Improving Palliative Care Through Teamwork: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Improving Palliative Care Through Teamwork (NCT01990742), a randomized na study. The primary outcome is Hospital Site of Death: Number of Decedents Cared for in an Enrolled and Assigned NH Facility, Who Were Transferred to a Hospital and Died in the Hospital During the Study Period.

2. Clinical Question

The registry objective was to evaluate whether nursing-home palliative care practice guidelines implemented through teams would improve care processes and outcomes for residents at the end of life. The primary outcome operationalizes one aspect of that objective: hospital death after transfer from a nursing home. It does not, by itself, describe symptoms, comfort, the quality of communication, or the reasons for a transfer. The posted results also include hospitalizations as a secondary outcome, measured as the mean number during the last 90 days of life. These outcomes therefore address related but distinct questions: one concerns where death occurred after transfer, and the other concerns hospital use near the end of life.

3. Trial Design

The registry record describes a randomized, parallel-group trial with no masking and lists the phase as NA. Its title is “Improving Palliative Care Through Teamwork,” and the official title is “Improving Palliative and End-of-Life Care in Nursing Homes.” The University of Rochester is listed as the sponsor. The record gives a start date of 2013-09-27, a primary completion date of 2016-12-31, and a completed status.

The intervention was a Palliative Care Team. The arm description says teams would be established and would round with residents in intervention homes. The standard-care arm is described as usual care for residents in control homes. The registry summary frames the study as a randomized controlled trial of nursing-home-based teams and says its objective was to use a difference-in-differences analytic approach. The reported analysis entries, however, identify Monte Carlo simulation and do not provide an effect estimate, confidence interval, or test statistic. Further details of the analysis used for the posted sponsor results are not reported in the registry record.

The record lists enrollment as 31. The posted primary outcome results give participant counts of 2852 for PCTeam and 2978 for standard care, and the secondary outcome gives counts of 1907 and 1930. The relationship between the enrollment figure and these outcome counts, including the units represented by enrollment and how participants were included in each outcome, is not reported in the registry record. The group descriptions refer to intervention and control homes, but the record does not give enough information to characterize the allocation unit or the handling of residents nested within homes. That matters statistically because outcomes from residents at the same home may be related; whether and how such dependence was incorporated into the analyses is not reported in the registry record.

The trial is marked completed, but completion alone does not provide details about adherence, follow-up completeness, changes to the analysis plan, or the timing of any analyses. Those details are not reported in the registry record. The posted results are therefore best read alongside the stated design and the specific outcome definitions, while keeping the unreported design and analysis details in view.

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: Nursing Homes, not individual participants, were enrolled in this RCT study, and were the unit of intervention. However, the outcomes analysis was performed on decedents who were cared for in the enrolled Nursing Homes.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Nursing home residents cared for in an assigned facility, who died during the study timeframe

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: All decedents with data who resided in assigned facilities.

6. Primary Endpoint

The registry lists the primary outcome as Hospital Site of Death: Number of Decedents Cared for in an Enrolled and Assigned NH Facility, Who Were Transferred to a Hospital and Died in the Hospital During the Study Period, measured over 1 year.

The primary endpoint is the number of decedents cared for in an enrolled and assigned nursing-home facility who were transferred to a hospital and died there during the study period. The registry defines the population as nursing-home residents cared for in an assigned facility who died during the study timeframe, and gives a timeframe of 1 year. The unit is participants, and the measure is a count of participants meeting the endpoint definition.

This endpoint is used to describe one care process and outcome relevant to end-of-life care: whether death occurred in the nursing home or in a hospital following transfer from the nursing home. It is narrower than a general measure of hospital use. A resident may be hospitalized without dying in hospital, and a death in hospital after transfer is only one possible pathway. The endpoint does not identify why a transfer occurred, whether it was consistent with the resident’s preferences, or whether the care received was appropriate. Those details are not reported in the registry record.

The posted results list 579 participants in the PCTeam group and 441 in standard care for this outcome, with group counts of 2852 and 2978, respectively. The posted comparison also gives proportions of 0.203 and 0.148, and a risk difference of 0.055. These are descriptive estimates from the posted summaries; the sponsor’s own analysis is reported separately and takes precedence where it conflicts with a recalculation. The registry record does not describe how deaths or transfers were verified, how residents with incomplete information were handled, or whether the endpoint was analyzed with adjustment for the nursing-home structure.

7. Results Posted in the Registry

• Palliative Care Team (PCTeam): 20.3% (579/2852) versus 14.8% (441/2978) on control; risk difference 5.5 percentage points (95% CI 3.5 to 7.4), risk ratio 1.37 (1.22–1.53); unadjusted two-proportion 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.

Hospitalizations

• 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 deviations.
• Result:
  • Palliative Care Team (PCTeam): mean 0.34 versus 0.29 on control; difference 0.05 (95% CI 0.01 to 0.09), standardized effect d = 0.08; z-test p = 0.018.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Palliative Care Team (PCTeam) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The registry lists hospitalizations as the secondary outcome. Its posted summary gives the mean number of hospitalizations during the last 90 days of life, and the simple recalculation gives an unadjusted mean difference of 0.05. The registry’s own analysis reports Monte Carlo simulation with p >0.05, while the simple recalculation gives p = 0.018. The sponsor analysis takes precedence because the record identifies a discrepancy.

The secondary outcome is related to the primary question because both concern aspects of hospital use near the end of life. They are not interchangeable outcomes: the primary endpoint concerns death in hospital after transfer, while hospitalizations measures a mean count during a specified period. A difference in one does not necessarily imply a difference in the other, and the secondary outcome does not directly explain why a resident was transferred or where death occurred.

Unadjusted secondary results carry less weight than a prespecified primary analysis for several reasons. Secondary outcomes may be among multiple questions examined, raising the chance of an apparently notable result by chance when no multiplicity adjustment is used. The registry record does not report whether adjustment was made. The secondary outcome also has its own population, measurement, and analysis choices, and the record does not describe how missing observations or clustering were handled. The recalculated p-value alone therefore should not be treated as confirmatory evidence.

Across the outcomes with posted sponsor analyses, the reported p-values are both >0.05, while unadjusted recalculations produce smaller p-values. This is a pattern of discordant inference rather than a consistent basis for choosing the recalculations over the trial’s own analysis. The record does not report other secondary outcome results in sufficient detail to characterize a broader pattern across all outcomes. No conclusion about the overall quality of end-of-life care can be drawn from the hospitalization outcome alone.

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

• Hospitalizations

9. Statistical Interpretation

10. Statistical Methods Explained

Several statistical ideas help distinguish what the posted numbers describe from what the registry’s reported analysis establishes. The endpoint is a count of participants meeting a binary definition: each person either met the definition of hospital death after transfer during the study period or did not. For such an outcome, a group proportion expresses the share of the group with the event. A risk difference compares those proportions by subtraction; its sign and size depend on which group is treated as the intervention and which as the comparator. The posted recalculation uses standard care as the reference and reports a risk difference of 0.055 for PCTeam.

What does an unadjusted recalculation represent?

An unadjusted recalculation uses the group counts and denominators in the posted summaries without accounting for other variables or design features. It is useful as a transparent description of the contrast implied by those summaries. It is not necessarily the same estimand or analysis used by the trial sponsor. An adjusted or model-based analysis may account for factors such as the way residents are grouped within homes, changes over time, or other covariates. Whether those features were used in the sponsor’s analysis is not reported in the registry record.

The primary recalculation reports a risk difference and confidence interval, and an unadjusted p-value of <0.001. A p-value summarizes how compatible the observed data are with a specified null hypothesis under the assumptions of the test; it is not the probability that the null hypothesis is true, nor does it describe clinical importance. The registry’s reported Monte Carlo simulation result is p >0.05. Since the record flags a discrepancy between the sponsor result and a recalculation, the sponsor’s own analysis takes precedence. The record does not report enough detail to identify why these methods yield different inferences.

Why might the analysis approach matter?

The registry summary describes the planned objective as using a difference-in-differences analytic approach. In general, difference-in-differences compares changes over time between groups, rather than relying only on a single end-of-study comparison. Its interpretation depends on the data available over time and on assumptions about how the groups would otherwise have changed. The registry record does not report the time-specific outcome data or the assumptions assessed for this trial. The sponsor’s reported analysis entry names Monte Carlo simulation, but provides no model specification, estimate, or confidence interval. The record therefore does not show how the stated objective and reported method relate in the final analysis.

The intervention and control descriptions refer to nursing homes, while outcome summaries report participant counts. In general, residents within the same home may share care processes, staffing, and local circumstances. If outcomes within a home are correlated, an analysis that treats every resident as independent can understate uncertainty. Analyses can account for such clustering, but the registry record does not report the allocation unit or whether clustering was incorporated. This is a limitation in interpreting the precision of any resident-level comparison, not evidence that a particular adjustment was or was not used.

What does the confidence interval add?

A confidence interval describes the range of effect estimates that are reasonably compatible with the data under the method’s assumptions. A narrower interval indicates greater statistical precision than a wider interval, but precision is distinct from freedom from bias. A confidence interval does not account automatically for unmeasured confounding, incorrect outcome classification, clustering ignored by the method, or incomplete follow-up. The intervals reported for the recalculations are based on simple unadjusted methods; they should not be assumed to represent the sponsor’s analysis.

For the primary outcome, the posted recalculation gives a confidence interval from 0.035 to 0.074 around the risk difference of 0.055. For hospitalizations, the unadjusted mean difference is 0.05 with a confidence interval from 0.009 to 0.091. These intervals characterize uncertainty around the recalculated contrasts only. They do not resolve the difference from the sponsor’s p >0.05 results or establish which analysis assumptions are appropriate.

How should the secondary continuous outcome be read?

The hospitalizations outcome is reported as the mean number of hospitalizations during the last 90 days of life, with standard deviations of 0.69 in PCTeam and 0.62 in standard care. A mean difference summarizes the contrast between group averages. The standard deviation describes spread among observations within a group; it is not the same as the standard error, which describes uncertainty in an estimated mean. The recalculation uses a pooled standard deviation of 0.656 and reports a standardized effect size of 0.076. Those are descriptive quantities from the unadjusted calculation and are not the sponsor’s reported effect estimate.

The posted secondary outcome also has different group counts from the primary outcome. The timeframe is the last 90 days of life, so its analytic population and observation window differ from the primary endpoint’s 1 year timeframe. The registry record does not explain the selection or availability of observations for either outcome, or how deaths and follow-up were handled in the hospitalization analysis. Such details matter when interpreting why outcome denominators differ and whether the groups remain comparable in the analysis.

What cannot be concluded from the reported methods?

The registry record does not report a full statistical analysis plan, the original sample-size justification, interim analyses, missing-data methods, or the population used for each analysis. It also does not provide enough information to assess covariate adjustment, clustering, model diagnostics, or the assumptions of the Monte Carlo simulation. The trial summary’s mention of difference-in-differences is not a substitute for those specifications. The available figures allow a reader to describe the posted group results and understand the simple recalculations, but not to reconstruct the sponsor’s full analysis.

These distinctions are especially important when a small p-value from a simple recalculation differs from the sponsor’s reported result. Statistical methods answer questions under their own assumptions; they are not interchangeable labels for the same calculation. The unadjusted recalculation should be read as a secondary analytical aid, while the sponsor analysis remains the primary reported inference for this trial. Neither resolves broader questions about outcome meaning, generalizability, or data completeness that the registry record does not address.

11. Confidence Intervals and Precision

The primary outcome recalculation estimates a risk difference of 0.055, with a confidence interval from 0.035 to 0.074. The interval conveys uncertainty around this unadjusted contrast under the calculation’s assumptions. It is not a range guaranteed to contain the true effect, and it does not account for possible bias, outcome misclassification, or dependence among residents in the same home. The registry record does not state whether the sponsor’s analysis produced a confidence interval; none is reported with its Monte Carlo result.

For hospitalizations, the unadjusted mean difference is 0.05, with a confidence interval from 0.009 to 0.091. This interval describes uncertainty around the difference in group means, not the range of individual residents’ hospitalization counts. The group standard deviations, 0.69 and 0.62, describe observed spread; they do not replace the interval around the estimated mean difference.

Precision and validity are separate. A calculation can have a relatively narrow interval and still be affected by a mismatch between the analysis and trial design, incomplete outcome data, or other biases. In this trial, the reported discrepancy between recalculations and sponsor analyses is particularly relevant: the intervals belong to the recalculated contrasts and should not be read as supporting the sponsor’s result. The registry record does not report enough detail to quantify how accounting for homes or other design features would change uncertainty.

The posted denominators are substantial at the participant level, but the enrollment figure is 31 and its relationship to the outcome counts is not explained. Larger participant counts do not, by themselves, establish greater precision if the analysis structure or effective amount of independent information is unclear. The confidence intervals are useful descriptions of the unadjusted estimates, while the registry’s limited reporting constrains how confidently they can be interpreted as trial-level uncertainty.

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 figures are illustrations based on the observed effect, not the trial’s original design assumptions. The registry record does not report the trial’s own sample-size justification or original planning assumptions. These illustrations use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set at the stated fraction of the observed effect.

For these outcomes, the planning values are participants per arm, not total events. The 80% power column gives the estimated participants per arm for a design with an 80% chance of detecting the stated effect under the illustration’s assumptions, if that effect and assumptions were correct. The 90% power column gives the corresponding participants per arm for a design with a 90% chance. Higher target power requires more participants under otherwise unchanged assumptions.

Rows that plan for 100%, 75%, or 50% of the observed effect show how the required sample size changes as the target effect becomes smaller. Smaller effects are harder to distinguish from random variation, so the planning illustration requires more participants per arm to detect them at the same target power. The figures are not observed power: they are prospective planning illustrations, not a measure of the probability that this completed trial detected an effect.

The illustrations are tied to the simple observed-effect calculations and do not resolve the discrepancy with the sponsor’s reported analyses. They also do not account for any design features that are not included in the stated assumptions, such as possible dependence among residents in the same home. The registry record does not report the original sample-size calculation, so these values should not be interpreted as evidence that the trial was designed around them.

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, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.148; treatment proportion p₁ = 0.203; average p̄ = 0.176.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5380 = 1.0545.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5366 = 0.4516.
4. n = (1.0545 + 0.4516)² / (0.0549)² = 751.8, rounded up to 752 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
p0 <- 0.14808596373405                          # control proportion
p1 <- 0.203015427769986                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^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 (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(1990742)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 752 (80% power), 1,006 (90% power)
 75% of observed effect: n per arm = 1,296 (80% power), 1,735 (90% power)
 50% of observed effect: n per arm = 2,822 (80% power), 3,777 (90% power)
power.prop.test: 751.8 
Simulated power at n = 752 per arm: 0.804 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-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 reports Monte Carlo simulation for the primary outcome and gives p >0.05, without an effect estimate or confidence interval. For hospitalizations, the reported analysis is also Monte Carlo simulation with p >0.05. Simple recalculations from the posted summaries instead give p <0.001 for the primary outcome and 0.018 for hospitalizations. The registry record flags a discrepancy between sponsor analyses and recalculations.

Because of that discrepancy, the sponsor’s own analysis takes precedence. The recalculations are unadjusted comparisons derived from posted group summaries; they may not reproduce the sponsor’s model, assumptions, or treatment of the trial design. The record does not provide enough detail to explain why the results differ, including the simulation procedure, model specification, or any adjustment for nursing homes or other factors.

The sponsor results should not be treated as proof that the groups were identical, just as the unadjusted p-values should not replace the sponsor’s reported analysis. A p-value does not quantify the size or practical importance of an effect. Without the sponsor’s effect estimates and uncertainty intervals, the magnitude and precision of its analysis are not reported in the registry record. The most transparent reading is to keep the sponsor result distinct from the descriptive summaries and to recognize that the registry does not supply sufficient methodological detail to reconcile them.

15. Multiplicity, Missing Data and Analysis Populations

The registry reports one primary outcome and lists hospitalizations as a secondary outcome, while also stating that there were three outcome measures and three analyses. The other outcomes and their results are not described in the record excerpt. A multiplicity strategy—such as whether outcomes were ranked, adjusted, or treated as exploratory—is not reported in the registry record. Without that information, unadjusted secondary p-values should be interpreted cautiously because examining multiple outcomes can increase the chance of finding a result that appears unusual by chance.

Missing-data methods are not reported in the registry record. It does not state how residents with unavailable outcome information were handled, whether the outcome counts include all eligible residents, or whether any imputation or sensitivity analysis was performed. The group counts differ between the primary and secondary outcomes, but the record does not explain why. Different outcome definitions and timeframes may be relevant, but the specific reason for the difference is not reported.

The analysis populations are also not specified. The record does not say whether analyses followed an intention-to-treat principle, included only residents with complete outcome information, or used another population definition. It gives an enrollment of 31 and participant counts in the posted outcomes that are much larger; the relationship between these figures is not reported. The unit represented by enrollment and the role of nursing homes in the analysis therefore remain unclear.

Finally, the registry does not report whether statistical methods accounted for residents being grouped within homes, how the Monte Carlo simulation was constructed, or whether the stated difference-in-differences objective was implemented in the final analysis. These omissions matter for uncertainty, comparability, and interpretation. They do not establish that any particular method was absent; they mean the relevant details are not reported in the registry record.

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 0% reporting threshold.

Most frequent serious adverse event terms

Serious events with the highest arm-level rate: death from causes not associated with the study (100.0%).

The serious adverse event entries list 2852 participants at risk and affected in the PCTeam group, and 2978 at risk and affected in standard care. The registry also lists 2852 and 2978 deaths, respectively. The leading serious event term is “Death from causes not associated with the study,” reported as 100.0%. No other adverse events are listed in the other-event fields.

These are registry reporting figures and should be read in the context of the outcome and population entries. The primary outcome itself concerns deaths and their location after transfer. The adverse-event entries do not, by themselves, establish that the intervention caused deaths or that all deaths were attributable to an adverse event. The listed term explicitly characterizes the deaths as from causes not associated with the study.

The record does not provide a narrative explaining how adverse events were collected, how seriousness or relatedness was determined, or how the entries correspond to the outcome population and follow-up. Nor does it report additional detail about event timing or individual causes. The figures therefore describe the registry’s adverse-event reporting, but they do not support a causal safety comparison beyond what is explicitly recorded.

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

The posted results describe differences between the PCTeam and standard-care groups for the recorded outcomes. For the primary endpoint, the posted counts and unadjusted proportions yield a risk difference of 0.055 in the PCTeam direction. The sponsor’s own analysis reports p >0.05, and the registry flags the discrepancy with the unadjusted recalculation. That sponsor analysis takes precedence; the available registry information does not provide an estimate or interval from it.

This does not establish that the groups were identical, nor does it establish a definitive effect in either direction. A non-small p-value is not proof of no difference, and a smaller p-value from a simple recalculation is not proof that the sponsor analysis is wrong. Interpretation depends on the design and assumptions behind each analysis, several of which are not reported in the registry record.

The endpoint is limited to hospital death after transfer from a nursing home. It does not measure all dimensions of palliative or end-of-life care, and it cannot explain whether transfers were appropriate, whether care matched residents’ preferences, or why particular decisions were made. The registry objective included care processes and resident and staff outcomes, but the results described in the record do not provide a complete evaluation of those domains.

The trial’s setting and intervention are relevant to nursing-home practice, but the record does not report enough information about the facilities, residents, implementation, or analysis to judge how broadly the findings apply. The enrollment figure and posted participant counts are not reconciled, and details of missing data and clustering are not reported. The secondary hospitalization result is also subject to the sponsor-versus-recalculation discrepancy and is not a substitute for the primary outcome.

Accordingly, the most defensible interpretation separates the descriptive posted summaries from the sponsor’s reported inference. The summaries show the observed group data as recorded; the sponsor analysis is the trial’s own reported statistical conclusion. Neither alone answers broader questions about quality of care, resident preferences, or the mechanisms behind transfers. Those conclusions require information that is not reported in the registry record.

18. Why This Trial Matters Statistically

This trial raises a central statistical issue in applied clinical research: a result can look different when analyzed from simple group summaries than when evaluated using the sponsor’s own method. The primary outcome recalculation gives p <0.001, while the registry’s reported Monte Carlo analysis gives p >0.05. The hospitalizations recalculation gives 0.018, while its reported sponsor analysis also gives p >0.05. The registry flags this discrepancy, so the sponsor analysis takes precedence, but the record does not provide enough detail to understand the source of the difference.

The trial also illustrates why the level of observation and the level of intervention need to be clear. The intervention is described in terms of teams and nursing homes, while results are reported as participant counts. Residents within a home may share care context, and the registry does not state how that structure was handled statistically. Clear reporting of allocation units, analytic populations, clustering, and outcome denominators is important for interpreting both effect estimates and their precision.

Another lesson is the need to distinguish outcomes that are related but not equivalent. Hospital death after transfer and the mean number of hospitalizations in the last 90 days of life address different aspects of end-of-life care. Neither alone captures the full quality of care or residents’ preferences. Reporting outcome definitions and analysis methods together helps readers understand what a trial result does and does not represent.

Finally, planning illustrations based on an observed effect are not a substitute for the original sample-size rationale. The registry record does not report that rationale, and the illustrations are not observed power. Transparent records of prespecified analyses, missing-data handling, multiplicity, and design assumptions would help readers distinguish planned inference from post hoc recalculation and assess the reliability of reported conclusions.

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.

Several trial-specific limitations arise from gaps in the registry record. The record lists enrollment as 31, whereas the posted primary outcome group counts are 2852 and 2978. The secondary outcome counts are 1907 and 1930. The relationship among these figures, including the units each represents and how each outcome population was defined, is not reported. This makes it difficult to interpret the flow from enrollment to analysis.

The intervention and control descriptions refer to nursing homes, but the record does not specify the allocation unit or explain whether statistical analyses accounted for residents grouped within homes. If residents in the same home have related outcomes, that structure can affect uncertainty estimates. The record does not say whether such dependence was addressed, so the recalculated intervals should be understood as unadjusted rather than assumed to represent the full design.

The registry describes a difference-in-differences objective but reports Monte Carlo simulation for the sponsor analyses without an estimate, confidence interval, or model details. It does not explain how the final analysis relates to the stated objective, why the sponsor results differ from simple recalculations, or what assumptions were used. Missing-data handling, analysis populations, multiplicity procedures, interim analyses, and the original sample-size justification are not reported in the registry record.

The primary outcome is narrowly defined as death in hospital following transfer. It does not measure all aspects of palliative care quality or explain the circumstances of transfer. The record also does not give detailed information about facility characteristics, implementation fidelity, resident preferences, outcome verification, or follow-up completeness. These limitations restrict interpretation of the posted results and their applicability beyond the outcomes and setting described in the registry.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Improving Palliative Care Through Teamwork — NCT01990742.
• Registry record last updated 2018-09-17; status last verified 2018-08.
• Results first posted 2018-08-20.
• 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

The registry record describes a completed, randomized, parallel-group, unmasked trial sponsored by the University of Rochester. It compares Palliative Care Teams, whose members would round with residents in intervention homes, with standard care in control homes. The primary outcome is the count of residents who were transferred from a nursing home and died in hospital during the 1 year timeframe.

The posted primary group summaries yield an unadjusted risk difference of 0.055, while the sponsor’s reported Monte Carlo analysis gives p >0.05. Hospitalizations during the last 90 days of life is listed as a secondary outcome; its sponsor analysis also reports p >0.05. The registry flags discrepancies between the sponsor analyses and simple recalculations, so the sponsor analyses take precedence.

The record does not report enough detail to reconcile enrollment with outcome counts or to assess the analysis population, missing data, multiplicity, or handling of residents within homes. The results are therefore best interpreted as outcome-specific registry reports, not as a complete assessment of end-of-life care quality.