Risk Factors for Hospital Readmission Post-Transcatheter Aortic Valve Implantation in the Contemporary Era: A Systematic Review

Background Despite transcatheter aortic valve implantation (TAVI) becoming a widely accepted therapeutic option for the management of aortic stenosis, post-procedure readmission rates remain high. Rehospitalization is associated with negative patient outcomes, as well as increased healthcare costs, and has therefore been identified as an important target for quality improvement. Strategies to reduce the post-TAVI readmission rate are needed but require the identification of patients at high risk for rehospitalization. Our systematic review aims to identify predictors of post-procedure readmission in patients eligible for TAVI. Methods We conducted a comprehensive search of the MEDLINE, Embase, and Cochrane Central Register of Controlled Trials (CENTRAL) databases for the time period from 2015 to the present for articles evaluating risk factors for rehospitalization post-TAVI with a follow-up period of at least 30 days in adults age ≥ 70 years with aortic stenosis. The quality of included studies was evaluated using the Newcastle-Ottawa Scale. We present the results as a qualitative narrative review. Results We identified 49 studies involving 828,528 patients. Post-TAVI readmission is frequent, and rates vary (14.9% to 54.3% at 1 year). The most-frequent predictors identified for both 30-day and 1-year post-TAVI readmission are atrial fibrillation, lung disease, renal disease, diabetes mellitus, in-hospital life-threatening bleeding, and non-femoral access. Conclusions This systematic review identifies the most-common predictors for 30-day and 1-year readmission post-TAVI, including comorbidities and potentially modifiable procedural approaches and complications. These predictors can be used to identify patients at high-risk for readmission who are most likely to benefit from increased support and follow-up post-TAVI.


R ESUM E
Contexte : Bien que l'implantation valvulaire aortique par cath eter (IVAC) soit maintenant une option th erapeutique largement accept ee pour la prise en charge de la st enose aortique, les taux de r eadmission des patients après cette intervention demeurent elev es. La r ehospitalisation est associ ee à des r esultats de sant e d efavorables pour les patients ainsi qu'à des coûts de soins de sant e plus elev es, ce qui en fait une cible importante pour l'am elioration de la qualit e des The prevalence of aortic stenosis, the most common form of valvular heart disease, is expected to grow significantly as the population ages, representing a major challenge to healthcare systems around the world. 1 In Canada, transcatheter aortic valve implantation (TAVI) has become the standard of care for inoperable or high-surgical-risk patients with symptomatic severe aortic stenosis, 1 with more recent evidence supporting use of TAVI as a reasonable alternative in intermediate-and low-risk patients. 2,3 Despite the dramatic improvements in the safety and efficacy of TAVI, post-procedure early and late readmissions remain a concern. Recent studies have reported that up to 22.4% and 54.3% of patients are readmitted at 30 days and 1 year after TAVI, respectively. 4,5 Preventing avoidable rehospitalizations has emerged as an important quality-improvement initiative, with policies such as the Hospital Readmissions Reduction Program in the US offering financial incentives to minimize unnecessary hospitalizations. 6 In TAVI, both cardiac and noncardiac causes have been identified as important contributors. 5,7 Rehospitalization has been associated with an increase in all-cause mortality 8 and healthcare costs. 9 Reducing readmission rates requires a rigorous means of identifying patients at high risk for readmission. Accordingly, in this systematic review, we aim to summarize the current evidence on predictors of postprocedure readmission in patients with aortic stenosis eligible for TAVI, with the goal of identifying pre-procedural comorbidities, as well as potentially modifiable periprocedural predictors. This identification in turn would facilitate interventions to reduce readmissions, which would need to be independently tested.

Eligibility criteria
We included randomized controlled trials (RCTs), observational cohort studies, and retrospective case-control studies of adults (age !70 years) with aortic stenosis eligible for TAVI. We excluded studies in which the primary intervention was surgical aortic valve replacement or valvein-valve TAVI.
To be included, a study had to evaluate one or more risk factors for rehospitalization post-TAVI, with a follow-up period of at least 30 days. Our primary outcome of interest is risk factors for early (at < 30 days) and late (at !30 days) all-cause readmission post-TAVI, and secondary outcomes include cardiac vs noncardiac causes of both early and late hospital readmission rates, as well as all-cause mortality in readmitted patients.
We excluded case reports, conference abstracts, and review articles. We included only studies written in the English language, and only those published in 2015 or later, as this period saw the expansion of TAVI to intermediate-and lowrisk patients.

Information sources
The databases, platforms, and coverage were MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials (CENTRAL), from 2015 to present.

Search strategy
The search strategy was developed by a medical librarian in consultation with a study investigator (R.P.). An exploratory literature review was conducted to find relevant articles, to mine key words. The search strategy was formed based on the following key words: "transcatheter aortic valve implantation," "transcatheter aortic valve replacement," "readmission," and "rehospitalization." The following medical subject heading (MeSH) terms were exploded: "transcatheter aortic valve replacement," "heart valve prosthesis implantation," "patient readmission," and "hospital readmission." The full search strategies are available in Supplemental Table S1.

Selection process
After de-duplication of search results, 2 independent investigators (R.P. and R.M.) used predefined inclusion and exclusion criteria to review titles and abstracts. Afterward, the full texts of eligible studies were reviewed for inclusion by 2 independent investigators (R.P. and M.R.). Throughout the screening process, all disagreements were resolved via consensus or via a third investigator (H.W.) when consensus could not be reached. De-duplication, screening, and data abstraction were performed using the Web-based Covidence software (Veritas Health Innovation, Melbourne, Australia). 11

Data collection process
We prepared a data extraction form on Covidence, and 2 independent investigators (R.P. and M.R.) carried out data patients at high risk for rehospitalization. Our systematic review aims to identify predictors of post-procedure readmission in patients eligible for TAVI. Methods: We conducted a comprehensive search of the MEDLINE, Embase, and Cochrane Central Register of Controlled Trials (CENTRAL) databases for the time period from 2015 to the present for articles evaluating risk factors for rehospitalization post-TAVI with a follow-up period of at least 30 days in adults age ! 70 years with aortic stenosis. The quality of included studies was evaluated using the Newcastle-Ottawa Scale. We present the results as a qualitative narrative review. Results: We identified 49 studies involving 828,528 patients. Post-TAVI readmission is frequent, and rates vary (14.9% to 54.3% at 1 year). The most-frequent predictors identified for both 30-day and 1year post-TAVI readmission are atrial fibrillation, lung disease, renal disease, diabetes mellitus, in-hospital life-threatening bleeding, and non-femoral access.
Conclusions: This systematic review identifies the most-common predictors for 30-day and 1-year readmission post-TAVI, including comorbidities and potentially modifiable procedural approaches and complications. These predictors can be used to identify patients at high-risk for readmission who are most likely to benefit from increased support and follow-up post-TAVI.
soins. Des strat egies de r eduction des taux de r eadmission après l'IVAC sont n ecessaires, mais elles exigent de rep erer les patients qui pr esentent des risques de r ehospitalisation plus elev es. Notre revue syst ematique vise à cerner les facteurs pr edictifs de r ehospitalisation après l'intervention chez les patients admissibles à une IVAC. extraction from eligible studies. The 2 investigators met to identify and resolve discrepancies through discussion. Please see Supplemental Table S2 for data elements.

Variables
In studies for which both univariate and multivariable analyses were performed to determine predictors for readmission, we included only predictors that were statistically significant in the multivariable analysis. We categorized predictors for early and late hospital readmission into 7 categories: (i) demographic factor; (ii) clinical characteristic; (iii) cardiac comorbidity or previous intervention; (iv) medical comorbidity; (v) laboratory marker; (vi) procedural characteristic; and (vii) procedural complication. Cardiac causes were those pathologies that are intrinsic to the heart, such as arrhythmia or heart failure, whereas noncardiac causes were all other medical comorbidities, such as lung or renal pathology. Procedural characteristics and complications were considered to be potentially modifiable. The Elixhauser Comorbidity Index quantifies a patient's comorbidity based on 30 International Classification of Diseases (ICD), ninth revision, Clinical Modification and ICD, tenth revision diagnosis codes, which are weighted based on the association of each comorbidity with death, to produce a summary index. 12 Study quality assessment Two authors (R.P. and M.R) independently evaluated study quality using the Newcastle-Ottawa Scale, and discrepancies were resolved by discussion. The quality assessment is presented in Supplemental Table S3.

Statistical analysis
We extracted the absolute number of events when it was available, or calculated it based on the statistical measures reported. Given the substantial heterogeneity among the included studies in the statistical measures used to report outcome data, a meta-analysis was not performed.

Results
Study selection (flow of studies) Figure 1 illustrates the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram for study selection. Database searching retrieved 1566 records, of which 413 were duplicates. We excluded 919 records after title and abstract screening, and assessed the full texts of 231 records. Ultimately, we included 49 studies in the review.

Study characteristics
Study characteristics are summarized in Table 1. The design of all included studies was either cohort (45) or RCT (4), and all were published between 2015 and 2022. The study period for included studies ranged from 1 to 10 years. Seventeen of the included studies were reports on early hospitalizations, 21 were reports on late hospitalizations, and 11 were reports on both. The definitions of early and late followup periods were similar across studies. Across all included studies, 828,528 patients underwent TAVI, and sample sizes ranged from 63 to 171,361 patients. The percentage of female patients ranged from 27% to 69.8% across studies. Study patients were older adults (age 74-84 years). Only 18 studies reported baseline surgical risk. Most patients were at intermediate or significant risk for in-hospital mortality and morbidity. The Society of Thoracic Surgeons (STS) score, the European System for Cardiac Operative Risk Evaluation (EuroSCORE), and the EuroSCORE II ranged from 3.2% to 10%, 13.3% to 23%, and 0.07% to 5.9%, respectively. Of the 15 studies reporting access site, the transfemoral approach was most frequently used, with the percentage of patients having a transfemoral TAVI ranging from 45% to 97.4%. The type of valve used was reported by 13 studies. Balloonexpandable valves were most common (6% to 100%), followed by self-expandable valves (0% to 90%).

Readmission rates
Only 17 of the 49 included studies reported readmission rate data ( Table 2). Readmission rates were divided into 2 categoriesdearly and latedand 3 subcategoriesdall-cause, cardiac cause, and noncardiac cause. Post-TAVI readmission rates are variable but high. All-cause early readmission rates range from 3.5% to 22.4%, and all-cause late readmission rates range from 14.9% to 54.3%. Despite the variability in rehospitalization rates, the causes of hospital readmission are consistent. Noncardiac causes are more common than cardiac causes for readmission. Cardiac causes were responsible for 31.8% to 44.6% of early readmissions, and 32% to 53.7% of late readmissions. Noncardiac causes were responsible for 55.4% to 68.2% of early readmissions, and 46.3% to 68% of late readmissions. The most common noncardiac causes are respiratory and bleeding events, and infection. 5,8,13,14 The most common cardiac causes for readmission are heart failure and arrhythmia. 5,8,[13][14][15]

Mortality rates
Eight studies (35,552 patients) reported the all-cause mortality rate, which ranged from 1.1% to 44.4%, with the mean follow-up duration ranging from 1 to 32 months (Table 3). Four studies (14,794 patients) reported mortality rate in readmitted patients, and 2 studies (1859 patients) reported mortality rate in non-readmitted patients. The mortality rate in readmitted patients was higher (30.2%), compared with that in patients without hospital readmission (19.2%). Multiple studies reported a statistically significant increase in all-cause mortality in readmitted patients, compared with that in patients who were not readmitted. 5,[16][17][18][19] Mortality risk also correlated with number of hospital readmissions and time of readmission. Patients with multiple hospital readmissions or a late readmission were at higher risk for mortality compared to patients with a single or early readmission. 16,17 Predictors for early hospital readmission Of the 49 included studies, 18 assessed risk factors for early hospital readmission post-TAVI. In Figure 2, we summarize the most frequently identified predictors for early readmission, and a list of all identified predictors can be found in Supplemental Table S4. A total of 37 unique predictors were identified among the 18 studies. Across the 37 predictors, the most-reported risk factor was atrial fibrillation/flutter (5 studies; 55,085 patients). The next most common predictors (4 studies each) were discharge to a skilled nursing facility (68,282 patients), chronic lung disease (39,595 patients), and non-femoral access (38,620 patients). The next most reported risk factors were as follows: diabetes mellitus (3 studies; 34,615 patients); in-hospital life-threatening bleeding, vascular complication, or transfusion (3 studies; 66,543 patients); and more than 4 Elixhauser comorbidities (2 studies; 12,562 patients). Taken together, renal pathologies (acute kidney injury, chronic kidney disease, and renal failure) were identified by 10 studies (96,625 patients) as risk factors for readmission post-TAVI. In the demographic factors category, the most identified risk factor for readmission was increasing age (3 studies; 74,814 patients). For laboratory markers, 2 studies (1434 patients) found an association between hemoglobin level and rehospitalization risk. A low hemoglobin level at discharge increased risk for readmission, whereas a high preoperative hemoglobin level was protective against readmission.

Predictors for late hospital readmission
Of the 49 included studies, 34 identified risk factors for late hospital readmission post-TAVI. Figure 2 depicts the most frequently identified predictors for late readmission, and a list of all identified predictors can be found in Supplemental Table S5. Across the 34 studies, a total of 59 unique predictors were reported. The predictor most frequently identified by the included studies was atrial fibrillation (11 studies; 157,601 patients)dboth preexisting and new-onset atrial fibrillation. Other cardiac risk factors included heart failure (5 studies; 79,563 patients) and post-TAVI aortic and pre-TAVI mitral regurgitation (4 studies; 37,408 patients). Among the medical comorbidities category, the most common predictors were those related to lung pathology (including chronic obstructive pulmonary disease, interstitial lung disease, pulmonary hypertension and chronic lung disease   male sex, anemia, nonfemoral access, and discharge to a skilled nursing facility remained predictors for readmission.

Discussion
The purpose of this systematic review was to identify predictors of both early and late readmission in patients post-TAVI. Our main findings are as follows: (i) post-TAVI readmission is common and has both cardiac and noncardiac causes; (ii) a possible association exists between post-TAVI readmission and subsequent mortality, requiring further investigation; and (iii) several comorbidities and procedural variables predict post-TAVI readmission with significant overlap between predictors for early and late readmission. These findings highlight the importance of identifying patients at elevated risk for post-TAVI readmission who may benefit from additional management or follow-up post-procedure; however, the current lack of tools available to identify patients most at risk for readmission may make this challenging. The ability to stratify patients according to readmission risk is necessary because initiatives aimed at lowering rehospitalization rates may be more successful when targeted toward those patients who are most at risk. 20 Similar work in this area has been done by Goldsweig et al. (2020) 21 and Li et al. (2021) 22 Goldsweig et al. 21 conducted a rapid review to identify predictors for post-TAVI readmission. They reported on 10 studies and 24 unique predictors. In their systematic review, Li et al. 22 also reported on 10 studies and found a total of 15 unique predictors. Both of these studies identify the following predictors as being the most common: non-transfemoral valve delivery, kidney disease, lung disease, major bleeding, atrial fibrillation, and left ventricular systolic dysfunction. Our current work builds on the findings of both these previous reviews. We performed a more comprehensive systematic search encompassing studies published to date. In comparison, Li et al. 22 searched only the literature up to 2018 and found 13 studies. As a result, compared to both these previous reviews, we were able to identify substantially more relevant articles at 49. In addition, our search identified several novel predictors commonly reported in the literatured heart failure, post-procedure aortic and pre-procedure mitral regurgitation, diabetes mellitus, peripheral vascular disease, frailty, and discharge to a skilled nursing facility. We also report on factors found to be protective against readmission, such as a high preoperative hemoglobin level and the absence of anemia prior to TAVI. Lastly, we present mortality data demonstrating an increased risk for mortality in TAVI patients with a post-procedure readmission, compared to the risk in those without one. This association may simply reflect the fact that the comorbidities that drive mortality and readmission are likely the same; moreover, this finding is limited, as only a small number of all included studies present mortality data.
The evidence is conflicting for several predictors, such as female sex, heart failure, valve type, pacemaker implantation, left bundle branch block, age, coronary artery disease, and post-procedure aortic regurgitation. Dodson et al. (2017) 23 found female sex to be protective against post-procedure readmission, whereas Doshi et al. (2019) 24 found female sex to increase the likelihood of post-procedure readmission. Similarly, multiple studies have shown that nonagenarians have a risk level for post-procedure readmission similar to that of a younger population. Moreover, we found one study that revealed an association of balloon-expanding valves with a decrease in readmission risk, compared with the risk with selfexpanding valves, but this association was not found in the Comparison of Balloon-Expandable vs Self-expandable Valves in Patients Undergoing Transcatheter Aortic Valve Replacement (CHOICE) 25 and Repositionable Percutaneous Replacement of Stenotic Aortic Valve Through Implantation of Lotus Valve SystemdRandomized Clinical Evaluation (REPRISE III) 26 RCTs. In addition, we found evidence for increased readmission risk in patients requiring pacemaker implantation, but once more, this association was not seen in the REPRISE III RCT. 26 Given the large number of studies, we believe the value of our work is in summarizing the breadth of the literature and identifying those risk factors for which the findings show broad consistency.
Our findings have significant implications with respect to post-TAVI readmission rates. By identifying common predictors for early and late post-procedure readmission, our work can inform risk stratification. This benefit will aid both primary care physicians and interdisciplinary TAVI teams, allowing them to follow high-risk patients more closely, which may reduce rehospitalization risk. Moreover, future projects may focus on developing and evaluating interventions to reduce readmission rates among the high-risk cohort. Decreased post-TAVI readmission rates may benefit both patientsdthrough reduced risk for mortality and morbiditydand the healthcare systemdthrough reduced costs from rehospitalizations. 8,9 In the context of heart failure, postdischarge remote monitoring has proven effective at identifying early health deterioration, allowing for prompt intervention to prevent rehospitalization. 27 As part of our future research, we plan to apply this strategy to TAVI patients at high risk for readmission at our centre.

Limitations
Our study findings should be evaluated in the context of several limitations. First, we did not explore studies published prior to 2015. Although this strategy could have resulted in oversight of relevant studies, we believe this to be unlikely given that our initial literature search found that the earliest relevant paper was published in 2015. Additionally, by restricting our search to this timeframe, our results are more in line with current TAVI practices and patient populations, enhancing their generalizability in the modern era. And second, our systematic review study design is limited in that it does not provide a summary estimate for our primary or secondary outcomes, as would have been possible with a metaanalysis design. We deemed a qualitative review to be more appropriate given the substantial heterogeneity among studies with regard to populations and reporting of outcome data.

Conclusion
This review demonstrates that 30-day and 1-year readmission rates post-TAVI are high, and that increased mortality is associated with readmission. In addition, it identifies the most common risk factors for both 30-day and 1-year hospital readmission. The results of this study can be used to identify patients at high-risk for hospital readmission post-TAVI. Tailored strategies can be developed accordingly to reduce readmission rates in high-risk cohorts.

Funding Sources
H.W. is supported by a Canada Research Chair in Structural Heart Disease Policy and Outcomes.

Disclosures
The authors have no conflicts of interest to disclose.