Acute kidney injury after aortic valve replacement in a nationally representative cohort in the USA

Acute kidney injury after aortic valve replacement in a nationally representative cohort in the USA Abstract Background Randomized trials have consistently shown lower rates of acute kidney injury (AKI) with transcatheter aortic valve replacement (TAVR) compared with surgical aortic valve replacement (SAVR). Comparative rates of AKI for TAVR versus SAVR, and predictors and prognostic implications of AKI after aortic valve replacement (AVR) have not been well studied in nationally representative real-world data. Objectives First, to compare rates of AKI and dialysis requiring AKI in TAVR versus SAVR. Second, to determine predictors of AKI and prognostic implications of AKI in patients undergoing TAVR or SAVR. Methods We used the 2011–14 National Inpatient Sample to identify all patients undergoing isolated TAVR or SAVR using validated international classification of diseases, ninth revision ICD-9 codes. Rates of AKI and AKI requiring dialysis (AKI-D) were compared between the two groups using a propensity-matched design. Predictors of AKI and prognostic impact of AKI on in-hospital outcomes were ascertained using multivariate logistic regression. Results A total of 8004 unweighted TAVR procedures and 29 355 unweighted SAVR procedures representative of 39 898 TAVR and 143 608 SAVR procedures nationwide were included in the analysis. Mean age of all patients undergoing AVR was 70.9 years and 42.3% were females. In a propensity-matched cohort of 4889 pairs of TAVR and SAVR procedures, TAVR was associated with significantly lower rates of AKI [odds ratio (OR) 0.73, 95% confidence interval (CI) 0.66–0.80, P < 0.001] and AKI-D (OR 0.69, 95% CI 0.50–0.96, P = 0.03) compared with SAVR. AKI was associated with significantly higher rates of in-hospital mortality for TAVR (OR 7.16, 95% CI 5.52–9.29, P < 0.001) as well as SAVR (OR 9.43, 95% CI 7.71–11.55, P < 0.001). Conclusions In a large propensity-matched cohort of TAVR and SAVR procedures, TAVR was associated with significantly lower rates of AKI and AKI-D compared with SAVR. AKI and AKI-D are predictors of poor in-hospital outcomes in TAVR as well as SAVR. AKI, cardiac surgery, outcomes research, SAVR, TAVR INTRODUCTION Transcatheter aortic valve replacement (TAVR) is a recent alternative to surgical aortic valve replacement (SAVR) in patients with an intermediate to high surgical risk of perioperative mortality. Acute kidney injury (AKI) is a frequent complication after aortic valve replacement (AVR) with rates ranging from 3 to 50% in different studies depending on the definition of AKI used [1–3]. Randomized trials have consistently shown a lower risk of AKI with TAVR compared with SAVR [4–7]. However, there are few data on comparative rates of AKI and AKI requiring dialysis (AKI-D) in real-world data. In contrast to the lower rates of AKI seen in TAVR versus SAVR in randomized trials, comparative rates of AKI for TAVR versus SAVR have been found to be heterogeneous in observational studies, with reports of lower, similar as well as higher rates of AKI after TAVR compared with SAVR [8]. A recent large observational study found rates of AKI to be similar for the two procedures [9]. While there is extensive literature on risk factors of AKI after cardiopulmonary bypass (CPB) and cardiac surgery as a whole, few studies have evaluated risk factors for AKI after AVR in the contemporary era. Moreover, the prognostic impact of AKI and AKI-D has not been well studied in patients undergoing TAVR or SAVR. To address these knowledge gaps, we used a large nationally representative cohort of TAVR and SAVR procedures to study comparative rates of AKI, predictors of AKI and prognostic impact of AKI on in-hospital outcomes. MATERIALS AND METHODS Data source We used the 2011–14 National Inpatient Sample (NIS) for this study. The NIS is the largest all-payer database of inpatient hospital stays in the USA developed by the Agency for Healthcare Research and Quality (AHRQ) as part of its Healthcare Cost and Utilization Project (HCUP). The NIS contains nearly 8 million hospitalizations per year representative of ∼35 million hospitalizations nationwide. The NIS is representative of  >95% of the target universe, which includes all US community hospital discharges [10]. Identification of cases We used validated ICD-9 codes to include adults (≥18 years) with a procedure code for transfemoral (TF) or transapical (TA) TAVR (35.05 and 35.06) or SAVR (35.21 and 35.22). We excluded discharge records with a diagnosis of end-stage renal disease. We also excluded records with procedure codes for coronary artery bypass grafting, aortic root replacement, mitral valve, tricuspid valve and pulmonic valve procedures (Supplementary data, Table S1 and Figure S1). Outcomes Co-primary outcomes of interest in propensity-matched analysis were AKI and AKI-D. To assess the prognostic impact of AKI and AKI-D on TAVR and SAVR procedures, in-hospital mortality, cost and length of stay (LOS) were used as outcome variables. Statistical analysis We used survey analysis techniques that accounted for clustering and stratification of data to produce national estimates as recommended by the AHRQ. Baseline patient and hospital characteristics were expressed as means for continuous variables and proportions for categorical variables. Differences in baseline characteristics were tested using linear and logistic regression for continuous and categorical variables, respectively. To test for differences in rates of AKI and AKI-D between TAVR and SAVR, we used a propensity-matched design to assemble a 1:1 matched cohort of TAVR and SAVR procedures. Propensity matching has the advantage of enhancing causal inferences in treatment outcomes by adjusting for all confounders simultaneously and, therefore, minimizing the risk of confounding by indication [11, 12]. To generate a propensity score, we built a logistic regression model with TAVR versus SAVR as the outcome variable and 23 baseline pretreatment covariates as predictor variables. A matched cohort of TAVR and SAVR procedures was assembled using the greedy matching algorithm (gmatch command in Stata) using a caliper of 0.1. Balance of covariates before and after propensity score matching was tested using standardized mean differences and P-values (pbalchk command in Stata). To test for predictors of AKI, we built logistic regression models with AKI as the outcome variable and age, sex, nonwhite race, diabetes, a (CKD), congestive heart failure (CHF), atrial fibrillation (AF), obesity, peripheral vascular disease (PVD), Charlson Comorbidity Index (CCI), transapical versus transfemoral approach, blood transfusion, cardiac complications and vascular complications as predictor variables. Analysis for predictors of AKI was stratified by TAVR and SAVR. Choice of predictor variables was based on biological plausibility and data availability. Finally, to evaluate the prognostic impact of AKI and AKI-D on in-hospital outcomes, we built logistic models with relevant hospitalization outcomes as the outcome variable (in-hospital death, LOS categorized at median and cost categorized at median) and AKI or AKI-D as predictor variables in separate models. Analyses were stratified by TAVR and SAVR. All models were adjusted for age, sex, CCI, CKD, postoperative stroke, bleeding or transfusion, cardiac complications and vascular complications. All analyses were conducted using Stata MP 13.0 (Statacorp, College Station, TX, USA) statistical software. RESULTS Baseline characteristics A total of 8004 TAVR and 29 355 SAVR procedures representative of 39 898 TAVR and 143 608 SAVR procedures nationally were included in our study. Mean age of the entire cohort was 70.9 years and 42.3% were female. Patients undergoing TAVR were significantly older (mean age 81.5 versus 68.0 years, P < 0.001), more likely to be female (48.2% versus 40%, P < 0.001) with a higher overall burden of comorbidities (mean CCI 2.79 versus 1.52, P < 0.001). Patients undergoing TAVR had a higher burden of hypertension, diabetes, CKD and CHF compared with those undergoing SAVR (Table 1). A significantly greater proportion of TAVR procedures were done at urban teaching centers compared with SAVR (Table 2). Table 1 Differences in baseline characteristics between TAVR versus SAVR groups Baseline characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Age (years) mean (SE)  81.5 (0.1)  68.0 (0.2)  <0.001  Female (%)  48.2  40.6  <0.001  Nonwhite race (%)  18.2  22.9  <0.001  Primary payer (%)   Medicare  90.4  62.6  <0.001   Medicaid  1.0  4.9  <0.001   Private insurance  6.9  27.9  <0.001  Socioeconomic status (%)   Quartile 1  20.2  21.3  0.16   Quartile 2  24.5  24.8  0.61   Quartile 3  25.2  25.8  0.41   Quartile 4  28.4  26.0  0.03  Comorbidities (%)   Hypertension  79.4  71.7  <0.001   Diabetes without complications  28.5  24.8  <0.001   Diabetes with complications  5.4  4.3  0.001   CKD  35.2  13.8  <0.001   CHF  11.7  1.5  <0.001   AF/flutter  46.2  45.1  0.11   Acute myocardial infarction  2.6  2.5  0.65   Coronary artery disease  71.4  40.6  <0.001   Coagulopathy  23.3  30.4  <0.001   Pulmonary hypertension  3.6  0.5  <0.001   PVD  29.5  14.1  <0.001   Obesity  14.1  21.1  <0.001   Chronic lung disease  33.7  21.1  <0.001   Neurological diseases  6.3  5.0  <0.001   Chronic liver disease  2.4  1.9  <0.001   Alcohol abuse  1.2  2.7  <0.001   Anemia  23.9  18.4  <0.001  CCI mean (SE)  2.79 (0.03)  1.52 (0.01)  <0.001  Baseline characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Age (years) mean (SE)  81.5 (0.1)  68.0 (0.2)  <0.001  Female (%)  48.2  40.6  <0.001  Nonwhite race (%)  18.2  22.9  <0.001  Primary payer (%)   Medicare  90.4  62.6  <0.001   Medicaid  1.0  4.9  <0.001   Private insurance  6.9  27.9  <0.001  Socioeconomic status (%)   Quartile 1  20.2  21.3  0.16   Quartile 2  24.5  24.8  0.61   Quartile 3  25.2  25.8  0.41   Quartile 4  28.4  26.0  0.03  Comorbidities (%)   Hypertension  79.4  71.7  <0.001   Diabetes without complications  28.5  24.8  <0.001   Diabetes with complications  5.4  4.3  0.001   CKD  35.2  13.8  <0.001   CHF  11.7  1.5  <0.001   AF/flutter  46.2  45.1  0.11   Acute myocardial infarction  2.6  2.5  0.65   Coronary artery disease  71.4  40.6  <0.001   Coagulopathy  23.3  30.4  <0.001   Pulmonary hypertension  3.6  0.5  <0.001   PVD  29.5  14.1  <0.001   Obesity  14.1  21.1  <0.001   Chronic lung disease  33.7  21.1  <0.001   Neurological diseases  6.3  5.0  <0.001   Chronic liver disease  2.4  1.9  <0.001   Alcohol abuse  1.2  2.7  <0.001   Anemia  23.9  18.4  <0.001  CCI mean (SE)  2.79 (0.03)  1.52 (0.01)  <0.001  Table 1 Differences in baseline characteristics between TAVR versus SAVR groups Baseline characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Age (years) mean (SE)  81.5 (0.1)  68.0 (0.2)  <0.001  Female (%)  48.2  40.6  <0.001  Nonwhite race (%)  18.2  22.9  <0.001  Primary payer (%)   Medicare  90.4  62.6  <0.001   Medicaid  1.0  4.9  <0.001   Private insurance  6.9  27.9  <0.001  Socioeconomic status (%)   Quartile 1  20.2  21.3  0.16   Quartile 2  24.5  24.8  0.61   Quartile 3  25.2  25.8  0.41   Quartile 4  28.4  26.0  0.03  Comorbidities (%)   Hypertension  79.4  71.7  <0.001   Diabetes without complications  28.5  24.8  <0.001   Diabetes with complications  5.4  4.3  0.001   CKD  35.2  13.8  <0.001   CHF  11.7  1.5  <0.001   AF/flutter  46.2  45.1  0.11   Acute myocardial infarction  2.6  2.5  0.65   Coronary artery disease  71.4  40.6  <0.001   Coagulopathy  23.3  30.4  <0.001   Pulmonary hypertension  3.6  0.5  <0.001   PVD  29.5  14.1  <0.001   Obesity  14.1  21.1  <0.001   Chronic lung disease  33.7  21.1  <0.001   Neurological diseases  6.3  5.0  <0.001   Chronic liver disease  2.4  1.9  <0.001   Alcohol abuse  1.2  2.7  <0.001   Anemia  23.9  18.4  <0.001  CCI mean (SE)  2.79 (0.03)  1.52 (0.01)  <0.001  Baseline characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Age (years) mean (SE)  81.5 (0.1)  68.0 (0.2)  <0.001  Female (%)  48.2  40.6  <0.001  Nonwhite race (%)  18.2  22.9  <0.001  Primary payer (%)   Medicare  90.4  62.6  <0.001   Medicaid  1.0  4.9  <0.001   Private insurance  6.9  27.9  <0.001  Socioeconomic status (%)   Quartile 1  20.2  21.3  0.16   Quartile 2  24.5  24.8  0.61   Quartile 3  25.2  25.8  0.41   Quartile 4  28.4  26.0  0.03  Comorbidities (%)   Hypertension  79.4  71.7  <0.001   Diabetes without complications  28.5  24.8  <0.001   Diabetes with complications  5.4  4.3  0.001   CKD  35.2  13.8  <0.001   CHF  11.7  1.5  <0.001   AF/flutter  46.2  45.1  0.11   Acute myocardial infarction  2.6  2.5  0.65   Coronary artery disease  71.4  40.6  <0.001   Coagulopathy  23.3  30.4  <0.001   Pulmonary hypertension  3.6  0.5  <0.001   PVD  29.5  14.1  <0.001   Obesity  14.1  21.1  <0.001   Chronic lung disease  33.7  21.1  <0.001   Neurological diseases  6.3  5.0  <0.001   Chronic liver disease  2.4  1.9  <0.001   Alcohol abuse  1.2  2.7  <0.001   Anemia  23.9  18.4  <0.001  CCI mean (SE)  2.79 (0.03)  1.52 (0.01)  <0.001  Table 2 Differences in hospital characteristics between TAVR versus SAVR groups Hospital characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Region (%)         Northeast  25.8  23.1  0.10   Midwest  22.2  22.8  0.67   South  33.9  32.9  0.53   West  18.0  21.2  0.09  Location and teaching status (%)         Rural  0.7  2.2  <0.001   Urban nonteaching  10.1  24.2  <0.001   Urban teaching  89.1  73.5  <0.001  Hospital ownership (%)         Government, nonfederal  7.8  6.9  0.27   Private, nonprofit  85.2  83.2  0.12   Private, investor owned  7.0  9.9  0.01  Hospital characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Region (%)         Northeast  25.8  23.1  0.10   Midwest  22.2  22.8  0.67   South  33.9  32.9  0.53   West  18.0  21.2  0.09  Location and teaching status (%)         Rural  0.7  2.2  <0.001   Urban nonteaching  10.1  24.2  <0.001   Urban teaching  89.1  73.5  <0.001  Hospital ownership (%)         Government, nonfederal  7.8  6.9  0.27   Private, nonprofit  85.2  83.2  0.12   Private, investor owned  7.0  9.9  0.01  Table 2 Differences in hospital characteristics between TAVR versus SAVR groups Hospital characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Region (%)         Northeast  25.8  23.1  0.10   Midwest  22.2  22.8  0.67   South  33.9  32.9  0.53   West  18.0  21.2  0.09  Location and teaching status (%)         Rural  0.7  2.2  <0.001   Urban nonteaching  10.1  24.2  <0.001   Urban teaching  89.1  73.5  <0.001  Hospital ownership (%)         Government, nonfederal  7.8  6.9  0.27   Private, nonprofit  85.2  83.2  0.12   Private, investor owned  7.0  9.9  0.01  Hospital characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Region (%)         Northeast  25.8  23.1  0.10   Midwest  22.2  22.8  0.67   South  33.9  32.9  0.53   West  18.0  21.2  0.09  Location and teaching status (%)         Rural  0.7  2.2  <0.001   Urban nonteaching  10.1  24.2  <0.001   Urban teaching  89.1  73.5  <0.001  Hospital ownership (%)         Government, nonfederal  7.8  6.9  0.27   Private, nonprofit  85.2  83.2  0.12   Private, investor owned  7.0  9.9  0.01  Comparative rates of AKI and AKI-D for TAVR versus SAVR in propensity-matched analysis The observed unadjusted rate of AKI in the unmatched cohort was 15.4% for SAVR and 18.7% for TAVR (P < 0.001). Observed rate of AKI-D was 1.26% in both the TAVR and SAVR groups (P = 0.99). Our propensity score model to predict TAVR versus SAVR using 23 baseline covariates had an excellent predictive power with a c-statistic of 0.905 [95% confidence interval (CI) 0.901–0.908]. Prior to propensity matching, there were significant imbalances in covariates between the two treatment groups. After matching, the imbalance in covariates between groups as measured by standardized mean difference was reduced to  ≤10% with P ≥0.05 for all covariates (Figure 1, Supplementary data, Table S2). FIGURE 1 View largeDownload slide Balance of covariates before and after propensity score matching. CAD, coronary artery disease; SES, socioeconomic status; AMI, acute myocardial infarction. FIGURE 1 View largeDownload slide Balance of covariates before and after propensity score matching. CAD, coronary artery disease; SES, socioeconomic status; AMI, acute myocardial infarction. In the cohort of 4889 propensity-matched pairs of TAVR and SAVR procedures, mean age was 78.9 years and 46.4% were females. Rates of AKI were 22.6% for SAVR and 17.6% for TAVR [odds ratio (OR) 0.73, 95% CI 0.66–0.81, P < 0.001]. Rates of AKI-D were 1.8% in the SAVR group and 1.3% in the TAVR group (OR 0.69, 95% CI 0.50–0.96, P < 0.001) (Figure 2). FIGURE 2 View largeDownload slide Comparative rates of AKI and AKI-D in 4889 propensity-matched pairs of TAVR and SAVR procedures. SAVR is the reference category for comparisons. FIGURE 2 View largeDownload slide Comparative rates of AKI and AKI-D in 4889 propensity-matched pairs of TAVR and SAVR procedures. SAVR is the reference category for comparisons. Predictors of AKI In analyses stratified by TAVR and SAVR, we found several significant predictors of in-hospital AKI. For TAVR procedures, significant predictors of AKI included male sex, CKD, CHF, AF, CCI ≥3, a transapical versus a transfemoral approach, blood transfusions, cardiac complications and vascular complications (Table 3). Definitions for cardiac and vascular complications are listed in Supplementary data, Table S1. Table 3 Predictors of AKI in multivariate logistic regression models stratified by TAVR and SAVR Predictors  TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  Demographic factors           Age  0.99 (0.98–1.00)  0.10  1.003 (1.0006–1.007)  0.02   Female sex  0.83 (0.73–0.95)  0.01  0.74 (0.69–0.80)  <0.001   Nonwhite race  1.18 (0.99–1.41)  0.06  1.27 (1.14–1.40)  <0.001  Comorbidity burden           Diabetes  1.03 (0.89–1.18)  0.72  1.05 (0.97–1.14)  0.23   CKD  3.75 (3.14–4.46)  <0.001  4.91 (4.40–5.48)  <0.001   CHF  1.51 (1.28–1.78)  <0.001  8.72 (6.97–10.91)  <0.001   AF  1.20 (1.06–1.35)  0.004  1.39 (1.28–1.50)  <0.001   Obesity  1.06 (0.90–1.26)  0.48  1.24 (1.13–1.36)  <0.001   PVD  0.95 (0.83–1.09)  0.44  1.06 (0.96–1.18)  0.23   CCI ≥3  1.22 (1.002–1.48)  0.05  1.72 (1.55–1.91)  <0.001  Procedural factors           Transapical approach (ref. transfemoral approach)  1.62 (1.39–1.88)  <0.001       Blood transfusion  2.03 (1.76–2.34)  <0.001  1.37 (1.25–1.50)  <0.001   Cardiac complications  1.62 (1.33–1.98)  <0.001  1.15 (1.03–1.29)  0.01   Vascular complications  1.92 (1.48–2.48)  <0.001  1.46 (1.26–1.70)  <0.001  Predictors  TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  Demographic factors           Age  0.99 (0.98–1.00)  0.10  1.003 (1.0006–1.007)  0.02   Female sex  0.83 (0.73–0.95)  0.01  0.74 (0.69–0.80)  <0.001   Nonwhite race  1.18 (0.99–1.41)  0.06  1.27 (1.14–1.40)  <0.001  Comorbidity burden           Diabetes  1.03 (0.89–1.18)  0.72  1.05 (0.97–1.14)  0.23   CKD  3.75 (3.14–4.46)  <0.001  4.91 (4.40–5.48)  <0.001   CHF  1.51 (1.28–1.78)  <0.001  8.72 (6.97–10.91)  <0.001   AF  1.20 (1.06–1.35)  0.004  1.39 (1.28–1.50)  <0.001   Obesity  1.06 (0.90–1.26)  0.48  1.24 (1.13–1.36)  <0.001   PVD  0.95 (0.83–1.09)  0.44  1.06 (0.96–1.18)  0.23   CCI ≥3  1.22 (1.002–1.48)  0.05  1.72 (1.55–1.91)  <0.001  Procedural factors           Transapical approach (ref. transfemoral approach)  1.62 (1.39–1.88)  <0.001       Blood transfusion  2.03 (1.76–2.34)  <0.001  1.37 (1.25–1.50)  <0.001   Cardiac complications  1.62 (1.33–1.98)  <0.001  1.15 (1.03–1.29)  0.01   Vascular complications  1.92 (1.48–2.48)  <0.001  1.46 (1.26–1.70)  <0.001  Table 3 Predictors of AKI in multivariate logistic regression models stratified by TAVR and SAVR Predictors  TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  Demographic factors           Age  0.99 (0.98–1.00)  0.10  1.003 (1.0006–1.007)  0.02   Female sex  0.83 (0.73–0.95)  0.01  0.74 (0.69–0.80)  <0.001   Nonwhite race  1.18 (0.99–1.41)  0.06  1.27 (1.14–1.40)  <0.001  Comorbidity burden           Diabetes  1.03 (0.89–1.18)  0.72  1.05 (0.97–1.14)  0.23   CKD  3.75 (3.14–4.46)  <0.001  4.91 (4.40–5.48)  <0.001   CHF  1.51 (1.28–1.78)  <0.001  8.72 (6.97–10.91)  <0.001   AF  1.20 (1.06–1.35)  0.004  1.39 (1.28–1.50)  <0.001   Obesity  1.06 (0.90–1.26)  0.48  1.24 (1.13–1.36)  <0.001   PVD  0.95 (0.83–1.09)  0.44  1.06 (0.96–1.18)  0.23   CCI ≥3  1.22 (1.002–1.48)  0.05  1.72 (1.55–1.91)  <0.001  Procedural factors           Transapical approach (ref. transfemoral approach)  1.62 (1.39–1.88)  <0.001       Blood transfusion  2.03 (1.76–2.34)  <0.001  1.37 (1.25–1.50)  <0.001   Cardiac complications  1.62 (1.33–1.98)  <0.001  1.15 (1.03–1.29)  0.01   Vascular complications  1.92 (1.48–2.48)  <0.001  1.46 (1.26–1.70)  <0.001  Predictors  TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  Demographic factors           Age  0.99 (0.98–1.00)  0.10  1.003 (1.0006–1.007)  0.02   Female sex  0.83 (0.73–0.95)  0.01  0.74 (0.69–0.80)  <0.001   Nonwhite race  1.18 (0.99–1.41)  0.06  1.27 (1.14–1.40)  <0.001  Comorbidity burden           Diabetes  1.03 (0.89–1.18)  0.72  1.05 (0.97–1.14)  0.23   CKD  3.75 (3.14–4.46)  <0.001  4.91 (4.40–5.48)  <0.001   CHF  1.51 (1.28–1.78)  <0.001  8.72 (6.97–10.91)  <0.001   AF  1.20 (1.06–1.35)  0.004  1.39 (1.28–1.50)  <0.001   Obesity  1.06 (0.90–1.26)  0.48  1.24 (1.13–1.36)  <0.001   PVD  0.95 (0.83–1.09)  0.44  1.06 (0.96–1.18)  0.23   CCI ≥3  1.22 (1.002–1.48)  0.05  1.72 (1.55–1.91)  <0.001  Procedural factors           Transapical approach (ref. transfemoral approach)  1.62 (1.39–1.88)  <0.001       Blood transfusion  2.03 (1.76–2.34)  <0.001  1.37 (1.25–1.50)  <0.001   Cardiac complications  1.62 (1.33–1.98)  <0.001  1.15 (1.03–1.29)  0.01   Vascular complications  1.92 (1.48–2.48)  <0.001  1.46 (1.26–1.70)  <0.001  For SAVR procedures, significant predictors of AKI included age, male sex, nonwhite race, CKD, CHF, AF, obesity, CCI  ≥3, blood transfusion, cardiac complications and vascular complications. The strength of association between CKD and CHF with AKI was higher in the SAVR group than in the TAVR group (Table 3). Impact of AKI and AKI-D on in-hospital outcomes In TAVR procedures, observed rates of in-hospital mortality were 11.7% with AKI and 2% without AKI (OR 6.58, 95% CI 5.25–8.24, P < 0.001). In SAVR procedures, observed rates of in-hospital mortality were 8.2% with AKI and 0.9% without AKI (OR 9.43, 95% CI 7.94–11.20, P < 0.001). In models adjusted for age, sex, CCI, CKD, postoperative stroke, bleeding or transfusion, vascular complications and cardiac complications, we found AKI and AKI-D to be powerful predictors of in-hospital mortality, LOS > median and cost > median. The strength of the association between AKI and in-hospital outcomes was similar for TAVR and SAVR. However, the strength of the association between AKI-D and in-hospital outcomes was stronger in the SAVR group than in the TAVR group (Table 4). Table 4 Prognostic impact of AKI and AKI-D on outcomes in patients undergoing AVR   TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  AKI           In-hospital death  7.16 (5.52–9.29)  <0.001  9.43 (7.71–11.55)  <0.001   LOS > median  5.58 (4.81–6.47)  <0.001  4.73 (4.35–5.15)  <0.001   Cost > median  3.19 (2.76–3.69)  <0.001  3.47 (3.16–3.82)  <0.001  AKI-D           In-hospital death  17.03 (10.18–28.54)  <0.001  20.76 (15.54–27.74)  <0.001   LOS > median  10.97 (5.46–22.06)  <0.001  11.85 (8.08–17.37)  <0.001   Cost > median  10.84 (4.62–25.43)  <0.001  23.27 (12.74–42.50)  <0.001    TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  AKI           In-hospital death  7.16 (5.52–9.29)  <0.001  9.43 (7.71–11.55)  <0.001   LOS > median  5.58 (4.81–6.47)  <0.001  4.73 (4.35–5.15)  <0.001   Cost > median  3.19 (2.76–3.69)  <0.001  3.47 (3.16–3.82)  <0.001  AKI-D           In-hospital death  17.03 (10.18–28.54)  <0.001  20.76 (15.54–27.74)  <0.001   LOS > median  10.97 (5.46–22.06)  <0.001  11.85 (8.08–17.37)  <0.001   Cost > median  10.84 (4.62–25.43)  <0.001  23.27 (12.74–42.50)  <0.001  ORs represent comparison for AKI versus no AKI stratified by TAVR versus SAVR procedure. All models adjusted for age, sex, CCI, CKD, postoperative stroke, bleeding or transfusion, vascular complications and cardiac complications. Table 4 Prognostic impact of AKI and AKI-D on outcomes in patients undergoing AVR   TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  AKI           In-hospital death  7.16 (5.52–9.29)  <0.001  9.43 (7.71–11.55)  <0.001   LOS > median  5.58 (4.81–6.47)  <0.001  4.73 (4.35–5.15)  <0.001   Cost > median  3.19 (2.76–3.69)  <0.001  3.47 (3.16–3.82)  <0.001  AKI-D           In-hospital death  17.03 (10.18–28.54)  <0.001  20.76 (15.54–27.74)  <0.001   LOS > median  10.97 (5.46–22.06)  <0.001  11.85 (8.08–17.37)  <0.001   Cost > median  10.84 (4.62–25.43)  <0.001  23.27 (12.74–42.50)  <0.001    TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  AKI           In-hospital death  7.16 (5.52–9.29)  <0.001  9.43 (7.71–11.55)  <0.001   LOS > median  5.58 (4.81–6.47)  <0.001  4.73 (4.35–5.15)  <0.001   Cost > median  3.19 (2.76–3.69)  <0.001  3.47 (3.16–3.82)  <0.001  AKI-D           In-hospital death  17.03 (10.18–28.54)  <0.001  20.76 (15.54–27.74)  <0.001   LOS > median  10.97 (5.46–22.06)  <0.001  11.85 (8.08–17.37)  <0.001   Cost > median  10.84 (4.62–25.43)  <0.001  23.27 (12.74–42.50)  <0.001  ORs represent comparison for AKI versus no AKI stratified by TAVR versus SAVR procedure. All models adjusted for age, sex, CCI, CKD, postoperative stroke, bleeding or transfusion, vascular complications and cardiac complications. DISCUSSION In this study using a nationally representative sample of TAVR and SAVR procedures, we made the following noteworthy observations. First, in patients undergoing TAVR or SAVR, nearly a fifth developed AKI. Second, in a propensity-matched cohort of 4889 pairs of TAVR and SAVR procedures, TAVR was associated with significantly lower odds of AKI and AKI-D compared with SAVR. Third, in stratified analyses, we identified multiple risk factors for the development of AKI after AVR. These included male sex, CKD, CHF, AF and periprocedural complications. Lastly, we found AKI and AKI-D to be powerful predictors of mortality, longer LOS and higher costs in patients undergoing TAVR and SAVR. Taken together, these data represent the most comprehensive analysis of comparative rates, predictors and prognostic implications of AKI and AKI-D in a large nationally representative cohort of TAVR and SAVR procedures using a rigorous propensity-matched design. The observed rates of AKI in our study were 15.4% for SAVR and 18.7% for TAVR. This estimate is significantly higher than observed rates of AKI in landmark clinical trials, where rates of AKI were ∼1–2% for TAVR and 5% for SAVR procedures [5, 6, 13]. Prior observational studies have found similarly high rates of AKI and AKI-D compared with randomized trials, which may reflect the higher prevalence of CKD and overall comorbidity burden in real-world patients undergoing these procedures compared with patients enrolled in trials [3, 9, 14, 15]. Our findings of lower rates of AKI and AKI-D for TAVR compared with SAVR are in agreement with findings from major randomized trials [4–6, 13]. In a propensity-matched analysis of 195 pairs of TAVR and SAVR procedures at a tertiary care center, rates of postoperative AKI were not significantly different for TAVR versus SAVR (24.1% for TAVR versus 29.7% for SAVR, P = 0.21) [9]. In contrast to these findings, another propensity-matched analysis of 1077 TAVR and 944 SAVR procedures by Thourani et al. found a significantly lower incidence of early AKI with TAVR compared with SAVR (OR 0.14, 95% CI 0.05–0.35) [16]. In recent meta-analyses of randomized trials, TAVR was found to have less than half the odds of AKI compared with SAVR [17, 18]. Similarly, in other meta-analyses that included both randomized trials and observational studies, the odds of AKI were significantly lower for TAVR compared with SAVR [8, 19]. Our findings add incrementally to the literature by unequivocally demonstrating a lower rate of AKI and AKI-D for TAVR versus SAVR in a robust propensity-matched analysis. There are several possible explanations for the lower rates of AKI for TAVR compared with SAVR. First, CPB may be associated with hemodynamic instability, which may induce ischemic renal injury [20]. Surgical trauma, contact with CPB surface and transient ischemia secondary to hemodynamic instability may all lead to an activation of the inflammatory cascade eventually causing AKI. Second, SAVR is associated with a greater requirement for blood transfusions compared with TAVR. The pathophysiological mechanisms by which blood transfusion causes AKI may involve activation of the pro-inflammatory cascade, impaired tissue oxygen delivery and exacerbation of oxidative stress brought about by morphologically and biochemically altered stored red blood cells [21]. Other intraoperative pathogenic mechanisms such as generation of microemboli and free radicals because of blood contact with CPB surface may cause SAVR to be associated with a higher risk of AKI compared with TAVR [20]. We found male sex, CKD, CHF, AF, CCI ≥3, a transapical (versus transfemoral) approach, blood transfusion and cardiac or vascular complications to be significant predictors of AKI for patients undergoing TAVR. We also found age, male sex, nonwhite race, CKD, CHF, AF, obesity, CCI ≥3, blood transfusion, cardiac complications and vascular complications to be risk factors for AKI after SAVR. Baseline CKD [3, 22, 23], blood transfusion [14, 24, 25], TA versus TF approach [26, 27] and heart failure [2] have previously been shown to be associated with postoperative AKI in TAVR. Data on the effect of sex on AKI risk have been conflicting. Female sex has traditionally been considered an AKI risk factor after cardiothoracic surgery [20, 28]. However, several recent studies have shown female sex to be protective of AKI, similar to our findings [1, 28]. In a recent meta-analysis by Neugarten et al., the authors found no sex-specific differences in AKI incidence when meta-analysis was restricted to studies that reported sex-specific odds of AKI after multivariate adjustment [28]. These disparate findings may reflect differences in risk profiles of males versus females in different studies or indicate a true biologic effect of sex on AKI risk and should be evaluated in future studies. We found AKI and AKI-D to be powerful predictors of adverse in-hospital outcomes including a higher risk of in-hospital mortality, longer LOS and higher costs for both TAVR and SAVR procedures. These findings are similar to those seen in multiple previous studies that have evaluated the prognostic impact of AKI on in-hospital outcomes [14, 24, 27, 29]. Our study adds to this previous literature by using a larger nationally representative database and evaluating the effect of AKI on in-hospital mortality as well as resource utilization. Several limitations to our study must be noted. First, the NIS is an administrative database and case-finding was based on ICD-9 codes. Therefore, inaccurate coding for procedures and complications has the potential to impact our results. However, we used codes for TAVR, SAVR as well as complications including AKI and AKI-D that have been extensively used in the literature in the past. Second, while we used a robust propensity-matched design to compare rates of AKI and AKI-D for TAVR versus SAVR, residual confounding because of unmeasured confounding is still a possibility. Third, the NIS does not supply detailed clinical variables, including vital signs in the intra- and perioperative periods and time spent on CPB; therefore, our models could not account for such variables to calculate propensity score or predictors and prognostic impact of AKI. Lastly, our analyses were limited to the index hospitalization and long-term outcomes were not evaluated in this study. In conclusion, we found TAVR to be associated with significantly lower risk of AKI and AKI-D compared with SAVR in a robust propensity-matched analysis in a large national database. AKI and AKI-D were associated with higher in-hospital mortality, longer LOS and higher cost for TAVR as well as SAVR. Future studies should evaluate whether TAVR is associated with better long-term outcomes compared with SAVR in patients with a high preoperative risk of AKI. SUPPLEMENTARY DATA Supplementary data are available at ndt online. CONFLICT OF INTEREST STATEMENT None declared. REFERENCES 1 Ram P, Mezue K, Pressman G et al.   Acute kidney injury post-transcatheter aortic valve replacement. Clin Cardiol  2017; 40: 1357– 1362 Google Scholar CrossRef Search ADS PubMed  2 Najjar M, Salna M, George I. Acute kidney injury after aortic valve replacement: incidence, risk factors and outcomes. Expert Rev Cardiovasc Ther  2015; 13: 301– 316 Google Scholar CrossRef Search ADS PubMed  3 Elhmidi Y, Bleiziffer S, Deutsch M-A et al.   Acute kidney injury after transcatheter aortic valve implantation: incidence, predictors and impact on mortality. Arch Cardiovasc Dis  2014; 107: 133– 139 Google Scholar CrossRef Search ADS PubMed  4 Smith CR, Leon MB, Mack MJ et al.   Transcatheter versus surgical aortic-valve replacement in high-risk patients. N Engl J Med  2011; 364: 2187– 2198 Google Scholar CrossRef Search ADS PubMed  5 Leon MB, Smith CR, Mack MJ et al.   Transcatheter or surgical aortic-valve replacement in intermediate-risk patients. N Engl J Med  2016; 374: 1609– 1620 Google Scholar CrossRef Search ADS PubMed  6 Reardon MJ, Van Mieghem NM, Popma JJ et al.   Surgical or transcatheter aortic-valve replacement in intermediate-risk patients. N Engl J Med  2017; 376: 1321– 1331 Google Scholar CrossRef Search ADS PubMed  7 Reardon MJ, Adams DH, Kleiman NS et al.   2-year outcomes in patients undergoing surgical or self-expanding transcatheter aortic valve replacement. J Am Coll Cardiol  2015; 66: 113– 121 Google Scholar CrossRef Search ADS PubMed  8 Villablanca PA, Mathew V, Thourani VH et al.   A meta-analysis and meta-regression of long-term outcomes of transcatheter versus surgical aortic valve replacement for severe aortic stenosis. Int J Cardiol  2016; 225: 234– 243 Google Scholar CrossRef Search ADS PubMed  9 Thongprayoon C, Cheungpasitporn W, Srivali N et al.   AKI after transcatheter or surgical aortic valve replacement. J Am Soc Nephrol  2016; 27: 1854– 1860 Google Scholar CrossRef Search ADS PubMed  10 NIS Related Reports. hcup-us.ahrq.gov. http://www.hcup-us.ahrq.gov/db/nation/nis/nisrelatedreports.jsp (15 January 2018, date last accessed) 11 D’Agostino RB. Propensity scores in cardiovascular research. Circulation  2007; 115: 2340– 2343 Google Scholar CrossRef Search ADS PubMed  12 Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res  2011; 46: 399– 424 Google Scholar CrossRef Search ADS PubMed  13 Thyregod HGH, Steinbrüchel DA, Ihlemann N et al.   Transcatheter versus surgical aortic valve replacement in patients with severe aortic valve stenosis: 1-year results from the all-comers NOTION randomized clinical trial. J Am Coll Cardiol  2015; 65: 2184– 2194 Google Scholar CrossRef Search ADS PubMed  14 Bagur R, Webb JG, Nietlispach F et al.   Acute kidney injury following transcatheter aortic valve implantation: predictive factors, prognostic value, and comparison with surgical aortic valve replacement. Eur Heart J  2010; 31: 865– 874 Google Scholar CrossRef Search ADS PubMed  15 Van Linden A, Kempfert J, Rastan AJ et al.   Risk of acute kidney injury after minimally invasive transapical aortic valve implantation in 270 patients. Eur J Cardiothorac Surg  2011; 39: 835– 842 Google Scholar CrossRef Search ADS PubMed  16 Thourani VH, Kodali S, Makkar RR et al.   Transcatheter aortic valve replacement versus surgical valve replacement in intermediate-risk patients: a propensity score analysis. Lancet  2016; 387: 2218– 2225 Google Scholar CrossRef Search ADS PubMed  17 Indraratna P, Tian DH, Yan TD et al.   Transcatheter aortic valve implantation versus surgical aortic valve replacement: a meta-analysis of randomized controlled trials. Int J Cardiol  2016; 224: 382– 387 Google Scholar CrossRef Search ADS PubMed  18 Siemieniuk RA, Agoritsas T, Manja V et al.   Transcatheter versus surgical aortic valve replacement in patients with severe aortic stenosis at low and intermediate risk: systematic review and meta-analysis. BMJ  2016; 354: i5130 Google Scholar CrossRef Search ADS PubMed  19 Gargiulo G, Capodanno D, Sannino A et al.   Moderate and severe preoperative chronic kidney disease worsen clinical outcomes after transcatheter aortic valve implantation: meta-analysis of 4992 patients. Circ Cardiovasc Interv  2015; 8: e002220 Google Scholar CrossRef Search ADS PubMed  20 Rosner MH, Okusa MD. Acute kidney injury associated with cardiac surgery. Clin J Am Soc Nephrol  2006; 1: 19– 32 Google Scholar CrossRef Search ADS PubMed  21 Karkouti K. Transfusion and risk of acute kidney injury in cardiac surgery. Br J Anaesth  2012; 109: i29– i38 Google Scholar CrossRef Search ADS PubMed  22 Mohananey D, Griffin BP, Svensson LG et al.   Comparative outcomes of patients with advanced renal dysfunction undergoing transcatheter aortic valve replacement in the United States from 2011 to 2014. Circ Cardiovasc Interv  2017; 10: e005477 Google Scholar CrossRef Search ADS PubMed  23 Gupta T, Goel K, Kolte D et al.   Association of chronic kidney disease with in-hospital outcomes of transcatheter aortic valve replacement. JACC Cardiovasc Interv  2017; 10: 2050– 2060 Google Scholar CrossRef Search ADS PubMed  24 Nuis R-JM, Van Mieghem NM, Tzikas A et al.   Frequency, determinants, and prognostic effects of acute kidney injury and red blood cell transfusion in patients undergoing transcatheter aortic valve implantation. Catheter Cardiovasc Interv  2011; 77: 881– 889 Google Scholar CrossRef Search ADS PubMed  25 Aalaei-Andabili SH, Pourafshar N, Bavry AA et al.   Acute kidney injury after transcatheter aortic valve replacement. J Card Surg  2016; 31: 416– 422 Google Scholar CrossRef Search ADS PubMed  26 Saia F, Ciuca C, Taglieri N et al.   Acute kidney injury following transcatheter aortic valve implantation: incidence, predictors and clinical outcome. Int J Cardiol  2013; 168: 1034– 1040 Google Scholar CrossRef Search ADS PubMed  27 Barbash IM, Ben-Dor I, Dvir D et al.   Incidence and predictors of acute kidney injury after transcatheter aortic valve replacement. Am Heart J  2012; 163: 1031– 1036 Google Scholar CrossRef Search ADS PubMed  28 Neugarten J, Sandilya S, Singh B et al.   Sex and the risk of AKI following cardio-thoracic surgery: a meta-analysis. Clin J Am Soc Nephrol  2016; 11: 2113– 2122 Google Scholar CrossRef Search ADS PubMed  29 Elhmidi Y, Bleiziffer S, Piazza N et al.   Incidence and predictors of acute kidney injury in patients undergoing transcatheter aortic valve implantation. Am Heart J  2011; 161: 735– 739 Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nephrology Dialysis Transplantation Oxford University Press

Acute kidney injury after aortic valve replacement in a nationally representative cohort in the USA

Loading next page...
 
/lp/ou_press/acute-kidney-injury-after-aortic-valve-replacement-in-a-nationally-3ULP3kR0ou
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.
ISSN
0931-0509
eISSN
1460-2385
D.O.I.
10.1093/ndt/gfy097
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Randomized trials have consistently shown lower rates of acute kidney injury (AKI) with transcatheter aortic valve replacement (TAVR) compared with surgical aortic valve replacement (SAVR). Comparative rates of AKI for TAVR versus SAVR, and predictors and prognostic implications of AKI after aortic valve replacement (AVR) have not been well studied in nationally representative real-world data. Objectives First, to compare rates of AKI and dialysis requiring AKI in TAVR versus SAVR. Second, to determine predictors of AKI and prognostic implications of AKI in patients undergoing TAVR or SAVR. Methods We used the 2011–14 National Inpatient Sample to identify all patients undergoing isolated TAVR or SAVR using validated international classification of diseases, ninth revision ICD-9 codes. Rates of AKI and AKI requiring dialysis (AKI-D) were compared between the two groups using a propensity-matched design. Predictors of AKI and prognostic impact of AKI on in-hospital outcomes were ascertained using multivariate logistic regression. Results A total of 8004 unweighted TAVR procedures and 29 355 unweighted SAVR procedures representative of 39 898 TAVR and 143 608 SAVR procedures nationwide were included in the analysis. Mean age of all patients undergoing AVR was 70.9 years and 42.3% were females. In a propensity-matched cohort of 4889 pairs of TAVR and SAVR procedures, TAVR was associated with significantly lower rates of AKI [odds ratio (OR) 0.73, 95% confidence interval (CI) 0.66–0.80, P < 0.001] and AKI-D (OR 0.69, 95% CI 0.50–0.96, P = 0.03) compared with SAVR. AKI was associated with significantly higher rates of in-hospital mortality for TAVR (OR 7.16, 95% CI 5.52–9.29, P < 0.001) as well as SAVR (OR 9.43, 95% CI 7.71–11.55, P < 0.001). Conclusions In a large propensity-matched cohort of TAVR and SAVR procedures, TAVR was associated with significantly lower rates of AKI and AKI-D compared with SAVR. AKI and AKI-D are predictors of poor in-hospital outcomes in TAVR as well as SAVR. AKI, cardiac surgery, outcomes research, SAVR, TAVR INTRODUCTION Transcatheter aortic valve replacement (TAVR) is a recent alternative to surgical aortic valve replacement (SAVR) in patients with an intermediate to high surgical risk of perioperative mortality. Acute kidney injury (AKI) is a frequent complication after aortic valve replacement (AVR) with rates ranging from 3 to 50% in different studies depending on the definition of AKI used [1–3]. Randomized trials have consistently shown a lower risk of AKI with TAVR compared with SAVR [4–7]. However, there are few data on comparative rates of AKI and AKI requiring dialysis (AKI-D) in real-world data. In contrast to the lower rates of AKI seen in TAVR versus SAVR in randomized trials, comparative rates of AKI for TAVR versus SAVR have been found to be heterogeneous in observational studies, with reports of lower, similar as well as higher rates of AKI after TAVR compared with SAVR [8]. A recent large observational study found rates of AKI to be similar for the two procedures [9]. While there is extensive literature on risk factors of AKI after cardiopulmonary bypass (CPB) and cardiac surgery as a whole, few studies have evaluated risk factors for AKI after AVR in the contemporary era. Moreover, the prognostic impact of AKI and AKI-D has not been well studied in patients undergoing TAVR or SAVR. To address these knowledge gaps, we used a large nationally representative cohort of TAVR and SAVR procedures to study comparative rates of AKI, predictors of AKI and prognostic impact of AKI on in-hospital outcomes. MATERIALS AND METHODS Data source We used the 2011–14 National Inpatient Sample (NIS) for this study. The NIS is the largest all-payer database of inpatient hospital stays in the USA developed by the Agency for Healthcare Research and Quality (AHRQ) as part of its Healthcare Cost and Utilization Project (HCUP). The NIS contains nearly 8 million hospitalizations per year representative of ∼35 million hospitalizations nationwide. The NIS is representative of  >95% of the target universe, which includes all US community hospital discharges [10]. Identification of cases We used validated ICD-9 codes to include adults (≥18 years) with a procedure code for transfemoral (TF) or transapical (TA) TAVR (35.05 and 35.06) or SAVR (35.21 and 35.22). We excluded discharge records with a diagnosis of end-stage renal disease. We also excluded records with procedure codes for coronary artery bypass grafting, aortic root replacement, mitral valve, tricuspid valve and pulmonic valve procedures (Supplementary data, Table S1 and Figure S1). Outcomes Co-primary outcomes of interest in propensity-matched analysis were AKI and AKI-D. To assess the prognostic impact of AKI and AKI-D on TAVR and SAVR procedures, in-hospital mortality, cost and length of stay (LOS) were used as outcome variables. Statistical analysis We used survey analysis techniques that accounted for clustering and stratification of data to produce national estimates as recommended by the AHRQ. Baseline patient and hospital characteristics were expressed as means for continuous variables and proportions for categorical variables. Differences in baseline characteristics were tested using linear and logistic regression for continuous and categorical variables, respectively. To test for differences in rates of AKI and AKI-D between TAVR and SAVR, we used a propensity-matched design to assemble a 1:1 matched cohort of TAVR and SAVR procedures. Propensity matching has the advantage of enhancing causal inferences in treatment outcomes by adjusting for all confounders simultaneously and, therefore, minimizing the risk of confounding by indication [11, 12]. To generate a propensity score, we built a logistic regression model with TAVR versus SAVR as the outcome variable and 23 baseline pretreatment covariates as predictor variables. A matched cohort of TAVR and SAVR procedures was assembled using the greedy matching algorithm (gmatch command in Stata) using a caliper of 0.1. Balance of covariates before and after propensity score matching was tested using standardized mean differences and P-values (pbalchk command in Stata). To test for predictors of AKI, we built logistic regression models with AKI as the outcome variable and age, sex, nonwhite race, diabetes, a (CKD), congestive heart failure (CHF), atrial fibrillation (AF), obesity, peripheral vascular disease (PVD), Charlson Comorbidity Index (CCI), transapical versus transfemoral approach, blood transfusion, cardiac complications and vascular complications as predictor variables. Analysis for predictors of AKI was stratified by TAVR and SAVR. Choice of predictor variables was based on biological plausibility and data availability. Finally, to evaluate the prognostic impact of AKI and AKI-D on in-hospital outcomes, we built logistic models with relevant hospitalization outcomes as the outcome variable (in-hospital death, LOS categorized at median and cost categorized at median) and AKI or AKI-D as predictor variables in separate models. Analyses were stratified by TAVR and SAVR. All models were adjusted for age, sex, CCI, CKD, postoperative stroke, bleeding or transfusion, cardiac complications and vascular complications. All analyses were conducted using Stata MP 13.0 (Statacorp, College Station, TX, USA) statistical software. RESULTS Baseline characteristics A total of 8004 TAVR and 29 355 SAVR procedures representative of 39 898 TAVR and 143 608 SAVR procedures nationally were included in our study. Mean age of the entire cohort was 70.9 years and 42.3% were female. Patients undergoing TAVR were significantly older (mean age 81.5 versus 68.0 years, P < 0.001), more likely to be female (48.2% versus 40%, P < 0.001) with a higher overall burden of comorbidities (mean CCI 2.79 versus 1.52, P < 0.001). Patients undergoing TAVR had a higher burden of hypertension, diabetes, CKD and CHF compared with those undergoing SAVR (Table 1). A significantly greater proportion of TAVR procedures were done at urban teaching centers compared with SAVR (Table 2). Table 1 Differences in baseline characteristics between TAVR versus SAVR groups Baseline characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Age (years) mean (SE)  81.5 (0.1)  68.0 (0.2)  <0.001  Female (%)  48.2  40.6  <0.001  Nonwhite race (%)  18.2  22.9  <0.001  Primary payer (%)   Medicare  90.4  62.6  <0.001   Medicaid  1.0  4.9  <0.001   Private insurance  6.9  27.9  <0.001  Socioeconomic status (%)   Quartile 1  20.2  21.3  0.16   Quartile 2  24.5  24.8  0.61   Quartile 3  25.2  25.8  0.41   Quartile 4  28.4  26.0  0.03  Comorbidities (%)   Hypertension  79.4  71.7  <0.001   Diabetes without complications  28.5  24.8  <0.001   Diabetes with complications  5.4  4.3  0.001   CKD  35.2  13.8  <0.001   CHF  11.7  1.5  <0.001   AF/flutter  46.2  45.1  0.11   Acute myocardial infarction  2.6  2.5  0.65   Coronary artery disease  71.4  40.6  <0.001   Coagulopathy  23.3  30.4  <0.001   Pulmonary hypertension  3.6  0.5  <0.001   PVD  29.5  14.1  <0.001   Obesity  14.1  21.1  <0.001   Chronic lung disease  33.7  21.1  <0.001   Neurological diseases  6.3  5.0  <0.001   Chronic liver disease  2.4  1.9  <0.001   Alcohol abuse  1.2  2.7  <0.001   Anemia  23.9  18.4  <0.001  CCI mean (SE)  2.79 (0.03)  1.52 (0.01)  <0.001  Baseline characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Age (years) mean (SE)  81.5 (0.1)  68.0 (0.2)  <0.001  Female (%)  48.2  40.6  <0.001  Nonwhite race (%)  18.2  22.9  <0.001  Primary payer (%)   Medicare  90.4  62.6  <0.001   Medicaid  1.0  4.9  <0.001   Private insurance  6.9  27.9  <0.001  Socioeconomic status (%)   Quartile 1  20.2  21.3  0.16   Quartile 2  24.5  24.8  0.61   Quartile 3  25.2  25.8  0.41   Quartile 4  28.4  26.0  0.03  Comorbidities (%)   Hypertension  79.4  71.7  <0.001   Diabetes without complications  28.5  24.8  <0.001   Diabetes with complications  5.4  4.3  0.001   CKD  35.2  13.8  <0.001   CHF  11.7  1.5  <0.001   AF/flutter  46.2  45.1  0.11   Acute myocardial infarction  2.6  2.5  0.65   Coronary artery disease  71.4  40.6  <0.001   Coagulopathy  23.3  30.4  <0.001   Pulmonary hypertension  3.6  0.5  <0.001   PVD  29.5  14.1  <0.001   Obesity  14.1  21.1  <0.001   Chronic lung disease  33.7  21.1  <0.001   Neurological diseases  6.3  5.0  <0.001   Chronic liver disease  2.4  1.9  <0.001   Alcohol abuse  1.2  2.7  <0.001   Anemia  23.9  18.4  <0.001  CCI mean (SE)  2.79 (0.03)  1.52 (0.01)  <0.001  Table 1 Differences in baseline characteristics between TAVR versus SAVR groups Baseline characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Age (years) mean (SE)  81.5 (0.1)  68.0 (0.2)  <0.001  Female (%)  48.2  40.6  <0.001  Nonwhite race (%)  18.2  22.9  <0.001  Primary payer (%)   Medicare  90.4  62.6  <0.001   Medicaid  1.0  4.9  <0.001   Private insurance  6.9  27.9  <0.001  Socioeconomic status (%)   Quartile 1  20.2  21.3  0.16   Quartile 2  24.5  24.8  0.61   Quartile 3  25.2  25.8  0.41   Quartile 4  28.4  26.0  0.03  Comorbidities (%)   Hypertension  79.4  71.7  <0.001   Diabetes without complications  28.5  24.8  <0.001   Diabetes with complications  5.4  4.3  0.001   CKD  35.2  13.8  <0.001   CHF  11.7  1.5  <0.001   AF/flutter  46.2  45.1  0.11   Acute myocardial infarction  2.6  2.5  0.65   Coronary artery disease  71.4  40.6  <0.001   Coagulopathy  23.3  30.4  <0.001   Pulmonary hypertension  3.6  0.5  <0.001   PVD  29.5  14.1  <0.001   Obesity  14.1  21.1  <0.001   Chronic lung disease  33.7  21.1  <0.001   Neurological diseases  6.3  5.0  <0.001   Chronic liver disease  2.4  1.9  <0.001   Alcohol abuse  1.2  2.7  <0.001   Anemia  23.9  18.4  <0.001  CCI mean (SE)  2.79 (0.03)  1.52 (0.01)  <0.001  Baseline characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Age (years) mean (SE)  81.5 (0.1)  68.0 (0.2)  <0.001  Female (%)  48.2  40.6  <0.001  Nonwhite race (%)  18.2  22.9  <0.001  Primary payer (%)   Medicare  90.4  62.6  <0.001   Medicaid  1.0  4.9  <0.001   Private insurance  6.9  27.9  <0.001  Socioeconomic status (%)   Quartile 1  20.2  21.3  0.16   Quartile 2  24.5  24.8  0.61   Quartile 3  25.2  25.8  0.41   Quartile 4  28.4  26.0  0.03  Comorbidities (%)   Hypertension  79.4  71.7  <0.001   Diabetes without complications  28.5  24.8  <0.001   Diabetes with complications  5.4  4.3  0.001   CKD  35.2  13.8  <0.001   CHF  11.7  1.5  <0.001   AF/flutter  46.2  45.1  0.11   Acute myocardial infarction  2.6  2.5  0.65   Coronary artery disease  71.4  40.6  <0.001   Coagulopathy  23.3  30.4  <0.001   Pulmonary hypertension  3.6  0.5  <0.001   PVD  29.5  14.1  <0.001   Obesity  14.1  21.1  <0.001   Chronic lung disease  33.7  21.1  <0.001   Neurological diseases  6.3  5.0  <0.001   Chronic liver disease  2.4  1.9  <0.001   Alcohol abuse  1.2  2.7  <0.001   Anemia  23.9  18.4  <0.001  CCI mean (SE)  2.79 (0.03)  1.52 (0.01)  <0.001  Table 2 Differences in hospital characteristics between TAVR versus SAVR groups Hospital characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Region (%)         Northeast  25.8  23.1  0.10   Midwest  22.2  22.8  0.67   South  33.9  32.9  0.53   West  18.0  21.2  0.09  Location and teaching status (%)         Rural  0.7  2.2  <0.001   Urban nonteaching  10.1  24.2  <0.001   Urban teaching  89.1  73.5  <0.001  Hospital ownership (%)         Government, nonfederal  7.8  6.9  0.27   Private, nonprofit  85.2  83.2  0.12   Private, investor owned  7.0  9.9  0.01  Hospital characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Region (%)         Northeast  25.8  23.1  0.10   Midwest  22.2  22.8  0.67   South  33.9  32.9  0.53   West  18.0  21.2  0.09  Location and teaching status (%)         Rural  0.7  2.2  <0.001   Urban nonteaching  10.1  24.2  <0.001   Urban teaching  89.1  73.5  <0.001  Hospital ownership (%)         Government, nonfederal  7.8  6.9  0.27   Private, nonprofit  85.2  83.2  0.12   Private, investor owned  7.0  9.9  0.01  Table 2 Differences in hospital characteristics between TAVR versus SAVR groups Hospital characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Region (%)         Northeast  25.8  23.1  0.10   Midwest  22.2  22.8  0.67   South  33.9  32.9  0.53   West  18.0  21.2  0.09  Location and teaching status (%)         Rural  0.7  2.2  <0.001   Urban nonteaching  10.1  24.2  <0.001   Urban teaching  89.1  73.5  <0.001  Hospital ownership (%)         Government, nonfederal  7.8  6.9  0.27   Private, nonprofit  85.2  83.2  0.12   Private, investor owned  7.0  9.9  0.01  Hospital characteristics  TAVR  SAVR  P-value  unweighted n = 8004,  unweighted n = 29 355,  weighted n = 39 898  weighted n = 143 608  Region (%)         Northeast  25.8  23.1  0.10   Midwest  22.2  22.8  0.67   South  33.9  32.9  0.53   West  18.0  21.2  0.09  Location and teaching status (%)         Rural  0.7  2.2  <0.001   Urban nonteaching  10.1  24.2  <0.001   Urban teaching  89.1  73.5  <0.001  Hospital ownership (%)         Government, nonfederal  7.8  6.9  0.27   Private, nonprofit  85.2  83.2  0.12   Private, investor owned  7.0  9.9  0.01  Comparative rates of AKI and AKI-D for TAVR versus SAVR in propensity-matched analysis The observed unadjusted rate of AKI in the unmatched cohort was 15.4% for SAVR and 18.7% for TAVR (P < 0.001). Observed rate of AKI-D was 1.26% in both the TAVR and SAVR groups (P = 0.99). Our propensity score model to predict TAVR versus SAVR using 23 baseline covariates had an excellent predictive power with a c-statistic of 0.905 [95% confidence interval (CI) 0.901–0.908]. Prior to propensity matching, there were significant imbalances in covariates between the two treatment groups. After matching, the imbalance in covariates between groups as measured by standardized mean difference was reduced to  ≤10% with P ≥0.05 for all covariates (Figure 1, Supplementary data, Table S2). FIGURE 1 View largeDownload slide Balance of covariates before and after propensity score matching. CAD, coronary artery disease; SES, socioeconomic status; AMI, acute myocardial infarction. FIGURE 1 View largeDownload slide Balance of covariates before and after propensity score matching. CAD, coronary artery disease; SES, socioeconomic status; AMI, acute myocardial infarction. In the cohort of 4889 propensity-matched pairs of TAVR and SAVR procedures, mean age was 78.9 years and 46.4% were females. Rates of AKI were 22.6% for SAVR and 17.6% for TAVR [odds ratio (OR) 0.73, 95% CI 0.66–0.81, P < 0.001]. Rates of AKI-D were 1.8% in the SAVR group and 1.3% in the TAVR group (OR 0.69, 95% CI 0.50–0.96, P < 0.001) (Figure 2). FIGURE 2 View largeDownload slide Comparative rates of AKI and AKI-D in 4889 propensity-matched pairs of TAVR and SAVR procedures. SAVR is the reference category for comparisons. FIGURE 2 View largeDownload slide Comparative rates of AKI and AKI-D in 4889 propensity-matched pairs of TAVR and SAVR procedures. SAVR is the reference category for comparisons. Predictors of AKI In analyses stratified by TAVR and SAVR, we found several significant predictors of in-hospital AKI. For TAVR procedures, significant predictors of AKI included male sex, CKD, CHF, AF, CCI ≥3, a transapical versus a transfemoral approach, blood transfusions, cardiac complications and vascular complications (Table 3). Definitions for cardiac and vascular complications are listed in Supplementary data, Table S1. Table 3 Predictors of AKI in multivariate logistic regression models stratified by TAVR and SAVR Predictors  TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  Demographic factors           Age  0.99 (0.98–1.00)  0.10  1.003 (1.0006–1.007)  0.02   Female sex  0.83 (0.73–0.95)  0.01  0.74 (0.69–0.80)  <0.001   Nonwhite race  1.18 (0.99–1.41)  0.06  1.27 (1.14–1.40)  <0.001  Comorbidity burden           Diabetes  1.03 (0.89–1.18)  0.72  1.05 (0.97–1.14)  0.23   CKD  3.75 (3.14–4.46)  <0.001  4.91 (4.40–5.48)  <0.001   CHF  1.51 (1.28–1.78)  <0.001  8.72 (6.97–10.91)  <0.001   AF  1.20 (1.06–1.35)  0.004  1.39 (1.28–1.50)  <0.001   Obesity  1.06 (0.90–1.26)  0.48  1.24 (1.13–1.36)  <0.001   PVD  0.95 (0.83–1.09)  0.44  1.06 (0.96–1.18)  0.23   CCI ≥3  1.22 (1.002–1.48)  0.05  1.72 (1.55–1.91)  <0.001  Procedural factors           Transapical approach (ref. transfemoral approach)  1.62 (1.39–1.88)  <0.001       Blood transfusion  2.03 (1.76–2.34)  <0.001  1.37 (1.25–1.50)  <0.001   Cardiac complications  1.62 (1.33–1.98)  <0.001  1.15 (1.03–1.29)  0.01   Vascular complications  1.92 (1.48–2.48)  <0.001  1.46 (1.26–1.70)  <0.001  Predictors  TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  Demographic factors           Age  0.99 (0.98–1.00)  0.10  1.003 (1.0006–1.007)  0.02   Female sex  0.83 (0.73–0.95)  0.01  0.74 (0.69–0.80)  <0.001   Nonwhite race  1.18 (0.99–1.41)  0.06  1.27 (1.14–1.40)  <0.001  Comorbidity burden           Diabetes  1.03 (0.89–1.18)  0.72  1.05 (0.97–1.14)  0.23   CKD  3.75 (3.14–4.46)  <0.001  4.91 (4.40–5.48)  <0.001   CHF  1.51 (1.28–1.78)  <0.001  8.72 (6.97–10.91)  <0.001   AF  1.20 (1.06–1.35)  0.004  1.39 (1.28–1.50)  <0.001   Obesity  1.06 (0.90–1.26)  0.48  1.24 (1.13–1.36)  <0.001   PVD  0.95 (0.83–1.09)  0.44  1.06 (0.96–1.18)  0.23   CCI ≥3  1.22 (1.002–1.48)  0.05  1.72 (1.55–1.91)  <0.001  Procedural factors           Transapical approach (ref. transfemoral approach)  1.62 (1.39–1.88)  <0.001       Blood transfusion  2.03 (1.76–2.34)  <0.001  1.37 (1.25–1.50)  <0.001   Cardiac complications  1.62 (1.33–1.98)  <0.001  1.15 (1.03–1.29)  0.01   Vascular complications  1.92 (1.48–2.48)  <0.001  1.46 (1.26–1.70)  <0.001  Table 3 Predictors of AKI in multivariate logistic regression models stratified by TAVR and SAVR Predictors  TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  Demographic factors           Age  0.99 (0.98–1.00)  0.10  1.003 (1.0006–1.007)  0.02   Female sex  0.83 (0.73–0.95)  0.01  0.74 (0.69–0.80)  <0.001   Nonwhite race  1.18 (0.99–1.41)  0.06  1.27 (1.14–1.40)  <0.001  Comorbidity burden           Diabetes  1.03 (0.89–1.18)  0.72  1.05 (0.97–1.14)  0.23   CKD  3.75 (3.14–4.46)  <0.001  4.91 (4.40–5.48)  <0.001   CHF  1.51 (1.28–1.78)  <0.001  8.72 (6.97–10.91)  <0.001   AF  1.20 (1.06–1.35)  0.004  1.39 (1.28–1.50)  <0.001   Obesity  1.06 (0.90–1.26)  0.48  1.24 (1.13–1.36)  <0.001   PVD  0.95 (0.83–1.09)  0.44  1.06 (0.96–1.18)  0.23   CCI ≥3  1.22 (1.002–1.48)  0.05  1.72 (1.55–1.91)  <0.001  Procedural factors           Transapical approach (ref. transfemoral approach)  1.62 (1.39–1.88)  <0.001       Blood transfusion  2.03 (1.76–2.34)  <0.001  1.37 (1.25–1.50)  <0.001   Cardiac complications  1.62 (1.33–1.98)  <0.001  1.15 (1.03–1.29)  0.01   Vascular complications  1.92 (1.48–2.48)  <0.001  1.46 (1.26–1.70)  <0.001  Predictors  TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  Demographic factors           Age  0.99 (0.98–1.00)  0.10  1.003 (1.0006–1.007)  0.02   Female sex  0.83 (0.73–0.95)  0.01  0.74 (0.69–0.80)  <0.001   Nonwhite race  1.18 (0.99–1.41)  0.06  1.27 (1.14–1.40)  <0.001  Comorbidity burden           Diabetes  1.03 (0.89–1.18)  0.72  1.05 (0.97–1.14)  0.23   CKD  3.75 (3.14–4.46)  <0.001  4.91 (4.40–5.48)  <0.001   CHF  1.51 (1.28–1.78)  <0.001  8.72 (6.97–10.91)  <0.001   AF  1.20 (1.06–1.35)  0.004  1.39 (1.28–1.50)  <0.001   Obesity  1.06 (0.90–1.26)  0.48  1.24 (1.13–1.36)  <0.001   PVD  0.95 (0.83–1.09)  0.44  1.06 (0.96–1.18)  0.23   CCI ≥3  1.22 (1.002–1.48)  0.05  1.72 (1.55–1.91)  <0.001  Procedural factors           Transapical approach (ref. transfemoral approach)  1.62 (1.39–1.88)  <0.001       Blood transfusion  2.03 (1.76–2.34)  <0.001  1.37 (1.25–1.50)  <0.001   Cardiac complications  1.62 (1.33–1.98)  <0.001  1.15 (1.03–1.29)  0.01   Vascular complications  1.92 (1.48–2.48)  <0.001  1.46 (1.26–1.70)  <0.001  For SAVR procedures, significant predictors of AKI included age, male sex, nonwhite race, CKD, CHF, AF, obesity, CCI  ≥3, blood transfusion, cardiac complications and vascular complications. The strength of association between CKD and CHF with AKI was higher in the SAVR group than in the TAVR group (Table 3). Impact of AKI and AKI-D on in-hospital outcomes In TAVR procedures, observed rates of in-hospital mortality were 11.7% with AKI and 2% without AKI (OR 6.58, 95% CI 5.25–8.24, P < 0.001). In SAVR procedures, observed rates of in-hospital mortality were 8.2% with AKI and 0.9% without AKI (OR 9.43, 95% CI 7.94–11.20, P < 0.001). In models adjusted for age, sex, CCI, CKD, postoperative stroke, bleeding or transfusion, vascular complications and cardiac complications, we found AKI and AKI-D to be powerful predictors of in-hospital mortality, LOS > median and cost > median. The strength of the association between AKI and in-hospital outcomes was similar for TAVR and SAVR. However, the strength of the association between AKI-D and in-hospital outcomes was stronger in the SAVR group than in the TAVR group (Table 4). Table 4 Prognostic impact of AKI and AKI-D on outcomes in patients undergoing AVR   TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  AKI           In-hospital death  7.16 (5.52–9.29)  <0.001  9.43 (7.71–11.55)  <0.001   LOS > median  5.58 (4.81–6.47)  <0.001  4.73 (4.35–5.15)  <0.001   Cost > median  3.19 (2.76–3.69)  <0.001  3.47 (3.16–3.82)  <0.001  AKI-D           In-hospital death  17.03 (10.18–28.54)  <0.001  20.76 (15.54–27.74)  <0.001   LOS > median  10.97 (5.46–22.06)  <0.001  11.85 (8.08–17.37)  <0.001   Cost > median  10.84 (4.62–25.43)  <0.001  23.27 (12.74–42.50)  <0.001    TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  AKI           In-hospital death  7.16 (5.52–9.29)  <0.001  9.43 (7.71–11.55)  <0.001   LOS > median  5.58 (4.81–6.47)  <0.001  4.73 (4.35–5.15)  <0.001   Cost > median  3.19 (2.76–3.69)  <0.001  3.47 (3.16–3.82)  <0.001  AKI-D           In-hospital death  17.03 (10.18–28.54)  <0.001  20.76 (15.54–27.74)  <0.001   LOS > median  10.97 (5.46–22.06)  <0.001  11.85 (8.08–17.37)  <0.001   Cost > median  10.84 (4.62–25.43)  <0.001  23.27 (12.74–42.50)  <0.001  ORs represent comparison for AKI versus no AKI stratified by TAVR versus SAVR procedure. All models adjusted for age, sex, CCI, CKD, postoperative stroke, bleeding or transfusion, vascular complications and cardiac complications. Table 4 Prognostic impact of AKI and AKI-D on outcomes in patients undergoing AVR   TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  AKI           In-hospital death  7.16 (5.52–9.29)  <0.001  9.43 (7.71–11.55)  <0.001   LOS > median  5.58 (4.81–6.47)  <0.001  4.73 (4.35–5.15)  <0.001   Cost > median  3.19 (2.76–3.69)  <0.001  3.47 (3.16–3.82)  <0.001  AKI-D           In-hospital death  17.03 (10.18–28.54)  <0.001  20.76 (15.54–27.74)  <0.001   LOS > median  10.97 (5.46–22.06)  <0.001  11.85 (8.08–17.37)  <0.001   Cost > median  10.84 (4.62–25.43)  <0.001  23.27 (12.74–42.50)  <0.001    TAVR  P-value  SAVR  P-value  OR (95% CI)  OR (95% CI)  AKI           In-hospital death  7.16 (5.52–9.29)  <0.001  9.43 (7.71–11.55)  <0.001   LOS > median  5.58 (4.81–6.47)  <0.001  4.73 (4.35–5.15)  <0.001   Cost > median  3.19 (2.76–3.69)  <0.001  3.47 (3.16–3.82)  <0.001  AKI-D           In-hospital death  17.03 (10.18–28.54)  <0.001  20.76 (15.54–27.74)  <0.001   LOS > median  10.97 (5.46–22.06)  <0.001  11.85 (8.08–17.37)  <0.001   Cost > median  10.84 (4.62–25.43)  <0.001  23.27 (12.74–42.50)  <0.001  ORs represent comparison for AKI versus no AKI stratified by TAVR versus SAVR procedure. All models adjusted for age, sex, CCI, CKD, postoperative stroke, bleeding or transfusion, vascular complications and cardiac complications. DISCUSSION In this study using a nationally representative sample of TAVR and SAVR procedures, we made the following noteworthy observations. First, in patients undergoing TAVR or SAVR, nearly a fifth developed AKI. Second, in a propensity-matched cohort of 4889 pairs of TAVR and SAVR procedures, TAVR was associated with significantly lower odds of AKI and AKI-D compared with SAVR. Third, in stratified analyses, we identified multiple risk factors for the development of AKI after AVR. These included male sex, CKD, CHF, AF and periprocedural complications. Lastly, we found AKI and AKI-D to be powerful predictors of mortality, longer LOS and higher costs in patients undergoing TAVR and SAVR. Taken together, these data represent the most comprehensive analysis of comparative rates, predictors and prognostic implications of AKI and AKI-D in a large nationally representative cohort of TAVR and SAVR procedures using a rigorous propensity-matched design. The observed rates of AKI in our study were 15.4% for SAVR and 18.7% for TAVR. This estimate is significantly higher than observed rates of AKI in landmark clinical trials, where rates of AKI were ∼1–2% for TAVR and 5% for SAVR procedures [5, 6, 13]. Prior observational studies have found similarly high rates of AKI and AKI-D compared with randomized trials, which may reflect the higher prevalence of CKD and overall comorbidity burden in real-world patients undergoing these procedures compared with patients enrolled in trials [3, 9, 14, 15]. Our findings of lower rates of AKI and AKI-D for TAVR compared with SAVR are in agreement with findings from major randomized trials [4–6, 13]. In a propensity-matched analysis of 195 pairs of TAVR and SAVR procedures at a tertiary care center, rates of postoperative AKI were not significantly different for TAVR versus SAVR (24.1% for TAVR versus 29.7% for SAVR, P = 0.21) [9]. In contrast to these findings, another propensity-matched analysis of 1077 TAVR and 944 SAVR procedures by Thourani et al. found a significantly lower incidence of early AKI with TAVR compared with SAVR (OR 0.14, 95% CI 0.05–0.35) [16]. In recent meta-analyses of randomized trials, TAVR was found to have less than half the odds of AKI compared with SAVR [17, 18]. Similarly, in other meta-analyses that included both randomized trials and observational studies, the odds of AKI were significantly lower for TAVR compared with SAVR [8, 19]. Our findings add incrementally to the literature by unequivocally demonstrating a lower rate of AKI and AKI-D for TAVR versus SAVR in a robust propensity-matched analysis. There are several possible explanations for the lower rates of AKI for TAVR compared with SAVR. First, CPB may be associated with hemodynamic instability, which may induce ischemic renal injury [20]. Surgical trauma, contact with CPB surface and transient ischemia secondary to hemodynamic instability may all lead to an activation of the inflammatory cascade eventually causing AKI. Second, SAVR is associated with a greater requirement for blood transfusions compared with TAVR. The pathophysiological mechanisms by which blood transfusion causes AKI may involve activation of the pro-inflammatory cascade, impaired tissue oxygen delivery and exacerbation of oxidative stress brought about by morphologically and biochemically altered stored red blood cells [21]. Other intraoperative pathogenic mechanisms such as generation of microemboli and free radicals because of blood contact with CPB surface may cause SAVR to be associated with a higher risk of AKI compared with TAVR [20]. We found male sex, CKD, CHF, AF, CCI ≥3, a transapical (versus transfemoral) approach, blood transfusion and cardiac or vascular complications to be significant predictors of AKI for patients undergoing TAVR. We also found age, male sex, nonwhite race, CKD, CHF, AF, obesity, CCI ≥3, blood transfusion, cardiac complications and vascular complications to be risk factors for AKI after SAVR. Baseline CKD [3, 22, 23], blood transfusion [14, 24, 25], TA versus TF approach [26, 27] and heart failure [2] have previously been shown to be associated with postoperative AKI in TAVR. Data on the effect of sex on AKI risk have been conflicting. Female sex has traditionally been considered an AKI risk factor after cardiothoracic surgery [20, 28]. However, several recent studies have shown female sex to be protective of AKI, similar to our findings [1, 28]. In a recent meta-analysis by Neugarten et al., the authors found no sex-specific differences in AKI incidence when meta-analysis was restricted to studies that reported sex-specific odds of AKI after multivariate adjustment [28]. These disparate findings may reflect differences in risk profiles of males versus females in different studies or indicate a true biologic effect of sex on AKI risk and should be evaluated in future studies. We found AKI and AKI-D to be powerful predictors of adverse in-hospital outcomes including a higher risk of in-hospital mortality, longer LOS and higher costs for both TAVR and SAVR procedures. These findings are similar to those seen in multiple previous studies that have evaluated the prognostic impact of AKI on in-hospital outcomes [14, 24, 27, 29]. Our study adds to this previous literature by using a larger nationally representative database and evaluating the effect of AKI on in-hospital mortality as well as resource utilization. Several limitations to our study must be noted. First, the NIS is an administrative database and case-finding was based on ICD-9 codes. Therefore, inaccurate coding for procedures and complications has the potential to impact our results. However, we used codes for TAVR, SAVR as well as complications including AKI and AKI-D that have been extensively used in the literature in the past. Second, while we used a robust propensity-matched design to compare rates of AKI and AKI-D for TAVR versus SAVR, residual confounding because of unmeasured confounding is still a possibility. Third, the NIS does not supply detailed clinical variables, including vital signs in the intra- and perioperative periods and time spent on CPB; therefore, our models could not account for such variables to calculate propensity score or predictors and prognostic impact of AKI. Lastly, our analyses were limited to the index hospitalization and long-term outcomes were not evaluated in this study. In conclusion, we found TAVR to be associated with significantly lower risk of AKI and AKI-D compared with SAVR in a robust propensity-matched analysis in a large national database. AKI and AKI-D were associated with higher in-hospital mortality, longer LOS and higher cost for TAVR as well as SAVR. Future studies should evaluate whether TAVR is associated with better long-term outcomes compared with SAVR in patients with a high preoperative risk of AKI. SUPPLEMENTARY DATA Supplementary data are available at ndt online. CONFLICT OF INTEREST STATEMENT None declared. REFERENCES 1 Ram P, Mezue K, Pressman G et al.   Acute kidney injury post-transcatheter aortic valve replacement. Clin Cardiol  2017; 40: 1357– 1362 Google Scholar CrossRef Search ADS PubMed  2 Najjar M, Salna M, George I. Acute kidney injury after aortic valve replacement: incidence, risk factors and outcomes. Expert Rev Cardiovasc Ther  2015; 13: 301– 316 Google Scholar CrossRef Search ADS PubMed  3 Elhmidi Y, Bleiziffer S, Deutsch M-A et al.   Acute kidney injury after transcatheter aortic valve implantation: incidence, predictors and impact on mortality. Arch Cardiovasc Dis  2014; 107: 133– 139 Google Scholar CrossRef Search ADS PubMed  4 Smith CR, Leon MB, Mack MJ et al.   Transcatheter versus surgical aortic-valve replacement in high-risk patients. N Engl J Med  2011; 364: 2187– 2198 Google Scholar CrossRef Search ADS PubMed  5 Leon MB, Smith CR, Mack MJ et al.   Transcatheter or surgical aortic-valve replacement in intermediate-risk patients. N Engl J Med  2016; 374: 1609– 1620 Google Scholar CrossRef Search ADS PubMed  6 Reardon MJ, Van Mieghem NM, Popma JJ et al.   Surgical or transcatheter aortic-valve replacement in intermediate-risk patients. N Engl J Med  2017; 376: 1321– 1331 Google Scholar CrossRef Search ADS PubMed  7 Reardon MJ, Adams DH, Kleiman NS et al.   2-year outcomes in patients undergoing surgical or self-expanding transcatheter aortic valve replacement. J Am Coll Cardiol  2015; 66: 113– 121 Google Scholar CrossRef Search ADS PubMed  8 Villablanca PA, Mathew V, Thourani VH et al.   A meta-analysis and meta-regression of long-term outcomes of transcatheter versus surgical aortic valve replacement for severe aortic stenosis. Int J Cardiol  2016; 225: 234– 243 Google Scholar CrossRef Search ADS PubMed  9 Thongprayoon C, Cheungpasitporn W, Srivali N et al.   AKI after transcatheter or surgical aortic valve replacement. J Am Soc Nephrol  2016; 27: 1854– 1860 Google Scholar CrossRef Search ADS PubMed  10 NIS Related Reports. hcup-us.ahrq.gov. http://www.hcup-us.ahrq.gov/db/nation/nis/nisrelatedreports.jsp (15 January 2018, date last accessed) 11 D’Agostino RB. Propensity scores in cardiovascular research. Circulation  2007; 115: 2340– 2343 Google Scholar CrossRef Search ADS PubMed  12 Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res  2011; 46: 399– 424 Google Scholar CrossRef Search ADS PubMed  13 Thyregod HGH, Steinbrüchel DA, Ihlemann N et al.   Transcatheter versus surgical aortic valve replacement in patients with severe aortic valve stenosis: 1-year results from the all-comers NOTION randomized clinical trial. J Am Coll Cardiol  2015; 65: 2184– 2194 Google Scholar CrossRef Search ADS PubMed  14 Bagur R, Webb JG, Nietlispach F et al.   Acute kidney injury following transcatheter aortic valve implantation: predictive factors, prognostic value, and comparison with surgical aortic valve replacement. Eur Heart J  2010; 31: 865– 874 Google Scholar CrossRef Search ADS PubMed  15 Van Linden A, Kempfert J, Rastan AJ et al.   Risk of acute kidney injury after minimally invasive transapical aortic valve implantation in 270 patients. Eur J Cardiothorac Surg  2011; 39: 835– 842 Google Scholar CrossRef Search ADS PubMed  16 Thourani VH, Kodali S, Makkar RR et al.   Transcatheter aortic valve replacement versus surgical valve replacement in intermediate-risk patients: a propensity score analysis. Lancet  2016; 387: 2218– 2225 Google Scholar CrossRef Search ADS PubMed  17 Indraratna P, Tian DH, Yan TD et al.   Transcatheter aortic valve implantation versus surgical aortic valve replacement: a meta-analysis of randomized controlled trials. Int J Cardiol  2016; 224: 382– 387 Google Scholar CrossRef Search ADS PubMed  18 Siemieniuk RA, Agoritsas T, Manja V et al.   Transcatheter versus surgical aortic valve replacement in patients with severe aortic stenosis at low and intermediate risk: systematic review and meta-analysis. BMJ  2016; 354: i5130 Google Scholar CrossRef Search ADS PubMed  19 Gargiulo G, Capodanno D, Sannino A et al.   Moderate and severe preoperative chronic kidney disease worsen clinical outcomes after transcatheter aortic valve implantation: meta-analysis of 4992 patients. Circ Cardiovasc Interv  2015; 8: e002220 Google Scholar CrossRef Search ADS PubMed  20 Rosner MH, Okusa MD. Acute kidney injury associated with cardiac surgery. Clin J Am Soc Nephrol  2006; 1: 19– 32 Google Scholar CrossRef Search ADS PubMed  21 Karkouti K. Transfusion and risk of acute kidney injury in cardiac surgery. Br J Anaesth  2012; 109: i29– i38 Google Scholar CrossRef Search ADS PubMed  22 Mohananey D, Griffin BP, Svensson LG et al.   Comparative outcomes of patients with advanced renal dysfunction undergoing transcatheter aortic valve replacement in the United States from 2011 to 2014. Circ Cardiovasc Interv  2017; 10: e005477 Google Scholar CrossRef Search ADS PubMed  23 Gupta T, Goel K, Kolte D et al.   Association of chronic kidney disease with in-hospital outcomes of transcatheter aortic valve replacement. JACC Cardiovasc Interv  2017; 10: 2050– 2060 Google Scholar CrossRef Search ADS PubMed  24 Nuis R-JM, Van Mieghem NM, Tzikas A et al.   Frequency, determinants, and prognostic effects of acute kidney injury and red blood cell transfusion in patients undergoing transcatheter aortic valve implantation. Catheter Cardiovasc Interv  2011; 77: 881– 889 Google Scholar CrossRef Search ADS PubMed  25 Aalaei-Andabili SH, Pourafshar N, Bavry AA et al.   Acute kidney injury after transcatheter aortic valve replacement. J Card Surg  2016; 31: 416– 422 Google Scholar CrossRef Search ADS PubMed  26 Saia F, Ciuca C, Taglieri N et al.   Acute kidney injury following transcatheter aortic valve implantation: incidence, predictors and clinical outcome. Int J Cardiol  2013; 168: 1034– 1040 Google Scholar CrossRef Search ADS PubMed  27 Barbash IM, Ben-Dor I, Dvir D et al.   Incidence and predictors of acute kidney injury after transcatheter aortic valve replacement. Am Heart J  2012; 163: 1031– 1036 Google Scholar CrossRef Search ADS PubMed  28 Neugarten J, Sandilya S, Singh B et al.   Sex and the risk of AKI following cardio-thoracic surgery: a meta-analysis. Clin J Am Soc Nephrol  2016; 11: 2113– 2122 Google Scholar CrossRef Search ADS PubMed  29 Elhmidi Y, Bleiziffer S, Piazza N et al.   Incidence and predictors of acute kidney injury in patients undergoing transcatheter aortic valve implantation. Am Heart J  2011; 161: 735– 739 Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Nephrology Dialysis TransplantationOxford University Press

Published: Apr 18, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off