects who remained in treatment for the full 24-week period and provided at least 4 liver-function blood samples. The record does not report how many participants met those requirements in either group.

The registry record does not report the category counts or a treatment-group comparison for this endpoint. It therefore does not establish from the posted primary outcome information whether the observed patterns differed between groups, nor how precisely any difference was estimated.

3. Trial Design

START is recorded as a Phase 4 randomized trial with a parallel-group model and no masking. Its planned comparison was between buprenorphine/naloxone and methadone. The registry lists an enrollment of 1269 and identifies the condition as opiate-related disorders. The sponsor is the University of California, Los Angeles. The recorded study start is 2006-04, primary completion is 2010-08, and the status is completed.

Random allocation means that assignment to the study groups is determined by a chance-based process; in general, this is intended to reduce systematic differences between groups at the point of assignment. The registry record does not report the sequence-generation method, whether allocation was concealed before assignment, or the procedures used to implement randomization. Masking is explicitly recorded as none, so the trial was open label in the sense that masking was not used. The record does not specify which people knew the assigned treatment or how awareness was handled in outcome assessment.

The parallel design means participants were assigned to a treatment group rather than being described as crossing between treatments. The registry includes arm descriptions for dosing, but those descriptions are incomplete in the record. Further details about treatment adjustment, adherence, discontinuation, and other aspects of treatment delivery are not reported in the registry record.

Although the enrollment figure is recorded, it should not be treated as the number contributing to the primary outcome. The outcome population required remaining in treatment for 24 weeks and providing at least 4 blood samples. Group-specific numbers meeting those criteria, and the numbers excluded from the analysis, are not reported in the registry record. A sample-size justification and the trial's own statistical analysis plan are also not reported.

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: evaluable subjects stayed in treatment for 24 weeks and gave at least 4 blood samples for liver function tests during the treatment period

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 Hepatic Safety, measured over 24 Weeks.

The primary endpoint is labeled Hepatic Safety and is measured in participants over 24 Weeks. It uses blood levels of ALT and AST, two liver transaminases, and compares baseline measurements with any and all later samples. Participants are placed in one of four categories. Category A describes both enzymes starting at or below 2 times the upper limit of normal (ULN) and remaining at or below that level. Category B describes either enzyme starting at or below 2 times ULN and exceeding that level at some point. Category C describes either enzyme starting above 2 times ULN, both enzymes later falling below 2 times ULN, and remaining below that level. Category D describes either enzyme starting above 2 times ULN and remaining above that level throughout the study.

As general context, repeated laboratory measurements can describe whether an abnormality appears, persists, or resolves during follow-up. Categorizing measurements can make those patterns easier to describe, but it also reduces continuous laboratory values to a small set of groups. People close to a threshold may be classified differently because of small measurement changes, while the categories do not show the full size of changes in ALT or AST. The registry record does not report the category counts, so the distribution of these patterns cannot be assessed from the recorded primary outcome information.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The methods information in the registry record is limited. It records random assignment, a parallel-group structure, no masking, one primary outcome measure, and one analysis. It does not describe the statistical model, comparison procedure, covariate adjustment, handling of repeated measurements, or criteria for interpreting the primary result. The explanations below describe general statistical ideas and distinguish them from the methods actually documented for START.

What does randomization contribute?

In general, randomization assigns participants using chance rather than clinician or participant choice. This can reduce systematic differences in measured and unmeasured characteristics between groups at the start of a trial. It does not guarantee that the groups will be identical, and it does not correct problems caused by missing outcome data, nonadherence, or differences in measurement. The registry records randomized allocation but does not report how the sequence was generated, how assignments were concealed, or whether randomization was stratified.

START is recorded as parallel and unmasked. In a parallel trial, each group receives its assigned treatment over the comparison period. In an unmasked trial, masking is not used; the registry does not specify exactly who was aware of assignments. General concerns in an open-label setting include whether treatment knowledge affects behavior, follow-up, or assessment. These are potential considerations, not evidence that bias occurred in this trial.

What does the endpoint's categorization mean?

The primary measure uses repeated ALT and AST observations to assign each evaluable participant to one of four patterns. In general, categorical outcomes summarize observations into named groups. This can make the main patterns easier to communicate, but it omits detail about the exact values and the timing or magnitude of changes. Here, the definitions distinguish measurements that remain at or below 2 times ULN, cross above that threshold, begin above it and later remain below it, or remain above it. The category definitions are reported; the category frequencies are not.

The endpoint description refers to baseline and any and all subsequent samples. Repeated observations from the same participant are related rather than independent. General analyses of repeated measurements need to account for that relationship if they use the measurements directly. A participant-level category can instead summarize a sequence into one outcome, but the registry does not explain the statistical procedure used to compare those participant-level categories. It also does not state whether the four categories were analyzed as a single multinomial outcome, through separate contrasts, or by another method. No such procedure should be assumed.

What can be said about the analysis plan?

The record lists one analysis, but does not report its test, model, estimand, adjustment variables, or inferential framework. It does not provide a p-value, treatment-effect estimate, or confidence interval for the primary comparison. General statistical interpretation would normally consider both the estimated group contrast and its precision. A p-value, when reported, addresses compatibility with a specified statistical model and null hypothesis; it does not measure the size or importance of an effect. A confidence interval, in general, conveys uncertainty around an estimate. Neither can be interpreted for this primary endpoint from the registry information.

The trial's sample-size justification is not reported in the registry record. The record does state assumptions for a planning illustration: a two-sided significance level of 0.05, equal allocation between arms, a normal approximation, and an effect to detect defined as a stated fraction of the observed effect. Such calculations are planning illustrations, not observed power. They do not establish the trial's original design rationale, and they cannot replace the actual outcome estimate and its uncertainty. The numerical planning estimates are not reported in the registry record.

In general, power is a design concept describing the chance that a study will detect a specified effect under specified assumptions. It is not a property that can be read from a nonsignificant result, and a planning calculation based on an observed effect is not evidence that the trial was prospectively powered for that effect. The registry does not report the original assumptions used to design the trial, or whether interim analyses were planned or conducted.

Why can over-dispersion matter?

General count-data models sometimes assume that variability follows a particular relationship to the average count. Over-dispersion means the observed variability is greater than that assumption allows. If it is ignored, uncertainty can be understated and inferential results can be unreliable. The primary START outcome is described as a categorical participant count, but the registry does not identify a count model or report a dispersion parameter. There is therefore no basis for saying that over-dispersion was present or that a particular correction was used.

How do missing observations affect interpretation?

The endpoint population requires staying in treatment for 24 weeks and providing at least 4 blood samples. This makes completion and measurement availability part of the analysis population definition. In general, excluding people without sufficient follow-up can affect a comparison if the reasons for missingness are related to treatment or outcome. The record does not state how many participants failed to meet the criteria, why observations were absent, whether missing values were imputed, or whether sensitivity analyses were conducted. It also does not report an intention-to-treat analysis or another analysis population beyond the evaluable-subject definition.

10. Multiplicity, Missing Data and Analysis Populations

The registry record identifies one primary outcome measure and one analysis. At the same time, the primary outcome contains four categories, and the record notes that it was not reduced to one comparison. The statistical handling of those categories is not reported in the registry record. It is therefore not possible to determine which contrasts were tested, whether all categories contributed to one overall test, or whether any adjustment for multiple comparisons was used.

Multiplicity arises in general when several hypotheses or comparisons are examined. Testing several outcomes or contrasts can increase the chance of a false-positive finding unless the analysis plan and interpretation account for that structure. The presence of four outcome categories alone does not establish how many statistical tests were performed. The registry does not provide enough analysis detail to assess the multiplicity implications for START.

The primary endpoint population consisted of evaluable subjects who stayed in treatment for 24 weeks and gave at least 4 blood samples for liver function tests during treatment. This is the only population definition described for the primary outcome. The numbers meeting the criteria, the reasons other participants may not have met them, and the handling of missing measurements are not reported in the registry record. Imputation, sensitivity analyses, and a full randomized-population analysis are also not reported.

These omissions limit assessment of whether the reported analysis, if any, represents everyone randomized and how robust it would be to missing data or alternative analysis choices. The record does not report interim analyses, a multiplicity strategy, or a statistical analysis plan. No conclusions about those procedures should be inferred from the count of listed outcomes or analyses.

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 1% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hospitalization (8.7%), disability (0.8%), life-threatening (0.6%), death (0.4%), congenital anomaly (0.1%).

The registry lists serious adverse event figures separately for the buprenorphine/naloxone and methadone groups. For buprenorphine/naloxone, it records 38 serious events affected among 727 at risk. For methadone, it records 45 serious events affected among 520 at risk. These are the posted group figures; they are not the primary hepatic-safety endpoint results. No group-specific death count is recorded in the serious-event entries.

The record also lists serious-event terms with percentages: hospitalization, 8.7%; disability, 0.8%; life-threatening, 0.6%; death, 0.4%; and congenital anomaly, 0.1%. The registry does not specify in this information how those percentages are denominated or attribute each term to a treatment group. They should therefore be read as listed serious-event terms and percentages, not as group-specific risks or as a basis for calculating a treatment comparison.

In general, serious adverse event summaries describe reported events under the record's definitions; they do not by themselves establish that treatment caused an event. The registry does not report event timing, relatedness assessments, or a comparative statistical analysis for these figures. They should not be combined with, or treated as a substitute for, the missing ALT and AST category results.

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

The registry record allows a clear description of the trial question: a randomized comparison of buprenorphine/naloxone and methadone on a hepatic-safety outcome over 24 weeks. It also defines how participants were to be classified using ALT and AST measurements. Those details describe the study's purpose and endpoint; they do not show the primary results.

The record does not report how many participants fell into each hepatic-safety category, the treatment-group comparison, or an estimate with an uncertainty interval. As a result, it cannot establish from the posted primary outcome information whether the groups differed, the direction or size of any difference, or the precision of a comparison. Neither a claim of benefit nor a claim of no difference can be drawn from the endpoint definition or trial status alone.

Even if results were available, their interpretation would depend on the analysis population and missing data. The endpoint applies only to evaluable subjects who stayed in treatment for 24 weeks and provided at least 4 blood samples. The record does not report how many participants met that definition by group or how participants without sufficient follow-up were handled. The outcome categories also summarize transaminase patterns rather than showing every measured value or the magnitude of any change.

More generally, a statistically detectable group difference and a clinically important difference are not interchangeable concepts. Statistical uncertainty concerns how compatible the data are with a range of effects; clinical importance requires context about the outcome and the consequences of the observed patterns. Since the primary results are not reported in the registry record, neither kind of judgment can be made from the posted primary endpoint information.

The serious adverse event entries provide separate descriptive information, but their denominators and comparative analysis are not fully described. They cannot fill the gap in the primary endpoint reporting. The appropriate reading is that the registry documents a completed comparison and its planned hepatic-safety classification, while the results necessary to judge that comparison are not reported in the registry record.

13. Why This Trial Matters Statistically

START is statistically notable for defining hepatic safety through repeated measurements rather than a single laboratory reading. The endpoint uses baseline and subsequent ALT and AST values to classify patterns over 24 weeks. This approach captures whether values remain within a threshold, cross it, or move from above it to below it. In general, a repeated-measure pattern can provide different information from a single measurement, although the categorical summary loses some detail about the actual values and timing.

The endpoint also illustrates why analysis-population definitions matter. The primary population requires remaining in treatment for the full follow-up period and providing at least 4 blood samples. Such criteria are relevant to interpreting who contributes to an outcome and may distinguish the analysis group from everyone randomized. The record does not provide the group counts meeting these criteria, so the scope of the analyzed population cannot be quantified.

The randomized, parallel-group design gives the comparison a clear structure, while the absence of masking is a consideration when assessing measurement and follow-up procedures. Neither design feature by itself determines the result. Reliable interpretation also requires transparent outcome counts, a defined comparison, and an account of uncertainty. Those primary results and methods are not reported in the registry record.

The record also shows the importance of separating endpoint definitions from endpoint findings. A detailed classification rule can make a measure understandable, but it cannot answer whether the treatment groups differed without the category results and analysis. The serious adverse event figures are another distinct layer of information and do not supply the missing primary outcome comparison. For statistical readers, the key lesson is to distinguish what the trial planned to measure from what its posted results permit one to conclude.

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.

A specific limitation is that the primary outcome is defined for evaluable subjects who stayed in treatment for 24 weeks and provided at least 4 blood samples. The registry record does not report group-specific numbers for this population, the reasons participants may not have qualified, or how missing measurements were handled. This prevents assessment of how the evaluable group relates to the randomized groups.

The outcome uses four threshold-based categories derived from ALT and AST measurements. In general, categorization can obscure the size and timing of laboratory changes and can make classification sensitive to values near a threshold. The record does not report the underlying measurement distributions or category counts, so those issues cannot be evaluated for this trial.

The trial is recorded as unmasked. The record does not describe who was aware of assignment, how blood sampling was scheduled, whether laboratory assessment was blinded, or how outcome classification was checked. These details would help readers assess possible sources of measurement or follow-up differences, but they are not reported in the registry record.

The arm descriptions are incomplete in the registry record, and detailed information about dose adjustment, adherence, discontinuation, and treatment received is not available in the listed details. The analysis method, statistical plan, sample-size justification, interim analyses, and multiplicity procedures are also not reported. Most importantly, the primary outcome counts and treatment comparison are absent. These limitations are specific to what the registry record documents and do not establish that the underlying trial lacked such procedures or data.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Starting Treatment With Agonist Replacement Therapies (START) — NCT00315341.
• Registry record last updated 2017-01-06; status last verified 2016-11.
• Results first posted 2017-01-06.
• 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

START is a completed, Phase 4 randomized parallel-group trial with no masking, sponsored by the University of California, Los Angeles. It recorded enrollment of 1269 and compared buprenorphine/naloxone with methadone in people with opiate-related disorders. The registry lists a study start of 2006-04 and primary completion of 2010-08.

The primary outcome, Hepatic Safety, classifies ALT and AST patterns over 24 weeks. Its evaluable population required remaining in treatment for 24 weeks and providing at least 4 blood samples. The registry record defines four categories but does not report their group-specific results or a treatment comparison. It also does not report the statistical analysis method, uncertainty estimates, missing-data procedures, or multiplicity approach. Serious adverse event figures are listed separately and do not replace the missing primary outcome results.