Intention to Prescribe/Take OAC Depending on the Number of Risk Diagrams , and Period for the Estimation of the Risk.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Intention to Prescribe/Take OAC Depending on the Number of Risk Diagrams , and Period for the Estimation of the Risk. (NCT02746107), a randomized na study. The primary outcome is Number of Participants Who Prescribed Oral Anticoagulants (OAC) According to the Number of Decision Aid Diagrams.

2. Clinical Question

The central question is whether the number of risk diagrams in a decision aid is associated with physicians’ stated intention to prescribe oral anticoagulants (OAC) for atrial fibrillation. One condition presented stroke risk on one diagram showing risk under OAC treatment; the other used two diagrams, presenting risk without and with treatment. The primary outcome was the number of participants who prescribed OAC according to the number of diagrams, measured after participants saw the decision aid.

3. Trial Design

The registry identifies this study as randomized, with no masking and a factorial model. It is marked completed, with an enrollment of 968. The record lists atrial fibrillation and stroke as conditions and a decision aid as the intervention. The study dates recorded are 2016-03 for start and 2016-06 for primary completion. Its phase is listed as NA.

The factorial structure is reflected in the listed factors: whether stroke risk appeared on one or two diagrams; whether risk was estimated over one year or five years; whether the decision concerned a virtual patient or the physician; and CHA2DS2-VASc risk-score categories from 1 to 5. These factors describe the dimensions represented in the registry record. The posted primary comparison isolates the one-diagram and two-diagram groups; the record does not provide enough detail to reconstruct the full allocation scheme or the analyses of every factorial combination.

The primary outcome is reported for 486 participants in the one-diagram group and 482 in the two-diagram group. The registry gives the number choosing OAC as 400 and 406, respectively. Its primary-outcome timeframe is after viewing the decision aid, described as 5 min. The record lists four outcome measures and twelve analyses, but it does not provide a full account of how the analyses correspond to each outcome or factorial factor.

The registry does not report the trial’s original sample-size justification, detailed allocation procedures, or a fuller protocol description. The sample-size estimates accompanying the results are planning illustrations based on the observed effect, not a record of the trial’s design assumptions.

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.

Population: age: median (min, max): 39 (25, 83)

6. Primary Endpoint

The registry lists the primary outcome as Number of Participants Who Prescribed Oral Anticoagulants (OAC) According to the Number of Decision Aid Diagrams, measured over after seeing the decision aid (5 min).

The primary endpoint is the number of participants who prescribed OAC according to whether the decision aid used one diagram or two. The registry describes the decision as occurring after the participant had viewed the risk diagram, with the timeframe recorded as 5 min. The endpoint therefore captures a stated prescribing choice made in response to a presentation of stroke-risk information.

A binary decision is useful for this question because it distinguishes participants who chose to prescribe or take OAC from those who did not. Comparing the frequency of that choice between randomized presentation groups can describe whether the diagram format was associated with different stated intentions. The endpoint does not measure actual prescribing, whether a prescription was filled, whether treatment was taken, or subsequent stroke outcomes. Those outcomes are not reported as the primary endpoint in the registry record.

The registry’s primary outcome title refers to participants who prescribed OAC, while its description uses the broader wording “prescribe/take or not the treatment.” The record does not further clarify how those wordings were reconciled in outcome collection. The results should therefore be interpreted as a decision or intention recorded after the decision aid, rather than as evidence about clinical events or treatment use over time.

7. Results Posted in the Registry

• Two Diagrams: 84.2% (406/482) versus 82.3% (400/486) on control; risk difference 1.9 percentage points (95% CI -2.8 to 6.6), risk ratio 1.02 (0.97–1.08); unadjusted two-proportion z-test p = 0.422.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is binary: each participant’s recorded decision falls into a choice to prescribe or take OAC, or a choice not to do so. For a binary outcome, a risk is the proportion of participants with the event—in this case, the prescribing choice—within a group. A risk difference compares those group risks by subtraction. A value of zero represents no difference in the two group risks; positive and negative values indicate opposite directions according to the order of the groups. This is a general explanation of the measure.

The posted recalculation gives a risk difference of 0.019 for the two-diagram comparison, with a confidence interval from -0.028 to 0.066. The point estimate is the single estimate from the posted data, while the interval expresses uncertainty around it under the method used for that recalculation. Because the interval crosses zero, the results are compatible with no difference as well as with differences in either direction. The p-value for this unadjusted recalculation is 0.422. A p-value describes how unusual data at least as incompatible with a specified null hypothesis would be if that null hypothesis and the statistical model were correct; it is not the probability that the null hypothesis is true.

The registry also reports relative measures for the recalculation: a risk ratio of 1.023 with interval 0.967 to 1.083, and an odds ratio of 1.149 with interval 0.819 to 1.611. A risk ratio compares group risks as a ratio, while an odds ratio compares group odds. A ratio of one corresponds to no relative difference. These measures use different scales from a risk difference, so their numerical values should not be compared as if they were in the same units. The intervals for both relative measures include one, indicating uncertainty that encompasses no relative difference.

The sponsor’s analysis is reported as a chi-squared analysis of risk difference. The chi-squared test is a general method for assessing association between categorical variables. The sponsor reports a risk difference of 0.5, a 95% confidence interval from -4 to 5.4, and p = 0.83. The registry does not describe the calculation details or clarify the scale of the risk-difference estimate. The unadjusted recalculation is a separate analysis, and its p-value and interval should not be substituted for the sponsor’s own reported analysis.

Randomization is intended, in general, to make treatment or presentation groups comparable in expectation, reducing systematic differences in assignment. This record identifies the study as randomized, but does not report the sequence-generation or allocation-concealment procedures. The study is listed as unmasked. In this context, the presentation format is the comparison of interest; the registry does not give further information about who knew which assignment or how outcome collection was conducted.

The factorial label means that the study considered multiple design factors. The record lists the number of diagrams, the time horizon used for stroke-risk estimates, the decision target, and risk-score categories. Factorial designs can study more than one factor within a study, but interpretation of one factor may depend on how the other factors are handled. The registry does not report the model specification, adjustment variables, interaction tests, or the exact analysis population for the primary comparison. It is therefore not possible to determine from the record whether the posted primary comparison was adjusted for the other factors.

General statistical context: confidence intervals describe a range of estimates compatible with the data and the assumptions of a method. Their width reflects both the amount of information and the variability in the observations. A narrow interval indicates greater precision for the estimated quantity; a wide interval indicates more uncertainty. An interval that includes the null value does not prove that the groups are identical. It means that the data do not distinguish the estimate from that null value at the interval’s stated confidence level, under the method’s assumptions.

For count outcomes, over-dispersion means that observed variation exceeds what a chosen probability model expects. Methods that account for over-dispersion can be useful when that feature is present. The primary endpoint is a binary participant decision, and the registry does not report an over-dispersion assessment or a model that addresses it. This general issue should not be mistaken for a reported feature of this study.

The available results do not identify how missing or incomplete primary-outcome observations were handled, whether any exclusions occurred after randomization, or whether analyses followed an intention-to-treat population. Those details are important for understanding what population the estimates represent. The registry also does not report a full statistical analysis plan, so the exact prespecified status of each analysis cannot be confirmed from this record.

10. Confidence Intervals and Precision

The recalculated risk difference is 0.019, with a confidence interval from -0.028 to 0.066. This interval crosses zero, the value representing no risk difference. It therefore does not rule out no difference, and it also leaves room for a difference in either direction. The interval conveys more information than the point estimate alone: the estimate’s direction and size remain uncertain.

The recalculated relative-effect intervals also include the no-effect value for their scales. The risk-ratio estimate is 1.023, with an interval from 0.967 to 1.083; the odds-ratio estimate is 1.149, with an interval from 0.819 to 1.611. A ratio of one is the no-relative-difference value. These intervals consequently do not establish a clear relative effect in either direction.

The sponsor’s analysis reports a 95% confidence interval from -4 to 5.4 for its risk-difference estimate of 0.5. That interval also spans zero. The registry does not clarify the scale used for these sponsor-reported values, so their numeric scale should not be treated as directly interchangeable with the recalculated risk difference. The posted analyses nevertheless both show substantial uncertainty around the question of whether the diagram formats differ.

Confidence intervals describe precision under the statistical method and assumptions used; they do not represent a range guaranteed to contain the true value. An interval that includes zero is not proof of equivalence. The record does not report the margin needed to interpret the sponsor’s non-inferiority or equivalence label, so that label cannot by itself establish that differences were sufficiently small.

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 estimates are participants per arm, not total events. They are illustrations based on the observed effect, not the trial’s own original design assumptions. The registry does not report the original design assumptions or sample-size justification.

The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is specified as a fraction of the observed effect. The estimates therefore illustrate the participant count per arm under those assumptions, rather than documenting how the randomized study was planned.

For planning for 100% of the observed effect, the estimates are 5,884 participants per arm for 80% power and 7,876 participants per arm for 90% power. For planning for 75% of the observed effect, they are 10,581 per arm for 80% power and 14,164 per arm for 90% power. For planning for 50% of the observed effect, they are 24,075 per arm for 80% power and 32,229 per arm for 90% power.

In plain terms, the 80% power column gives the estimated participants per arm for a design intended to have an 80% chance of detecting the chosen effect under the stated assumptions, if that effect is present. The 90% column uses a higher chance of detection and therefore gives larger planning estimates in these scenarios. Smaller effects need more participants to distinguish them from random variation with the same power and significance threshold. These figures are planning illustrations, not observed power.

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 = 0.823; treatment proportion p₁ = 0.842; average p̄ = 0.833.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5279 = 1.0346.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5277 = 0.4441.
4. n = (1.0346 + 0.4441)² / (0.0193)² = 5,883.4, rounded up to 5,884 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 <- 0.823045267489712                          # control proportion
p1 <- 0.842323651452282                          # 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(2746107)
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 = 5,884 (80% power), 7,876 (90% power)
 75% of observed effect: n per arm = 10,581 (80% power), 14,164 (90% power)
 50% of observed effect: n per arm = 24,075 (80% power), 32,229 (90% power)
power.prop.test: 5883.4 
Simulated power at n = 5,884 per arm: 0.801 (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. Analyses Reported by the Sponsor

The registry reports the sponsor’s analysis as a chi-squared test of risk difference. It gives an estimate of 0.5, a 95% confidence interval from -4 to 5.4, and p = 0.83. That result is the sponsor’s reported analysis for the comparison of one diagram with two diagrams.

A separate unadjusted recalculation from the posted group summaries gives a risk difference of 0.019, with a confidence interval from -0.028 to 0.066 and p = 0.422. This calculation is not a replacement for the sponsor’s analysis. The values are displayed on apparently different numerical scales, and the registry does not explain the scale or the methods in enough detail to reconcile the estimates and p-values. The discrepancy flag is not marked true in the record, but that does not supply an explanation for the different presentations.

Both analyses report uncertainty that includes no difference, but they should remain distinguishable: one is the sponsor’s own reported chi-squared analysis, and the other is a simple unadjusted recalculation from posted summary statistics. The record does not report enough methodological detail to determine whether differences arise from scaling, analysis choices, or another reason.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists four outcome measures and twelve analyses. It does not report a multiplicity-adjustment strategy, a hierarchy of outcomes, or which analyses were designated confirmatory. When multiple outcomes, comparisons, or subgroup analyses are examined, the chance of finding at least one apparently notable result can increase. Statistical procedures can account for that issue, but whether and how such procedures were used is not reported in the registry record.

The factorial study includes several dimensions: diagram count, the time horizon of the risk estimate, the target of the prescribing decision, and CHA2DS2-VASc score categories from 1 to 5. The registry does not report whether the primary comparison was adjusted for these factors, whether interactions were evaluated, or how results across the listed combinations were interpreted. It is therefore not possible to infer from the factorial label alone that all factors or interactions were included in the primary analysis.

The record reports group sizes and outcome counts for the primary comparison, but it does not describe missing-outcome handling, exclusions, or the analysis population. It does not state whether all randomized participants were included in the analysis or how participants with incomplete information were treated. These omissions limit assessment of how closely the reported comparison follows the randomized allocation.

The registry also does not report the original statistical analysis plan, interim analyses, or a prespecified multiplicity procedure. Those elements should be treated as unreported, not assumed to have been absent or present. Without that information, the posted primary outcome can be described, but the record does not establish the full set of planned analyses or how selection among multiple results was controlled.

15. Safety Results

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

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The adverse-event record lists zero serious adverse events for the one-diagram group, with 486 participants at risk, and zero for the two-diagram group, with 482 participants at risk. It also lists zero other adverse events in each group. These are recorded counts in the registry, not evidence that no event could occur outside the reported collection or reporting process.

Deaths are recorded as null for both groups. A null entry is not a count of zero; the registry record does not report a death count for these groups. The record lists no top serious adverse-event terms.

This was a study of a decision aid and stated prescribing intentions. The adverse-event figures should be read as the registry’s posted event counts, not as a comparison of clinical safety outcomes after taking OAC. The record does not provide further detail about event definitions, collection procedures, or the period over which events were monitored.

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

The posted primary comparison concerns the number of participants who stated that they would prescribe or take OAC after seeing a decision aid with one diagram or two. The registry reports 400 such choices in the one-diagram group and 406 in the two-diagram group. The recalculated estimate and the sponsor’s reported analysis both have uncertainty spanning the no-difference value. The results do not establish a clear difference in this immediate decision outcome.

That interpretation is narrower than a claim that the formats are equivalent. Failure to establish a difference does not show that any possible difference is negligible. To support an equivalence or non-inferiority conclusion, the analysis would need a stated margin and an appropriate interval interpretation; the margin is not reported in the registry record. The sponsor’s analysis is tagged as non-inferiority or equivalence, but the available record does not provide the information needed to assess that label fully.

The outcome is a stated prescribing choice after viewing the decision aid, not actual prescribing behavior or patient treatment. It does not measure adherence, clinical events, stroke incidence, or whether OAC changes those outcomes. The trial’s recorded intention outcome should not be used to make claims about treatment benefit or harm.

The result also does not resolve whether the time horizon used to describe stroke risk or the target of the hypothetical prescription changed decisions. Those are listed factorial factors, but their outcome results are not included in the posted primary comparison. The registry does not report enough detail to interpret all factorial effects or interactions.

The appropriate statistical reading is therefore limited: in the reported one-versus-two-diagram comparison, the estimates remain uncertain and do not demonstrate a clear difference in stated intention. They also do not prove equality, establish clinical consequences, or answer questions beyond the measured decision.

17. Why This Trial Matters Statistically

This trial illustrates how the format of risk communication can be studied as an experimental factor rather than treated only as background material. The primary outcome is a decision made after viewing a decision aid, which connects the presentation of risk information to an observable response while keeping the comparison focused on the diagram format.

It also illustrates why a point estimate should not be read alone. The raw counts are close in scale, but group sizes differ, and the relevant comparison is an estimate of the difference in group risks with its uncertainty—not a comparison of counts in isolation. The recalculated interval crosses zero, so the data do not pin down the direction or size of a difference with confidence.

The factorial structure is statistically relevant because a decision aid can vary along more than one dimension. Diagram number, risk-estimation period, decision target, and risk-score category are all represented in the record. Analysis of such a design depends on the model and comparison strategy, which the registry does not fully report. The study therefore highlights the importance of making analysis specifications and multiplicity handling transparent alongside the outcome estimates.

Finally, the sponsor’s analysis and the unadjusted recalculation show why statistical reporting should identify scales and methods clearly. The estimates are presented differently, and the record does not explain the difference. Readers benefit when an effect measure’s units, analysis population, adjustment choices, and uncertainty are stated in a way that allows the result to be interpreted without reconstruction.

18. 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 gives limited detail about who participated beyond the description of physicians making a prescribing decision. Eligibility criteria, recruitment procedures, and participant characteristics are not reported. The decision could concern a virtual patient or the physician imagining she or he had atrial fibrillation, but the record does not give enough detail to determine how those targets were allocated or analyzed in relation to the primary comparison.

The endpoint is an immediate stated decision after exposure to the decision aid. The record does not report whether the decision predicted actual prescribing or treatment use. It also does not report subsequent clinical outcomes for the primary comparison. Conclusions should therefore remain confined to the recorded intention outcome.

The factorial design includes several factors, but the registry does not provide the full statistical model, adjustment strategy, interaction analyses, or results for each factorial dimension. It lists four outcome measures and twelve analyses without reporting a multiplicity plan. Missing-data procedures, analysis populations, and exclusions are also not described. These gaps limit appraisal of the analysis choices and the population represented by the estimates.

The sponsor’s reported risk-difference scale is not explained, and its figures differ in presentation from the unadjusted recalculation. The record does not provide enough detail to reconcile the estimates or p-values. The sponsor’s non-inferiority or equivalence classification is not accompanied by a reported margin, limiting interpretation of that classification.

The original sample-size justification and design assumptions are not reported. The planning estimates accompanying the results are based on the observed effect and should not be mistaken for the trial’s original planning. The registry also does not report detailed event-collection methods; null death entries should not be interpreted as zero deaths.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Intention to Prescribe/Take OAC Depending on the Number of Risk Diagrams , and Period for the Estimation of the Risk. — NCT02746107.
• Registry record last updated 2016-10-31; status last verified 2016-09.
• Results first posted 2016-10-31.
• 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.

22. Record Summary

The registry record describes a completed randomized, unmasked factorial study of a decision aid for stroke-risk presentation in atrial fibrillation. Its primary comparison is one diagram versus two diagrams, with the outcome recorded after viewing the aid as the number choosing to prescribe or take OAC. The record lists an enrollment of 968 and reports 400 choices among 486 in the one-diagram group and 406 among 482 in the two-diagram group.

The sponsor reports a chi-squared risk-difference analysis with estimate 0.5, 95% confidence interval -4 to 5.4, and p = 0.83. The unadjusted recalculation reports a risk difference of 0.019, confidence interval -0.028 to 0.066, and p = 0.422. Both analyses leave uncertainty around whether the diagram formats differ. The record does not report the original sample-size justification, multiplicity plan, missing-data procedures, or full analysis specification.