This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
EXPEDITE was a randomized, parallel-group, quadruple-masked phase 3 trial comparing MVI 200 with dinoprostone vaginal insert for reducing time to vaginal delivery during the first hospital admission. The registry reports 1358 participants, two treatment arms, two primary endpoints, 11 posted outcome measures, and 10 posted statistical analyses.
| Feature | EXPEDITE |
|---|---|
| Phase | Phase 3 |
| Status | Completed |
| Start | 2010-09 |
| Primary completion | 2012-03 |
| Enrollment | 1358 |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Lead sponsor | Ferring Pharmaceuticals |
| Primary endpoints | 2 |
| Statistical analyses posted | 10 |
| ClinicalTrials.gov | NCT01127581 |
2. Clinical Question
The central question was whether MVI 200 could reduce the time to vaginal delivery compared with dinoprostone vaginal insert during the first hospital admission, while also assessing the incidence of cesarean delivery and other delivery-related outcomes.
Population
Participants enrolled in a phase 3 randomized study addressing cervical ripening and induction of labor, with outcomes assessed during the first hospital admission.
Intervention
Misoprostol vaginal insert (MVI 200).
Comparator
Dinoprostone vaginal insert (DVI).
Primary questions
Does MVI 200 change time to vaginal delivery, and is its cesarean-delivery rate non-inferior to DVI under the registry's prespecified decision rule?
3. Trial Design
MVI 200
- Misoprostol vaginal insert
- Compared with DVI in a randomized parallel-group design
- Primary time-to-vaginal-delivery analysis reported as MVI 200 minus DVI
- Secondary delivery outcomes included time-to-event and binary endpoints
Dinoprostone Vaginal Insert (DVI)
- Dinoprostone vaginal insert
- Comparator in the randomized parallel-group design
- Used as the reference group for the reported treatment differences
- Cesarean-delivery rate registry-reported the reference for the non-inferiority framework
4. Registered Primary Endpoints
| Primary endpoint | Registry time frame | Endpoint type | Primary analysis |
|---|---|---|---|
| Time to Vaginal Delivery During the First Hospital Admission | Interval from study drug administration to vaginal delivery (average 24 hours) | Time-to-event | Log-rank test; median difference |
| Incidence of Cesarean Delivery During the First Hospital Admission | Interval from study drug administration to cesarean delivery (average 24 hours) | Binary | Chi-squared test; difference in populations |
The two primary endpoints use different statistical structures. Time to vaginal delivery is a time-to-event outcome because both the occurrence of delivery and the time until delivery matter. Cesarean delivery is treated as a binary outcome for its primary analysis, with the comparison framed around the between-treatment-group difference in event rates.
5. Statistical Methodology
Time-to-event analysis
The registry reports a log-rank analysis for time to vaginal delivery. Participants who underwent cesarean delivery during the first hospitalization were censored using the longest time interval from study drug administration to cesarean delivery during the first hospitalization, independent of treatment group. Participants discharged before delivery or who withdrew consent before delivery were also censored according to the registry's stated censoring rules.
The registry reports the treatment contrast as a median difference in minutes and evaluates the time-to-event distributions using a log-rank test.
Categorical analysis
The incidence of cesarean delivery was analyzed using a chi-squared test. The registry defines the analysis as a between-treatment-group difference in the safety population. Participants discharged before delivery, who withdrew early without a cesarean delivery, or who were lost to follow-up were classified as not having the event.
Fisher exact analyses
Secondary binary outcomes were analyzed with Fisher exact tests. The registry reports differences in proportions, expressed as risk differences, for these endpoints.
Intention-to-treat analysis
The registry identifies the ITT population for several secondary efficacy analyses, including time to active labor and delivery-within-12-hours or delivery-within-24-hours endpoints. The registry also states that the ITT population was used for all secondary efficacy analyses.
| Method | Endpoints using the method | Statistical purpose |
|---|---|---|
| Log-rank test | Time to vaginal delivery; time to any delivery; time to active labor | Compare time-to-event distributions between randomized groups |
| Chi-squared test | Incidence of cesarean delivery | Compare categorical event rates between treatment groups |
| Fisher exact test | Pre-delivery oxytocin and delivery-within-time-window endpoints; incidence of vaginal delivery | Compare binary outcomes between treatment groups |
6. Primary Result: Time to Vaginal Delivery
The primary time-to-event analysis compared MVI 200 with DVI over the interval from study drug administration to vaginal delivery during the first hospital admission.
Median difference in time to vaginal delivery
95% two-sided confidence interval: not numerically reported in the registry analysis entry
Log-rank P < 0.001
Hypothesis type: Superiority
The registry's treatment contrast is MVI 200 minus DVI. A median difference of −677 minutes therefore represents a shorter reported median time to vaginal delivery with MVI 200 relative to DVI by the amount represented by that negative difference.
The result is a time-to-event comparison, not a statement that every participant delivered 677 minutes earlier. Individual delivery times vary, and the analysis accounts for censoring according to the registry-defined rules.
The registry reports a P < 0.001 log-rank result. A p-value addresses compatibility with the null hypothesis under the specified statistical test; it does not measure the size or clinical importance of the treatment effect.
The registry does not numerically report a confidence interval for this median difference in the posted analysis entry. Consequently, the precision of the −677-minute estimate cannot be summarized numerically from that entry alone.
Why censoring matters here
Cesarean delivery creates a particularly important analytic distinction. The primary endpoint is time to vaginal delivery, so a participant who has a cesarean delivery does not experience the vaginal-delivery event. The registry therefore specifies censoring rules for such participants rather than treating cesarean delivery as a vaginal-delivery event.
This is conceptually different from the secondary endpoint of time to any delivery, for which both vaginal and cesarean delivery constitute the delivery event.
7. Primary Result: Cesarean Delivery
The second primary endpoint evaluated the incidence of cesarean delivery during the first hospital admission. The registry specifies that the analysis was based on the between-treatment-group difference in the safety population.
Difference in cesarean-delivery rates
95% two-sided CI: −5.79 to 3.59
Chi-squared analysis
Hypothesis type: Non-inferiority or equivalence
The reported contrast is MVI 200 − DVI. The estimate of −1.10 is therefore a negative between-group difference, with the registry expressing the endpoint as a difference in populations.
The 95% CI of −5.79 to 3.59 describes the uncertainty around that treatment difference under the analysis framework. Because the interval crosses zero, the data are compatible with both a negative and a positive difference in cesarean-delivery rates at the conventional two-sided 95% confidence level.
The registry's non-inferiority framework is different from simply asking whether a conventional two-sided p-value is below 0.05. The prespecified decision rule uses the upper limit of the two-sided 95% confidence interval and compares it with the calculated non-inferiority margin.
The registry defines the non-inferiority limit as 10% relative to the DVI rate, expressed in the analysis description as 0.1 times the DVI rate. The registry analysis entry does not provide the DVI event rate needed to convert that relative margin into a specific percentage-point value. The statistical logic therefore should not be replaced with a simple comparison of 3.59 against 10 percentage points.
The confidence interval is about the treatment difference, whereas a p-value would address a hypothesis-testing question. Neither quantity alone describes the clinical consequences of individual cesarean deliveries.
Non-inferiority logic
The registry specifies a 10% non-inferiority limit relative to the DVI rate. The comparison is based on the upper confidence-limit criterion rather than on a standalone superiority p-value.
This distinction is central to interpreting the cesarean endpoint. A non-inferiority analysis asks whether the data exclude an unacceptable disadvantage, rather than requiring the experimental treatment to demonstrate superiority.
8. Secondary Endpoint Results
The registry posts eight secondary statistical analyses in addition to the two primary analyses. They cover time to any delivery, time to active labor, pre-delivery oxytocin, and several delivery time-window outcomes.
| Secondary endpoint | Analysis | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Time to Any Delivery (Vaginal or Cesarean) During the First Hospital Admission | Log-rank; median difference | −543 minutes | Not numerically reported | <0.001 |
| Time to Active Labor During the First Hospital Admission | Log-rank; median difference | −390 minutes | Not numerically reported | <0.001 |
| Incidence of Pre-delivery Oxytocin During the First Hospital Admission | Fisher exact; risk difference | −26 | Not numerically reported | <0.001 |
| Incidence of Vaginal Delivery Within 12 Hours | Fisher exact; risk difference | 9.67 | Not numerically reported | <0.001 |
| Incidence of Any Delivery Within 24 Hours | Fisher exact; risk difference | 26.96 | Not numerically reported | <0.001 |
| Incidence of Any Delivery Within 12 Hours | Fisher exact; risk difference | 13.90 | Not numerically reported | <0.001 |
| Incidence of Vaginal Delivery Within 24 Hours | Fisher exact; risk difference | 29.80 | Not numerically reported | <0.001 |
| Incidence of Vaginal Delivery | Fisher exact; risk difference | 1.68 | Not numerically reported | <0.001 |
All of the reported secondary superiority analyses have P-values below 0.001. The registry expresses the binary secondary effects as differences in proportions, with MVI 200 minus DVI for the reported risk differences.
Time to any delivery
The reported median difference was −543 minutes. Unlike time to vaginal delivery, the event is delivery of the neonate by either vaginal or cesarean route.
Time to active labor
The reported median difference was −390 minutes. This analysis used the ITT population and treated participants who never entered active labor as censored under the registry's stated rule.
12-hour outcomes
The registry reports risk differences of 9.67 for vaginal delivery within 12 hours and 13.90 for any delivery within 12 hours.
24-hour outcomes
The registry reports risk differences of 29.80 for vaginal delivery within 24 hours and 26.96 for any delivery within 24 hours.
9. Understanding the Secondary Risk Differences
A risk difference is an absolute difference in event proportions between treatment groups. Here, the registry defines the reported contrast as MVI 200 minus DVI.
A positive risk difference indicates a higher event proportion in the MVI 200 group; a negative risk difference indicates a lower event proportion in the MVI 200 group.
For example, the reported risk difference of 29.80 for vaginal delivery within 24 hours is an absolute between-group difference in the percentage of participants meeting that endpoint. It is not a relative risk, odds ratio, or hazard ratio.
Similarly, the reported −26 difference for pre-delivery oxytocin indicates a negative treatment-group difference under the registry's MVI 200 minus DVI convention. The exact underlying percentages are not reported in the statistical-analysis entry summarized here, so the absolute difference should not be converted into a relative effect measure.
10. Analysis Populations and Censoring
| Endpoint | Analysis population / censoring rule |
|---|---|
| Time to vaginal delivery | Cesarean delivery, discharge before delivery, or withdrawal before delivery were handled through the registry-specified censoring rules. |
| Cesarean delivery | Safety population; participants discharged before delivery, withdrawing early without cesarean delivery, or lost to follow-up were classified as not having the event. |
| Time to any delivery | Participants who did not deliver during the first hospitalization were censored at the time of labor and delivery discharge. |
| Time to active labor | ITT population; participants who never entered active labor during the first hospitalization were censored using the registry-specified longest time interval. |
| Secondary binary efficacy endpoints | The ITT population was used for all secondary efficacy analyses. |
These definitions illustrate why an endpoint's statistical method cannot be separated from its event definition. A vaginal-delivery endpoint treats cesarean delivery differently from an any-delivery endpoint, and those different event definitions lead to different censoring or classification rules.
11. Statistical Methods Explained
Why was a log-rank test used for time to vaginal delivery?
The endpoint records both whether vaginal delivery occurs and the time until it occurs. The log-rank test is designed to compare time-to-event distributions while incorporating information from participants who are censored before experiencing the specified event.
What does a median difference of −677 minutes mean?
The registry reports the contrast as MVI 200 minus DVI. The negative value indicates a shorter reported median time to vaginal delivery in the MVI 200 group by the magnitude of the reported difference. It does not mean that every participant experienced a delivery 677 minutes earlier.
Why is cesarean delivery analyzed differently from vaginal delivery?
Cesarean delivery is a binary event for the primary cesarean endpoint: each participant is classified according to whether a cesarean delivery occurred during the defined first-hospital-admission interval. Time to vaginal delivery is instead a time-to-event endpoint in which cesarean delivery is handled through censoring.
Why is the non-inferiority margin different from zero?
For superiority, a difference of zero represents no treatment-group difference. In non-inferiority testing, the question is whether the experimental treatment could be worse than the comparator by more than an unacceptable amount. The registry defines that amount as 10% relative to the DVI rate and specifies an upper-confidence-limit decision rule.
Why does the confidence interval matter more than the p-value for the non-inferiority decision?
The registry's non-inferiority rule explicitly uses the upper limit of the two-sided 95% confidence interval. This focuses the analysis on whether the plausible treatment difference extends beyond the prespecified unacceptable disadvantage.
Why are Fisher exact tests used for several secondary outcomes?
These secondary endpoints are binary outcomes. Fisher exact testing provides an exact approach to comparing categorical outcomes between treatment groups. The registry reports the corresponding treatment effect as a difference in proportions.
Why can the time-to-any-delivery and time-to-vaginal-delivery results differ?
They define the event differently. Time to vaginal delivery counts only vaginal delivery as the event, whereas time to any delivery counts either vaginal or cesarean delivery. Consequently, the two analyses answer different clinical questions even though both use time-to-event methodology.
12. Primary Endpoint Interpretation: Effect Size, Precision, and P-values
The primary time-to-vaginal-delivery analysis reports a median difference of −677 minutes with P < 0.001. The effect estimate describes the direction and magnitude of the reported treatment contrast; the p-value describes the evidence against the null hypothesis under the log-rank testing framework.
For the cesarean-delivery endpoint, the 95% CI extends from −5.79 to 3.59. This interval describes uncertainty around the reported treatment difference and includes zero. For the time-to-vaginal-delivery endpoint, the registry analysis entry reports the confidence level as 95% but does not provide numerical interval limits.
The trial reports a median difference for time-to-event outcomes and a risk difference for binary outcomes. These measures should not be placed on a common scale or interpreted as interchangeable versions of the same treatment effect.
13. Multiplicity and Multiple Secondary Endpoints
The registry posts two primary analyses and eight secondary statistical analyses. Many of the secondary analyses have P-values below 0.001, but a collection of nominally significant secondary endpoints should not automatically be interpreted as eight independent confirmatory discoveries.
Each additional hypothesis test creates another opportunity for a statistically unusual result under a global null framework. Formal interpretation therefore depends on the prespecified testing hierarchy, multiplicity-control strategy, and whether particular endpoints were confirmatory or supportive.
| Analysis family | Role in the registry record | Interpretive issue |
|---|---|---|
| Time to vaginal delivery | Primary | Superiority time-to-event analysis |
| Cesarean delivery | Primary | Non-inferiority/equivalence framework |
| Time to any delivery | Secondary | Superiority time-to-event analysis |
| Time to active labor | Secondary | Superiority time-to-event analysis in ITT population |
| Delivery and oxytocin binary endpoints | Secondary | Fisher exact tests and risk differences |
The primary endpoints also involve different hypothesis types: superiority for time to vaginal delivery and non-inferiority or equivalence for cesarean delivery. Those hypotheses should not be treated as though they were identical statistical questions.
14. Safety Results
The registry reports serious adverse events by treatment arm using affected participants divided by participants at risk.
| Treatment group | Serious adverse events | At risk |
|---|---|---|
| MVI 200 | 130 | 678 |
| Dinoprostone Vaginal Insert (DVI) | 107 | 680 |
The registry's posted serious-adverse-event information therefore provides the number of affected participants and the corresponding at-risk denominator for each treatment group. These data are descriptive safety information and should be interpreted separately from the efficacy hypothesis tests.
15. Power and Sample-Size Logic
The registry's primary time-to-vaginal-delivery analysis notes a planned sample size of 675 per group to provide 90% power to detect an improvement of ≥320 minutes, described as a 20% improvement from DVI, assuming a median time of 1600 minutes for DVI and a 34% dropout rate. The calculation used a 5% 2-sided test.
Planned detectable difference
The registry specifies an improvement of ≥320 minutes as the target effect for the time-to-vaginal-delivery sample-size calculation.
Power
The stated design target was 90% power under the specified assumptions.
Assumed reference median
The DVI median time was assumed to be 1600 minutes for the sample-size calculation.
Dropout assumption
The calculation incorporated a 34% dropout-rate assumption.
For the cesarean-delivery endpoint, the registry describes a sample size of 675 subjects per group with 80% power and a 5% alpha level for a two-sided approach, using a 10% non-inferiority limit relative to the DVI rate. The planning assumptions included a 30% cesarean-delivery rate in the DVI group and 26% in the MVI 200 group.
16. Why Time-to-Event Analysis Is Useful Here
A simple proportion of participants who eventually deliver cannot capture the timing of delivery. Two treatment groups could eventually have similar proportions delivering while differing substantially in how quickly delivery occurs.
Time-to-event analysis retains information about when an event happens. It also provides a principled framework for participants whose outcome is not observed within the defined hospitalization period. In EXPEDITE, that framework is particularly important because the primary endpoint distinguishes vaginal delivery from cesarean delivery.
The first treats vaginal delivery as the event and uses specified censoring rules for other outcomes. The second treats either vaginal or cesarean delivery as the event. The resulting treatment contrasts therefore answer different questions.
17. Why the Non-Inferiority Endpoint Requires Special Care
The cesarean-delivery endpoint is not a conventional superiority test in which the objective is simply to show a statistically significant difference. The registry instead describes a non-inferiority framework designed to determine whether MVI 200 is not unacceptably worse than DVI with respect to cesarean-delivery rate.
The direction of the treatment difference is important. Because the registry defines the contrast as MVI 200 − DVI, increasingly positive values represent increasingly higher cesarean-delivery rates for MVI 200 relative to DVI. The non-inferiority decision therefore focuses on the upper confidence limit.
This is why a confidence interval that includes zero is not, by itself, sufficient to conclude that a non-inferiority test failed. Conversely, a small p-value from an ordinary superiority test would not be the appropriate criterion for declaring non-inferiority. The margin and the prespecified confidence-interval rule define the relevant decision framework.
18. Statistical Interpretation of the Posted Results
| Finding | Statistical interpretation |
|---|---|
| −677-minute median difference for time to vaginal delivery | The reported median time-to-event contrast favors a shorter time with MVI 200 under the MVI 200 minus DVI convention. |
| P < 0.001 for time to vaginal delivery | The log-rank test provides strong evidence against its null hypothesis under the specified analysis. |
| −1.10 cesarean-delivery difference | The reported MVI 200 minus DVI treatment difference is negative. |
| 95% CI −5.79 to 3.59 for cesarean delivery | The treatment-difference interval includes zero and provides the uncertainty range relevant to the non-inferiority framework. |
| −543-minute median difference for time to any delivery | The reported time-to-event contrast favors shorter time to any delivery with MVI 200. |
| 9.67 risk difference for vaginal delivery within 12 hours | The reported absolute event-proportion difference is positive for MVI 200 minus DVI. |
| 29.80 risk difference for vaginal delivery within 24 hours | The reported absolute event-proportion difference is positive for MVI 200 minus DVI. |
Across the posted efficacy analyses, the direction of the treatment contrasts is consistent with faster delivery-related outcomes in the MVI 200 group. The statistical interpretation remains endpoint-specific: a shorter time-to-event estimate, a binary risk difference, and a non-inferiority comparison are not interchangeable measures.
19. Important Limitations and Interpretation Issues
- Registry-level reporting: the statistical-analysis entries provide estimates and p-values for many endpoints, but numerical confidence-interval limits are not reported for most of the posted secondary analyses.
- Primary time-to-event precision: the registry identifies a 95% two-sided confidence level for the −677-minute estimate but does not provide the numerical confidence limits in the posted analysis entry.
- Non-inferiority margin: the cesarean-delivery margin is defined relative to the DVI rate. The registry describes the decision rule but does not provide the DVI rate needed to express that relative margin as a percentage-point threshold in the analysis entry.
- Censoring: time-to-vaginal-delivery results depend on the prespecified handling of cesarean delivery, discharge, and withdrawal. These rules determine which participants contribute time at risk and when they are censored.
- Endpoint definitions: time to vaginal delivery and time to any delivery are related but distinct estimands because they use different event definitions.
- Multiple analyses: the registry contains two primary analyses and eight secondary statistical analyses. Individual nominal p-values should be interpreted in the context of the overall testing strategy.
- Analysis populations: the primary cesarean endpoint uses the safety population, whereas the registry identifies the ITT population for several secondary efficacy analyses.
- Reported safety data: serious adverse-event counts by arm are descriptive and should not be combined with efficacy endpoints into a single statistical measure.
20. Why This Trial Matters Statistically
EXPEDITE is a useful statistical teaching case because the trial places several important clinical-trial methods side by side within a single randomized comparison.
| Concept | How it appears in EXPEDITE |
|---|---|
| Randomization | Participants were randomly allocated to MVI 200 or DVI in a parallel-group phase 3 design. |
| Blinding | The registry classifies the study as quadruple-masked. |
| Time-to-event analysis | Time to vaginal delivery, any delivery, and active labor were analyzed with log-rank tests. |
| Censoring | The registry specifies different censoring rules according to the delivery endpoint. |
| Risk difference | Binary secondary outcomes were reported as MVI 200 minus DVI differences in proportions. |
| Fisher exact test | Used for several binary secondary efficacy outcomes. |
| Chi-squared test | Used for the primary cesarean-delivery analysis. |
| Non-inferiority | The cesarean-delivery endpoint uses a prespecified relative margin and confidence-interval decision rule. |
| ITT analysis | The registry identifies ITT populations for several secondary efficacy analyses. |
| Power calculation | The registry describes separate planning assumptions for time to vaginal delivery and cesarean delivery. |
21. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
22. Related Statistical Calculators
23. Sources
- ClinicalTrials.gov: NCT01127581.
- PubMed: PMID 25758619.
- PubMed: PMID 23857539.
Continue through the Clinical Biostats knowledge graph
Connect the trial's endpoints and statistical methods to deeper tutorials, calculators, and other clinical-trial analyses.
24. Record Summary
EXPEDITE provides a compact example of how different statistical frameworks can be used within the same randomized clinical trial. The primary time-to-vaginal-delivery endpoint uses a log-rank comparison and a reported median difference, while the primary cesarean-delivery endpoint uses a chi-squared comparison framed around non-inferiority. Secondary endpoints extend the analysis to time to any delivery, time to active labor, pre-delivery oxytocin, and delivery within defined time windows using log-rank or Fisher exact methods.
The most informative statistical reading therefore keeps four elements together: the endpoint definition, the analysis population and censoring rules, the effect measure, and the hypothesis framework. In EXPEDITE, those distinctions are especially important because a shorter time to vaginal delivery, a difference in cesarean-delivery rates, and a positive risk difference for delivery within 24 hours represent different estimands and require different interpretations.