This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for HiLo: Pragmatic Trial of Higher vs Lower Serum Phosphate Targets in Patients Undergoing Hemodialysis (NCT04095039), a randomized na study. The primary outcome is Hierarchical Composite Mortality and All Cause Hospitalization.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT04095039 |
| Phase | NA |
| Condition | All-cause Mortality, Hospitalization |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None (open label) |
| Enrollment | 793 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Duke University |
| Results posted | Yes |
2. Clinical Question
Population
Patients with end-stage renal disease undergoing hemodialysis. The primary-outcome description says analysis was restricted to the 249 participants individually randomized when the trial was terminated.
Intervention
The Hi Arm allowed serum phosphate levels to rise to 6.5 mg/dl or above.
Comparator
The Lo Arm aimed to titrate serum phosphate levels to the standard level below 5.5 mg/dl.
Primary question
Was the less stringent phosphate target associated with a different hierarchical composite of time to all-cause mortality and all-cause hospitalization rate?
HiLo was designed to compare two approaches to serum phosphate targets in people receiving maintenance hemodialysis. Its primary hypothesis concerned the hierarchical composite of time to all-cause mortality and all-cause hospitalization. The registry description frames the comparison as testing whether allowing higher phosphate levels would reduce the burden represented by that composite. That is the intended research question, not a result established by the record.
The trial also specified questions about all-cause mortality and all-cause hospitalization considered individually, as well as serum albumin and protein catabolic rate. These outcomes address different aspects of the study question and should not be treated as interchangeable with the primary composite. The registry record does not report results for these comparisons.
The record describes a pragmatic trial embedded in dialysis care and states that randomization was changed from cluster-level to individual-level because of DSMB concerns about self-selection. The primary-outcome population was restricted to those individually randomized participants. The clinical question therefore needs to be read alongside this stated population restriction; the record does not establish that the analysis represents every person counted in registry enrollment.
3. Trial Design
HiLo is listed as a randomized, parallel-group trial with no masking. Hemodialysis is the registered intervention type, while the two arms differed in their serum phosphate targets: the Hi Arm allowed levels to rise to 6.5 mg/dl or above, and the Lo Arm aimed for the standard level below 5.5 mg/dl. Because masking was listed as none, the record describes an open-label comparison.
The brief summary describes a pragmatic, multicenter study intended to involve about 4400 patients receiving in-center maintenance hemodialysis at 120-150 units maintained by two dialysis organizations. The registry enrollment is 793. These are distinct record entries: the brief summary describes the intended scale, while enrollment gives the registered enrollment figure. The record does not explain the difference between them.
The study started on 2020-03-11, and the primary completion date is 2023-11-17. Its status is terminated. The termination date and reason are not reported in the registry record. The primary-outcome description states that, at termination, only 249 participants had been individually randomized and that the analysis was restricted to that population.
The record says the randomization scheme was adjusted from cluster-randomized to individual randomization based on DSMB concerns about self-selection. It does not give further detail about the decision process or its timing. The trial is listed with phase NA. The sponsor is Duke University. No further protocol details about implementation, allocation procedures, or follow-up conduct are 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.
Overall Study
| Stage | Lo Arm | Hi Arm |
|---|---|---|
| STARTED | 441 | 352 |
| COMPLETED | 368 | 261 |
| NOT COMPLETED | 73 | 91 |
Reasons for not completing this period:
| Reason | Lo Arm | Hi Arm |
|---|---|---|
| Withdrawal by Subject | 2 | 16 |
| Transplanted | 21 | 24 |
| Transferred to Non-study Facility | 47 | 44 |
| Withdrawal from internvention target | 0 | 2 |
| Switch to peritoneal dialysis or home hemodialysis | 2 | 4 |
| Cacliphylaxis | 0 | 1 |
| Pregnancy | 1 | 0 |
Analysis population for the primary outcome, as defined in the registry: Based on DSMB concerns we adjusted the randomization scheme from cluster-randomized to individual randomization due to concerns of self-selection. At the time that the trial was terminated, only 249 participants were individually randomized so we restricted our analysis to that population.
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.
| Characteristic | Unit | Lo Arm | Hi Arm | Total |
|---|---|---|---|---|
| Participants analysed | 441 | 352 | 793 | |
| Age, Continuous | years | 61.1 (SD 14.29) | 59.2 (SD 15.00) | 60.3 (SD 14.63) |
| Sex: Female, Male: Female | Participants | 177 | 152 | 329 |
| Sex: Female, Male: Male | Participants | 264 | 200 | 464 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 47 | 45 | 92 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 12 | 12 | 24 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 382 | 295 | 677 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 15 | 19 | 34 |
| Race (NIH/OMB): Asian | Participants | 3 | 4 | 7 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 4 | 6 | 10 |
| Race (NIH/OMB): Black or African American | Participants | 181 | 146 | 327 |
| Race (NIH/OMB): White | Participants | 180 | 129 | 309 |
| Race (NIH/OMB): More than one race | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 58 | 48 | 106 |
| Region of Enrollment: United States | Participants | 441 | 352 | 793 |
| PCR | g/kg/day | 1.0 (SD 0.29) | 1.1 (SD 0.32) | 1.1 (SD 0.3) |
6. Primary Endpoint
The registry lists the primary outcome as Hierarchical Composite Mortality and All Cause Hospitalization, measured over up to 27 (enter at enrollment end) - 45 (enter at enrollment start) months.
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | not applicable |
| Unit | total number of events |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint is described as a hierarchical composite involving time to all-cause mortality and all-cause hospitalization rate. Hospitalization is represented as total counts per person-years of follow-up, and the registered unit for the outcome is total number of events. The intended analysis population is the individually randomized cohort, further restricted to the 249 participants individually randomized at the time the trial was terminated.
A composite endpoint brings multiple clinical events into one outcome framework. A hierarchical composite can distinguish events by priority, rather than treating every event as having identical importance. In general, this can retain information about more than one type of event while giving the comparison an ordered structure. The registry description does not give the hierarchy’s operational rules, including how event pairs were ranked or how the final comparison was calculated.
Using mortality together with hospitalization reflects the trial’s stated interest in both survival and hospitalization burden. Expressing hospitalization as counts relative to person-time can account for differing observation time in a rate-based analysis, in general. Interpretation still depends on the specified analytic method and follow-up information. Those details, and a numerical primary-outcome result, are not reported in the registry record.
The listed timeframe is up to 27 months for enrollment end and up to 45 months for enrollment start. This describes a range tied to enrollment timing, not an identical follow-up duration for every participant. The record does not provide the event counts or comparative estimate needed to quantify the primary endpoint.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry record establishes the study question and design but does not provide a primary-outcome comparison that can be interpreted as an estimated treatment effect. It describes the primary outcome as a hierarchical composite of time to all-cause mortality and all-cause hospitalization rate, but the posted information does not give a numerical estimate, confidence interval, or p-value for that comparison. Without those results, the direction and precision of the primary comparison cannot be assessed.
The record explicitly says that the primary outcome has several categories or time points and was not reduced to one comparison. That matters because there is no single summarized result to characterize. The absence of a single comparison is not evidence of no difference, nor is it evidence that the approaches are equivalent. It means the registry record does not present a reduced result from which either conclusion could be drawn.
The registry lists secondary outcomes including time to all-cause mortality, hospitalization days, serum albumin, and protein catabolic rate. The brief summary also describes individual mortality and hospitalization hypotheses. No numerical results for these outcomes are reported in the registry record. Their listing indicates planned questions, not findings.
The adverse-event listing contains death counts of 94 for the Lo Arm and 52 for the Hi Arm. These counts appear in the adverse-event data and are not a reported comparative analysis of the primary endpoint. The registry record does not provide the corresponding follow-up time, outcome definitions for interpreting these counts as an endpoint comparison, or an adjusted or unadjusted effect estimate. They should therefore not be read as a standalone estimate of the effect of assignment on mortality.
Enrollment was 793, but the primary-outcome description states that analysis was restricted to the 249 participants individually randomized when the trial was terminated. The record does not provide a numerical account of the primary outcome within that group. The difference in scope between the enrollment entry and the stated primary analysis population makes it especially important not to treat the enrollment figure as the analysis denominator.
Overall, the posted record supports a description of HiLo’s intended comparison, trial structure, termination status, and restricted analysis population. It does not support a quantitative conclusion about the primary or secondary comparisons. Any interpretation of benefit, harm, or similarity would require outcome estimates and uncertainty measures that are not reported in the registry record.
9. Statistical Methods Explained
The registry identifies HiLo as a randomized, parallel-group trial with no masking. Randomization is intended to make the assigned groups comparable, on average, with respect to factors that may affect outcomes. In an open-label study, participants and care teams know the assigned target. In general, this can matter for outcomes influenced by clinical decisions or reporting, such as hospitalization, although the registry record does not quantify any such influence in HiLo.
What does the hierarchical composite represent?
A composite combines more than one outcome domain. A hierarchical composite also gives the domains an order or priority in the comparison. In general, this can ensure that outcomes designated as more important are not treated as equivalent to less important events. The HiLo record names time to all-cause mortality and all-cause hospitalization rate, but does not report the detailed hierarchy, the comparison algorithm, or the statistical model. Those operational details are not reported in the registry record.
The hospitalization component is described using total counts per person-years of follow-up. In general, a rate uses the amount of time participants are observed as part of its interpretation, whereas a raw count alone does not describe how long each person was followed. The registry lists the primary outcome unit as total number of events, so the stated endpoint description contains both a rate-based component and an event-count unit. The exact derivation of the final composite is not reported.
Why can counts need a model that allows over-dispersion?
For count outcomes, a simple model may assume that the variability is closely tied to the average count. In general, over-dispersion occurs when observed variation is greater than that assumption allows. A model that does not account for over-dispersion can give uncertainty estimates that are too narrow. Whether over-dispersion was assessed or how counts were modeled in HiLo is not reported in the registry record. No specific count model should be inferred from the endpoint label alone.
What can a confidence interval tell readers?
In general, a confidence interval describes the range of effect estimates reasonably compatible with the data and the assumptions of the analysis. A narrow interval indicates greater precision than a wide interval, while an interval that includes a null value may be compatible with a range of effects, including little or no difference. This is general statistical interpretation; the registry record does not report a confidence interval for HiLo’s primary comparison.
A p-value, in general, measures how incompatible the observed data are with a specified null model, under the analysis assumptions. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The record does not report a primary-outcome p-value or specify the test used for that outcome.
How should the planning assumptions be understood?
The record describes sample-size planning illustrations based on the observed effect, with a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation. It states that the effect to detect is a specified fraction of the observed effect. For time-to-event tables, the planning quantity is the total number of events using the Schoenfeld approach; for other outcomes, it is participants per arm. These are planning illustrations, not observed power and not evidence that the study achieved a particular precision. The trial’s own sample-size justification is not reported in the registry record.
More generally, sample-size planning depends on assumptions about the effect, variability, allocation, and analysis. If those assumptions differ from the observed setting, the planning illustration may not describe the precision of the completed study. The registry record does not provide enough primary-outcome results to evaluate precision in HiLo from the posted data.
For this trial, the registry states that the randomization plan changed from cluster-level to individual-level because of DSMB concerns about self-selection. Individual randomization assigns participants, while cluster randomization assigns groups such as sites. In general, the design choice can affect how dependence among participants is handled in analysis. The record does not report an analytic adjustment for clustering or provide further methods for the changed allocation scheme.
10. Multiplicity, Missing Data and Analysis Populations
HiLo posed a primary composite question and several secondary questions. The record describes individual hypotheses for mortality and hospitalization, along with serum albumin and protein catabolic rate, and lists hospitalization days among the secondary outcomes. When multiple outcomes or hypotheses are examined, the chance of at least one apparently notable result can increase. Statistical approaches can address this multiplicity, but the choice depends on the analysis plan. A multiplicity strategy is not reported in the registry record.
The registry describes the primary outcome as involving several categories or time points and says it was not reduced to one comparison. It does not provide a single primary estimate for interpretation. The statistical analysis plan, including the treatment of multiple outcomes and any prespecified ordering or adjustment, is not reported in the registry record.
The primary-outcome description identifies 249 individually randomized participants as the analysis population at termination. It does not describe a broader analysis population, such as whether all randomized participants were to be analyzed according to assigned group, nor how participants with incomplete follow-up were handled. Missing-data procedures, censoring rules, and sensitivity analyses are not reported in the registry record.
The record states that the randomization scheme was changed from cluster-randomized to individual randomization because of DSMB concerns about self-selection. It does not specify interim analyses, stopping boundaries, or how the change was incorporated into the statistical plan. These details are not reported in the registry record and should not be inferred from the terminated status alone.
Termination may limit the amount of outcome information available, but the record does not state the reason for termination or provide enough detail to assess its statistical consequences. The entries do not explain how the enrollment count of 793 relates to the 249-person individually randomized analysis population.
11. What the Result Does, and Does Not, Mean
The registry record describes a randomized comparison of two serum phosphate target strategies and identifies a primary question involving mortality and hospitalization. It also reports that the trial was terminated and that the primary analysis was restricted to 249 individually randomized participants. These details define the scope of the available record; they do not establish the result of the comparison.
The record does not report a primary effect estimate, confidence interval, or p-value. As a result, the posted information cannot establish whether the primary composite differed between arms, how large any difference was, or how precisely it was estimated. It also cannot establish that the strategies were equivalent. A lack of a reported comparative result is not the same as evidence of no effect.
The death counts listed in the adverse-event information are not a substitute for a mortality analysis. They are not accompanied by an endpoint comparison or the information needed to interpret them as a difference in risk or rate. The record also does not provide comparative estimates for hospitalization, serum albumin, or protein catabolic rate.
General statistical principles help explain what estimates and uncertainty measures would contribute, but they cannot fill in missing trial results. A point estimate would describe the observed contrast under a specified analysis; an interval would convey uncertainty around that estimate. Neither is reported for the primary comparison in the registry record.
Accordingly, the supported conclusion is limited: HiLo was designed to compare more permissive and more stringent phosphate targets using a hierarchical mortality and hospitalization outcome, and the primary-outcome population was restricted after termination. The posted record does not support a claim that either target improved outcomes, worsened outcomes, or produced no difference.
12. Why This Trial Matters Statistically
HiLo is statistically notable because its design and analysis population changed during the trial. The registry describes an adjustment from cluster-randomized to individual randomization due to DSMB concerns about self-selection. Changes in the unit of randomization can affect the assumptions needed for analysis. A cluster design may create dependence among participants assigned within the same group, while individual assignment has a different structure. The registry does not report the detailed analysis used after the change, so the methods cannot be reconstructed from the record.
The stated restriction to 249 individually randomized participants is also important for interpreting the scope of any result. Registry enrollment is 793, but those figures describe different entries in the record. Without the analysis results and an explanation of how the populations relate, readers should not assume that the primary comparison covers everyone counted as enrolled.
The primary endpoint combines mortality and hospitalization in a hierarchical framework and includes a rate-based hospitalization component. Such an endpoint can bring several clinically relevant event types into one analysis, but it requires clear definitions of hierarchy, follow-up, and the statistical comparison. Those details are not reported in the registry record.
Finally, the record marks the study as terminated but does not report the termination reason or a primary-outcome estimate. This makes the distinction between a research question and a reported result especially consequential. The entry documents the planned question and the limited analysis population, but does not provide the quantitative evidence needed to assess the comparison.
13. 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 most consequential limitation for interpretation is that the registry record does not report a numerical result for the primary comparison. It identifies a hierarchical endpoint and says the outcome was not reduced to one comparison, leaving no single estimate or uncertainty measure to interpret. The record also omits comparative results for the secondary outcomes.
The trial is marked terminated, and the primary-outcome description says the analysis was restricted to the 249 participants individually randomized at that time. Registry enrollment is 793, while the brief summary describes an intended study of about 4400 participants. The registry does not explain the relationship among these figures or report outcome data for the different populations. The reason and date of termination are not reported in the registry record.
The randomization strategy changed from cluster-level to individual-level assignment because of DSMB concerns about self-selection. The record does not provide further information about how this change affected implementation or analysis. Detailed methods for the hierarchical composite, count analysis, multiplicity, missing data, and censoring are also not reported in the registry record.
The adverse-event listing includes death counts by arm, but does not present these as a comparative endpoint analysis or provide the accompanying information required to interpret them as such. They cannot fill the gap left by the missing primary-outcome result.
14. Related Tutorials
15. Related Statistical Calculators
16. Sources
- ClinicalTrials.gov: HiLo: Pragmatic Trial of Higher vs Lower Serum Phosphate Targets in Patients Undergoing Hemodialysis — NCT04095039.
- Registry record last updated 2025-04-24; status last verified 2023-07.
- Results first posted 2025-01-08.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
17. Record Summary
HiLo is a randomized, parallel-group, open-label trial sponsored by Duke University. It compared a Hi Arm that allowed serum phosphate levels to reach 6.5 mg/dl or above with a Lo Arm targeting the standard level below 5.5 mg/dl among patients receiving hemodialysis. The stated primary endpoint was a hierarchical composite of time to all-cause mortality and all-cause hospitalization rate.
The trial started on 2020-03-11, has a primary completion date of 2023-11-17, and is listed as terminated. Registry enrollment is 793. The primary-outcome description says only 249 participants were individually randomized at termination and that analysis was restricted to them. The brief summary describes an intended study of about 4400 patients at 120-150 units.
The registry record does not report a primary comparative estimate or results for the stated secondary outcomes. It also does not report the termination reason, detailed analysis plan, or procedures for missing data and multiplicity. The record supports describing the design and analysis population, but not drawing a quantitative conclusion about the target comparison.