Antidepressant Treatment Plus Cognitive Behavioral Therapy for Generalized Anxiety Disorder in Older Adults: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Antidepressant Treatment Plus Cognitive Behavioral Therapy for Generalized Anxiety Disorder in Older Adults (NCT00601965), a randomized na study. The primary outcome is Hamilton Anxiety Rating Scale.

2. Clinical Question

The registry’s stated purpose was to assess whether adding CBT to escitalopram reduces anxiety in older adults with generalized anxiety disorder. The posted outcome and the sponsor-reported analysis address relapse, not a general measure of every aspect of anxiety or functioning. The arm summaries and the sponsor’s time-to-event analysis are not identical forms of evidence: the former give counts, while the latter reports a log-rank test across the four groups. The record does not report a sponsor effect estimate or confidence interval for that test. The unadjusted recalculations compare selected pairs using the posted counts and should be read separately from the sponsor analysis.

3. Trial Design

This was a randomized, parallel-group trial sponsored by the Veterans Medical Research Foundation. The registry records quadruple masking, while the treatment descriptions identify an initial open-label escitalopram phase. The trial is marked completed. Its recorded start was 2007-10, and primary completion was 2011-09.

The registry lists four combinations: escitalopram with CBT, escitalopram without CBT, placebo with CBT, and placebo without CBT. The arm descriptions outline an initial 12 weeks of open-label escitalopram, followed by 16 weeks of continuation treatment, with CBT included in some groups. Later treatment descriptions distinguish maintenance escitalopram from pill placebo. The described treatment course extends through a 28-week maintenance or placebo phase. The exact dose in the initial phase is recorded as 10-20 mg/day as tolerated in some descriptions; later escitalopram dosing is described as continuing at the dose reached at the end of the initial phase.

The registry lists arm outcome counts of 18 for escitalopram plus CBT, 16 for escitalopram without CBT, 19 for placebo plus CBT, and 17 for placebo without CBT. These are arm-specific values, not a basis for revising the registry’s overall enrollment figure. The record does not report the randomization procedure, allocation concealment details, or the identities of people included in the masking description. It also does not report further detail about the sequence of treatment assignment beyond the listed randomized groups. The posted sponsor analysis is a log-rank test across all four groups; the record does not provide a corresponding effect estimate or confidence interval.

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.

Population: please see Wetherell et al. (2013), Antidepressant Medication Augmented with Cognitive-Behavioral Therapy for Generalized Anxiety Disorder in Older Adults, American Journal of Psychiatry, 170:782-9.

6. Primary Endpoint

The registry lists the primary outcome as Hamilton Anxiety Rating Scale, measured over 56 weeks.

The primary endpoint is the number of participants who relapsed, measured over a timeframe of 56 weeks. The Hamilton Anxiety Rating Scale is a clinician-rated assessment of cognitive and somatic anxiety symptoms. Its 14 items are scored from 0 to 4 and summed to a total ranging from 0 to 56; higher scores indicate greater anxiety. The study required a score of at least 17 for inclusion.

Relapse was defined using several conditions together. A participant had to have a Hamilton score increase of at least 5 points relative to the end of the continuation phase at week 28, reach a total Hamilton score of at least 14, meet the duration requirement of at least 2 weeks, and have both the clinician and participant judge that anxiety symptoms had recurred. This definition combines a change from the participant’s earlier score, an absolute symptom threshold, persistence, and clinical judgment.

Using a defined relapse outcome gives the analysis a concrete event to compare across treatment groups. A time-to-event analysis can also account for when an event is recorded, rather than reducing the follow-up period to a single end-of-study status. The registry does not report additional endpoint details about assessment schedules, censoring rules, or how missing endpoint assessments were handled.

7. Results Posted in the Registry

• Escitalopram + CBT: 100.0% (18/18) versus 73.7% (14/19) on control; risk difference 26.3 percentage points (95% CI 6.5 to 46.1), risk ratio 1.36 (1.04–1.78); unadjusted two-proportion z-test p = 0.019.
• Escitalopram + no CBT: 93.8% (15/16) versus 73.7% (14/19) on control; risk difference 20.1 percentage points (95% CI -3.0 to 43.1), risk ratio 1.27 (0.95–1.71); unadjusted two-proportion z-test p = 0.117.
• Placebo + no CBT: 47.1% (8/17) versus 73.7% (14/19) on control; risk difference -26.6 percentage points (95% CI -57.5 to 4.3), risk ratio 0.64 (0.36–1.13); unadjusted two-proportion z-test p = 0.102.

8. Statistical Interpretation

9. Statistical Methods Explained

Different statistical summaries answer different questions. The registry reports a sponsor log-rank test across the four treatment groups, while the posted pairwise risk differences and their intervals are unadjusted recalculations from arm-level counts. They should not be treated as interchangeable analyses. Because the record flags a discrepancy, the sponsor’s own analysis takes precedence.

What does a risk difference describe?

In general, a risk difference subtracts the event proportion in one group from the event proportion in another. A positive estimate means the event proportion is higher in the group named first in that contrast; a negative estimate means it is lower. The recalculated comparison of escitalopram plus CBT with placebo plus CBT is 0.263. The comparison of escitalopram without CBT with placebo plus CBT is 0.201, and the comparison of placebo without CBT with placebo plus CBT is -0.266. These estimates are unadjusted and describe the posted counts, not a causal effect guaranteed by the analysis.

The recalculated relative measures provide another way to express the same group contrasts. For escitalopram plus CBT versus placebo plus CBT, the risk ratio is 1.357, with an interval from 1.037 to 1.776. For escitalopram without CBT versus placebo plus CBT, the risk ratio is 1.272, with an interval from 0.945 to 1.712; the odds ratio is 5.357, with an interval from 0.555 to 51.71. For placebo without CBT versus placebo plus CBT, the risk ratio is 0.639, with an interval from 0.361 to 1.131, and the odds ratio is 0.318, with an interval from 0.079 to 1.283. Relative measures can look less intuitive than absolute differences, especially when event proportions are high. They do not resolve the discrepancy with the sponsor analysis.

What does the log-rank test assess?

In general, a log-rank test compares the pattern of time to an event across groups over follow-up. It uses information about when events occur, rather than only comparing the final event counts. The registry reports p < .001 for a test involving all four groups. This result is evidence against the test’s null hypothesis of no difference in the time-to-event patterns across groups. It is a global result; it does not establish which pairwise comparisons account for the difference, nor does it quantify the size of any contrast. The registry does not report an effect estimate or interval for the log-rank result.

How should the recalculated p-values be read?

In general, a p-value describes how compatible data are with a specified null hypothesis under the assumptions of the analysis. The recalculated values are p = 0.019, p = 0.117, and p = 0.102 for the three listed pairwise comparisons. They are unadjusted values, and the registry does not report whether or how the sponsor adjusted for multiple comparisons. They also do not measure the probability that a treatment is effective, or the probability that the null hypothesis is true.

What do the planning illustrations mean?

The planning illustrations are based on observed effects and are not observed power or a reconstruction of the trial’s original design. The stated 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 a stated fraction of the observed effect. For escitalopram without CBT versus placebo plus CBT, planning for 100% of the observed effect gives 52 participants per arm for 80% power and 70 participants per arm for 90% power. Planning for 75% gives 105 participants per arm for 80% power and 140 for 90% power; planning for 50% gives 261 and 348, respectively. For placebo without CBT versus placebo plus CBT, planning for 100% gives 52 participants per arm for 80% power and 69 for 90% power; planning for 75% gives 90 and 120, and planning for 50% gives 195 and 260, respectively. These are illustrations under the stated assumptions, not evidence about the study’s design adequacy.

10. Confidence Intervals and Precision

A confidence interval expresses the uncertainty around an estimate under the method used to construct it. In general, a narrower interval indicates greater precision, while a wider interval indicates that a broader range of values is compatible with the data and method. The registry does not identify the confidence level or calculation method for the recalculated intervals, so their precision should be interpreted without assuming details not in the record.

For escitalopram plus CBT compared with placebo plus CBT, the recalculated risk difference is 0.263, with an interval from 0.065 to 0.461. This interval remains above zero. For escitalopram without CBT compared with placebo plus CBT, the estimate is 0.201, with an interval from -0.030 to 0.431. For placebo without CBT compared with placebo plus CBT, it is -0.266, with an interval from -0.575 to 0.043. The latter two intervals include zero and extend across values in both directions. These intervals reflect uncertainty in the simple count-based comparisons; they are not intervals for the sponsor’s log-rank analysis.

The relative-measure intervals also vary in width. The recalculated odds ratio for escitalopram without CBT versus placebo plus CBT is 5.357, with an interval from 0.555 to 51.71. This wide range indicates substantial uncertainty in the relative estimate. The registry does not provide a sponsor confidence interval for its global test, so the precision of the sponsor-reported analysis cannot be assessed from an effect interval in the record.

Arm-level counts are modest, and estimates based on counts can consequently be imprecise. That general point does not resolve the flagged discrepancy between the count-based recalculations and the sponsor’s analysis. The confidence intervals should be read together with the endpoint definition, comparison being made, and difference in analytic approach, rather than as a standalone verdict about the study.

11. Analyses Reported by the Sponsor

The sponsor-reported analysis is a log-rank test across all four groups, with p < .001. The registry does not give an effect estimate or confidence interval for this analysis, and the reported test is global rather than a set of pairwise comparisons. It therefore indicates evidence of a difference in time-to-event patterns across the groups under the reported analysis, but does not state the magnitude or direction of each pairwise difference.

The recalculated comparisons use the posted event counts and are unadjusted. They yield risk differences of 0.263, 0.201, and -0.266 for the listed comparisons. These do not reproduce the sponsor’s analysis: they address selected pairs using counts rather than the sponsor’s reported time-to-event test. The registry explicitly flags a discrepancy. Under that circumstance, the sponsor’s own analysis takes precedence, and the recalculations should be treated only as a separate check of the posted summaries.

The record does not explain the discrepancy or report the additional details needed to determine its source. It would therefore be inappropriate to declare the analyses equivalent, infer that one method caused the difference, or use the recalculated values to override the sponsor result. The sponsor result should also be read within its scope: it is a global test and does not supply pair-specific effect sizes or precision.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists two outcome measures and two analyses, and reports a log-rank test across the four groups. It does not report a multiplicity strategy, such as an adjustment to account for testing several outcomes, groups, or pairwise contrasts. The recalculated pairwise p-values are unadjusted. Without a reported analysis plan, it is not possible to determine whether the sponsor treated the global test as primary, whether additional comparisons were prespecified, or how any multiplicity was handled.

Details about missing data are also not reported in the registry record. The record does not describe how participants with incomplete follow-up, missing Hamilton assessments, or unknown relapse status were handled. It does not report censoring rules for the log-rank analysis, how withdrawal or loss to follow-up was treated, or whether any sensitivity analyses addressed missing outcomes. The posted arm counts should not be assumed to represent a particular missing-data strategy.

The analysis populations are not reported in the registry record. In particular, the record does not say whether the sponsor analysis followed an intention-to-treat, per-protocol, or another population definition. It does not report which participants were included in the log-rank test or whether any exclusions occurred after randomization. These omissions limit assessment of how the analysis relates to the randomized groups.

These reporting gaps matter because the interpretation of a time-to-event test depends on how event times, censoring, and incomplete follow-up are handled. They also matter when several outcomes or contrasts are considered. No assumption about the sponsor’s statistical plan, interim analyses, or sample-size justification can be made from the registry entry.

13. 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.

The serious adverse event summary records zero affected participants in each listed group: escitalopram plus CBT, escitalopram without CBT, placebo plus CBT, and placebo without CBT. The corresponding numbers at risk are 18, 18, 19, and 18. These are registry figures for serious adverse events and should be read as counts in the recorded categories, not as proof that no serious events occurred outside the record’s reporting window or definitions.

The registry does not list serious adverse event terms. Death counts are recorded as null for each group, so they are not reported in the registry record; null should not be read as zero. The record also does not provide a detailed description of adverse-event collection, follow-up completeness, or event attribution. The zero serious-event counts therefore offer limited information on their own and do not establish the broader safety profile of the treatments.

These figures are separate from the relapse endpoint. They do not change the interpretation of the sponsor’s log-rank result or the recalculated relapse comparisons. They should be considered only as the posted serious adverse event summary, with the limitations in reporting kept in view.

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

The posted results describe relapse counts under a specific definition and a sponsor-reported comparison of time-to-event patterns across four groups. The sponsor’s log-rank test has p < .001. This is evidence against equal time-to-event patterns under the test’s assumptions, but the registry does not provide an effect estimate or confidence interval for that analysis. The result alone does not say which group differences account for the global finding, or how large those differences are.

The unadjusted recalculations give pairwise risk differences based on the posted counts. The interval for escitalopram plus CBT versus placebo plus CBT excludes zero, while the intervals for the other listed comparisons include zero. Because the registry flags a discrepancy, the sponsor’s own analysis takes precedence. The recalculated comparisons are not a substitute for the sponsor result and should not be used to make a definitive claim about pairwise effects.

In general, statistical evidence is not the same as proof of causation, practical importance, or clinical benefit. A p-value is not the probability that a hypothesis is true, and a confidence interval does not remove uncertainty about study conduct or analysis. The randomized design is relevant, but incomplete information about analysis populations, missing data, multiplicity, and the sponsor analysis’s effect estimates limits what can be concluded from the registry record.

The findings apply to the recorded study population and treatment schedule, not automatically to other populations, settings, or treatment strategies. The registry record does not supply enough detail to assess all aspects of generalizability. The results should therefore be described as the evidence recorded for this trial, with the distinction between the sponsor’s global test and the count-based recalculations kept explicit.

15. Why This Trial Matters Statistically

This trial is statistically instructive because it presents two different kinds of result for the same clinical question: a sponsor-reported log-rank test and recalculated pairwise comparisons from event counts. The log-rank test considers timing of events across all groups, while a risk difference based on counts compares event proportions for selected groups. These analyses answer related but distinct questions, so their estimates and p-values should not be treated as interchangeable.

The registry also flags a discrepancy between the recalculations and the sponsor analysis. That makes it especially important to distinguish the source and purpose of each statistic. The sponsor analysis takes precedence, but its global p-value is not accompanied by an effect estimate or interval. The recalculated estimates add descriptive context but cannot fill that reporting gap or replace the sponsor’s analysis.

The record illustrates why effect size and precision matter alongside a test result. A global p-value does not indicate the size or direction of each group contrast. Pairwise intervals show the range of uncertainty around the count-based estimates, but they do not resolve the discrepancy or establish a treatment effect. Planning illustrations based on observed effects are also not a substitute for the original design rationale, which is not reported in the registry record.

Finally, the endpoint definition combines a score change, a threshold, duration, and clinical judgment. Interpreting its counts requires attention to that definition and to the registry’s limited reporting on missing data and analysis populations. The trial therefore demonstrates how careful reading of endpoint construction, analysis type, and reporting gaps can prevent a global test, a pairwise recalculation, or a planning illustration from being given more meaning than it supports.

16. 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 leaves important analytical details unreported. It does not provide the sponsor’s effect estimate or confidence interval for the log-rank analysis, explain the discrepancy flagged between that analysis and the count-based recalculations, or identify the analysis population. Details of censoring and missing endpoint data are also not reported in the registry record. These omissions restrict assessment of the sponsor result and prevent a full reconciliation with the posted arm counts.

The primary outcome depends on several components: change in Hamilton score from week 28, a total-score threshold, persistence for at least 2 weeks, and both clinician and participant judgment of recurrence. The record does not describe how disagreements in judgment were handled, the assessment schedule, or how incomplete assessments affected relapse classification. It reports that raters were blind, but does not give further information about the rating process.

Some treatment-description details are incomplete in the registry entry. The CBT plus escitalopram description ends after the mention of up to 3 booster sessions, so further booster details are not reported in the registry record. The duration of the taper in the placebo description is not reported in the registry record. The registry gives an overall enrollment figure and arm-level counts, but the analysis populations and the relationship between those counts and all randomized participants are not explained.

The design and reporting also limit interpretation of the recalculated pairwise results: they are unadjusted, and no multiplicity approach is reported. The planning illustrations are based on observed effects under stated assumptions and are not the trial’s own sample-size justification. The registry record does not report that justification, interim analyses, or a complete statistical plan. These limitations warrant caution without establishing that any specific analysis or treatment conclusion is invalid.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Antidepressant Treatment Plus Cognitive Behavioral Therapy for Generalized Anxiety Disorder in Older Adults — NCT00601965.
• Registry record last updated 2016-08-25; status last verified 2016-07.
• Results first posted 2016-08-25.
• 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.

20. Record Summary

The registry identifies a completed randomized, parallel-group study sponsored by the Veterans Medical Research Foundation. It enrolled 73 participants and compared four combinations of escitalopram or placebo with CBT or no CBT in older adults with generalized anxiety disorder. The primary endpoint was the number of participants meeting the registry’s relapse definition during a 56-week timeframe.

The sponsor-reported result is a log-rank test across all four groups, with p < .001. Unadjusted pairwise recalculations from posted arm summaries are also recorded, but the registry flags a discrepancy; the sponsor’s own analysis takes precedence. The record does not report the sponsor’s effect estimate or confidence interval, the analysis populations, or details of missing-data handling and multiplicity. Serious adverse event counts are zero in each listed group; deaths are not reported in the registry record.