A Study for the Treatment of Diabetic Peripheral Neuropathic Pain: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study for the Treatment of Diabetic Peripheral Neuropathic Pain (NCT00552175), a randomized phase 3 study. The primary outcome is Change From Baseline at Week 12 in Average Pain Severity Rating Using Diaries for the Combined Duloxetine Arms (40 mg + 60 mg).

2. Clinical Question

3. Trial Design

This was a phase 3, randomized, quadruple-masked, parallel-group study sponsored by Eli Lilly and Company. The registry reports enrollment of 339 participants and a completed status. The study began in 2007-11, and primary completion was recorded in 2009-03. Randomization indicates that treatment assignment was allocated by chance; quadruple masking indicates that four categories of people were masked, but the registry information does not identify those categories or describe how masking was maintained.

The intervention options were duloxetine 40 mg, duloxetine 60 mg, and placebo, each taken orally every day according to the arm descriptions. The primary outcome pooled the two duloxetine doses rather than presenting the main comparison by dose. For that endpoint, the registry reports 171 participants in the combined duloxetine group and 167 in placebo. Those outcome counts are not a replacement for the overall enrollment figure: the primary outcome population is described as Full Analysis Set participants with a post-baseline measurement in the variable being presented.

The registry record does not report the original sample-size justification, the assumptions used in the original design, or whether interim analyses were planned or conducted. Detailed information about randomization procedures, allocation concealment, masking procedures, protocol deviations, and the statistical analysis plan is not reported in the registry record. The posted information identifies a sponsor-reported mixed models analysis for the primary comparison, but it does not give enough detail to reconstruct that model or establish which adjustments it included.

The parallel-group structure means participants were assigned to a treatment group rather than crossing between treatments. The central posted comparison is the difference in average pain severity change at Week 12 between the combined duloxetine arms and placebo. The study also lists secondary outcomes involving diary scores, patient global impression, and Brief Pain Inventory scores. The registry lists 12 outcome measures and 49 analyses overall, while the detailed results considered in this statistical interpretation cover only the posted outcomes represented in the available results.

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: Full Analysis Set Population consisted of all randomized patients who had at least one post-baseline measure of the primary efficacy variable.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Patients in the Full Analysis Set Population who had a post-baseline measurement in the variable being presented.

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 Change From Baseline at Week 12 in Average Pain Severity Rating Using Diaries for the Combined Duloxetine Arms (40 mg + 60 mg), measured over Baseline, 12 weeks.

The primary endpoint is change from baseline at Week 12 in average pain severity measured using diaries, with the two duloxetine groups combined. Participants rated average pain over the past 24 hours on an 11-point numerical rating scale, from 0, no pain, to 10, pain as severe as you can imagine. The diary ratings were expressed as a weekly mean. A change score compares the Week 12 value with the baseline value, so a more negative change indicates a larger decrease in the recorded score.

This endpoint provides a repeated, patient-reported measure of pain severity over the study period rather than relying on a single retrospective rating at the end. Diary collection can capture ratings across time, while a weekly mean condenses those ratings into a summary for analysis. Those features explain the endpoint’s measurement role; they do not establish that diary responses were complete or free from measurement error. Completion rates and the handling of missing diary entries are not reported in the registry record.

The primary outcome is expressed as a least-squares mean, with a standard error reported for each group. The sponsor’s primary comparison is a least-squares mean difference from a mixed models analysis. The registry identifies the outcome population as Full Analysis Set participants with a post-baseline measurement for the variable being presented. The registry record does not report enough detail to determine how baseline values, repeated measurements, or other covariates were represented in the model.

7. Results Posted in the Registry

• Duloxetine 40 mg and 60 mg Combined: mean -2.47 versus -1.61 on control; difference -0.86 (95% CI -1.36 to -0.36), standardized effect d = -0.37; z-test p = <0.001.

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.

Change From Baseline at Week 12 in Average Pain Severity Rating Score Using Diaries

• 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:
  • Duloxetine 40 mg: mean -2.41 versus -1.61 on control; difference -0.80 (95% CI -1.36 to -0.24), standardized effect d = -0.35; z-test p = 0.005.
  • Duloxetine 60 mg: mean -2.53 versus -1.61 on control; difference -0.92 (95% CI -1.48 to -0.36), standardized effect d = -0.40; z-test p = 0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Duloxetine 40 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Duloxetine 60 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Patient Global Impression of Improvement Scale at Week 12 in Combined Duloxetine Arms (40 mg + 60 mg)

• 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:
  • Duloxetine 40 mg and 60 mg Combined: mean 2.53 versus 3.18 on control; difference -0.65 (95% CI -0.98 to -0.32), standardized effect d = -0.42; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Duloxetine 40 mg and 60 mg Combined than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Patient Global Impression of Improvement Scale at Week 12

• 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:
  • Duloxetine 40 mg: mean 2.53 versus 3.18 on control; difference -0.65 (95% CI -1.01 to -0.29), standardized effect d = -0.45; z-test p = <0.001.
  • Duloxetine 60 mg: mean 2.52 versus 3.18 on control; difference -0.66 (95% CI -1.02 to -0.30), standardized effect d = -0.45; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Duloxetine 40 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Duloxetine 60 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The posted secondary results broadly point in the same direction as the primary comparison: the reported diary-based average pain changes are more negative in each duloxetine dose arm than in placebo, and the patient global impression scores are lower in the duloxetine groups than in placebo. The registry’s patient global impression outcome is reported both for the combined duloxetine arms and by dose. These related findings offer context for the primary result but do not change the status of the combined-arm diary outcome as the primary endpoint.

Secondary outcomes address additional aspects of participants’ reported experience, including worst pain, night pain, global impression of improvement, and Brief Pain Inventory severity and interference. The registry lists 12 outcome measures and 49 analyses, and indicates 11 secondary outcomes as posted. The available detailed results include only some of those outcome entries. Where secondary estimates and sponsor analyses are posted, their interpretation remains specific to the measure, time point, population, and comparison stated in the registry.

Secondary analyses can help assess whether the primary finding sits alongside related patterns in other measures. However, they generally carry less confirmatory weight when many outcomes or comparisons are tested. Testing multiple hypotheses can increase the chance of apparently strong results arising without a corresponding underlying effect. The registry record does not report a multiplicity adjustment or a hierarchy for secondary hypotheses, so the posted unadjusted p-values should not be treated as if they were adjusted confirmatory tests.

The registry record also does not report a complete account of which secondary analyses were planned in advance, how all listed outcomes were handled, or whether any analyses were exploratory. Therefore, the secondary results are best regarded as supportive descriptive context rather than independent proof of the primary finding. Their consistency in direction is notable, but it does not remove uncertainty about multiplicity, missing data, or the methods used for each outcome.

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

• Change From Baseline at Week 12 in Average Pain Severity Rating Score Using Diaries
• Change From Baseline at Week 12 in Worst Pain Severity Score and Night Pain Severity Score Using Diaries for the Combined Duloxetine Arms (40 mg + 60 mg)
• Change From Baseline at Week 12 in Worst Pain Severity Score and Night Pain Severity Score Using Diaries
• Patient Global Impression of Improvement Scale at Week 12 in Combined Duloxetine Arms (40 mg + 60 mg)
• Patient Global Impression of Improvement Scale at Week 12
• Change From Baseline in Brief Pain Inventory Severity Scores at Week 12 for the Combined Duloxetine Arms (40 mg + 60 mg)
• Change From Baseline in Brief Pain Inventory Severity Scores at Week 12
• Change From Baseline in Brief Pain Inventory Interference Scores at Week 12 for the Combined Duloxetine Arms (40 mg + 60 mg)

9. Statistical Interpretation

10. Statistical Methods Explained

Statistical concepts in this explanation are general unless explicitly attributed to the registry record. The primary endpoint is continuous: it is a numerical score with a meaningful range, rather than a count of events or a yes/no outcome. The registry reports least-squares means, which are model-based group estimates. In general, a least-squares mean is an estimate from a statistical model and may account for features represented in that model. The registry names a mixed models analysis for the primary comparison but does not report enough model detail to state exactly which terms, covariance assumptions, or missing-data assumptions were used.

What does a mean difference express?

A mean difference compares the estimated average outcome in one group with that in another. For this endpoint, the sponsor reported a least-squares mean difference of -0.87 for the combined duloxetine group relative to placebo. Because a lower pain severity score represents a lower rating on the scale, a negative difference indicates a lower estimated change value in the combined duloxetine group. The interpretation needs to keep the outcome definition in view: these are changes from baseline, not the absolute Week 12 scores.

The simple unadjusted recalculation uses the posted arm means and their reported standard errors. It gives a difference of -0.86. This is not the same as reproducing the mixed models analysis. The registry also gives a pooled standard deviation estimate for the recalculation and marks it as approximate. In general, converting standard errors to an approximate standard deviation depends on assumptions about how the summaries were generated and the analysis population. The recalculated comparison is therefore a transparent, summary-statistic comparison, not an independent reconstruction of the sponsor’s full model.

What does a standard error describe?

A standard error describes the estimated sampling variability of an estimate under the assumptions of its method. It is not the spread of individual participant scores. The arm-level standard errors for the primary result are 0.18 in each group. A smaller standard error generally corresponds to a more precise estimate, all else equal. It does not capture every source of uncertainty: for example, it does not automatically address bias from missing data, limitations of outcome measurement, or unreported analysis choices.

How should a confidence interval be read?

The sponsor reported a 95% confidence interval from -1.17 to -0.56 for the primary difference. In general, a confidence interval describes a range of estimates compatible with the data and the assumptions of the statistical method. Its width communicates precision: a narrower interval indicates less sampling uncertainty than a wider interval, on the same scale and under comparable methods. The interval does not mean that 95% of participants had changes within its limits, nor does it assign a 95% probability to the specific fixed effect being inside this one calculated interval.

The sponsor interval does not include zero. In general, when a two-sided confidence interval for a difference excludes zero, that is consistent with a statistically significant difference at the corresponding confidence level under the analysis assumptions. This does not establish clinical importance, eliminate bias, or show that the model assumptions are correct. The registry does not state a clinical importance threshold for this endpoint.

What does a p-value contribute?

The sponsor reported p < 0.001 for the primary mixed models comparison. The unadjusted recalculation reports p < 0.001. In general, a p-value measures how incompatible the observed data, or more extreme data, are with a specified null model, assuming that model and the analysis assumptions hold. It is not the probability that the null hypothesis is true, the probability that the result occurred by chance, or a measure of the size or importance of the estimated difference.

The p-value for the recalculation is labeled unadjusted. That label matters because the registry lists multiple outcomes and analyses. If many hypotheses are tested, the chance of obtaining at least one small p-value can increase. Whether and how the trial controlled multiplicity is not reported in the registry record. Therefore, the primary result and secondary results should not be treated as if all tests had equal evidential status.

Why can modeling matter?

Mixed models are commonly used for continuous outcomes when data have a structure such as repeated observations or clustering; that is a general explanation, not a claim about the exact structure or specification used in this trial. A model can estimate group differences while accounting for correlations represented in its design. Its validity depends on the model specification, data, and assumptions. The registry’s method name alone does not disclose which observations were included at each time, how within-participant correlations were modeled, or how incomplete data were handled. Those details are not reported in the registry record.

For continuous outcomes, over-dispersion is not the usual framing used for count data. General considerations still include whether the outcome distribution and model assumptions are appropriate, whether residual variability is represented adequately, and whether unusual observations have undue influence. The registry results do not report diagnostic checks or sensitivity analyses, so no conclusion about those checks can be drawn.

In short, the sponsor’s model-based difference is the primary reported inferential result; the unadjusted recalculation is a simpler descriptive cross-check. Neither the p-value nor the interval addresses all possible sources of uncertainty. Model details, multiplicity procedures, and missing-data methods are not reported in the registry record.

11. Confidence Intervals and Precision

The primary sponsor-reported comparison is -0.87, with a 95% confidence interval from -1.17 to -0.56. The interval is entirely below zero and is reasonably informative about the direction of the model-based difference: the values within it all correspond to a more negative change estimate in the combined duloxetine group than in placebo. The interval also gives a range of plausible magnitudes under the mixed models analysis and its assumptions. It is not a range for individual participants’ outcomes.

Precision and statistical significance are related but distinct. The interval excluding zero indicates evidence against a zero difference under the model assumptions, while its width shows that the exact size remains uncertain. A single point estimate should not be read as an exact effect. The interval does not say whether the estimated difference is large enough to matter to patients; a clinically important threshold is not reported in the registry record.

The primary arm standard errors are 0.18 for placebo and 0.18 for the combined duloxetine group. These describe uncertainty in the reported arm means, not the variation in individual ratings. The sponsor’s confidence interval comes from the mixed models analysis, and should be interpreted as model-based. The unadjusted recalculation gives a separate point estimate of -0.86, but it is not a substitute interval or a replication of the sponsor model.

Confidence intervals address sampling uncertainty conditional on the analysis method. They do not automatically include uncertainty about missing observations, the choice of analysis model, measurement quality, outcome reporting, or multiple testing. The registry does not report enough detail to quantify these additional uncertainties. Thus, the interval offers useful precision information for the primary comparison, but it does not exhaust the limits on interpretation.

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 illustration translates the observed primary comparison into hypothetical future-trial planning values. The displayed figures are participants per arm, not total participants and not total events. The calculations assume a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They use an effect equal to a stated fraction of the observed effect. These are planning assumptions for the illustration; the trial’s own original design assumptions and sample-size justification are not reported in the registry record.

For planning at 100% of the observed effect, the illustration gives 117 participants per arm for 80% power and 156 participants per arm for 90% power. At 75% of the observed effect, it gives 207 participants per arm for 80% power and 277 participants per arm for 90% power. At 50% of the observed effect, it gives 465 participants per arm for 80% power and 623 participants per arm for 90% power. The curve should be read as showing the same planning relationship across effect sizes: as the effect assumed for planning becomes smaller, the required number of participants per arm becomes larger.

In general, 80% power means that, if the assumed effect and statistical assumptions were correct, a study of that planned size would have an 80% chance of meeting the specified statistical significance criterion. The 90% column describes the corresponding chance at the larger target. Power is a design property under assumptions, not a guarantee that a future study will produce a particular result. These values are not observed power for the completed trial. They are illustrative planning calculations based on the observed effect, and they may be optimistic if that estimate is larger than the effect a future study should plan to detect.

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 = -0.860; pooled standard deviation σ = 2.340.
2. n = 2 × (1.96 + 0.842)² × 2.340² / -0.860² = 116.2, rounded up to 117 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 <- -0.86                      # observed difference in means
sigma <- 2.34016482935955                     # 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(552175)
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 = 117 (80% power), 156 (90% power)
 75% of observed effect: n per arm = 207 (80% power), 277 (90% power)
 50% of observed effect: n per arm = 465 (80% power), 623 (90% power)
power.t.test: 117.2 
Simulated power at n = 117 per arm: 0.803 (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. Analyses Reported by the Sponsor

The sponsor’s primary analysis is a mixed models analysis with a least-squares mean difference of -0.87, a 95% confidence interval from -1.17 to -0.56, and p < 0.001. This is the trial’s own reported model-based analysis and is the primary result to use when describing the sponsor’s inferential conclusion.

A simple unadjusted recalculation from the posted arm summaries gives a mean difference of -0.86 and p < 0.001. The recalculation is close to the sponsor’s point estimate, but it uses a different method. It is based on the reported group summaries and an approximate pooled standard deviation rather than the sponsor’s mixed model. It does not reproduce adjustments, repeated-measure structure, or other features that may be part of the sponsor’s model; the registry does not provide the model specification needed to establish those details.

The close point estimates are a useful consistency check, not proof that the methods are interchangeable. The sponsor’s confidence interval and p-value belong to the reported mixed models analysis. The recalculation’s unadjusted p-value belongs only to the simpler summary-statistic calculation. The registry record marks no discrepancy, and the sponsor’s own reported analysis remains the appropriate reference for the primary trial result.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists multiple outcome measures and analyses, including several secondary measures and dose-specific comparisons. When many statistical tests are conducted, the probability of observing at least one small p-value can rise even if the hypotheses being tested are all null. General approaches to multiplicity include adjusting significance thresholds, controlling an overall error rate, or specifying a hierarchy of outcomes. The registry record does not report which, if any, approach was used for the primary and secondary analyses. The unadjusted secondary p-values therefore warrant more caution than a single prespecified primary test would ordinarily receive.

The registry names a Full Analysis Set population for the primary endpoint, defined as participants with a post-baseline measurement in the variable being presented. That description makes clear that participants without a post-baseline measurement for the outcome are not represented in that stated population. The registry does not report the numbers missing by group, reasons for missingness, completion rates for diary records, or the method used to address missing observations in the mixed models analysis. It is therefore not possible to judge from the record whether missingness differed between groups or how sensitive estimates might be to alternative assumptions.

The registry also does not report a full statistical analysis plan, the original sample-size justification, interim analysis procedures, or sensitivity analyses for missing data. These omissions limit assessment of whether all analyses followed prespecified plans and how robust the estimates are to analytic choices. They do not, on their own, show that the analyses were inappropriate; they mean that the relevant details cannot be evaluated from the registry record.

Finally, the primary endpoint is patient-reported and diary-based. The registry describes the instrument and scale, but does not report diary completion or validation details beyond that description. Outcome measurement and missingness are distinct concerns: a complete diary may still involve measurement variability, while an incomplete diary creates a separate analysis issue. Both should be considered when interpreting a model-based estimate.

16. 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 4% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: alcohol poisoning (1.2%), contusion (1.2%), ulna fracture (1.2%), hypoglycaemia (1.2%), epiglottitis (1.2%).

The registry reports serious adverse events affecting 6 of 167 participants at risk in the placebo group, 3 of 85 in the duloxetine 40 mg group, and 2 of 86 in the duloxetine 60 mg group. These are group-level counts of participants reported as affected, not a complete account of event timing, duration, severity, or causal attribution. They should be read alongside the number at risk in each group rather than as an isolated count.

The registry also lists the most frequent serious adverse event terms, each reported at 1.2%: alcohol poisoning, contusion, ulna fracture, hypoglycaemia, epiglottitis, and radiculopathy. The provided information does not establish that these events were caused by study treatment. Death counts are not reported in the registry record. The record also does not provide enough safety detail to compare overall adverse-event profiles or to assess statistical uncertainty around these counts.

Serious adverse events are distinct from all adverse events. The registry reports other adverse events affecting 121 of 167 participants in placebo, 71 of 85 in duloxetine 40 mg, and 73 of 86 in duloxetine 60 mg. These counts are descriptive and do not by themselves establish comparative risk, causality, or the clinical meaning of differences between groups. Safety interpretation also depends on event definitions, follow-up, and ascertainment, details not reported in the registry record.

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

The primary result describes a difference in diary-based average pain severity change at Week 12 between the combined duloxetine arms and placebo. The sponsor’s mixed models estimate is -0.87, with a 95% confidence interval from -1.17 to -0.56. The interval lies below zero, and the sponsor reported p < 0.001. These statistics describe the strength and precision of the observed comparison under the reported model; they do not independently establish how important the difference is to participants.

The result does not show the separate primary effect of duloxetine 40 mg or 60 mg, because the primary analysis combines those arms. Nor does it establish that the doses are equivalent or that one dose is better than the other. The secondary dose-specific results are not a substitute for a dedicated dose comparison. The registry record does not report a threshold for a clinically meaningful change, so the numerical result cannot be classified against such a threshold using the information available.

The result also does not resolve uncertainty about missing observations, multiplicity, model specification, or possible measurement limitations. The registry describes the Full Analysis Set as participants with a post-baseline measurement for the presented variable, but details of missing-data handling are not reported. Multiple secondary outcomes are listed, while multiplicity procedures are not reported. These are reasons to keep the interpretation bounded to the posted analysis rather than to treat one p-value as answering every question.

Finally, a registry result is not a complete account of all aspects of a clinical trial. It does not provide all protocol details, participant characteristics, or the full analysis plan in the information. The most defensible statistical statement is that the sponsor-reported Week 12 model estimated a greater decrease in the combined duloxetine group than in placebo, with the stated interval and p-value. This statement is descriptive of the posted data and is not medical advice or a recommendation about treatment.

18. Why This Trial Matters Statistically

This trial illustrates why the estimated group difference, its confidence interval, and the analysis method should be read together. A pair of arm means describes each group, but the contrast between groups is the direct comparison relevant to the randomized design. The sponsor’s reported contrast is accompanied by a confidence interval, which makes the uncertainty in the estimated difference more visible than a p-value alone.

It also illustrates that two analyses of the same outcome can be similar in point estimate while answering through different statistical procedures. The unadjusted recalculation from summaries is close to the sponsor’s mixed models estimate, but it does not reproduce the model. Distinguishing these methods prevents a simple calculation from being mistaken for the trial’s full analysis and clarifies why the sponsor-reported result is the primary inferential reference.

The combined-dose primary endpoint is another important interpretive feature. It provides an estimate for the two duloxetine arms together, while not establishing a separate primary effect for either dose. Related secondary results may add context, but multiple testing and the lack of reported multiplicity procedures limit how strongly they can confirm the primary result.

Finally, the planning illustration demonstrates how sample-size requirements depend on the effect size a future study aims to detect. The values are hypothetical planning outputs based on the observed comparison, not evidence about the completed trial’s power. Together, these features show why statistical significance, precision, analysis assumptions, endpoint structure, and planning assumptions answer different questions and should not be collapsed into a single verdict.

19. Limitations

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

The registry record does not report the original sample-size justification, so the planning assumptions behind enrollment cannot be assessed. The sample-size illustration is based on the observed effect and specified general assumptions, not on the trial’s documented design calculation. The record also does not report the detailed randomization or masking procedures, participant characteristics, eligibility criteria, or a complete protocol in the information.

For the primary result, the registry names a mixed models analysis but does not provide its full specification, including the exact model terms, covariance structure, or approach to incomplete observations. The primary population is described as Full Analysis Set participants with a post-baseline measurement, but missing counts, reasons, and sensitivity analyses are not reported. These gaps limit independent evaluation of the model-based estimate and its robustness.

There are numerous listed outcomes and analyses. The registry record does not report a multiplicity strategy, prespecified hierarchy, or enough information to distinguish all confirmatory from exploratory analyses. Consequently, secondary p-values should be interpreted cautiously. The record lists several secondary outcomes, but detailed results represented in the available information do not cover every listed outcome.

The primary outcome is based on self-reported diary ratings. The registry explains the scale and the weekly mean but does not report diary completion, adherence, or measurement-performance results. The clinical importance threshold for the score change is also not reported. Thus, the statistical evidence can be described, but the degree of patient importance cannot be determined from the registry results alone.

The safety information is limited to counts of participants affected and selected serious event terms. Death counts are not reported in the registry record, and the information does not establish causality or provide a complete assessment of event timing and severity. These registry-specific gaps narrow the conclusions that can be drawn from the posted results.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study for the Treatment of Diabetic Peripheral Neuropathic Pain — NCT00552175.
• Registry record last updated 2010-04-13; status last verified 2010-03.
• Results first posted 2010-04-13.
• 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.

23. Record Summary

NCT00552175 is a completed phase 3 randomized, quadruple-masked, parallel-group study sponsored by Eli Lilly and Company. The registry reports enrollment of 339 participants, a start date of 2007-11, and primary completion in 2009-03. The study compared duloxetine 40 mg and 60 mg with placebo in patients with diabetic peripheral neuropathic pain.

The primary outcome was change from baseline at Week 12 in diary-based average pain severity, analyzed for the combined duloxetine arms versus placebo. The sponsor reported a mixed models least-squares mean difference of -0.87, with a 95% confidence interval from -1.17 to -0.56 and p < 0.001. A simple unadjusted recalculation gave -0.86 with p < 0.001. The registry also reports secondary outcome results and adverse-event counts, but important details about multiplicity, missing-data handling, the full model specification, and the original sample-size design are not reported in the registry record.