Predicting hospital readmissions in patients receiving novel-dose Sacubitril/Valsartan therapy: A competing-risk, causal mediation analysis

Background: The effects of novel-dose Sacubitril/Valsartan (S/V) in patients with heart failure (HF) in the real world have not been adequately studied. We examined the risk for all-cause re-admission in the patients with HF taking novel-dose S/V and the possible mediator role of left ventricular reverse remodeling (LVRR).

and Results:There were 464 patients recruited from December 2017 to September 2021 in our hospital with a median follow-up of 660 days (range, 17-1494). Model 1 and 2 were developed based on the results of univariable competing risk analysis, least absolute shrinkage and selection operator approach, backward stepwise regression and multivariable competing risk analysis. The internal verification (data-splitting method) indicated that Model 1 had better discrimination, calibration, and clinical utility. The corresponding nomogram showed that patients aged 75 years and above, or taking the lowest-dose S/V (≤50mg twice a day), or diagnosed with ventricular tachycardia, or valvular heart disease, or chronic obstructive pulmonary disease, or diabetes mellitus were at the highest risk of all-cause readmission. In the causal mediation analysis, LVRR was considered as a critical mediator that negatively affected the difference of novel-dose S/V in readmission.

A significant association was detected between novel-dose S/V and all-cause readmission in HF patients, in part negatively mediated by LVRR. The web-based nomogram could provide individual prediction of all-cause readmission in HF patients receiving novel-dose S/V . The effects of different novel-dose S/V are still needed to be explored further in the future. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint

novel-dose Sacubitril/Valsartan; readmission; left ventricular reverse remodeling; competing-risk nomogram; causal mediation

Heart failure (HF) was a complex clinical syndrome that negatively impacted quality of life, and placed a huge and costly burden on global healthcare system probably causing by high readmission rates1-3. As the first agent of angiotensin receptor-neprilysin inhibitor (ARNI) , Sacubitril/Valsartan (S/V) has been proven to significantly reduce all-cause readmission and mortality of patients diagnosed with HF with reduced ejection fraction (HFrEF)4-7. In clinical practice, however, many patients cannot achieve standard dose (97/103 mg twice a day [b.i.d.]) , or even the lowest approved dose (24/26 mg b.i.d.) due to several factors (i.e., hypotension, hyperkalemia, renal dysfunction)8-10. Given novel-dose S/V (below the standard dose) would be used to treat HF patients in the real world, the clinical effects were of particular concern to clinicians. A clear and early benefit of S/V was to reduce hospital readmissions of HF patients due to any causes 11. McMurray et al. and Desai et al. pointed out that S/V was superior to enalapril in improving HF patient outcomes4,5. Lately, Carnicelli et al. reported that patients with higher adherence to S/V showed a significant reduction in all-cause readmission at 3 months or 1 year12. Unfortunately, there remained poor All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint understanding of novel-dose S/V13. It would be of importance and interest to estimate the effects of novel-dose S/V on patients’ readmissions. Moreover, left ventricular reverse remodeling (LVRR) has the potential to play a role in the beneficial effects of S/V14-16. Specially, some previous studies have demonstrated a strong favorable effect conferred by S/V on LVRR15,17,18, meanwhile some have reported a positive association between LVRR and clinical outcomes of HF patients19,20. It was well known that LVRR was pivotal to the progression of HFrEF patients21-23, while far fewer studies have examined its mediation effects on the relationship between S/V and patients’ outcomes. Herein, this study would estimate the effects of novel-dose S/V on hospital readmissions, construct a risk prediction model, as well as investigate the possible role of LVRR. The findings of our study would hopefully provide crucial insights into the novel-dose S/V , and assist clinicians in selecting treatment options for patients to improve therapeutic outcomes.

Data sources The present study complied with the Declaration of Helsinki and was approved by the ethics committee of Xinqiao Hospital, Army Medical University (Third Military Medical University) in Chongqing, China. We retrieved the data of 2424 patients diagnosed with HF from Xinqiao Hospital from December 29st, 2017 to September All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint 17st, 2021. Patients would be included in our study if the following criteria were met: (1) age ≥18 years, (2) LVEF <50%, (3) an echocardiography was performed as historical data, and another would be performed as a comparison, (4) receiving novel-dose S/V , and would not discontinue therapy during the study period. Exclusion criteria included:(1) underwent CRT or cardiac transplantation therapy before or during the follow-up, (2) loss of follow-up. The flowchart of patients’ selection was shown in Figure 1. Data collection HF-associated raw data was to be mainly collected from the medical records as following: (1) demographic characteristics, i.e., identification number, sex, date of birth, height, weight, etc., (2) comorbidities, i.e., dilated cardiomyopathy (DCM), myocardial infarction (MI), hypertension, coronary heart disease (CHD), valvular heart disease (VHD), chronic obstructive pulmonary disease (COPD), diabetes mellitus (DM), severe renal impairment (SRI), (3) electrocardiogram, i.e., premature atrial contractions (PACs), premature ventricular contractions (PVCs), ventricular tachycardia (VT), (4) echocardiography, i.e., left ventricular ejection fraction (LVEF), left ventricular end-diastolic diameter (LVEDD). All included patients were followed up about 1 year (as of September 1st, 2022). The primary outcome was all-cause readmission, and other outcomes of interest was all-cause death. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint Statistical analysis Given the study aimed to estimate the effects of different novel-dose levels, all patients were divided into three groups based on the time-weighted average dose24, that was, lowest dose (≤50mg b.i.d.), lower dose (50-<100mg b.i.d.) and low dose (100+mg b.i.d.), respectively. Then we split the dataset into training dataset and testing dataset with the ratio of 6:4 stratified by dose groups, which were used to develop predictive model (nomogram) and assess the performance, respectively. All statistical analyses were performed with R-4.2.1. In order to correctly estimate the probability of the primary outcome of interest (i.e., all-cause-readmission), an extension of survival analysis, competing risk analysis, would be employed to predict all-cause readmission associated with different novel-dose S/V , which took into account the fact that competing outcomes (i.e., all-cause death) could prevent the occurrence of primary endpoint25. Firstly, cumulative incidence function (CIF) was used for showing the probability of clinical outcomes (all-cause readmission or death) in those receiving novel-dose S/V over follow-up time, and Gray’s Test was employed to test the equality of CIF curves in novel-dose subgroups. Secondly, univariable competing risk analysis was applied to explore the associations between covariates (i.e., demographic characteristics, comorbidities, electrocardiogram) and all-cause readmission, and those with a P-value < 0.2 would be identified as the potential prognostic variables26. If one of dummy All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint variables was significant or met the inclusion criteria, the variable should be remained and considered as a unit in a model27. Thirdly, in order to avoid missing covariates that were clinically significant, two variable selection methods would be combined to determine the potential predictors for multivariable competing risk analysis, that was, least absolute shrinkage and selection operator (LASSO) approach with 5-fold cross validation and multivariable regression using stepwise backward selection based on Akaike’s Information Criterion (AIC)1. Fourthly, our study would produce two potential models with different predictors using multivariable competing risk analysis, that was, Model 2 (statistically significant variables with P-value < 0.05, and marginally significant with 0.05 < P-value < 0.1 according to the results of LASSO and AIC), Model 2 (statistically significant variables with P-value < 0.05), respectively. Meanwhile, two nomograms based on the multivariable regression models would be constructed to predict all-cause readmission of individual with HR receiving novel-dose S/V by weighting prognostic factors. Fifthly, a precise prognostic nomogram would be selected through evaluating the performance, including discrimination (i.e., concordance index (C-index) based on bootstrap, time-dependent area under the curve (AUC)), calibration (i.e., calibration curve), and net benefit (i.e., decision curve analysis (DCA)). C-index and time-dependent AUC values exceed 0.7 suggested a reasonable estimation28. Accordingly, a web-based dynamic nomogram would be built as a decision-making tool. Moreover, non-linear causal mediation analysis was used to further clarify the All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint association between novel-dose S/V and all-cause readmission through LVRR as the potential mediation (Figure S1). Specially, to promote LVRR was defined as an improvement ≥10% in left ventricular ejection fraction (LVEF) accompanied by a reduction of left ventricular end-diastolic diameter (LVEDD) ≥ 10%, or a LVEF increase ≥10% with at least 6-month follow-up29-31. First and foremost, S/V dosage was considered as a continuous variable to assess the significance of omnibus mediation effect. Then categorical novel-dose S/V was used to estimate the relative mediation effect, of which the lowest was the reference. Note that the estimated effect values were measured in terms of the log of hazards (odds of having all-cause readmission), which could be interpreted as the average differences in the log(hazards) between different levels of novel-dose S/V . Mediation analysis commonly decomposed the total effect into direct effect and indirect effect through mediators32-33, as shown in Figure S1.

Study population A total of 464 patients with a median follow-up of 660 days (interquartile range (IQR):17-1494) were entered into our dataset, which comprised 122, 173 and 169 receiving lowest-, lower- and low-dose S/V , respectively. 278 patients (median follow-up 646 days, IQR: 21-1487) and 186 patients (median follow-up 687 days, IQR: 17-1494) were divided into the training dataset (model development) and testing dataset (model performance), respectively. The baseline characteristics of these All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint patients were summarized in Table S1. In general, a greater proportion of these three groups of patients (lowest-, lower- and low-dose S/V) were male, or those aged <55 years old, or diagnosed with DCM or VHD. Novel-dose related cumulative incidence Gray’s Test showed that there was statistically significant difference in all-cause readmission (P-value < 0.001) across the three novel-dose groups, whereas no significant difference in all-cause death (P-value = 0.307). As illustrated in Figure 2, the estimated cumulative incidences of all-cause readmission of patients receiving the lowest-dose S/V were 4%, 10%, 15%, 25%, 30%, 43% at 30 days, 3 months, 6 months, 1 year, 2 years, 3 years, respectively. The corresponding estimates for patients taking lower-dose S/V were 0%, 1%, 2%, 4%, 11%, 18%, and for those with low-dose S/V were 0%, 2%, 3%, 4%, 6%, 8%, respectively. Risk prediction model Univariate competing risk analysis (Table 1) revealed that age, BMI, VHD, PACs, PVCs, VT, COPD, DM, SRI, dose were identified as the potential prognostic factors on all-cause readmission of HF patients. Then LASSO approach (Figure S2) and stepwise backward elimination method (Table S2) suggested that age, VT, DM and dose were significantly associated with patients’ readmission, while VHD and COPD were marginally significant variables. As shown in Table 1, our study constructed two models to predict primary outcome of HF patients with novel-dose S/V therapy, that All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint was, Model 1(age, VT, VHD, COPD, DM and dose) and Model 2 (age, VT, DM and dose). C-index was 0.752 ( 95% confidence interval (CI) : 0.724, 0.78 9) and 0.702 (95%CI: 0.663, 0.743 ) , time-dependent AUC ( Figure S3) was more than 0.677 and 0.556 over the follow-up period for Model 1 and 2 , respectively . The calibration curves depicted that observed and predicted values of two models were almost consistent (Figure S4) . Decision curves were illustrated in Figure S5 , and exhibited that clinical net benefit gained from Model 1 was higher than two hypothetical scenarios ( none readmission, all readmission ) and Model 2 when the threshold probabilities were within the range of 0%-31% and 0%-7% for 30-day and 6-month readmission, respectively. Model 1 also had a wider range of threshold probabilities (0%-10%, 17%-20%, 23%-33%, 39%-50%) in predicting 3-month readmission. Therefore , Model 1 had considerable discriminative and calibrating abilities, and clinical utility, which was displayed as a static nomogram (Figure3) and a web-based dynamic nomogram ( https://haoxx.shinyapps.io/DynNom_HF/). The Nomogram implied that patients aged 75 years and above were at the highest risk of having novel-dose related readmission, followed by those aged <55y, 65-<75y, 55-<65y. Then all-cause readmission mainly occurred in patients with the lowest-dose S/V , followed by the lower- and low-dose. Lastly, other factors associated with an increase in all-cause readmission included history of VT, VHD, COPD and DM. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint Causal mediation effects An omnibus mediation effect analysis (Figure 4) indicated that LVRR had a significant negative indirect effect on the difference of novel-dose S/V in all-cause readmission, accounting for 20.3% of the total effect. Then the relative mediation effect analysis showed that the odds of having readmission for the lower- and low-dose were respectively on average 0.45 (95% CI: 0.37, 0.56) and 0.14 (95%CI: 0.10, 0.19) times that for the lowest-dose in total effect, while the corresponding odds of indirect effect of LVRR were 0.92 (95%CI: 0.91, 0.94) and 0.978 (95%CI: 0.971, 0.984) times.

Although S/V has been recommended in clinical practice guidelines for patients with HFrEF, some were only able to tolerate novel dosage (below the standard) dominantly due to hypotension. Our study was an attempt to assess the effects of novel-dose S/V on hospital readmissions in these patients, establish clinical prediction models, as well as explore the possible mediation effect of LVRR. Berg et al stated that the efficacy and safety of different S/V levels seemed to be consistent no matter whether patients obtained the maximum dose34. Consistently, our study showed significantly different cumulative incidences of all-cause readmission. The risk of all-cause readmission for patients with lowest-dose S/V were much higher than those with the lower- and low-dose, and mostly occurred within 1 year. However, in our small sized cohort, no All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint difference in all-cause death among the three novel-dose S/V were observed. The clinical effects of novel-dose S/V remained to be further investigated. Reducing hospital readmission was one of the primary targets of the HF treatment. However, it remained unsolved that how to predict which patients with HF would suffer hospital readmission. In this ARNI therapy cohort, we built a nomogram model, which revealed that age, VT, VHD, COPD, DM and dose were associated with an increased risk of all-cause readmission. As expected, patients who taken the lowest-dose S/V were at the highest risk of all-cause readmission, followed by those who taken lower- and low-dose, which was in accordance with cumulative incidence curves. Previous studies have reported some factors associated with hospital readmission of HF patients, such as, male gender 35, age36, DM36, VT37, VHD38, COPD35,39, renal dysfunction40, psychiatric illness40, etc. Similarly, our study suggested that patients aged 75+y were the high-risk population, followed by those aged <55y, 65-<75y, 55-<65y. The elderly patients were mostly observed to have low degrees of physical activity and depression and anxiety, as well as concomitant chronic diseases, which were likely to result in increasing disease risk and aggravating clinical outcomes41,42. Whellan et al. pointed out a dichotomous relationship between age and risk of death or readmission, namely, risk of readmission or death decreased as age increased up to 55 years of age, then increased in those older than 55 years43. Thus, there might be a tick-shaped relationship between age and all-cause readmission in patients taking novel-dose. Future studies might help more finely identify the All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint high-risk population. In addition, patients with VT, or COPD, or DM were at significantly higher risk of all-cause readmission. This might indicate that targeted treatments of these comorbidities would provide a potential opportunity to improve outcomes in patients with HF. Reverse cardiac remodeling refers to the recovery of ventricular function and reduction in cardiac volumes. Reverse remodeling has become a primary objective in HF treatment. Guideline directed medical and device therapies has been proven to

in the reverse cardiac remodeling. Numerous studies have highlighted the role for RAAS blockade in promoting reverse remodeling in patients with HFrEF. The SOLVD and Val-HeFT studies have demonstrated the beneficial effects of ACEI and ARB on LVRR 44-47, respectively. Since the landmark of PARADIGM-HF study 4, the data support the link between ARNI and LVRR was growing. A meta-analysis including more than 10,000 population showed that ARNI has significant improvement over ACEI/ARB on the left LV dimensions and EF48. The EV ALUATE-HF and PROVE-HF trial both observed the reduction of multiple atrial and ventricular parameters of remodeling after initiation of ARNI 49,50. However, the relations between ARNI dosage and the LVRR are not addressed. In our study, we classified the patients into different dosage group and revealed that the under target novel dose remained effective on LVRR. Mechanistically, LVRR might be a part of the explanation of reduced readmission. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint The mediation effects of a variety of covariates on all-cause readmission was explored. The omnibus mediation effect analysis revealed that novel-dose S/V was negatively related to all-cause readmission through its effect on LVRR, which has been identified as a critical mediator accounting for approximately 20.3% of total effect. Meanwhile, our study demonstrated that the lower- and low-dose S/V decreased the risk of readmission by 55% (1-0.45) and 86% (1-0.14) as compared with the lowest-dose, respectively, whereas only on average 8% (1-0.92) and 2.2% (1-0.978) of the readmission risk were correspondingly reduced through the mediation of LVRR. Although LVRR had a significant negative indirect effect on the difference of novel-dose S/V in all-cause readmission, the relative effect caused by LVRR was similar between the lower- (or low-) and lowest-dose levels. It seemed that the novel-dose S/V might exert its effect through other undermined mechanisms. There were several limitations with our study. First, as a single-center retrospective study, our study performed internal validation for the prediction models, which still needed further external validation using an independent cohort. Second, VHD and COPD were marginally significant variables in our proposed Model 1, probably because the sample size was relatively small. Future studies with a larger sample size would be required to confirm the model.

Our study developed a clinically useful competing-risk nomogram for predicting All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint all-cause readmission in HF patients receiving novel-dose S/V therapy. The findings revealed that the novel-dose S/V , especially the lowest, significantly affected the hospital readmissions, in part mediated by LVRR. However, the clinical benefits of different novel-dose S/V and more potential novel mechanisms still required further in-depth examination. Supplementary Material The Supplementary Material for this article can be found online as additional file 1. Acknowledgments We would like to thank Lili Yang for her technical assistance in the data search and process. Sources of Funding This study was supported by the National Natural Science Foundation of China (No. 82000295) and Project for Young Talented Doctors of Xinqiao Hospital Affiliated to Army Medical University (No. 2022YQB001). Disclosures The authors declare that there is no conflict of interest. Authors' contributions All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint Zhexue Qin Xiaolin Luo and Xi Liu were involved in conception and design. Changchun Hou, Ling Chen, and Juhao Yang carried out the acquisition of data. Ning Sun, Enpu Yang, Zebi Wang, Yun Cui, Jing Zhong and Luyu Wang contributed to the follow up of the patients. Xinxin Hao and Changchun Hou performed the statistical analysis and drafted the manuscript. Xiaolin Luo and Zhichun Gao contributed to interpretation of data. All authors contributed to critical revision of the manuscript for important intellectual content. Xinxin Hao, Changchun Hou, Xi Liu and Zhexue Qin were involved in final revision of manuscript. All authors read and approved the final manuscript.

1. Gao S, Yin G, Xia Q, et al. Development and validation of a nomogram to predict the 180-day readmission risk for chronic heart failure: a multicenter prospective study. Frontiers in cardiovascular medicine 2021: 1033. 2. Virani SS, Alonso A, Benjami n EJ, et al. Heart disease and stroke statistics—2020 update: a report from the American Heart Association. Circulation 2020; 141(9): e139-e596. 3. Heidenreich P, Albert N, Allen L, et al. Forecasting the imp act of heart failure in the United States: a policy statement from the American Heart Association. Circ Heart Fail 2013; 6(3): 606-19. 4. McMurray JJV , Packer M, Desai AS, et al. Angiotensin–neprily sin inhibition versus enalapril in heart failure. N Engl J Med 2014; 371: 993-1004. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint 5. Desai AS, Claggett BL, Pack er M, et al. Influence of sacubitril/valsartan (LCZ696) on 30-day readmission after heart failure hospitalization. Journal of the American College of Cardiology 2016; 68(3): 241-8. 6. Mazza A, Townsend DM, Torin G, et al. The role of sacubitril /valsartan in the treatment of chronic heart failure with reduced ejection fraction in hypertensive patients with comorbidities: from clinical trials to real-world settings. Biomedicine Pharmacotherapy 2020; 130: 110596. 7. Giovinazzo S, Carmisciano L, Toma M, et al. Sacubitril/valsa rtan in real‐life European patients with heart failure and reduced ejection fraction: a systematic review and meta‐analysis. ESC heart failure 2021; 8(5): 3547-56. 8. Luo N, Fonarow GC, Lippmann SJ, et al. Early adoption of sac ubitril/valsartan for patients with heart failure with reduced ejection fraction: insights from Get With the Guidelines–Heart Failure (GWTG-HF). JACC: Heart Failure 2017; 5(4): 305-9. 9. Norberg H, Bergdahl E, Lindmark K. Eligibility of sacubitril –valsartan in a real‐ world heart failure population: a community‐based single‐centre study. ESC heart failure 2018; 5(2): 337-43. 10. Andries G, Yandrapalli S, Aronow WS. Benefit–risk review of different drug classes used in chronic heart failure. Expert opinion on drug safety 2019; 18(1): 37-49. 11. Nikolic M, Srejovic I, Jovic J, et al. Sacubitril/valsartan in Heart Failure and Beyond—From Molecular Mechanisms to Clinical Relevance. Reviews in Cardiovascular Medicine 2022. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint 12. Carnicelli AP, Li Z, Greiner MA, et al. Sacubitril/valsartan adherence and postdischarge outcomes among patients hospitalized for heart failure with reduced ejection fraction. Heart Failure 2021; 9(12): 876-86. 13. Pandey AC, Jer D, Kuo RS, et al. Novel doses of sacubitril/ valsartan in patients unable to tolerate traditional therapy: effects on N‐terminal pro B‐type natriuretic peptide levels. Clinical Cardiology 2021; 44(1): 85-90. 14. Choi H-M, Shin M-S. Angiotensin receptor-neprilysin inhibit or for the treatment of heart failure: a review of recent evidence. The Korean Journal of Internal Medicine 2020; 35(3): 498. 15. Paolini C, Mugnai G, Dalla Valle C, et al. Effects and clinical implications of sacubitril/valsartan on left ventricular reverse remodeling in patients affected by chronic heart failure: A 24-month follow-up. IJC Heart Vasculature 2021; 35: 100821. 16. Xie X, Yang M, Xie S, et al. Early Prediction of Left Ventricular Reverse Remodeling in First-Diagnosed Idiopathic Dilated Cardiomyopathy: A Comparison of Linear Model, Random Forest, and Extreme Gradient Boosting. Frontiers in cardiovascular medicine 2021; 8: 684004. 17. Almufleh A, Marbach J, Chih S, et al. Ejection fraction improvement and reverse remodeling achieved with Sacubitril/Valsartan in heart failure with reduced ejection fraction patients. American journal of cardiovascular disease 2017; 7(6): 108. 18. Martens P, Beliën H, Dupont M, Vandervoort P, Mullens W. The reverse remodeling response to sacubitril/valsartan therapy in heart failure with reduced ejection fraction. Cardiovascular therapeutics 2018; 36(4): e12435. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint 19. Wong M, Johnson G, Shabetai R, et al. Echocardiographic var iables as prognostic indicators and therapeutic monitors in chronic congestive heart failure. Veterans Affairs cooperative studies V-HeFT I and II. V-HeFT V A Cooperative Studies Group. Circulation 1993; 87(6 Suppl): VI65-70. 20. Yu C-M, Bleeker GB, Fung JW-H, et al. Left ventricular reverse remodeling but not clinical improvement predicts long-term survival after cardiac resynchronization therapy. Circulation 2005; 112(11): 1580-6. 21. Vasan RS, Larson MG, Benjamin EJ, Evans JC, Levy D. Left ve ntricular dilatation and the risk of congestive heart failure in people without myocardial infarction. New England Journal of Medicine 1997; 336(19): 1350-5. 22. Monzo L, Lanzillo C, Tota C, et al. Sacubitril/valsartan ef fect on left ventricular remodeling: the case of a super-responder. Current medical research and opinion 2019; 35(sup3): 3-6. 23. Correale M, Mallardi A, Tricarico L, et al. Remodelling is inversely proportional to left ventricular dimensions in a real-life population of patients with chronic heart failure after therapy with sacubitril/valsartan. Acta Cardiologica 2022; 77(5): 416-21. 24. Zhang Y , Pletcher MJ, Vittinghoff E, et al. Association Between Cumulative Low-Density Lipoprotein Cholesterol Exposure During Young Adulthood and Middle Age and Risk of Cardiovascular Events. JAMA Cardiology 2021; 6(12): 1406. 25. Wolbers M, Koller MT, Stel VS, et al. Competing risks analyses: objectives and approaches. European heart journal 2014; 35(42): 2936-41. 26. Kuo SZ, Lizaola B, Hayssen H, Lai JC. Beta-blockers and phy sical frailty in All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint patients with end-stage liver disease. World journal of gastroenterology 2018; 24(33): 3770. 27. Bursac Z, Gauss CH, Williams DK, Hosmer DW, medicine. Purposeful selection of variables in logistic regression. Source code for biology and medicine 2008; 3(1): 1-8. 28. Jiang S, Zhang J, Zhang L, et al. A novel nomogram based on log odds of positive lymph nodes to predict survival for non-metastatic gallbladder adenocarcinoma after surgery. Scientific Reports 2022; 12(1): 1-11. 29. Merlo M, Pyxaras SA, Pinamonti B, Barbati G, Di Lenarda A, Sinagra G. Prevalence and prognostic significance of left ventricular reverse remodeling in dilated cardiomyopathy receiving tailored medical treatment. Journal of the American College of Cardiology 2011; 57(13): 1468-76. 30. Kimura Y , Okumura T, Morimoto R, et al. A clinical score for predicting left ventricular reverse remodelling in patients with dilated cardiomyopathy. ESC Heart Failure 2021; 8(2): 1359-68. 31. Díez-López C, Salazar-Mendiguchía J, García-Romero E, et al . Clinical Determinants and Prognosis of Left Ventricular Reverse Remodelling in Non-Ischemic Dilated Cardiomyopathy. Journal of Cardiovascular Development and Disease 2022; 9(1): 20. 32. Yu Q, Fan Y , Wu X. General multiple mediation analysis with an application to explore racial disparities in breast cancer survival. Journal of Biometrics and Biostatistics 2014; 5(2): 1-9. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint 33. Rijnhart JJ, Lamp SJ, Valente MJ, MacKinnon DP, Twisk JW, Heymans MW. Mediation analysis methods used in observational research: a scoping review and recommendations. BMC medical research methodology 2021; 21(1): 1-17. 34. Berg DD, Braunwald E, DeV ore AD, et al. Efficacy and Safety of Sacubitril/Valsartan by Dose Level Achieved in the PIONEER-HF Trial. JACC Heart Fail 2020; 8(10): 834-43. 35. Giakoumis M, Sargsyan D, K ostis JB, et al. Readmission and mortality among heart failure patients with history of hypertension in a statewide database. The Journal of Clinical Hypertension 2020; 22(7): 1263-74. 36. Kaneko H, Suzuki S, Goto M, et al. Incidence and predictors of rehospitalization of acute heart failure patients. International heart journal 2015; 56(2): 219-25. 37. Ahmed R, Sawatari H, Deshpande SA, et al. HF-564-01 READMIS SION OUTCOMES IN PATIENTS WITH CARDIAC SARCOIDOSIS. Heart Rhythm 2022; 19(5): S15. 38. Bottle A, Aylin P, Bell D. Effect of the readmission primary diagnosis and time interval in heart failure patients: analysis of English administrative data. European journal of heart failure 2014; 16(8): 846-53. 39. Gulea C, Zakeri R, Quint JK. Impact of chronic obstructive pulmonary disease on readmission after hospitalization for acute heart failure: A nationally representative US cohort study. International journal of cardiology 2019; 290: 113-8. 40. Wideqvist M, Cui X, Magnusson C, Schaufelberger M, Fu M. Ho spital readmissions of patients with heart failure from real world: timing and associated risk All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint factors. ESC heart failure 2021; 8(2): 1388-97. 41. Sangprasert P, Palangrit S, Tiyoa N, Pattaraarchachai J. Effects of mindfulness-based health education practice on health behaviors and quality of life among hypertensive patients: A quasi-experimental research. Journal of Health Research 2019; 33(3): 186-96. 42. Verulava T, Jorbenadze R, Lordkipanidze A, Gongadze A, Tsverava M, Donjashvili M. Readmission after hospitalization for heart failure in elderly patients in Chapidze Emergency Cardiology Center, Georgia. Journal of Health Research 2021; 36(3): 575-83. 43. Whellan DJ, Stebbins A, Hernandez AF, et al. Dichotomous relationship between age and 30-day death or rehospitalization in heart failure patients admitted with acute decompensated heart failure: results from the ASCEND-HF trial. Journal of cardiac failure 2016; 22(6): 409-16. 44. Konstam MA, Kronenberg MW, Rousseau MF, et al. Effects of the angiotensin converting enzyme inhibitor enalapril on the long-term progression of left ventricular dilatation in patients with asymptomatic systolic dysfunction. SOLVD (Studies of Left Ventricular Dysfunction) Investigators. Circulation 1993; 88(5 Pt 1): 2277-83. 45. Greenberg B, Quinones MA, Koilpillai C, et al. Effects of l ong-term enalapril therapy on cardiac structure and function in patients with left ventricular dysfunction.

of the SOLVD echocardiography substudy. Circulation 1995; 91(10): 2573-81. 46. Wong M, Staszewsky L, Latini R, et al. Valsartan benefits left ventricular All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint structure and function in heart failure: Val-HeFT echocardiographic study. J Am Coll Cardiol 2002; 40(5): 970-5. 47. Wong M, Staszewsky L, Latini R, et al. Severity of left ven tricular remodeling defines outcomes and response to therapy in heart failure: Valsartan heart failure trial (Val-HeFT) echocardiographic data. J Am Coll Cardiol 2004; 43(11): 2022-7. 48. Wang Y , Zhou R, Lu C, Chen Q, Xu T, Li D. Effects of the Angiotensin-Receptor Neprilysin Inhibitor on Cardiac Reverse Remodeling: Meta-Analysis. J Am Heart Assoc 2019; 8(13): e012272. 49. Januzzi JL, Jr., Prescott MF, Butler J, et al. Association of Change in N-Terminal Pro-B-Type Natriuretic Peptide Following Initiation of Sacubitril-Valsartan Treatment With Cardiac Structure and Function in Patients With Heart Failure With Reduced Ejection Fraction. JAMA 2019; 322(11): 1085-95. 50. Desai AS, Solomon SD, Shah AM, et al. Effect of Sacubitril-Valsartan vs Enalapril on Aortic Stiffness in Patients With Heart Failure and Reduced Ejection Fraction: A Randomized Clinical Trial. JAMA 2019; 322(11): 1077-84. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint What is new? The study was the first attempt to explore the effects of the novel-dose Sacubitril/Valsartan (S/V) on all-cause readmission in patients with heart failure (HF) using the competing-risk analysis, as well as investigate the possible mediator role of left ventricular reverse remodeling (LVRR). A nomogram based on the competing-risk regression model was established to predict all-cause readmission of patients who taken novel-dose S/V . Moreover, our study was useful in gaining a better understanding of the relationship and mediating mechanism between novel-dose S/V , LVRR, all-cause readmission. What are the clinical implications? So far there was little data on the novel-dose S/V in a real world setting. The findings of our study would provide new insights on the effects of novel-dose S/V , as well as useful information for health decision-making. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint Table 1. Univariate and multivariate competing risk analysis of patients’ characteristics influencing primary outcome (all-cause readmission) based on training dataset Characteristics Univariable competing risk analysis Multivariable competing risk analysis HR (95%CI, P-value) HR (95%CI, P-value)a HR (95%CI, P-value)b HR (95%CI, P-value)c Gender (Male) 0.97 (0.48-1.94, p=0.930) - - - <55y Reference Reference Reference Reference Age 55-<65y 0.42 (0.14-1.22, p=0.110) 0.32 (0.11-0.96, p=0.04 1) 0.34 (0.12-0.95, p=0.040) 0.38 (0.13-1.12, p=0.080) 65-<75y 1.61 (0.73-3.55, p=0.240) 1.13 (0.50-2.55, p=0.770) 1.23 (0.54-2.83, p=0.620) 1.53 (0.71-3.29, p=0.270) 75+y 3.09 (1.21-7.84, p=0.018) 1.72 (0.55-5.42, p=0.350) 2 .42 (0.94-6.19, p=0.065) 2.95 (1.18-7.35, p=0.021) <18.5kg/m 2 Reference Reference - - BMI 18.5-<24kg/m 2 0.66 (0.18-2.39, p=0.530) 1.31 (0.30-5.76, p=0.720) - - 24-<28kg/m 2 0.23 (0.06-0.94, p=0.041) 0.48 (0.09-2.62, p=0.390) - - 28+kg/m 2 0.41 (0.10-1.60, p=0.200) 1.26 (0.26-6.16, p=0.770) - - DCM (Presence) 0.96 (0.41-2.25, p=0.920) - - - Hypertension (Presence) 1.13 (0.56-2.27, p=0.730) - - - CHD (Presence) 1.13 (0.45-2.83, p=0.800) - - - MI (Presence) 1.51 (0.37-6.27, p=0.570) - - - VHD (Presence) 2.80 (1.25-6.28, p=0.013) 2.35 (0.94-5.86, p=0 .068) 2.14 (0.89-5.17, p=0.090) - Bradycardia (Presence) 0.94 (0.28-3.17, p=0.920) - - - PACs (Presence) 2.12 (0.97-4.62, p=0.060) - - - PVCs (Presence) 2.95 (1.57-5.55, p=0.001) 1.77 (0.75-4.17, p= 0.190) - - VT (Presence) 4.38 (1.81-10.55, p=0.001) 2.93 (1.05-8.23, p=0 .041) 3.59 (1.33-9.69, p=0.012) 3.50 (1.32-9.32, p=0.012) COPD (Presence) 3.45 (1.27-9.38, p=0.015) 2.92 (0.91-9.42, p= 0.073) 2.70 (0.97-7.51, p=0.056) - DM (Presence) 2.28 (1.02-5.10, p=0.045) 3.22 (1.35-7.66, p=0. 008) 2.70 (1.17-6.22, p=0.020) 2.26 (1.05-4.86, p=0.038) SRI (Presence) 1.93 (0.31-11.86, p=0.480) - - - ≤ 50mg Reference Reference Reference Reference Dose 50-<100mg 0.38 (0.20-0.74, p=0.004) 0.51 (0.24-1.07, p=0 .077) 0.45 (0.23-0.89, p=0.021) 0.39 (0.20-0.80, p=0.009) 100+mg 0.14 (0.05-0.38, p<0.001) 0.24 (0.08-0.77, p=0.016 ) 0.21 (0.07-0.59, p=0.003) 0.16 (0.06-0.45, p<0.001) Note: a Variables from least absolute shrinkage and selection operator approach and Akaike’s Information Criterion, b Model 1, c Model 2. Abbreviations: BMI, body mass index; DCM, dilated cardiomyopathy; CHD, coronary heart disease; MI, myocar dial infarction; VHD, valvular heart disease; PACs, premature a trial contractions; PVCs, premature ventricular contractions; VT, ventricular tachycardia; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; SRI, severe renal impairment; HR, hazard ratio; CI, confidence interval. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint Figure 1. The flowchart of patients’ selection All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint Figure 2. Cumulative incidence curves for all-cause readmission and death stratified by novel-dose Sacubitril/Valsartan on the basis of all dataset. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint Figure 3. Nomogram for predicting 30-day, 3-month and 6-month cumulative incidence of patients with heart failure (in accordance with Model 1 developed by training dataset). Abbreviations: VHD, valvular heart disease; VT, ventricular tachycardia; COPD, chronic obstructive pulmonary disease; DM, diabetes mell All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint Figure 4. Estimated mediation effects (with 95% confidence intervals) of left ventricular reverse remodeling on the association between novel -dose Sacubitril/Valsartan (S/V) and hospital readmission of patients with heart failure. Note: Predictor variable (S/V dosage) was continuous and categorical in omnibus and relative mediation effect analysis. Black line with error bars represented the 95% confidence intervals corresponding to quantile estimators from the bootstrap. Additionally, total effect = indirect effect + direct effect (see Figure S1). # Lowest -dose S/V was the reference. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted February 10, 2023. ; https://doi.org/10.1101/2023.02.08.23285680doi: medRxiv preprint