Application of Bioimpedance Spectroscopy in Taiwan Dialysis Patients: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Application of Bioimpedance Spectroscopy in Taiwan Dialysis Patients (NCT02325856), a randomized na study. The primary outcome is All-cause Hospitalization.

2. Clinical Question

The central statistical question is how compatible the posted hospitalization data are with different relative event rates between the groups. The point estimate is a comparison of rates, not a count of participants who were hospitalized. A patient-year rate expresses events relative to accumulated exposure time. The accompanying confidence interval describes uncertainty around the rate ratio, so the point estimate should not be read in isolation.

The registry also lists complications during dry-weight adjustment as a secondary outcome. The available record does not provide a posted result for that outcome in the registry record. The primary result therefore answers a narrow question about the recorded all-cause hospitalization rate, rather than summarizing every outcome or aspect of the intervention.

3. Trial Design

The registry describes a randomized, parallel-group trial with no masking. It lists 298 participants enrolled and the status as completed. The sponsor is An Hsin QingShui Clinic. The registry identifies fluid overload as the condition and records the phase as NA. The listed start and primary-completion dates are 2013-10 and 2014-10.

Participants were assigned to a Study Group or a Control Group. In the Study Group, dry weight was determined by performing bioimpedance spectroscopy. In the Control Group, dry weight was determined by clinical symptoms. The study summary says that an algorithm was used to guide dry-weight determination with a Body Composition Monitor based on bioimpedance spectroscopy. The comparison is therefore between two approaches to dry-weight determination, rather than between two named medicines.

Randomization can help make groups comparable at the start of a trial, but the registry record does not report how the sequence was generated, whether assignment was concealed, or whether any stratification was used. It also does not report the protocol details needed to evaluate adherence to the assigned approach. Masking is listed as none; the record does not explain whether outcome assessment or data handling included any separate masking procedures.

The primary outcome was all-cause hospitalization, expressed as hospitalizations per patient-year over 1 year. The posted outcome groups include 128 participants in the Study Group and 123 in the Control Group. These are the group counts associated with the result, while the overall enrollment is listed as 298. The registry record does not explain why these counts differ or specify the analysis population used to generate the outcome results.

Complications during dry-weight adjustment is the listed secondary outcome. The record indicates one secondary outcome posted, but details of its result are not included in the registry record. A full protocol, planned analysis population, and detailed statistical analysis plan 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.

Overall Study

5. Baseline Characteristics

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

6. Primary Endpoint

The registry lists the primary outcome as All-cause Hospitalization, measured over 1 year.

All-cause hospitalization counts hospitalizations from any cause rather than limiting the endpoint to a particular diagnosis. The result is reported as a rate per patient-year. This form of endpoint accounts for the amount of participant follow-up contributing to the rate, which is useful when event counts are interpreted relative to exposure time. It is not the same as reporting the proportion of participants who had at least one hospitalization.

The registry gives a 1-year timeframe and reports a rate of 0.52 hospitalizations per patient-year for the Study Group and 0.54 for the Control Group. The comparison is a rate ratio, with the Study Group as the comparison group against the Control Group. A rate ratio summarizes relative rates: a value of 1 would represent equal rates, while values on either side represent a lower or higher rate in the Study Group relative to the Control Group.

All-cause hospitalization is a broad measure of healthcare use and clinical events. Its breadth avoids restricting the count to a selected cause, but it does not explain why hospitalizations occurred. The registry record does not give a breakdown by cause, the definition of a qualifying hospitalization, or further details about exposure-time construction. Those details are important when judging how completely the endpoint captures the intended clinical question.

The endpoint is one component of the trial's assessment. It does not, by itself, summarize complications during dry-weight adjustment or other clinical outcomes. The posted result should therefore be read as evidence about the specified hospitalization rate over the stated timeframe, not as a complete evaluation of every consequence of the two dry-weight approaches.

7. Results Posted in the Registry

• Study Group: 0.52 versus 0.54 (hospitalizations per patient-year) on control; rate ratio 0.96 (95% CI 0.77–1.21, derived from the posted confidence limits); p ≈ 0.744.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is an event rate: hospitalizations per patient-year. An event rate relates the number of recorded events to the time participants contribute to observation. Unlike a proportion, it can include repeated events for a participant and uses exposure time in its denominator. The registry reports group rates of 0.52 and 0.54, with 95% confidence intervals, and a rate ratio comparing the Study Group with the Control Group.

What does a rate ratio express?

A rate ratio is the rate in one group divided by the rate in the comparison group. A ratio of 1 represents equal rates. A ratio below 1 represents a lower estimated rate in the group named in the comparison, and a ratio above 1 represents a higher estimated rate. Here, the posted rate ratio is 0.963. This is a relative measure and should not be mistaken for an absolute difference in rates or a probability that an individual will be hospitalized.

The ratio is accompanied by a 95% confidence interval from 0.768 to 1.208. The interval provides a range of values that are reasonably compatible with the data under the statistical procedure used. Its inclusion of 1 means that equal rates remain compatible with the interval, while values below and above 1 are also compatible. The interval does not assign a probability to the true value being within its particular endpoints.

Why is a Poisson approach relevant?

Poisson methods are commonly used for event counts and rates, particularly when event counts are related to exposure time. The comparison is marked as Poisson in the posted information. In a basic Poisson framework, the count is modeled relative to exposure, and uncertainty in the estimated rate or rate ratio is reflected in confidence intervals and a test statistic. The exact implementation, model specification, and any adjustments used for this trial are not reported in the registry record.

Event data can be more variable than a simple Poisson model anticipates. This situation is often called over-dispersion: the observed variation in counts exceeds the model's expected variation. If over-dispersion is present and not adequately handled, standard errors and confidence intervals may be too narrow. General statistical practice may address this through alternative variance estimates or other models, but the registry record does not report whether over-dispersion was assessed or how it was addressed. It would be inappropriate to assume a particular adjustment.

How should the p-value be read?

The posted unadjusted p-value is 0.744. It summarizes how unusual results at least as discrepant as the observed comparison would be under a specified null model and its assumptions. It is not the probability that the null is true, and it does not measure the probability that the observed result is due to chance. It also does not establish whether a difference is clinically important. Interpretation should consider the estimate, interval, endpoint definition, and trial context together.

The term unadjusted indicates that this p-value is presented without a reported multiplicity adjustment. The record does not describe the overall testing strategy or whether the primary outcome was protected from other comparisons through a prespecified hierarchy. The listed secondary outcome is complications during dry-weight adjustment, but a complete statistical plan is not reported.

What can and cannot be inferred from the group rates?

The posted rates and confidence intervals describe the results for the analyzed groups as recorded. They do not reveal the number of unique participants hospitalized, the distribution of repeated hospitalizations, or the length of follow-up contributed by each participant. Nor do the summaries show the cause or severity of events. The record does not state how censoring, missing observations, or incomplete follow-up were managed.

Randomization is listed, but additional design details that can affect interpretation are absent. The registry record does not report the sequence-generation method, allocation concealment, detailed eligibility criteria, or a full analysis plan. It also does not state whether the comparison was adjusted for any baseline factors. The posted comparison should therefore be described as the reported rate-ratio result, not as an independently verified model reconstruction.

For planning illustrations, the record describes calculations using a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation. The effect to detect is expressed as a stated fraction of the observed effect. These are planning assumptions for the displayed illustrations, not reported assumptions from the trial's original design. They do not alter the interpretation of the observed result.

10. Confidence Intervals and Precision

Precision is reflected most directly in the confidence intervals. The Study Group rate is 0.52 hospitalizations per patient-year, with a 95% confidence interval from 0.44 to 0.61. The Control Group rate is 0.54, with a 95% confidence interval from 0.46 to 0.63. Each interval conveys uncertainty around its own group's estimated rate; neither interval alone answers the comparative question.

For the comparison, the rate ratio is 0.963, and its 95% confidence interval runs from 0.768 to 1.208. This interval includes 1, the value corresponding to equal rates. It also includes values below 1 and above 1, so the observed data do not pin down the direction of the relative rate with precision. The width of the interval is important: the point estimate alone gives a more definite impression than the range of compatible values warrants.

A confidence interval is not a guarantee that the true rate ratio lies between its endpoints, nor is it a range containing 95% of individual participant outcomes. In general, a 95% confidence interval is produced by a procedure that, under its assumptions, would cover the true parameter in 95% of repeated studies. This is a general explanation; the interval is the one posted in the registry record.

The unadjusted p-value of 0.744 is consistent with the fact that the confidence interval includes the equal-rate value. Neither the p-value nor the interval establishes equivalence. A claim of equivalence would require an appropriate prespecified margin and analysis, which are not reported in the registry record. The posted uncertainty should therefore be communicated directly rather than converted into a categorical claim that the rates are the same.

11. 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 sample-size table and its accompanying curve are planning illustrations based on the observed effect, not a report of the trial's original design calculation. The figures are participants per arm for 80% power and participants per arm for 90% power; they are not total event counts. The planning assumptions listed are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set as the stated fraction of the observed effect.

The table shows scenarios for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. For 100% of the observed effect, the entries are 20,803 participants per arm for 80% power and 27,848 participants per arm for 90% power. For 75% of the observed effect, they are 36,805 and 49,271 participants per arm, respectively. For 50% of the observed effect, they are 82,416 and 110,332 participants per arm, respectively.

In plain language, the 80% power column gives the planning number per arm under which the stated calculation would have an 80% chance of detecting the specified effect if the assumptions and effect were correct. The 90% column applies the corresponding 90% target. The higher power target requires a larger planned sample in these illustrations. Smaller effects also require more participants: detecting 50% of the observed effect needs a larger planned sample than detecting the full observed effect.

These figures are not observed power and do not describe the probability that this completed trial detected an effect. They are conditional planning illustrations. The assumptions used for the original trial's design, including its target effect and sample-size justification, are not reported in the registry record.

12. 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, λ₀ and λ₁ are event rates per patient-year, and T is follow-up per participant in years. This assumes Poisson counts and equal follow-up.

Worked calculation for this trial

1. Control rate λ₀ = 0.5400; treatment rate λ₁ = 0.5200 events per patient-year; follow-up T = 1.00 years.
2. n = (1.96 + 0.842)² × (1/0.5400 + 1/0.5200) / (1.00 × [ln(0.963)]²) = 20,802.0, rounded up to 20,803 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
lambda0 <- 0.54                      # control rate, events per patient-year
lambda1 <- 0.52                      # treatment rate
tt      <- 1                       # follow-up per participant, years
n_per_arm <- function(power, frac = 1) {
  l1 <- lambda0 * (lambda1 / lambda0)^frac   # smaller effect = fraction of the observed log rate ratio
  ceiling(round((z_a + qnorm(power))^2 * (1 / lambda0 + 1 / l1) / (tt * log(l1 / lambda0)^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 = ",")))

# Simulation 1: Poisson counts. Simulation 2: over-dispersed counts (variance = 1.5 x mean), analysed with the matching variance.
set.seed(2325856)
n    <- n_per_arm(0.80)
reps <- 20000
m0 <- lambda0 * n * tt; m1 <- lambda1 * n * tt
x0 <- rpois(reps, m0); x1 <- rpois(reps, m1)
z  <- log(x1 / x0) / sqrt(1 / x0 + 1 / x1)
cat(sprintf("Simulated power, Poisson counts, n = %s per arm: %.3f\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE)))
phi <- 1.5
y0 <- rnbinom(reps, mu = m0, size = m0 / (phi - 1)); y1 <- rnbinom(reps, mu = m1, size = m1 / (phi - 1))
z2 <- log(y1 / y0) / sqrt(phi * (1 / y0 + 1 / y1))
cat(sprintf("Simulated power, over-dispersed counts (phi = 1.5), same n: %.3f\n", mean(abs(z2) > z_a, na.rm = TRUE)))

Output from running the code

100% of observed effect: n per arm = 20,803 (80% power), 27,848 (90% power)
 75% of observed effect: n per arm = 36,805 (80% power), 49,271 (90% power)
 50% of observed effect: n per arm = 82,416 (80% power), 110,332 (90% power)
Simulated power, Poisson counts, n = 20,803 per arm: 0.800
Simulated power, over-dispersed counts (phi = 1.5), same n: 0.627

The first simulation draws Poisson counts and should land close to the target power. The second allows extra variability (variance 1.5 times the mean, a purely illustrative value), which lowers the power at the same sample size, so rate trials usually need more participants than the Poisson formula suggests.

13. Multiplicity, Missing Data and Analysis Populations

The posted primary outcome is all-cause hospitalization. The record also lists complications during dry-weight adjustment as a secondary outcome and indicates one secondary outcome posted. No result for that secondary outcome is included in the registry record. The number of outcome measures is listed as 2, and the registry lists no analysis entries. The available record does not provide a complete prespecified list of comparisons or a testing hierarchy.

Multiplicity arises when several outcomes, timepoints, groups, or analyses are tested. The more comparisons considered, the greater the chance that at least one result may appear statistically notable by chance alone. The primary outcome is identified, but the registry record does not state whether multiplicity adjustment was planned or used, how secondary analyses were handled, or whether any subgroup analyses were prespecified. The posted p-value is described as unadjusted, so it should not be interpreted as a multiplicity-adjusted finding.

The registry record does not report how missing outcome data were handled. It does not state whether participants with incomplete follow-up were excluded, whether exposure time was censored at a particular point, or whether any imputation or sensitivity analysis was conducted. These details matter for rates because the interpretation depends on both event counts and time at risk.

The analysis population is also not reported. The overall enrollment is 298, while the primary outcome results list 128 participants in the Study Group and 123 in the Control Group. The record does not explain how participants were assigned to these outcome groups, whether these counts reflect all randomized participants, or why they differ from the enrollment figure. It does not report an intention-to-treat analysis or another named analysis population. Consequently, the result can be described as posted, but the relationship between the reported outcome counts and all randomized participants cannot be established from the registry record.

14. 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 0% reporting threshold.

The registry lists serious adverse events affecting 0 participants in the Study Group, among 128 at risk, and 0 participants in the Control Group, among 123 at risk. These are participant counts for the serious adverse event summary, not event rates per patient-year. The figures describe what is recorded in the registry for the stated groups.

Zero affected participants in the posted summary does not establish that no serious adverse events could occur, or that the approaches have identical safety. Safety interpretation depends on the way events were defined, collected, and reported, as well as the amount of observation. The registry record does not provide serious adverse event terms in the listed top-term field, and it does not report death counts. Those items should not be inferred from the zero figures.

The record also lists 0 participants with other adverse events in each group, with 128 and 123 at risk, respectively. The available information does not give event descriptions or further details about surveillance. These counts are therefore best read as registry summaries rather than as a comprehensive account of all possible harms.

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

The posted primary result describes all-cause hospitalization rates of 0.52 hospitalizations per patient-year in the Study Group and 0.54 in the Control Group over 1 year. The rate ratio of 0.963 places the point estimate slightly below equal rates, but the 95% confidence interval from 0.768 to 1.208 includes 1. The unadjusted p-value is 0.744. Together, these summaries show that the comparison is uncertain and do not establish a clear difference in relative hospitalization rates.

This does not mean the approaches have been shown to be equivalent. Equivalence requires a defined margin and an analysis designed to assess whether the result falls within that margin; neither is reported in the registry record. Nor does the result demonstrate that any observed difference is clinically important. The interval includes a range of values in both directions, and the registry does not specify a threshold for a meaningful difference.

The endpoint concerns hospitalizations from any cause, expressed relative to patient-year exposure. It does not tell us which causes contributed to the events, how many distinct participants experienced hospitalization, or whether some participants had repeated events. The posted outcome result also does not establish effects on the listed secondary outcome, complications during dry-weight adjustment, because its result is not included.

Interpretation is further limited by reporting gaps. The registry record does not explain the relationship between the overall enrollment of 298 and the outcome group counts of 128 and 123. The analysis population, missing-data handling, detailed endpoint definition, and statistical analysis plan are not reported. The result is therefore best understood as a description of the posted comparison, with its stated uncertainty, rather than a complete account of all trial analyses or outcomes.

16. Why This Trial Matters Statistically

This trial is statistically notable because its primary outcome is a rate rather than a simple count of participants with an event. Rates use exposure time and can include repeated events, so readers need to distinguish them from risks or proportions. The registry reports hospitalizations per patient-year and a rate ratio, making the scale of the outcome central to interpretation.

The result also illustrates why a point estimate should be read together with its confidence interval. The estimated rate ratio is 0.963, but its 95% confidence interval extends from 0.768 to 1.208. The interval includes equal rates as well as relative rates in either direction. Reporting the estimate alone could overstate how precisely the comparison is known; reporting only the p-value would also hide the range of uncertainty.

The registry's planning illustrations show how sample-size requirements change with the effect targeted. Smaller effects require more participants under the listed assumptions. These calculations are useful for understanding planning trade-offs, but they are not the original trial's design justification and are not observed power. Keeping those categories separate prevents a retrospective calculation from being mistaken for a feature of the trial protocol.

Finally, the record highlights the importance of complete reporting. Enrollment and outcome-group counts are both present, but their relationship is not explained. Details of analysis populations, missing data, and the statistical plan are not reported. Such information is needed to assess how the posted result was produced and how broadly it represents the randomized trial.

17. 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.
• Rate comparisons use a Poisson approximation unless the sponsor posted confidence limits; real event counts are often over-dispersed.

The registry record leaves several trial-specific questions unresolved. It identifies randomization and a parallel-group design but does not report sequence generation, allocation concealment, eligibility criteria in detail, or whether randomization used stratification. Masking is listed as none, but the record does not describe any separate masking of outcome assessment or analysis.

The overall enrollment is 298, whereas the primary outcome groups list 128 and 123 participants. The record does not explain the difference, describe participant flow, or state whether all randomized participants contributed outcome data. The analysis population and handling of missing follow-up are not reported. This makes it difficult to assess the extent to which the posted result represents the full randomized population.

The primary outcome is named and its timeframe and rate unit are stated, but the detailed definition of a qualifying hospitalization and the construction of exposure time are not reported. There is no breakdown of events by cause, no account of repeated events by participant, and no information about follow-up completeness. The exact statistical model implementation and any assessment or handling of over-dispersion are also not reported.

The registry lists complications during dry-weight adjustment as a secondary outcome, but its result is not included in the registry entries. The record does not provide a full statistical analysis plan, a multiplicity strategy, or the trial's original sample-size justification. The displayed planning calculations use assumptions for illustration, not the trial's own reported design assumptions. These limitations constrain assessment of the full trial and do not change the numerical description of the posted primary result.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Application of Bioimpedance Spectroscopy in Taiwan Dialysis Patients — NCT02325856.
• Registry record last updated 2016-02-02; status last verified 2015-07.
• Results first posted 2015-08-17.
• 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.

21. Record Summary

NCT02325856 is a completed randomized, parallel-group trial with no masking, titled “Application of Bioimpedance Spectroscopy in Taiwan Dialysis Patients.” It compared dry-weight determination using bioimpedance spectroscopy with determination based on clinical symptoms or clinical judgement in hemodialysis patients. The registry lists fluid overload as the condition, An Hsin QingShui Clinic as the sponsor, and 298 participants enrolled.

The primary outcome was all-cause hospitalization, reported as hospitalizations per patient-year over 1 year. The posted rates were 0.52 in the Study Group and 0.54 in the Control Group. The rate ratio was 0.963, with a 95% confidence interval from 0.768 to 1.208 and an unadjusted p-value of 0.744. The interval includes equal rates and values on either side.

Complications during dry-weight adjustment is listed as a secondary outcome, but its result is not reported in the registry record. The record also does not report the analysis population, missing-data handling, or the trial's original sample-size assumptions.