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
PRESERVE was a randomized, double-blind, phase 3 factorial trial evaluating two separate interventions in participants undergoing angiography: intravenous sodium bicarbonate versus intravenous sodium chloride, and oral N-acetylcysteine versus oral placebo. The primary outcomes were serious, adverse, patient-centered events within 90 days following angiography.
| Feature | PRESERVE |
|---|---|
| Phase | Phase 3 |
| Therapeutic area | Cardiovascular |
| Conditions | Acute Renal Failure; Kidney Disease; Coronary Artery Disease |
| Design | Randomized, double-blind, factorial |
| Primary purpose | Prevention |
| Enrollment | 5177 |
| Primary endpoints | 2 binary endpoints |
| Primary comparisons | Intravenous sodium bicarbonate vs intravenous sodium chloride; oral N-acetylcysteine vs oral placebo |
| Primary analysis | Wald's Chi-Square; odds ratio |
| Registry status | Completed |
| Lead sponsor | VA Office of Research and Development |
2. Clinical Question
The trial addressed two related prevention questions following angiography. First, does intravenous sodium bicarbonate differ from intravenous sodium chloride with respect to serious, adverse, patient-centered events within 90 days? Second, does oral N-acetylcysteine differ from oral placebo for the same endpoint?
Population
Participants enrolled in the phase 3 PRESERVE trial in the setting of angiography, with conditions including acute renal failure, kidney disease, and coronary artery disease.
Intravenous intervention
IV isotonic bicarbonate, compared with IV isotonic saline.
Oral intervention
N-acetylcysteine, compared with oral placebo.
Primary questions
Whether either intervention changed the odds of the registered serious, adverse, patient-centered composite endpoint within 90 days following angiography.
3. Trial Design
The factorial structure is central to the statistical interpretation. The four arms combine two levels of the intravenous intervention with two levels of the oral intervention. This allows the registry to report a sodium bicarbonate-versus-saline comparison and a N-acetylcysteine-versus-placebo comparison within the same overall trial population.
IV isotonic saline + oral placebo
- IV isotonic saline
- Oral placebo
IV isotonic saline + oral N-acetylcysteine
- IV isotonic saline
- Oral N-acetylcysteine
IV isotonic bicarbonate + oral placebo
- IV isotonic bicarbonate
- Oral placebo
IV isotonic bicarbonate + oral N-acetylcysteine
- IV isotonic bicarbonate
- Oral N-acetylcysteine
4. Why the Factorial Design Matters
A factorial trial evaluates more than one intervention simultaneously. Here, the intravenous factor has two levels—sodium bicarbonate and sodium chloride—and the oral factor has two levels—N-acetylcysteine and placebo. Their combination produces four treatment cells.
The primary registry analyses then address the two principal intervention contrasts separately: sodium bicarbonate versus saline and N-acetylcysteine versus placebo.
The factorial structure is statistically efficient because two intervention questions can be addressed in one randomized trial. It also creates an important interpretive question: whether the effect of one intervention depends on the level of the other intervention. The registry information in the ClinicalTrials.gov record does not report an interaction analysis, so the two primary comparisons are the documented statistical results.
5. Primary Endpoints
| Endpoint | Time frame | Definition | Analysis |
|---|---|---|---|
| Serious, adverse, patient-centered events comparing IV sodium bicarbonate with IV sodium chloride | Within 90 days following angiography | Death, need for acute dialysis, or persistent decline in kidney function | Wald's Chi-Square; odds ratio |
| Serious, adverse, patient-centered events comparing oral N-acetylcysteine with oral placebo | Within 90 days following angiography | Death, need for acute dialysis, or persistent decline in kidney function | Wald's Chi-Square; odds ratio |
Endpoint components
The registry defines death using medical record and/or vital status registry documentation. Need for acute dialysis is defined as initiation of any modality of renal replacement, including intermittent hemodialysis, peritoneal dialysis, continuous renal replacement therapy, or sustained low-efficiency dialysis.
Persistent decline in kidney function is defined as an increase in serum creatinine of at least 50% from the baseline value collected before angiography to the measurement taken 90 days following angiography.
6. Statistical Methodology
Wald's Chi-Square test
The registry reports Wald's Chi-Square as the formal method for both primary endpoint comparisons. This is a hypothesis-testing framework for assessing the treatment contrast represented by the fitted odds-ratio estimate.
An odds ratio of 1 represents equal odds between the compared groups. Values below 1 indicate lower estimated odds in the numerator group, while values above 1 indicate higher estimated odds.
Odds ratio
The effect measure reported for both primary endpoints is the odds ratio. Unlike a risk ratio, an odds ratio compares odds rather than probabilities directly. For relatively uncommon events, odds ratios and risk ratios can be numerically similar, but they are not the same quantity.
Two-sided 95% confidence intervals
Both primary analyses report two-sided 95% confidence intervals. The interval describes statistical uncertainty around the estimated odds ratio under the analysis framework. It does not describe the range of effects experienced by individual participants.
Superiority hypothesis
The registry identifies both primary analyses as superiority analyses. The inferential question is therefore whether the randomized intervention comparison provides evidence of a difference rather than whether two treatments satisfy a prespecified non-inferiority margin.
7. Primary Results: Sodium Bicarbonate vs Sodium Chloride
The first primary endpoint compared intravenous sodium bicarbonate with intravenous sodium chloride for serious, adverse, patient-centered events within 90 days following angiography.
Odds ratio for the primary endpoint
95% CI: 0.73–1.24 · P = 0.70
Wald's Chi-Square; superiority analysis; two-sided 95% confidence interval
An odds ratio of 0.95 means that the estimated odds of the composite primary endpoint were 0.95 times the odds in the sodium chloride comparison group under the reported analysis. Expressed descriptively, this estimate is close to 1.
The estimate does not mean that the two interventions are proven equivalent. A superiority analysis with a confidence interval that includes 1 does not establish equivalence or non-inferiority; it indicates that the data are compatible with a range of effects that includes no difference on the odds-ratio scale.
The 95% CI of 0.73–1.24 communicates the precision of the estimated odds ratio. The interval extends below and above 1, so the uncertainty around the estimate includes both a possible reduction and a possible increase in odds.
The P = 0.70 value addresses the statistical evidence against the null hypothesis represented by the analysis. It is not a measure of effect size, clinical importance, or the probability that the treatment works. The odds ratio and its confidence interval are needed to understand the magnitude and precision of the estimated treatment difference.
| Primary comparison | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Sodium bicarbonate vs saline | Odds ratio | 0.95 | 0.73–1.24 | 0.70 |
8. Primary Results: N-Acetylcysteine vs Placebo
The second primary endpoint compared oral N-acetylcysteine with oral placebo for serious, adverse, patient-centered events within 90 days following angiography.
Odds ratio for the primary endpoint
95% CI: 0.78–1.34 · P = 0.86
Wald's Chi-Square; superiority analysis; two-sided 95% confidence interval
An odds ratio of 1.02 means that the estimated odds of the primary endpoint were 1.02 times the odds in the placebo comparison group under the reported analysis. The point estimate is therefore very close to the null value of 1.
The 95% CI of 0.78–1.34 indicates uncertainty around the point estimate. It includes 1 and extends in both directions, meaning the confidence interval is compatible with a range of lower and higher odds relative to placebo.
The P = 0.86 value is evidence from the specified superiority test, not a measure of how large or clinically meaningful an effect is. A high P-value does not establish that the intervention and comparator have identical effects, just as a low P-value by itself would not describe the size of an effect.
Because this was a superiority analysis rather than a non-inferiority or equivalence analysis, the absence of statistically significant evidence of superiority should not be rephrased as proof that the two interventions are equivalent.
| Primary comparison | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| N-acetylcysteine vs placebo | Odds ratio | 1.02 | 0.78–1.34 | 0.86 |
9. Comparing the Two Primary Analyses
| Feature | Sodium bicarbonate vs saline | N-acetylcysteine vs placebo |
|---|---|---|
| Endpoint | Serious, adverse, patient-centered events | Serious, adverse, patient-centered events |
| Time frame | Within 90 days following angiography | Within 90 days following angiography |
| Analysis | Wald's Chi-Square | Wald's Chi-Square |
| Effect measure | Odds ratio | Odds ratio |
| Estimate | 0.95 | 1.02 |
| 95% CI | 0.73–1.24 | 0.78–1.34 |
| P-value | 0.70 | 0.86 |
| Hypothesis | Superiority | Superiority |
The two estimates are close to the null value of 1, but they represent separate randomized comparisons. The sodium bicarbonate analysis estimates an odds ratio below 1, whereas the N-acetylcysteine analysis estimates an odds ratio slightly above 1. Neither confidence interval excludes 1.
It is important not to interpret the numerical difference between 0.95 and 1.02 as evidence that one intervention has a different effect from the other. They are different treatment contrasts, and the registry information does not report a formal statistical comparison of those two treatment effects.
10. Safety: Serious Adverse Events by Arm
The registry reports serious adverse events by the four factorial treatment combinations. These figures are presented separately from the two primary efficacy analyses because the primary endpoint is a specific patient-centered composite, whereas the safety information summarizes serious adverse events by randomized arm.
| Randomized arm | Affected | At risk |
|---|---|---|
| IV isotonic saline & oral placebo | 941 | 1300 |
| IV isotonic saline & oral N-acetylcysteine | 1040 | 1292 |
| IV isotonic bicarbonate & oral placebo | 1083 | 1294 |
| IV isotonic bicarbonate & oral N-acetylcysteine | 941 | 1291 |
The four-arm presentation is useful for understanding the factorial structure. It preserves the individual treatment combinations rather than collapsing the trial into only two groups. The registry figures should not be substituted for the primary composite endpoint results because serious adverse events and the registered patient-centered composite are distinct outcome definitions.
11. Statistical Methods Explained
Why was a factorial design used?
A factorial design permits two intervention questions to be investigated within one randomized trial. In PRESERVE, the intravenous intervention and oral intervention each had two levels, creating four combinations. The design therefore supports separate comparisons of sodium bicarbonate versus saline and N-acetylcysteine versus placebo.
What does an odds ratio of 0.95 mean?
An odds ratio of 0.95 means the estimated odds in the sodium bicarbonate group were 0.95 times those in the saline group. Because the odds ratio is close to 1, the point estimate represents only a small relative difference on the odds scale. The confidence interval is essential because it shows that the point estimate is uncertain.
What does an odds ratio of 1.02 mean?
An odds ratio of 1.02 means the estimated odds in the N-acetylcysteine group were 1.02 times those in the placebo group. Again, the estimate is close to the null value of 1. The corresponding confidence interval extends from 0.78 to 1.34, so the uncertainty is substantially broader than the point estimate alone suggests.
Why use a Chi-squared test for a binary endpoint?
The primary endpoints are binary: participants either experienced the registered serious, adverse, patient-centered event within the 90-day window or they did not. A Chi-squared framework is therefore appropriate for testing a treatment comparison involving categorical outcome information. The registry specifically reports Wald's Chi-Square for both primary analyses.
Why is the confidence interval more informative than the P-value alone?
The P-value addresses evidence against the null hypothesis under the specified test. The confidence interval additionally shows the range of odds-ratio values compatible with the statistical uncertainty. For PRESERVE, the intervals of 0.73–1.24 and 0.78–1.34 provide information about the possible magnitude and direction of the treatment contrast that a P-value alone cannot convey.
Does a confidence interval containing 1 prove equivalence?
No. A confidence interval containing 1 means that no difference on the odds-ratio scale is among the values compatible with the analysis. It does not demonstrate equivalence or non-inferiority. Those conclusions require prespecified equivalence or non-inferiority criteria and margins.
Why does the four-arm structure matter when the reported results are two comparisons?
The four arms represent every combination of the two intravenous levels and the two oral levels. The primary results then summarize the two intervention contrasts. Keeping the four-arm structure visible is important because it shows how the randomized treatment combinations were constructed and prevents the factorial design from being mistaken for a simple two-arm trial.
12. Confidence Intervals and Statistical Precision
Sodium bicarbonate
The estimated odds ratio was 0.95, with a two-sided 95% CI of 0.73–1.24. The interval includes the null value of 1 and therefore does not establish superiority under the reported test.
N-acetylcysteine
The estimated odds ratio was 1.02, with a two-sided 95% CI of 0.78–1.34. The interval also includes 1 and therefore does not establish superiority under the reported test.
Confidence intervals should be read together with the point estimates. The point estimate is the single best estimate produced by the reported analysis, while the confidence interval communicates the uncertainty surrounding that estimate. Neither should be interpreted as an individual-level prediction.
For both primary analyses, the two-sided 95% confidence interval contains 1. The corresponding P-values are 0.70 and 0.86.
13. P-values and Superiority Testing
The registry identifies both primary analyses as superiority tests. The sodium bicarbonate comparison has a P-value of 0.70, while the N-acetylcysteine comparison has a P-value of 0.86.
| Comparison | P-value | What the P-value addresses |
|---|---|---|
| Sodium bicarbonate vs saline | 0.70 | Evidence against the null hypothesis in the reported superiority analysis |
| N-acetylcysteine vs placebo | 0.86 | Evidence against the null hypothesis in the reported superiority analysis |
A P-value is not the probability that the null hypothesis is true. It is also not the probability that the observed effect occurred by chance, and it does not quantify the clinical importance of the treatment difference. The odds ratio and confidence interval provide the complementary information needed to describe the estimated magnitude and uncertainty.
14. Multiplicity and the Factorial Structure
Two primary endpoints were registered, corresponding to two principal intervention comparisons. The registry reports formal analyses for both endpoints but does not provide additional multiplicity-adjustment details in the information summarized here.
That distinction matters because a trial with more than one confirmatory question requires an explicit statistical framework for interpreting the collection of hypothesis tests. Without additional documented information, the two reported P-values should be described as the results of their respective registered analyses rather than assigning an unreported multiplicity procedure to the trial.
15. Interaction in a Factorial Trial
Factorial designs create an additional statistical concept: interaction. An interaction would mean that the effect of one intervention depends on which level of the other intervention is present.
The reciprocal question is whether the effect of N-acetylcysteine differs according to sodium bicarbonate assignment. The registry information for this record does not report a formal interaction analysis.
This is an important distinction between a factorial design and two completely separate randomized trials. The four treatment combinations create an opportunity to examine whether the two interventions interact. A formal interaction test, however, is required before concluding that the effect of one intervention differs according to the other intervention.
16. Endpoint Definition and Clinical Meaning
The primary endpoint is intentionally broader than a single renal laboratory measurement. It combines serious, patient-centered outcomes: death, need for acute dialysis, or persistent decline in kidney function.
| Component | Registry definition |
|---|---|
| Death | Based on medical record and/or vital status registry documentation |
| Need for acute dialysis | Initiation of any modality of renal replacement, including intermittent hemodialysis, peritoneal dialysis, continuous renal replacement therapy, or sustained low-efficiency dialysis |
| Persistent decline in kidney function | Increase in serum creatinine of at least 50% from the baseline value collected pre-angiography to the measurement taken 90 days following angiography |
Composite endpoints require careful interpretation because an overall treatment comparison combines potentially different clinical events. A participant meeting any registered component contributes to the binary endpoint, but the statistical result does not by itself reveal which component drove the overall comparison.
17. Trial Timeline
Trial start
The registry lists October 7, 2013 as the study start date.
Primary completion
The registry lists September 29, 2017 as the primary completion date.
Registry status
The study is listed as completed, with 5177 participants enrolled.
18. Safety Results in the Four Randomized Arms
The reported serious adverse-event counts provide a four-cell view of the factorial treatment structure.
The affected counts should be read alongside their respective at-risk denominators. The registry reports 941/1300 for saline plus placebo, 1040/1292 for saline plus N-acetylcysteine, 1083/1294 for bicarbonate plus placebo, and 941/1291 for bicarbonate plus N-acetylcysteine.
19. What the Odds Ratio Does — and Does Not — Mean
An odds ratio of 0.95 for sodium bicarbonate versus saline means that the estimated odds of the registered primary endpoint were 0.95 times the comparator odds under the reported analysis.
An odds ratio of 1.02 for N-acetylcysteine versus placebo means that the estimated odds were 1.02 times the comparator odds.
Neither odds ratio is a percentage of participants who benefited, a probability that treatment works, or a direct estimate of relative risk. An odds ratio should not be described as though it were automatically a risk ratio.
The 95% confidence intervals of 0.73–1.24 and 0.78–1.34 show the statistical uncertainty around the two point estimates. Both intervals contain 1, the null value for an odds ratio.
The P-values of 0.70 and 0.86 address the statistical tests associated with the two superiority analyses. They do not tell the reader how large an effect is. Effect size is described by the odds ratio, while precision is described by the confidence interval.
20. Limitations and Interpretation Issues
- Composite endpoint: the primary outcome combines death, acute dialysis, and persistent decline in kidney function. The overall odds ratio does not identify which component contributed most to the result.
- Factorial interpretation: the design permits two intervention comparisons within one trial, but the registry information does not report a formal interaction analysis.
- Multiplicity: two primary analyses are reported, but the registry information summarized here does not specify an alpha-allocation or multiplicity-adjustment procedure.
- Odds ratio interpretation: an odds ratio should not be treated as a risk ratio without additional information about event probabilities.
- Superiority versus equivalence: a confidence interval containing 1 and a non-significant superiority test do not establish equivalence or non-inferiority.
- Endpoint timing: the primary endpoint is defined specifically within 90 days following angiography; the estimate should therefore be interpreted in that time window.
- Safety versus efficacy endpoint: serious adverse-event counts are reported separately from the registered serious, adverse, patient-centered composite endpoint and should not be assumed to be identical outcomes.
- Missing statistical detail: the registry information does not provide additional details on stratification, missing-data imputation, or Bayesian methods for the primary analyses.
21. Why This Trial Matters Statistically
PRESERVE is a useful teaching example because its statistical structure is different from a conventional two-arm trial. Two independent intervention questions are embedded within a four-arm factorial design, and both primary outcomes are binary composite endpoints analyzed with odds ratios and Wald's Chi-Square tests.
| Concept | How it appears in PRESERVE |
|---|---|
| Randomization | Randomized phase 3 trial with four factorial treatment combinations |
| Blinding | Double-blind design |
| Factorial design | Two intravenous levels crossed with two oral levels |
| Binary endpoint | Primary outcomes classify serious patient-centered events within 90 days |
| Composite endpoint | Death, acute dialysis, or persistent decline in kidney function |
| Chi-squared testing | Wald's Chi-Square used for both primary comparisons |
| Odds ratio | Reported effect measure for both primary endpoints |
| Confidence interval | Two-sided 95% intervals accompany both odds-ratio estimates |
| Superiority testing | Both primary analyses are identified as superiority hypotheses |
| Interaction | Factorial structure creates an interaction question between the two interventions |
22. Statistical Methods in Practice
The statistical story can be summarized as a sequence from trial architecture to estimand and inference.
This structure illustrates why statistical interpretation should not stop at the P-value. The design determines the comparisons, the endpoint determines the appropriate outcome scale, the odds ratio describes the estimated contrast, and the confidence interval provides the corresponding measure of precision.
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Statistical Calculators
25. Sources
- ClinicalTrials.gov: NCT01467466 — PRESERVE: Prevention of Serious Adverse Events Following Angiography.
- PubMed: PRESERVE publication indexed under PMID 29130810.
Continue through Clinical Biostats
Explore statistical tutorials, calculators, and additional clinical-trial analyses covering study design, endpoints, effect measures, and statistical inference.
26. Record Summary
PRESERVE provides a clear example of how a factorial randomized trial can address two intervention questions within one four-arm design. The two registered primary endpoints were binary serious, adverse, patient-centered outcomes assessed within 90 days following angiography. Both were analyzed using Wald's Chi-Square tests with odds ratios and two-sided 95% confidence intervals.
The reported sodium bicarbonate comparison produced an odds ratio of 0.95 with a 95% CI of 0.73–1.24 and P = 0.70. The N-acetylcysteine comparison produced an odds ratio of 1.02 with a 95% CI of 0.78–1.34 and P = 0.86. Both confidence intervals include the null odds ratio of 1. These results should be interpreted as the reported superiority analyses rather than as equivalence or non-inferiority demonstrations.