Effectiveness of Lithium Plus Optimized Medication in Treating People With Bipolar Disorder: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Effectiveness of Lithium Plus Optimized Medication in Treating People With Bipolar Disorder (NCT00667745), a randomized phase 4 study. The primary outcome is Overall Change in Bipolar Illness Severity as Measured by Clinical Global Impression for Bipolar Disorder Severity (CGI-BP-S) Score.

2. Clinical Question

3. Trial Design

The registry record describes a phase 4, randomized, single-masked, parallel-group trial sponsored by the National Institute of Mental Health. It lists enrollment of 283 and a completed study status. The recorded start is 2008-04, and primary completion is listed as 2010-03. These dates describe the registry record; they do not establish when each participant began or ended treatment.

Participants were assigned to one of two groups. The lithium group received lithium plus optimized medication treatment, as needed. The comparison group received optimized medication treatment, as needed, without lithium. The record identifies lithium carbonate and optimized treatment as the interventions, but does not describe the medication choices, dose adjustments, adherence procedures, or other details of treatment implementation.

The primary outcome was continuous: change in the Clinical Global Impression for Bipolar Disorder Severity score, reported as a mean with standard deviation. Its relevant time points were baseline and week 24. The registry also lists depression symptoms, mania symptoms, and suicidality as secondary outcomes, each measured over 6 months. The number of posted outcome measures is 5, while 3 secondary outcomes are included in the information summarized for this analysis.

Single masking is reported, but the identity of the masked party is not reported in the registry record. Details of sequence generation, concealment of assignment, the original sample-size justification, and the prespecified analysis population are also not reported in the registry record. These omissions limit what can be inferred about trial conduct beyond the design features that are explicitly recorded.

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 Overall Change in Bipolar Illness Severity as Measured by Clinical Global Impression for Bipolar Disorder Severity (CGI-BP-S) Score, measured over Relevant time points: baseline and week 24.

The primary endpoint was change in the Clinical Global Impression for Bipolar Disorder Severity score between baseline and week 24. The scale is intended to represent overall bipolar illness severity, drawing together a clinician’s assessment of depression, mania, and overall illness rather than focusing on only one symptom domain. Its stated range runs from 0, not assessed, through 1, normal or not at all ill, to 7, among the most extremely ill patients.

Because the outcome is a change score, negative group means indicate a decrease in the recorded severity score over the specified period. The registry lists mean change and standard deviation for each group. The posted means are -1.22 for optimized treatment with lithium and -1.48 for optimized treatment without lithium. These summaries describe the observed groups; by themselves, they do not establish whether the between-group difference is distinguishable from chance variation.

A global severity measure is useful when a trial’s question concerns overall illness rather than a single symptom. It can capture a broad clinical assessment, but it does not show which component—depression, mania, or another aspect of illness—accounts for a score change. The secondary symptom measures offer separate views of selected domains, but they do not replace the primary endpoint or eliminate uncertainty about the trial’s overall comparison.

The registry record does not report a minimal clinically important difference for this scale or a threshold defining a meaningful change. Statistical evidence about a group difference should therefore be read separately from judgments about clinical importance.

7. Results Posted in the Registry

• OPT Without Lithium: mean -1.48 versus -1.22 on control; difference -0.26 (95% CI -0.59 to 0.07), standardized effect d = -0.18; z-test p = 0.128.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Depression Symptoms as Measured Self Report Montgomery Asberg Depression Rating Scale (MADRS)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • OPT Without Lithium: mean 8.84 versus 8.20 on control; difference 0.64 (95% CI -2.38 to 3.66), standardized effect d = 0.05; z-test p = 0.678.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between OPT Without Lithium and the control. The interval gives the range of effects that remain compatible with the data.

Mania Symptoms as Measured by the Young Mania Rating Scale (YMRS)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • OPT Without Lithium: mean 5.79 versus 6.35 on control; difference -0.56 (95% CI -2.73 to 1.61), standardized effect d = -0.06; z-test p = 0.613.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between OPT Without Lithium and the control. The interval gives the range of effects that remain compatible with the data.

Suicidality as Measured by the Modified Scale for Suicidal Ideation (MSSI)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • OPT Without Lithium: mean 1.10 versus 0.73 on control; difference 0.37 (95% CI -0.55 to 1.29), standardized effect d = 0.09; z-test p = 0.431.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between OPT Without Lithium and the control. The interval gives the range of effects that remain compatible with the data.

The posted secondary outcomes cover depression symptoms, mania symptoms, and suicidality. Across these measures, the unadjusted recalculated intervals include zero, and the reported mixed-model p-values do not indicate clear evidence of a group difference. The overall pattern is therefore consistent with the primary outcome’s lack of clear evidence for a difference, rather than providing a contrasting signal across the symptom domains.

This broad consistency should not be mistaken for proof that the groups had identical outcomes. Each secondary comparison has uncertainty, and the registry reports no confidence intervals for the mixed-model results. Further, these measures address selected symptoms and do not constitute an exhaustive assessment of bipolar illness. They complement the broad primary severity measure but cannot independently establish the overall effect of the treatment strategies.

Secondary findings generally carry less weight than a prespecified primary endpoint because they may be exploratory and because examining multiple outcomes increases the opportunity for chance findings. In general, multiplicity adjustment or a clearly prespecified hierarchy can help control interpretation across many tests. Whether either approach was used is not reported in the registry record. The number of posted outcome measures is 5, and the information analyzed includes 3 secondary outcomes.

The secondary outcomes were measured over 6 months, while the primary outcome specifies baseline and week 24 as relevant time points. The registry record does not report enough detail to determine whether the modeling and observation schedules were identical across outcomes. The secondary results should thus be read as additional descriptive and inferential evidence, not as a reason to replace the primary analysis or to make claims beyond the measurements recorded.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Depression Symptoms as Measured Self Report Montgomery Asberg Depression Rating Scale (MADRS)
• Mania Symptoms as Measured by the Young Mania Rating Scale (YMRS)
• Suicidality as Measured by the Modified Scale for Suicidal Ideation (MSSI)

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous, and the registry reports it as a mean with a standard deviation. The mean describes the center of the observed scores or changes; the standard deviation describes how dispersed participant values are around that mean. Neither statistic alone describes the uncertainty in the difference between groups. That uncertainty is represented by a confidence interval or assessed through a statistical test.

What does the mixed models result represent?

The registry reports a mixed models analysis for the primary outcome, with an F value for the main effect of 0.94 and p = 0.33. In general, a mixed model can account for related measurements from the same participant and can represent patterns over time, depending on how the model is specified. The registry record does not report the model formula, covariance structure, included covariates, or detailed handling of repeated observations. Therefore, the particular features of the fitted model cannot be reconstructed from the reported label alone.

The F value and its p-value summarize evidence against the model’s null hypothesis for the stated main effect. A p-value is not the probability that the null hypothesis is true, nor is it the probability that the observed difference arose only by chance. It measures how unusual data at least as incompatible with the null as those observed would be under the test’s assumptions. The registry record does not give a confidence interval for the mixed-model result, so the size and precision of that model’s estimated effect are not available from the reported fields.

What is the unadjusted recalculation?

The recalculated mean difference is based on the posted means, standard deviations, and arm sizes. It compares optimized treatment without lithium against optimized treatment with lithium, yielding an estimate of -0.26. The accompanying confidence interval and p-value are from a simple, unadjusted comparison rather than the registry’s mixed models analysis. This kind of recalculation is useful as a transparent check of the summary statistics, but it does not reproduce a model that may account for the trial’s longitudinal data or other features.

In general, a confidence interval gives a range of effect estimates compatible with the data and the method’s assumptions. Its width reflects statistical precision: wider intervals indicate greater uncertainty, while narrower intervals indicate greater precision. A confidence interval that includes zero does not establish that the true difference is zero; it indicates that zero is among the values compatible with the interval’s method and data.

Why can analysis methods differ?

The simple calculation treats the posted group summaries as the inputs to a direct comparison. A mixed model may use repeated observations and can incorporate modeling assumptions that are not reflected in those summaries. The registry record does not report enough detail to identify exactly why the reported analysis and the unadjusted calculation differ in their estimates or p-values. They should therefore be presented as distinct analyses, not combined into one conclusion.

The arm means and standard deviations also do not disclose the distribution of scores. In general, means and standard deviations can hide skew, clusters, floor or ceiling effects, and unusual observations. The registry record does not report distributional diagnostics or alternative analyses for the primary endpoint. This limits evaluation of how well the model assumptions fit the observed data.

What can the p-values say?

The reported p = 0.33 belongs to the mixed models analysis; p = 0.128 belongs to the unadjusted recalculation. They answer questions under different methods, so the difference between them should not be interpreted as a contradiction by itself. Neither p-value quantifies the clinical importance of a difference, and neither indicates the probability that one treatment strategy is better. Interpretation should consider the estimated effect, its uncertainty, the endpoint’s meaning, and the completeness of the analysis information.

For the recalculation, the confidence interval ranges from -0.595 to 0.075. Since the comparison is without lithium minus with lithium, negative values indicate a more negative mean change in the no-lithium group; positive values indicate a more negative mean change in the lithium group. The interval covers both directions and zero. This is an estimate-level description, not a conclusion that either strategy caused a particular outcome.

General statistical caution applies when multiple outcomes are examined: the chance of at least one apparently unusual result can increase as the number of tests increases. The registry record does not report whether the secondary-outcome analyses were adjusted for multiplicity. The primary outcome remains the main stated endpoint, while secondary findings are best treated as additional, less definitive evidence.

11. Confidence Intervals and Precision

The unadjusted recalculation of the primary mean difference is -0.26, with a 95% confidence interval from -0.595 to 0.075. The comparison is optimized treatment without lithium minus optimized treatment with lithium. The interval includes zero and also includes estimates on either side of zero. Consequently, the posted summary statistics do not pin down the direction of the underlying mean difference with high precision.

In general, a confidence interval is a range generated by a statistical method that, over repeated use under its assumptions, would cover the target parameter at the stated rate. It is not a range containing a known proportion of individual participant outcomes, and it is not a probability statement about a fixed parameter after observing the data. The interval’s interpretation also depends on the assumptions used to construct it.

The registry’s mixed models primary analysis reports an F value and p-value but no confidence interval. The unadjusted interval therefore describes the recalculation, not the mixed-model estimate. Because those analyses use different methods, the interval should not be presented as uncertainty around the sponsor-reported model result.

The arm standard deviations are 1.5 for optimized treatment with lithium and 1.37 for optimized treatment without lithium. These describe the spread of participant-level changes, not the precision of the estimated group difference. Precision of the comparison depends on the amount of information and the statistical method. The registry record does not report a model-based interval or enough model details to assess that interval’s precision. Overall, the recalculation is compatible with a range of small differences in either direction, and the posted model result does not supply an interval that would narrow that uncertainty.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are participants per arm, not total event counts. They are planning illustrations based on the observed effect, not the trial’s original design calculations. The original sample-size justification and its assumptions are not reported in the registry record.

The illustration assumes a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It considers detecting the stated fraction of the observed effect. The 80% power column describes the number of participants per arm that the calculation estimates would give an 80% chance of detecting the targeted effect under those assumptions, if that effect were the true one. The 90% column uses the same interpretation with a higher target chance of detection.

Rows for 100%, 75%, and 50% of the observed effect show how the planning result changes when the target effect is smaller than the observed estimate. Smaller effects require more participants per arm to achieve the same target power under the stated assumptions. For the primary outcome, the entries range from 480 and 642 participants per arm for planning for 100% of the observed effect to 1,917 and 2,566 for planning for 50% of the observed effect, at the respective 80% and 90% targets.

These are hypothetical planning calculations, not observed power. They do not tell how likely the trial’s result was to be correct, and they do not revise the trial’s actual enrollment. Their usefulness depends on whether the assumed effect and calculation assumptions are appropriate for a future study.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -0.260; pooled standard deviation σ = 1.436.
2. n = 2 × (1.96 + 0.842)² × 1.436² / -0.260² = 479.0, rounded up to 480 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- -0.26                      # observed difference in means
sigma <- 1.43624027272474                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(667745)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 480 (80% power), 642 (90% power)
 75% of observed effect: n per arm = 852 (80% power), 1,141 (90% power)
 50% of observed effect: n per arm = 1,917 (80% power), 2,566 (90% power)
power.t.test: 480 
Simulated power at n = 480 per arm: 0.801 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The registry reports the primary outcome analysis as a mixed models analysis, with an F value for the main effect of 0.94 and p = 0.33. This is the analysis reported by the trial. It does not include a confidence interval in the registry record.

The unadjusted recalculation uses the posted means and standard deviations to estimate the difference between groups. It gives a mean difference of -0.26, with a 95% confidence interval from -0.595 to 0.075 and an unadjusted p-value of 0.128. The estimate is for optimized treatment without lithium minus optimized treatment with lithium. This calculation is not the same as the mixed models analysis and does not reproduce its modeling choices.

The registry record does not report enough detail about the mixed model to explain exactly how it relates to the simple comparison, including whether and how repeated measurements or other variables entered the model. The two results should therefore be read as outputs of different methods, rather than as competing versions of one identical calculation. The recalculation is an aid to understanding the posted arm summaries; the trial’s reported model remains the sponsor’s own analysis. No discrepancy is flagged in the registry record.

15. Multiplicity, Missing Data and Analysis Populations

The registry record does not report whether statistical adjustments were made for testing multiple outcomes, whether a hierarchy among endpoints was prespecified, or whether the secondary analyses were considered confirmatory or exploratory. This matters because repeated testing can increase the chance of obtaining a seemingly notable result even when no true effect exists. Without a reported multiplicity strategy, secondary p-values should be interpreted cautiously and should not be treated as if they were the only tests performed.

The record lists a primary outcome and three secondary outcomes in the information, while the number of posted outcome measures is 5. It does not report a full statistical analysis plan or identify how the remaining measures relate to the reported analyses. The choice of statistical tests, model specification, covariate adjustment, handling of repeated measurements, and planned contrasts are not reported in the registry record in sufficient detail for independent reconstruction.

Missing-data handling is not reported in the registry record. The reported arm sizes are 141 for optimized treatment with lithium and 142 for optimized treatment without lithium, but those numbers alone do not explain how missing observations at follow-up were handled or whether all randomized participants contributed to each analysis. The record also does not report reasons for missing observations, sensitivity analyses, or assumptions about missingness.

The analysis population is not reported in the registry record. It is therefore not possible to determine from the posted information whether the analysis followed all randomized participants, included only participants with available outcome data, or used another population definition. Details of interim analyses, stopping rules, and the trial’s original sample-size justification are also not reported. These gaps limit assessment of prespecification, robustness to missing data, and the extent to which reported analyses reflect the full randomized comparison.

16. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: suicidal ideation (4.2%), cholecystitis (1.4%), mania (1.4%), alcohol detoxification (0.7%), chron's disease (0.7%).

The serious adverse event figures are reported separately from the efficacy outcomes. For optimized treatment with lithium, 141 participants were listed as at risk and 14 as affected by a serious adverse event. For optimized treatment without lithium, 142 were listed as at risk and 12 as affected. These are counts, not estimates of comparative risk, and the registry record does not provide a statistical comparison or confidence interval for them.

The listed serious-event terms and percentages include suicidal ideation at 4.2%, cholecystitis at 1.4%, mania at 1.4%, alcohol detoxification at 0.7%, Chron's Disease at 0.7%, and dehydration at 0.7%. The record does not indicate in this listing how each term is distributed between the treatment groups. It also does not provide additional detail on event timing, severity beyond the serious-event classification, or relatedness to treatment.

Other adverse events affected 69 participants at risk in the lithium group and 61 in the group without lithium, with 141 and 142 listed as at risk, respectively. Death counts are not reported in the registry record. These summaries do not establish whether the groups differ in safety, and they should not be read as a complete account of the benefits and harms of either strategy.

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

The registry’s primary analysis reports no clear statistical evidence of a between-group difference in change in overall bipolar illness severity. The simple recalculation is also uncertain: its confidence interval includes zero and extends in both directions. These findings describe the evidence available in the posted outcomes; they do not establish that the treatment strategies are equivalent or that any difference is clinically unimportant.

The observed mean changes were negative in both groups, indicating decreases in the recorded severity score over the specified period. The registry’s scale description places more severe illness at higher scores. However, group averages do not reveal the range of individual experiences, and the posted results do not show whether individual participants improved, worsened, or remained unchanged. The magnitude considered meaningful is not reported in the registry record.

The outcome combines assessments of depression, mania, and overall illness. A difference or lack of difference on this combined measure would not necessarily imply the same pattern on every symptom domain. Secondary results provide additional measurements, but they do not remove uncertainty about the primary comparison or resolve the limitations in analysis reporting.

These results concern the trial’s stated comparison: lithium added to optimized medication treatment versus optimized medication treatment without lithium. They do not establish effects of lithium in other settings, treatment regimens, or populations. The registry record does not report enough detail about participant eligibility, treatment delivery, missing observations, or analysis populations to extend interpretation beyond the recorded trial description. No individual treatment recommendation follows from these group-level summaries.

18. Why This Trial Matters Statistically

This trial is statistically informative because it asks a comparative effectiveness question using a randomized, parallel-group design and a broad measure of illness severity. The primary endpoint combines depression, mania, and overall illness into one continuous score, making the target of inference an average change in a holistic measure rather than a single symptom. The registry also records separate secondary assessments that help describe distinct symptom domains.

The contrast between an arm-level recalculation and the reported mixed models analysis illustrates why method matters. A simple comparison of posted summary statistics is transparent and easy to interpret, but it does not reproduce a model that may account for repeated observations or other features. The registry’s limited model detail means readers cannot fully assess how the sponsor’s analysis was specified, while the reported result remains the trial’s stated analysis.

The confidence interval for the recalculated difference is an important complement to the p-value. It shows that the estimate is uncertain and compatible with values on either side of zero. This is more informative than treating a p-value as a binary verdict, but the missing model-based interval and unreported analysis details constrain conclusions about precision.

The trial also highlights a common interpretive distinction: absence of clear statistical evidence is not evidence of equivalence. Equivalence requires an explicit margin and a suitable design, neither reported in the registry record. The secondary outcomes and adverse-event counts add context, yet unreported multiplicity and missing-data procedures limit their inferential weight. Careful reading separates observed summaries, model-based evidence, and questions that the registry record does not answer.

19. 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 specific to this trial follow from the available registry information. The primary mixed-model result is reported as an F value and p-value without a confidence interval. The model’s specification, including its handling of repeated observations and any covariate adjustment, is not reported in the registry record. This limits independent assessment of the primary analysis and prevents direct evaluation of its estimate’s precision.

The posted arm summaries support a simple recalculation, but that calculation is unadjusted and does not necessarily match the trial’s longitudinal analysis. It should not be treated as a substitute for the model-based result. The registry record does not report an analysis plan detailed enough to determine which result was prespecified or how the reported analysis relates to all planned analyses.

Missing-data procedures, reasons for missing observations, sensitivity analyses, and analysis population are not reported in the registry record. The arm sizes alone cannot establish whether all randomized participants contributed data at every relevant time point. The original sample-size justification and interim-analysis procedures are also not reported.

The record describes single masking without identifying who was masked, and it does not provide detailed eligibility criteria or medication implementation procedures in the information. The registry does not report a clinically meaningful threshold for the primary scale. Finally, multiplicity handling for the secondary outcomes is not reported. Together, these omissions limit evaluation of robustness, clinical importance, and the extent to which the reported findings generalize beyond the specific trial comparison.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Effectiveness of Lithium Plus Optimized Medication in Treating People With Bipolar Disorder — NCT00667745.
• Registry record last updated 2018-02-13; status last verified 2010-07.
• Results first posted 2013-06-28.
• 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.

23. Record Summary

NCT00667745 is a completed phase 4 randomized, single-masked, parallel-group effectiveness trial sponsored by the National Institute of Mental Health. It compared lithium plus optimized medication treatment, as needed, with optimized medication treatment, as needed, without lithium in people with bipolar disorder.

The primary outcome was change in Clinical Global Impression for Bipolar Disorder Severity score from baseline to week 24. The registry reports a mixed models main-effect F value of 0.94 and p = 0.33. An unadjusted recalculation gives a mean difference of -0.26, with a 95% confidence interval from -0.595 to 0.075. The registry record does not report a confidence interval for its mixed-model result.

Secondary outcomes addressed depression symptoms, mania symptoms, and suicidality. Their overall pattern does not provide clear evidence of group differences, but multiplicity handling and missing-data procedures are not reported in the registry record. Serious adverse-event counts are posted by group, without a reported comparative analysis. The findings should be read as uncertain group-level evidence, not as proof of equivalence or a clinical recommendation.