Treatment of Depression in Parkinson's Disease Trial: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Treatment of Depression in Parkinson's Disease Trial (NCT00062738), a randomized phase 2 study. The primary outcome is Hamilton Depression Scale.

2. Clinical Question

This is a comparison of average scores at a specified follow-up, not a comparison of individual changes or the proportion of participants meeting a response definition. The registry also lists Percent Responders as a secondary outcome, but the primary question concerns the continuous Hamilton Depression Scale measure. The arm summaries and comparisons can describe the direction and uncertainty of differences in average scores. They do not, by themselves, establish why any difference occurred or whether it is important to each participant.

The two drug comparisons share the same placebo arm. Each estimate therefore addresses a drug-versus-placebo contrast; the registry does not post a direct comparison between nortriptyline and paroxetine. The record identifies the study as randomized and quadruple masked, but details of the randomization procedure and which groups of people were masked are not reported in the registry record.

3. Trial Design

The Treatment of Depression in Parkinson's Disease Trial is listed as a phase 2, randomized, quadruple-masked, parallel study. It enrolled patients with Parkinson disease and depression and evaluated nortriptyline, paroxetine, and placebo. The registry describes the study as completed, with a start of 2003-06 and primary completion of 2009-01. Rutgers, The State University of New Jersey, is listed as the sponsor.

A parallel design assigns participants to an arm and compares outcomes across arms at the designated assessment. Randomization is intended, in general, to make treatment groups comparable on average at the point of assignment. Masking is intended, in general, to limit the influence of knowledge about assignment on participants' experiences, care, or outcome assessment. The registry labels this study quadruple masked, but the identities of the masked parties and the operational details of masking are not reported in the registry record.

The posted primary measure is the Hamilton Depression Scale at 8 weeks. The registry identifies the summary parameter as a least-squares mean and the dispersion measure as a standard deviation. It describes the outcome population as intent to treat, with total patients in the arm. The posted arm sizes are 17.0 for nortriptyline, 18.0 for paroxetine, and 17.0 for placebo. The record does not explain how the analysis population was formed or how participants without an 8-week score, if any, were handled.

The study compares each drug arm with placebo. Although the record lists two interventions of drug type and one placebo intervention, it does not report the treatment allocation ratio or detailed procedures for assigning participants. The original design assumptions, including any sample-size justification, are not reported in the registry record. The planning illustrations derived from the observed comparisons are separate from the original design and should not be taken as a description of the trial's planned enrollment target.

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: Patients who met inclusion and not exclusion criteria were enrolled.

Overall Study

Analysis population for the primary outcome, as defined in the registry: intent to treat - total patients in the arm

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 Hamilton Depression Scale, measured over 8 weeks.

The primary endpoint is the Hamilton Depression Scale total score at 8 weeks. The registry describes it as a score from 0 to 54, with higher scores indicating worse depression. It is a continuous outcome: each participant contributes a score on a numeric scale, and the posted arm summaries describe average scores rather than the number of people meeting a threshold.

A continuous symptom score can retain information across the scale instead of reducing results to a binary category such as responder or non-responder. In general, comparing average scores can help describe whether the observed score distribution is shifted between arms. A mean difference alone, however, does not show how scores vary among individuals or whether the difference is meaningful to every participant. The registry separately lists Percent Responders as a secondary outcome.

The record identifies the reported summary as a least-squares mean and reports standard deviations for dispersion. Least-squares means are often model-based adjusted means, but the model and any covariates used for this endpoint are not reported in the registry record. The posted endpoint is therefore best read as an 8-week arm-level summary, with the details of how the means were estimated left unspecified. The registry's description of the analysis population is intent to treat, with total patients in the arm; further details about missing scores and inclusion rules are not reported in the registry record.

The endpoint addresses symptom severity at the specified follow-up. It does not, on its own, summarize safety, functional outcomes, longer-term course, or individual experiences. Those questions require outcomes and analyses beyond the primary score.

7. Results Posted in the Registry

• Nortriptyline: mean 10.84 versus 15.81 on control; difference -4.97 (95% CI -9.00 to -0.94), standardized effect d = -0.83; z-test p = 0.016.
• Paroxetine: mean 12.45 versus 15.81 on control; difference -3.36 (95% CI -7.38 to 0.66), standardized effect d = -0.55; z-test p = 0.102.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous, and the posted arm summaries are means with standard deviations. In general, a mean describes the center of a set of scores, while a standard deviation describes how spread out individual scores are around that center. These are different quantities: the standard deviation reflects participant-to-participant variability, whereas the uncertainty in a mean difference depends on the variability and the number of observations contributing to each arm estimate.

What does a mean difference express?

A mean difference expresses the contrast between two arm averages on the original outcome scale. For the reported drug-versus-placebo comparisons, negative estimates correspond to lower scores in the drug arm because the outcome is expressed as drug-arm mean minus placebo-arm mean. Since higher Hamilton Depression Scale scores indicate worse symptoms, the negative direction is toward lower average scores. This directional interpretation describes the posted summaries; it is not a claim about the cause or importance of the difference.

The nortriptyline mean difference is -4.97, and the paroxetine mean difference is -3.36. These are in units on the scale, not standardized units. Keeping the estimate in the original scale makes its direction and scale units clear, but judging its practical importance would require context not reported in the registry record. The posted estimates compare each drug with placebo; they do not provide a direct contrast between the drugs.

What do the p-values say?

A p-value measures, under a specified null model and analysis, how incompatible the observed data are with that model. It is not the probability that the null hypothesis is true, nor is it a measure of the size or importance of an effect. The posted p-values are unadjusted: 0.016 for nortriptyline versus placebo and 0.102 for paroxetine versus placebo. An unadjusted value does not account for the possibility that multiple outcomes or comparisons were examined. The registry does not report a multiplicity adjustment or a broader analysis plan.

In general, a threshold such as 0.05 is a convention for making a decision under a specified testing plan. Interpreting a p-value without knowing that plan is limited, particularly when more than one comparison or outcome may be considered. The p-values should therefore be read together with the estimated differences, their intervals, the small arm sizes, and the record's missing methodological details. They should not be treated as a ranking of the two drugs.

Why do intervals matter?

A confidence interval gives a range of effect estimates that are compatible with the data under the interval's statistical assumptions and method. In repeated use of the same procedure, a confidence interval at a chosen confidence level would cover the target value at that level over repeated samples. It is not a guarantee that a particular interval contains the true value, and it is not a range containing a specified proportion of individual participant scores.

The interval for the nortriptyline comparison lies below zero, while the interval for the paroxetine comparison crosses zero. The first pattern indicates that the posted interval excludes no mean difference under its method; the second indicates that the posted interval includes no mean difference. Neither pattern alone establishes the clinical relevance of the estimate. Interval width is also important: a wide interval reflects substantial uncertainty about the size of the difference, even when its direction is more clearly separated from zero.

What is known about the calculation?

The registry reports least-squares means and standard deviations for the primary outcome, but does not describe the model used to estimate the least-squares means. The comparison summaries include pooled standard deviations and standard errors, and the calculations indicate approximate standard-deviation inputs. Details of the precise interval procedure, degrees of freedom, and any covariate adjustment are not reported in the registry record. This means the numerical comparisons should not be read as a fully documented model-based analysis.

The registry labels the primary population intent to treat, with total patients in the arm. In general, an intention-to-treat approach aims to analyze participants according to their randomized assignment, but the label alone does not specify how missing outcome values, protocol deviations, or other analysis issues were handled. Those implementation details are not reported in the registry record. Nor does the record describe whether any sensitivity analyses were conducted.

The trial is listed as randomized and quadruple masked. Randomization and masking can reduce certain sources of bias in general, but the registry record does not identify the allocation procedure, masking roles, or any departures from assigned treatment. These design labels do not replace a description of the analysis. The record also does not report the original sample-size assumptions or a formal statistical analysis plan, so the analysis methods cannot be fully reconstructed from the reported summaries alone.

All comparisons are against the same placebo arm. In general, a shared comparator can make each contrast useful for describing a separate treatment-versus-control question, but the contrasts are not independent of one another because they use the same control information. The registry does not provide a direct between-drug test. The two p-values should not be compared as though they test whether nortriptyline and paroxetine differ from each other.

10. Confidence Intervals and Precision

Precision concerns how narrowly the available data constrain an estimate. The mean difference for nortriptyline versus placebo is -4.97, with an interval that stays below zero. The mean difference for paroxetine versus placebo is -3.36, with an interval that crosses zero. These interval patterns are consistent with the respective unadjusted p-values, but they communicate more than a threshold decision: they show the range of differences that remains compatible with the data under the interval method.

The nortriptyline interval is not so narrow that the magnitude can be treated as exact. Its lower limit is -8.999, and its upper limit remains below zero. The paroxetine interval extends from -7.383 to a positive upper limit, so it includes both a negative difference and no difference. The width of these intervals matters: multiple effect sizes remain plausible, especially for paroxetine. No interval can guarantee the true effect's location, and an interval for a mean difference does not describe the spread of individual outcomes.

The arm standard deviations—6.2 for nortriptyline, 6.36 for paroxetine, and 5.78 for placebo—show substantial score variation around the respective means. The posted arm sizes are 17.0, 18.0, and 17.0. With limited observations in each arm and variation in scores, the estimates are inherently uncertain. The exact interval calculation and any adjustments to the least-squares means are not reported in the registry record.

Confidence intervals are useful for considering both direction and plausible magnitude, but should be interpreted with the analysis plan in mind. The registry does not report whether these comparisons were designated as confirmatory, how multiple comparisons were handled, or whether additional analyses informed interpretation. Consequently, the intervals describe the posted comparisons rather than resolving every uncertainty about the study's analysis.

11. 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 illustrations translate the observed mean differences into planning scenarios. For these continuous outcomes, the displayed numbers are participants per arm, not total events. Each contrast has scenarios based on planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. They are illustrations based on the observed results, not the trial's own enrollment rationale.

The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect. For each assumed effect, the 80% power column gives the participants per arm indicated by that planning calculation for an 80% chance of detecting an effect of the assumed size under those assumptions. The 90% power column gives the corresponding per-arm planning figure for a 90% chance. These are design targets under a model, not promises about the results of a future study.

Smaller effects generally require more participants to distinguish them from zero with the same assumptions and target power. Thus, planning for a reduced fraction of the observed effect leads to larger per-arm requirements than planning for the full observed effect. The two drug-versus-placebo contrasts have different observed effect estimates, so their planning illustrations differ; neither set identifies an appropriate target without additional design choices and clinical judgment.

The trial's original sample-size justification and design assumptions are not reported in the registry record. The planning values are not observed power: they do not tell how likely this completed trial was to detect an effect after its results were known. The associated curve should be read as a visual representation of the planning relationship under the stated assumptions, not as a record of the study's original planning or a prediction of clinical outcomes.

12. 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 = -4.970; pooled standard deviation σ = 5.994.
2. n = 2 × (1.96 + 0.842)² × 5.994² / -4.970² = 22.8, rounded up to 23 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 <- -4.97                      # observed difference in means
sigma <- 5.99368000480506                     # 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(62738)
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 = 23 (80% power), 31 (90% power)
 75% of observed effect: n per arm = 41 (80% power), 55 (90% power)
 50% of observed effect: n per arm = 92 (80% power), 123 (90% power)
power.t.test: 23.8 
Simulated power at n = 23 per arm: 0.783 (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.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary Hamilton Depression Scale outcome and a secondary outcome, Percent Responders. It also posts two treatment-versus-placebo comparisons for the primary outcome. Considering several outcomes or comparisons can increase the chance of at least one apparently notable result when each is tested separately. In general, multiplicity methods address that issue by adjusting the testing procedure or by defining a hierarchy of questions in advance.

The posted p-values are unadjusted. Whether the trial used a prespecified multiplicity strategy, which outcomes were included in such a strategy, and whether the secondary outcome was analyzed are not reported in the registry record. The record also does not provide a statistical analysis plan identifying primary contrasts or the order in which outcomes were tested. The two p-values should therefore not be interpreted as if the record established that they were the only analyses considered.

The outcome population is described as intent to treat, with total patients in the arm. The record does not describe the number of participants with an observed 8-week score, the extent or pattern of missing outcome data, or the method used to address any missing scores. It also does not state whether participants were analyzed according to assigned group in all circumstances or describe sensitivity analyses for missingness. These details matter because an analysis can depend on assumptions about the unobserved outcomes.

The registry reports least-squares means but does not identify the fitted model or any adjustment variables. It does not report whether the comparisons were prespecified as adjusted or unadjusted at the model level, nor does it provide model diagnostics. The posted p-values are identified as unadjusted, but that label alone does not answer every question about the underlying outcome model.

These omissions do not establish that a method was absent; they mean that it is not reported in the registry record. The estimates can be described as posted, but the statistical analysis cannot be fully evaluated for multiplicity, missing-data assumptions, or population implementation from the registry information alone.

14. 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: muscle rigidity (5.9%), delirium (5.9%), fainting (5.6%).

The registry reports serious adverse events by arm using the number at risk and the number affected. The serious-event figures are 0 of 17 at risk in the nortriptyline arm, 1 of 18 in the paroxetine arm, and 2 of 17 in the placebo arm. These are counts of affected participants as posted, not estimates of comparative risk with uncertainty intervals.

The record lists muscle rigidity and delirium at 5.9% each, and fainting at 5.6%, among the top serious adverse event terms. The information does not map each listed term to a specific arm in the registry record for interpretation, so no arm-specific attribution should be inferred from those percentages. The record does not report the number of deaths; that information is not reported in the registry record.

Small counts are difficult to compare reliably. In general, a difference of one or a few events can be highly sensitive to the experience of individual participants, and counts alone do not account for uncertainty, timing, duration, or severity. The posted figures therefore describe what was recorded, but do not by themselves establish whether arms differ in safety. The registry's other adverse-event counts are separate from the serious-event counts and should not be treated as interchangeable.

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

The posted results describe 8-week Hamilton Depression Scale averages in a randomized, quadruple-masked phase 2 trial. The nortriptyline arm's mean score was lower than the placebo arm's, and its unadjusted interval lies below zero. The paroxetine arm's mean was also lower than placebo's, but its interval includes zero. The estimates are uncertain, and the registry record does not provide enough detail to assess every aspect of the analysis plan.

These findings do not establish that either treatment will produce a particular result for an individual. An arm average does not show how each participant's score changed, how many participants improved, or how outcomes varied across subgroups. Although Percent Responders is listed as a secondary outcome, its result is not reported in the registry record. The score comparisons also do not answer questions about longer-term outcomes or other aspects of health.

The results should not be interpreted as a direct comparison between nortriptyline and paroxetine. Each posted estimate compares one drug arm with the shared placebo arm, and the record does not provide a between-drug test. Nor does a difference in the significance status of the two placebo comparisons establish that the drug effects differ.

Statistical uncertainty and clinical importance are distinct. The intervals express uncertainty about the mean differences under the statistical method; the registry does not report a threshold for a clinically important difference. The figures therefore cannot, on their own, determine how consequential a difference might be in practice. The reported serious adverse event counts likewise offer limited information for comparing safety because they are sparse and do not include comparative uncertainty estimates.

In short, the registry record offers a narrow account of outcome scores and adverse-event counts for this study. It supports description of the posted estimates and their uncertainty, but does not resolve missing-data, multiplicity, model-specification, or broader clinical questions.

16. Why This Trial Matters Statistically

This trial raises several statistical issues that are common in clinical studies. It has two treatment-versus-placebo contrasts, a continuous primary outcome, and a shared comparator. Each contrast can be described separately, but the shared placebo arm means the estimates are related. A direct question about whether the two drugs differ requires a direct contrast and its uncertainty; the two separate placebo comparisons do not substitute for that test.

The results also show why estimates and intervals add context beyond a binary significance label. The nortriptyline comparison's interval lies below zero, while the paroxetine interval crosses zero. This difference in interval position does not itself show a difference between the drugs. Reading the point estimates and intervals together makes clear that both magnitude and precision matter, and that a non-significant result is not proof of no effect.

The planning illustrations provide another useful lesson. Sample-size requirements depend on the effect chosen for planning and on assumptions about the test and allocation. Planning for a smaller effect requires more participants per arm under the listed assumptions. The illustrations are based on observed effects, however, and cannot reconstruct the trial's original rationale, which is not reported in the registry record.

Finally, the record demonstrates the limits of interpreting summary results without methodological detail. The outcome is labelled as a least-squares mean, but the model is not described; the analysis population is labelled intent to treat, but missing-data handling is not described; and the p-values are unadjusted, while multiplicity procedures are not reported. These gaps do not negate the posted estimates, but they constrain what can be concluded from them. Clear reporting of analysis decisions is essential for evaluating how robustly a clinical result follows from the data.

17. 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 posted arm groups are small: the registry gives 17.0 for nortriptyline, 18.0 for paroxetine, and 17.0 for placebo. The estimates consequently have appreciable uncertainty, reflected in the comparison intervals. The interval for paroxetine includes zero, and the interval for nortriptyline, while below zero, still spans a range of possible mean differences. These results should not be treated as precise estimates of a single underlying effect.

The registry identifies the primary summary as a least-squares mean but does not report the model or adjustment variables. It is therefore not possible from the record to determine precisely how those means relate to the raw arm summaries or what assumptions informed the estimates. The comparisons are labelled with unadjusted p-values, and the record does not describe any adjustment for examining multiple comparisons or outcomes.

Although the primary population is labelled intent to treat, the record does not describe missing outcome data, the number with an observed 8-week score, or the method used to handle missingness. The original sample-size justification and analysis plan are also not reported in the registry record. These are reporting limitations; they should not be taken as evidence that the trial did not have procedures for these issues.

The endpoint is measured at 8 weeks, so the posted primary result addresses that follow-up only. The registry lists Percent Responders as a secondary outcome but does not report its result in the registry record. The record also provides only limited serious adverse event information: counts are small, death figures are not reported in the registry record, and the listed serious-event terms are not assigned to arms in the available information.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Treatment of Depression in Parkinson's Disease Trial — NCT00062738.
• Registry record last updated 2015-11-24; status last verified 2015-10.
• Results first posted 2013-03-05.
• 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.

21. Record Summary

The registry record identifies NCT00062738 as the Treatment of Depression in Parkinson's Disease Trial, a completed randomized, quadruple-masked, parallel phase 2 study sponsored by Rutgers, The State University of New Jersey. It lists patients with Parkinson disease and depression, an enrollment of 52, a study start of 2003-06, and primary completion of 2009-01.

The primary endpoint is the Hamilton Depression Scale total score at 8 weeks, with higher scores indicating worse symptoms. The posted means are 10.84 for nortriptyline, 12.45 for paroxetine, and 15.81 for placebo. The corresponding mean differences versus placebo are -4.97 and -3.36, with unadjusted p-values of 0.016 and 0.102.

The record also lists Percent Responders as a secondary outcome and reports serious adverse event counts by arm. Model details, multiplicity procedures, missing-data handling, the original sample-size rationale, and death counts are not reported in the registry record.