Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

VA Health Systems Research

Go to the VA ORD website
Go to the QUERI website

IIR 04-046 – HSR Study

IIR 04-046
Optimal Use and Cost-Effectiveness of ICDS in the VA Health Care System
Barry M Massie, MD
San Francisco VA Medical Center, San Francisco, CA
San Francisco, CA
Funding Period: July 2005 - September 2011
Until 2004, most implantable cardioverter-defibrillators (ICDs) were implanted in patients who survived episodes of sudden cardiac death (SCD) or other life-threatening arrhythmias (e.g., secondary prevention). Several subsequent trials demonstrated statistically significant, but relatively modest, improvements in survival in both myocardial infarction survivors and patients with heart failure with impaired left ventricular function without documented arrhythmias (primary prevention). These indications increased the pool of patients eligible for ICDs by 5-fold, and ICD implantation rates rose precipitously. As a consequence, ICD implantation costs are projected to grow from $1.5 billion in 2005 to $10-20 billion over 10 years.

Although statistically significant, the survival benefit in the primary prevention trials has been modest, ranging from 7-10 lives saved over 3-5 years per 100 patients implanted, with important subgroups showing no benefit. Furthermore, these clinical trials enrolled populations that differed from the "real-world patients" at high risk for SCD. Trial populations were younger, predominantly men, and had fewer comorbid conditions and less severe heart failure.

Therefore, it was critical that further studies delineate the characteristics of real-world patients who receive ICDs, the subsequent incidence of outcomes among these patients, and patient characteristics predictive of these outcomes. This information is critical to optimizing the utilization of ICDs.

We utilized the unique information available through the VA National Cardiac Device Surveillance Program (VANCDSP) based at the San Francisco VAMC to identify optimal and appropriate candidates for ICD implantation.

Specific aims were:
1. To determine the demographic and clinical characteristics of Veteran ICD recipients, the indications for ICD placement, and the appropriateness of ICD placement in relation to guideline recommendations.

2. To determine patient characteristics that predict low and high rates of appropriate ICD discharges.

3. To identify baseline characteristics that stratify patients based upon their likelihood of benefit from ICD implantation (defined as having received an appropriate ICD intervention followed by a subsequent survival period of at least 6 months) and develop a prediction model incorporating these characteristics.

4. To estimate the cost-effectiveness of ICD implantation in veterans across a broad spectrum of arrhythmia and mortality risk.

5. To compare the characteristics and outcomes of patients receiving ICDs within the VA system to those of the broader U.S. health care system, using available registry data.

We conducted a prospective cohort study by utilizing VA Health Information systems to phenotype the characteristics and clinical outcomes of nearly 5,000 Veterans who underwent ICD implantation and were enrolled in the VANCDSP between 2003 and 2009. This cohort was not subject to the restrictive entry criteria or selection biases of randomized clinical trials or to either the reimbursement restrictions or the financial incentives that apply to non-VA patients.

The following was collected from review of electronic medical records, periodic downloading of stored ICD data with independent adjudication, and available VA databases to track patient outcomes: information concerning patients' demographic and clinical characteristics; indication for ICD placement; ICD type, programming, and activity; clinical outcomes; and resource utilization.

Multi-predictor regression models were used to assess secular trends in patient characteristics and indications for ICD placement, as well as to determine independent predictors of clinical outcomes. Cox and tree-structured survival methods were used to develop prediction rules with potential policy implications, using generalized cross-validation methods to obtain reliable estimates of the likely prediction error of the derived rules. Cost-effectiveness will be assessed in patients across a range of indications and risk predictors by projecting life expectancy and lifetime cost of care using actual survival and cost information from the period of observation.

Approximately 5,000 patients met our entry criteria: namely, implantation of a first ICD within the VA system with enrollment in the VANCDSP monitoring program within 3 months of implantation. This did not include veterans whose devices were implanted at non-VA hospitals or at VA hospitals not yet participating in VANCDSP.

We found that the overwhelming majority (89%) of devices were implanted for primary prevention of SCD. As anticipated, the age and level of comorbidity of VA recipients was substantially higher than that of patients entered into randomized trials. After a mean follow-up of 2.5 years, we observed a rate of all-cause mortality of 6.5 per 100 person-years. The 11% mortality we observed at two years compares well with the range of 2-year mortality observed in the major randomized trials of primary prevention ICD therapy: MADIT (12%), MADIT-2 (16%), and SCD-HeFT (11%), despite older subjects and greater burden of comorbidity among veterans with ICDs. Remarkably, the 4-year mortality rates in our cohort were less than 26%.

Mortality was much higher among patients with more severe underlying cardiac disease and comorbid conditions. Mortality by NYHA functional class ranged from 3.9 per 100 person-years (Class I) to 15.7 per 100 person-years (Class IV) (p < 0.0001). After adjusting for potential confounders, Class IV symptoms were associated with nearly four-fold greater mortality (hazard ratio 3.9; 95% CI, 1.9 - 7.8; p < 0.001) compared with Class I (very mild symptoms). Similarly, hospitalization within the preceding 6 months was associated with higher mortality (HR, 1.7; 95% CI, 1.3-2.3; p < 0.001). Patients with chronic kidney disease had a much higher risk of death (HR, 1.7; 95% CI, 1.3 - 2.1; p < 0.001).

In analyses examining guideline concordance of indications for implantation of both ICDs and cardiac resynchronization (CRT) ICDs, there was less overuse in the VA system compared with non-VA data derived from the National Cardiovascular Data Registry (NCDR, a national registry). In NCDR, 22.5% of all ICDs were found not to meet guideline indications, compared to 12.8% of ICDs implanted in the VA. Similarly, 23.7% of non-VA implanted CRT ICDs were outside clinical guidelines; "off-label" devices represented only 19.6% of VA recipients.

Our findings, derived from a population-based cohort of unselected Veterans receiving ICDs with robust long-term clinical and device performance data, have confirmed that veterans receiving these expensive devices are older and have a greater burden of comorbidity than trial populations. These characteristics among veterans were associated with higher mortality, but surprisingly, overall mortality in our cohort was similar to that in trials enrolling less ill subjects. Importantly, we found that there was less overuse of these devices in the VA system than in the fee-for-service population evaluated by NCDR, and that rates of use of optimal medical therapy were extremely high, indicative of the high quality of care among veterans receiving ICDs.

Importantly, this well-phenotyped cohort has created a unique data resource that will be available for future analyses, and provides a data standard for future care of veterans with arrhythmia devices. As part of the highly successful CART-CL program that is currently employed VA-wide to document the care of greater than 30,000 cardiac catheterization and coronary intervention procedures annually, the data standard and data collection processes established in this study have already heavily informed the structure of CART-EP, the next generation VA clinical documentation platform for the comprehensive care of patients with implantable arrhythmia devices, expected to be implemented in the VA system in 2012-2013.

External Links for this Project

Dimensions for VA

Dimensions for VA is a web-based tool available to VA staff that enables detailed searches of published research and research projects.

Learn more about Dimensions for VA.

VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address.
    Search Dimensions for this project


None at this time.

DRA: Health Systems, Cardiovascular Disease
DRE: Treatment - Observational, Technology Development and Assessment
Keywords: Cardiac procedures, Cost, Research measure
MeSH Terms: none

Questions about the HSR website? Email the Web Team

Any health information on this website is strictly for informational purposes and is not intended as medical advice. It should not be used to diagnose or treat any condition.