A Study to Investigate the Safety and Efficacy of ZS in Patients With Hyperkalemia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Investigate the Safety and Efficacy of ZS in Patients With Hyperkalemia (NCT02875834), a randomized phase 3 study. The primary outcome is Least Square Mean S-K Level on Days 8-29.

2. Clinical Question

The central statistical question concerns maintenance of normokalemia after participants had first received the initial treatment and achieved normokalemia. It is therefore a question about S-K levels during the subsequent randomized phase, not a direct comparison of initial treatment against placebo. The registry describes sequential comparisons of each active dose with placebo, beginning with the highest dose and then moving to the lowest. The registry record does not report further detail about the testing procedure used for that sequence.

This design gives the randomized comparisons a common placebo group and focuses the primary outcome on repeated S-K measurements over the specified treatment period. The results apply to the enrolled study population and its stated treatment sequence. They do not, on their own, address treatment outside that study context or establish how results might differ for patients who did not reach normokalemia during the initial phase.

3. Trial Design

This completed Phase 3 study was randomized, double-blind, placebo-controlled, and used a parallel-group design. It enrolled 267 participants with hyperkalemia. The registry lists AstraZeneca as sponsor. The study began on 2017-03-03, and primary completion was 2018-02-14.

All participants first entered a 48-hour open-label initial phase with sodium zirconium cyclosilicate 10g three times daily. The randomized treatment phase then assigned eligible participants to sodium zirconium cyclosilicate 5g once daily, sodium zirconium cyclosilicate 10g once daily, or placebo once daily for 28 days. The brief study description specifies that this maintenance comparison was in initially hyperkalemic patients who achieved normokalemia after the initial two days of treatment.

The primary outcome is reported for the Full Analysis Set. The primary endpoint timeframe is the randomized treatment phase on days 8-29. In the posted arm results, the 5g and 10g groups each have 99 participants, and the placebo group has 50. A separate 10g three-times-daily entry has no evaluable participants for the reported primary outcome. It represents the initial open-label treatment rather than a randomized maintenance arm in the reported comparison.

The registry lists 13 outcome measures and 8 analyses, and 12 secondary outcomes are posted. It does not report further details about the allocation procedure, the method used to maintain masking, participant flow through each study phase, or the statistical plan beyond the outcome descriptions and analyses recorded. Those details should not be inferred from the parallel-group, randomized, and double-blind labels alone.

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: the study consisted of two phases: an open-label phase (single arm, 48 hours, ZS 10 g TID; OLP) and a randomized treatment phase (3 arms, 28 days; RTP). 267 patients entered OLP. Out of these, 248 endered RTP and were randomized to placebo, ZS 5 g QD or ZS 10 g QD in a 1:2:2 ratio.

Open-label Phase

Reasons for not completing this period:

Randomized Treatment Phase

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full Analysis Set

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: In RTP there were no subjects analyzed in Sodium Zirconium Cyclosilicate (ZS) 10g tid arm. In OLP, only ZS 10 g TID arm was present. All subjects in RTP took part in OLP.

6. Primary Endpoint

The registry lists the primary outcome as Least Square Mean S-K Level on Days 8-29, measured over Through 28-day randomized treatment study phase day 8-29..

The primary endpoint is the least-squares mean S-K level, measured in mmol/L, across randomized treatment days 8-29. S-K is the outcome abbreviation used in the registry. The endpoint summarizes the average level over a specified period rather than a single measurement at one visit. The registry states that the analysis compares placebo with each active dose, with the dose comparisons performed sequentially from the highest dose to the lowest.

Using an average across a period can describe the overall level during treatment and is suited to a question about maintaining normokalemia after the initial phase. It can incorporate repeated measurements in a way that a single endpoint measurement would not. The registry names a mixed models analysis for the reported primary comparisons, but it does not report the model specification, the covariance structure, or how missing measurements were handled.

The posted least-squares mean levels are 4.811 mmol/L for the 5g group, 4.381 mmol/L for the 10g group, and 5.321 mmol/L for placebo. Their reported 95% confidence intervals are 4.685 to 4.94, 4.265 to 4.501, and 5.155 to 5.493, respectively. These arm-level summaries describe the endpoint estimates; the model-based comparisons are a separate part of the registry results. A confidence interval conveys uncertainty around an estimate under the model and analysis used. It does not by itself describe the range of individual participant measurements.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a continuous measure: S-K level, reported in mmol/L. Its arm summaries are least-squares means, which are model-based means adjusted according to the fitted analysis. The registry reports 95% confidence intervals for these means. In general, a confidence interval expresses the uncertainty of an estimate under the assumptions of the chosen model and sampling process. It does not show the full spread of participant-level values, and it does not guarantee that the interval contains the true value.

What does the mixed models analysis contribute?

The sponsor’s reported primary comparisons use a mixed models analysis. In general, mixed models can analyze repeated or grouped observations while accounting for dependence among observations from the same participant. The primary endpoint covers days 8-29, but the registry record does not report whether or how repeated measurements were modeled, the fixed or random effects, covariance structure, covariates, or handling of incomplete follow-up. It is therefore not possible to reconstruct the exact model from the registry description alone.

The primary analysis reports mean differences, 95% confidence intervals, and p-values for comparisons with placebo. A mean difference expresses a contrast between groups on the outcome scale, but its sign depends on the order used for subtraction. The registry describes the groups being compared without clearly documenting the direction convention for the values. The arm means are also reported, so readers can inspect their ordering without treating the contrast’s sign as self-explanatory.

Why are recalculated comparisons separate?

A recalculated comparison, when present, is a simple unadjusted calculation based on posted summary statistics. It is not a reproduction of a fitted model, and it cannot account for details that are not available in summary values. No recalculated primary comparison is estimable from the posted primary summaries. Consequently, the primary comparison should be read from the sponsor’s reported mixed models analysis, alongside the reported arm estimates. The registry does not flag a discrepancy between the sponsor’s analysis and a recalculated primary result.

How should p-values be read?

A p-value describes how incompatible the observed data are with a specified null hypothesis under the statistical model and its assumptions. It is not the probability that the null hypothesis is true, nor a measure of effect size, clinical importance, or the probability that a result will replicate. The registry reports p < 0.001 for each primary comparison. The stated sequential comparison order is highest dose first and lowest dose second, but the exact multiplicity procedure and any adjustment to the significance threshold are not reported.

What do the secondary analyses add?

Secondary outcomes include binary counts or proportions, time-to-event outcomes, and continuous summaries. For binary outcomes, an unadjusted risk difference is a comparison of the observed proportions between two groups. A confidence interval around a risk difference conveys uncertainty in that contrast. Odds ratios compare odds rather than probabilities, and hazard ratios compare event rates over time under the assumptions of a time-to-event model. These measures are not interchangeable, and their numerical scales differ.

Some posted secondary comparisons are simple recalculations from the reported group summaries; they are explicitly unadjusted. The sponsor’s own reported analyses use regression methods, including logistic, linear, and Cox regression for different outcomes. In general, regression estimates can reflect modeling choices and any included covariates, while a simple unadjusted calculation uses only the summarized group data. The registry does not report enough detail to determine the full specification of those regressions.

For continuous secondary outcomes, the recalculated mean difference and confidence interval are based on posted group summaries. In general, over-dispersion means that variability is greater than a model assumes; if a model underestimates that variability, uncertainty intervals can be too narrow and tests can appear more decisive than warranted. Whether over-dispersion is relevant depends on the outcome and model. The registry does not report diagnostic checks that would establish its presence or absence for these analyses.

The sample-size figures associated with recalculated secondary comparisons are planning illustrations based on the observed effect and stated assumptions. They are not observed power and are not the trial’s original sample-size justification. The assumptions specify a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; the effect to detect is a stated fraction of the observed effect. Time-to-event illustrations give total events, while the others give participants per arm. The registry record does not report the original sample-size rationale, recruitment target justification, or any interim-analysis plan.

10. Analyses Reported by the Sponsor

The primary comparisons reported by the sponsor use a mixed models analysis. The reported mean difference for 5g versus placebo is 0.904, with a 95% confidence interval of 0.876 to 0.933 and p < 0.001. For 10g versus placebo, the reported mean difference is 0.823, with a 95% confidence interval of 0.797 to 0.850 and p < 0.001. The registry does not clearly state the subtraction order that determines the sign of these estimates. The arm-level least-squares means are reported separately, with the 5g and 10g estimates below the placebo estimate.

No recalculated primary comparison is estimable from the posted summary statistics. The sponsor’s model-based results therefore stand as the reported inferential analyses; a simple summary-statistic calculation would not replicate the mixed model. The registry record does not report enough model detail to reproduce or independently audit the sponsor’s estimates, including the handling of repeated measurements and missing values.

For secondary outcomes, the sponsor also reports regression-based estimates. Those analyses are not necessarily expected to match unadjusted recalculations from group counts or summaries, because the methods and possibly their assumptions differ. The registry does not fully describe their model specifications. The posted record does not flag a discrepancy between sponsor and recalculated results. Taken together, the reported model-based comparisons and arm summaries should be considered in light of the undefined direction convention and the limits of the available model documentation.

11. Multiplicity, Missing Data and Analysis Populations

The registry describes sequential primary comparisons, starting with the highest dose and ending with the lowest. This is relevant because testing more than one dose against a shared placebo group creates a multiplicity question: the interpretation of p-values depends partly on whether and how the testing sequence controls the chance of false-positive findings across comparisons. The registry record does not report the specific procedure, significance thresholds at each step, or whether the stated p-values are adjusted. The sequence itself is documented, but its statistical consequences cannot be fully assessed from that description.

There are also multiple secondary measures and analyses. The record lists 13 outcome measures, 8 analyses, and 12 posted secondary outcomes, but it does not report a complete strategy for multiplicity across the secondary family. Unadjusted p-values and confidence intervals do not account for the number of analyses. This does not make the results uninformative, but it reduces the strength of claims based on isolated secondary findings, particularly when several outcomes are examined.

Missing-data handling is not reported in the registry record. The primary endpoint spans days 8-29, but the registry does not explain how participants with incomplete measurement sequences were included, whether imputation was used, or what assumptions supported the analysis. It also does not report sensitivity analyses for departures from those assumptions. Without those details, the potential influence of missing observations on the estimates cannot be evaluated from the posted description.

The primary outcome population is identified as the Full Analysis Set. The record does not report the exact membership rules for that set or a participant-flow account that would clarify how participants moved from the open-label phase into randomized treatment and analysis. The initial-phase group and randomized arms should not be treated as interchangeable populations. The registry also does not report an interim-analysis plan, original sample-size justification, or a complete protocol-level statistical analysis plan. These omissions limit assessment of prespecification and robustness, without establishing that any particular analysis choice was inappropriate.

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: ankle fracture (2.0%), cystitis (1.0%), infectious colitis (1.0%), pneumonia (1.0%), upper respiratory tract infection (1.0%).

The registry lists serious adverse events for the randomized groups and the initial-treatment entry. There were 4 serious adverse events among 99 participants in the 5g group, 3 among 99 in the 10g group, and 1 among 50 in the placebo group. The initial 10g three-times-daily entry lists 0 serious adverse events among 267 at risk. No deaths are listed in any of these groups.

For other adverse events, the registry lists 6 among 99 participants in the 5g group, 21 among 99 in the 10g group, and 0 among 50 in the placebo group; the initial-treatment entry lists 0 among 267 at risk. The registry does not explain in this outcome listing how the “other” category is defined or whether the counts represent unique participants, event episodes, or a particular coding convention. The figures should be read as the registry’s reported counts, not converted into rates by inference.

Listed leading serious-event terms include ankle fracture at 2.0%, and cystitis, infectious colitis, pneumonia, upper respiratory tract infection, and cardiac failure at 1.0% each. The record does not assign these listed terms to a specific randomized arm in the information summarized. These event counts are descriptive and do not establish that treatment caused or prevented any event. Interpretation is limited by the group sizes, the study duration, and the absence of further detail about event definitions and ascertainment in the registry record.

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

The primary results describe mean S-K levels during randomized treatment days 8-29 after participants had received an initial open-label phase and achieved normokalemia. The arm-level least-squares mean estimates are 4.811 mmol/L for 5g, 4.381 mmol/L for 10g, and 5.321 mmol/L for placebo. The sponsor reports statistically significant mixed models comparisons for both active doses, with p < 0.001. These results characterize the analyzed trial population and treatment schedule.

They do not describe the effect of the initial 48-hour therapy versus placebo, because the posted design describes that initial phase as open-label treatment. Nor do they establish the outcomes for patients who did not reach normokalemia during that phase. The study’s maintenance comparison is conditional on the described initial response, so generalization beyond that sequence requires caution.

A statistically significant result is not by itself evidence of clinical importance. The registry does not provide a threshold for a clinically meaningful S-K difference or a decision rule for interpreting one. It also does not report the full model specification, missing-data methods, or a comprehensive multiplicity plan. Those omissions constrain how confidently the model estimates can be independently evaluated.

The reported group averages do not imply that all participants had the same S-K level, remained normokalemic for the same duration, or had the same safety experience. Secondary outcomes address related questions, but they involve distinct definitions and statistical methods. The posted data should therefore be understood as a set of trial estimates rather than an individual prediction. They do not constitute medical advice and do not resolve questions outside the population, treatment sequence, outcomes, and follow-up described in the registry record.

14. Why This Trial Matters Statistically

This trial illustrates why the estimand and treatment sequence matter when interpreting randomized evidence. The randomized comparison concerns maintenance treatment after an open-label initial phase, not the full treatment pathway against placebo from the start. The distinction shapes which participants and clinical question the primary estimate represents. A randomized label alone does not make every phase of a study randomized or every comparison applicable to the same population.

The primary endpoint also shows the difference between an arm summary and a model-based contrast. The least-squares means describe estimated group levels, while the mixed models analysis reports contrasts and uncertainty intervals. Both are useful, but their interpretation depends on knowing the model and the direction convention for contrasts. The registry provides some of the numerical results but not enough detail to reproduce the model or fully resolve the sign convention.

The secondary outcomes provide a broader view of maintenance, changes in S-K, and time spent normokalemic. Their variety also creates a multiplicity challenge. Unadjusted analyses can provide informative exploratory evidence, but they do not have the same confirmatory strength as a prespecified primary comparison with a fully described error-control strategy. The registry does not document that strategy for the full set of secondary findings.

Finally, a complete statistical reading requires more than p-values. Confidence intervals show precision under stated analytical assumptions; design details and missing-data methods determine how credible those assumptions are. This record offers a useful account of the randomized comparison and its reported estimates, while leaving important planning and analysis details unreported. Recognizing both the evidence and its limits is central to sound interpretation.

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 details about the primary model unreported, including its specification, the handling of repeated S-K measurements, covariates, and missing observations. The lack of a clearly stated direction convention for the reported primary mean differences also complicates interpretation of those estimates alongside the arm-level means.

The population is defined by the treatment sequence: participants were initially hyperkalemic and achieved normokalemia after two days of open-label therapy. The record does not report detailed participant flow or the rules determining entry into the Full Analysis Set. This limits assessment of who contributed to each randomized comparison and how broadly the estimates apply beyond those who reached the stated initial response.

The registry describes sequential primary comparisons but does not report the exact multiplicity-control procedure. It also does not document a comprehensive plan for secondary-outcome multiplicity. The posted secondary findings therefore have less secure confirmatory interpretation, particularly where analyses are unadjusted or differ in method from the sponsor’s regression analyses.

Safety reporting is limited to the listed counts and event terms. The registry information summarized does not clarify whether “other” adverse-event counts represent participants or event episodes, does not map the leading serious-event terms to particular arms, and does not provide further details about event ascertainment. The record also does not report the original sample-size rationale, interim-analysis plan, or full statistical analysis plan. These limitations do not negate the reported estimates, but they restrict independent evaluation of analysis choices, robustness, and the completeness of the trial account.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study to Investigate the Safety and Efficacy of ZS in Patients With Hyperkalemia — NCT02875834.
• Registry record last updated 2020-08-19; status last verified 2020-08.
• Results first posted 2020-08-19.
• 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

NCT02875834 is a completed Phase 3 randomized, double-blind, placebo-controlled, parallel-group study sponsored by AstraZeneca. It enrolled 267 participants with hyperkalemia. After a 48-hour open-label initial phase using sodium zirconium cyclosilicate 10g three times daily, participants received 5g once daily, 10g once daily, or placebo for 28 days.

The primary endpoint was least-squares mean S-K level in mmol/L during randomized treatment days 8-29, reported for the Full Analysis Set. The registry reports mixed models comparisons for both active doses against placebo, with p < 0.001 for each. Secondary outcomes cover normokalemia, S-K changes, and hyperkalemia. The record describes a sequential primary comparison order but does not report full model details, missing-data methods, or a comprehensive multiplicity plan.