Left ventricular assist devices (LVADs) are an established therapy for patients with advanced chronic heart failure (ACHF), significantly improving survival and quality of life as a bridge to transplantation or as a destination therapy [1]. According to recent registry data, the number of LVAD implantations has steadily increased, with more than 15,000 patients supported [2]. However, the success of LVAD therapy is often compromised by the early morbidity and mortality caused by right ventricular failure (RVF) [3]. Severe RVF exhibits an incidence as high as 50% after LVAD implantation, resulting in perioperative mortality rates of 19% to 43% and driving significant complications, including end-organ dysfunction and the need for prolonged intensive care and hospitalization [4, 5].

The most severe form of RVF necessitates the implantation of a right ventricular assist device (RVAD). Meanwhile, the pathophysiology of post-LVAD RVF is multifactorial. The abrupt increase in venous return following LVAD initiation increases RV preload, while the leftward shift of the interventricular septum can alter RV geometry and impair contractility. These changes can overwhelm a ventricle that has already been compromised by chronic heart failure [6]. Thus, recognizing this risk, numerous studies have sought to identify preoperative predictors of RVF to improve the selection of candidates for isolated LVAD therapy and to identify those who might benefit from a planned biventricular ventricular assist device (BiVAD) strategy. These efforts have led to the development of several clinical risk scores, such as the Michigan michigan right ventricular failure risk score (RVFRS), the CRITT score, and the EUROMACS-right sided heart failure (RHF) score [7]. However, these scores have often yielded inconsistent predictors, and external validation studies have demonstrated only modest predictive power, limiting their widespread clinical application [8, 9].

Collectively, this systematic review and meta-analysis were conducted to aggregate the current evidence given the profound impact of post-LVAD and RVAD requirements and the ongoing challenges in risk stratification. Finally, this review aimed to identify the most robust and consistent preoperative predictors of postoperative RVAD requirement in patients undergoing durable, continuous-flow LVAD (CF-LVAD) implantation.

RVF remains the Achilles’ heel of contemporary durable or temporary LVAD therapy and is strongly associated with peri-operative morbidity, prolonged resource use, and excess mortality [32]. In its severest form, RVF mandates RVAD support, a scenario repeatedly linked with worse early and late outcomes compared with isolated LVAD [6]. Patients requiring RVAD after LVAD implantation experience higher peri-operative mortality, more bleeding and neurological complications, and worse long-term survival compared with those managed with isolated LVAD support [33], underscoring the critical importance of early risk recognition and proactive management.

Several prior systematic reviews and meta-analyses have sought to identify predictors of right ventricular dysfunction after LVAD implantation. However, most have considered RVF as a broad and heterogeneous entity, often combining transient inotrope dependence, prolonged nitric oxide therapy, and RVAD implantation into a single outcome. Bellavia et al. [34] synthesized data from 36 studies and identified haemodynamic and echocardiographic predictors of RVF, including elevated central venous pressure, lower right ventricular stroke work index, and increased RV/LV diameter ratio, but RVAD implantation was not assessed separately as a hard endpoint. Frankfurter et al. [7] instead focused on the performance of dedicated RVF risk scores, such as the Michigan, CRITT, and EUROMACS-RHF models, and demonstrated their limited discriminatory ability when externally validated, with most models showing modest predictive value only. Other reviews have concentrated on specific preoperative domains. For example, Chai et al. [35] examined echocardiographic markers, reporting that impaired RV function and significant tricuspid regurgitation were associated with post-LVAD RVF.

Importantly, none of these prior studies have synthesized predictors specifically for RVAD implantation, the most clinically decisive manifestation of RVF. This distinction matters as patients requiring RVAD after LVAD implantation have a three- to fourfold higher early mortality and significantly worse long-term outcomes compared with those supported with isolated LVAD [36]. By treating RVAD implantation as a discrete endpoint, our study moves beyond the heterogeneity of prior RVF definitions and provides a more actionable evidence base for surgical decision-making.

The present meta-analysis therefore builds on the existing literature by systematically evaluating preoperative predictors across all major domains, clinical and demographic factors, haemodynamic, echocardiography, and laboratory parameters, and quantifying their association with the subsequent need for RVAD support. This integrated approach allows for a more holistic risk profile and provides cardiac surgeons with practical, clinically relevant predictors that can guide both preoperative optimization and intraoperative strategy. Synthesizing the evidence from 20 studies, this comprehensive meta-analysis identified a robust profile of preoperative risk factors for RVAD requirement after LVAD implantation. The results clearly indicate that the need for RVAD is driven by a syndrome of global illness severity, characterized by advanced clinical acuity, profound haemodynamic compromise, and significant end-organ dysfunction, rather than any single isolated parameter.

Our analysis reaffirms the central role of haemodynamic assessment in risk stratification. An elevated CVP is a consistent and powerful predictor, reflecting the degree of systemic venous congestion that the RV must overcome. However, isolated CVP can be misleading. Indices that provide broader context, such as the CVP/PCWP ratio, which reflects the balance of right- versus left-sided filling pressures, and the RVSWI, which quantifies the effective work of the RV, were also highly significant in our analysis. This emphasizes that a dynamic assessment of RV function relative to its loading conditions is more informative than static pressure measurements alone.

Recent evidence further supports the prognostic importance of the right atrial pressure to pulmonary capillary wedge pressure (RAP/PCWP) ratio in the LVAD population. A 2025 study by Keskin et al. [37] demonstrated that an elevated RAP/PCWP ratio strongly predicts in-hospital mortality following LVAD implantation. In that cohort, patients with a high RAP/PCWP ratio ($\ge$0.47) had a markedly higher in-hospital mortality rate compared with those below this threshold (46% vs. 10%), and the RAP/PCWP ratio showed excellent discriminatory ability (AUC 0.829). Importantly, RAP/PCWP remained an independent predictor of early mortality on multivariate analysis (OR 3.48 per 0.1 increase), outperforming other coupling indices such as PAPi and TAPSE/PASP. Although our review could not quantitatively pool RAP/PCWP due to heterogeneous reporting across studies, our findings align with this physiologic relationship and further underscore the central role of venous congestion and RV–pulmonary arterial uncoupling in determining early postoperative outcomes. Integration of RAP/PCWP into standardized preoperative evaluation may enhance risk stratification, particularly when used alongside established clinical predictors such as INTERMACS profile.

A key theme emerging from this analysis is the paramount importance of systemic and end-organ function. The strong association with high-acuity INTERMACS profiles and the need for preoperative life support (mechanical ventilation, IABP, ECMO) underscores that patients in cardiogenic shock are at the highest risk. This is further supported by the laboratory findings. Elevated bilirubin and INR, and low albumin are not just markers of hepatic dysfunction; they are powerful surrogates for the systemic effects of chronic right-sided congestion and low cardiac output. These findings validate the components of widely used prognostic scores like the MELD score, which Yost et al. [30] found to be predictive of RVF.

The challenge of risk prediction is highlighted by the inconsistent performance of dedicated RVF risk scores. While scores developed in one cohort may perform well internally, studies by Pettinari et al. [26] and Rivas-Lasarte et al. [8] demonstrate that their predictive power often diminishes upon external validation. Rivas-Lasarte et al. [8] found the EUROMACS-RHF score had an AUC of only 0.64 in a large external cohort, concluding its clinical utility remains to be determined. This suggests that static scoring systems may fail to capture the complex, dynamic pathophysiology of individual patients and underscores the need for a more integrated, clinician-driven assessment.

Previous systematic reviews have largely focused on RVF as a broader construct. Bellavia et al. [34] identified elevated CVP, impaired RVSWI, and RV dilation as predictors of RVF, but did not differentiate between mild, moderate, or severe failure, nor did they isolate RVAD implantation as an endpoint. Frankfurter et al. [7] reviewed existing risk scores (e.g., Michigan, CRITT, EUROMACS-RHF) and demonstrated their limited discriminatory ability when externally validated. More recently, Rodenas-Alesina et al. [38] and Wang et al. [39] provided narrative reviews of RVF incidence, pathophysiology, and prevention strategies, but again encompassed heterogeneous outcomes. Importantly, Reid et al. [36] specifically showed that patients requiring RVAD after LVAD implantation had more than a threefold higher 30-day mortality (31.9% vs 6.7%) compared with isolated LVAD recipients, reinforcing the prognostic weight of this endpoint. Our study advances the field by being the first meta-analysis to quantitatively synthesize preoperative predictors of RVAD implantation as a distinct and clinically decisive outcome.

The findings of this meta-analysis have direct clinical implications. Patient selection for LVAD therapy must involve a comprehensive, multi-parameter risk assessment. A patient presenting with multiple risk factors identified in this analysis, such as an INTERMACS Profile 1, a history of prior cardiac surgery, need for preoperative IABP, an elevated CVP, low TAPSE, and elevated bilirubin, should be considered at very high risk for requiring postoperative RVAD support. In these patients, the clinical team should consider strategies to mitigate risk, such as intensive preoperative medical optimization to decongest the patient and improve end-organ function. Furthermore, there should be a low threshold for a planned biventricular support strategy, either with a temporary RVAD or consideration of a primary BiVAD or total artificial heart. As shown by Ahmed et al. [14], a sequential, rescue RVAD implantation after a period of deterioration is associated with significantly worse survival and higher adverse event rates compared to a concurrent, planned BiVAD implantation. Therefore, anticipating the need for RV support and acting proactively is critical.

This meta-analysis is subject to several limitations inherent to the source literature. The majority of the included studies were retrospective, and significant heterogeneity existed in patient populations, definitions of RVF, and eras of LVAD technology. The decision to implant an RVAD is also subject to institutional practice patterns, which introduces variability. While we focused on the hard endpoint of RVAD implantation, this does not capture the full spectrum of RVF, which includes patients managed with prolonged inotropes who may also have poor outcomes.

Furthermore, a key limitation of this meta-analysis is the probable inclusion of overlapping patient cohorts from large, registry-based studies. Specifically, major studies such as Kiernan et al. [13] and Ahmed et al. [14] both draw from the INTERMACS registry during overlapping time periods. This can potentially skew the summary effect estimates and lead to an overestimation of the precision of our findings. While it was not possible to identify and exclude overlapping patients from the primary data, this inherent limitation should be considered when interpreting the results of this analysis.

Finally, the analysis is limited by the variables reported in the primary studies, and the potential for publication bias cannot be excluded.

In conclusion, this systematic review and meta-analysis identifies a clear and consistent profile of preoperative risk for RVAD requirement after LVAD implantation. This profile is characterized by high clinical acuity, the need for preoperative mechanical support, haemodynamic evidence of severe RV dysfunction, and laboratory signs of systemic congestion and end-organ injury. No single variable is sufficient for risk prediction; rather, a comprehensive, multi-parameter assessment is essential to guide patient selection, facilitate informed decision-making, and plan perioperative strategies to improve the outcomes of this high-risk patient population.

The data are available from the corresponding author after reasonable demand.

MC conceptualised the study and drafted the manuscript. MC, AD, CC, IZ, AS, AP, EK, MO, MMV, NM, and HSA performed data acquisition, data extraction, and data analysis. SGR contributed to the study conceptualisation and provided senior review. AS, AP, EK, MO, MMV, NM, HSA, AD, CC, IZ, and SGR contributed to critical revision of the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work in accordance with the ICMJE authorship criteria.

Not applicable.

Not applicable.

This research received no external funding.

The authors declare no conflict of interest. Shahzad G. Raja is serving as one of the Editorial Board members of this journal. We declare that Shahzad G. Raja had no involvement in the peer review of this article and has no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to Tadahisa Sugiura.

Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.31083/HSF50915.