Safety and Efficacy Study of Roxadustat to Treat Anemia in Patients With Chronic Kidney Disease (CKD), Not on Dialysis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Safety and Efficacy Study of Roxadustat to Treat Anemia in Patients With Chronic Kidney Disease (CKD), Not on Dialysis (NCT02174627), a randomized phase 3 study. The primary outcome is Mean Change From Baseline in Hb Averaged Over Week 28 to Week 52.

2. Clinical Question

3. Trial Design

This was a phase 3, multicenter, randomized, double-blind, placebo-controlled, parallel-group study. It compared roxadustat with placebo in participants with anemia and chronic kidney disease who were not on dialysis. The sponsor was AstraZeneca. The registry record lists enrollment of 2781 and a completed study, with a start date of 2014-06-26 and primary completion on 2018-10-04.

Random allocation is intended to make the treatment groups comparable at the time of assignment, on average, including for factors that are not measured. Blinding is intended to reduce the influence of knowledge of assigned treatment on participant experience, care, and outcome assessment. These design features support a comparative interpretation, but do not remove the need to examine missing data, analysis choices, and the outcomes actually reported.

The primary outcome analysis set is described as the ITT set: randomized participants who received investigational product, irrespective of protocol adherence or continued participation, except for participants excluded from analysis. The registry record does not describe the number or reasons for those exclusions. Arm-specific primary outcome counts are reported, but they do not by themselves establish the reason for any difference between those counts and other reported denominators.

The protocol’s original sample-size justification, planned interim analyses, and detailed allocation procedures are not reported in the registry record. The planning illustrations associated with the posted results are separate from the trial’s original design assumptions. Details of treatment dose, duration beyond the stated outcome windows, and concomitant care are also not reported in the registry record.

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: Study had a screening period (up to 6 weeks), treatment period (up to 4 years) and follow-up period (4 weeks after treatment period for those who completed treatment). 2781 participants were randomized, of which 2761 were included in the analysis and 20 were excluded. Only participants included in the analysis are presented in the participant flow.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The ITT analysis set included all participants who were randomized to IP irrespective of their protocol adherence and continued participation in the study, except for those excluded from analysis.

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: The Intention-To-Treat (ITT) analysis set included all participants who were randomized to investigational product (IP) irrespective of their protocol adherence and continued participation in the study, except for those excluded from analysis.

6. Primary Endpoint

The registry lists the primary outcome as Mean Change From Baseline in Hb Averaged Over Week 28 to Week 52, measured over Baseline (Day 1, Week 0) and Week 28 to Week 52..

The primary endpoint was mean change from baseline in hemoglobin (Hb), averaged over Week 28 to Week 52, in g/dL. It compares each participant’s baseline value with an average over the later measurement period, then summarizes the adjusted average change within each treatment group. The reported contrast is the difference between those adjusted group averages.

Baseline Hb was defined as the mean of the last 3 central laboratory Hb values collected at screening and randomization. The timeframe is stated as Baseline (Day 1, Week 0) and Week 28 to Week 52. The registry description says that the analysis used multiple imputation under a missing-at-random assumption and an ANCOVA model, with baseline Hb, baseline estimated glomerular filtration rate, cardiovascular history, geographic region, and treatment group as fixed-effect covariates.

General statistical context: an average over a specified follow-up window can represent the group’s typical measured change during that period, rather than relying on a single follow-up measurement. Adjusting for prespecified baseline characteristics can account for their association with the outcome in the analysis. The endpoint quantifies a laboratory measure; the reported group contrast alone does not establish the clinical importance of a particular change.

7. Results Posted in the Registry

• Roxadustat: mean 1.75 versus 0.40 on control; difference 1.35 (95% CI 1.26 to 1.44), standardized effect d = 1.10; z-test p = <0.001.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Percentage of Participants With Hb Response During the First 24 Weeks of Treatment

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Roxadustat: 77.0% (1056/1371) versus 8.5% (115/1357) on control; risk difference 68.5 percentage points (95% CI 65.9 to 71.2), risk ratio 9.09 (7.61–10.85); unadjusted two-proportion z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Roxadustat than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Mean Change From Baseline in Hb Averaged Over Week 28 to Week 52 in Participants With Baseline High Sensitivity C-Reactive Protein (hsCRP) Greater Than the Upper Limit of Normal (ULN)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Roxadustat: mean 1.75 versus 0.62 on control; difference 1.13 (95% CI 0.88 to 1.38), standardized effect d = 0.89; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Roxadustat than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Proportion of Total Time of Interpolated Hb Values Greater Than or Equal To 10 g/dL From Week 28 to Week 52

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Roxadustat: mean 0.82 versus 0.33 on control; difference 0.49 (95% CI 0.46 to 0.52), standardized effect d = 1.29; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Roxadustat than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes cover several kinds of information, including Hb response during the first 24 weeks, Hb change in a subgroup with baseline hsCRP above the upper limit of normal, time spent at Hb thresholds during a later period, LDL cholesterol, rescue therapy, transfusion, and vitality. These outcomes address dimensions that are related to, but not identical with, the primary average Hb-change endpoint.

Among the secondary comparisons with numerical results reported, the direction of the estimates is consistent with a greater measured Hb response or Hb-related measure in the roxadustat group than in the placebo group. This pattern is supportive of the primary endpoint’s direction, but it does not make the secondary outcomes independent confirmation of the same question: they use different definitions, periods, populations, or outcome scales. The hsCRP analysis is a subgroup result and should be interpreted as applying to that stated subgroup, not automatically to all participants.

The registry record lists 10 secondary outcomes as posted, but numerical comparative details for all of them are not reported in the registry record. A complete pattern across every secondary outcome cannot therefore be established from these results. The listed outcomes also include rescue therapy and patient-reported vitality, which cannot be inferred from a change in mean Hb alone.

Unadjusted secondary results generally carry less confirmatory weight when many outcomes are examined, because the chance of at least one apparently notable result can increase across multiple comparisons. General statistical context: multiplicity procedures can control this issue by adjusting testing or interpretation across a defined family of outcomes. Whether a multiplicity adjustment was prespecified or applied to the secondary analyses is not reported in the registry record. Secondary findings should consequently be read as additional descriptive evidence rather than as a set of independent definitive tests.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Percentage of Participants With Hb Response During the First 24 Weeks of Treatment
• Mean Change From Baseline in Hb Averaged Over Week 28 to Week 52 in Participants With Baseline High Sensitivity C-Reactive Protein (hsCRP) Greater Than the Upper Limit of Normal (ULN)
• Proportion of Total Time of Interpolated Hb Values Greater Than or Equal To 10 g/dL From Week 28 to Week 52
• Proportion of Total Time of Interpolated Hb Values Within the Interval of 10 to 12 g/dL From Week 28 to Week 52
• Mean Change in Low-Density Lipoprotein (LDL) Cholesterol From Baseline to Week 24
• Time-To-First Instance of Receiving IV Iron, RBC Transfusion or Erythropoietin Analogue as Rescue Therapy
• Time-To-First Instance of Receiving a RBC Transfusion As Rescue Therapy
• Mean Change From Baseline in Short Form 36 (SF-36) Vitality Sub-Score From Week 12 to Week 28

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous: it is a measured quantity expressed in g/dL. The reported treatment contrast is a difference in adjusted group means, not a risk ratio or a count of participants who crossed a threshold. The sponsor reports least-squares means and an ANCOVA analysis.

What does ANCOVA do?

General statistical explanation: analysis of covariance (ANCOVA) estimates a group contrast while accounting for selected covariates. In this study, the registry description lists baseline Hb, baseline estimated glomerular filtration rate, cardiovascular history, geographic region, and treatment group as fixed-effect covariates. Adjustment can improve precision when covariates are related to the outcome and can account for chance imbalances in measured baseline characteristics. It does not adjust for unmeasured factors simply by virtue of being an ANCOVA.

The resulting least-squares means are model-adjusted estimates. They need not equal the raw arithmetic means calculated directly from participant-level endpoint values. The registry reports adjusted least-squares mean change for each group and an adjusted difference between groups. The raw participant-level data and full model output are not reported in the registry record.

What does multiple imputation under MAR mean?

The registry states that the primary model used missing-at-random (MAR) based multiple imputation. General statistical explanation: multiple imputation replaces missing values with several sets of plausible values based on observed information, analyzes each completed set, and combines the resulting estimates to reflect uncertainty from both sampling and imputation. The MAR assumption means that, conditional on information included in the imputation model, the probability that a value is missing does not depend on the unseen value itself. MAR is an assumption, not a directly observable property of the data.

The registry record does not report the imputation variables, number of imputed data sets, diagnostics, or sensitivity analyses under alternative missing-data assumptions. It also does not report the amount of missingness for the primary endpoint. Those omissions limit independent assessment of how strongly the primary result depends on the missing-data approach.

What does the recalculation represent?

The recalculated comparison is a simple unadjusted calculation based on the posted group summaries. For this continuous endpoint, the recalculation uses the group estimates and their reported standard errors to produce a mean difference, an interval, and a p-value. It does not reproduce the covariate adjustment or imputation procedure described for the sponsor’s primary ANCOVA.

General statistical explanation: a standard error describes the estimated sampling variability of an estimate under a particular calculation. A confidence interval combines an estimate and its uncertainty to show a range of values compatible with the analysis at the stated confidence level. The recalculated interval and the sponsor’s interval can differ because they arise from different analysis procedures and inputs; neither should be interpreted as a range containing a known true value with certainty.

How should the p-value be read?

General statistical explanation: a p-value measures how incompatible the observed data, or more extreme data, are with a specified null model, under the assumptions of the analysis. It is not the probability that the null hypothesis is true, nor the probability that the result is due to chance. The reported p-value is below 0.001 for the primary comparison. That value does not measure the size or clinical importance of the estimated difference.

The trial also reports binary and continuous secondary outcomes. A binary outcome can be summarized as a proportion or as a relative measure; a continuous outcome can be summarized as a difference in means. The reported methods differ across those outcomes, so estimates should be interpreted using the metric and analysis identified for each result rather than treating all outcomes as interchangeable.

The registry record reports a superiority analysis for the primary comparison. The complete statistical analysis plan, including any prespecified multiplicity procedure, is not reported in the registry record. The registry’s description of the primary model is therefore informative but does not establish all protocol-level analysis decisions.

11. Confidence Intervals and Precision

The sponsor-reported adjusted primary difference is 1.35 g/dL, with a 95% confidence interval from 1.27 to 1.43. This interval describes the statistical precision of the estimated adjusted contrast under the reported model and its assumptions. Because the interval does not include zero, the result is consistent with a positive adjusted difference for roxadustat versus placebo in this endpoint.

General statistical explanation: a confidence interval gives a range of estimates compatible with the observed data and specified analysis at a stated confidence level. It communicates uncertainty more fully than a point estimate alone. It is not a guarantee that the true effect lies within the interval, and it does not express the probability that a particular treatment effect is correct.

The recalculated unadjusted interval is 1.257 to 1.443, around the recalculated mean difference of 1.35. It is not identical to the sponsor’s adjusted interval of 1.27 to 1.43. The two intervals come from different analyses, so their endpoints should not be compared as though they were generated by the same model. The sponsor’s ANCOVA result is the registry-reported primary analysis.

Precision in the estimated group contrast does not address every source of uncertainty. It does not resolve possible sensitivity to missing-data assumptions, selection into the analysis set, or multiplicity across outcomes. It also does not tell whether a difference is important to patients. The registry record does not report a threshold for clinical importance or a minimally important difference for this endpoint.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning table and curve are illustrations based on the observed effect, not reconstructions of the trial’s original enrollment plan. For the primary endpoint, the illustration lists participants per arm for 80% power and participants per arm for 90% power. At 100% of the observed effect, the figures are 13 and 18; at 75% of the observed effect, they are 23 and 31; and at 50% of the observed effect, they are 52 and 69, respectively.

The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the effect to detect is the stated fraction of the observed effect. These are planning assumptions for the illustration. The trial’s own original sample-size assumptions and justification are not reported in the registry record. For time-to-event outcomes, the planning figures refer to total events; for the other outcomes, they refer to participants per arm.

In plain language, 80% power means that a study planned under the stated assumptions would have an 80% chance of detecting an effect of the specified size at the stated significance level, if that effect were the true one. The 90% column makes the same kind of planning statement with a higher target chance of detection. Smaller effects are harder to detect and therefore require more participants under otherwise similar assumptions.

These figures are not observed power. They are prospective planning illustrations calculated from an observed estimate, and should not be used to judge whether the completed trial was adequately designed. The observed study enrollment and its reported analysis remain distinct from these hypothetical sample-size calculations.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 1.350; pooled standard deviation σ = 1.223.
2. n = 2 × (1.96 + 0.842)² × 1.223² / 1.350² = 12.9, rounded up to 13 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 1.35                      # observed difference in means
sigma <- 1.22271800635525                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(2174627)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 13 (80% power), 18 (90% power)
 75% of observed effect: n per arm = 23 (80% power), 31 (90% power)
 50% of observed effect: n per arm = 52 (80% power), 69 (90% power)
power.t.test: 13.9 
Simulated power at n = 13 per arm: 0.767 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The registry-reported primary analysis is the sponsor’s adjusted ANCOVA comparison of least-squares means. It gives a difference of 1.35 g/dL, a 95% confidence interval from 1.27 to 1.43, and p < 0.001. This is the primary analysis described for the endpoint, including the listed covariates and the missing-at-random based multiple-imputation approach.

The recalculated result is an unadjusted comparison derived from posted summary statistics. It gives the same point estimate at the reported precision, 1.35 g/dL, but a 95% interval from 1.257 to 1.443. The p-value is also reported as <0.001. Similar point estimates do not make the methods equivalent: the recalculation does not reproduce the sponsor’s covariate adjustment or multiple imputation.

The registry record does not flag a discrepancy between the posted and recalculated results. The sponsor’s result remains the appropriate reference for the stated primary analysis because it is the model-based analysis identified in the registry record. The recalculation is a secondary check on the posted summaries, not a replacement for the sponsor’s analysis or a reanalysis of participant-level data.

15. Multiplicity, Missing Data and Analysis Populations

The primary endpoint is described with a specific ANCOVA model and a missing-at-random based multiple-imputation approach. The analysis population is described as randomized participants who received investigational product, regardless of protocol adherence or continued participation, except for those excluded from analysis. The registry record does not report the number of exclusions or their reasons. It also does not provide the amount of missing primary-outcome data or the detailed imputation procedure.

Missing-data assumptions matter because the outcome values for participants who lack measurements are not observed. General statistical explanation: an analysis under MAR relies on the assumption that missingness can be accounted for by observed information used in the model. If that assumption is not plausible, estimates can be sensitive to the missing-data mechanism. Sensitivity analyses can examine how conclusions vary under alternative assumptions, but such analyses are not reported in the registry record.

The study has a primary outcome and multiple secondary outcomes. The full statistical analysis plan, the planned hierarchy of outcomes, any family of hypotheses, and any multiplicity adjustment are not reported in the registry record. It is therefore not possible to determine from the registry record whether the secondary tests were intended to be confirmatory, how their type I error was controlled, or whether the displayed p-values were adjusted across outcomes.

The primary p-value should be interpreted in the context of the stated primary analysis, not as if it automatically accounts for testing every secondary endpoint. Likewise, a small p-value on a secondary outcome does not establish that the finding has the same confirmatory status as the primary result. These limits do not negate the reported estimates; they clarify which questions the available statistical record can answer and which remain uncertain.

16. 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: end stage renal disease (14.6%), pneumonia (8.2%), azotaemia (4.4%), sepsis (3.5%), death (3.1%).

The registry reports serious adverse events for 1384 participants at risk in the roxadustat group and 1377 in the placebo group. The number recorded as affected was 795 and 749, respectively. Other adverse events were recorded for 839 and 757 participants, with the same respective at-risk denominators. Deaths were recorded for 284 participants in the roxadustat group and 245 in the placebo group.

These are descriptive counts, not estimates of causal treatment effects. The adverse-event denominators differ from the primary outcome analysis counts, and the registry record does not explain the relationship between those populations. Counts alone do not account for differences in exposure time, event recurrence, severity, or participant characteristics. They should not be converted into comparative risks without a suitable analysis and denominator.

The registry lists end stage renal disease at 14.6%, pneumonia at 8.2%, azotaemia at 4.4%, sepsis at 3.5%, death at 3.1%, and hyperkalaemia at 3.0% among the leading serious terms. The record does not specify in this listing the denominator or whether terms represent participants, events, or a particular coding convention. The counts and terms do not establish that an event was caused by a study drug, nor do they by themselves provide a complete benefit-risk assessment.

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

The primary result describes a difference between randomized groups in adjusted mean change in Hb averaged over Week 28 to Week 52. The estimated difference is 1.35 g/dL, with the sponsor-reported 95% confidence interval from 1.27 to 1.43 and p < 0.001. This is evidence of a group difference for the specified outcome, analysis population, and time window.

It does not show that every participant experienced the group average, or that individual responses were uniform. A mean summarizes the center of a distribution and can conceal variation between people. The registry record does not report the distribution of individual changes in the primary endpoint, so the group estimate cannot be used to infer the range of individual outcomes.

The result does not by itself establish the clinical importance of the estimated difference, effects on symptoms or quality of life, or the balance of benefits and harms. Although secondary outcomes include Hb response and a vitality measure, their presence in the outcome list does not supply a complete account of their results. The registry record does not report a clinical-importance threshold for the primary endpoint.

Randomization and blinding strengthen the basis for comparing groups, but interpretation still depends on the analysis population, missing-data assumptions, outcome reporting, and conduct of the study. Details of the original statistical plan, the handling of all secondary comparisons, and sensitivity analyses are not reported in the registry record. The reported finding should therefore be kept specific to the endpoint and analysis rather than generalized to every outcome or to all people with chronic kidney disease.

This statistical summary is not medical advice and does not determine an individual treatment choice. It describes what the registry reports about the comparative Hb outcome and the limits of that evidence.

18. Why This Trial Matters Statistically

This trial is statistically informative because it combines random allocation and blinding with a large enrolled study population and a prespecified primary comparison described in the registry. The primary endpoint is a continuous measure, and the analysis reports both an adjusted effect estimate and its confidence interval. These features allow readers to consider not only whether the comparison is statistically distinguishable from zero, but also the magnitude and precision of the estimated difference.

The analysis also illustrates why the statistical method matters. The sponsor’s primary result accounts for baseline Hb, estimated glomerular filtration rate, cardiovascular history, and geographic region, and uses multiple imputation under a missing-at-random assumption. A simple recalculation from group summaries is useful as a check, but it does not recreate those model choices. Similar point estimates do not erase the methodological distinction.

The record also shows why a primary result cannot stand in for the whole trial. Secondary outcomes address response, later Hb patterns, laboratory measures, rescue interventions, and participant-reported vitality. Safety counts form another evidence stream. Each requires its own analysis and context, and multiplicity and incomplete reporting affect how strongly secondary findings can be interpreted.

Finally, the registry record documents important limitations in what can be assessed independently: detailed missing-data procedures, sensitivity analyses, multiplicity planning, and the original sample-size justification are not reported. Recognizing these gaps is part of statistical interpretation, not a reason to disregard the posted primary estimate. It helps keep conclusions aligned with the design, estimand, and information actually recorded.

19. 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 provides the primary model description but not a full statistical analysis plan. The original sample-size justification, interim analyses, multiplicity procedures, and any hierarchy for secondary hypotheses are not reported in the registry record. This limits assessment of whether the reported analyses followed all planned decisions and how secondary tests were intended to be interpreted.

The registry states that the primary endpoint used multiple imputation under a missing-at-random assumption, but the extent of missing outcome data, imputation implementation, and sensitivity analyses are not reported in the registry record. The number and reasons for participants excluded from the described ITT analysis are also not reported. These details matter when considering the robustness and generalizability of the model-based result.

The numerical comparative results do not cover all listed secondary outcomes. Results for the remaining outcomes are not reported in the registry record. The safety information is presented as counts and leading serious-event terms; details needed to interpret exposure time, event recurrence, attribution, and comparative risk are not reported in the registry record.

The registry describes the study population as people with chronic kidney disease not on dialysis, but full eligibility criteria and participant characteristics are not reported in the registry record. Treatment dosing and details of adherence or discontinuation are also not reported. These limitations constrain how far the results can be extended beyond the analyzed endpoint and the trial population.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Safety and Efficacy Study of Roxadustat to Treat Anemia in Patients With Chronic Kidney Disease (CKD), Not on Dialysis — NCT02174627.
• Registry record last updated 2019-12-16; status last verified 2019-11.
• Results first posted 2019-12-16.
• 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.

23. Record Summary

NCT02174627 is a completed phase 3, multicenter, randomized, double-blind, placebo-controlled, parallel-group study sponsored by AstraZeneca. It enrolled 2781 participants with anemia and chronic kidney disease who were not on dialysis. The study began on 2014-06-26, and primary completion was 2018-10-04.

The primary outcome was mean change from baseline in Hb averaged over Week 28 to Week 52. The registry-reported adjusted difference for roxadustat versus placebo was 1.35 g/dL, with a 95% confidence interval from 1.27 to 1.43 and p < 0.001. The described analysis used ANCOVA with multiple imputation under a missing-at-random assumption. Interpretation remains specific to that endpoint and analysis; key details about missingness, multiplicity, and the original sample-size plan are not reported in the registry record.