Does Duloxetine Reduce Sub-Acute Pain After Knee Arthroplasty?: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Does Duloxetine Reduce Sub-Acute Pain After Knee Arthroplasty? (NCT02005601), a randomized phase 4 study. The primary outcome is NRS Pain With Ambulation at 2 Weeks.

2. Clinical Question

The posted results address a focused question about pain while walking at a defined postoperative time, rather than pain in every setting or across the full recovery period. The comparison is between group averages. It does not show how individual participants’ scores changed, how many participants experienced a particular degree of improvement, or whether participants considered any score difference important. Those details are not reported in the registry record.

3. Trial Design

This was a Phase 4, randomized, quadruple-masked, parallel-group trial. The registry record reports enrollment of 106 participants and gives an equal number in each arm: 53 assigned to duloxetine and 53 to control. The record marks the trial as completed. Its listed start is 2013-11 and its primary completion is 2015-05.

A randomized design assigns participants to treatment groups by chance, which is intended to support a fair comparison of outcomes between groups. Masking is intended to limit the influence of knowledge about assignment on experiences, assessments, or study conduct. The record labels the trial quadruple-masked but does not identify which groups of people were masked or describe how masking was maintained. These are details not reported in the registry record.

The parallel model means the record presents separate groups receiving the duloxetine and control regimens, with outcomes compared between those groups. The duloxetine arm description gives a preoperative capsule followed by daily capsules from POD1 through POD14. The control description says 0 mg of duloxetine; the record also lists placebo as an intervention, but does not give the comparator’s dosing schedule or explain the relationship between the listed placebo and control descriptions.

The sponsor is Hospital for Special Surgery, New York. The registry record does not report the trial’s original sample-size justification, the statistical analysis plan, details of eligibility criteria, or the intended analysis population. The record reports a completed status and posted outcome summaries, but it does not include posted analyses for the primary or secondary endpoints. This distinction matters: design labels and outcome summaries do not by themselves disclose the methods used to obtain inferential 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.

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 NRS Pain With Ambulation at 2 Weeks, measured over 2 weeks after surgery.

The primary endpoint is NRS pain with ambulation at two weeks after surgery. It records participants’ descriptions of pain in the knee undergoing or having undergone surgery, during ambulation in the last 24 hours. The stated scale runs from 0, meaning no pain, to 10, meaning pain as bad as the participant can imagine. The registry reports the group mean and standard deviation for this continuous measure.

Using a rating scale gives a way to summarize a subjective experience numerically and compare average scores between randomized groups. The endpoint is specific to pain during movement and to the stated assessment time. It should not be interpreted as a measure of pain at rest, pain at other times, overall recovery, or every participant’s individual experience.

For this endpoint, the posted group means are 3.5 in the duloxetine arm and 3.8 in the control arm. The standard deviations describe the spread of scores within each group; they are not confidence intervals for the group means. The record does not report the distribution of individual scores, a threshold for an important difference, or a responder analysis. Accordingly, the mean comparison is informative about the average score but cannot alone describe how many participants had a meaningful change.

7. Results Posted in the Registry

• Duloxetine: mean 3.50 versus 3.80 on control; difference -0.30 (95% CI -1.14 to 0.54), standardized effect d = -0.14; z-test p = 0.483.

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.

Total Daily Opioid Use (mg Oral Morphine Equivalents)

• 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 deviations.
• Result:
  • Duloxetine: mean 57.80 versus 72.70 on control; difference -14.90 (95% CI -27.01 to -2.79), standardized effect d = -0.47; z-test p = 0.016.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Duloxetine than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Nausea Severity

• 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 deviations.
• Result:
  • Duloxetine: mean 0.90 versus 2.30 on control; difference -1.40 (95% CI -2.49 to -0.31), standardized effect d = -0.49; z-test p = 0.012.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Duloxetine than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes provide additional context about the posted results, but they do not change the primary endpoint’s uncertainty. The tables report comparisons for total daily opioid use in oral morphine equivalents and nausea severity, both at POD1. In each comparison, the duloxetine group’s posted mean is lower than the control group’s mean. This is a descriptive pattern across the two outcomes, not by itself a conclusion about why the outcomes differ or how they relate to pain two weeks after surgery.

The primary question concerns NRS pain with ambulation two weeks after surgery, whereas these secondary measures concern different outcomes at an earlier postoperative time. Agreement in the direction of group means does not establish that changes in these measures explain the primary result. Nor does it establish a general effect across other outcomes or assessment times. The registry record does not report analyses connecting the secondary outcomes to the primary endpoint.

The secondary comparisons are unadjusted. General statistical guidance is that when several outcomes are examined, nominal p-values can overstate the strength of evidence if the possibility of multiple testing is not addressed. The registry record does not report a multiplicity adjustment, the planned hierarchy among outcomes, or the analysis plan. These results should therefore be read as exploratory context rather than as independent confirmation of the primary comparison.

The posted summaries also do not include endpoint-specific participant counts beyond the arm summaries, individual outcome distributions, or details on how the measures were assessed. Those limitations make it important not to overread the numerical pattern. The tables show what was recorded and the uncertainty displayed for each comparison; they do not establish a broader causal explanation or settle whether the differences are important.

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

• Total Daily Opioid Use (mg Oral Morphine Equivalents)
• Nausea Severity

9. Statistical Interpretation

10. Statistical Methods Explained

The registry record identifies the primary outcome as a continuous measure summarized by a mean and standard deviation. The comparison shown is a mean difference between the duloxetine and control groups. These are familiar ways of summarizing a numerical rating score: the mean represents the group’s average, while the standard deviation describes how dispersed individual observations are around that average.

What does a mean difference represent?

A mean difference subtracts one group mean from the other according to a specified ordering. In the posted comparison, the estimate is -0.3 points, with the duloxetine group’s mean lower than the control group’s. The sign is meaningful only when the group ordering is understood. A negative difference does not itself tell whether the magnitude is important, and the estimate does not imply that each participant experienced a change of that size.

The group standard deviations are 2.1 for duloxetine and 2.3 for control. These describe variation among the recorded scores; they do not describe uncertainty in the estimated difference. The confidence interval addresses uncertainty in the comparison estimate, whereas the standard deviations describe the observed spread within each arm. Confusing these quantities can make group variability look like precision of the estimated treatment contrast.

How should the confidence interval be read?

A confidence interval gives a range of values compatible with the data and the method used to calculate it, under the method’s assumptions. The interval for the primary comparison extends from -1.138 to 0.538. It includes zero, the value representing no mean difference. It also includes negative and positive values, so the posted data do not pin down the direction with confidence. The registry record does not describe the interval-construction method, so its stated endpoints can be interpreted, but the underlying calculation cannot be independently assessed from this record alone.

A confidence interval is not a guarantee that the true value lies inside the displayed range, and it is not a range containing a known proportion of participants’ scores. It refers to uncertainty in the estimated group difference. Its width reflects limited precision, but deciding whether values in that range are important requires a context-specific threshold; one is not reported in the registry record.

What does the p-value add?

The posted unadjusted p-value is 0.483. In general, a p-value is calculated by comparing the observed result with results expected under a null hypothesis and a statistical model. A larger p-value is not proof that there is no difference, and it does not quantify the probability of a treatment effect. It is best read alongside the estimated difference and confidence interval, not as a standalone verdict.

The adjective unadjusted indicates that the displayed p-value is not presented as corrected for multiple comparisons. The registry record does not report a multiplicity procedure or a detailed analysis plan. It also does not post a separate primary analysis record. The displayed result therefore should not be assumed to incorporate adjustments or analysis choices that are not described.

Why do assumptions matter?

Statistical comparisons rely on assumptions about how observations are generated and how uncertainty is represented. For a mean comparison, considerations can include the distribution of scores, whether observations are independent, and how missing measurements are treated. General statistical guidance is that substantial skew or unusual observations can affect the usefulness of a mean and a standard deviation, while unequal variability or dependence can affect standard errors. The registry record does not give participant-level data or enough methodological detail to assess these matters for this trial.

Randomization supports comparison of groups, but it does not eliminate uncertainty, guarantee balance on every characteristic, or settle how missing outcomes should be handled. The record does not report baseline characteristics, participant flow after enrollment, or endpoint-specific analysis counts. Although 53 participants are listed in each arm, those counts alone do not confirm that all contributed primary outcome measurements.

What is the role of the posted secondary comparisons?

The registry reports means, standard deviations, mean differences, confidence intervals, and unadjusted p-values for two secondary outcomes. These are descriptive and inferential summaries, but the record lists no posted analysis records for them. General statistical guidance is to treat several unadjusted comparisons cautiously because examining more outcomes creates more opportunities for a nominally unusual result. The record does not say whether the secondary outcomes were prespecified in a formal analysis plan or whether any multiplicity strategy was used.

Finally, the sample-size illustrations are distinct from the analysis of the completed trial. They describe hypothetical planning based on the observed effect and stated assumptions. They do not change the uncertainty in the observed comparison or provide evidence that the completed study had a particular probability of detecting an effect. The original design assumptions are not reported in the registry record.

11. Confidence Intervals and Precision

The primary confidence interval ranges from -1.138 to 0.538 points for the duloxetine-versus-control mean difference. Because it crosses zero, the posted estimate is compatible with a lower average score in the duloxetine group, no mean difference, or a higher average score in that group. The point estimate of -0.3 is only one value within this broader range, not a precise statement of the true difference.

Precision refers to how narrowly the data constrain an estimate. The interval’s span illustrates that the trial does not resolve the direction of the primary group difference. The estimate is based on 53 participants in each arm, and the observed standard deviations are 2.1 and 2.3. The record does not report endpoint-specific sample sizes, so the arm enrollment figures should not be taken as confirmation that all participants had an observed primary score.

Confidence intervals are most useful when read against a meaningful range of values. The score itself ranges from 0 to 10, but the registry record does not identify what size difference would be clinically important. Without such a threshold, it is possible to describe the interval’s direction and uncertainty, but not to determine from the record whether every value it contains would be important or unimportant.

The interval also does not show the distribution of individual outcomes. It describes uncertainty in the average difference, not the range of scores among participants and not the likelihood of benefit for a particular person. The posted result is therefore best summarized as an estimated mean difference of -0.3 with substantial uncertainty about its direction and magnitude. That is a statement about precision, not a conclusion about individual experiences.

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 table and curve are planning illustrations based on the observed effect, not the trial’s original design calculation. The registry record’s stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The planning rows consider detecting 100%, 75%, and 50% of the observed effect. The trial’s original sample-size justification and design assumptions are not reported in the registry record.

For these continuous outcomes, the table gives participants per arm, not a combined total and not total events. Read each row as a hypothetical effect size to detect and each power column as the number of participants per arm indicated by the planning calculation. The 80% power column describes a design that would have an 80% chance of producing a statistically significant result under the assumed effect and model, if those assumptions held. The 90% power column describes the corresponding planning target at a higher chance of detection. These are design probabilities under assumptions, not guarantees about an actual study.

The rows show that smaller effects need more participants to detect under the same planning framework. The curve provides a visual way to see how the required sample size changes as the target effect changes; it should be read as an illustration of the calculation, not as additional trial evidence. The figures are not observed power. They do not describe the probability that the completed trial’s result is correct, and they do not replace the confidence interval for the observed comparison.

Because the calculation uses an observed effect as its starting point, it is a retrospective planning illustration rather than proof that the trial was designed around that effect. The table should not be used to infer what assumptions the investigators used before enrollment; those assumptions are not reported in the registry record.

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 = -0.300; pooled standard deviation σ = 2.202.
2. n = 2 × (1.96 + 0.842)² × 2.202² / -0.300² = 845.9, rounded up to 846 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 <- -0.3                      # observed difference in means
sigma <- 2.20227155455452                     # 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(2005601)
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 = 846 (80% power), 1,133 (90% power)
 75% of observed effect: n per arm = 1,504 (80% power), 2,014 (90% power)
 50% of observed effect: n per arm = 3,384 (80% power), 4,530 (90% power)
power.t.test: 846.9 
Simulated power at n = 846 per arm: 0.797 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study examines more than one outcome, comparison, time point, or analysis. In general, the more statistical tests considered, the greater the opportunity for at least one result to appear unusual by chance. This does not make secondary outcomes invalid, but it affects how much weight should be placed on unadjusted p-values and confidence intervals when several results are read together.

The registry record lists one primary outcome and two secondary outcomes. It reports unadjusted comparisons, but does not describe a multiplicity adjustment, a formal testing hierarchy, or which outcomes would be considered confirmatory. It also lists zero reported analysis entries. The statistical analysis plan is not reported in the registry record. Consequently, readers cannot determine whether the posted comparisons follow a prespecified inferential strategy or how the investigators intended to interpret multiple outcomes.

Missing-data handling is also not reported. The registry gives 53 participants in each arm, but does not provide endpoint-specific analysis counts, the number of missing scores, reasons for missingness, or methods for handling missing observations. General statistical guidance is that missingness can affect a comparison if the availability of measurements is related to group assignment or outcome. Without participant flow and analysis details, the impact in this trial cannot be assessed from the posted record.

The analysis population is not reported in the registry record. It is therefore unclear from the posted summaries whether the estimates use all randomized participants, only participants with available outcome measurements, or another population. The record also does not describe whether any protocol deviations or exclusions affected the analyses. These details matter for judging how closely the analyzed groups preserve the original randomized comparison.

These absences should be treated as limits on interpretation, not as evidence that particular procedures were or were not used. The available record supports description of the posted summaries, while leaving the multiplicity strategy, analysis population, and missing-data approach unresolved.

15. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry reports zero serious adverse events among the 53 participants at risk in the duloxetine group and zero among the 53 participants at risk in the control group. It also reports zero deaths in each arm. These figures describe the serious adverse events recorded in the registry for the groups shown.

A count of zero observed events does not establish that an event is impossible or that the groups have identical safety profiles. With no serious events recorded in either arm, this trial’s registry summary does not provide a comparative estimate of serious-event risk. It also does not describe the types of serious events that were monitored beyond listing no top serious terms.

The record includes other adverse-event counts separately, but the serious-event figures should not be combined with those counts. The registry summary does not provide enough detail to assess event definitions, timing, severity criteria, or how events were collected. These data are a limited account of reported events in the trial, not a complete characterization of safety.

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

The primary posted comparison is a mean NRS pain score with ambulation at two weeks after surgery. The duloxetine group’s average was lower than the control group’s, but the confidence interval includes zero and extends to values in both directions. The unadjusted p-value is 0.483. Taken together, these results do not give a precise estimate of the direction or size of the group difference.

This does not mean that the groups have been shown to be equivalent. Demonstrating equivalence or noninferiority requires a prespecified margin and an analysis designed for that question; such a margin and analysis are not reported in the registry record. Conversely, a lower point estimate in one arm is not enough to establish a meaningful difference. The record does not state what size change would be important to participants or clinicians.

The endpoint describes average self-reported pain during ambulation over the last 24 hours at a specified time. It does not answer questions about pain at rest, other stages of recovery, individual responses, or outcomes not listed in the registry. The secondary outcomes add information about other measures at POD1, but they cannot substitute for the primary result or establish a mechanism connecting outcomes.

The findings should also be read in light of reporting gaps. The registry record does not report the analysis population, missing-data methods, detailed statistical plan, or multiplicity strategy. Those omissions limit assessment of how the posted inferential results were generated and how multiple results should be weighed.

Overall, the record supports a cautious description: the primary point estimate favors a lower average score in the duloxetine group, while the uncertainty interval includes no difference and values in either direction. It does not provide a definitive answer about the size, importance, or individual relevance of any difference. This is a statistical interpretation of the posted results, not medical advice.

17. Why This Trial Matters Statistically

This trial is statistically useful as an example of why a point estimate should not be read alone. The primary mean difference is -0.3, but its confidence interval extends across zero. Looking at both the estimate and its interval makes the uncertainty visible: the posted result points in one direction, yet the data do not narrowly determine the underlying group difference.

The trial also illustrates the distinction between within-group variability and precision of a comparison. The group standard deviations describe the spread of pain scores, while the confidence interval describes uncertainty in the difference between group means. Both kinds of information matter, but they answer different questions. Neither tells whether a difference would be important without an appropriate clinical threshold, which is not reported in the registry record.

Its secondary outcome summaries demonstrate another general issue: several unadjusted comparisons can create a collection of results that looks more conclusive than the underlying evidence warrants. The registry lists two secondary outcomes but does not report a multiplicity strategy or analysis plan. Readers therefore need to distinguish the primary question from additional descriptive findings and avoid treating every nominal result as independent confirmation.

Finally, the sample-size illustrations show how planning depends on the effect one aims to detect. Smaller target effects require more participants under the stated planning assumptions. Those illustrations are not a reconstruction of the investigators’ original design and are not observed power. The record does not report the original sample-size justification, so the calculations should be used to understand planning principles rather than to judge the trial against an unknown target.

18. Limitations

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

The registry record does not give eligibility criteria or detailed participant characteristics. This limits the ability to determine how closely the enrolled participants represent any broader group of people undergoing TKA. The sponsor is listed as Hospital for Special Surgery, New York, but the record does not establish from that fact alone where all trial activity took place or how widely the results may apply.

The control arm is described as receiving 0 mg of duloxetine, and placebo is listed as an intervention. However, the comparator’s dosing schedule and the details of matching or masking are not reported. The trial is labeled quadruple-masked, but the identities of the masked groups and the procedures used to maintain masking are not described. These gaps make it difficult to assess the practical implementation of the design from the registry record alone.

For the primary endpoint, the record reports group means and standard deviations but no participant-level distribution, responder definition, or threshold for a meaningful difference. Endpoint-specific sample sizes and missing-data handling are also not reported. Thus, it is unclear whether the arm enrollment counts correspond to the number of observations used in every outcome comparison.

The statistical analysis plan, original sample-size justification, analysis population, and multiplicity approach are not reported. There are no posted analysis entries. These omissions constrain assessment of whether the displayed results match prespecified methods and how the secondary results should be weighted. They do not prove that such methods were absent; they mean they cannot be evaluated using this registry record.

Finally, the primary endpoint is restricted to pain with ambulation at two weeks after surgery, and the secondary outcomes shown are at POD1. The registry summaries do not describe outcomes across other times or settings. The results therefore answer a narrow set of questions and should not be generalized beyond those measures without further evidence.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Does Duloxetine Reduce Sub-Acute Pain After Knee Arthroplasty? — NCT02005601.
• Registry record last updated 2017-08-29; status last verified 2017-07.
• Results first posted 2017-08-29.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

22. Record Summary

The registry record describes a completed, randomized, quadruple-masked, parallel Phase 4 trial of duloxetine after total knee arthroplasty. It reports enrollment of 106 participants, with 53 in each arm. The primary endpoint was mean NRS pain with ambulation two weeks after surgery.

The posted primary mean difference was -0.3 points, with a confidence interval from -1.138 to 0.538 and an unadjusted p-value of 0.483. The interval includes zero, leaving uncertainty about the direction and size of the group difference. Two secondary outcomes at POD1 are also posted, but the registry record does not report a multiplicity strategy or analysis plan.

The record reports zero serious adverse events and zero deaths in each arm. It does not report the original sample-size justification, analysis population, or missing-data methods. The posted summaries describe the trial results, while those reporting gaps limit fuller evaluation.