Double Blind Trial of Duloxetine in Chronic Fatigue Syndrome: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Double Blind Trial of Duloxetine in Chronic Fatigue Syndrome (NCT00375973), a randomized phase 2, phase 3 study. The primary outcome is Change From Baseline in Multidimensional Fatigue Inventory (MFI)--General Fatigue Subscale Score.

2. Clinical Question

3. Trial Design

The registry describes a randomized, parallel-group, placebo-controlled trial of duloxetine in patients with chronic fatigue syndrome. Masking is recorded as triple, and the official title describes the study as double-blind. The sponsor was the University of Cincinnati. The registry lists the phase as Phase 2 and Phase 3, enrollment as 60, and study status as completed. The recorded start was 2006-09 and primary completion was 2012-06.

The intervention was oral duloxetine at 60-120 mg/day for 12 weeks. The comparator was placebo. The primary endpoint was change from baseline to the 12-week endpoint on the general fatigue subscale of the Multidimensional Fatigue Inventory (MFI). The registry also lists outcomes concerning pain, depression, clinical global severity, patient global improvement, and discontinuation.

For the primary efficacy analysis, the registry reports 27 patients in the duloxetine group and 30 in the placebo group. It says three duloxetine patients were not included: one patient's treatment assignment was unblinded because of a serious adverse event of suicidal ideation, and two patients did not have compliant postbaseline visits. The registry record does not report the trial's original sample-size assumptions, allocation ratio, or a fuller description of the analysis population. Enrollment should therefore not be treated as an analysis denominator.

The registry reports a linear regression analysis for the primary comparison, with a p-value of 0.23. It does not report the corresponding model estimate or interval. A separate simple, unadjusted calculation from the posted group summaries gives a different estimate of uncertainty and should be distinguished from that reported analysis.

4. Participant Flow and Accounting

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

Overall Study

Analysis population for the primary outcome, as defined in the registry: For the efficacy analysis, in the duloxetine group, 3 patients were not included in analysis for the following reasons: one patient's treatment group assignment was unblinded owing to a serious adverse event of suicidal ideation; 2 other patients did not have compliant postbaseline visits.

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 in Multidimensional Fatigue Inventory (MFI)--General Fatigue Subscale Score, measured over Baseline to endpoint at 12 weeks.

The primary endpoint is change from baseline to the 12-week endpoint in the MFI general fatigue subscale. The MFI is a self-reported instrument with 20 statements across five subscales: general fatigue, physical fatigue, mental fatigue, reduced activity, and reduced concentration. Each subscale has four items with five-point Likert scales; scores range from 4-20, with higher scores indicating greater fatigue. A decrease indicates improvement.

The general fatigue subscale focuses on statements about tiredness, feeling rested, and overall feelings of being fit. It is used as the primary measure because it directly captures the fatigue domain the study was designed to assess. As a patient-reported measure, it reflects participants' reported experience rather than an objective test of physical performance.

The outcome is summarized as a mean change, with standard deviations describing variation among the analyzed participants. The group means indicate average change within each arm, while the between-group estimate addresses the comparative question. A within-group decrease by itself does not establish a difference between treatments; the comparative estimate and its uncertainty are needed to assess how well the posted results distinguish the groups.

7. Results Posted in the Registry

• Duloxetine: mean -3.30 versus -1.80 on control; difference -1.50 (95% CI -3.37 to 0.37), standardized effect d = -0.42; z-test p = 0.117.

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 in Brief Pain Inventory (BPI) --Average Pain Severity Score

• 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 -1.60 versus -0.80 on control; difference -0.80 (95% CI -1.80 to 0.20), standardized effect d = -0.41; z-test p = 0.116.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Duloxetine and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline in the Hospital Anxiety and Depression Scale (HADS) --Depression Subscale

• 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 -1.60 versus -1.90 on control; difference 0.30 (95% CI -1.23 to 1.83), standardized effect d = 0.10; z-test p = 0.701.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Duloxetine and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline in the Clinical Global Impression of Severity (CGI-S)

• 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 -1.10 versus -0.40 on control; difference -0.70 (95% CI -1.28 to -0.12), standardized effect d = -0.64; z-test p = 0.017.
• 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 posted secondary results cover changes in average pain severity, the depression subscale of the Hospital Anxiety and Depression Scale, and clinical global severity. Across these measures, the directions and sizes of the group comparisons are not uniform. The secondary findings therefore do not form a single consistent pattern that can be treated as confirmation of the primary fatigue result.

These outcomes address related but distinct aspects of participants' experience. Pain, depressive symptoms, and clinician-rated global severity are not interchangeable with self-reported general fatigue. Secondary results may add context, but they do not replace the primary comparison or directly resolve its uncertainty. The registry lists patient global improvement and discontinuation outcomes as well, but numerical results for those outcomes are not reported in the registry record.

The posted secondary p-values are unadjusted. When multiple outcomes are evaluated, some apparently notable results can arise by chance even if there is no underlying effect on those outcomes. Unless a prespecified multiplicity strategy is reported, isolated secondary p-values carry less weight than a result supported by a clearly described primary analysis. The registry record does not report whether multiplicity was addressed in the trial's analysis plan.

The secondary estimates should consequently be treated as exploratory context rather than a set of independent confirmations. Their interpretation also depends on the analysis populations, missing data, and outcome measurement details, which are not fully reported for every measure. Taken together, the secondary information may help describe what was measured, but it does not overcome the uncertainty in the primary comparison.

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

• Change From Baseline in Brief Pain Inventory (BPI) --Average Pain Severity Score
• Change From Baseline in the Hospital Anxiety and Depression Scale (HADS) --Depression Subscale
• Change From Baseline in the Clinical Global Impression of Severity (CGI-S)
• Patient Global Impression of Improvement (PGI-I)
• Number of Participants Who Discontinued the Study for Any Reason
• Number of Participants Who Discontinued Use of Treatment Due to Adverse Events

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous and is reported as a mean change with a standard deviation in each group. A mean describes the group's average change; a standard deviation describes how spread out individual changes are around that average. These summaries answer different questions. The mean captures the direction and size of average change, while the standard deviation indicates the variability within the group.

What does the between-group estimate represent?

The recalculated mean difference compares the average change in the duloxetine group with the average change in the placebo group. The estimate is -1.5 units. Since decreases indicate improvement on the MFI general fatigue subscale, a negative estimate favors a larger decrease in the duloxetine group. This is a contrast between group averages, not the change for a typical individual, and it does not mean that every participant changed by that amount.

The unadjusted calculation uses the posted group means and standard deviations. It does not incorporate covariates or reproduce the registry's linear regression model. The recalculated estimate and its interval are therefore a descriptive statistical comparison based on the posted summaries, while the registry's regression result is the sponsor-reported analysis. The registry record does not report the regression coefficient or its interval, so the two analyses cannot be compared in every modeling detail.

Why is an interval useful?

An estimate is a single value, but sampling variability means another study of the same question could yield a different estimate. An interval expresses a range of values compatible with the data and the calculation's assumptions. A wider interval signals less precision than a narrower interval. For the primary comparison, the reported interval extends from -3.374 to 0.374. Because it covers values on either side of no difference, the data do not sharply determine the direction of the comparative effect.

The confidence level for this interval is not reported in the registry record. The interval should therefore be interpreted using the stated bounds and the calculation that produced them, without attributing an unreported confidence level. A confidence interval is not a guarantee that the true effect lies within its bounds; it is a statistical procedure for expressing uncertainty under its assumptions.

How should the p-values be read?

The unadjusted p-value for the recalculated primary comparison is 0.117. A p-value measures how inconsistent the observed data, or data more extreme, would be with a specified null model, given the analysis assumptions. It does not give the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The registry's reported linear regression p-value is 0.23. These values arise from different analyses and should not be treated as interchangeable.

The posted p-value for the recalculation is labeled unadjusted. It does not account for the set of primary and secondary outcomes as a family. When multiple outcomes are examined, the chance of obtaining at least one apparently notable result can increase. Whether or how the trial's own analysis addressed multiplicity is not reported in the registry record. This matters especially when interpreting secondary outcomes in isolation.

Why can over-dispersion matter?

General statistical explanation: a simple comparison of means is most informative when the model's assumptions reasonably describe the outcome data. If observations vary more than an assumed model allows, uncertainty can be understated and tests can appear more decisive than warranted. This is often called over-dispersion. The registry reports group standard deviations but does not provide participant-level data or diagnostics that would establish whether over-dispersion affected this trial's analyses. The posted summaries alone cannot resolve that question.

What do the analysis counts mean?

The primary efficacy summaries include 27 duloxetine patients and 30 placebo patients. The registry says three duloxetine patients were excluded from the efficacy analysis: one after unblinding due to a serious adverse event of suicidal ideation and two because they lacked compliant postbaseline visits. The randomized group counts, if different from these reported efficacy counts, are not reported in the registry record. The method used for missing outcomes and the precise analysis-population definition are also not reported. These omissions limit assessment of whether exclusions could have affected the comparison.

In general, missing outcome data can weaken a comparison if the probability of missingness is related to treatment or outcome. An analysis that includes only participants with usable measurements can be biased when those participants differ systematically from those without measurements. This is a general concern, not evidence that bias occurred in this trial. The registry's stated reasons for the exclusions are relevant, but do not by themselves establish the effect of missing data on the estimate.

The statistical methods should thus be read at two levels: the group summaries and unadjusted recalculation show a transparent comparison based on posted values, while the registered regression result records the sponsor's own analysis. Neither removes the need to consider interval width, missingness, analysis choices, and the set of outcomes examined.

11. Confidence Intervals and Precision

The primary recalculated mean difference is -1.5 units, with an interval from -3.374 to 0.374. The range is wide enough to include both a larger average decrease with duloxetine and values consistent with a smaller decrease relative to placebo. This means the point estimate is not a precise determination of the direction or size of the comparative effect.

Confidence intervals express uncertainty around an estimate under a specified procedure and its assumptions. The interval's width reflects sampling uncertainty; it does not describe the full range of possible participant responses or account automatically for bias, missing data, or model misspecification. The confidence level for the recalculated interval is not reported in the registry record, so a confidence percentage should not be inferred from the bounds alone.

The registry's own linear regression analysis reports p=0.23 but does not report its estimated effect or interval. That result cannot be used to assess the model-based effect's precision directly. The unadjusted recalculation and the registered regression analysis answer related, but not identical, statistical questions.

Precision is also constrained by the analyzed group counts and observed variability. The primary standard deviation was 4.2 for duloxetine and 2.8 for placebo. These values describe variation in the analyzed groups; they do not indicate how much variability would occur in another population. The posted estimate should therefore be understood as informative but uncertain, rather than as a precise measurement of a treatment effect.

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 planning figures are participants per arm for a continuous outcome, not total events. They are illustrations calculated from the observed effect, rather than the trial's original design assumptions. The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; each planning scenario uses the stated fraction of the observed effect.

The 80% power column indicates the illustrative number of participants per arm that the calculation associates with an 80% chance of detecting the chosen effect under those assumptions. The 90% power column gives the corresponding planning figure for a 90% chance. These are prospective planning quantities under a model, not probabilities that the trial's conclusion is correct. The registry record does not report the original sample-size justification or the assumptions used when the study was designed.

The rows consider planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. Smaller effects need more participants to detect because they are harder to distinguish from random variation. The estimates can therefore rise substantially when planning for a smaller fraction of the observed effect. They are sensitive to the observed effect used as the starting point, which is itself uncertain in a trial with imprecise results.

These planning figures are not observed power. Observed power calculated after a study is complete largely restates the result and does not resolve uncertainty better than the estimate and interval. The figures are best used to understand how assumptions about effect size and desired power influence planning, not to judge the trial's results by whether a particular target was met.

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 = -1.500; pooled standard deviation σ = 3.532.
2. n = 2 × (1.96 + 0.842)² × 3.532² / -1.500² = 87.0, rounded up to 88 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 <- -1.5                      # observed difference in means
sigma <- 3.53167485376659                     # 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(375973)
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 = 88 (80% power), 117 (90% power)
 75% of observed effect: n per arm = 155 (80% power), 208 (90% power)
 50% of observed effect: n per arm = 349 (80% power), 466 (90% power)
power.t.test: 88 
Simulated power at n = 88 per arm: 0.799 (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 registry reports a linear regression analysis for the primary comparison, with p=0.23. It does not report the regression estimate, its interval, or details sufficient to reconstruct that model. The separate recalculation uses the posted group means and standard deviations and gives a mean difference of -1.5 units, an interval from -3.374 to 0.374, and an unadjusted p-value of 0.117.

The recalculation is a simple, unadjusted analysis of summary statistics. It should not be presented as if it were the sponsor's regression result, nor should the two p-values be expected to match. Differences in model specification and analysis choices can affect estimates and uncertainty. The registry does not give enough information to determine exactly why the reported p-values differ.

The registry's own analysis remains the reported trial analysis. The recalculation offers a transparent, complementary view of the group summaries, but it does not supersede the regression analysis or establish that the regression was incorrect. Because the registry record does not report the regression coefficient or interval, the model-based magnitude and precision remain unclear. Both results should be read with that distinction in mind.

15. Multiplicity, Missing Data and Analysis Populations

The registry identifies a primary fatigue outcome and lists secondary outcomes concerning pain, depression, clinical global severity, patient global improvement, and discontinuation. It reports analyses for the primary endpoint and three secondary measures. The registry also records seven outcome measures and seven analyses. Whether the trial used a prespecified hierarchy or multiplicity adjustment is not reported in the registry record.

Multiplicity matters because examining several outcomes creates multiple opportunities for a result to appear notable by chance. An unadjusted p-value evaluates a particular comparison without accounting for the broader set of tests. This does not make secondary results unusable, but it means they should be interpreted cautiously and in the context of the primary objective. The registry does not report whether any secondary outcomes were designated as confirmatory.

For the primary efficacy analysis, the registry states that three duloxetine patients were not included: one patient's treatment assignment was unblinded following a serious adverse event of suicidal ideation, and two did not have compliant postbaseline visits. The reported primary analysis counts are 27 for duloxetine and 30 for placebo. The record does not report the method used to handle missing measurements, whether any imputation was performed, or a complete accounting of analysis populations by outcome.

Missingness can affect interpretation if the participants with observed outcomes differ from those without them in ways related to treatment or fatigue. The stated reasons for exclusion provide useful context but do not establish whether the resulting analysis is unbiased. The registry record also does not report a sensitivity analysis evaluating alternative assumptions about missing outcomes. These gaps limit assessment of robustness, without demonstrating that the results are necessarily biased.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: suicidal ideation (3.4%).

For serious adverse events, the registry reports 29 duloxetine patients at risk, with 1 affected, and 30 placebo patients at risk, with 0 affected. The leading serious adverse event term listed is suicidal ideation, at 3.4%. The registry also states that one duloxetine patient's treatment assignment was unblinded because of a serious adverse event of suicidal ideation.

These figures describe events recorded in the registry and the group counts at risk; they do not establish that treatment caused or prevented an event. The counts are small, and the record does not provide enough detail to compare the likelihood of uncommon harms precisely. The deaths field is not reported in the registry record.

The registry separately records other adverse events for 29 of 29 duloxetine patients at risk and 30 of 30 placebo patients at risk. That information should not be conflated with the serious adverse event figures. Adverse-event summaries are descriptive and should be interpreted alongside the limited group counts and the incomplete detail available in the registry record.

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

The primary result describes average change in self-reported general fatigue over the 12-week study period among participants included in the efficacy analysis. Both groups had a mean decrease, and the unadjusted between-group estimate was -1.5 units in favor of a larger decrease with duloxetine. The interval extends from -3.374 to 0.374, leaving substantial uncertainty about the comparative effect. The registry's linear regression analysis reports p=0.23.

These data do not establish a clear difference between duloxetine and placebo on the primary measure. They also do not establish that the treatments have identical effects: the interval includes a range of possible comparative values. Statistical uncertainty should not be confused with proof of no effect, just as a point estimate in one direction is not proof of a treatment benefit.

The result concerns one subscale of a self-reported instrument at a 12-week endpoint. It does not by itself answer questions about every fatigue domain, longer-term outcomes, or all people with chronic fatigue syndrome. The registry record does not report a threshold for a clinically meaningful change, so statistical estimates cannot be translated into a conclusion about the importance of the observed difference to patients.

Nor does the primary endpoint alone summarize safety. The serious adverse event counts are descriptive and limited, and they do not establish causation. The trial's exclusions from the efficacy analysis, incomplete missing-data detail, and unreported multiplicity approach further qualify interpretation. The most defensible reading is that the posted results provide an uncertain estimate of a specific comparative outcome, not a definitive broad claim about treatment effects or safety.

18. Why This Trial Matters Statistically

This trial illustrates why the comparison between groups matters more than the direction of change within either group alone. Average fatigue scores decreased in both arms, but the clinical question concerns whether the change differed between duloxetine and placebo. The recalculated between-group estimate is accompanied by an interval that covers values on both sides of no difference, making precision central to interpretation.

It also shows why reported analyses and recalculations should be kept distinct. The registry reports a linear regression p-value, while the summary-based unadjusted calculation gives a different p-value and a directly interpretable mean difference with an interval. Neither should be silently substituted for the other. Statistical reports are more informative when they include the effect estimate, uncertainty, model details, and analysis population, not only a p-value.

The analysis exclusions demonstrate that the population represented by a result depends on how outcomes are handled. The registry explains three exclusions from the duloxetine efficacy analysis, but does not report a full missing-data strategy or sensitivity analyses. Such details help readers assess whether conclusions might depend on assumptions about participants without compliant postbaseline data.

Finally, the multiple secondary outcomes and unadjusted results show why outcome hierarchy and multiplicity planning matter. Secondary measures can add context, but isolated findings are less persuasive when the overall testing strategy is not described. Together, these features make the trial useful for understanding how effect size, precision, analysis choices, and incomplete reporting shape the interpretation of clinical trial results.

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.

Several trial-specific limitations constrain interpretation. The primary efficacy analysis excluded three duloxetine patients, including one whose assignment was unblinded following suicidal ideation and two without compliant postbaseline visits. Although these reasons are stated, the registry does not report a detailed missing-data method or sensitivity analyses. The impact of the exclusions on the primary estimate therefore cannot be assessed from the registry record.

The primary group summaries and recalculated comparison describe analyzed participants, while the registered enrollment is 60. The registry does not provide a complete accounting of randomized participants by group or a full analysis-population definition. The registry's reported linear regression analysis gives p=0.23 but does not provide its coefficient or interval, limiting evaluation of the sponsor-reported model result.

The primary recalculated interval is broad and spans values favoring either direction. The registry record does not report the confidence level for this interval or a threshold for clinically meaningful change. This makes it difficult to connect the statistical estimate to the importance of a difference in fatigue scores.

The record also does not report whether multiplicity was addressed, the original sample-size justification, or enough model detail to reproduce the regression analysis. Numerical results are available for only some of the listed secondary outcomes. These reporting gaps should be distinguished from evidence that a particular analytic choice was flawed: the missing information limits independent assessment but does not, on its own, show that the findings are biased.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Double Blind Trial of Duloxetine in Chronic Fatigue Syndrome — NCT00375973.
• Registry record last updated 2015-08-21; status last verified 2015-07.
• Results first posted 2015-07-27.
• 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

NCT00375973 is a completed randomized, parallel-group, placebo-controlled trial of duloxetine for chronic fatigue syndrome, sponsored by the University of Cincinnati. The registry lists Phase 2 and Phase 3, triple masking, and enrollment of 60. Duloxetine was given orally at 60-120 mg/day for 12 weeks.

The primary outcome was change from baseline to the 12-week endpoint in the MFI general fatigue subscale. Mean changes were -3.3 for duloxetine and -1.8 for placebo. The unadjusted recalculated mean difference was -1.5, with an interval from -3.374 to 0.374 and p=0.117. The registry-reported linear regression analysis gives p=0.23.

The primary comparison is uncertain, and the registry record does not report the regression estimate, a full missing-data strategy, or whether multiplicity was addressed. Three patients were excluded from the duloxetine efficacy analysis for reasons described in the record. Secondary and adverse-event findings add context but do not resolve the uncertainty in the primary comparison.