Patient factors associated with left ventricular assist device infections: A scoping review

Abstract

Infections are widely prevalent in left ventricular assist device (LVAD) recipients and associated with adverse events including mortality and rehospitalizations. Current evidence examining factors associated with infections in this setting predominantly comprises single-center observational data. We performed a scoping review to systematically summarize all existing studies examining patient-related factors associated with infections after LVAD implantation. Studies published between 01/06 and 02/19 were identified through searching 5 bibliographic databases: PubMed, Scopus, EMBASE, CINAHL, and Web of Science Core Collection. Inclusion criteria required examination of patient-related factors associated with infections among recipients of contemporary implantable, continuous flow LVADs. Key study characteristics were extracted by four independent reviewers and current literature described narratively. All analyses took place between February 2019 and May 2021. A total of 31 studies met inclusion criteria. All included studies were observational, and most commonly focused on driveline infections (n = 17). Factors studied most commonly included body composition (n = 8), diabetes and other comorbidities (n = 8), and psychosocial/socio-economic factors (n = 6). Studies were frequently single-center with heterogeneity in definition of infectious outcomes as well as exposure variables. Patient race and sex did not correlate with infection risk. There was no consistent association noted between obesity, diabetes, or psychosocial/socio-economic factors and infections in LVAD recipients. Two studies reported a significant association between malnutrition and hypoalbuminemia and post implant infections. This review summarizes 31 studies that described patient-related factors associated with infection after LVAD implantation. Patient related comorbidities, especially body composition and diabetes, were most commonly evaluated, but were not consistently associated with infections after LVAD implantation.

Advancements in durable left ventricular assist device (LVAD) technology have contributed to improved survival and broader utilization for LVADs as treatment for end stage heart failure, particularly as permanent or destination therapy.^1,2 Although technological advances have reduced the incidence of certain complications, notably pump thrombosis, patients remain at-risk for infection in both early and late stages of LVAD therapy.¹ Infections in LVAD recipients contribute to an increased risk of rehospitalization and expenditure as well as decreased quality of life and survival.^1,3

Given the elevated risk of postimplant infections and their associated adverse outcomes, several studies have attempted to evaluate determinants of infections. In particular, understanding patient factors associated with infections after LVAD implantation may help improve patient selection for LVAD therapy and identify high risk cohorts for targeted infection prevention strategies. However, studies examining patient-level risk factors have been limited by design (e.g., case series, single center studies), scope (e.g., limited sample size or infection subtypes) and lack of granularity of outcomes. Accordingly, the synthesis of this literature has been challenging.⁴

A scoping review was performed to systematically summarize all published literature examining patient factors associated with infections after LVAD implantation. A scoping review maps literature on a specific topic to provide an opportunity to identify key concepts and gaps in research to help inform practice.⁵

Methods

A scoping review was designed and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses Extension for Scoping Review (PRISMA-SCr) guidelines (Appendix 1).⁶ Scoping review methodology was utilized rather than systematic review methodology due to wide variability in study designs and lack of high quality studies.⁷

Literature search

Information sources were searched for studies including contemporary LVADs published between January 2006 and February 2019. The list of sources included: Pubmed.gov, Scopus (scopus.com), EMBASE (including Embase Classic, on Embase.com), CINAHL Complete (EbscoHost), and Web of Science Core Collection (SCI-EXPANDED, SSCI, A&HCI, CPCI-SSH, BKCI-SSH, ESCI, CCR-EXPANDED).⁸ A comprehensive and sensitive search strategy (Appendix 2) was created in concert with a health sciences informationist (WT). References of all included manuscripts were screened for other relevant manuscripts. Search keywords were determined by 3 authors (SS, TW, WT) using the population (adults with durable LVAD), concept (LVAD infection), and context (risk factors for infection) approach to designing scoping review search strategy.⁷ The Covidence systematic review software (Veritas Health Information, Melbourne, Australia) was used for deduplication.

Screening evidence

Three authors (SS, TW, DL) independently screened the title and abstract of each manuscript identified in the initial search to determine suitability for inclusion for full-text review. Studies were included at this stage if the population was ≥18 years and underwent durable continuous flow LVAD implantation. Any manuscript deemed appropriate by either reviewer was included for the next stage of review. Four authors (SS, TW, GY, DL) then reviewed the full text of selected studies according to the following inclusion criteria: Adults ≥18 years, contemporary LVADs (HeartMate II (Abbott Labs, Chicago, IL), HeartMate 3 (Abbott Labs, Chicago, IL), and HeartWare HVAD (Medtronic, Inc., Minneapolis, MN)), published full text manuscript in English language, studies with at least 10 patients, described rate of any infection. Exclusion criteria included: narrative and systematic reviews, editorials, LVAD type not described.

Data charting and synthesis

Four independent reviewers (SS, MP, GY, DL) performed in-depth review of the selected manuscripts. Data from each text were abstracted independently by 2 reviewers. A standardized review form and data dictionary were utilized to collect elements from each study (Appendix 3): first and last author, study design, country, study population parameters, demographics, follow-up duration, and use of standardized infection definitions as described by the International Society for Heart and Lung Transplant (ISHLT) Consensus statement.⁹ Disagreements were resolved through consensus among reviewers.

Statistical analysis

Summary statistics were generated where appropriate using Stata IC (StataCorp, College Station, TX). Interrater reliability for reviewer screening and data abstraction was assessed by the Kappa Statistic.¹⁰ A heat map was created to characterize the type of infection evaluated and if the patient risk factor significantly affected infection risk. Studies were evaluated for adherence to standardized infection definitions as per the 2011 International Society for Heart and Lung Transplant (ISHLT) Consensus Statement.⁹ Infections were accordingly classified as LVAD-specific (pump or cannula, pocket or driveline infections), LVAD-related (mediastinitis, infective endocarditis or bloodstream infections), or non-LVAD infections (pneumonia, urinary tract infections, etc). For this review, infections in studies not adhering to the ISHLT criteria were reported as described in the study.

Results

A total of 9,680 studies were evaluated using titles and abstracts, with 480 undergoing full-text review. Overall, 137 studies described infections in LVAD recipients and/or identified associated risk factors. This manuscript examines findings from 31 unique studies evaluating preoperative patient risk factors associated with infections after LVAD implantation (Figure 1). Interrater reliability for screening and abstraction was high (Ƙ 0.71). The full list of studies along with summarized findings is included in Appendix 4. All included studies were observational, with the majority being single center series (n = 24, 77.4%). The most assessed factors were obesity (n = 8), diabetes and other comorbidities (n = 8), and psychosocial/socioeconomic evaluations (n = 6). In all studies, exposure-outcome relationships were evaluated on a univariate basis. The most studied infections were driveline infections (n = 17), any infection (n = 11) and bloodstream infection (n = 8). Overall findings from this scoping review are summarized in Figure 2.

Figure 1.

Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Review (PRISMA-ScR) Diagram

Figure 2. Heat Map of Patient Factors and Studied Infections in LVAD recipients.

Heat Map of Patient Factors and Studied Infections in LVAD recipients. Yellow: No significant change in infection. Green: Significant decrease in infection. Red: Significant increase in infection. Gray areas indicate exposure-outcome relationships that were not evaluated.

Abbreviations: HM2: Heart Mate 2; LVAD: left ventricular assist devices

Patient demographics

One study evaluated the relationship between patient sex and postimplant infections.¹¹ Bogaev et al reported a lower incidence of device-related infections among women as compared to men; however, there was no difference in other infection subtypes.¹¹ Tsiouris et al evaluated postimplant outcomes by race and demonstrated that African American patients (relative to White patients) had a higher incidence of pneumonia but noted no difference in the incidence of driveline infections.¹²

Obesity

Eight studies evaluated the impact of body mass index, obesity, or body surface area on infections.^13–20 Four studies reported no association between obesity and infection after LVAD implantation,^13,15,16,20 although 2 of these studies had overlapping patient populations ^15,16 and all 4 used different BMI cut points for modeling obesity. Akay et al demonstrated a higher incidence of driveline infections in patients with BMI >30.¹⁷ Brewer et al reported a higher incidence of sepsis and device-related infection in a multicenter study among patients with BMI >35.^15,18 Two studies solely focused on patients with small body surface area (BSA <1.5m2). Volkovicher et al demonstrated a lower incidence of pump infection and LVAD-related sepsis, although a higher incidence of non-LVAD related sepsis among small BSA patients - there was no noted statistically significant association noted for driveline infections.¹⁴ Ono et al identified a higher incidence of driveline infection among small BSA patients.¹⁹

Diabetes and other comorbidities

A number of preimplant comorbidities have been evaluated. ^21–27 Three studies reported no significant relationship between diabetes and risk of infection after LVAD implantation.^21,22,27 Similarly, prior reports have found no significant risk of infection associated with preoperative atrial fibrillation, type of cardiomyopathy, patient age, and renal dysfunction.^23–26 One study reported no significant association between preoperative multidrug resistant bacteria colonization and LVAD-specific/related and non-LVAD infections.²⁸ However, colonization was associated with a risk of infection-related mortality.

Preoperative temporary mechanical circulatory support

Two studies evaluated patients who received temporary mechanical circulatory support prior to LVAD implantation. In a single center study evaluating TandemHeart (CardiacAssist Inc., Pittsburgh, PA, USA) or Impella (Abiomed, Danvers, MA, USA), Kurihara et al reported no significant association with their use and driveline infection, pump infection or sepsis.²⁹ In a 5-center study, Shah et al found patients receiving temporary mechanical circulatory support (Impella, TandemHeart, extracorporeal membrane oxygenation, temporary extracorporeal ventricular assist devices) had a significantly higher rate of infection compared to INTERMACS Profiles 1–3.³⁰

Psychological and socioeconomic factors

Six studies evaluated various psychosocial factors or assessments. ^31–36 There was no significant association between preimplant Psychosocial Assessment of Candidates for Transplantation (PACT) and postoperative infection.³¹ Likewise, neither the Stanford integrated psychosocial assessment for transplantation (SIPAT) nor Transplant Evaluation Rating Scale (TERS) were associated with postoperative infection.^33,34 Cogswell et al found that active substance abuse at the time of LVAD implantation was associated with higher risk of driveline infection.³² An analysis of United Network for Organ sharing data and the Agency for Healthcare Research and Quality Socioeconomic Status (SES) index found no association between SES and device infection.³⁵ Koeckert et al found that there was no significant association between caregiver support status (having a consistent caregiver, a caregiver that quit, or having no caregiver) at the time of LVAD implantation and in the postoperative period and risk of device-related infections.³⁶

Laboratory assessment

Five studies specifically evaluated laboratory assessments or risk stratification.^37–41 Critsinelis et al evaluated the Model of End-Stage Liver Disease eXcluding International Normalized Ratio (MELD-XI) and found that MELD-XI ≥14 was a significant predictor of pump/driveline infection or bacteremia.³⁷ Likewise, Uribarri et al evaluated the Nutritional Risk Index (NRI) and found that worsening levels of malnutrition were associated with higher incidence of postoperative infection.⁴¹ Similarly, preoperative hypoalbuminemia (<3.5 g/dL) was associated with a higher incidence of pump-related infections and LVAD-related sepsis; however, there was no significant difference in driveline infection.³⁸ An analysis of preimplant red blood cell distribution width (RDW) found no significant difference in postoperative infections across RDW tertiles.³⁹ Abou Obeid et al reported that the majority of LVAD recipients had insufficient or deficient Vitamin D levels.⁴⁰ As compared to patients with normal Vitamin D levels, those with insufficient or deficient levels had a significantly higher risk of driveline infections.

Discussion

This scoping review identified 31 studies examining the association between preoperative patient factors and post-LVAD infections. The vast majority of included studies were single-center experiences (77.4%). While several areas were evaluated by multiple studies, notably obesity and other comorbidities, this review also identified areas that may benefit from additional research. Additionally, preoperative laboratory evaluation and risk assessments provide clinicians with important information related to infection risk (Figure 3).

Figure 3.

This scoping review included 31 studies evaluating patient factors associated with infections in LVAD recipients. Several factors were found to be associated with LVAD infections however, comparison was limited by the lack of standardization of exposure and outcome definitions

Although several studies evaluated obesity, the relationship between obesity and infections after LVAD implantation remains an area requiring further evaluation. The studies evaluating obesity demonstrate conflicting results about the risk of infection subtypes across BMI categories. Importantly, the ability to compare study findings is limited by the lack of standardization of infection measures and modeling of obesity. Only 1 study reported driveline infection by Intermacs criteria¹⁷ and 1 reported using International Society for Heart and Lung Transplantation (ISHLT) criteria.¹⁸ The remaining studies examining obesity reported a variety of outcomes including “infection,” “sternal infection,” “driveline/pocket infection,” “device infection,” “systemic infection,” and “sepsis.” A similar relationship was demonstrated regarding diabetes. Although none of the 3 studies found an association between diabetes and postoperative infection, only 2 used Intermacs criteria to define infection outcomes,^22,27 and none of the studies used the same criteria to define diabetes. The variability in both risk factor and outcome definitions limits: (1) the generalizability of each study’s results and (2) further summarizing of findings in the form of meta-analyses.

Several preoperative laboratory assessments including MELD-XI, the Nutritional Risk Index, albumin, and Vitamin D were associated with infection risk. The NRI, which is calculated from serum albumin along with patient weight relative to ideal body weight, is utilized to assess a patient’s nutritional status, with a higher NRI associated with less nutritional risk. Both higher BMI and higher albumin will result in a higher calculated NRI (less nutritional risk),⁴¹ however in a separate study evaluating preoperative albumin levels, BMI was not a significant predictor of albumin levels.³⁸ These findings suggest that the relationship between BMI, albumin, and postoperative infection is not yet fully understood and that additional markers of preoperative nutritional status may be useful for stratifying patient risk for postoperative adverse event risk assessment. Although each of these studies were single center series, they highlight several areas for additional research and possible interventions to improve or supplement nutrition and optimize body weight.

Knowledge gaps and future directions

This review identifies several areas that may affect postoperative infection risk but has been inadequately evaluated. Our findings suggest the need for further investigation of preoperative temporary mechanical circulatory support, patient caregiver status/engagement, discharge location (home, inpatient rehabilitation, skilled nursing facility, etc.), and socioeconomic/education status. Several studies also demonstrate ongoing interest in body composition and outcomes after LVAD implantation. These studies could be extended to further evaluate measures of frailty and interventions to improve functional status or resolve malnutrition. Addressing the above topics would collectively inform clinical teams about mechanisms for optimizing post-LVAD implantation infections, and other outcomes more broadly, while delineating high-risk patient subgroups.

The investigatory landscape related to LVAD infections could be improved through standardization of reporting for exposures/outcomes and appropriately powered investigations that move beyond associated studies towards evaluations of mechanisms and pathways underlying various outcomes. Such mechanistic investigations would highlight possible areas for interventions to mitigate infection risks and improve outcomes for patients. Standardization of reporting for exposures and outcomes represents an opportunity for professional organizations (ISHLT, STS-Intermacs) and journal editors to advance the quality of investigations related to LVAD infections.

Limitations

This scoping review has several limitations. The search was limited to English language publications, as such, it is possible that relevant studies were excluded. The exclusion of non-English language studies has been identified as a necessary tradeoff to enhance feasibility.⁴² Abstracts were excluded from consideration because data presented in abstracts are often preliminary or are inconsistent with final publications.⁴³ Additionally, this review did not include formal evaluations of study quality, nor was a meta-analysis performed (due to the wide variation in study designs and outcomes). The lack of quality assessment has been recognized as a limitation of scoping review methodology but coincides with the goal of assessing the current scope of literature related to a topic rather than producing a synthesized answer to a particular question.⁷

Summary

Stratification of postoperative infection risk should be an important component when evaluating the risks and benefits of pursuing LVAD therapy given a patient’s risk of postimplant infections and associated sequelae. Several patient factors associated with infection after LVAD implantation have been evaluated. Obesity was the most studied comorbidity; however, the effect of obesity on infection after LVAD implantation is not clear. Comparison of outcomes across studies has been limited by heterogeneity of comorbidity classification and infection classification. Future studies would benefit from standardization to enhance generalizability.