Study to Develop a Reliable Nomogram That Incorporates Clinical and Genetic Information: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study to Develop a Reliable Nomogram That Incorporates Clinical and Genetic Information (NCT00401414), a non-randomized na study. The primary outcome is Mean Percentage of Time That INR Within Therapeutic Range Using Linear Interpolation (Rosendaal et al)..

2. Clinical Question

3. Trial Design

The CReating an Optimal Warfarin Nomogram trial is listed as a completed study sponsored by Brigham and Women's Hospital. Its registry identifier is NCT00401414. The record gives a start date of 2007-01, a primary completion date of 2011-06 and an enrollment of 344. The phase is listed as NA.

The listed allocation is non-randomized, masking is none, and the model is single group. These design descriptors matter for interpretation: the record does not describe randomized assignment that would make groups comparable by design. Although Dosing Algorithm A, Dosing Algorithm B and Dosing Algorithm C are listed in the adverse-event information, the record explicitly cautions that they are not a like-for-like control comparison. Differences among those groups therefore cannot be read as unbiased treatment contrasts from the information available.

The intervention is warfarin, and the study aim was to develop a nomogram for warfarin dosing using clinical and genetic information. The arm description says rapid-turnaround genetic testing and monthly nomogram modification, if necessary, were intended to guide induction and maintenance. The brief summary describes the broader challenge of finding and maintaining an appropriate dose and explains the study's interest in using health and genetic information to guide dosing. The detailed protocol, including procedures beyond those descriptions, is not reported in the registry record.

The primary endpoint concerns time in the therapeutic INR range over 90 days, estimated by linear interpolation. The listed secondary outcomes concern time to a first therapeutic INR, the per-patient percentage of INRs outside the therapeutic range, time to stable anticoagulation in days, and the proportion of patients with serious adverse clinical events. The record lists five outcome measures but does not report the primary endpoint estimate or the results for these secondary outcomes. It also records zero analyses. Consequently, the study's design and intended outcomes can be described, but the registry information does not support an assessment of efficacy or comparative performance.

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: We enrolled patients over 9 months, following each patient for 3 months with twice weekly coagulation testing of the prothrombin time standardized to the International Normalized Ratio, and adjusted monthly the nomogram (if necessary) to improve the fit with emerging data from the cohort.

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 Mean Percentage of Time That INR Within Therapeutic Range Using Linear Interpolation (Rosendaal et al)., measured over 90 Days.

The primary endpoint is the mean percentage of time that INR was within the therapeutic range during a 90-day timeframe. INR is a standardized way of presenting prothrombin-time test results for people taking warfarin; the registry description notes that the formula adjusts for variation in laboratory testing methods so that results can be compared. The target INR was 2.0-3.0, while the study defined the therapeutic range more broadly as 1.8-3.2 to allow for measurement error and avoid problems inherent in overcorrection.

Linear interpolation, the method named in the endpoint, estimates how the measured INR changes between recorded observations. This permits an estimate of the amount of time spent in range rather than treating each test as if it represented only the instant it was taken. As a general statistical concept, a time-in-range measure can capture both the frequency and duration of control within a prespecified interval. Its meaning depends on the chosen interval, observation period and method for handling values between tests.

The endpoint is continuous, and the record specifies a mean and standard deviation as its summary measures. However, the observed mean, standard deviation, sample size for the endpoint analysis and any uncertainty interval are not reported in the registry record. The endpoint definition therefore identifies what was intended to be measured, but does not give the numerical result or show how precisely it was estimated.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the primary endpoint summarize?

The primary outcome is a continuous measure: the mean percentage of time that INR was within the defined therapeutic range. A continuous endpoint can take many values across a scale, rather than being limited to categories such as event or no event. The registry specifies the mean as the summary measure and the standard deviation as the dispersion measure. The mean describes the central value across observations; the standard deviation describes how spread out the observations are around that mean. Neither numerical summary is reported in the registry record.

The endpoint uses a 90-day timeframe and linear interpolation, as named in the registry. In general, interpolation estimates values between observations according to an assumed path between them. For an INR time-in-range measure, the method is intended to represent estimated time within the range between measurement points, rather than count only the measurements that fall in range. The record identifies the method by reference to Rosendaal et al., but does not provide additional details about its implementation, data handling or sensitivity analyses.

Why distinguish the target from the therapeutic range?

The registry specifies a target INR of 2.0-3.0 but defines therapeutic INR as 1.8-3.2 for this endpoint. The description says that this wider range allows for INR measurement error and avoids problems inherent in overcorrection. These are distinct concepts in the endpoint definition: the target is the stated goal, while the operational range determines which values count as therapeutic for calculating the primary measure. Interpretation of a time-in-range result must therefore use the endpoint's stated range, not silently substitute the target range.

What can the design say about comparisons?

The record lists non-randomized allocation, no masking and a single-group model. As a general statistical principle, random allocation helps balance both measured and unmeasured characteristics between groups, on average. A non-randomized study does not get that protection from its allocation process, so observed group differences may reflect differences in who entered each group, other factors, or the intervention itself. The registry specifically warns that the algorithm groups are not a like-for-like control comparison. Thus, their safety counts should be treated as descriptive, not as randomized evidence that one algorithm causes more or fewer events than another.

The presence of three algorithm labels alongside a single-group model does not establish a controlled comparison. The registry does not report how participants came to be represented in each algorithm group, how the algorithms differed in use, or whether formal between-group analyses were planned or performed. Those details are not reported in the registry record.

How should precision be assessed?

A point estimate alone describes a central result but does not show how uncertain it is. A confidence interval is a general statistical tool for expressing uncertainty around an estimate, given a model and its assumptions. The record does not include a primary endpoint estimate, standard deviation, confidence interval or p-value. It is therefore not possible to assess the numerical precision of the endpoint from the posted results. The count of zero recorded analyses reinforces that the registry record does not contain a documented statistical analysis for the outcomes.

The registry includes a sample-size-planning description based on an observed effect, with a two-sided significance level of 0.05, equal allocation between two arms and a normal approximation. It says that the effect to detect is a stated fraction of the observed effect. For time-to-event outcomes, the planning tables use total events under the Schoenfeld approach; for other outcomes, they use participants per arm. These are planning illustrations, not observed power and not evidence that the trial was designed around those assumptions. No specific planning values or trial-specific sample-size justification are reported in the registry record.

As a general concept, power is the probability that a specified analysis will detect an effect of a particular size under its assumptions. Planning calculations can illustrate the sample size associated with assumptions, but they do not describe the probability that a completed study has found a true effect. In particular, a calculation based on an observed effect is not a substitute for the trial's analysis and should not be called observed power. The record does not report a protocol-based sample-size justification or explain whether the planning illustration reflects the original study design.

What does over-dispersion mean in general?

For count or event data, over-dispersion means that the variability in the data is greater than a statistical model expects. When that occurs, a model that assumes too little variability can produce uncertainty estimates that are too narrow. This general concept is relevant when selecting models for event counts, but the registry does not report a count model, an over-dispersion assessment or any related analysis for this study. It would not be appropriate to infer that a particular model or correction was used.

What analytic details remain unknown?

The registry record does not report the primary endpoint analysis population, handling of missing INR measurements, treatment of participants who stopped contributing measurements, covariate adjustment, or a full statistical analysis plan. It does not describe interim analyses, subgroup analyses, or adjustments for multiple outcome tests. These omissions limit reproducibility and make it difficult to assess whether the listed endpoint and secondary outcomes were analyzed as planned. They do not show that any particular procedure was or was not used outside what is recorded; they show that those procedures are not reported in the registry record.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Dosing Algorithm A", "Dosing Algorithm B", "Dosing Algorithm C")
n   <- c(118, 147, 79)
m   <- c(58.9, 59.7, 65.8)
sp  <- c(22, 23, 16.9)
sd  <- sp      # posted standard deviations
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Dosing Algorithm A  118  58.900  22.000  54.889  62.911
CARM  Dosing Algorithm B  147  59.700  23.000  55.951  63.449
CARM  Dosing Algorithm C  79  65.800  16.900  62.015  69.585

11. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary endpoint and secondary outcomes covering first therapeutic INR, the per-patient percentage of INR measurements outside the therapeutic range, time to stable anticoagulation in days, and the proportion of patients with serious adverse clinical events. Multiple outcomes create a general multiplicity issue: when several statistical tests are conducted, the chance of at least one apparently unusual result can increase. Statistical plans may address this through a prespecified hierarchy, adjustment or other strategy. Whether any such approach was used is not reported in the registry record. The record lists zero analyses, so it does not document outcome-specific tests or results.

Missing-data handling is also not reported in the registry record. For a time-in-range measure based on INR observations, missing or irregularly spaced measurements can affect how much of the observation period contributes to the estimate. As a general principle, the impact depends on why measurements are missing and how the analysis handles gaps; no particular missingness pattern or method can be inferred from the record. The primary endpoint analysis population and the number of participants contributing to that endpoint are likewise not reported.

The registry gives an overall enrollment of 344 and separate safety denominators for the three algorithm groups, but it does not explain how those figures relate to the primary endpoint population. It does not report an intention-to-treat population, a per-protocol population, exclusions from analysis, or a participant flow for the primary outcome. These details are important for knowing which participants inform a result and how deviations from the intended approach are addressed. Their absence from the registry record limits assessment of selection and missing-data effects; it does not establish what methods may have been used in unreported study materials.

12. 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

Serious events with the highest arm-level rate: inr ≥ 4 (33.1%), major bleeding events (timi definition) (2.5%), vitamin k administration (1.7%), thrombotic events (mi, stroke, vte) (1.4%).

The registry reports serious adverse-event counts for three groups: Dosing Algorithm A had 43 affected among 118 at risk; Dosing Algorithm B had 41 among 147 at risk; and Dosing Algorithm C had 24 among 79 at risk. The other-adverse-event field is 0 in each group. Death counts are not reported in the registry record. These fields should be read as recorded counts, not as a complete account of every safety outcome or a comparison adjusted for differences between groups.

The record also lists serious-event terms and percentages: INR of 4 or higher, 33.1%; major bleeding events (TIMI definition), 2.5%; vitamin K administration, 1.7%; and thrombotic events (MI, stroke, VTE), 1.4%. The registry does not provide further denominator or analysis details for these term percentages. The terms describe events recorded in the safety information; they do not establish that an event was caused by a dosing algorithm.

Because allocation was non-randomized and the registry cautions that the groups are not like-for-like controls, raw counts cannot be interpreted as a reliable ranking of algorithm safety. Differences in group size and participant characteristics can affect descriptive counts. These safety figures should be considered alongside the absence of a reported primary endpoint analysis and the limited analytic detail in the registry record.

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

The registry record establishes that the study aimed to develop a warfarin dosing nomogram incorporating clinical and genetic information and identifies a primary endpoint for evaluating time in the therapeutic INR range. It also states a 90-day timeframe, a linear-interpolation method and the therapeutic range of 1.8-3.2, distinct from the target INR of 2.0-3.0. Those details clarify the intended measurement.

The record does not report the observed mean percentage of time in range, its standard deviation, a confidence interval or a p-value. Therefore, it does not show the size or precision of the primary endpoint result. The study's stated aims, including the aspiration to improve dosing performance, cannot be treated as findings. The absence of a reported endpoint result is not evidence that the approach did or did not achieve the intended performance.

The adverse-event counts and terms provide descriptive safety information, but the non-randomized design and explicit warning about the groups prevent a straightforward causal comparison. The record does not report an analysis establishing whether observed differences were due to the algorithms or to other factors. Nor do the available safety fields provide a complete analysis of risk over time or all possible harms.

More broadly, this record supports a description of the intended study question, its design labels, outcome definitions and some safety counts. It does not support a conclusion that the nomogram is more accurate, safer, or more effective than another dosing approach. Such conclusions would require results and analyses that are not reported in the registry record. The appropriate interpretation is therefore bounded: the record documents the study's purpose and selected descriptive data, while leaving the primary result and its uncertainty unavailable.

14. Why This Trial Matters Statistically

This trial is statistically interesting because it links a practical dosing problem to a continuously measured outcome: the estimated percentage of time within a therapeutic INR range. That endpoint illustrates how a clinical question can be translated into a measurable quantity over a defined timeframe. The registry also makes the operational definition explicit by distinguishing the target INR of 2.0-3.0 from the therapeutic range of 1.8-3.2 and naming linear interpolation as the calculation method.

The record also shows why design labels matter. A non-randomized, single-group study can describe how a nomogram was developed or used, but it does not automatically support a causal comparison between groups. The registry's warning that the algorithm groups are not like-for-like controls is particularly important when readers encounter separate adverse-event counts. The grouping labels do not, by themselves, justify treating the counts as comparative evidence.

Finally, the record illustrates the difference between a study's aim and its statistical result. The stated ambition to improve dosing performance is not a substitute for a reported estimate, uncertainty measure and analysis. Without the primary endpoint result, readers cannot evaluate the magnitude or precision of the outcome. The listed planning assumptions are also distinct from evidence that a sample-size rationale was part of the original design. These distinctions are useful when interpreting registry records: design, measurement, planning and observed results answer different questions, and only some of those elements are reported for this trial.

15. 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 main limitation for statistical interpretation is that the registry record does not report the primary endpoint result. Although it specifies the endpoint, timeframe, therapeutic range and summary measures, it gives no observed mean percentage of time in range or standard deviation. It also does not report a confidence interval or p-value. Consequently, neither the direction nor magnitude of the observed primary outcome, nor its precision, can be evaluated.

The record identifies a non-randomized allocation, no masking and a single-group model. It also cautions that the algorithm groups are not a like-for-like control comparison. This limits interpretation of group-specific adverse-event counts as evidence of differences attributable to an algorithm. The record does not describe group assignment or provide adjustment methods that might address group differences.

Important analysis details are not reported in the registry record, including the primary endpoint population, missing-data handling, participant flow, covariate adjustment and a complete statistical analysis plan. The registry lists zero analyses. It is therefore unclear from this record how the outcome measures were analyzed, whether planned analyses were completed, or how uncertainty and multiple outcomes were handled.

The safety information is also limited: death counts are not reported, and additional denominator and analytic details for the listed event-term percentages are not given. The sample-size description refers to planning illustrations, but the record does not provide a trial-specific sample-size justification. These limitations concern what can be learned from the registry record; they should not be taken as proof that further information does not exist elsewhere.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Study to Develop a Reliable Nomogram That Incorporates Clinical and Genetic Information — NCT00401414.
• Registry record last updated 2013-08-30; status last verified 2013-07.
• Results first posted 2013-08-30.
• 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.

19. Record Summary

NCT00401414 is the completed CReating an Optimal Warfarin Nomogram (CROWN) Trial, sponsored by Brigham and Women's Hospital. The registry lists a start date of 2007-01, primary completion date of 2011-06, enrollment of 344 and phase NA. The study concerned warfarin dosing with a nomogram incorporating clinical and genetic information.

The primary endpoint was the mean percentage of time INR was within the therapeutic range, assessed by linear interpolation over 90 days. The target INR was 2.0-3.0, and the endpoint's therapeutic range was 1.8-3.2. The design is listed as non-randomized, unmasked and single group. The registry reports selected serious adverse-event counts but no primary endpoint estimate, uncertainty interval or recorded analyses. It cautions that the algorithm groups are not a like-for-like control comparison.