Does Atorvastatin Prevent Post Operative Atrial Fibrillation After Pulmonary Resection: A Randomized Clinical Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Does Atorvastatin Prevent Post Operative Atrial Fibrillation After Pulmonary Resection: A Randomized Clinical Study (NCT00756886), a randomized na study. The primary outcome is Atrial Fibrillation.

2. Clinical Question

The clinical question focuses on a postoperative rhythm outcome following non-cardiac thoracic surgery. The comparison is between assignment to atorvastatin and assignment to placebo, with the primary result expressed as participants with atrial fibrillation rather than as the number of separate episodes. This distinction matters: a participant with repeated events is still represented in a participant-count endpoint as a participant with an event.

The registry describes a randomized, double-masked, parallel-group study. Random allocation is intended to support a comparison between assigned groups, while masking is intended to limit influence on participants or study conduct; the record does not specify who was masked or how masking was maintained. It does not report additional eligibility criteria, how atrial fibrillation was identified, or details of outcome assessment. Those omissions limit how specifically the clinical question and its measurement can be characterized.

3. Trial Design

The registry describes a randomized, double-masked, parallel-group study of atorvastatin and placebo in the setting of pulmonary resection. The study is listed as completed, with University of Alabama at Birmingham as sponsor. The registry lists enrollment as 73. The primary outcome record gives an arm denominator of 30 for atorvastatin and 25 for placebo; the record does not explain the difference between those outcome denominators and the enrollment figure.

The atorvastatin arm description specifies 40 mg once daily for 7 days before surgery, followed by the same dosage for 14 days after surgery. The placebo arm description repeats the same dose and schedule. Since that wording is what the registry records, it cannot be used to infer the actual composition or appearance of the placebo; those details are not reported in the registry record. The registry lists allocation as randomized, masking as double, and the model as parallel. It does not name the people or roles included in masking, describe sequence generation or concealment, or state whether allocation was stratified.

The primary outcome is atrial fibrillation, recorded as participants with an event during 0-21 days post-operative. The record lists one outcome measure and no posted analyses. It lists no secondary outcomes. Phase is recorded as NA. The start and primary completion entries are 2008-09 and 2011-05, respectively. The record does not give further information about recruitment sites, participant characteristics, adherence, withdrawals, or follow-up beyond the primary outcome timeframe.

These design features define what can be interpreted from the record, but do not establish details that are absent. In particular, the registry does not report a protocol-level statistical analysis plan, a sample-size justification, or the original design assumptions. The sample-size calculations discussed elsewhere are planning illustrations based on the observed effect, not documentation of the trial's original design.

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

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 Atrial Fibrillation, measured over 0-21 days post-operative.

The primary endpoint counts participants with atrial fibrillation during 0-21 days post-operative. The registry identifies the unit as participants and the outcome type as a number of people with an event. It therefore represents whether participants experienced the event within the specified postoperative window, not how many atrial fibrillation episodes occurred or how long any episode lasted.

A participant-count endpoint is one way to describe how often an event occurred in each assigned group. It can support a comparison of the observed proportions with an event, while keeping the outcome at the participant level. It does not, by itself, describe event severity, timing within the follow-up window, recurrence, or duration. Those characteristics are not reported in the registry record.

A postoperative timeframe makes the endpoint specific to the period stated in the record, rather than to long-term rhythm outcomes. The registry's primary question concerns atrial fibrillation following pulmonary resection, and this event measure provides the outcome used for that comparison. The record does not explain how the endpoint was defined clinically, how it was detected, whether confirmation required a particular test, or whether outcome assessment was centrally reviewed. Consequently, the event counts can be interpreted as posted, but their ascertainment procedures cannot be assessed from the registry entry.

7. Results Posted in the Registry

• Atorvastatin: 20.0% (6/30) versus 4.0% (1/25) on control; risk difference 16.0 percentage points (95% CI -0.2 to 32.2), risk ratio 5.00 (0.64–38.82); unadjusted two-proportion z-test p = 0.076.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is binary at the participant level: each participant either appears among those with atrial fibrillation during the stated follow-up period or does not. For this type of endpoint, group counts can be expressed as event proportions, and comparisons can be presented on absolute or relative scales. The registry posts the event counts and denominators, while the displayed unadjusted comparisons are recalculations from those summary values.

What does the risk difference describe?

The risk difference subtracts the event proportion in the reference group from the event proportion in the comparison group. Its scale is an absolute difference in proportions. The posted estimate is 0.16, with a confidence interval from -0.002 to 0.322. A positive value indicates a larger observed event proportion in the atorvastatin arm; a negative value would indicate a larger proportion in the placebo arm. Values around zero represent little or no difference on this scale. The interval spans zero, so the data are compatible with a range that includes no difference as well as differences in either direction.

What do the risk ratio and odds ratio describe?

The risk ratio compares the event proportions between groups by division, whereas the odds ratio compares the event odds. Both are relative measures, but they are not interchangeable, especially when outcomes are not rare. The recalculated unadjusted risk ratio is 5.0, with a confidence interval from 0.644 to 38.82. The recalculated unadjusted odds ratio is 6.0, with a confidence interval from 0.671 to 53.68. For either ratio, a value of 1 represents no relative difference. Each interval includes 1 and is broad, indicating substantial uncertainty in the relative comparison.

These recalculated measures use only the posted group summaries and are unadjusted. They do not account for participant characteristics, clinical factors, or analysis choices that are not represented in the counts. No adjusted model or sponsor analysis is reported in the registry record. The discrepancy indicator is false, but that does not supply a missing analysis plan or make the recalculated comparisons adjusted estimates.

How should the p-value be read?

The posted unadjusted p-value is 0.076. In general, a p-value describes how compatible the observed data are with a specified null model, under the assumptions of the test used. It is not the probability that the null hypothesis is true, nor is it a measure of treatment benefit, harm, or clinical importance. The registry record does not identify the test used or provide a more detailed statistical analysis description. The p-value should therefore be read alongside the group counts and confidence intervals, rather than as a standalone verdict.

Why do the estimates have substantial uncertainty?

In general, estimates based on small numbers of observed events are often imprecise. A small change in the event count can materially change an event proportion or relative measure when denominators and event counts are limited. Confidence intervals express uncertainty around an estimate under the method used to calculate them. Broad intervals indicate that a range of effects remains consistent with the data and method. The registry does not report the interval-construction method, so no more specific claim about that method is possible.

The group denominators in the primary outcome are 30 and 25, and the recorded event counts are 6 and 1. These counts explain why the relative estimates are particularly unstable: the placebo event count is low, so the relative comparison is sensitive to the observed count. That is a general statistical point about sparse binary outcomes, not evidence about why the particular counts arose.

For binary outcomes, over-dispersion is not a feature of the posted participant-count summary in the same way it can be for count outcomes such as repeated events per participant. In general, over-dispersion means variability greater than a model assumes; if present for an outcome and analysis, it can make standard errors too small and intervals too narrow. The registry record does not report a model or evidence of over-dispersion. Since the primary endpoint counts participants with an event, it should not be interpreted as a count of recurrent episodes.

Randomization is a design feature that can support an unbiased comparison in expectation, while masking may reduce some influences on care, measurement, or reporting. Those general properties do not establish that all potential biases were eliminated in this particular trial. The registry does not describe the randomization process, who was masked, treatment adherence, or missing-outcome handling. Statistical interpretation should remain tied to the posted summaries and their uncertainty.

10. Confidence Intervals and Precision

The risk-difference estimate is 0.16, with a confidence interval from -0.002 to 0.322. The interval is wide relative to the point estimate and crosses zero. On the absolute scale, the posted data therefore do not pin down whether the event proportion was higher or lower in the atorvastatin arm, or by how much. Its endpoints allow a range from a value very close to no difference to a considerably larger positive difference.

The recalculated risk ratio is 5.0, with a confidence interval from 0.644 to 38.82. The recalculated odds ratio is 6.0, with a confidence interval from 0.671 to 53.68. For these relative measures, the null value is 1, and both intervals include it. Their wide ranges show that the relative magnitude is also uncertain. In particular, a point estimate alone would conceal how much uncertainty is attached to the comparison.

In general, a confidence interval describes a range of values reasonably compatible with the observed data under the interval method and its assumptions. It is not a guarantee that the true effect lies within the interval, nor does it express the probability that a particular effect is true. The registry record does not report the method used to construct these intervals, so their exact statistical properties cannot be assessed from this entry.

The observed event counts are 6 in the atorvastatin arm and 1 in the placebo arm. With few events, estimates can move substantially when counts change, and relative measures can be especially unstable when one group has a low event count. The intervals communicate that imprecision; they do not identify its cause beyond the limited event information. The posted results support describing the observed direction, but not a precise estimate of the size of a treatment effect.

11. 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 events. They are illustrations based on the observed effect, using a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each scenario asks how many participants per arm would be needed to detect the stated fraction of the observed effect under those assumptions. The trial's own original sample-size assumptions are not reported in the registry record.

For 80% power, the calculation describes the chance of detecting an effect of the specified size if that effect were present and the planning assumptions held. The 90% power column represents a higher target chance under the same general assumptions. These are planning concepts, not guarantees about the results of a future study. The listed scenarios range from planning for 100% of the observed effect to planning for 75% or 50% of it. Smaller effects require more participants per arm to detect at the same power target, which is why the required counts increase as the assumed effect becomes smaller.

For example, the listed planning figures for 100% of the observed effect are 64 participants per arm for 80% power and 85 for 90% power. For 75% of the observed effect, the corresponding figures are 97 and 130; for 50% of the observed effect, they are 180 and 240. These values are planning illustrations only, not observed power and not evidence about the probability that the trial's result is true. The calculations do not substitute for a prospective design based on a justified target effect and an explicitly specified analysis.

12. 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, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.040; treatment proportion p₁ = 0.200; average p̄ = 0.120.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.4596 = 0.9007.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.4454 = 0.3749.
4. n = (0.9007 + 0.3749)² / (0.1600)² = 63.6, rounded up to 64 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
p0 <- 0.04                          # control proportion
p1 <- 0.2                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^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 (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(756886)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 64 (80% power), 85 (90% power)
 75% of observed effect: n per arm = 97 (80% power), 130 (90% power)
 50% of observed effect: n per arm = 180 (80% power), 240 (90% power)
power.prop.test: 63.6 
Simulated power at n = 64 per arm: 0.835 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-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.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists one outcome measure and no secondary outcomes. It also lists an analysis count of 0. Thus, the record does not provide a set of multiple posted efficacy analyses that could be reviewed for multiplicity, nor does it describe how multiple endpoints, time points, or subgroup comparisons were handled. The absence of posted secondary results does not establish whether any were collected or analyzed outside the registry record.

In general, multiplicity arises when several statistical questions are examined, such as multiple outcomes or repeated comparisons. More testing can increase the chance of apparently notable findings under a null model, so a planned approach to multiplicity can matter for interpretation. No multiplicity adjustment or prespecified testing hierarchy is reported in the registry record. It would be inappropriate to infer that no such plan existed; its details are simply not reported.

The record gives primary outcome counts by arm and unadjusted recalculated comparisons, but it does not identify a statistical analysis population. It does not state whether participants were analyzed according to their randomized assignment, whether the analysis was per protocol, or how treatment changes and non-adherence were addressed. Those distinctions can affect interpretation because an analysis based on original assignment and one based on treatment received answer different questions.

Missing outcome data and withdrawals are not described. The registry does not report how many participants had complete follow-up for the endpoint, why any data might be absent, or what assumptions or methods were used to handle missingness. The arm denominators recorded for the endpoint are 30 and 25, while the overall enrollment entry is 73; the reason for the difference is not explained. It should not be treated as evidence of a particular pattern of withdrawal or exclusion.

These omissions limit assessment of the robustness of the primary comparison. They do not establish that multiplicity, missingness, or analysis-population concerns affected the result; they mean the registry entry lacks the information needed to evaluate them. A protocol or full analysis report would be needed to characterize these decisions, but such documentation is not reported in the registry record.

14. 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 1% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: respiratory distress (6.7%), pericarditis (6.7%), subcutaneous air (4.0%), pneumothorax (4.0%), elevated liver function tests (3.3%).

The adverse-event record lists serious adverse events affecting 11 participants among 30 at risk in the atorvastatin group and 3 among 25 at risk in the placebo group. It also lists other adverse events affecting 25 among 30 at risk in the atorvastatin group and 17 among 25 at risk in the placebo group. These are group-level counts of participants affected, as presented in the registry record; they are not a treatment-effect analysis.

The serious-event terms with the largest listed percentages are Respiratory Distress and Pericarditis at 6.7% each, followed by Subcutaneous Air and Pneumothorax at 4.0% each, then Elevated Liver Function Tests and Airleak at 3.3% each. The record does not provide a comparative estimate or confidence interval for each term. The percentages should not be interpreted as evidence that a particular event was caused by a study drug.

Deaths are not reported in the registry record. The record does not give further detail on event definitions, severity grading beyond the serious classification, timing, relatedness assessment, overlap between event categories, or how events were collected. The counts alone cannot establish comparative safety or explain the clinical context of an event.

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

The primary outcome counts show 6 participants with atrial fibrillation in the atorvastatin arm and 1 in the placebo arm during 0-21 days post-operative. The unadjusted risk difference is positive, while the recalculated risk ratio and odds ratio also have point estimates above their null values. These describe the observed data; they do not establish that atorvastatin increased risk or that placebo prevented the event.

The confidence intervals are broad and include the null value for each of the absolute and relative comparisons. The unadjusted p-value is 0.076. Taken together, the posted summaries do not provide a precise estimate of the size or direction of a treatment effect. The data remain compatible with a range of possibilities, and the sparse event counts make the relative estimates especially uncertain.

Randomized assignment is a useful design feature for comparing interventions, but the registry record does not provide enough methodological detail to assess randomization implementation, analysis populations, adherence, missing data, or outcome ascertainment. No sponsor analysis is posted, and no secondary outcomes are listed. The unadjusted calculations should therefore be understood as a limited reading of the available primary summary, not as a complete account of the trial's statistical analysis.

The record also does not establish how broadly the result applies beyond the study setting. Detailed participant characteristics and eligibility criteria are not reported. The result addresses atrial fibrillation in the specified postoperative window, not other postoperative outcomes, longer-term rhythm outcomes, or treatment effects in populations not described in the registry.

Accordingly, the result does not show that atorvastatin works or does not work, and it is not a basis for clinical advice. It shows the recorded event counts and an uncertain unadjusted comparison in this registry entry. Judging clinical importance would require more context about participants, outcome assessment, analysis plans, and the intended decision, none of which is fully in the registry record.

16. Why This Trial Matters Statistically

This trial illustrates why a randomized design and a treatment-effect estimate do not, by themselves, guarantee a precise answer. The endpoint is a binary participant-level event, and the observed counts are small, particularly in the placebo arm. A few events can have a large influence on estimated proportions and relative comparisons when the outcome is sparse.

Different effect measures communicate different aspects of the same comparison. The risk difference expresses an absolute contrast, while the risk ratio and odds ratio express relative contrasts. Their point estimates can look large, yet their confidence intervals are broad and include the corresponding null values. Reading estimates alongside intervals helps prevent an overly confident interpretation of a striking point estimate.

The example also shows why the p-value should not replace effect estimates and uncertainty. The unadjusted p-value of 0.076 does not say how clinically important the observed contrast is, and it is not a probability that a treatment effect exists. The registry's lack of a posted sponsor analysis or detailed analysis plan further limits interpretation of the recalculated comparisons.

Finally, registry completeness is statistically important. The record does not describe missing-data handling, the analysis population, endpoint ascertainment, or original sample-size planning. These details shape how a result should be understood and whether an estimate is robust. The primary lesson is not a treatment conclusion, but the need to distinguish observed group counts, uncertain estimates, and unanswered methodological questions.

17. 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 contains limited information about the trial population. It identifies pulmonary resection and non-cardiac thoracic surgery as the clinical setting but does not report detailed eligibility criteria or participant characteristics. This makes it difficult to judge how representative the participants were of other surgical populations.

The primary endpoint is recorded as participants with atrial fibrillation during 0-21 days post-operative, but the record does not explain the event definition or ascertainment process. It does not state how rhythm events were detected, confirmed, or adjudicated. Nor does it report recurrence, episode duration, or event timing within the follow-up window. The endpoint therefore supports interpretation only as a participant event count.

The outcome denominators differ from the enrollment entry, and the registry does not explain why. Withdrawals, missing outcome data, adherence, and treatment changes are not described. The record does not specify the analysis population or whether participants were analyzed according to random assignment. These omissions prevent assessment of potential bias from incomplete follow-up or departures from assigned treatment.

Although the study is described as randomized and double-masked, the registry does not describe sequence generation, allocation concealment, who was masked, or how masking was maintained. The placebo regimen wording repeats the dose and schedule listed for atorvastatin, and the record does not clarify that wording. Original sample-size assumptions and a detailed statistical plan are also not reported. The posted comparisons are unadjusted recalculations, and no sponsor analysis is listed.

Safety information is limited to recorded counts and selected serious-event terms. Deaths are not reported in the registry record, and details about event ascertainment, relatedness, and timing are absent. These limitations do not prove that bias or safety concerns occurred; they identify information needed for a fuller appraisal that this registry entry does not provide.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Does Atorvastatin Prevent Post Operative Atrial Fibrillation After Pulmonary Resection: A Randomized Clinical Study — NCT00756886.
• Registry record last updated 2013-04-02; status last verified 2012-12.
• Results first posted 2013-04-02.
• 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.

21. Record Summary

NCT00756886 is a completed randomized, double-masked, parallel-group study sponsored by University of Alabama at Birmingham. It compared atorvastatin with placebo in the setting of pulmonary resection. The registry lists enrollment as 73 and records the primary outcome as participants with atrial fibrillation during 0-21 days post-operative.

The posted outcome counts are 6 among 30 in the atorvastatin arm and 1 among 25 in the placebo arm. The unadjusted risk difference is 0.16, with a confidence interval from -0.002 to 0.322, and the unadjusted p-value is 0.076. Recalculated relative estimates are imprecise, with confidence intervals that include their null values.

The registry lists no posted analyses and no secondary outcomes. It does not report the original sample-size assumptions, detailed analysis plan, missing-data methods, or analysis populations. The observed difference should therefore be interpreted as an uncertain unadjusted comparison, not as a definitive treatment conclusion.