True Biventricular Pacing Favors Outcomes of Cardiac Resynchronization

The 15th Annual Scientific Meeting

245 The PROTECT In-Hospital Risk Model: 7-Day Outcome in Patients Hospitalized with Acute Heart Failure (AHF) and Renal Dysfunction R.J. Mentz1, M. Fiuzat1, K. Chiswell1, D. Wojdyla1, B. Davison2, J. Cleland3, H.C. Dittrich4, M.M. Givertz5, G. Mansoor4, P. Ponikowski6, J.R. Teerlink7, A.A. Voors8, G. Cotter2, M. Metra9, B.M. Massie7, C.M. O’Connor1; 1Duke, Durham; 2 Momentum Res, Durham; 3Univ of Hull, Hull, United Kingdom; 4Merck Res, Rahway; 5BWH, Boston; 6Med Univ, Wroclaw, Poland; 7UCSF & SF VAMC, San Francisco; 8Univ Med Ctr, Groningen, Netherlands; 9Univ Brescia, Brescia, Italy Background: Predictive models for AHF patients may assist clinicians with patient triage, but short-term prognostic variables and stratification of patients with concomitant renal dysfunction are less well characterized. We aimed to identify the predictors of 7-day outcomes in patients hospitalized with AHF and renal dysfunction. Methods: A predictive model for short-term morbidity and mortality (7-day post-admission death, worsening HF or rehospitalization for HF) was created using variables collected within 24 hours of admission from patients enrolled in the PROTECT trial, a randomized, double-blind, multi-center study of the A1-adenosine receptor antagonist, rolofylline vs. placebo in AHF patients with renal dysfunction (CrCl 20-80 ml/min). Internal validation of the model was performed using bootstrap resampling. Results: 2,015 patients with complete data were included. 14.6% experienced the primary composite endpoint. At multivariate analysis, the strongest predictor of short-term morbidity and mortality was higher blood urea nitrogen level. Predictors of a better outcome were higher values of serum albumin, serum cholesterol, and systolic blood pressure and lower heart rate and respiratory rate as well as the absence of diabetes or hospitalization for HF in the past year (Table). Conclusions: The PROTECT predictive model demonstrates the baseline determinants of short-term morbidity and mortality in AHF patients with renal dysfunction. The current study confirms the role of previously described AHF risk factors and shows the importance of markers of nutritional status (serum albumin and cholesterol levels). The role of nutritional markers for short-term HF risk assessment is a novel finding that warrants further prospective investigations. Table. Multivariate predictors for 7-day death, worsening HF or HF rehospitalization Variable

Unit or levels

HR represents.

HR

95% CI

Chi-Square

P-value

BUN (log) Respiratory Rate SBP Heart Rate Albumin Cholesterol Diabetes Hong for HF

mg/dL breaths/min mmHG bgsm g/dL mg/dL Yes Yes

Doubling in BUN 5 units increase 10 units increase 5 units increase 0.5 units increase 40 units increase Yes vs. No Yes vs No

1.76 1.17 0.90 1.05 0.84 0.88 1.26 1.23

1.48-2.11 1.06-1.30 0.83-0.96 1.01 - 1.09 0.73-0.96 0.78-1.00 0.99-1.60 097-156

39.03 947 9.18 6.53 6.52 3.91 3.63 3.03

!0.0001 !0.01 !0.0l 0.01 0.01 0.05 0.06 0.08



S77

HFSA

and 149 projects ongoing not in response to VA H2H (versus 6 projects at facilities without OL). Conclusions: Facilities with OLs were very successful in implementing the VA H2H initiative. Impact: Use of OLs may be effective in implementing nonmandated QI initiatives to improve care for all VA HF patients.

247 Association of Heart Failure Symptoms Identified by Home Monitoring with Long-Term Mortality and Shock Incidence Paul A. Heidenreich1, John P. Boehmer2, David L. Hayes3, F.R. Gilliam4, John D. Day5, Milan Seth6, Leslie A. Saxon7; 1VA Palo Alto Health Care System, Palo Alto, CA; 2Penn State, Hershey, PA; 3Mayo Clinic, Rochester, MN; 4Cardiology Associates of NE Arkansas, Jonesboro, AR; 5Intermountain Health System, Salt Lake City, UT; 6Boston Scientific, St. Paul, MN; 7USC, Los Angeles, CA Background: The ALTITUDE study group uses data from the LATITUDEÒ remote monitoring system (Boston Scientific) to evaluate patient outcomes. The LATITUDE system allows patients to record the severity of HF-related symptoms. This analysis evaluated the association between baseline HF symptoms and long-term incidence of shock or mortality in patients with ICD or CRT-Ds. Methods: Patients (n55,685 with ICDs and 7,220 with CRT-Ds) entered weekly symptom data between March 2006 and February 2010 using 6 HF symptom questions in the LATITUDE system during a baseline period from 2 to 4 weeks after first transmission. Time 0 for follow-up began at the end of the baseline period. Survival status was obtained from the Social Security Death Index. Kaplan-Meier and Cox regression models were used to determine the association between symptom status and shock or death at up to 3 years of follow-up. Results: In ICD patients, mean age was 66.1 6 13.0 yrs and 73.6% were male; in CRT-D patients, mean age was 69.6 6 11.3 yrs and 71.5% were male. In ICD patients, interval reporting of any symptom predicted shock, and all symptoms except orthopnea also predicted mortality. In CRT-D patients, only new onset faintness/dizziness and nocturnal dyspnea significantly predicted shock, but all symptoms except faintness/dizziness significantly predicted death. Conclusions: In this study, several simple questions collected via remote monitoring identified patients at increased long-term risk of mortality and/or shock and demonstrated differences in the association between baseline symptoms and shock/ death between patients with ICDs and CRT-Ds. Hazard Rates and P values For Subsequent Shock or Death With Worsened Versus Stable HF-Related Symptoms in Patients with ICDs or CRT-Ds Shock Symptom Fatigue Faint/dizzy Edema Exercise capacity Orthopnea Nocturnal dyspnea

ICD 1.25 1.50 1.41 1.38 1.16 1.36

(P!0.01) (P!0.01) (P50.01) (P50.02) (P50.03) (P!0.01)

Mortality CRT-D

1.10 1.14 1.12 1.11 1.10 1.19

(P50.15) (P50.04) (P50.31) (P50.33) (P50.12) (P50.01)

ICD 1.54 1.35 1.89 1.97 1.09 1.74

(P!0.01) (P!0.01) (P!0.01) (P!0.01) (P50.33) (P!0.01)

CRT-D 1.62 1.10 1.60 1.67 1.24 1.78

(P!0.01) (P50.10) (P!0.01) (P!0.01) (P!0.01) (P!0.01)

246 A Blended Facilitation To Implement the VA Hospital-to-Home (H2H) Initiative: CHF QUERI Anju Sahay, Paul A. Heidenreich; CHF QUERI, Dept of Veterans Affairs, Palo Alto, CA Objectives: Reducing readmission rates for heart failure (HF) patients is the primary goal of the Department of Veteran Affairs and its’ Chronic Heart Failure (CHF) QUERI. It is also the goal of Hospital to Home (H2H) national quality improvement (QI) initiative cosponsored by the American College of Cardiology and the Institute for Healthcare. The CHF QUERI has an existing HF Provider Network which also focuses on implementing non-mandated QI initiatives to reduce HF readmission rates. In January 2010 the CHF QUERI through its HF Network launched the “VA H2H” QI initiative to implement the national H2H initiative at the VA facilities with O100 admissions in 2007-2008. We used a blended facilitation approach with the HF Network’s leadership being the “external” facilitators and the opinion leaders (OLs) at each facility being the “internal” facilitators to successfully facilitate the implementation of the VA H2H initiative at the facilities. Methods: We used the sociometric method to identify OLs who would be the “internal” facilitators. We asked the members of the HF Network at 122 facilities to nominate up to 2 providers at own facility (physician/nurse/other/self) as the OL. These OL teams were asked to participate in the VA H2H. As “external” facilitators the HF Network’s leadership conducted web-based meetings and provided tool-kit and consultation to members at all 122 facilities. Periodic surveys were used to track projects (or interventions) recently initiated based on VA H2H, planned based on VA H2H, or ongoing to reduce HF readmissions not based on VA H2H initiative. Results: Regarding identifying the “internal” facilitators members from 66% facilities (n581) responded with 1-2 nominations for OLs. Members from the remaining 34% facilities (n541) either responded with no nominations (n53) or were non-responders (n538). Overall, 47 facilities (39%) reported a total of 243 projects. Among them, 44 facilities had OLs (94%) with 234 projects (96%). In sharp contrast, the remaining 3 facilities (6%) with no OL had only 9 projects (4%). These 44 facilities with OLs reported 37 recently initiated VA H2H projects (versus 3 VA H2H projects at facilities without OL); 40 VA H2H projects being planned (versus 0 project at facilities without OL);

248 True Biventricular Pacing Favors Outcomes of Cardiac Resynchronization Kelly L. Brooke, Tracy A. Webster, Heather Wiste, Samuel J. Asirvatham, David L. Hayes, Yong-Mei Cha; Cardiovascular Diseases, Mayo Clinic, Rochester, MN Background: Maximizing biventricular pacing is required in patients (pts) receiving CRT. Certain contemporary CRT devices have a feature of ventricular sensed response (VSR) intended to provide biventricular pacing after a sensed ventricular event. The sensed ventricular events often occur in pts with atrial fibrillation (AF) with fast ventricular rates and patients with frequent premature ventricular complexes. The extent to which true biventricular pacing (capturing biventricles before the onset of intrinsic QRS) impacts CRT outcomes has not been well studied. Methods: One hundred thirty-one consecutive pts who received CRT were studied. All pts received devices that had the capability of VSR pacing and had this feature programmed on. The percentage of VSR pacing was recorded during regular device follow-up by CareLink or LATITUDE remote monitoring. NYHA class and echocardiography were assessed before and 6-months after CRT. The clinical outcomes of CRT was compared between pts with VSR pacing $ 1% (VSR+ group, n562) and VSR pacing 0% (all true biv pacing, VSR- group, n569). Results: Baseline demographics were similar in NYHA class, LVEF, LV end-diastolic diameter, and the severity of mitral regurgitation, although AF was more common in VSR+ group compared to VSR- group (63% vs. 17%, P!0.001). The median (Q1, Q3) % biventricular pacing (excluding VSR pacing) was 100% (99%, 100%) in VSR- group and 95% (88%, 98%) in VSR+ group (P!0.001). The median (Q1, Q3) % VSR pacing in VSR+ group was 3% (1%, 9%). After CRT, improvement in NYHA class and left ventricular (LV) function was observed in both VSR- and VSR+ groups as shown in table (* P!0.01 compared to pre-CRT). However, LV reverse remodeling was more significant (P50.03), and there was a statistical trend in LVEF improvement (P50.08) in VSR- group compared to VSR+ group. Conclusions: VSR pacing is a beneficial feature of CRT device for improving HF symptoms and LV function. However,

S78 Journal of Cardiac Failure Vol. 17 No. 8S August 2011 maximizing true biventricular pacing (before the onset of QRS) appears to achieve more favorable CRT outcomes compared with VSR pacing (after the onset of QRS).

VSR50% Variable Change Change Change Change Change

(N569) in in in in in

NYHA Class (post-pre) LVEF (post-pre) LVED size (post-pre) PASP (post-pre) MR Severity (post-pre)

-0.8 9.8 -5.6 -4.7 -0.2

6 6 6 6 6

0.8* 12.2* 7.2* 11.7* 0.7*

VSR$1% (N562) -0.6 -6.0 -2.6 -7.2 -0.4

6 6 6 6 6

0.9* 9.7* 5.4* 11.8* 0.5*

P-value 0.19 0.08 0.033 0.34 0.27

follow-up across groups, the analysis was limited to patients with $1 year of follow-up. Hospitalizations, inpatient days, and inpatient costs were compared between patients experiencing SCD vs other causes of death using generalized linear models with adjustment for patient characteristics. Results: 231 patients died after the first 12 months of follow-up: 72 from SCD; 80 due to pump failure, 34 from other CV causes; and 45 patients from non-CV causes. At one year prior to death, SCD patients were younger and had less severe HF compared to patients dying from other causes. They were also significantly less likely to have an ICD (22% vs. 53%, P!0.0001) or a biventricular pacemaker at baseline (8% vs 23%, P50.007) compared to the other groups combined. SCD patients incurred significantly less inpatient care and lower costs compared to patients dying from other causes.

Inpatient Resource Use and Cost in Year Prior to Death

249 Peripartum Cardiomyopathy e Is the Current Definition Too Restrictive? Kismet Rasmusson1, Kim Brunisholz1, Deborah Budge1, Rami Alharethi1, Jenny Connolly1, Benjamin Horne1, Kurt Jensen1, Josef Stehlik2, Abdallah Kfoury1; 1 Heart Failure Prevention and Treatment Program, Intermountain Medical Center, Salt Lake City, UT; 2Heart Failure and Transplant, University Hospital, Salt Lake City, UT Purpose: Peripartum cardiomyopathy (PPCM) is an uncommon condition defined as new symptomatic heart failure of unknown etiology with left ventricular ejection fraction [LVEF] ! 45% diagnosed in the last month of pregnancy and up to the 5th postpartum month. The purpose of this study was to determine how many patients referred with the diagnosis of PPCM met all current defining criteria and whether outcomes were similar in patients who did not. Methods: The Utah PPCM Registry was queried. Patients age $18 years with PPCM were identified by 3 means: direct referral, by ICD-9 codes within our hospital system, or self-referral. Patients were categorized into those meeting (Group 1) vs. not meeting (Group 2) PPCM criteria. Reasons for exclusion included; LVEF O 45%, outside the timeframe, or prior known cardiac history. Outcomes studied included internal cardioverter defibrillators (ICD) use, need for mechanical circulatory support (MCS) and/or heart transplantation, and death. Results: 204 patients with clinical characteristics of PPCM were identified; 108 patients met strict PPCM criteria while 96 did not. Reasons for not meeting PPCM criteria included: LVEF out of range (36.5%); prior cardiac history (36.5%); and the timeframe (13.5%). 13 (13.5%) patients had 2 reasons for not meeting PPCM criteria. 162 patients were referred or identified by ICD-9 codes, and 42 were self-referred. PPCM criteria were met in 50.6 % of referred patients versus 61.9% of self-referred patients. Basic demographics and outcomes studied were uniformly similar between both groups in those with available records (Table).

Demographics and Outcomes Age at Diagnosis (years) Caucasian Gravida Parity Cesarean Section Delivery Subsequent Pregnancy (%) ICD (%) MCS (%) Transplant (%) Death (%)

Group 1 (n588)

Group 2 (n525)

P-value

30.3 6 6.3 84.5% 2.7 6 1.9 (n579) 2.3 6 1.5 (n583) 59% (n578) 3.4% 15.9% 4.5% 5.7% 2.3%

30.4 6 5.7 88.0% 2.9 6 2.2 (n525) 2.5 6 1.9 50% (n522) 16.0% 20.0% 0% 4.0% 0%

0.921 1.00 0.56 0.475 0.453 0.042 0.762 0.574 1.00 1.00

Hospitalized, n (%) Hospitalizations, mean (SD) Inpatient days, mean (SD) Inpatient costs, mean (SD)

SCD (n572)

Pump Failure (n580)

Other CV (n534)

Non-CV (n545)

45 (63) 1.4 (1.6) 11.8 (19.5) $23,328 (48,003)

76 (95)* 3.5 (2.2)* 32.3 (28.5)* $70,224 (85,154)*

31 (91)* 2.4 (1.9)* 42.8 (55.0)* $119,374 (159,398)*

42 (93)* 3.3 (2.2)* 36.0 (37.1)* $76,313 (96,019)*

Reference group is SCD; *P!0.05 Conclusion: After adjustment, inpatient resource use varied between w2-4 fold across modes of death. Assigning rates of resource use by mode of death in an economic model of heart failure is warranted.

251 Regadenoson Stress SPECT MPI for Prediction of Outcomes in Patients with Impaired Left Ventricular Function: A Comparison with Adenosine Stress Salvador Borges-Neto, Linda K. Shaw, Mona Fiuzat, Robert Pagnanelli, Christopher M. O’Connor; Medicine/Cardiology, Duke University, Durham, NC Background: There is no information on the performance of the new A2a receptor agonist stressor Regadenoson in predicting outcomes in patients with decreased LV systolic function. Furthermore, no prognostic comparison data with other stressor agents have been reported. The purpose of this study was to evaluate the prognostic power of Regadenoson in patients with decreased LV function and to compare with Adenosine. Methods: We studied 703 consecutive patients with decreased LV function (LVEF!50%) that underwent rest and either Regadenoson (n5435) or Adenosine stress (n5268) SPECT MPI. Follow-up information to assess mortality and non-fatal myocardial infarction was collected and analyzed over a period of 1 year. Cox proportional hazards models were generated to determine the contribution of the stress MPI results and stress agent in predicting outcomes. One-year KaplanMeier survival curves were generated and stratified by normal/mild abnormal versus moderate/severe abnormal SPECT for comparison between stressors. Results: There were 53 deaths and 71 death or non-fatal myocardial infarction events at 1 year. Overall SPECT MPI study results were significantly associated with all outcomes (LR Chi-square for mortality: 8.08, p5.0045 and mortality/non-fatal MI LR Chisquare: 12.21; p5.0005). Kaplan-Meier time to event curves demonstrated worse outcomes for patients with moderate/severe study results compared to patients with normal/mild abnormal study results.

Conclusions: A significant number of patients labeled as PPCM do not fit strict defining criteria despite otherwise similar clinical characteristics and outcomes. If validated on a larger scale, these results would raise the question whether current defining criteria for PPCM need to be expanded.

250 Resource Use and Medical Costs by Cause of Death in HF-ACTION Yanhong Li1, David J. Whellan2, Mark E. Dunlap3, William E. Kraus1, Gregory Samsa1, Kevin A. Schulman1, Michael Zile4, Shelby D. Reed1; 1Duke Clinical Research Institute, Duke University, Durham, NC; 2Jefferson Medical Center, Philadelphia, PA; 3Case Western Reserve University, Cleveland, OH; 4Medical University of South Carolina, Charleston, OH Background: Economic evaluations of heart failure interventions require accurate estimates of costs and survival. To inform the development of an economic model, we compared patterns of inpatient resource use and costs between patients who died from sudden cardiac death (SCD) and those who died from pump failure, other cardiovascular (CV) causes and non-CV causes over the year prior to death. Hypothesis: We hypothesized that patients who die from pump failure, other CV and non-CV causes incur greater medical resource use and higher costs compared to patients who die from SCD. Methods: The study sample was derived from HF-ACTION, a randomized trial of exercise training in 2331 patients with NYHA Class II-IV. Cause of death was adjudicated by the trial’s endpoint committee. To ensure equal

The log-rank test for difference across groups was p5.0171 for mortality and p5.0056 for the composite death/MI. but no significant difference was documented between the two stress agents (pO0.90 for both endpoints). Furthermore, the addition of stressor type to each of the outcome models did not provide significant