Prevention of Stroke and Systemic Embolic Events in Patients With Atrial Fibrillation: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Prevention of Stroke and Systemic Embolic Events in Patients With Atrial Fibrillation (NCT00684307), a randomized phase 2 study. The primary outcome is Bleeding Events.

2. Clinical Question

3. Trial Design

This was a randomized, triple-masked, parallel-group, multicentre phase 2 study. Its official title describes an extended-release formulation of the oral direct thrombin inhibitor AZD0837 and a comparison with well-controlled, dose-adjusted vitamin-K antagonists. The registry lists a start in 2007-02, primary completion in 2008-06, and a completed status. Enrollment was 1084.

Four AZD0837 dose regimens were compared with VKA therapy targeting INR 2-3: 150 mg od, 300 mg od, 450 mg od, and 200 mg bd. The registry’s primary endpoint results list 164.0 participants in the 150 mg od group, 151.0 in the 300 mg od group, 156.0 in the 450 mg od group, 160.0 in the 200 mg bd group, and 318.0 in the VKA group. The record also lists adverse-event denominators that differ slightly from these endpoint denominators. Their definitions and reasons for the difference are not reported in the registry record.

The stated objective was to assess safety and tolerability and to inform dose selection in people with nonvalvular atrial fibrillation and at least one additional stroke risk factor. Although prevention of stroke and systemic embolic events appears in the study title, the posted primary outcome is bleeding events. The record names several secondary measures, including coagulation tests, D-dimer, and AZD0837 or metabolite concentrations. It does not report posted results for those outcomes in the information.

Further details about randomization procedures, masking procedures, treatment allocation ratios, protocol deviations, and the trial’s original sample-size justification are not reported in the registry record. The listed allocation is randomized, masking is triple, and the model is parallel; those labels do not specify all operational details.

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: Participants were enrolled in the study up to two weeks before randomisation and treatment assignment. Participants that were already treated with Vitamin K Antagonists (VKA) at the time of enrollment had their dose adjusted to achive INR \<2.0 at the time of randomisation. If this was not achieved the participant was discontinued from the study.

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 Bleeding Events, measured over 36 weeks according to protocol. For patients who discontinued treatment the time frame was <36 weeks. Mean number of weeks was 21 weeks (baseline to end of treatment visit).

The primary endpoint is the number of participants with a bleeding event while on study drug. A participant with multiple bleeding events is counted once, so this is a participant-level count of whether an event occurred, not a count of all bleeding episodes or a rate per unit of follow-up. It does not distinguish event severity in the primary outcome description.

For this endpoint, the registry gives a protocol timeframe of 36 weeks. For participants who discontinued treatment, follow-up for the endpoint was shorter than 36 weeks; the listed mean time from baseline to the end-of-treatment visit was 21 weeks. That variation in time at risk matters when interpreting the crude group counts: participants with longer observation have more opportunity to experience an event. The posted figures do not adjust for differing follow-up duration.

A participant-level bleeding outcome is relevant to the stated safety and tolerability purpose and can support dose-guiding comparisons. It is not, on its own, a measure of stroke prevention, systemic embolic events, all-cause mortality, or the total burden of bleeding events. The registry record does not report a more detailed definition of bleeding, event adjudication procedures, or a primary analysis population.

7. Results Posted in the Registry

• 300 mg od: 5.3% (8/151) versus 11.0% (18/164) on control; risk difference -5.7 percentage points (95% CI -11.6 to 0.3), risk ratio 0.48 (0.22–1.08); unadjusted two-proportion z-test p = 0.067.
• 450 mg od: 14.1% (22/156) versus 11.0% (18/164) on control; risk difference 3.1 percentage points (95% CI -4.1 to 10.4), risk ratio 1.28 (0.72–2.30); unadjusted two-proportion z-test p = 0.398.
• 200 mg bd: 10.6% (17/160) versus 11.0% (18/164) on control; risk difference -0.4 percentage points (95% CI -7.1 to 6.4), risk ratio 0.97 (0.52–1.81); unadjusted two-proportion z-test p = 0.919.
• VKA INR 2-3: 14.5% (46/318) versus 11.0% (18/164) on control; risk difference 3.5 percentage points (95% CI -2.7 to 9.6), risk ratio 1.32 (0.79–2.20); unadjusted two-proportion z-test p = 0.285.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is binary at the participant level: each person is counted as having or not having a bleeding event while on study drug. The reported group values are counts of participants with an event, accompanied by the number of participants in each group. Because multiple events in one person are counted once, the result does not measure recurrent-event burden. It also does not incorporate the exact length of each participant’s time at risk.

What does a risk difference describe?

A risk difference contrasts the event proportions in two groups by subtracting one proportion from the other. In the listed comparisons, the reference is 150 mg od and the comparison is the named alternative regimen. Negative estimates indicate a lower observed event proportion in the comparison group; positive estimates indicate a higher one. This measure is expressed on an absolute scale, rather than as a ratio. The listed estimates are -0.057 for 300 mg od, 0.031 for 450 mg od, -0.004 for 200 mg bd, and 0.035 for VKA INR 2-3.

What do the ratio measures add?

The registry calculations also list risk ratios and odds ratios with confidence intervals. A risk ratio compares event proportions as a ratio, while an odds ratio compares event odds. These measures can describe relative contrasts, but they are not interchangeable with the absolute risk difference. For example, the listed risk ratio for 300 mg od is 0.483, with a confidence interval from 0.216 to 1.077; its odds ratio is 0.454, with a confidence interval from 0.191 to 1.077. The estimates are recalculated from posted summary statistics and are unadjusted.

How should the p-values be read?

An unadjusted p-value describes how unusual data at least as incompatible with a specified null comparison would be under the test’s assumptions. It is not the probability that the null hypothesis is true, and it does not measure the clinical importance of a difference. The listed values are 0.067 for 300 mg od, 0.398 for 450 mg od, 0.919 for 200 mg bd, and 0.285 for VKA INR 2-3. These values do not account for the set of multiple comparisons. The registry record does not report a multiplicity strategy or identify a prespecified hierarchy for these contrasts.

Why does follow-up duration matter?

The endpoint timeframe was 36 weeks according to protocol, but participants who discontinued treatment could have less than 36 weeks of observation. The mean time from baseline to the end-of-treatment visit was 21 weeks. A simple participant count treats each participant as an event or non-event and does not directly account for unequal time at risk. Time-to-event methods can incorporate event timing and censoring, but the registry record does not report such an analysis for the primary endpoint. It also does not report how missing endpoint status was handled.

General statistical context: estimates based on small numbers of events can be imprecise, and a few participants changing event status can meaningfully alter an estimated group contrast. Confidence intervals communicate uncertainty in the estimate under the calculation’s assumptions. The interval is not a range containing a fixed proportion of individual participant outcomes.

General statistical context: when several treatment groups are compared, conducting multiple tests can raise the chance of at least one apparently notable result by chance. Multiplicity procedures can address that issue, but the registry record does not report whether a procedure was planned or used. Therefore the listed p-values should be understood as unadjusted pairwise calculations, rather than as multiplicity-controlled evidence.

General statistical context: an adjusted analysis could account for prognostic variables or design features, while an unadjusted summary-count calculation does not. The registry record does not report covariate adjustment for these estimates, an analysis population definition, or a model that accounts for the different follow-up periods. These omissions limit what can be inferred about the analysis strategy; they do not establish that no such methods were used outside the registry record.

10. Confidence Intervals and Precision

Confidence intervals show the range of effect estimates reasonably compatible with the observed data under the method’s assumptions. They communicate precision: narrower intervals represent more precise estimates, while wider intervals leave more uncertainty about the size and direction of the contrast. A confidence interval is not a range of outcomes for individual participants and does not assign a probability to the true effect after the data have been observed.

For 300 mg od compared with 150 mg od, the risk difference was -0.057, with a 95% confidence interval from -0.117 to 0.003. For 450 mg od, the estimate was 0.031, with a 95% confidence interval from -0.041 to 0.104. For 200 mg bd, it was -0.004, with a 95% confidence interval from -0.071 to 0.064. For VKA INR 2-3, it was 0.035, with a 95% confidence interval from -0.027 to 0.096.

All four risk-difference intervals include zero, so the posted calculations do not rule out no difference under these unadjusted methods. They also include nonzero values in both directions, leaving uncertainty about whether event proportions were higher or lower in the comparison group. The registry record does not report a prespecified clinically important margin, so the intervals cannot be judged against a stated threshold for an important difference.

The endpoint counts are modest, and the interval widths show that the observed contrasts are not precisely determined. In addition, differing observation time after treatment discontinuation is not incorporated into these simple comparisons. The intervals should therefore be interpreted as uncertainty around unadjusted contrasts in the posted participant-level counts, not as a complete account of uncertainty about treatment effects over a common follow-up period.

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 illustrations describe participants per arm for 80% power and participants per arm for 90% power. They are calculated under the stated assumptions of a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The assumed effect to detect is the stated fraction of the observed effect. The original trial’s sample-size justification and design assumptions are not reported in the registry record. These figures are planning illustrations, not observed power.

For the comparison of 300 mg od with 150 mg od, planning for 100% of the observed effect gives 363 participants per arm for 80% power and 486 participants per arm for 90% power. Planning for 75% of the observed effect gives 697 and 933 participants per arm, respectively; planning for 50% gives 1,683 and 2,252.

For 450 mg od, the corresponding figures are 1,760 and 2,355 for 100% of the observed effect; 3,045 and 4,076 for 75%; and 6,662 and 8,918 for 50%. For 200 mg bd, they are 123,023 and 164,692 for 100%; 219,488 and 293,831 for 75%; and 495,601 and 663,468 for 50%. For VKA INR 2-3, they are 1,430 and 1,914 for 100%; 2,468 and 3,304 for 75%; and 5,384 and 7,207 for 50%.

In plain language, 80% power is the probability of detecting an effect of the assumed size at the stated significance level if that effect is the true one and the planning assumptions hold. The 90% column uses a higher probability of detection and consequently requires more participants for each comparison. Smaller assumed effects require more participants because they are harder to distinguish from random variation. These illustrations use the observed, unadjusted effect estimates; they do not establish what effect the trial was designed to detect.

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 = 0.110; treatment proportion p₁ = 0.053; average p̄ = 0.081.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.3866 = 0.7578.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.3846 = 0.3237.
4. n = (0.7578 + 0.3237)² / (-0.0568)² = 362.8, rounded up to 363 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 <- 0.109756097560976                          # control proportion
p1 <- 0.0529801324503311                          # 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(684307)
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 = 363 (80% power), 486 (90% power)
 75% of observed effect: n per arm = 697 (80% power), 933 (90% power)
 50% of observed effect: n per arm = 1,683 (80% power), 2,252 (90% power)
power.prop.test: 362.8 
Simulated power at n = 363 per arm: 0.807 (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 four AZD0837 regimens and a VKA comparator, with four pairwise calculations against the 150 mg od group. The listed p-values are unadjusted. The registry record does not report a multiplicity adjustment, a primary contrast among the four regimens, or a testing hierarchy. When several comparisons are considered, interpreting each p-value as if it were the only test may overstate the strength of evidence. The posted calculations should therefore be treated as exploratory, unadjusted contrasts unless an analysis plan outside the registry record establishes otherwise.

The record does not report how missing primary endpoint data were handled, whether participants were analyzed according to randomized assignment, or whether the analysis was restricted to participants who took study drug. It also does not define an intention-to-treat, per-protocol, or safety analysis population. These details matter because the primary endpoint is described as bleeding while on study drug, while the timeframe could be shorter than 36 weeks after treatment discontinuation. No assumptions about missing outcomes or censoring can be inferred from the reported group counts.

The primary endpoint counts use denominators of 164.0, 151.0, 156.0, 160.0, and 318.0 across the five groups. The adverse-event data use denominators of 166, 152, 157, 161, and 319. The record does not explain why these sets differ. They may reflect different populations or endpoint-specific availability, but the reason is not reported in the registry record and should not be guessed.

Although the record describes the study as randomized and triple-masked, it does not report the randomization method, concealment details, masking success, or the analysis model. It also does not report how outcome assessment was conducted. These gaps prevent a full assessment of whether the posted comparisons preserve the benefits of randomization or address possible differences in observation time. They do not show that methods were absent; they indicate that the registry record does not document them.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: atrial fibrillation (2.5%), cardiac failure (2.0%), syncope (1.3%), appendicitis (1.2%), haemoglobin decreased (1.2%).

The registry’s adverse-event data report serious adverse events affecting 12 participants among 166 at risk in the AZD0837 150 mg od group, 20 among 152 in the 300 mg od group, 31 among 157 in the 450 mg od group, 27 among 161 in the 200 mg bd group, and 50 among 319 in the VKA INR 2-3 group. These are counts of participants reported as affected, alongside the stated at-risk denominators. The record also reports other adverse events affecting 46, 62, 46, 59, and 67 participants, respectively, with those same group denominators.

These figures describe the posted adverse-event counts; they do not establish that every event was caused by treatment. The record does not provide event-specific exposure-adjusted rates, a comparison adjusted for follow-up duration, or the criteria and analysis population used for these summaries. The listed denominators differ from those used for the primary bleeding endpoint, and the reason is not reported in the registry record.

The most frequent serious-event terms listed include Atrial Fibrillation at 2.5%, Cardiac Failure at 2.0%, Syncope at 1.3%, Appendicitis at 1.2%, Haemoglobin Decreased at 1.2%, and Ischaemic Stroke at 1.2%. Death counts are not reported in the registry record. These summaries should be read as descriptive safety information, not as proof that one regimen has a higher or lower overall safety risk.

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

The posted primary data describe the number of participants in each group who had at least one bleeding event while receiving study drug. The observed counts were 18.0 for 150 mg od, 8.0 for 300 mg od, 22.0 for 450 mg od, 17.0 for 200 mg bd, and 46.0 for VKA INR 2-3. The recalculated risk differences compare each listed regimen with 150 mg od, and every corresponding 95% confidence interval includes zero. These results leave substantial uncertainty about the size and direction of each contrast.

The estimates do not establish equivalence, superiority, or noninferiority. The registry record does not report an equivalence margin, noninferiority margin, or multiplicity adjustment. Nor do these unadjusted comparisons account for differences in time at risk after discontinuation. A nonsmall p-value does not demonstrate that two groups are the same, and a confidence interval that includes zero does not by itself establish that clinically important differences are absent.

The endpoint is bleeding during treatment, not stroke or systemic embolism. The study title and objective refer to stroke prevention, but the primary posted outcome does not measure the prevention of those events. The reported bleeding counts therefore cannot establish whether AZD0837 or VKA therapy prevented stroke or systemic embolic events.

The record describes a phase 2 dose-guiding study focused on safety and tolerability. The posted results can be understood as descriptive information about participant-level bleeding and adverse events across the listed regimens. They should not be interpreted as definitive comparative evidence about benefits or overall safety. The registry record does not report complete outcome definitions, a formal analysis plan, missing-data methods, or a full account of follow-up and analysis populations; each of those limits interpretation.

16. Why This Trial Matters Statistically

This study illustrates how the endpoint chosen for a clinical question shapes the interpretation of trial results. Although prevention of stroke and systemic embolic events is central to the study title, the posted primary endpoint is the number of participants with bleeding while on study drug. The primary counts therefore answer a safety-oriented question about bleeding, rather than directly addressing whether embolic events were prevented.

It also shows why counts alone need context. Participants who discontinued treatment could have less than 36 weeks of follow-up, and the listed mean time to the end-of-treatment visit was 21 weeks. A participant-level event count does not account for unequal time at risk. Comparing such counts can be informative descriptively, but it does not replace an analysis that handles event timing and differing observation periods.

The four unadjusted risk-difference intervals all include zero and span values in both directions. That pattern emphasizes the importance of reading estimates and confidence intervals together instead of treating p-values as a verdict. It also illustrates that a lack of a small p-value is not evidence of equivalence, particularly when the registry record does not report a prespecified equivalence framework or multiplicity strategy.

Finally, the design has several treatment regimens and a common reference group. This creates a family of comparisons that should be interpreted with attention to multiplicity and the study’s dose-guiding purpose. The registry record does not document how the comparisons were prioritized or adjusted. Statistically, the study is a useful example of how trial purpose, endpoint choice, follow-up, precision, and reporting completeness all constrain the claims that can be made from observed results.

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 primary endpoint is defined as participants with a bleeding event while on study drug, with multiple events in one participant counted once. This loses information about recurrent events and does not distinguish severity in the endpoint description. The detailed bleeding definition and event-adjudication process are not reported in the registry record.

Follow-up was planned for 36 weeks, but participants who discontinued treatment could have less than 36 weeks; the listed mean time to the end-of-treatment visit was 21 weeks. The posted comparisons use participant counts and do not adjust for the time each participant remained at risk. The record does not report time-to-event estimates or how participants with incomplete observation were handled.

The comparisons are unadjusted recalculations from group summary counts. The record does not report a sponsor’s inferential analysis, covariate adjustment, or the analysis population for the primary endpoint. Although the discrepancy indicator is false, that does not supply missing details about the original analysis. The p-values are unadjusted, and multiplicity handling is not reported.

The endpoint denominators and adverse-event denominators differ. The reason is not reported in the registry record. The record also does not report deaths, posted results for the named secondary outcomes, or a sample-size justification for the original trial. These limitations restrict interpretation to the descriptive counts and recalculated unadjusted comparisons available in the record; they do not establish that the omitted methods or outcomes were not collected elsewhere.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Prevention of Stroke and Systemic Embolic Events in Patients With Atrial Fibrillation — NCT00684307.
• Registry record last updated 2012-03-23; status last verified 2012-03.
• Results first posted 2011-09-22.
• 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

NCT00684307 is a completed randomized, triple-masked, parallel, multicentre phase 2 study of extended-release AZD0837 in people with nonvalvular atrial fibrillation. Enrollment was 1084. The study compared four AZD0837 regimens—150 mg od, 300 mg od, 450 mg od, and 200 mg bd—with vitamin-K antagonist therapy targeting INR 2-3.

The primary posted endpoint counts participants with a bleeding event while on study drug. The recalculated unadjusted risk differences compare each regimen with 150 mg od; all four 95% confidence intervals include zero. These estimates are uncertain and do not establish equivalence or a treatment effect on stroke prevention. The registry record does not report the original sample-size justification, multiplicity methods, missing-data handling, or a complete inferential analysis plan.