Beyond the Composite: Enhancing Trial Analysis through a Divide & Conquer Approach to 'Days Alive and at Home': Insights from the NOTACS trial

Dominique-Laurent Couturier; Letao Yuan; Sof\'ia S. Villar

arxiv: 2605.16606 · v1 · pith:JPBBLNWHnew · submitted 2026-05-15 · 📊 stat.ME · stat.AP

Beyond the Composite: Enhancing Trial Analysis through a Divide & Conquer Approach to 'Days Alive and at Home': Insights from the NOTACS trial

Letao Yuan , Sof\'ia S. Villar , Dominique-Laurent Couturier This is my paper

Pith reviewed 2026-05-20 15:14 UTC · model grok-4.3

classification 📊 stat.ME stat.AP

keywords Days Alive and at HomeDAHDivide and Conquersample size calculationtrial designzero-inflated distributionbimodal distributionsimulation-based planning

0 comments

The pith

A divide-and-conquer model decomposes 'Days Alive and at Home' into individually modeled parts to improve simulation-based sample size calculations in trials.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper seeks to improve the statistical handling of 'Days Alive and at Home' (DAH), a complex patient-centered outcome in perioperative trials that typically shows a zero-inflated, left-skewed, bi-modal distribution without a clear standard form. By breaking DAH into distinct parts and modeling each separately using data from the NOTACS trial, the authors create a 'Divide & Conquer' approach that fits the data better than previous methods. This matters because accurate modeling supports reliable simulation studies for determining appropriate sample sizes, particularly when the central limit theorem does not apply due to the distribution's features. If the method works as described, it opens the door to more robust trial designs not only for DAH but also for similar composite endpoints.

Core claim

Using 200 data points from the interim data of the NOTACS trial, whose primary endpoint was DAH, we developed a novel 'Divide & Conquer' model that breaks DAH into distinct parts modeled individually. We demonstrate that our approach significantly improves model fit compared to existing alternatives, enabling more suitable DAH data generation that can be used for simulation-based sample size calculations and evaluation of operating characteristics of the statistical test(s). Beyond NOTACS, our work has large potential to inform the design and analysis of other trials using DAH or similar complex endpoints.

What carries the argument

The 'Divide & Conquer' model, which breaks DAH into distinct parts modeled individually to capture its zero-inflated bimodal features better than standard approaches.

If this is right

More suitable DAH data can be generated for simulation-based sample size calculations.
The operating characteristics of statistical tests can be evaluated more realistically.
The method can inform the design and analysis of trials using DAH or similar complex endpoints.
Improved model fit compared to existing alternatives for handling such distributions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

This decomposition approach may apply to other composite endpoints that combine survival and count data.
Further work could explore whether the individual components offer additional clinical insights.
Validation on complete trial data or external datasets would confirm the method's robustness.

Load-bearing premise

That the DAH distribution can be decomposed into distinct, separately modelable components in a way that preserves the joint behavior needed for accurate simulation of trial outcomes.

What would settle it

Simulations using the divided model producing sample size estimates or power that substantially differ from those based on the actual observed distribution in the NOTACS trial data.

read the original abstract

"Days alive and at home" (DAH) is a recent patient-centered outcome measure for perioperative trials, defined as the number of days a patient spends at home during the follow-up period. DAH typically follows a zero-inflated, left-skewed, bi-modal distribution. Other increasingly used complex endpoints, such as days alive without a ventilator, share these statistical features arising from combining survival with another clinically relevant count outcome into a single, comprehensive measure. A key challenge for DAH and similar endpoints is the lack of a readily identifiable distributional form, which complicates the statistical design of trials using it as the primary endpoint, particularly regarding the robustness of sample size calculations and final analyses where the central limit theorem might not be suitable. Using 200 data points from the interim data of the NOTACS trial (ISRCTN14092678), whose primary endpoint was DAH, we developed a novel 'Divide & Conquer' model that breaks DAH into distinct parts modeled individually. To our knowledge, such a model has not been used before for DAH. We demonstrate that our approach significantly improves model fit compared to existing alternatives, enabling more suitable DAH data generation that can be used for simulation-based sample size calculations and evaluation of operating characteristics of the statistical test(s). Beyond NOTACS, our work has large potential to inform the design and analysis of other trials using DAH or similar complex endpoints.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper applies a component-wise decomposition to the zero-inflated bimodal DAH endpoint and reports better in-sample fit on 200 interim points, but leaves the joint-distribution preservation for simulation untested.

read the letter

The core idea is to split DAH into separate pieces—survival, hospital-stay counts, zero-inflation—and model each on its own before putting them back together for data generation. That is the new piece they claim has not been tried for this endpoint before. It targets a real headache in perioperative trials where the composite does not follow any standard distribution and normal approximations can fail for sample-size work. If the recombination step works, it could give more realistic synthetic datasets than trying to force a single parametric form on the whole thing. The abstract shows they compared fit on the NOTACS interim data and found an improvement, which is at least a concrete starting point for this kind of applied work. The approach is straightforward enough that trial statisticians could try it on similar endpoints like days alive without ventilation. The main limitation is that everything rests on the same 200 observations used both to build the model and to claim better performance. No out-of-sample predictive checks, no external dataset, and no direct verification that the recombined samples recover the empirical tails or dependence structure appear in the description. That matters because the whole point is to use the generated data for simulation-based sample-size calculations where the central limit theorem is unreliable. If the parts are treated as conditionally independent in ways that do not match reality, the operating characteristics could be off exactly where robustness is needed most. The paper is aimed at statisticians who design trials with these patient-centered composites and want better simulation tools. A reader already working on DAH or similar endpoints would find the decomposition worth looking at for ideas, even if the current evidence is preliminary. It is worth sending to referees so the methods section and any additional validation can be examined in detail.

Referee Report

3 major / 2 minor

Summary. The manuscript introduces a 'Divide & Conquer' modeling strategy for the 'Days Alive and at Home' (DAH) endpoint, which exhibits zero-inflation, left-skewness, and bimodality. Using 200 interim observations from the NOTACS trial, the authors decompose DAH into separate components (survival, hospital-stay counts, zero-inflation), fit models to each, and recombine them to generate synthetic data intended for simulation-based sample size calculations in settings where the central limit theorem is unreliable. They claim this yields a significantly better fit than existing alternatives and has broad applicability to similar composite endpoints.

Significance. If the recombination step can be shown to recover the empirical joint distribution (including tails and dependence) and if out-of-sample performance is demonstrated, the approach would offer a practical tool for designing and analyzing perioperative trials that use DAH or analogous non-standard endpoints, improving the reliability of simulation studies for sample-size determination and operating-characteristic evaluation.

major comments (3)

[Methods and Results] The model is developed and evaluated exclusively on the same 200 interim data points from the NOTACS trial that it is intended to simulate for future trials. No out-of-sample testing, external validation cohort, or cross-validation procedure is described, which directly undermines the claim that the generated data are suitable for robust simulation-based sample size calculations.
[Abstract and Results] No quantitative evidence (e.g., AIC/BIC differences, likelihood-ratio tests, Kolmogorov-Smirnov statistics, or predictive calibration metrics) is supplied to support the assertion of 'significantly improved model fit' relative to existing alternatives. Without these numbers, the magnitude and statistical significance of the improvement cannot be assessed.
[Results] The manuscript provides no multivariate diagnostics or joint-distribution checks (e.g., comparison of empirical vs. simulated higher-order moments, tail probabilities, or dependence measures) after recombining the separately modeled components. This is load-bearing for the central claim, because accurate simulation when the CLT fails requires that the joint behavior, not merely the marginals, is preserved.

minor comments (2)

[Abstract] The abstract would be strengthened by reporting at least one concrete fit statistic or cross-validation result rather than the qualitative statement 'significantly improves model fit'.
[Methods] Notation for the recombination step (how the survival, count, and zero-inflation components are combined) should be made explicit, ideally with a small algorithmic or equation block.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for their detailed and constructive comments, which highlight important aspects of model validation and diagnostics. We address each major comment below and indicate the revisions planned for the next version of the manuscript.

read point-by-point responses

Referee: [Methods and Results] The model is developed and evaluated exclusively on the same 200 interim data points from the NOTACS trial that it is intended to simulate for future trials. No out-of-sample testing, external validation cohort, or cross-validation procedure is described, which directly undermines the claim that the generated data are suitable for robust simulation-based sample size calculations.

Authors: We agree that development and evaluation on the identical interim dataset represents a limitation for claims of robustness in simulation studies. In the revised manuscript we will add a k-fold cross-validation procedure that holds out portions of the 200 observations, refits the component models on the training folds, and evaluates the recombined simulations against the held-out data using appropriate discrepancy measures. An independent external validation cohort is not available for the NOTACS trial at present, as these are interim observations from a single study; we will therefore note this constraint explicitly while emphasizing the internal validation results. revision: partial
Referee: [Abstract and Results] No quantitative evidence (e.g., AIC/BIC differences, likelihood-ratio tests, Kolmogorov-Smirnov statistics, or predictive calibration metrics) is supplied to support the assertion of 'significantly improved model fit' relative to existing alternatives. Without these numbers, the magnitude and statistical significance of the improvement cannot be assessed.

Authors: We accept that the current version lacks explicit numerical comparisons. The revised manuscript will report AIC and BIC values for the Divide & Conquer decomposition versus standard zero-inflated and hurdle models fitted to the same data, together with Kolmogorov-Smirnov statistics and calibration plots for the marginal distributions of each component. These quantitative results will be placed in the Results section and referenced in the Abstract. revision: yes
Referee: [Results] The manuscript provides no multivariate diagnostics or joint-distribution checks (e.g., comparison of empirical vs. simulated higher-order moments, tail probabilities, or dependence measures) after recombining the separately modeled components. This is load-bearing for the central claim, because accurate simulation when the CLT fails requires that the joint behavior, not merely the marginals, is preserved.

Authors: We recognize that preservation of the joint distribution after recombination is central to the method's utility. The revised Results section will include direct comparisons of empirical and simulated pairwise correlations, selected higher-order moments, and tail probabilities (e.g., P(DAH = 0) and upper-tail quantiles) between the observed data and multiple replicates generated from the recombined model. These checks will be presented alongside the marginal fit metrics. revision: yes

standing simulated objections not resolved

An independent external validation cohort for the NOTACS trial interim data is not currently accessible.

Circularity Check

1 steps flagged

Model fitted and evaluated on interim NOTACS data then repurposed as generator for future-trial simulations

specific steps

fitted input called prediction [Abstract]
"Using 200 data points from the interim data of the NOTACS trial (ISRCTN14092678), whose primary endpoint was DAH, we developed a novel 'Divide & Conquer' model that breaks DAH into distinct parts modeled individually. ... We demonstrate that our approach significantly improves model fit compared to existing alternatives, enabling more suitable DAH data generation that can be used for simulation-based sample size calculations and evaluation of operating characteristics of the statistical test(s)."

Parameters of the Divide & Conquer model are estimated from the identical 200 interim observations that the subsequent 'more suitable DAH data generation' is intended to reproduce. The reported improvement in fit and the synthetic data produced for simulations are therefore direct consequences of the in-sample estimation step; no independent data or external validation is invoked to break the dependence.

full rationale

The paper's central advance is a Divide & Conquer decomposition fitted to 200 interim observations from the same NOTACS trial whose data it is later used to simulate. The abstract explicitly states that the model is developed on these points, its fit is demonstrated on them, and the fitted object is then offered for 'DAH data generation' in simulation-based sample-size work. Because the generation step re-uses the very parameters estimated from the input sample, any claimed improvement in distributional fidelity is forced by construction rather than independently validated. This matches the 'fitted input called prediction' pattern; no external benchmark, hold-out set, or out-of-sample calibration is described in the provided text. The remainder of the derivation (component-wise modeling and recombination) does not introduce additional circularity beyond this core reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review provides no explicit list of free parameters, axioms, or invented entities. The approach implicitly assumes the DAH distribution admits a useful decomposition whose components can be modeled independently while still reproducing the joint distribution for simulation purposes.

pith-pipeline@v0.9.0 · 5804 in / 1351 out tokens · 48986 ms · 2026-05-20T15:14:09.478930+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

we developed a novel 'Divide & Conquer' model that breaks DAH into distinct parts modeled individually... y_Ii = P_i 1(Ei=0) (p̃ + y_Ei) 1(Ei=1) − y_Ci 1(Ci=1) ... logistic for death, ZICPIG for extended stay, ZABB for post-discharge
IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

enabling more suitable DAH data generation that can be used for simulation-based sample size calculations

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

Component over Composite: Mitigating Type I Error Inflation when Imputing "Days Alive and at Home"
stat.ME 2026-05 unverdicted novelty 5.0

Simulation study finds that imputing missing DAH components separately controls type I error better than imputing the composite outcome directly with predictive mean matching.
Component over Composite: Mitigating Type I Error Inflation when Imputing "Days Alive and at Home"
stat.ME 2026-05 accept novelty 5.0

Simulation shows multiple imputation at the DAH component level controls type I error and maintains power better than imputation at the composite level for Mann-Whitney-Wilcoxon analysis.

Reference graph

Works this paper leans on

28 extracted references · 28 canonical work pages · cited by 1 Pith paper

[1]

D., Lanphear, B

Alampi, J. D., Lanphear, B. P., and McCandless, L. C. (2025). Performance of quantile regression methods with discrete outcomes: A simulation study with applications to environmental epidemiology. Environmental Epidemiology 9(6), e432

work page 2025
[2]

A., Cleland, J

Ariti, C. A., Cleland, J. G., Pocock, S. J., Pfeffer, M. A., Swedberg, K., Granger, C. B., et al. (2011). Days alive and out of hospital and the patient journey in patients with heart failure: Insights from the candesartan in heart failure: assessment of reduction in mortality and morbidity (CHARM) program. American Heart Journal 162(5), 900--906

work page 2011
[3]

S., Wang, Y., Chen, J., Vidán, M

Bueno, H., Ross, J. S., Wang, Y., Chen, J., Vidán, M. T., Normand, S. L., et al. (2010). Trends in length of stay and short-term outcomes among Medicare patients hospitalized for heart failure, 1993--2006. JAMA 303(21), 2141--2147

work page 2010
[4]

and Lin, M

Carey, K. and Lin, M. Y. (2014). Hospital length of stay and readmission: An early investigation. Medical Care Research and Review 71(1), 99--111

work page 2014
[5]

M., Faridi, K

Chung, M., Butala, N. M., Faridi, K. F., Almarzooq, Z. I., Liu, D., Xu, J., et al. (2023). Days at home after transcatheter or surgical aortic valve replacement in high-risk patients. American Heart Journal 255, 125--136

work page 2023
[6]

N., Chiu, Y

Dawson, S. N., Chiu, Y. D., Klein, A. A., Earwaker, M., and Villar, S. S. (2022). Effect of high-flow nasal therapy on patient-centred outcomes in patients at high risk of postoperative pulmonary complications after cardiac surgery: A statistical analysis plan for NOTACS, a multicentre adaptive randomised controlled trial. Trials 23(1), 699

work page 2022
[7]

Dunn, P. K. and Smyth, G. K. (1996). Randomised quantile residuals. Journal of Computational and Graphical Statistics 5, 236--244

work page 1996
[8]

Earwaker, M., Villar, S., Fox-Rushby, J., Duckworth, M., Dawson, S., Steele, J., et al. (2022). Effect of high-flow nasal therapy on patient-centred outcomes in patients at high risk of postoperative pulmonary complications after cardiac surgery: A study protocol for a multicentre adaptive randomised controlled trial. Trials 23(1), 232

work page 2022
[9]

Fagerland, M. W. and Sandvik, L. (2009). The Wilcoxon-Mann-Whitney test under scrutiny. Statistics in Medicine 28(10), 1487--1497

work page 2009
[10]

C., Cyr, D., Neely, M

Fanaroff, A. C., Cyr, D., Neely, M. L., Bakal, J., White, H. D., Fox, K. A. A., et al. (2018). Days alive and out of hospital: Exploring a patient-centered, pragmatic outcome in a clinical trial of patients with acute coronary syndromes. Circulation: Cardiovascular Quality and Outcomes 11(12), e004755

work page 2018
[11]

E., Bradshaw, L

Goldberg, S. E., Bradshaw, L. E., Kearney, F. C., Russell, C., Whittamore, K. H., Foster, P. E., et al. (2013). Care in specialist medical and mental health unit compared with standard care for older people with cognitive impairment admitted to general hospital: Randomised controlled trial (NIHR TEAM trial). BMJ 347, f4132

work page 2013
[12]

Z., and Cheung, Y

Ling, W., Cheng, B., Wei, Y., Willey, J. Z., and Cheung, Y. K. (2022). Statistical inference in quantile regression for zero-inflated outcomes. Statistica Sinica 32(3), 1411--1433

work page 2022
[13]

L., McGuinness, S

Litton, E., Parke, R. L., McGuinness, S. P., Dawson, S. N., Villar, S. S., Shetty, S. S., et al. (2026). High-flow nasal oxygen therapy after cardiac surgery: A randomized clinical trial. JAMA Network Open 9(4), e265447

work page 2026
[14]

and Agresti, A

Min, Y. and Agresti, A. (2005). Random effect models for repeated measures of zero-inflated count data. Statistical Modelling 5(1), 1--19

work page 2005
[15]

S., Shulman, M

Myles, P. S., Shulman, M. A., Heritier, S., Wallace, S., McIlroy, D. R., McCluskey, S., et al. (2017). Validation of days at home as an outcome measure after surgery: A prospective cohort study in Australia. BMJ Open 7(8), e015828

work page 2017
[16]

S., Dieleman, J

Myles, P. S., Dieleman, J. M., Forbes, A., Heritier, S., and Smith, J. A. (2018). Dexamethasone for Cardiac Surgery trial (DECS-II): Rationale and a novel, practice preference-randomized consent design. American Heart Journal 204, 52--57

work page 2018
[17]

S., Richards, T., Klein, A., Smith, J., Wood, E

Myles, P. S., Richards, T., Klein, A., Smith, J., Wood, E. M., Heritier, S., et al. (2021). Rationale and design of the intravenous iron for treatment of anemia before cardiac surgery trial. American Heart Journal 239, 64--72

work page 2021
[18]

F., Barat, I., Riis, A

Rasmussen, L. F., Barat, I., Riis, A. H., Gregersen, M., and Grode, L. (2023). Effects of a transitional care intervention on readmission among older medical inpatients: A quasi-experimental study. European Geriatric Medicine 14(1), 131--144

work page 2023
[19]

R., Myles, P

Reilly, J. R., Myles, P. S., Wong, D., Heritier, S. R., Brown, W. A., Richards, T., et al. (2022). Hospital costs and factors associated with days alive and at home after surgery (DAH30). The Medical Journal of Australia 217(6), 311--317

work page 2022
[20]

C., Jr, Martin, S

Shinall, M. C., Jr, Martin, S. F., Karlekar, M., Hoskins, A., Morgan, E., Kiehl, A., et al. (2023). Effects of specialist palliative care for patients undergoing major abdominal surgery for cancer: A randomized clinical trial. JAMA Surgery 158(7), 747--755

work page 2023
[21]

D., Rigby, R

Stasinopoulos, M. D., Rigby, R. A., Heller, G. Z., Voudouris, V., and De Bastiani, F. (2017). Flexible Regression and Smoothing: Using GAMLSS in R . Chapman & Hall/CRC, Boca Raton

work page 2017
[22]

A., Soukkio, P

Suikkanen, S. A., Soukkio, P. K., Aartolahti, E. M., Kautiainen, H., Kääriä, S. M., Hupli, M. T., et al. (2021). Effects of home-based physical exercise on days at home and cost-effectiveness in pre-frail and frail persons: Randomized controlled trial. Journal of the American Medical Directors Association 22(4), 773--779

work page 2021
[23]

S., Dawson, S., Yuan, L., Couturier, D.-L., and Villar, S

Tackney, M. S., Dawson, S., Yuan, L., Couturier, D.-L., and Villar, S. S. (2026). Component over composite: Mitigating type I error inflation when imputing ``Days Alive and at Home'' (Working paper, submitted)

work page 2026
[24]

Food and Drug Administration (2019)

U.S. Food and Drug Administration (2019). Adaptive Designs for Clinical Trials of Drugs and Biologics: Guidance for Industry. https://www.fda.gov/media/78495/download (accessed March 7, 2026)

work page 2019
[25]

H., Smith, V

Van Houtven, C. H., Smith, V. A., Lindquist, J. H., Chapman, J. G., Hendrix, C., Hastings, S. N., et al. (2019). Family caregiver skills training to improve experiences of care: A randomized clinical trial. Journal of General Internal Medicine 34(10), 2114--2122

work page 2019
[26]

Waddingham, E., Phillips, R., and Cornelius, V. (2025). PANTHER Statistical Design Appendix V1.0. https://panthertrial.org/assets/images/uploads/doc/PANTHER_Statistical_design_appendix_V1.0.docx (accessed June 25, 2025)

work page 2025
[27]

Wong, S. S. Y., Cheung, H. H. T., Ng, F. F., Yau, D. K. W., Wong, M. K. H., Lau, V. N. M., et al. (2022). Effect of a patient education video and prehabilitation on the quality of preoperative person-centred coordinated care experience: Protocol for a randomised controlled trial. BMJ Open 12(9), e063583

work page 2022
[28]

T., Cui, D., El-Behesy, B., and Story, D

Wu, A., Fahey, M. T., Cui, D., El-Behesy, B., and Story, D. A. (2022). An evaluation of the outcome metric 'days alive and at home' in older patients after hip fracture surgery. Anaesthesia 77(8), 901--909

work page 2022

[1] [1]

D., Lanphear, B

Alampi, J. D., Lanphear, B. P., and McCandless, L. C. (2025). Performance of quantile regression methods with discrete outcomes: A simulation study with applications to environmental epidemiology. Environmental Epidemiology 9(6), e432

work page 2025

[2] [2]

A., Cleland, J

Ariti, C. A., Cleland, J. G., Pocock, S. J., Pfeffer, M. A., Swedberg, K., Granger, C. B., et al. (2011). Days alive and out of hospital and the patient journey in patients with heart failure: Insights from the candesartan in heart failure: assessment of reduction in mortality and morbidity (CHARM) program. American Heart Journal 162(5), 900--906

work page 2011

[3] [3]

S., Wang, Y., Chen, J., Vidán, M

Bueno, H., Ross, J. S., Wang, Y., Chen, J., Vidán, M. T., Normand, S. L., et al. (2010). Trends in length of stay and short-term outcomes among Medicare patients hospitalized for heart failure, 1993--2006. JAMA 303(21), 2141--2147

work page 2010

[4] [4]

and Lin, M

Carey, K. and Lin, M. Y. (2014). Hospital length of stay and readmission: An early investigation. Medical Care Research and Review 71(1), 99--111

work page 2014

[5] [5]

M., Faridi, K

Chung, M., Butala, N. M., Faridi, K. F., Almarzooq, Z. I., Liu, D., Xu, J., et al. (2023). Days at home after transcatheter or surgical aortic valve replacement in high-risk patients. American Heart Journal 255, 125--136

work page 2023

[6] [6]

N., Chiu, Y

Dawson, S. N., Chiu, Y. D., Klein, A. A., Earwaker, M., and Villar, S. S. (2022). Effect of high-flow nasal therapy on patient-centred outcomes in patients at high risk of postoperative pulmonary complications after cardiac surgery: A statistical analysis plan for NOTACS, a multicentre adaptive randomised controlled trial. Trials 23(1), 699

work page 2022

[7] [7]

Dunn, P. K. and Smyth, G. K. (1996). Randomised quantile residuals. Journal of Computational and Graphical Statistics 5, 236--244

work page 1996

[8] [8]

Earwaker, M., Villar, S., Fox-Rushby, J., Duckworth, M., Dawson, S., Steele, J., et al. (2022). Effect of high-flow nasal therapy on patient-centred outcomes in patients at high risk of postoperative pulmonary complications after cardiac surgery: A study protocol for a multicentre adaptive randomised controlled trial. Trials 23(1), 232

work page 2022

[9] [9]

Fagerland, M. W. and Sandvik, L. (2009). The Wilcoxon-Mann-Whitney test under scrutiny. Statistics in Medicine 28(10), 1487--1497

work page 2009

[10] [10]

C., Cyr, D., Neely, M

Fanaroff, A. C., Cyr, D., Neely, M. L., Bakal, J., White, H. D., Fox, K. A. A., et al. (2018). Days alive and out of hospital: Exploring a patient-centered, pragmatic outcome in a clinical trial of patients with acute coronary syndromes. Circulation: Cardiovascular Quality and Outcomes 11(12), e004755

work page 2018

[11] [11]

E., Bradshaw, L

Goldberg, S. E., Bradshaw, L. E., Kearney, F. C., Russell, C., Whittamore, K. H., Foster, P. E., et al. (2013). Care in specialist medical and mental health unit compared with standard care for older people with cognitive impairment admitted to general hospital: Randomised controlled trial (NIHR TEAM trial). BMJ 347, f4132

work page 2013

[12] [12]

Z., and Cheung, Y

Ling, W., Cheng, B., Wei, Y., Willey, J. Z., and Cheung, Y. K. (2022). Statistical inference in quantile regression for zero-inflated outcomes. Statistica Sinica 32(3), 1411--1433

work page 2022

[13] [13]

L., McGuinness, S

Litton, E., Parke, R. L., McGuinness, S. P., Dawson, S. N., Villar, S. S., Shetty, S. S., et al. (2026). High-flow nasal oxygen therapy after cardiac surgery: A randomized clinical trial. JAMA Network Open 9(4), e265447

work page 2026

[14] [14]

and Agresti, A

Min, Y. and Agresti, A. (2005). Random effect models for repeated measures of zero-inflated count data. Statistical Modelling 5(1), 1--19

work page 2005

[15] [15]

S., Shulman, M

Myles, P. S., Shulman, M. A., Heritier, S., Wallace, S., McIlroy, D. R., McCluskey, S., et al. (2017). Validation of days at home as an outcome measure after surgery: A prospective cohort study in Australia. BMJ Open 7(8), e015828

work page 2017

[16] [16]

S., Dieleman, J

Myles, P. S., Dieleman, J. M., Forbes, A., Heritier, S., and Smith, J. A. (2018). Dexamethasone for Cardiac Surgery trial (DECS-II): Rationale and a novel, practice preference-randomized consent design. American Heart Journal 204, 52--57

work page 2018

[17] [17]

S., Richards, T., Klein, A., Smith, J., Wood, E

Myles, P. S., Richards, T., Klein, A., Smith, J., Wood, E. M., Heritier, S., et al. (2021). Rationale and design of the intravenous iron for treatment of anemia before cardiac surgery trial. American Heart Journal 239, 64--72

work page 2021

[18] [18]

F., Barat, I., Riis, A

Rasmussen, L. F., Barat, I., Riis, A. H., Gregersen, M., and Grode, L. (2023). Effects of a transitional care intervention on readmission among older medical inpatients: A quasi-experimental study. European Geriatric Medicine 14(1), 131--144

work page 2023

[19] [19]

R., Myles, P

Reilly, J. R., Myles, P. S., Wong, D., Heritier, S. R., Brown, W. A., Richards, T., et al. (2022). Hospital costs and factors associated with days alive and at home after surgery (DAH30). The Medical Journal of Australia 217(6), 311--317

work page 2022

[20] [20]

C., Jr, Martin, S

Shinall, M. C., Jr, Martin, S. F., Karlekar, M., Hoskins, A., Morgan, E., Kiehl, A., et al. (2023). Effects of specialist palliative care for patients undergoing major abdominal surgery for cancer: A randomized clinical trial. JAMA Surgery 158(7), 747--755

work page 2023

[21] [21]

D., Rigby, R

Stasinopoulos, M. D., Rigby, R. A., Heller, G. Z., Voudouris, V., and De Bastiani, F. (2017). Flexible Regression and Smoothing: Using GAMLSS in R . Chapman & Hall/CRC, Boca Raton

work page 2017

[22] [22]

A., Soukkio, P

Suikkanen, S. A., Soukkio, P. K., Aartolahti, E. M., Kautiainen, H., Kääriä, S. M., Hupli, M. T., et al. (2021). Effects of home-based physical exercise on days at home and cost-effectiveness in pre-frail and frail persons: Randomized controlled trial. Journal of the American Medical Directors Association 22(4), 773--779

work page 2021

[23] [23]

S., Dawson, S., Yuan, L., Couturier, D.-L., and Villar, S

Tackney, M. S., Dawson, S., Yuan, L., Couturier, D.-L., and Villar, S. S. (2026). Component over composite: Mitigating type I error inflation when imputing ``Days Alive and at Home'' (Working paper, submitted)

work page 2026

[24] [24]

Food and Drug Administration (2019)

U.S. Food and Drug Administration (2019). Adaptive Designs for Clinical Trials of Drugs and Biologics: Guidance for Industry. https://www.fda.gov/media/78495/download (accessed March 7, 2026)

work page 2019

[25] [25]

H., Smith, V

Van Houtven, C. H., Smith, V. A., Lindquist, J. H., Chapman, J. G., Hendrix, C., Hastings, S. N., et al. (2019). Family caregiver skills training to improve experiences of care: A randomized clinical trial. Journal of General Internal Medicine 34(10), 2114--2122

work page 2019

[26] [26]

Waddingham, E., Phillips, R., and Cornelius, V. (2025). PANTHER Statistical Design Appendix V1.0. https://panthertrial.org/assets/images/uploads/doc/PANTHER_Statistical_design_appendix_V1.0.docx (accessed June 25, 2025)

work page 2025

[27] [27]

Wong, S. S. Y., Cheung, H. H. T., Ng, F. F., Yau, D. K. W., Wong, M. K. H., Lau, V. N. M., et al. (2022). Effect of a patient education video and prehabilitation on the quality of preoperative person-centred coordinated care experience: Protocol for a randomised controlled trial. BMJ Open 12(9), e063583

work page 2022

[28] [28]

T., Cui, D., El-Behesy, B., and Story, D

Wu, A., Fahey, M. T., Cui, D., El-Behesy, B., and Story, D. A. (2022). An evaluation of the outcome metric 'days alive and at home' in older patients after hip fracture surgery. Anaesthesia 77(8), 901--909

work page 2022