A Study of Dog Adoption in Veterans With Posttraumatic Stress Disorder (PTSD): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Dog Adoption in Veterans With Posttraumatic Stress Disorder (PTSD) (NCT01729026), a randomized early/phase 1 study. The primary outcome is Change in PTSD Checklist (PCL-5) Score Between Baseline and 3-month Follow-up.

2. Clinical Question

3. Trial Design

This was a randomized, single-masked, parallel-group early-phase study. The registry record lists 19 participants and identifies the condition as PTSD. It is marked completed. The recorded study start is 2013-10, and the primary completion is 2014-12. The sponsor was the VA Office of Research and Development.

One group adopted a shelter dog at the outset. Veterans selected a dog from the San Antonio Humane Society with assistance from an adoption counselor and study staff. The dog was intended as a pet, not a service dog. The intervention description says that veterans and their dogs would receive eight weeks of free obedience training. The other group remained on a wait-list for three months, then chose and adopted a dog. Its description also identifies the dog as a pet rather than a service dog, but the training details in that description are incomplete in the registry record.

The study summary describes adoption as a supplement to usual care for veterans with PTSD who wanted a pet. It does not detail usual care, explain how participants were assigned, or identify who was masked. The meaning of single masking and the practical steps used to maintain it are not reported in the registry record. Neither the intended sample-size rationale nor the study's full protocol is reported there. The early-phase designation and the small enrollment are important context: this record describes a preliminary study, not a large confirmatory comparison.

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: Two of the 22 consented subjects were not randomized: one left before he was evaluated and the second did not meet entry criteria. A third subject was randomized to the Adoption group but subsequently found not to meet criteria for PTSD, leaving 9 in this group and 10 in the Wait-List group.

Randomization to 3-month Follow-up

Reasons for not completing this period:

3-month to 6-month (Final) Follow-up

Reasons for not completing this period:

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 in PTSD Checklist (PCL-5) Score Between Baseline and 3-month Follow-up, measured over Baseline and 3-month follow-up visit..

The primary endpoint was change in the PTSD Checklist for DSM-5 (PCL-5) score between baseline and the three-month follow-up visit. This is a self-rating measure based on DSM-5 PTSD diagnostic criteria. Its range is 0 to 80, with 0 indicating no symptoms and 80 indicating maximal symptoms. A change score describes how a participant's measured score differed across the specified visits; it is not itself a diagnosis.

The registry reports mean changes of -15.2 for shelter-dog adoption and -7.77 for wait-list, then adoption. The accompanying measures of spread are standard errors: 3.52 and 3.29, respectively. These group summaries describe the recorded changes, but the registry does not report separate baseline and follow-up means or participant-level data. The summaries therefore do not show how individual participants' scores varied or whether all participants had measurements at both visits.

In general, a symptom scale can provide a consistent quantitative outcome for comparing measurements over time. A self-rating scale records participants' own responses, so its score is a measure rather than a complete account of a person's experience. The registry record does not state a threshold for interpreting the size of a PCL-5 change, nor does it report a clinically meaningful-change criterion. The endpoint should therefore be read as a scale-score outcome, not as proof of a clinical benefit or a diagnosis.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it is a score on the PCL-5 scale, and the registry identifies the reported group values as means. A mean describes the center of a set of scores. It can be useful for comparing groups, but by itself it does not show the distribution of participants' scores, how many improved, or whether the same pattern applied to each participant. The registry reports standard errors alongside the two mean changes. A standard error describes uncertainty in an estimated mean; it is not the same as the spread of individual participants' scores.

What does the model-based comparison represent?

The sponsor's reported analysis used mixed effects regression models and gives an effect size expressed as Cohen's d. In general, Cohen's d expresses a group contrast on a standardized scale, while a mixed effects regression model can account for repeated measurements or other grouped data structures, depending on how the model is specified. Those are general descriptions, not claims about the exact model specification in this trial. The registry record does not report the model formula, included covariates, covariance structure, or other details needed to reproduce the analysis.

The reported effect size is -0.75, with a 95% confidence interval from -1.85 to 0.35 and p = 0.141. The interval represents uncertainty in the estimated contrast under the analysis. Because it spans zero, the data are compatible with effects in either direction as well as no difference under that model. This does not mean that every value in the interval is equally plausible, and it does not tell readers the probability that a particular effect is true. The p-value summarizes how incompatible the observed result is with a specified null model, under the analysis assumptions; it is not the probability that the null hypothesis is true.

Why does the small enrollment matter?

With 9 participants in the shelter-dog adoption group and 10 in the wait-list group, estimates can be imprecise. In general, fewer observations provide less information about a population mean or group contrast, which can yield wider confidence intervals. The width of the reported interval is relevant to interpretation: the point estimate is not precise enough to settle the direction of the between-group effect. This is an issue of uncertainty, not proof that the estimated effect is absent.

The registry also contains planning illustrations based on the observed effect, with assumptions of a two-sided significance level of 0.05, equal allocation, and a normal approximation. Such calculations are prospective planning exercises: they explore the sample size that would be needed under stated assumptions to detect an effect of a specified fraction of the observed effect. They are not observed power and do not establish that the trial was adequately designed. The registry record does not report the trial's own sample-size justification, and no additional planning values are described in the record text.

What can be inferred from the design?

Random allocation is intended, in general, to reduce systematic differences between groups at the outset. The registry identifies the allocation as randomized, but does not describe sequence generation, concealment, stratification, or how the randomization was implemented. It identifies the study as single-masked without stating who was masked or how masking was maintained. Readers should not assume those details from the design label alone.

The parallel-group arrangement means participants were assigned to one of the two study paths rather than crossing between them during the primary comparison. The paths differ in when adoption occurred: one group adopted at the outset, while the other waited for three months and then adopted. That comparison is not the same as adoption versus no adoption over the full study. The registry does not provide detailed follow-up timing beyond the baseline and three-month primary endpoint or explain how analyses handled participants with missing measurements.

What is not available for independent checking?

The posted group means and standard errors are summaries, not participant-level records. The registry does not provide separate visit means, change-score distributions, or a reported unadjusted two-group estimate with its uncertainty. The reported mixed effects regression result is therefore the main inferential comparison available in the registry record. Detailed model specifications, assumptions checks, analysis-population definitions, and missing-data methods are not reported there. Without those details, independent reproduction or assessment of sensitivity to analytic choices is limited.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Shelter Dog Adoption", "Wait-list, Then Adoption After 3 Months")
n   <- c(9, 10)
m   <- c(-15.2, -7.77)
sp  <- c(3.52, 3.29)
sd  <- sp * sqrt(n)      # standard errors converted to SDs
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Shelter Dog Adoption  9  -15.200  10.560  -23.317  -7.083
CARM  Wait-list, Then Adoption After 3 Months  10  -7.770  10.404  -15.212  -0.328

11. Analyses Reported by the Sponsor

The registry reports the sponsor's own primary comparison as a mixed effects regression analysis. It gives a Cohen's d effect size of -0.75, a 95% confidence interval from -1.85 to 0.35, and p = 0.141. That is the inferential result to use when describing the sponsor's analysis. The interval includes zero, and its breadth indicates considerable uncertainty about the direction and size of the contrast.

The registry also gives each group's mean change and standard error, but it does not report an unadjusted re-calculation of the between-group contrast. The posted group summaries do not, on their own, identify the sponsor model's exact contrast or allow its result to be reconstructed. The record does not explain the effect-size contrast coding or provide sufficient model detail to establish why the reported estimate has its particular direction. Thus, the two group mean changes can be described separately, while the model-based effect should be quoted as reported rather than treated as an independently verified arithmetic comparison.

There is no recorded discrepancy between the posted analysis and a recalculation. That absence does not add information about the model or make an unavailable re-calculation possible. The sponsor analysis takes precedence over a simple unadjusted calculation when the two differ; for this endpoint, the registry's reported model result is the available formal comparison. Its uncertainty and the limitations of the small enrollment remain central to interpretation.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists 12 outcome measures and 11 analyses. Among the named secondary outcomes are changes in PTSD Checklist-related symptoms, alcohol use, depression, community integration, sleep, and physical activity, as well as improvement in quality of life after dog adoption. The record also indicates 11 secondary outcomes posted. These counts describe the registered outcome and analysis record, but do not establish that every outcome had a complete result or that all were tested in the same way.

When a study assesses multiple outcomes or makes multiple comparisons, the chance of at least one apparently notable result can increase. In general, multiplicity procedures can address that issue by adjusting inference or by specifying which outcomes are primary and which are exploratory. The registry record does not report a multiplicity adjustment or a detailed hierarchy for interpreting the secondary outcomes. The primary endpoint should therefore be distinguished from the broader set of outcomes, and isolated secondary results should not be treated as confirmatory without further context.

Important analysis details are also not reported in the registry record. These include the definition of the analysis population, the number of participants with evaluable primary endpoint data, how missing measurements were handled, and whether any participants were excluded from the analysis. The record does not report an intention-to-treat approach, a per-protocol approach, sensitivity analyses, or the reasons for missingness. The number enrolled and the group counts alone do not reveal whether every enrolled participant contributed data at both visits.

The protocol's statistical analysis plan, interim analyses, and sample-size justification are not reported in the registry record. Those omissions limit assessment of whether the reported analyses followed a prespecified plan and how the study addressed multiple outcomes. They do not show that a particular method was or was not used; they mean the relevant details are unavailable in the registry record.

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 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry reports no serious adverse events among the 9 participants at risk in the shelter-dog adoption group and no serious adverse events among the 10 at risk in the wait-list, then adoption group. It also reports 5 participants with other adverse events in the adoption group and 4 in the wait-list group. These are counts of participants affected as recorded, not a description of event severity, duration, or cause. The registry does not list serious-event terms.

The death fields are not populated with a numeric count; deaths are not reported in the registry record. That is different from a documented count of zero. The record also does not provide event-level descriptions for the other adverse events, their timing, or an assessment of whether they were related to study participation or dog adoption. The figures therefore cannot establish the nature or cause of the events.

In general, safety counts in a small study are descriptive. They can indicate what was recorded during the study, but they do not by themselves estimate risk precisely or rule out uncommon harms. The group counts should be read in the context of the limited enrollment and the incomplete event details, rather than as a comprehensive safety assessment.

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

The primary result describes a comparison of PCL-5 score changes over the first three months between veterans who adopted a shelter dog and veterans assigned to wait before adopting. Both group mean changes were negative. The sponsor's model-based estimate was uncertain, with a confidence interval extending from -1.85 to 0.35 and p = 0.141. These statistics describe the observed study data and the analysis recorded in the registry; they do not establish a clear difference between the groups.

The finding does not demonstrate that dog adoption caused the changes in either group. Nor does an uncertain between-group result prove that there is no effect. The estimate's confidence interval allows a range of possible contrasts, and the small study leaves substantial uncertainty. The registry's model details are limited, so the reported analysis cannot be fully assessed or reproduced from the available summaries alone.

The result is also specific to the study's design and population: veterans with PTSD who wanted a pet, with a comparison between adopting at the outset and adopting after a three-month wait. It does not directly answer what would happen for people who do not want a dog, in different care settings, or over a longer period. Since both study paths included planned adoption, the comparison is not adoption versus no adoption.

The endpoint is a self-rated symptom scale, not a complete assessment of PTSD or a measure of every aspect of wellbeing. The registry does not report a criterion for interpreting a clinically meaningful PCL-5 change. Secondary outcomes address other domains, but their existence does not resolve the uncertainty in the primary comparison, especially when multiplicity handling is not reported. The record is best understood as preliminary evidence from a small randomized study, with a formal estimate that remains imprecise.

15. Why This Trial Matters Statistically

This study is statistically instructive because it separates descriptive change within groups from an estimated comparison between groups. The mean PCL-5 changes are easy to read as summaries, but they do not answer the randomized comparison on their own. The sponsor's mixed effects regression result is the formal contrast reported in the registry. Keeping those quantities distinct helps prevent a group-level change from being mistaken for a treatment effect.

The study also illustrates why effect estimates should be read alongside their uncertainty. The reported Cohen's d is -0.75, but its 95% confidence interval, from -1.85 to 0.35, is broad and includes zero. The point estimate alone may appear decisive if its precision is ignored; the interval shows that the direction and magnitude are not settled by these data. A p-value of 0.141 should likewise be interpreted as part of the model-based evidence, not as a verdict about whether an intervention works.

The randomized design is relevant, but a design label is not a substitute for implementation and analysis details. The registry does not report who was masked, how allocation was conducted, how missing outcomes were handled, or the model specification. Those omissions matter when assessing the credibility and reproducibility of a result. The early-phase designation and enrollment of 19 participants also place the estimate in a preliminary setting where uncertainty is expected to be important.

Finally, the wait-list design has a time-specific interpretation: the primary contrast is between adopting at the outset and waiting before adoption. It is not a long-term comparison with no adoption. This distinction, alongside the range of secondary outcomes and unreported multiplicity methods, helps frame what the trial can contribute: an initial randomized comparison whose estimate invites cautious interpretation rather than a definitive conclusion.

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 several trial-specific questions unanswered. It does not describe eligibility criteria beyond the focus on veterans with PTSD who wanted a pet, nor does it report participant characteristics that would help readers assess how closely the enrolled group represents other populations. The enrollment was 19, with 9 in the shelter-dog adoption group and 10 in the wait-list group, limiting the information available for precise comparisons.

The study is marked randomized and single-masked, but the registry does not explain sequence generation, allocation concealment, who was masked, or how masking was maintained. It also does not provide the full protocol or a statistical analysis plan. The model formula and its assumptions are not reported, and the registry does not explain the coding direction for the Cohen's d contrast. These gaps constrain independent evaluation of the sponsor's analysis.

The primary result is reported for baseline to three-month follow-up, but separate baseline and follow-up group means, participant-level data, and the number with complete outcome data are not reported. Missing-data methods and the analysis population are also not reported. The registry does not state whether deviations, exclusions, or sensitivity analyses were considered. As a result, readers cannot determine from this record how incomplete observations may have affected the estimate.

The wait-list group was scheduled to adopt after three months, making the main comparison time-limited and different from adoption versus no adoption. The intervention description for the wait-list group's post-adoption training is incomplete. Usual care is mentioned in the study summary but is not described in detail. Finally, the registry does not report multiplicity procedures, a sample-size justification, or interim analyses. These limits do not invalidate the reported result, but they narrow what can be concluded from the record.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: A Study of Dog Adoption in Veterans With Posttraumatic Stress Disorder (PTSD) — NCT01729026.
• Registry record last updated 2020-06-02; status last verified 2020-05.
• Results first posted 2016-04-27.
• 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.

20. Record Summary

NCT01729026 is a completed, randomized, single-masked, parallel-group early-phase study sponsored by the VA Office of Research and Development. It enrolled 19 veterans with PTSD who wanted a pet. One group adopted a shelter dog, while the other waited three months before adoption. The primary endpoint was mean change in PCL-5 score from baseline to the three-month follow-up.

The registry reports mean changes of -15.2 (standard error 3.52) for shelter-dog adoption and -7.77 (standard error 3.29) for wait-list, then adoption. The sponsor's mixed effects regression analysis reports Cohen's d of -0.75, with a 95% confidence interval from -1.85 to 0.35 and p = 0.141. The interval includes zero, and the estimate is imprecise. The record also reports no serious adverse events in either group, other adverse events affecting 5 and 4 participants, and no numeric death counts. Missing-data methods, model specifications, and multiplicity handling are not reported in the registry record.