Chronic Pain Master Protocol (CPMP): A Study of LY3016859 in Participants With Osteoarthritis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Chronic Pain Master Protocol (CPMP): A Study of LY3016859 in Participants With Osteoarthritis (NCT04456686), a randomized phase 2 study. The primary outcome is Change From Baseline in Average Pain Intensity as Measured by the Numeric Rating Scale (NRS).

2. Clinical Question

3. Trial Design

This was a randomized, double-masked, parallel-group Phase 2 clinical trial comparing LY3016859 with placebo in participants with osteoarthritis. Random allocation is intended to make treatment groups comparable on average, while masking is intended to reduce the influence of treatment knowledge on participants’ reports and other trial assessments. The registry does not report the allocation ratio or describe how the randomization sequence was generated or concealed.

The LY3016859 regimen started at 750 mg and continued with 500 mg intravenous infusions every 2 weeks, for a total of 4 doses. The placebo group received intravenous infusions on the same stated schedule. The registry lists Eli Lilly and Company as the sponsor and marks the study completed. The recorded start date is 2020-07-01, and the primary completion date is 2021-04-22.

Enrollment was 117. The primary outcome analysis description specifies participants who received at least one dose, and says that the overall number analyzed includes participants with a non-missing Week 8 value. The posted arm counts for the primary outcome are 63 for LY3016859 and 31 for placebo. These counts should be read as the endpoint analysis counts, not as a replacement for the overall enrollment figure. The registry does not report why the endpoint counts differ from enrollment or provide a detailed account of participants’ progress through the trial.

The study’s stated purpose was to evaluate safety and efficacy for osteoarthritis pain. It was part of a chronic pain master protocol described in the registry. Details of the broader protocol do not establish additional features of this particular trial’s design; information not specified for this trial is not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All enrolled or randomized participants who received at least one dose of study drug. Here, the overall number of participants analyzed includes the number of participants with non-missing value at Week 8.

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: All enrolled or randomized participants who received at least one dose of study drug.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in Average Pain Intensity as Measured by the Numeric Rating Scale (NRS), measured over Baseline, up to Week 8.

The primary endpoint was change from baseline in average pain intensity measured by the Numeric Rating Scale. Participants described their average pain over the past 24 hours on a scale from 0 to 10, where 0 means no pain and 10 means pain as bad as you can imagine. The measure was used during the preliminary data-entry period and daily throughout the study. The stated timeframe for the primary outcome was baseline through Week 8.

A change-from-baseline endpoint expresses how a participant’s score at follow-up differs from their own starting score. A group mean summarizes these changes, while an estimate comparing groups addresses whether their average changes differed. This endpoint is a continuous measure: it can retain differences in reported symptom intensity rather than reducing responses to categories such as improved or not improved. In general, repeated symptom ratings can capture variation in pain over time, though the registry does not describe how daily reports were combined to define the analysis measure beyond the stated endpoint.

The registry describes the primary result as a posterior mean change and a 95% credible interval from a Bayesian mixed model repeated-measures analysis. The arm-level mean changes were -2.04 for LY3016859 and -2.0 for placebo. The reported modelled contrast was -0.03, with a 95% credible interval from -0.77 to 0.70. The registry does not report a separate threshold for a clinically important change on this scale.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary analysis is described as a Bayesian mixed model repeated-measures analysis. That description identifies a model-based analysis of repeated observations and a Bayesian approach to uncertainty. The registry reports a posterior mean change for each arm and a posterior mean difference between arms, along with 95% credible intervals. It does not report enough detail to reconstruct the model or independently assess its assumptions.

What does the mean change describe?

The mean is the average of a continuous measure across participants in a group. Here, the measure is change from baseline in average pain intensity on the NRS, and the reported group means are -2.04 for LY3016859 and -2.0 for placebo. These are within-group summaries. A change from baseline is calculated at the participant level before being summarized, but the registry does not provide the individual observations or the exact handling of daily scores.

The group contrast answers a different question: how the model estimates the average change in one randomized group relative to the other. The reported posterior mean difference is -0.03. A contrast can be much closer to zero than either group’s change from baseline when both groups have similar average changes. That distinction is important: the within-group values do not establish that the treatment groups differed by those amounts.

How should the credible interval be read?

The 95% credible interval for the difference extends from -0.77 to 0.70. In general, a Bayesian credible interval represents an interval containing a stated proportion of the posterior distribution for a parameter, conditional on the model, data, and prior assumptions. It differs in interpretation from a confidence interval, which is defined through the behavior of an estimation procedure over repeated samples. The record’s primary outcome description explicitly identifies the interval as a credible interval. It also states that the Bayesian analysis uses posterior probabilities instead of p-values.

The interval includes values on both sides of zero. This means that the posted model result does not isolate the difference to one direction. The interval is also a reminder that the posterior mean is not the only relevant feature of the analysis: uncertainty around that mean affects how narrowly the result can be interpreted. The record does not give posterior probabilities for specific values, so such probabilities should not be inferred from the interval alone.

What does a repeated-measures model do in general?

When measurements are collected repeatedly from the same participants, those measurements are related rather than independent. A repeated-measures model can account for the structure of observations over time and estimate group patterns while using repeated outcome information. In general, such models can accommodate incomplete measurement patterns under assumptions about the missingness process. Those general properties do not establish which exact model structure or missing-data assumptions were used in this trial.

The registry does not report the covariance structure, fixed and random effects, covariates, prior distributions, convergence checks, or sensitivity analyses. It does not describe whether the model used every daily NRS response or a derived summary for each visit. Those details are not reported in the registry record, so the model’s fit and robustness cannot be fully evaluated from the posted summary.

Why is a simple recalculation not the same as the reported model?

A simple unadjusted recalculation from posted summary statistics would not reproduce a Bayesian repeated-measures model. The model may use repeated observations and assumptions that are not represented in an arm-level mean alone. In this record, values are missing or not estimable for a two-arm comparison of the continuous endpoint using the simple recalculation approach. No recalculated contrast should therefore be substituted for the reported Bayesian analysis.

What do planning illustrations mean?

In general, a sample-size illustration asks how many participants per arm might be needed under specified assumptions to detect a chosen effect with a chosen power. The registry’s stated illustration assumptions use a two-sided significance level of 0.05, equal allocation, a normal approximation, and effects expressed as fractions of the observed effect. Such planning quantities are not observed trial outcomes, and they are not observed power. The registry does not report that these assumptions were the trial’s original design basis or provide a sample-size justification in this record.

10. Analyses Reported by the Sponsor

The sponsor’s reported primary analysis is the Bayesian mixed model repeated-measures comparison: a posterior mean difference of -0.03, with a 95% credible interval from -0.77 to 0.70. It is this model-based result that should anchor interpretation of the primary endpoint.

A basic unadjusted recalculation from posted summary statistics is not available for this two-arm continuous endpoint because the registry record marks the values as missing or not estimable for that comparison. Such a recalculation would also not be equivalent to the sponsor’s model, which is described as Bayesian and repeated-measures. No alternate recalculated treatment contrast is reported, and the record does not identify a discrepancy between the sponsor’s analysis and a recalculation.

The arm means, -2.04 for LY3016859 and -2.0 for placebo, provide context about within-group average change. They should not be mistaken for the sponsor’s between-group estimate. The modelled contrast and its credible interval address the treatment-group comparison directly. A posterior mean near zero accompanied by an interval spanning values on either side of zero indicates uncertainty about the direction and size of the difference, rather than proving that the groups were identical.

The registry does not report enough model detail to reproduce the sponsor’s analysis independently. The prior distributions, covariates, repeated-measures structure, and sensitivity analyses are not reported in the registry record. The posted result can be interpreted as stated, while recognizing that its statistical assumptions cannot be fully examined from the available description.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 9 posted secondary outcomes in addition to the primary endpoint. They cover WOMAC pain, stiffness and physical function; patient global impression of change; worst pain intensity; pain on a visual analog scale; sleep; and rescue medication use. The record also indicates 10 outcome measures and 10 analyses. It does not report a multiplicity adjustment, a hierarchical testing plan, or another procedure for controlling the chance of false-positive findings across outcomes. Without such information, secondary results should be interpreted as separate analyses rather than as a prespecified family with a known error-control strategy.

The primary outcome is described as analyzed among participants who received at least one dose of study drug, with the overall number analyzed including participants with a non-missing Week 8 value. The primary endpoint counts posted by arm are 63 for LY3016859 and 31 for placebo. The overall enrollment is 117. The record does not give a participant flow account explaining the difference between enrollment and the endpoint counts, nor does it describe the amount or pattern of missing outcome data.

Mixed models can, in general, use available repeated observations under assumptions about missingness. Whether that approach is appropriate depends on the model and the missing-data process. The registry does not report the assumptions used, imputation procedures, sensitivity analyses for missing outcomes, or how missing daily ratings were treated. Therefore, the potential effect of missing data on the result cannot be assessed from this record alone.

The adverse-event populations have different posted counts from the primary endpoint counts: the serious and other adverse-event summaries use 77 participants at risk in the LY3016859 group and 36 in the placebo group. These are safety figures and should not be substituted for the endpoint analysis population. The registry does not explain the relationship between the analysis populations. Detailed analysis-population definitions beyond those stated, and the protocol’s multiplicity and missing-data plans, are not reported in the registry record.

12. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 3% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: supraventricular tachycardia (1.3%), nephrolithiasis (1.3%).

The registry reports serious adverse events for a safety population of 77 participants at risk in the LY3016859 group and 36 at risk in the placebo group. Serious adverse events affected 2 participants in the LY3016859 group and 0 in the placebo group. The listed serious-event terms are supraventricular tachycardia and nephrolithiasis, each reported as 1.3%. The record does not specify in the registry record how those terms were distributed by treatment group, so they should not be assigned to an arm.

Other adverse events affected 18 participants in the LY3016859 group and 11 in the placebo group. Deaths were reported as 0 in each group. These counts describe events recorded during the trial; they do not by themselves establish that treatment caused an event or that an event was unrelated to treatment. The registry summary does not provide event-level timing, severity details beyond the serious-event classification, exposure-adjusted rates, or a causal assessment.

The at-risk counts differ from the primary endpoint analysis counts. Safety summaries and efficacy analyses can use different populations and definitions, so the figures should be read within their stated context rather than combined. The registry does not report further detail explaining the difference. The event counts are descriptive and should not be treated as a complete assessment of safety or as proof of a between-group safety difference.

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

The primary result is an estimated difference in average pain-score change between LY3016859 and placebo through Week 8. The Bayesian model reports a posterior mean difference of -0.03 and a 95% credible interval from -0.77 to 0.70. The interval covers values in both directions, so the posted estimate does not resolve whether the group difference favored one arm or the other. Its width also signals uncertainty about the magnitude of any difference.

The result does not mean that every participant’s pain changed by the group average. It does not show that the groups had exactly equal outcomes, because the interval allows a range of differences. Nor does it establish a treatment effect beyond the stated population, endpoint, and follow-up period. The record does not give a clinically important threshold for interpreting score differences, so the clinical importance of values within the interval cannot be classified using this record alone.

Both arm-level means are changes from baseline, not comparisons against an untreated natural history. A within-group change can reflect multiple influences, including time and trial participation, and is not an estimate of the randomized treatment contrast. The between-group modelled result is the relevant posted comparison for the primary endpoint.

The registry lists secondary measures, but the primary result should not be generalized to those measures without their own estimates and uncertainty. The record does not report the secondary outcome results in the registry record for this summary, or how multiplicity was addressed. Likewise, the primary endpoint findings do not resolve safety questions. The adverse-event counts are descriptive, and their interpretation is limited by the detail reported.

Overall, the result is a model-based estimate with uncertainty, not a definitive statement that the treatments are identical or that one treatment does or does not work. Its interpretation depends on the Bayesian model and its assumptions, which are not fully documented in the registry record.

14. Why This Trial Matters Statistically

This trial illustrates why a randomized comparison should be interpreted through the between-group estimate rather than through change within either arm alone. The posted mean changes were -2.04 for LY3016859 and -2.0 for placebo, while the reported modelled difference was -0.03. The within-group summaries and the randomized contrast answer different statistical questions. Keeping them separate prevents improvement from baseline in both groups from being misread as evidence of a difference between groups.

It also shows the importance of presenting uncertainty alongside a point estimate. The 95% credible interval from -0.77 to 0.70 includes values on both sides of zero, so the posterior mean should not be treated as a precise, conclusive description of the group difference. A single estimate cannot convey the range of values compatible with the model and data.

The analysis also raises practical questions about reproducibility and interpretation. The registry names a Bayesian mixed model repeated-measures method but does not report the priors, model structure, or sensitivity analyses. Those details matter because model-based results depend on assumptions. The missing detail does not invalidate the posted result, but it limits independent evaluation of how robust it is.

Finally, the record distinguishes an efficacy endpoint population from safety counts and lists multiple secondary outcomes. Those features make clear why population definitions, missing-data handling, and multiplicity plans are important. Their absence from the registry description leaves limits on what can be inferred from the summarized results. These are general statistical considerations illustrated by the reporting available for this trial, not conclusions about the study’s conduct.

15. Limitations

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

The primary endpoint counts are smaller than the overall enrollment figure, and the registry describes the endpoint analysis as including participants who received at least one dose and had non-missing Week 8 data. It does not provide a participant flow explanation or clarify the reasons values were unavailable. The safety at-risk counts are different again, without an explanation of how the safety and primary analysis populations relate.

Although the primary result is identified as Bayesian, the registry does not report prior distributions or detailed model specifications. It does not describe the repeated-measures covariance structure, covariates, model diagnostics, or sensitivity analyses. Without those details, independent reproduction and evaluation of robustness are limited.

The primary result concerns average pain intensity measured by the NRS through Week 8. The record does not provide a clinically important difference threshold, so the clinical meaning of the range of differences in the credible interval cannot be determined from this record alone. Information about longer follow-up is not reported in the registry record.

The registry lists 9 secondary outcomes, but the multiplicity strategy is not reported. Secondary findings, if considered, could not be interpreted as a controlled set of tests using details in this record. The record also does not report full event-level safety information, exposure-adjusted summaries, or a causal assessment of adverse events.

Finally, the enrollment figure does not establish the original target or the basis for the trial’s design. A sample-size justification, interim-analysis plan, and detailed statistical analysis plan are not reported in the registry record. Planning illustrations based on observed effects are not evidence of the study’s original design choices or observed power.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Chronic Pain Master Protocol (CPMP): A Study of LY3016859 in Participants With Osteoarthritis — NCT04456686.
• Registry record last updated 2023-11-14; status last verified 2023-11.
• Results first posted 2022-06-23.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT04456686 is a completed randomized, double-masked, parallel-group Phase 2 trial of LY3016859 versus placebo in participants with osteoarthritis. The registry reports enrollment of 117 and lists Eli Lilly and Company as sponsor. The active regimen began with 750 mg followed by 500 mg intravenous infusions every 2 weeks for a total of 4 doses; placebo was infused on the same stated schedule.

The primary outcome was change from baseline in average pain intensity on the NRS through Week 8. The reported Bayesian mixed model posterior mean difference was -0.03, with a 95% credible interval from -0.77 to 0.70. The arm-level mean changes were -2.04 for LY3016859 and -2.0 for placebo. The interval spans values in both directions and conveys uncertainty around the estimated contrast.

The registry lists secondary outcomes and descriptive safety counts, but does not report detailed multiplicity handling, missing-data sensitivity analyses, or full model specifications. Those omissions limit independent assessment of the analysis and the wider set of outcomes.