Cost Comparison of Cardiac Magnetic Resonance Imaging (MRI) Use in Emergency Department (ED) Patients With Chest Pain: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Cost Comparison of Cardiac Magnetic Resonance Imaging (MRI) Use in Emergency Department (ED) Patients With Chest Pain (NCT00678639), a randomized na study. The primary outcome is Cost of Index Hospitalization.

2. Clinical Question

This is a cost comparison, not a test of whether either care approach improves health. The registry also includes a secondary question about whether the admission decision was correct against a reference standard of acute coronary syndrome at 30 days. That question concerns classification and follow-up, while the primary endpoint concerns the cost of the initial hospital visit. Keeping these questions separate helps avoid treating a difference in costs as evidence about diagnostic accuracy, later health events, or the overall value of care.

3. Trial Design

The registry describes NCT00678639 as a randomized, single-masked, parallel-group study. It lists the phase as not applicable and the status as completed. Enrollment was 110. The listed study start was 2008-01, and primary completion was 2009-03. The sponsor is Wake Forest University Health Sciences.

Participants were assigned to emergency-department observation-unit care with a stress cardiac MRI protocol or to hospital admission with usual care. The registry does not identify who was masked, how randomization was generated or concealed, or whether masking could be maintained during care. Those details are not reported in the registry record. The usual-care arm allowed the treating doctor to determine testing, so it was not described as a single specified diagnostic procedure.

The primary endpoint used intention-to-treat analysis: participants were analyzed according to their assigned group. The primary result lists 53 participants in the observation-unit group and 57 in usual care. The secondary 30-day admission-decision result has denominators of 52 and 54, respectively; the reason these denominators differ from those in the primary result is not reported in the registry record. The record describes follow-up for heart-related events, but further details of that follow-up are not reported in the registry record.

These design features define the scope of the comparison. Randomization supports a comparison between assigned strategies, while the parallel-group structure means each participant belongs to one assigned group. The available record does not give enough detail to assess the randomization procedures, masking roles, or handling of missing observations beyond the stated intention-to-treat approach for the primary endpoint.

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: Once consented, no participants were excluded from the trial prior to group assignment.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All participants were analyzed based on intention to treat.

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 Cost of Index Hospitalization, measured over Emergency Department (ED) arrival through hospital discharge, median length of stay was 28.1 hours.

The primary endpoint was the cost, in US dollars, of the index hospitalization: the hospital visit during which a participant was enrolled. The registry defines the assessment window as emergency-department arrival through hospital discharge and reports a median length of stay of 28.1 hours. The outcome therefore concerns the cost of the initial episode of care, rather than all later healthcare costs or the cost of follow-up.

Cost is often highly uneven across patients: a small number of lengthy or resource-intensive visits can sit far above the typical observation. A median summarizes the middle of the observed cost distribution and is less sensitive to extreme values than a mean. The registry reports inter-quartile range information with the group medians, describing the spread of the central portion of the costs. This gives readers information about both the typical recorded cost and its variability.

A cost endpoint is relevant to a comparison of care pathways because the approaches may involve different settings, tests, and hospital use. It does not, by itself, establish that one approach produces better health outcomes, is safer, or is preferable overall. Nor does the endpoint include costs beyond the index hospitalization unless those are separately measured. The registry’s stated primary analysis population was all participants analyzed by intention to treat.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Why report a median for cost?

General statistical explanation: Cost data can be asymmetric. Many observations may be near a typical level while a smaller number are much larger. In that setting, the median is a useful description of the middle observation and is less influenced by extreme values than the mean. It does not summarize total expenditure across participants; it summarizes a typical participant’s position in the cost distribution.

For this endpoint, the registry reports medians and inter-quartile range information. The inter-quartile range describes the central spread of the distribution, rather than the uncertainty in the estimated median or the uncertainty in a group comparison. The confidence interval for the difference addresses a different question: how precisely the comparison estimate is determined by the data and method.

What does the Hodges-Lehmann estimate represent?

General statistical explanation: The Hodges-Lehmann method is a rank-based way to estimate a difference in location between two distributions. In a two-group setting, it is commonly understood through the collection of pairwise differences between observations in one group and observations in the other. It can be useful when the outcome is not well represented by a normal distribution. It is not generally identical to subtracting one group median from the other.

Here, the registry identifies the primary comparison method as Hodges-Lehmann and reports an estimated median cost difference of 588, with a 95% confidence interval from 336 to 811. The interval gives a range of values compatible with the data under the method’s assumptions and construction. A confidence interval is not a guarantee that the true value lies within the displayed endpoints, and it is not a range containing 95% of individual costs. The record does not report a p-value for this comparison, so the estimate and interval are the available inferential summary.

How should the confidence interval be read?

General statistical explanation: A confidence interval conveys precision as well as direction. A narrower interval indicates greater precision than a wider interval on the same scale, while an interval spanning both negative and positive values would leave the direction of a difference less settled. The interval posted is entirely positive. That supports reading the reported contrast as positive under the registry’s comparison convention. Readers should still check which group is subtracted from which before assigning a plain-language direction to a signed estimate.

The interval is an estimate-based uncertainty summary, not a prediction interval for a future participant’s cost. Individual costs may vary much more widely than the uncertainty around the group comparison. The displayed group medians and spread information are therefore useful alongside the difference estimate.

What does intention to treat contribute?

General statistical explanation: An intention-to-treat analysis keeps participants in the groups to which they were randomized, regardless of subsequent treatment changes or adherence. This approach aims to preserve the benefits of randomized assignment and estimates the effect of assignment to a care strategy. It is different from an analysis restricted to participants who completed every assigned procedure.

The registry states that all participants were analyzed based on intention to treat for the primary endpoint. It does not describe how missing cost values were handled or whether every randomized participant contributed an observed cost. The primary group counts are 53 and 57, while the secondary 30-day outcome has denominators of 52 and 54. The record does not explain the difference, so it should not be attributed to a particular cause.

How do secondary comparisons differ?

General statistical explanation: Binary outcomes record whether an event or classification occurred. A Fisher exact test evaluates evidence of association in a two-group table and is often used when counts are limited. The secondary 30-day admission-decision analysis reports this test and a p-value of <0.001. That p-value concerns the specific secondary comparison; it is not a p-value for the primary cost endpoint.

A p-value measures how unusual the observed data, or more extreme data, would be under a specified null model. It is not the probability that the null hypothesis is true, nor does it state the size or practical importance of a difference. The secondary outcome also has its own denominators and definition. Its result should not be transferred to the cost endpoint.

Why distinguish reported analysis from planning illustrations?

General statistical explanation: A planning calculation can illustrate how a sample size might depend on an assumed effect and design choices. Such an illustration is not a count of participants actually observed and is not “observed power.” The registry notes that its planning illustration uses the observed effect, a two-sided significance level of 0.05, equal allocation, and a normal approximation. Its stated effect scenarios are fractions of the observed effect. These are planning assumptions, not evidence that the original trial was designed with those assumptions.

The trial’s own sample-size justification, statistical analysis plan, and any interim analyses are not reported in the registry record. Readers should therefore treat the planning display as a separate illustration and not as documentation of the original design. The posted primary analysis remains the relevant result for the cost endpoint.

10. Analyses Reported by the Sponsor

The registry reports the primary cost comparison using the Hodges-Lehmann method, with an estimated median cost difference of 588 and a 95% confidence interval from 336 to 811. The group medians and their spread information provide descriptive context, while the reported estimate and interval summarize the between-group analysis. No p-value is listed for this primary result.

The re-calculation shown on the page is a simple, unadjusted check based on the posted summary information. It should not be treated as a substitute for the registry’s stated analysis: the Hodges-Lehmann comparison uses information about the distributions that is not captured by subtracting the two medians. The discrepancy check does not flag a disagreement. In the absence of a flagged discrepancy, the registry’s own analysis is the appropriate reported inferential result, and the re-calculation is best read as a supplementary check rather than a competing estimate.

This comparison does not establish the trial’s original analysis plan or sample-size rationale. Those details are not reported in the registry record. The interpretation is also limited to the index-hospitalization cost endpoint and the participants included in the stated intention-to-treat analysis.

11. Multiplicity, Missing Data and Analysis Populations

The registry identifies one primary outcome: cost of the index hospitalization. It also lists four secondary outcomes, including the 30-day admission-decision measure, MRI completion, indicated healthcare procedures, and adverse events during MRI scanning. One secondary result is described in detail in the posted results; outcome-specific results for the other listed measures are not reported in the registry record.

The record does not report whether the investigators adjusted for multiple comparisons, specified a hierarchy among secondary outcomes, or treated secondary tests as exploratory. When several outcomes are examined, unadjusted p-values do not account for the increased opportunity to observe a small p-value by chance across the set. The secondary p-value should thus be understood as belonging to its particular comparison, not as a multiplicity-adjusted conclusion about the trial as a whole.

The primary population is described as intention to treat. The registry does not report details of missing-data handling, the number of participants with missing primary cost observations, or sensitivity analyses for missing data. The secondary result has different denominators from the primary cost result, but the reason is not reported in the registry record. A difference in denominators alone does not establish why observations were unavailable.

The registry record also does not report the original sample-size justification, the planned primary analysis in a protocol or statistical analysis plan, or interim analyses. The sample-size section on this page is a planning illustration based on the observed effect and stated assumptions; it does not document the trial’s original design and is not observed power. These omissions limit assessment of how outcomes and analyses were selected before the data were examined.

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

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hospitalization for chest pain (8.8%), hospitalization for abdominal pain due to cholecystitis (1.9%), hospitalization for neck surgery (1.9%), hospitalization for cerebral infarct (1.8%), hospitalization for chest pain and suicidal ideation (1.8%).

The registry lists serious adverse events affecting 2 of 53 participants in the emergency-department observation-unit group and 7 of 57 in usual care. It lists other adverse events affecting 6 of 53 and 4 of 57, respectively. These are descriptive participant counts, not a randomized estimate adjusted for follow-up duration or other factors. The record does not provide a comparative test or confidence interval for these safety counts.

The listed serious-event terms include hospitalization for chest pain at 8.8%, hospitalization for abdominal pain due to cholecystitis at 1.9%, hospitalization for neck surgery at 1.9%, hospitalization for cerebral infarct at 1.8%, and hospitalization for chest pain and suicidal ideation at 1.8%. These terms describe events reported in the record; they do not establish that an event was caused by the assigned care strategy. The record does not provide enough detail to interpret the event-term percentages as comparative risks between groups.

Deaths are not reported in the registry record. The safety figures should be read alongside the study’s size and the incomplete detail about event definitions, timing, and ascertainment. They are not sufficient on their own to establish whether the care approaches differ in safety.

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

The results describe a randomized comparison of two emergency-department care strategies for patients with chest pain. For the primary endpoint, the registry reports lower median index-hospitalization costs in the observation-unit group than in the usual-care group, together with a positive Hodges-Lehmann difference estimate and a confidence interval that does not include zero. This is evidence about the recorded cost of the initial hospital visit under the analysis described in the registry.

It does not show that one approach is clinically superior, safer, or more effective. The primary endpoint is cost, not a health outcome. The secondary 30-day admission-decision result addresses a separate diagnostic-classification question, and the safety counts address yet another question. They cannot be combined into a single verdict about the overall value of either strategy without additional evidence and a defined framework for balancing outcomes and costs.

The result is specific to this trial’s population, care settings, intervention descriptions, and index-hospitalization window. It does not establish what costs would be over longer follow-up, how results would change in other settings, or whether the same cost pattern would persist under different usual-care practices. The registry does not report all details needed to evaluate missing data, adjustment for multiple comparisons, or the original sample-size rationale.

Finally, a group median is not a prediction for an individual patient. The spread information indicates that costs varied within each group, while the confidence interval represents uncertainty in the estimated group comparison. Neither quantity supplies a personal forecast. The appropriate reading is bounded: the posted data describe a cost difference for this trial’s index hospitalization, with uncertainty summarized by the reported interval, and do not settle broader clinical or economic questions.

14. Why This Trial Matters Statistically

This trial illustrates why a cost comparison needs more than two group medians. Cost distributions can be uneven, so a median and inter-quartile range describe the observed groups in a way that is less dominated by unusually high values than a mean might be. A separate between-group method, here the Hodges-Lehmann estimate, gives an inferential comparison rather than a simple subtraction of the displayed medians.

It also shows the importance of identifying the exact estimand: the endpoint is the cost of the index hospitalization, not total costs over follow-up. A result can be precise for that defined window while leaving longer-term resource use and health outcomes unresolved. Distinguishing endpoint, time window, and analysis population is central to interpreting economic results.

Randomized assignment makes the comparison of care strategies informative, and intention-to-treat analysis retains participants in their assigned groups. At the same time, the registry does not report several design and analysis details that would help readers assess how the result was planned and how incomplete observations were handled. The stated secondary and safety outcomes broaden the questions, but they do not replace the primary analysis or answer every aspect of the care comparison.

For biostatisticians and clinical researchers, the case is a reminder to separate descriptive summaries, model-based or rank-based estimates, confidence intervals, and p-values. Each has a distinct role. It is also important not to treat planning calculations based on observed effects as proof of an original design choice or as evidence of observed power. Transparent reporting of the analysis plan, missing-data procedures, and multiplicity strategy would make the inferential context clearer.

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 registry record leaves important design details unspecified. It describes the trial as single-masked but does not identify who was masked or how masking was maintained. The method of sequence generation and allocation concealment is not reported in the registry record. These omissions make it difficult to assess potential sources of bias in assignment and outcome assessment.

The primary outcome is limited to the cost of the index hospitalization. Costs beyond discharge are not part of the stated endpoint, and the record does not provide a broader cost-effectiveness analysis that combines costs with health outcomes. The usual-care group could receive testing chosen by the treating doctor, so the comparator represents a care approach rather than one fixed testing protocol.

Missing-data handling and the reason for the different primary and secondary denominators are not reported in the registry record. Nor are a sample-size justification, the original statistical analysis plan, interim analyses, or an approach to multiplicity. These gaps constrain assessment of whether analyses followed prespecified plans and how much confidence to place in secondary findings.

Only one secondary outcome has a detailed comparison in the results; the outcome-specific results for other listed secondary measures are not reported in the registry record. Safety information is presented as counts and event terms without comparative uncertainty estimates or detailed event definitions in this record. Finally, the findings come from an enrolled sample of 110 in this trial, and the record alone does not establish how well they generalize to other settings or care systems.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Cost Comparison of Cardiac Magnetic Resonance Imaging (MRI) Use in Emergency Department (ED) Patients With Chest Pain — NCT00678639.
• Registry record last updated 2018-09-11; status last verified 2018-08.
• Results first posted 2010-04-28.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT00678639 is a completed randomized, parallel-group cost comparison sponsored by Wake Forest University Health Sciences. It enrolled 110 participants with chest pain and compared emergency-department observation-unit care with stress cardiac MRI against hospital admission with usual care. The listed study start was 2008-01, and primary completion was 2009-03.

The primary endpoint was the US-dollar cost of the index hospitalization, analyzed by intention to treat. The registry reports group medians and a Hodges-Lehmann median cost difference with a 95% confidence interval. A secondary result concerns correct admission decisions against an acute coronary syndrome reference standard at 30 days. Safety counts are also posted. Several design, missing-data, multiplicity, and analysis-plan details are not reported in the registry record, so the conclusions should remain focused on the outcomes and time windows explicitly described.