This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Comparing Treatments for Self-Injury and Suicidal Behavior in People With Borderline Personality Disorder (NCT00834834), a randomized phase 4 study. The primary outcome is Suicide Events.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT00834834 |
| Phase | Phase 4 |
| Condition | Borderline Personality Disorder, Suicide |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Single |
| Enrollment | 84 |
| Arms in the primary outcome | 2 |
| Lead sponsor | New York State Psychiatric Institute |
| Results posted | Yes |
2. Clinical Question
Population
People with borderline personality disorder, in a study that also lists suicide as a condition. The registry record reports enrollment of 84 participants. The posted primary outcome is measured after 6 months of treatment; the registry record does not give a more detailed description of the population contributing to that outcome.
Intervention
Dialectical behavior therapy (DBT), described as a behavioral treatment. The registry record lists DBT as an intervention and identifies the study arm as Dialectical Behavior Therapy.
Comparator
Fluoxetine with clinical management. The arm description says clinical management may involve switching medication to citalopram, another SSRI. The registry record lists citalopram as an intervention, but does not report how often a switch occurred.
Primary question
How many suicide events were recorded after 6 months of treatment in each arm? The primary outcome is a count of events, rather than a count only of participants who experienced at least one event. The registry record also posts a secondary outcome counting participants with suicide events.
The central comparison is between two treatment strategies, not simply between two individual drugs or between treatment and no treatment. One strategy is DBT; the other is fluoxetine with clinical management, which could include a change to citalopram. The record does not give enough detail to determine how clinical management was delivered in practice or whether medication changes were made for particular participants.
The primary question concerns suicidal behavior captured by the Columbia Suicide History Interview (CSHI). Its scope includes actual suicide attempts and suicide-related behaviors such as aborted and interrupted attempts. The count may therefore reflect more than completed or medically treated attempts. The interview's operational definition of an actual attempt requires a self-injurious act performed with at least some intent to die.
This is a question about the observed event count at a stated treatment time point. It is not, by itself, a comparison of time to first event, the severity of every event, or longer-term outcomes. The registry record does not report a primary comparative estimate, confidence interval, or statistical test for the event-count outcome. The secondary participant-count outcome offers a different view: whether participants had suicide events, rather than how many events occurred. These outcomes address related but distinct aspects of suicidal behavior.
3. Trial Design
The registry record identifies this as a phase 4, randomized, single-masked, parallel-group study. It is marked completed and reports enrollment of 84. The listed start is 2009-03 and the primary completion is 2015-08. The sponsor is New York State Psychiatric Institute. These registry entries describe the trial's recorded design and status; they do not establish how closely the trial was conducted to every planned procedure.
The two arms are Fluoxetine and Dialectical Behavior Therapy. The fluoxetine arm is described as fluoxetine with clinical management, which may involve switching to citalopram, another SSRI. The other arm receives dialectical behavior therapy. Although the intervention list includes fluoxetine, DBT, and citalopram, the registry record does not report how often citalopram was used or how medication decisions were made. It also does not specify what was masked in this single-masked design or who was masked.
The posted primary endpoint is the number of suicide events measured after 6 months of treatment. Its arm entries list 12.0 events for 38.0 participants in the Fluoxetine arm and 4.0 events for 39.0 participants in the Dialectical Behavior Therapy arm. The registry record also reports a secondary outcome, the number of participants with suicide events, measured after 6 months. Outcome entries can have different contributing numbers from the overall enrollment; the record does not explain the relationship between enrollment and the numbers listed for these outcomes.
Randomization supports a comparative design, but details needed to assess its implementation are not reported in the registry record. The randomization method, allocation concealment, masking details, and analysis population are not reported. The registry record also does not state the planned sample-size justification, whether interim analyses were planned, or the statistical analysis plan. Those omissions limit what can be inferred about the intended design and the handling of the observed data.
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: All participants are washed out of psychotropic medications. Benzodiazapines are permitted for sleep.
Overall Study
| Stage | Fluoxetine | Dialectical Behavior Therapy |
|---|---|---|
| STARTED | 42 | 42 |
| COMPLETED | 38 | 39 |
| NOT COMPLETED | 4 | 3 |
Reasons for not completing this period:
| Reason | Fluoxetine | Dialectical Behavior Therapy |
|---|---|---|
| Withdrawal by Subject | 4 | 3 |
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.
| Characteristic | Unit | Fluoxetine | Dialectical Behavior Therapy | Total |
|---|---|---|---|---|
| Participants analysed | 42 | 42 | 84 | |
| Age, Continuous | years | 27.9 (SD 7.8) | 30.6 (SD 10.8) | 29.3 (SD 9.4) |
| Sex: Female, Male: Female | Participants | 37 | 40 | 77 |
| Sex: Female, Male: Male | Participants | 5 | 2 | 7 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 9 | 11 | 20 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 33 | 31 | 64 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 0 | 0 | 0 |
6. Primary Endpoint
The registry lists the primary outcome as Suicide Events, measured over Measured after 6 months of treatment.
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | not applicable |
| Unit | suicide events |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint is the number of suicide events measured after 6 months of treatment. The registry record says suicidal behavior was collected using the Columbia Suicide History Interview (CSHI), a semi-structured interview developed by the study group. It elicits the history of actual suicide attempts and asks about the circumstances of suicidal behavior and its medical lethality.
The measure also captures suicide-related behaviors such as aborted and interrupted suicide attempts. The CSHI defines an actual suicide attempt as a self-injurious act performed with at least some intent to die. This definition makes the endpoint broader than a count limited to events resulting in medical treatment. At the same time, it is a count of events: a participant can contribute more than one event. It therefore answers a different question from a binary outcome that records only whether a participant had any event.
Counting events can describe the overall frequency of recorded behavior during the measurement period, while the secondary participant-count outcome describes how many people experienced events. Both perspectives can matter when events may recur. The registry record does not report details such as event adjudication procedures, interviewer masking, how incomplete interviews were handled, or whether all events were independently reviewed. It also does not report a measure of dispersion for the event counts. Those missing details make it harder to assess the precision and reproducibility of the primary count outcome.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
For the primary outcome, the posted arm entries are 12.0 suicide events among 38.0 participants in the Fluoxetine arm and 4.0 among 39.0 participants in the Dialectical Behavior Therapy arm, measured after 6 months of treatment. These are descriptive event counts. The registry record does not post an estimated between-arm difference, confidence interval, or p-value for this primary count outcome. It also provides no measure of variation around the counts. Consequently, the posted primary results do not quantify the precision of the arm contrast or establish how compatible the observed difference is with other possible underlying differences.
The difference between an event count and a participant count is important for interpretation. Several events could arise from one person, or events could be distributed across several people. The primary entries alone do not say how the events were distributed among participants. The secondary outcome records participants with suicide events and therefore supplies a related, but not interchangeable, summary. The registry's count outcome is not accompanied by a posted comparative analysis, and the secondary result should not be treated as a substitute for that missing primary analysis.
The secondary binary comparison reports a risk difference with a confidence interval that includes zero and an unadjusted p-value of 0.47. In general, a risk difference expresses the contrast in observed event proportions between arms; a value of zero corresponds to equal proportions. An interval spanning zero is compatible with no difference as well as with differences in either direction at the level of uncertainty represented by that interval. The point estimate is not a precise or conclusive statement about the treatment contrast. The unadjusted p-value also does not account for any multiplicity across outcomes or analyses.
The secondary comparison is based on 6 of 38 participants in the Fluoxetine arm and 4 of 39 in the Dialectical Behavior Therapy arm. These counts are modest, and the confidence interval is broad. The estimate therefore has limited precision. It should not be read as proof that the arms are equivalent, nor as proof that one strategy is superior. A p-value above a conventional threshold is not evidence that there is no difference; it indicates that the observed data do not provide strong evidence against the null hypothesis under the particular unadjusted test reported.
For the event-count outcome, the absence of a comparative estimate is the main statistical limitation. The posted arm counts may be described, but their contrast cannot be assigned a reliable uncertainty range from the registry record. The registry record notes that values were missing or not estimable for a two-arm comparison for this count-type endpoint. That means a standard comparison is not posted; it does not establish why a comparison could not be estimated. Any conclusion should remain limited to the descriptive entries and the imprecise secondary comparison, rather than treating either as a definitive treatment effect.
9. Statistical Methods Explained
Two kinds of outcomes appear in the registry record, and they call for different statistical summaries. The primary endpoint is a count of suicide events. The secondary endpoint is binary: whether a participant had suicide events. A count can include recurrent events from the same participant, whereas a binary outcome records a participant only as having or not having an event. Methods suitable for the binary outcome do not automatically answer the count-outcome question.
What does a confidence interval express?
General explanation: a confidence interval describes uncertainty around an estimated effect under the statistical method used. It is not a range that guarantees the true effect lies inside it, and it does not measure clinical importance. For the secondary risk difference, the reported interval crosses zero. Since zero represents no difference in proportions, the interval is compatible with no difference and with effects in either direction. Its breadth indicates that the estimate is not precise. The interval should be read alongside the point estimate and the underlying participant counts, not as a standalone verdict.
The secondary result also includes an unadjusted p-value of 0.47. General explanation: a p-value measures how unusual data at least as inconsistent with a specified null model would be if that model were true and the test assumptions held. It is not the probability that the null hypothesis is true, nor a direct measure of effect size. Because the value is unadjusted, it does not account for testing multiple endpoints or analyses. The registry record does not report a multiplicity adjustment or a prespecified hierarchy for interpreting secondary results.
Why is the primary event count difficult to compare?
The primary outcome is a count, and the registry record gives arm-level counts but no dispersion measure or comparative estimate. Counts often have variability that depends on their mean, and recurrent events within a participant may be statistically related. General explanation: a count model may need to account for over-dispersion, which occurs when the observed variability exceeds what a simple model assumes. If events cluster in a small number of participants, treating every event as independent may understate uncertainty. The registry record does not report the distribution of events across participants or the method used to account for recurrence, clustering, or over-dispersion. It is therefore not possible to assess whether a particular count model was appropriate or to reproduce the primary comparison from the posted summary.
The record indicates that values were missing or not estimable for a two-arm comparison for this count-type endpoint. It does not explain whether this arose from the count distribution, model fit, data availability, or another issue. No explanation should be inferred from that notation. The primary outcome remains descriptively reported, but a comparative uncertainty estimate is not available in the registry record.
How should the planning illustrations be read?
The sample-size planning illustrations are based on the observed secondary effect, rather than the trial's own design justification. They assume a two-sided significance level of 0.05, equal allocation, and a normal approximation. They express participants per arm for 80% or 90% power when planning for 100%, 75%, or 50% of the observed effect. For 100% of the observed effect, the figures are 580 participants per arm for 80% power and 776 for 90% power. For 75% of the observed effect, they are 1,078 and 1,443 participants per arm, respectively. For 50% of the observed effect, they are 2,528 and 3,384 participants per arm, respectively.
General explanation: these are planning illustrations, not evidence that the trial had a particular probability of detecting an effect. They do not describe observed power. Because they use an effect estimated from the posted data, they inherit uncertainty from that estimate; the broad interval around the secondary effect is especially relevant. Smaller assumed effects require larger planning samples under the stated assumptions. The registry record does not report the trial's own sample-size justification, so these illustrations should not be mistaken for the original target or a protocol requirement.
What do the methods not establish?
The registry record reports no formal analysis for the primary count outcome and no reported analyses in its analysis listing. It does not describe an intention-to-treat or per-protocol population, handling of missing outcome data, adjustment variables, or a statistical plan. General explanation: these choices can affect estimates, uncertainty, and the interpretation of randomized comparisons. Without them, the descriptive results cannot be converted into a complete account of the trial's inferential analysis. The posted secondary comparison is useful as a summary, but its unadjusted status and imprecision limit the weight it can carry.
10. Multiplicity, Missing Data and Analysis Populations
The registry record lists one secondary outcome and reports no analyses in its analysis listing. It does not state how many outcomes or statistical tests were planned, whether the secondary outcome was prespecified as confirmatory or exploratory, or whether a hierarchy of outcomes was used. A multiplicity adjustment is not reported. The secondary comparison therefore carries an unadjusted p-value, which should not be interpreted as though it accounts for testing across a broader set of possible outcomes or analyses.
General explanation: multiplicity arises when several hypotheses are examined. If each is assessed separately without adjustment, the overall chance of at least one false-positive result can be higher than the nominal level for a single test. Prespecification, hierarchical testing, or other approaches can help control interpretation, but the registry record does not say whether any such approach was planned or used. The lack of a reported adjustment does not prove that no plan existed; it means the record does not document one.
Missing-data handling is also not reported in the registry record. The record gives overall enrollment of 84 and outcome-specific participant counts, but it does not explain why those counts differ, identify whether any outcomes were missing, or describe how missing observations were handled. The outcome population is not specified as intention-to-treat, modified intention-to-treat, per-protocol, or another analysis set. No imputation method or sensitivity analysis is reported.
For the primary count outcome, a two-arm comparison is described as not estimable or missing, but the reason is not reported. The registry record provides arm-level counts without a comparative model, variance estimate, or confidence interval. It does not report whether this was a planned limitation, an analytic difficulty, or a consequence of the available data. These omissions constrain assessment of the primary result and of the robustness of the secondary comparison. They also prevent a full evaluation of whether the analysis followed the randomized allocation or excluded participants after randomization.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Fluoxetine | 14 | 38 | 36.8% |
| Dialectical Behavior Therapy | 5 | 39 | 12.8% |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Fluoxetine | 0 of 38 |
| Dialectical Behavior Therapy | 0 of 39 |
Most frequent serious adverse event terms
| Event term | Organ system | Fluoxetine | Dialectical Behavior Therapy |
|---|---|---|---|
| Serious Adverse Event | Psychiatric disorders | 14 of 38 | 5 of 39 |
Serious events with the highest arm-level rate: serious adverse event (36.8%).
The registry record lists serious adverse events for 14 of 38 participants in the Fluoxetine arm and 5 of 39 participants in the Dialectical Behavior Therapy arm. It also lists zero other adverse events for each arm. The record identifies “Serious Adverse Event” as a top serious-event term and gives 36.8% for that term; it does not state the denominator or the precise relationship between that percentage and the arm-specific figures.
These are descriptive safety entries. The registry record does not provide event-level details, definitions of seriousness, timing, severity grades, relatedness assessments, or an adjusted comparison between arms. It also does not report deaths in these entries. The figures should not be interpreted as proof that an event was caused by a treatment, or that a treatment caused no adverse events. A zero in the other-adverse-event field is limited to what was recorded in that field; it is not evidence that no non-serious symptoms occurred.
Different numbers of participants are listed for the two arms, and the posted figures do not establish why. The registry record does not describe completeness of safety follow-up or methods used to collect adverse events. These limitations matter when comparing safety summaries: differences in ascertainment, exposure, or follow-up could affect the counts, but the record does not provide details to evaluate those possibilities. The serious-event figures should therefore be read as registry summaries rather than as a complete comparative safety analysis.
12. What the Result Does, and Does Not, Mean
The posted results describe a randomized comparison of DBT with fluoxetine and clinical management, and report arm-level counts for suicidal behavior measured after 6 months of treatment. The primary endpoint captures events, including actual attempts as well as aborted and interrupted attempts. The record also reports a secondary count of participants with events. These summaries address clinically important aspects of the study question, but the available statistical detail is not sufficient for a precise primary comparative conclusion.
The primary event counts are descriptive. No comparative estimate, confidence interval, p-value, or measure of variability is posted for that outcome. The secondary participant-count analysis provides an unadjusted comparison, but its confidence interval includes zero and its p-value is 0.47. That result does not establish equivalence, prove that there is no effect, or demonstrate that one treatment strategy is better. It indicates uncertainty around the estimated contrast, and the broad interval leaves room for effects in either direction.
These results do not determine why the arms' recorded counts differ, whether any difference is caused by treatment, or how outcomes might compare over a different duration. They do not answer questions about event timing, severity beyond the interview's recorded features, or longer-term outcomes. The registry record also does not provide enough information to assess missing-data handling, the analysis population, multiplicity, or the model intended for the recurrent-event endpoint.
The safety entries are also descriptive and do not establish causation. The record does not report details that would support a complete comparative safety assessment. More broadly, registry results should be interpreted with the documented methods and omissions in view. The posted figures are evidence about the outcomes recorded in this trial, but they do not by themselves support broad claims about treatment effectiveness or safety outside the study. No medical recommendation follows from these registry summaries.
13. Why This Trial Matters Statistically
This trial raises a statistical issue that is easy to miss when two outcomes sound similar: the number of events and the number of participants with events are not the same estimand. An event count can reflect repeated events among some participants, while a participant count treats anyone with at least one event as an outcome. Reporting both can be informative, but the analyses need to match the specific question. The primary event-count result is particularly difficult to interpret because the registry record gives arm-level counts without a comparative estimate or uncertainty measure.
The outcome also illustrates why event definition and measurement matter. The CSHI captures actual attempts as well as aborted and interrupted attempts, and defines an actual attempt using intent to die. Such detail helps clarify what is being counted, while also making it important to understand how events were elicited, classified, and counted. Interviewer masking, adjudication, and incomplete assessment procedures are not reported in the registry record.
The secondary binary comparison illustrates the role of precision. Its confidence interval includes zero, and the unadjusted p-value is 0.47. A point estimate alone cannot settle the comparative question, especially when its uncertainty is broad. The planning illustrations further show how strongly sample-size requirements depend on the effect one seeks to detect, but they are not the trial's original design justification and should not be read as observed power.
Finally, a randomized label does not remove the need to understand analysis populations, missing data, and multiplicity. Those details are not reported in the registry record. This study is statistically noteworthy because its posted outcome types, incomplete primary comparison, and unadjusted secondary analysis require careful separation of descriptive findings from inferential conclusions.
14. 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 most important limitation specific to these results is that the primary endpoint is a recurrent-event count, yet the registry record does not report a two-arm estimate, a measure of dispersion, a confidence interval, or a statistical test. It states that values were missing or not estimable for a two-arm comparison. Without an explanation or a reported model, readers cannot assess the magnitude or precision of the primary contrast, or determine how repeated events within participants were handled.
The event measure is based on a semi-structured interview, and the record describes the behaviors included in the endpoint. It does not report the interview schedule beyond the six-month measurement point, interviewer masking, independent event review, classification rules for every event type, or procedures for missing interview information. These omissions limit assessment of measurement reliability and completeness.
The secondary outcome has an unadjusted comparison and an interval that includes zero. Its precision is limited, and multiplicity methods or a prespecified hierarchy are not reported. The posted planning illustrations use the observed secondary effect and are not a substitute for a trial-specific sample-size rationale. The registry record does not report the original sample-size assumptions or whether interim analyses were planned.
Other trial details that would help interpret the results are also absent: the randomization procedure, allocation concealment, who was masked, the analysis population, missing-data methods, and protocol deviations. The overall enrollment and outcome-specific participant numbers are not reconciled in the record. The medication-management description allows a switch to citalopram, but the frequency and circumstances of switching are not reported. Safety entries lack event-level detail and a comparative analysis. These limitations do not invalidate the posted data, but they restrict the strength and scope of conclusions that can be drawn from them.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Comparing Treatments for Self-Injury and Suicidal Behavior in People With Borderline Personality Disorder — NCT00834834.
- Registry record last updated 2017-08-18; status last verified 2017-07.
- Results first posted 2017-08-18.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
NCT00834834 is a completed phase 4 randomized, single-masked, parallel-group study sponsored by New York State Psychiatric Institute. It compares dialectical behavior therapy with fluoxetine and clinical management, which may involve switching to citalopram. The registry record reports enrollment of 84, a start of 2009-03, and primary completion in 2015-08.
The primary outcome is the number of suicide events measured after 6 months of treatment using the CSHI. The posted counts are 12.0 among 38.0 participants in the Fluoxetine arm and 4.0 among 39.0 in the Dialectical Behavior Therapy arm. A comparative analysis for this count outcome is not reported as estimable. The secondary outcome counts participants with suicide events and includes an unadjusted comparison with a confidence interval spanning zero and p-value of 0.47. The registry record does not report the analysis population, missing-data methods, multiplicity approach, or a detailed statistical plan.