Interhospital Variability in Infections After Durable Left Ventricular Assist Device: Analysis of The Society of Thoracic Surgeons Intermacs Database

S. Zhou¹, H. Hou¹, T. M.. Braun², K. D.. Aaronson¹, L. Cabrera³, C. E.. Chenoweth¹, A. Hider⁴, P. Malani¹, M. J.. Pienta¹, R. Cantor⁵, J. Kirklin⁵, J. Wolverton⁶, F. D.. Pagani⁷, D. Likosky^1
1Michigan Medicine, Ann Arbor, Michigan ²School of Public Health, University of Michigan, Ann Arbor, Michigan ³Michigan Society of Thoracic and Cardiovascular Surgeons Quality Coll, Ann Arbor, Michigan ⁴University of Colorado, Dearborn, Michigan ⁵University of Alabama in Birmingham, Birmingham, Alabama ⁶University of Michigan, Ann Arbor, Michigan ⁷University of Michigan Hospital, Ann Arbor, Michigan

Purpose: Inter-hospital variability in rates of infection following durable left ventricular assist device (dLVAD) implant in the contemporary era is understudied, despite infections being the leading cause of morbidity and mortality. This national study used a unique dataset of merged clinical and administrative data to investigate institutional determinants of dLVAD-associated infections.

Methods: Data on 10,632 dLVAD implants entered into The Society of Thoracic Surgeons Intermacs National Database were evaluated at 170 hospitals from 2018-22. STS Intermacs data were merged with the 2021 American Hospital Association “AHA” survey to assess hospital-level determinants (e.g., academic affiliation). The primary outcome was the incidence of 90-day infections post-implant per 100 person-months. Terciles based upon the rates of hospital-level infection were created for descriptive statistics (T1: low; T2: middle; T3: high). Nested Cox proportional hazards models were performed to evaluate both the overall as well as the incremental variability in interhospital infection rates across 7 models: M1: demographics; M2: M1 + risk factors (BMI, Intermacs profile, NYHA class, previous cardiac surgery, comorbidities); M3: M2 + hemodynamics, laboratory values; M4: M3 + device strategy; M5: M4 + pre-implantation events and interventions; M6: M5 + operative duration; M7: M6 + hospital characteristics.

Results: A total of 2,333 (22%) patients developed an infection within 90 days post-implant. Overall, the median (IQR) age was 60 (50-68yrs), 2,309 (22%) were female, 3,295 (31%) were of Black race and 695 (6.5%) were Hispanic. Infection rates varied by hospital (median, IQR): 8.4 (3.8-13.2). Patients in high (T3) versus low (T1) infection hospitals were more likely to have NYHA Class IV heart failure [3,938 (88%) versus 1,984 (78%)], baseline pulmonary disease [898 (20%) versus 297 (12%)], require a temporary mechanical circulatory support (MCS) device pre-implant [687 (15.4%) versus 270 (11%)], and receive a dLVAD for destination therapy [3,350 (75%) versus 1,790 (70%)], all p< 0.05. Hospitals in the T3 versus T1 were more often 500+ beds [3,952 (89%) versus 2,149 (85%)], academic [3,808 (85%) versus 1,881 (74%)] non-profit [4,043 (91%) versus 2,141 (84%)] and performed heart transplantation [4,012 (90%) versus 1,794 (71%)], all p< 0.0001. Patients in T3 versus T1 hospitals had higher rates of MCS (1.4 versus 0.7) and non-MCS (13.9 versus 2.0) infections, all p< 0.0001. Overall, 16.1% of interhospital variability in infection rates was explained by Model 7, leaving 83.9% of variation unexplained, Figure. Model 3 accounted for the most (17.8%) variability of all seven models.

Conclusion: Counter to conventional wisdom, traditional patient characteristics and the implanting hospital characteristics account for less than 20% of interhospital variation in infection rates following dLVAD implant. Efforts to prevent infections should consider other potential determinants (e.g., central line management, social determinants of health) to reduce unwarranted variation in hospital performance.

Identify the source of the funding for this research project: This project was supported by Grant Number R01HS026003 from the Agency for Healthcare Research and Quality.