Randomized Controlled Trial of Vitamin D3 in Diabetic Kidney Disease: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Randomized Controlled Trial of Vitamin D3 in Diabetic Kidney Disease (NCT00552409), a randomized phase 2, phase 3 study. The primary outcome is Change in Urine Albumin Excretion.

2. Clinical Question

This is a question about a biomarker of kidney disease, not a direct measure of kidney failure, symptoms, or cardiovascular events. The posted outcome focuses on a transformed urine measure. It does not by itself establish whether any observed difference corresponds to a change in how participants feel or function.

The registry also describes aims concerning serum calcium, circulating vitamin D levels, and markers of kidney and cardiovascular risk. However, the primary result reported for this analysis is change in urine albumin excretion. The other aims should not be mistaken for additional results in the posted primary outcome.

3. Trial Design

The registry identifies a randomized, parallel-group trial with quadruple masking. Participants were assigned to cholecalciferol or placebo, with each study medication taken by mouth once daily. The planned study duration was one year. The record lists the study as Phase 2 and Phase 3 and as completed.

The registry enrollment is 22. The primary outcome reports 11 participants in the cholecalciferol group and 11 in the placebo group. The analysis population is described as all participants analyzed. The registry does not give further detail about how masking was implemented, such as which roles or groups were masked, so the meaning of “quadruple” cannot be expanded beyond the designation recorded.

The stated population was people with type 2 diabetes and early kidney disease. The description characterizes early kidney disease as stage 1-2 chronic kidney disease, with a urine albumin-creatinine ratio of 30-300 mg/g and an estimated glomerular filtration rate ≥ 60 mL/min. The study medications were to be added to standard treatment, including an angiotensin converting enzyme inhibitor and/or an angiotensin II receptor blocker.

The primary endpoint was assessed using urine albumin and creatinine concentrations from 24hr urine collections. The record describes collection at baseline, 3 months after randomization, and one year after randomization. The post-randomization observations were analyzed together, using all available data, and compared with baseline.

The sponsor is the University of Washington. The registry lists a start date of 2007-12 and primary completion in 2010-06. These design features establish the context for the estimates, but they do not reveal the full protocol, allocation procedure, masking procedures, or statistical analysis plan. Those details are 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: Of the 93 participants who provided informed consent, 63 were ineligible, 8 withdrew prior to randomization, and 22 were randomized to treatment.

Overall Study

Analysis population for the primary outcome, as defined in the registry: All participants were analyzed

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.

6. Primary Endpoint

The registry lists the primary outcome as Change in Urine Albumin Excretion, measured over Baseline, 3 months, and one year.

The primary endpoint is change in urine albumin excretion, expressed as a percent difference in urine albumin-creatinine ratio (ACR). Urine albumin and creatinine concentrations were measured in 24hr urine collections. The analysis used the difference in log-transformed ACR after randomization compared with baseline. Measurements at 3 months and one year were analyzed together, with all available data included. The result was transformed to present a percent difference in urine ACR.

ACR is a measure derived from urinary albumin and creatinine concentrations. In general, a biomarker endpoint can provide a measurable way to compare biological changes between randomized groups. Its interpretation depends on what was measured, the timing of measurement, and how the data were analyzed. Here, the endpoint is a laboratory measure rather than a direct clinical event or a report of symptoms or functioning.

Combining post-randomization measurements means the reported primary analysis does not present separate primary estimates for the 3-month and one-year visits. It addresses the average treatment contrast across the included post-randomization data as described in the registry, rather than a distinct estimate for each visit.

The arm-level estimates are reported as mean percent differences, with 95% confidence intervals. The record also reports a regression-based comparison labeled as a mean difference in final values. The registry does not explain in further detail how that label relates to the log-transformed, jointly analyzed measurements. That limits the ability to reconstruct the analysis from the short description alone.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it is a quantitative measure reported as a percent difference in urine ACR. A continuous endpoint can take many values rather than representing only an event count or category. The registry reports a mean for each treatment group and a confidence interval around each mean. It also records a regression-based comparison between the groups.

What does the mean describe?

A mean is a way to describe the center of a set of measurements. In this trial, the posted arm estimates are -21.0 for cholecalciferol and -5.0 for placebo. They are reported on the transformed presentation scale described in the endpoint definition. A mean does not show the full spread of participant-level measurements, and the registry does not provide the individual observations or a separate measure of their dispersion.

The arm-level confidence intervals are -50.0 to 23.0 for cholecalciferol and -40.0 to 49.0 for placebo. A confidence interval conveys uncertainty in an estimate under the analysis and its assumptions. It is not a list of the observed participant values, and it does not mean that every value in the interval was observed. Wider intervals generally indicate less precision than narrower intervals, while their interpretation also depends on the model and data used.

Why analyze log-transformed ACR?

The record says the analysis used the difference in log-transformed ACR after randomization compared with baseline and then transformed results to present a percent difference in urine ACR. As a general statistical approach, taking logarithms can be useful for measurements whose variation is more naturally expressed proportionally than additively. It can also affect the influence of unusually large measurements. Those are general reasons for a transformation; the registry does not report the investigators' specific rationale for this choice.

A transformed analysis requires care in interpretation because the model works on a transformed scale while the results are presented on a percent-difference scale. The registry provides the intended presentation but does not specify the detailed transformation back to that scale. This limits independent reconstruction of the calculation from the record.

What does combining follow-up measurements mean?

The outcome description identifies measurements at baseline, 3 months, and one year. It says post-randomization measurements at 3 months and one year were analyzed together, using all available data, and compared with baseline. In general, an analysis that uses repeated measurements can incorporate information from more than one follow-up observation. Its interpretation depends on the model, the dependence between observations from the same participant, and the assumptions about incomplete data.

The registry identifies the reported comparison method as linear regression, but does not give the model formula, covariates, correlation structure, or handling of repeated observations. Those details are not reported in the registry record. The brief description alone therefore does not establish how each measurement contributed to the fitted model or how within-participant correlation was accounted for.

How should the between-group result be read?

The posted regression result is a mean difference in final values of -17, with a 95% confidence interval from -56 to 251. The interval crosses zero and spans a broad range of values. In general, a confidence interval that includes zero for a difference is compatible with no difference, as well as with differences in either direction, subject to the analysis assumptions. The p value is not reported in the registry record, so a p-value-based interpretation cannot be made from this result.

The sponsor's comparison is not interchangeable with simply comparing the two arm means by inspection. The result is labeled as a regression estimate, and the outcome description refers to transformed measurements and combined follow-up data. The registry does not include enough model detail to determine whether the contrast is adjusted or how it maps precisely to the displayed arm estimates.

Why does uncertainty matter?

Statistical uncertainty reflects the fact that estimates based on a sample can vary across repetitions of a study. It is distinct from biological variability among participants and from uncertainty about the model itself. Confidence intervals summarize sampling uncertainty under specified assumptions; they do not capture every possible source of uncertainty, including measurement issues, missing-data assumptions, or imperfect reporting.

In general, small datasets tend to provide less precise estimates when other factors are similar, but sample size alone does not determine precision. The outcome variability, design, and analysis also matter. The registry reports enrollment of 22 and 11 participants in each primary outcome group. It does not report a sample-size justification or the assumptions used to plan the trial.

Over-dispersion is a general concern when observed variation is greater than a model expects. For a continuous outcome, an unsuitable variance model can produce confidence intervals that are too narrow or too wide. The registry does not provide enough information about residual variation, model diagnostics, or variance estimation to assess whether over-dispersion was relevant.

These methods explain what can be learned from the posted estimates and what remains unclear. The intervals communicate imprecision; they do not settle the clinical importance of the endpoint. Without more complete statistical details, the reported values should be interpreted as registry-level results rather than a fully reproducible analysis.

10. Analyses Reported by the Sponsor

The registry reports the sponsor's regression analysis as a mean difference in final values of -17, with a 95% confidence interval from -56 to 251. The analysis method is listed as linear regression. The p value is not reported in the registry record. The interval spans negative and positive differences, including zero, and is broad relative to the point estimate. That means the reported comparison is imprecise and does not isolate a clear direction or magnitude of difference.

No unadjusted recalculation is reported for this continuous, two-arm outcome. The record indicates that values were missing or not estimable for that comparison. Accordingly, there is no recalculated estimate to set alongside the sponsor's result or use to explain a numerical discrepancy. The arm-level means and their confidence intervals remain descriptive summaries, not a replacement for the regression result.

The short registry description does not include enough detail to reproduce the sponsor's model. It identifies log-transformed ACR, combined post-randomization observations, and linear regression, but does not give the full model specification or explain the relationship between the group summaries and the reported comparison. The sponsor's own analysis is therefore the available inferential result in the record, while its interpretation remains constrained by missing methodological detail.

11. Multiplicity, Missing Data and Analysis Populations

The registry identifies one primary outcome measure: change in urine albumin excretion. It also describes study aims involving serum calcium, circulating vitamin D levels, and markers of kidney disease and cardiovascular risk. The record lists no secondary outcome results in the posted results. The presence of several aims does not establish how many statistical tests were planned or whether any adjustment for multiple testing was used.

Multiplicity is a general issue that arises when several hypotheses or outcomes are tested. As the number of tests increases, the chance of at least one apparently unusual result can increase if the tests are considered together. Statistical plans can address this through a prespecified hierarchy, adjusted thresholds, or other approaches. The registry record does not report whether such a strategy was used, whether any outcomes were designated as secondary in the original protocol, or whether there were interim analyses.

The outcome description says the 3-month and one-year post-randomization measurements were analyzed together, with all available data included. This indicates that the primary analysis was not reported as separate tests for each follow-up time. However, the record does not explain the model structure or whether separate time-specific analyses were also conducted. It does not report a detailed missing-data strategy, the amount or reasons for missing data, or sensitivity analyses for missing observations.

The analysis population is described as all participants analyzed. That statement does not fully specify the population definition or establish how participants with incomplete measurements were handled. In general, the interpretation of an analysis using available observations depends on whether the reasons data are missing are related to outcomes, treatment, or participant characteristics. The registry does not provide enough information to assess those assumptions.

For these reasons, the primary outcome should be treated as the posted result, without assuming that the multiplicity, missing-data, or analysis-population questions were resolved in a particular way. Those statistical-plan details are not reported in the registry record.

12. Safety Results

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

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 1% reporting threshold.

Most frequent non-serious adverse event terms

The registry reports serious adverse events for 0 of 11 participants at risk in the cholecalciferol group and 0 of 11 participants at risk in the placebo group. It lists other adverse events for 8 of 11 participants at risk in the cholecalciferol group and 10 of 11 participants at risk in the placebo group. The record does not list top serious adverse event terms.

These counts describe what is recorded in the registry; they do not establish that no other events occurred or that the treatments have equivalent safety. With no serious events recorded in either group, the data cannot characterize differences in serious-event rates with precision. The groups are small, and an absence of recorded events is not evidence that the risk is absent.

The registry does not report deaths for either group. Death counts are therefore not reported in the registry record, rather than being interpreted as zero. The record also does not provide further detail about event definitions, severity grading, follow-up completeness, or how other adverse events were collected.

In general, adverse-event counts should be considered alongside how events were defined and monitored, how long participants were followed, and whether reporting was complete. The posted counts are a limited description of recorded events and are not a standalone assessment of safety.

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

The result estimates a difference in a urine biomarker in a randomized comparison of cholecalciferol and placebo among people with diabetes and early kidney disease. The arm-level means are -21.0 for cholecalciferol and -5.0 for placebo, with broad 95% confidence intervals. The sponsor's reported regression comparison is -17, with a 95% confidence interval from -56 to 251. This estimate is imprecise, and the interval includes zero as well as values on either side.

The endpoint is urine ACR, presented as a percent difference after an analysis of log-transformed measurements. It does not directly measure kidney failure, cardiovascular events, symptoms, or quality of life. The registry's description of the study aims includes other markers, but results for those aims are not included in the primary outcome information.

The result does not show that cholecalciferol does or does not change health outcomes. It describes the observed estimates and their uncertainty for this biomarker analysis. A broad confidence interval leaves substantial uncertainty about both the direction and size of the comparison. The absent p value and limited model details further restrict what can be inferred from the registry record.

General statistical interpretation also distinguishes a lack of a precise estimate from proof that two treatments are the same. A result whose interval includes zero may reflect a small effect, no effect, or an effect in either direction, depending on the range of values and assumptions. The posted interval is too broad to support a narrow conclusion about the magnitude of the comparison.

The trial's enrollment and group counts describe a small study, but they do not alone determine what conclusions are warranted. The record does not report a sample-size justification. No sample-size illustration should be read as observed power; planning calculations, when presented, describe hypothetical future designs under stated assumptions rather than changing what was observed in this trial.

Interpretation should remain specific to the population, intervention, comparator, endpoint, and time course described in the registry. It should not be generalized to other populations or outcomes without evidence. The registry record does not report enough detail about missing data, statistical modeling, or other planned analyses to resolve those uncertainties.

14. Why This Trial Matters Statistically

This trial illustrates a central issue in biostatistics: a point estimate is only part of a result. The arm-level estimates and the regression comparison are accompanied by broad confidence intervals, so precision is limited. Interpreting the estimate without attending to the interval would overstate what the data establish.

It also illustrates how endpoint definitions shape conclusions. The primary outcome is based on urine ACR, with a log-transformed analysis and a percent-difference presentation. That can be informative as a biomarker comparison, but it is not the same as measuring a clinical event or a participant's experience. Clear distinction between biomarker and clinical outcomes helps prevent an estimate from being interpreted more broadly than the endpoint allows.

The record provides a short description of combining measurements at 3 months and one year, but not the complete model specification. This highlights why reporting the model, repeated-measures assumptions, missing-data approach, and analysis population matters. Without those details, readers can describe the posted estimates but cannot fully reproduce or evaluate their derivation.

The comparison also demonstrates that the sponsor's regression result and the arm-level summaries serve different purposes. A group mean describes an arm; a regression contrast estimates a between-group comparison according to a model. When there is no estimable unadjusted recalculation, the summaries should not be substituted for the sponsor's analysis.

Finally, the safety counts show why event counts need context. Zero recorded serious events in each group does not establish that risk is absent, particularly when the study is small. The registry record makes it possible to describe the counts, while leaving important questions about event monitoring and completeness unanswered. This is a useful case for separating what is recorded from what can be concluded.

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 provides limited statistical detail for the primary analysis. It reports log-transformed ACR, combined post-randomization measurements, and linear regression, but not the full model specification, covariates, correlation structure, or variance method. The result cannot be independently reproduced from the registry description alone.

The endpoint combines measurements at 3 months and one year. The record does not present separate estimates by follow-up time or explain how the combined analysis represents changes over time. A combined estimate may not describe whether the result differed between visits; time-specific conclusions are not reported in the registry record.

Although the outcome description says all available data were included and the population is described as all participants analyzed, the record does not state how many measurements were missing, why they were missing, or whether sensitivity analyses were conducted. Missingness can affect interpretation when the availability of observations is related to outcomes or group assignment. The available description does not allow that possibility to be evaluated.

The registry enrollment is 22, with 11 participants reported in each primary outcome group. The resulting estimates have broad confidence intervals. A sample-size justification, target effect, or planning assumptions for the original trial are not reported in the registry record. The listed phase is Phase 2 and Phase 3, but the record does not explain the phase designation further.

The comparison result has no reported p value, and the description does not clarify the meaning of the “final values” label in relation to the transformed, jointly analyzed endpoint. No unadjusted recalculation is available because values were missing or not estimable for that comparison. These factors restrict comparison between the posted arm summaries and the sponsor's regression result.

Finally, the safety information is limited to event counts. The registry does not provide serious-event terms, death counts, or further details on ascertainment and follow-up. These limitations do not invalidate the recorded results, but they narrow the conclusions that can be drawn from them.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Randomized Controlled Trial of Vitamin D3 in Diabetic Kidney Disease — NCT00552409.
• Registry record last updated 2014-04-28; status last verified 2014-04.
• Results first posted 2013-09-27.
• 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 University of Washington sponsored a completed randomized, parallel-group trial of cholecalciferol versus placebo in people with diabetes and early kidney disease. The registry lists quadruple masking, Phase 2 and Phase 3, and enrollment of 22. The study description specifies daily oral study medication for one year, added to standard treatment.

The primary outcome was change in urine albumin excretion, presented as a percent difference in urine ACR. The analysis used log-transformed measurements, combining post-randomization observations at 3 months and one year with baseline as the reference. The registry reports arm estimates of -21.0 for cholecalciferol and -5.0 for placebo, with 95% confidence intervals, and a regression comparison of -17 with a 95% confidence interval from -56 to 251.

The intervals indicate substantial uncertainty. The p value, detailed model specification, sample-size justification, and full missing-data strategy are not reported in the registry record. Serious adverse events are recorded as 0 in each group; deaths are not reported in the registry record.