Risk Factors and Outcomes of Gastrointestinal Bleeding in Left Ventricular Assist Device Recipients




Increasing use of left ventricular assist devices (LVADs) has been accompanied by rising incidence of gastrointestinal bleeding (GIB). Objectives of this study were to determine the yearly incidence of GIB in LVAD recipients, compare outcomes of continuous-flow (CF) and pulsatile-flow LVAD eras, and investigate for risk factors. The Healthcare Cost and Utilization Project–Nationwide Inpatient Sample database from 2005 to 2010 was analyzed. Primary outcome of interest was incidence of GIB in LVAD recipients. Multivariate logistic regression model was used to examine independent associations of GIB with risk factors and outcomes. An estimated 8,879 LVAD index admissions and 8,722 readmissions in LVAD recipients over 6 years were analyzed. The yearly incidence of GIB after LVAD implantation increased from 5% in 2005 to 10% in 2010. On multivariate regression analysis, the odds of GIB was 3.24 times greater (95% confidence interval 1.53 to 6.89) in the era of CF LVADs than in the era of pulsatile-flow LVADs. Compared to their younger counterparts, in LVAD recipients aged >65 years, the adjusted odds of GIB was 20.5 times greater (95% confidence interval 2.24 to 188). GIB did not significantly increase the inhospital mortality but increased the inpatient length of stay. In conclusion, the incidence of GIB in LVAD recipients has increased since the use of CF LVADs has increased, leading to greater inpatient lengths of stay and hospital charges. Older recipients of CF LVADs appear to be at a greater risk of GIB.


A growing population of patients with end-stage heart disease are being considered for left ventricular assist device (LVAD) implantation because of organ shortage. First-generation LVADs were pulsatile volume-displacement devices. Second-generation continuous-flow (CF) systems, such as the HeartMate-II (Thoratec, Pleasanton, California) and HeartWare (HeartWare, Massachusetts), based on rotary-pump technology offer improved reliability and allow implantation in smaller sized patients. The increasing use of LVADs has led to a significant rate of gastrointestinal bleeding (GIB) in their recipients. For patients with CF LVADs, reported incidence of GIB is 17.6% to 40% which is more than that for pulsatile-flow devices. Our aim was to evaluate outcomes and investigate for significant risk factors for GIB in these patients. After the Food and Drug Administration approved CF LVADs such as the HeartMate-II in April 2008 for use as “bridge to transplant,” a much wider patient population became eligible for CF devices and so, pulsatile-flow LVADs fell out of use in the United States. To best compare pulsatile-flow versus CF LVADs, we performed our analysis in 2 temporal subgroups: 2005 to 2007 representing the pulsatile LVAD era and 2008 to 2010 representing the CF LVAD era.


Methods


The Healthcare Cost and Utilization Project–Nationwide Inpatient Sample (NIS) database from the year 2005 to 2010 was used for this study. NIS was created by the Agency for Healthcare Research and Quality as part of the Healthcare Cost and Utilization Project and contains data on hospital stays from about 1,000 hospitals sampled to approximate a 20% stratified sample of US community hospitals. Each hospitalization is treated as an individual entry in the database and is coded with 1 principal diagnosis, up to 14 secondary diagnoses, and 15 procedural diagnoses associated with that stay. NIS includes information on all hospitalizations, regardless of the payer. To facilitate the production of national estimates, both hospital and discharge weights are provided. As the NIS database is publicly available and without patient identifiers, this study is deemed exempt from review by the institutional review board. Adults aged ≥18 years were considered for the study. The International Classification of Diseases, Ninth Revision , clinical modification code 37.66 (ICD-9-CM) was used to identify admissions during which LVAD was placed, which will be referred to as index admission in this study. During index admissions, upper and lower GI bleed, if they occurred after LVAD placement, were identified using procedure codes for esophagogastroduodenoscopy and colonoscopy. Readmissions were identified using ICD-9-CM code V43.21. During readmissions, GIB was identified using primary diagnosis of upper or lower GI bleed. The ICD-9-CM codes to identify upper and lower GI bleed are provided in Supplementary Table 1 . Known risk factors for GIB were identified using ICD-9-CM codes as presented in Supplementary Table 2 . NIS variables were used to identify patient age, gender, race, and primary payer. Age groups were divided into 4 categories: 18 to 34, 35 to 49, 50 to 64, and ≥65 years. Information on race was missing in 20% to 25% of records, and these are reported under “unknown” category. NIS classifications for a hospital’s teaching characteristics were also used.


Primary outcomes of interest were the frequency of GIB and all cause inhospital mortality. Secondary outcomes included length of hospital stay and hospital charges. The variables and outcomes related to GIB were examined in LVAD recipients in 2 groups: during index admission for LVAD implantation and readmissions. The 2 groups were further subdivided into 2 temporal subgroups: 2005 to 2007 and 2008 to 2010, respectively, representing the eras when pulsatile LVADs and CF LVADs were placed. Statistical analysis was performed in STATA IC 11.0 (STATA-Corp, College Station, Texas) using the strata weights and survey commands to generate national estimates of patients with LVADs during index admissions and readmissions. Pearson chi-square test and Wilcoxon rank test were used to compare categorical and continuous variables, respectively. Multivariate logistic regression model was used to examine independent associations of GIB with the 2 eras. Univariate model was initially used to identify clinically plausible risk factors associated with GIB. Variables that were significant in the univariate analysis at p <0.10 were included in the final multivariate model. Those variables, which have been shown to be associated with GIB in the literature, were kept in the model even if they did not meet the statistical significance. These independent variables were then checked for multicollinearity using variance inflation factor. Interaction between the 2 eras and the variables were checked and kept in the model if p value was <0.10. Similarly, multivariate logistic regression was used to examine if GIB was independently associated with an increase in hospital mortality in patients receiving LVADs.




Results


In the era of pulsatile-flow LVADs, 2,672 devices were implanted with 1,180 readmissions, whereas in the era of CF LVADs, 6,207 devices were implanted with 7,542 readmissions ( Tables 1 and 2 ). Demographic information of LVAD recipients in the 2 eras is provided in Table 1 (LVAD index admissions) and Table 2 (LVAD readmissions). Device recipients in the CF LVAD era were older, predominantly male, and had more co-morbidities such as obesity, malnutrition, and chronic anemia, compared to pulsatile-flow LVAD recipients ( Table 1 ). The trend of the yearly incidence of all episodes of GIB in LVAD recipients requiring inpatient care increased from 5% in 2005 to 10% in 2010 ( Figure 1 ). Incidence of GIB during LVAD index admissions was not significantly different in the 2 eras (7.7% vs 8.6%; Table 1 ). However, among the patients readmitted after LVAD implantation, the proportion of recipients admitted for GIB was more than 3 times greater in the CF LVAD era than the pulsatile-flow LVAD era (10% vs 2.9%, p <0.001; Table 2 ). After adjusting for other clinically plausible risk factors using multivariate regression analysis, the odds of GIB was 3.24 times greater (95% confidence interval [CI] 1.53 to 6.89) in the CF LVAD era compared to the era of pulsatile-flow LVAD implantation ( Figure 2 ). Age was the strongest independent risk factor associated with increased incidence of GIB in LVAD recipients. For readmissions after LVAD placement, compared to 18- to 34-year-old recipients, the adjusted odds of having GIB was 20.5 times greater (95% CI 2.24 to 188) for patients aged >65 years and 13.5 times greater (95% CI 1.47 to 124) for recipients aged 50 to 64 years ( Figure 2 ). Among other co-morbidities, history of peptic ulcer disease was independently associated with 20.5 times greater odds of GIB in LVAD recipients. The inhospital mortality in patients receiving LVADs has decreased in CF LVAD era compared to pulsatile-flow LVAD era. Presence of GIB did not significantly increase the inhospital mortality in either of the 2 eras; both during index admission and readmissions ( Table 3 ). Readmissions for GIB after LVAD placement had higher mortality in the CF LVAD era compared to pulsatile-flow LVAD era (2.1% vs 0%) though not reaching statistical significance. In the era of CF LVADs, presence of GIB during index admission was significantly associated with increased median inpatient length of stay (LOS; 44.5 vs 29 days, p <0.001). However, in the era of pulsatile-flow LVADs, presence of GIB during index admission did not significantly increase the median LOS (36 vs 31 days). In the absence of GIB, median LOS in the CF LVAD era was similar to the pulsatile-flow LVAD era (29 vs 31 days). When the index admission was complicated by GIB, the hospital charges were significantly greater for both pulsatile-flow LVADs ($714,567) and CF LVADs ($674,038) compared to implantations that were not complicated by GIB (p <0.05).



Table 1

Characteristics of Left Ventricular Assist Device recipients for index admission



























































































































































































































































Variable 2005-2007 2008-2010 P-value
Total LVADs 2672 6207
GI Bleed after LVAD placement 7.7% 8.6% 0.67
Age (years) 0.025
18-34 10.7% 10.4%
35-49 25.3% 19.1%
50-64 46.7% 46.7%
≥65 17.3% 23.8%
Males 73% 78% 0.025
Race 0.58
White 58.1% 54.5%
Black 14.4% 16.7%
Hispanic 6.5% 6.0%
Asian 1.7% 2.4%
Others 2.4% 3.8%
Unknown 16.8% 16.7%
Insurance <0.001
Medicare 26.1% 41.0%
Medicaid 12.7% 11.9%
Private 52.2% 42.7%
Uninsured 3.3% 2.1%
Others 5.6% 2.3%
Teaching hospital 91.3% 94.8% 0.40
Urban hospital 98.9% 99.6% 0.025
Hospital bed size 0.006
Small 2.2% 0.8%
Medium 19.3% 3.7%
Large 78.5% 95.5%
Hospital region 0.69
Northeast 27.8% 20.6%
Midwest 23.2% 34.4%
South 27.4% 23.8%
West 21.7% 21.1%
Co-morbidities
Atrial Fibrillation 26.8% 22.6% 0.13
Hypertension 25.9% 31.6% 0.10
Coronary artery disease 50.2% 46.2% 0.25
Peripheral vascular disease 3.2% 4.3% 0.29
Diabetes 17.2% 15.9% 0.6
Body Mass Index >40 1.0% 4.0% <0.001
Chronic obstructive pulmonary disease 8.7% 7.0% 0.25
Venous thromboembolism 6.9% 7.5% 0.79
Stroke 6.1% 6.9% 0.56
Cirrhosis 2.6% 1.7% 0.13
Chronic anemia 1.5% 3.5% 0.039
Cancer 1.8% 2.2% 0.75
End stage renal disease 2.2% 0.5% 0.001
Malnutrition 9.8% 21.1% 0.003
Peptic ulcer disease 0.6% 0.8% 0.69
Smoker 6.2% 7.4% 0.41

GI = gastrointestinal; LVAD = Left Ventricular Assist Device.

*p <0.05.


Table 2

Characteristics of Left Ventricular Assist Device recipients needing readmission



























































































































































































































































Variable 2005-2007 2008-2010 P value
Total admissions 1180 7542
GI bleed 2.9% 10% <0.001
Age 0.39
18-34 8.5% 8.6%
35-49 21.6% 15.2%
50-64 41.9% 48.1%
≥65 28.0% 28.2%
Males 68.2% 74.8% 0.17
Race 0.88
White 51.2% 53.7%
Black 23.4% 18.2%
Hispanic 5.5% 5.2%
Asian 0% 0.9%
Others 1.0% 2.3%
Unknown 18.5% 19.6%
Insurance 0.65
Medicare 44.1% 51.2%
Medicaid 14.0% 10.7%
Private 39.4% 35.4%
Uninsured 1.3% 0.9%
Others 1.2% 1.7%
Teaching hospital 85.7% 93.1% 0.22
Urban hospital 96.8% 99.7% <0.001
Hospital bed size 0.006
Small 5.8% 0.6%
Medium 16.7% 4.5%
Large 77.5% 94.9%
Hospital region 0.81
Northeast 18.2% 28.0%
Midwest 36.0% 36.2%
South 27.0% 19.5%
West 18.8% 16.4%
Co-morbidities
Atrial Fibrillation 20.3% 23.8% 0.34
Hypertension 42.0% 43.7% 0.76
Coronary artery disease 42.8% 54.6% 0.02
Peripheral vascular disease 3.3% 5.3% 0.23
Diabetes 24.2% 33.2% 0.048
Body Mass Index >40 2.2% 4.5% 0.18
Chronic obstructive pulmonary disease 6.7% 11.6% 0.025
Venous thromboembolism 1.1% 1.7% 0.54
Stroke 7.5% 5.3% 0.16
Cirrhosis 0% 1.8% 0.14
Chronic anemia 5.6% 6.9% 0.49
Cancer 3.7% 2.6% 0.44
End stage renal disease 2.8% 2.6% 0.93
Malnutrition 4.5% 6.1% 0.37
Peptic ulcer disease 0% 1.3% 0.21
Smoker 6.4% 13.3% 0.16

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Nov 27, 2016 | Posted by in CARDIOLOGY | Comments Off on Risk Factors and Outcomes of Gastrointestinal Bleeding in Left Ventricular Assist Device Recipients

Full access? Get Clinical Tree

Get Clinical Tree app for offline access