Improving Diabetes Risk Factors in Persons With Schizophrenia or Bipolar Disorder by Switching to Ziprasidone: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Improving Diabetes Risk Factors in Persons With Schizophrenia or Bipolar Disorder by Switching to Ziprasidone (NCT00538642), a randomized na study. The primary outcome is Insulin Sensitivity.

2. Clinical Question

3. Trial Design

This was a randomized, parallel-group trial with no masking. The registry lists an enrollment of 24 and marks the study completed. It lists the phase as NA. The study was sponsored by The University of Texas Health Science Center at San Antonio. The registry dates the study start to 2007-08 and primary completion to 2011-04.

The study description focuses on people with schizophrenia or bipolar disorder who were taking certain newer antipsychotics and who also had obesity and/or a family history of diabetes. The oral glucose tolerance test was used to identify prediabetes. The description says that eligible participants would have insulin resistance measured before being randomly assigned to either continue their current antipsychotic or switch to ziprasidone. The assigned treatment strategies were not blinded: the record explicitly lists masking as none.

In the continuation arm, participants stayed on their baseline daily oral antipsychotic, with clinically indicated dose adjustments allowed. In the switch arm, participants moved from their current antipsychotic treatment to daily oral ziprasidone, titrated to a clinically effective level. The registry reports an intended repeat insulin-resistance measurement after four months.

The posted primary endpoint summaries include 9 participants in each arm, while the registry enrollment is 24. The record does not explain the relationship between the enrollment figure and the numbers contributing to these endpoint summaries. It also does not report details such as the allocation sequence, concealment procedures, adherence, treatment discontinuation, or the analysis population. The trial’s open-label design is relevant when considering outcomes that can be influenced by awareness of treatment assignment, although the endpoint itself is a laboratory-based measurement.

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.

6. Primary Endpoint

The registry lists the primary outcome as Insulin Sensitivity, measured over Baseline.

The primary outcome is titled Insulin Sensitivity and is expressed in mg glucose/kg.min/μIU insulin. The registry says it was measured using the euglycemic clamp method. This method is used to assess how the body responds to insulin by measuring the glucose needed to maintain a target glucose level while insulin is administered. In general, greater insulin sensitivity indicates that a given insulin exposure is associated with a stronger glucose-handling response; insulin resistance describes a weaker response.

Insulin sensitivity is relevant to the study’s stated question because the registry describes prediabetes as elevated glucose levels below the diabetic range and identifies insulin resistance as the outcome expected to change after the treatment switch. The clamp measurement is a physiological endpoint rather than a clinical event such as a diabetes diagnosis. It can provide a quantitative measure of insulin action, but interpreting an intervention effect requires knowing which time point or change was compared.

The primary outcome entry lists its timeframe as Baseline. The study description, by contrast, says the measure would be repeated after four months. The posted results give a mean and standard deviation for each arm, but do not label those summaries as follow-up measurements or changes from baseline. The registry record therefore does not establish from the posted endpoint entry that the reported comparison measures change over the intended treatment period.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a mean and standard deviation describe?

The mean is a measure of the centre of a set of observations. The standard deviation describes how dispersed observations are around that mean. In the registry, the reported summaries are means with standard deviations for a continuous measurement. A standard deviation is not the same as a confidence interval: it describes participant-level variation, whereas a confidence interval describes uncertainty about an estimated quantity, such as a mean or between-arm contrast.

The posted summaries are 5.55 with standard deviation 3.51 in the continuation arm and 1.88 with standard deviation 1.08 in the ziprasidone arm. The reported participant count is 9 for each. These summaries alone do not reveal the distribution of individual values, whether there are influential observations, or how much uncertainty surrounds a treatment contrast. No confidence interval for the contrast is reported.

What does ANOVA contribute?

ANOVA, or analysis of variance, is a general method for comparing means across groups. In a two-arm setting, it can test whether the observed group means are compatible with a null hypothesis of equal means, subject to the model assumptions and the way the endpoint and analysis population are defined. The registry reports an ANOVA result of p < 0.05. It does not provide an estimated contrast, an interval estimate, details of model adjustment, or an explanation of the exact endpoint data used in the analysis.

A p-value measures how incompatible the observed data are with a specified null model, assuming that model and its assumptions are appropriate. It is not the probability that the null hypothesis is true, and it is not a measure of the probability that a result is due to chance. A threshold-based p-value also does not indicate how large a difference is, whether it matters clinically, or how reproducible the estimate may be. Because the record gives only a thresholded p-value, its exact evidential strength cannot be gauged more finely.

Why is the endpoint timeframe important?

A randomized trial intended to compare change over time needs a clearly defined comparison: for example, a difference between arms at follow-up, a contrast in change from baseline, or a model using repeated measurements. These approaches answer related but not identical questions. The study description says the measurement would be repeated after four months, but the primary endpoint is labelled Baseline. The registry does not state whether the ANOVA used baseline values, follow-up values, or another analysis. Without that detail, the reported p-value cannot be confidently interpreted as a test of the intended four-month change.

The timing issue is not a technical detail that can be resolved by the arm means. Even when treatment is randomized, baseline and follow-up comparisons have different interpretations. A baseline comparison describes groups before the assigned intervention could have produced a treatment effect. A follow-up comparison can reflect treatment and other post-randomization events, while a change analysis compares the evolution of the outcome. The record does not identify which of these is represented by the posted results.

What do confidence intervals add?

In general, a confidence interval presents a range of estimates compatible with the data and the chosen statistical procedure. Its width conveys precision: narrower intervals usually indicate greater precision, while wider intervals indicate more uncertainty. A confidence interval for a between-arm contrast would help readers understand both the range of plausible effect sizes and whether the data are compatible with materially different conclusions. No confidence interval is reported for the ANOVA comparison, so the precision of the contrast is not available from the registry entry.

Why can small samples make estimates unstable?

With limited numbers of observations, means and standard deviations may be sensitive to the particular participants included. Randomization does not guarantee that small groups will be balanced on every relevant characteristic. The registry lists enrollment of 24, but each primary arm summary includes 9 participants; the record does not explain this difference. The endpoint summaries therefore should not be treated as a complete account of participant flow or as evidence that all enrolled participants contributed to the analysis.

What does over-dispersion mean in general?

Over-dispersion occurs when observed variability is greater than a statistical model assumes. In general, ignoring over-dispersion can make standard errors too small and uncertainty appear less than it is. This concept is especially relevant for count outcomes or models with specific distributional assumptions. The primary outcome in this trial is continuous, and the registry does not report an over-dispersion assessment. It would not be appropriate to claim that over-dispersion occurred; the general concept simply illustrates why model assumptions matter.

How should planning calculations be understood?

Planning sample-size illustrations are not observed power. They describe the participant counts that would be needed under stated assumptions to detect an assumed effect with a specified testing approach. The registry’s planning assumptions describe a two-sided significance level of 0.05, equal allocation, a normal approximation, and effects expressed as a stated fraction of the observed effect. Such illustrations are conditional on those assumptions; they do not show the probability that the completed study detected an effect, and they do not establish the trial’s original design rationale. The registry does not report its own sample-size justification.

More broadly, a trial’s statistical interpretation depends on an explicit estimand: the population, treatment strategies, outcome, timing, and summary measure being compared. The registry gives the broad treatment strategies and describes a planned four-month repeat measurement, but the posted primary outcome timeframe is Baseline and the analysis description is sparse. That leaves important elements of the intended comparison unresolved.

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("Stay on Current Antipsychotic", "Ziprasidone Treatment")
n   <- c(9, 9)
m   <- c(5.55, 1.88)
sp  <- c(3.51, 1.08)
sd  <- sp      # posted standard deviations
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  Stay on Current Antipsychotic  9  5.550  3.510  2.852  8.248
CARM  Ziprasidone Treatment  9  1.880  1.080  1.050  2.710

11. Analyses Reported by the Sponsor

The registry reports the study analysis as ANOVA with p < 0.05. It does not report the corresponding between-arm estimate, confidence interval, model details, or a more precise p-value. The posted arm summaries are means and standard deviations, but the registry does not provide enough information to identify their role in a treatment-effect calculation with confidence.

A recalculated two-arm contrast is not available in the posted analysis because the record indicates that values are missing or not estimable for that comparison. The arm-level summaries should not be used to construct a new treatment-effect estimate without a clear endpoint time point and analysis definition. In particular, the outcome timeframe is labelled Baseline even though the study description anticipates repeat measurement after four months.

There is no reported discrepancy between the sponsor’s analysis and a recalculated estimate. The reported ANOVA result should therefore be presented as the registry’s own analysis, while recognizing that the record does not provide the estimate and precision needed to interpret its magnitude. The p-value alone cannot fill in those missing details or resolve whether the analysis addresses the intended follow-up question.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists 18 outcome measures, 16 posted secondary outcomes, and one reported analysis. These counts indicate that the record includes multiple outcome measures but reports only one analysis entry in the analysis summary. They do not establish how many statistical tests were conducted, which outcomes were formally tested, or whether any adjustment for multiple comparisons was used. The multiplicity plan is not reported in the registry record.

When multiple outcomes or repeated analyses are examined, the chance of obtaining at least one apparently notable result can increase. Statistical approaches to multiplicity can define which outcome is primary, control a family-wise error rate, or emphasize estimation rather than a collection of threshold-based tests. Those are general considerations; the registry does not state which approach, if any, was used. The single reported p-value should not be treated as proof that multiplicity was controlled.

The record also does not report an analysis population, such as whether results were analyzed according to randomized assignment or only among participants with available measurements. It does not give participant flow by arm, reasons for non-completion, the amount of missing endpoint data, or methods for handling missing observations. The enrollment is 24, while the primary arm summaries each report 9 participants; the record does not explain the difference. No assumptions about who was omitted or why should be made.

The statistical plan is not reported in the registry record. Details such as a prespecified model, covariate adjustment, handling of repeated measurements, interim analyses, and procedures for missing data are also not reported in the registry record. Without those details, readers cannot assess whether the posted analysis follows a prespecified plan or determine how robust its result is to alternative reasonable analyses.

These gaps matter particularly because the study description specifies a baseline measurement and a repeat measurement after four months, whereas the primary endpoint timeframe is listed as Baseline. Clear reporting of the analysis population, time point, missing-data handling, and multiplicity approach would be needed to interpret the reported test as a complete account of the primary question.

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 summaries report 0 affected participants among 10 at risk in the Stay on Current Antipsychotic group and 0 among 14 at risk in the Ziprasidone Treatment group. These figures describe what is posted in the registry for the specified safety groups. The safety denominators differ from the primary endpoint summaries, which report 9 participants in each arm; the record does not explain the difference.

No serious adverse event terms are listed among the top serious terms. The registry does not report death counts for either arm, so deaths should not be interpreted as absent. It also does not provide detail in these summaries about non-serious events, event timing, duration, severity beyond the serious-event classification, or whether events led to treatment changes.

Zero reported serious adverse events in these groups does not establish that serious events could not occur or that the two strategies have equivalent safety. The number of participants represented is limited, and the registry entry does not provide an uncertainty estimate for the event rates. These figures should be read narrowly as the serious adverse event counts recorded for the listed at-risk groups, not as a complete safety profile.

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

The registry record documents a randomized comparison between continuing a current antipsychotic and switching to ziprasidone, with insulin sensitivity named as the primary outcome. It reports arm-level means and standard deviations and an ANOVA p-value below 0.05. Those entries establish that a statistical comparison was reported, but they do not provide a complete estimate of the size or precision of the comparison.

The result does not establish from the posted endpoint entry that insulin sensitivity changed over the intended four-month period. The study description refers to baseline and repeat measurement, but the primary outcome timeframe is Baseline and the posted summaries are not explicitly identified as follow-up or change values. No between-arm estimate or confidence interval is reported. The ANOVA p-value cannot resolve this timing ambiguity or convey the magnitude of a treatment effect.

The reported group summaries are descriptive and should not be treated as proof of a clinically important effect. They are based on 9 participants per arm, while registry enrollment is 24, and the relationship between those counts is not explained. The record does not report participant flow, missing-data methods, an analysis population, or enough model details to judge the analysis fully.

The study’s stated hypothesis gives the clinical and statistical rationale for measuring insulin resistance, but a stated hypothesis is not the same as a clearly reported estimate of its result. The posted record also does not establish how the primary analysis handled the intended repeated measurements, whether other analyses were conducted, or whether multiplicity was addressed.

Accordingly, the results support a cautious description of what the registry reports: different arm-level summaries, with an ANOVA p-value of p < 0.05, for an endpoint whose reported timeframe is Baseline. They do not, as posted, quantify a four-month comparative change, provide its uncertainty, or justify a broader conclusion about treatment effects or clinical outcomes.

15. Why This Trial Matters Statistically

This trial is statistically notable for the gap between its stated longitudinal question and the way the primary outcome is represented in the registry. The study description says insulin resistance would be measured before random assignment and again after four months. Yet the primary outcome entry lists Baseline, and the arm summaries do not identify the measurement occasion. That difference illustrates why a clearly stated outcome time point is essential to interpreting a trial comparison.

The study also shows the distinction between a p-value and an effect estimate. A thresholded ANOVA result indicates that a statistical test was reported, but it does not communicate the size of the between-arm contrast or its precision. Without an estimate and confidence interval, readers cannot judge the range of values compatible with the data or compare statistical evidence with practical importance.

Another important issue is the relationship between enrollment and endpoint counts. The registry lists enrollment of 24, while the primary summaries include 9 participants in each arm. The record does not explain the difference. Transparent participant flow and reporting of the analysis population would help readers understand which participants contributed to the endpoint results and how missing observations were handled.

The posted safety denominators differ from the endpoint denominators as well, and the registry supplies no explanation. This is a reminder that outcome and safety populations can be distinct, but their relationship should be clearly described rather than inferred from totals alone.

More generally, randomized allocation can support a fair comparison, but the interpretation depends on the definition of the outcome, the timing of measurement, the analysis model, and clear reporting of uncertainty and missing data. Those details are especially important in a small study. The record’s limitations do not negate the reported analysis; they bound what can be concluded from it.

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.

Several limitations are specific to the information reported for this trial. The most consequential is the mismatch between the stated measurement schedule and the primary endpoint entry. The study description anticipates a repeat insulin-resistance measurement after four months, but the endpoint timeframe is listed as Baseline. The registry does not identify the time point underlying the posted arm summaries or the ANOVA result. This prevents a confident interpretation of that result as a treatment-related change.

The registry reports means and standard deviations but no between-arm estimate or confidence interval. It gives only the threshold p < 0.05 for the ANOVA analysis. The exact statistical model, whether baseline adjustment was used, and how repeated measurements were handled are not reported in the registry record. The recalculated two-arm comparison is not estimable from the posted analysis.

The listed enrollment is 24, while the primary endpoint summaries report 9 participants in each arm. The record does not explain participant flow, exclusions, missing measurements, or the analysis population. Safety groups have different at-risk counts, and the relationship between those counts and the endpoint summaries is not reported. These differences should not be resolved by assumption.

The record does not report a sample-size justification, detailed statistical plan, multiplicity adjustment, interim analyses, treatment adherence, discontinuation, or baseline characteristics. The registry also does not identify the specific antipsychotic medications participants were taking before assignment. Since masking was none, participants and treatment providers knew assignment; the degree to which that might affect care or outcome collection is not described. The endpoint is a clamp-based physiological measure, but its measurement procedures beyond the stated method are not reported.

These limitations concern the completeness and interpretability of the registry report. They do not establish that the trial was conducted incorrectly, but they constrain how far the posted results can be taken beyond the summaries and test result recorded.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Improving Diabetes Risk Factors in Persons With Schizophrenia or Bipolar Disorder by Switching to Ziprasidone — NCT00538642.
• Registry record last updated 2012-11-29; status last verified 2012-11.
• Results first posted 2012-11-19.
• 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

NCT00538642 is a completed, randomized, open-label, parallel-group study sponsored by The University of Texas Health Science Center at San Antonio. It lists enrollment of 24 and compares continuing a current antipsychotic with switching to daily oral ziprasidone among people with schizophrenia, schizoaffective disorder, or bipolar disorder who were at risk of diabetes.

The primary outcome is insulin sensitivity measured by the euglycemic clamp method. The registry reports a baseline-labelled mean of 5.55 with standard deviation 3.51 for 9 participants continuing current treatment, and a mean of 1.88 with standard deviation 1.08 for 9 participants assigned to ziprasidone. It reports an ANOVA result of p < 0.05 but no contrast estimate or confidence interval. The registry record does not clarify whether these summaries or the analysis represent follow-up or change after the stated four-month period.

Serious adverse event summaries report 0 affected participants in groups with 10 and 14 participants at risk, respectively. The record does not explain the different endpoint and safety counts or provide further analysis details needed to interpret the comparative result.