Comparison of Sacubitril/Valsartan Versus Enalapril on Effect on NT-proBNP in Patients Stabilized From an Acute Heart Failure Episode.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Comparison of Sacubitril/Valsartan Versus Enalapril on Effect on NT-proBNP in Patients Stabilized From an Acute Heart Failure Episode. (NCT02554890), a randomized phase 4 study. The primary outcome is N-terminal Pro-brain Natriuretic Peptide (NT-proBNP) Values and Time-averaged Change From Baseline.

2. Clinical Question

3. Trial Design

The registry record describes a Phase 4, randomized, quadruple-masked, parallel-group, active-controlled study. Its official title specifies a multicenter, double-blind, double-dummy study lasting 8 weeks. The active treatments were sacubitril/valsartan (LCZ696) and enalapril, with matching placebos listed for each. Enrollment was 887. The study is recorded as completed; its start date was 2016-04-29 and its primary completion date was 2018-06-29.

The question concerns treatment initiated in hospital after stabilization following hospitalization for acute decompensated heart failure. The eligibility description for the primary outcome specifies reduced ejection fraction, at most 40%. Starting doses and titration were determined by blood pressure at randomization and at visits, respectively. The target doses recorded were 97/103 mg twice daily for sacubitril/valsartan and 10 mg twice daily for enalapril. The available description of dose adjustments is incomplete, so further details of titration are not reported in the registry record.

The primary endpoint focused on NT-proBNP and used measurements at baseline, Week 4, and Week 8. The primary analysis population was the full analysis set, defined as randomized patients except those who had not qualified for randomization and had not received study treatment but were inadvertently randomized. The primary outcome description further specifies patients with both a baseline value and a value at Week 4 or Week 8. The number of patients contributing to the primary endpoint is not reported in the registry record beyond the arm counts displayed for the outcome.

The protocol, allocation concealment details, sample-size justification, interim-analysis plans, and a fuller statistical analysis plan are not reported in the registry record. Randomization and masking are features of the stated design; the record does not provide enough detail to evaluate how they were implemented.

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: Patients were randomized 1:1 to sacubitril/valsartan or enalapril. At the end of the 8-week treatment period, all patients had a 36-hour washout from study treatment prior to starting the open-label extension to ensure that the blinding of the core study was maintained.

Double-blind

Reasons for not completing this period:

Open-label

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full analysis set (FAS): consisted of all randomized patients with the exception for those patients who had not been qualified for randomization and had not received study treatment, but had been inadvertently randomized into the study.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Demographic and baseline characteristics are in the Full analysis set (FAS) of treatment groups. FAS: Patients who were randomized except those who were not qualified for randomization and have not received study treatment, but were inadvertently randomized.

6. Primary Endpoint

The registry lists the primary outcome as N-terminal Pro-brain Natriuretic Peptide (NT-proBNP) Values and Time-averaged Change From Baseline, measured over Baseline, Week 4 and Week 8.

The primary outcome is NT-proBNP, a peptide released by the heart in greater quantities when the heart senses that it must work harder, as in heart failure. The registry describes it as a marker related to heart function and specifies measurements in pg/ml at baseline, Week 4, and Week 8. The primary target was the time-averaged percentage change from baseline between Week 4 and Week 8 after in-hospital treatment initiation.

Using follow-up measurements over the stated interval can represent biomarker response across that period rather than relying on a single follow-up visit. The record says the Week 4 and Week 8 plasma values were averaged for patients with a baseline value and a value at either follow-up visit. The posted arm summaries use geometric means and 95% confidence intervals. Geometric summaries are often used when measurements are positive and their distribution is asymmetric; that is a general statistical rationale, not a reason documented for this trial.

NT-proBNP is a biomarker endpoint, not a direct measure of symptoms, survival, or quality of life. A difference in the biomarker estimate therefore describes the measured endpoint and does not by itself establish the size or importance of a difference in patient experience or other clinical outcomes. The registry record does not report a threshold for a clinically important NT-proBNP change.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Several statistical ideas help distinguish the primary model result from the arm summaries and the secondary-outcome comparisons. The registry reports geometric means for the primary outcome, a model-based comparison using ANCOVA, confidence intervals, and a p-value. These quantities answer related but different questions.

What does a geometric mean summarize?

A geometric mean is a multiplicative summary, rather than the arithmetic average obtained by adding measurements and dividing by the number of observations. It is commonly useful for positive measurements that are right-skewed or naturally interpreted on a ratio scale. A general way to understand it is that values are combined on a logarithmic scale and then expressed back on the original scale. This can make the summary more representative of proportional change than an arithmetic mean when the data are strongly asymmetric. The registry records geometric means but does not explain why that summary was selected.

For this endpoint, the registry describes time-averaged percentage change from baseline and gives geometric-mean arm values with 95% confidence intervals. Those figures should not be read as participant counts, raw means, or ranges of individual outcomes. The endpoint's unit is recorded as pg/ml, but the outcome description also emphasizes change from baseline and averaging follow-up measurements. The exact transformation and derivation of the displayed arm summaries are not fully explained in the registry record.

What does the ANCOVA comparison add?

ANCOVA is a general regression method that compares groups while adjusting for specified covariates. Adjustment can account for measured baseline or design variables and can improve precision when those variables are predictive of the outcome. The fitted model estimates adjusted group means or a contrast between them. In this trial, the registry labels the reported primary estimate as exponentiated and back-transformed from least-squares means. That indicates a model-based estimate expressed back on a multiplicative scale. The covariates, model specification, and transformation details are not reported in the registry record, so the exact calculation cannot be independently reconstructed from the posted arm summaries.

An adjusted model estimate need not equal a contrast calculated directly from unadjusted group summaries. The model may use covariates and a different scale, and its confidence interval reflects the fitted model. The registered comparison is therefore the relevant formal result for the primary endpoint; a simple comparison of arm summaries would not reproduce it.

How should the confidence interval and p-value be read?

A confidence interval conveys uncertainty about the estimated group contrast under the model and its assumptions. A narrower interval means greater precision than a wider interval on the same scale. It does not show the distribution of individual responses or guarantee that the true value lies within the particular calculated interval. The 95% label describes the procedure's long-run coverage property under repeated use of the same valid method and assumptions; it is not a statement that a specific participant has a 95% chance of an outcome.

The p-value measures how incompatible the observed data are with a specified null hypothesis under the analysis model. It is not the probability that the null hypothesis is true, nor a measure of the size or clinical importance of an effect. The primary reported p-value, <.0001, is evidence against the model's null of no group difference as tested. Interpretation still depends on the analysis population, endpoint definition, model assumptions, and the broader set of outcomes considered.

Why are the secondary recalculations different in status?

For the posted binary secondary outcomes, the unadjusted risk difference is a simple comparison of observed proportions between groups. Its confidence interval reflects uncertainty in that difference under the calculation used. These are recalculated, unadjusted comparisons and are not replacements for a sponsor-reported model analysis. The registry record lists no sponsor-reported analyses for these secondary endpoints.

A confidence interval that includes zero for a risk difference is compatible with no difference under that interval method, while an interval that excludes zero is evidence of a difference under the same method. These simple comparisons do not adjust for other variables or for the fact that several outcomes were examined. The p-values are unadjusted and should not be treated as if they were prespecified, multiplicity-controlled confirmatory tests.

What do the planning illustrations mean?

Planning figures based on observed effects are illustrations of the sample size that would be needed under the stated assumptions to detect a chosen fraction of an observed effect. The stated assumptions are a two-sided significance level of 0.05, equal allocation between the arms, and a normal approximation. The calculations are not the trial's original design justification and are not observed power. An effect estimated from a completed study is uncertain; using it as a planning target does not establish that a future study will have the displayed operating characteristics. The sample-size justification for this trial is not reported in the registry record.

Planning for a smaller fraction of an observed effect generally calls for a larger sample under otherwise similar assumptions. This is a general consequence of trying to detect a smaller signal against random variation, not a guarantee about any future trial. For binary endpoints, the planning calculation also depends on the event frequency and the method's assumptions. The posted planning figures should be viewed as scenarios rather than recommendations or evidence about the power of the completed study.

Why can over-dispersion matter?

Over-dispersion is a general issue in count or binary-outcome models when observed variability exceeds what a model assumes. If it is ignored, standard errors can be too small, confidence intervals too narrow, and p-values too optimistic. The registry record does not report whether an over-dispersion assessment was relevant or conducted for these analyses. It would therefore be inappropriate to claim that over-dispersion was present or that a particular adjustment was used.

10. Analyses Reported by the Sponsor

The registry reports the sponsor's primary comparison as an ANCOVA estimate, exponentiated and back-transformed from least-squares means. The estimate is 0.714, its 95% confidence interval is 0.631 to 0.808, and the reported p-value is <.0001. This is the formal model-based result for the primary endpoint. The arm-level geometric-mean summaries are 0.747 for enalapril and 0.533 for sacubitril/valsartan (LCZ696), but those descriptive summaries do not reproduce the adjusted ANCOVA contrast.

The distinction matters because the ANCOVA estimate can incorporate model adjustment and is reported on a transformed scale. The covariates and detailed model specification are not reported in the registry record, limiting independent reconstruction from the posted summaries. The primary registry analysis should be interpreted as the sponsor's specified result; an unadjusted contrast of summary values would not be a substitute for it. The record flags no discrepancy between a recalculation and the sponsor analysis. There is no recalculated primary contrast presented as a competing estimate, so the sponsor's model result remains the relevant reported inferential analysis.

The secondary binary-outcome calculations are a different matter: they are unadjusted recalculations from posted counts, and the registry lists no sponsor-reported analyses for those endpoints. They do not alter or validate the adjusted primary result. The statistical plan and a sample-size justification are not reported in the registry record, so the result cannot be assessed against unreported prespecification details.

11. Multiplicity, Missing Data and Analysis Populations

The registry record does not report whether the primary endpoint was protected by a prespecified testing hierarchy, whether secondary hypotheses were adjusted for multiplicity, or whether any interim analyses were planned or conducted. These omissions limit evaluation of how the reported p-values relate to the full family of outcomes and decisions. Multiplicity is a general concern when several hypotheses are tested: the chance of at least one apparently unusual result can increase as more tests are considered. A documented testing strategy can control or organize that risk; no such strategy is reported in the registry record.

The p-value for the primary ANCOVA comparison is reported, but the analysis plan details that would establish prespecification are not reported in the registry record. The recalculated secondary p-values are explicitly unadjusted, and no sponsor-reported secondary analyses are listed. They should therefore be treated as exploratory evidence rather than as a set of definitive confirmatory conclusions. No conclusion about an unreported adjustment or hierarchy is warranted.

Missing-data handling is also not reported in the registry record. The primary outcome description identifies patients with both a baseline value and a value at Week 4 or Week 8, but it does not state how missing follow-up values were modeled, imputed, or otherwise addressed. The endpoint population's exact observation count is not reported beyond the posted arm counts. Without further details, it is not possible to assess whether missingness could have affected the estimate or whether sensitivity analyses were conducted.

The full analysis set includes randomized patients except those who had not qualified for randomization and had not received study treatment but had been inadvertently randomized. This definition is not identical to an unrestricted analysis of every randomized record, and it is not described as an intention-to-treat analysis in the registry entry. The number excluded under the stated rule and the reasons for exclusion are not reported in the registry record. Whether other analysis populations were used is not reported.

Protocol deviations, adherence handling, censoring rules where applicable, interim monitoring, and sensitivity analyses are not reported in the registry record. These are not grounds for assuming a flaw, but their absence prevents a complete appraisal of robustness. The numerical results can be described as posted; the impact of unreported data-handling and analysis choices cannot be quantified from the record.

12. Safety Results

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

Deaths

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: cardiac failure congestive (6.7%), hypotension (3.9%), acute kidney injury (3.6%), cardiac failure (3.4%), cardiac failure acute (2.5%).

The registry reports serious adverse event counts separately by study phase and treatment group. In the double-blind phase, serious adverse events affected 132 of 436 at risk in the enalapril group and 117 of 439 at risk in the sacubitril/valsartan group. The listed deaths were 15 and 10, respectively. These are counts for the registry's stated groups, not adjusted comparisons, and no confidence intervals or statistical tests for these figures are reported.

The record also gives open-label and pooled sacubitril/valsartan figures: serious adverse events affected 104 of 828 at risk in the open-label phase and 191 of 851 at risk in the pooled phase; listed deaths were 6 and 16, respectively. Those groups represent different phase definitions and should not be read as interchangeable randomized comparisons with the double-blind groups. The record does not provide enough detail to assess overlap between phase-specific and pooled populations.

The most frequently listed serious-event terms include congestive cardiac failure at 6.7%, hypotension at 3.9%, acute kidney injury at 3.6%, cardiac failure at 3.4%, acute cardiac failure at 2.5%, and hyperkalaemia at 2.3%. The registry excerpt does not specify the denominator or analytic details for these listed percentages. Serious adverse event counts do not by themselves establish causation or comparative risk, and the full definitions and collection procedures are not reported in the registry record.

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

The primary result is evidence of a difference between the randomized treatment groups in the registry's time-averaged NT-proBNP endpoint. The adjusted estimate is 0.714, with a 95% confidence interval from 0.631 to 0.808 and p <.0001. The estimate is on the registry's exponentiated, back-transformed least-squares-means scale. It should be interpreted as a model-based comparison of the endpoint, not as a direct statement about raw concentrations, symptoms, survival, or an individual patient's response.

The confidence interval conveys uncertainty in the estimated contrast, while the p-value describes compatibility with the model's null hypothesis. Neither is a measure of clinical importance. The registry record does not report a threshold defining a clinically meaningful change in NT-proBNP, and the endpoint is a biomarker rather than a patient-centered outcome. A statistically clear difference in this biomarker does not, by itself, establish the magnitude of benefit or harm in other outcomes.

The result applies to the analysis population and conditions described in the registry: stabilized patients following hospitalization for acute decompensated heart failure with reduced ejection fraction, treatment initiated in hospital, and follow-up over the stated 8-week study period. The analysis population excludes patients who had not qualified and had not received study treatment but were inadvertently randomized. Generalization beyond this population, treatment period, or setting requires evidence not contained in this record.

The result also does not establish that all patients experienced a similar change, that the same estimate would be obtained under another analysis, or that differences in secondary outcomes are confirmed. The missing-data methods, covariates in the primary ANCOVA, multiplicity plan, and sample-size justification are not reported in the registry record. Those gaps limit assessment of the analysis's full robustness, even though the primary result itself is clearly reported.

14. Why This Trial Matters Statistically

This trial illustrates why the endpoint definition and analysis scale matter when reading a clinical-trial result. NT-proBNP is measured in concentration units, but the primary question concerns time-averaged change from baseline and the comparison is reported after exponentiating and back-transforming least-squares means. Reading the result as if it were a raw concentration difference would confuse distinct quantities. Keeping the outcome definition, model scale, and interpretation aligned is essential to a sound statistical reading.

It also shows why descriptive arm summaries and an adjusted model contrast should not be treated as interchangeable. Geometric means summarize the observed endpoint by group, whereas ANCOVA gives a model-based comparison. The adjusted estimate, confidence interval, and p-value are the formal primary inferential result; the underlying model details are needed to reproduce it. The registry record reports the estimate but not its covariates or complete model specification, leaving a limit to independent verification.

The secondary outcomes demonstrate a further statistical issue: several outcomes can be informative without carrying the same evidential weight as the primary test. Unadjusted comparisons across multiple endpoints can generate suggestive patterns, but interpretation depends on prespecification and multiplicity control. Because these details are not reported, secondary findings should be read cautiously and in their endpoint-specific context.

Finally, the record separates efficacy-related biomarker analysis from safety counts, and it reports safety by distinct study phases. Recognizing differences in population definitions and phase structure prevents inappropriate cross-group comparisons. Together, these features make the trial useful for understanding how model-based inference, biomarker endpoints, missing methodological details, and multiplicity shape the strength and scope of conclusions from a registry result.

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 entry does not provide the primary ANCOVA covariates, full model specification, transformation details, or enough information to reproduce the reported contrast from the arm summaries. The number of patients contributing observations to the primary outcome is not reported beyond the arm counts. Missing-data methods and sensitivity analyses are also not reported, which limits assessment of whether incomplete follow-up could affect the estimate.

The full analysis set definition excludes a specified category of inadvertently randomized patients who had not qualified and had not received study treatment. The number of patients in that category is not reported in the registry record. The record does not detail other analysis populations or the treatment of protocol deviations and adherence. These omissions make it difficult to assess how the analysis population relates to all randomized patients.

The primary endpoint is a biomarker measure over the stated study period. A clinically important threshold is not reported, and the registry result does not by itself describe symptoms, function, or longer-term outcomes. The trial's 8-week duration limits what can be concluded about persistence of the biomarker pattern beyond that period. The registry record does not provide the protocol or a complete statistical analysis plan for further context.

The secondary binary comparisons are unadjusted recalculations, and no sponsor-reported analyses are listed for them. The multiplicity strategy is not reported, so those results should not be treated as a coordinated confirmatory family. Safety data are listed for double-blind, open-label, and pooled phases with different at-risk counts. The record does not supply enough detail to make those phase-specific figures a single randomized safety contrast, and the serious-event definitions and percentage denominators are not fully described in the excerpt.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Comparison of Sacubitril/Valsartan Versus Enalapril on Effect on NT-proBNP in Patients Stabilized From an Acute Heart Failure Episode. — NCT02554890.
• Registry record last updated 2021-01-05; status last verified 2020-09.
• Results first posted 2019-09-24.
• 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

The registry describes a completed Phase 4, randomized, quadruple-masked, parallel-group comparison of in-hospital sacubitril/valsartan (LCZ696) and enalapril in patients stabilized after hospitalization for acute decompensated heart failure with reduced ejection fraction. Enrollment was 887, and the official title specifies an 8-week study.

The primary endpoint was time-averaged change in NT-proBNP from baseline between Week 4 and Week 8. The reported ANCOVA estimate was 0.714, with a 95% confidence interval of 0.631 to 0.808 and p <.0001. The result supports a difference on the reported model scale, while its clinical importance cannot be determined from the registry result alone.

Secondary recalculations are unadjusted, and the record does not report multiplicity procedures, missing-data methods, detailed model covariates, or the sample-size justification. Serious adverse-event counts are presented by study phase and should not be treated as a single randomized comparison.