[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure 1.
Risk-Standardized Annual All-Cause Mortality Rates Among Patients With Ischemic Heart Disease (IHD) Across Veterans Affairs Medical Centers (VAMCs) (N = 138), 2010-2014
Risk-Standardized Annual All-Cause Mortality Rates Among Patients With Ischemic Heart Disease (IHD) Across Veterans Affairs Medical Centers (VAMCs) (N = 138), 2010-2014

The x-axis denotes each VAMC ranked from lowest to highest risk-standardized IHD survival, and the y-axis indicates survival rates. The 95% confidence intervals for each hospital’s IHD survival estimate are indicated by the vertical whiskers.

Figure 2.
Risk-Standardized Annual All-Cause Mortality Rates Among Patients With Chronic Heart Failure (CHF) Across Veterans Affairs Medical Centers (VAMCs) (N = 138), 2010-2014
Risk-Standardized Annual All-Cause Mortality Rates Among Patients With Chronic Heart Failure (CHF) Across Veterans Affairs Medical Centers (VAMCs) (N = 138), 2010-2014

The x-axis denotes each VAMC ranked from lowest to highest risk-standardized CHF survival, and the y-axis indicates survival rates. The 95% confidence intervals for each hospital’s CHF survival estimate are indicated by the vertical whiskers.

Figure 3.
Comparisons With Published Hospital-Level Quality Measures
Comparisons With Published Hospital-Level Quality Measures

A, Correlation between the hospital-level ischemic heart disease (IHD) mortality calculated in the study (y-axis) and the 30-day posthospitalization mortality for acute myocardial infarction (x-axis). B, Correlation between the hospital-level chronic heart failure (CHF) mortality calculated in the study (y-axis) and the 30-day posthospitalization mortality for CHF (x-axis). C, Correlation between each hospital’s standardized, composite survival rate for IHD and CHF (y-axis) and the Department of Veterans Affairs’ (VA) Strategic Analytics for Improvement and Learning (SAIL) star rating system (x-axis).

Table 1.  
Characteristicsa of IHD and CHF Cohorts
Characteristicsa of IHD and CHF Cohorts
Table 2.  
Composite Standardized Survivala for IHD and CHF by VA Hospital Location and Complexityb
Composite Standardized Survivala for IHD and CHF by VA Hospital Location and Complexityb
1.
Kochanek  KD, Murphy  SL, Xu  J, Tejada-Vera  B. Deaths: final data for 2014. Centers for Disease Control and Prevention. https://www.cdc.gov/nchs/data/nvsr/nvsr65/nvsr65_04.pdf. Published June 30, 2016. Accessed December 14, 2017.
2.
Jha  AK, Perlin  JB, Kizer  KW, Dudley  RA.  Effect of the transformation of the Veterans Affairs Health Care System on the quality of care.  N Engl J Med. 2003;348(22):2218-2227.PubMedGoogle ScholarCrossref
3.
VA Office of Quality, Safety, and Value. 2012 VHA facility quality and safety report. https://www.va.gov/HEALTH/docs/2012_VHA_Facility_Quality_and_Safety_Report_FINAL508.pdf. Published 2012. Accessed December 4, 2017.
4.
Centers for Medicare and Medicaid Services. Hospital Compare overview. https://www.medicare.gov/hospitalcompare/about/what-is-HOS.html. Published 2017. Accessed December 14, 2017.
5.
Ashton  CM, Petersen  NJ, Souchek  J,  et al.  Geographic variations in utilization rates in Veterans Affairs hospitals and clinics.  N Engl J Med. 1999;340(1):32-39.PubMedGoogle ScholarCrossref
6.
Sohn  MW, Arnold  N, Maynard  C, Hynes  DM.  Accuracy and completeness of mortality data in the Department of Veterans Affairs.  Popul Health Metr. 2006;4:2.PubMedGoogle ScholarCrossref
7.
Elixhauser  A, Steiner  C, Harris  DR, Coffey  RM.  Comorbidity measures for use with administrative data.  Med Care. 1998;36(1):8-27.PubMedGoogle ScholarCrossref
8.
Keenan  PS, Normand  SL, Lin  Z,  et al.  An administrative claims measure suitable for profiling hospital performance on the basis of 30-day all-cause readmission rates among patients with heart failure.  Circ Cardiovasc Qual Outcomes. 2008;1(1):29-37.PubMedGoogle ScholarCrossref
9.
Krumholz  HM, Lin  Z, Drye  EE,  et al.  An administrative claims measure suitable for profiling hospital performance based on 30-day all-cause readmission rates among patients with acute myocardial infarction.  Circ Cardiovasc Qual Outcomes. 2011;4(2):243-252.PubMedGoogle ScholarCrossref
10.
American Community Survery (ACS). https://www.census.gov/programs-surveys/acs/. Published 2018. Accessed March 5, 2018.
11.
Krumholz  HM, Wang  Y, Mattera  JA,  et al.  An administrative claims model suitable for profiling hospital performance based on 30-day mortality rates among patients with an acute myocardial infarction.  Circulation. 2006;113(13):1683-1692.PubMedGoogle ScholarCrossref
12.
Ash  AS, Fienberg  SE, Louis  TA, Normand  SL, Stukel  TA, Utts  J. Statistical issues in assessing hospital performance. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/Downloads/Statistical-Issues-in-Assessing-Hospital-Performance.pdf. Published 2012. Accessed November 20, 2017.
13.
Finkelstein  DM.  A proportional hazards model for interval-censored failure time data.  Biometrics. 1986;42(4):845-854.PubMedGoogle ScholarCrossref
14.
U.S. Department of Veteran Affairs. End of year hospital star rating (FY2016). https://www.va.gov/qualityofcare/measure-up/End_of_Year_Hospital_Star_Rating.asp. Published 2016. Accessed December 8, 2017.
15.
Menendez  ME, Neuhaus  V, van Dijk  CN, Ring  D.  The Elixhauser comorbidity method outperforms the Charlson index in predicting inpatient death after orthopaedic surgery.  Clin Orthop Relat Res. 2014;472(9):2878-2886.PubMedGoogle ScholarCrossref
16.
Silber  JH, Rosenbaum  PR, Brachet  TJ,  et al.  The Hospital Compare mortality model and the volume-outcome relationship.  Health Serv Res. 2010;45(5, pt 1):1148-1167.PubMedGoogle ScholarCrossref
17.
Borzecki  AM, Christiansen  CL, Loveland  S, Chew  P, Rosen  AK.  Trends in the inpatient quality indicators: the Veterans Health Administration experience.  Med Care. 2010;48(8):694-702.PubMedGoogle ScholarCrossref
18.
Ho  PM, Luther  SA, Masoudi  FA,  et al.  Inpatient and follow-up cardiology care and mortality for acute coronary syndrome patients in the Veterans Health Administration.  Am Heart J. 2007;154(3):489-494.PubMedGoogle ScholarCrossref
19.
Keyhani  S, Cheng  E, Arling  G,  et al.  Does the inclusion of stroke severity in a 30-day mortality model change standardized mortality rates at Veterans Affairs hospitals?  Circ Cardiovasc Qual Outcomes. 2012;5(4):508-513.PubMedGoogle ScholarCrossref
20.
Wheeler  S, Bowen  JD, Maynard  C,  et al.  Women veterans and outcomes after acute myocardial infarction.  J Womens Health (Larchmt). 2009;18(5):613-618.PubMedGoogle ScholarCrossref
21.
Bush  RL, DePalma  RG, Itani  KM, Henderson  WG, Smith  TS, Gunnar  WP.  Outcomes of care of abdominal aortic aneurysm in Veterans Health Administration facilities: results from the National Surgical Quality Improvement Program.  Am J Surg. 2009;198(5)(suppl):S41-S48.PubMedGoogle ScholarCrossref
22.
Khuri  SF, Daley  J, Henderson  WG.  The comparative assessment and improvement of quality of surgical care in the Department of Veterans Affairs.  Arch Surg. 2002;137(1):20-27.PubMedGoogle ScholarCrossref
23.
Smith  T, Li  X, Nylander  W, Gunnar  W.  Thirty-Day postoperative mortality risk estimates and 1-year survival in veterans health administration surgery patients.  JAMA Surg. 2016;151(5):417-422.PubMedGoogle ScholarCrossref
24.
Groeneveld  PW, Kruse  GB, Chen  Z, Asch  DA.  Variation in cardiac procedure use and racial disparity among Veterans Affairs Hospitals.  Am Heart J. 2007;153(2):320-327.PubMedGoogle ScholarCrossref
25.
Chen  Q, Hanchate  A, Shwartz  M,  et al.  Comparison of the Agency for Healthcare Research and Quality patient safety indicator rates among veteran dual users.  Am J Med Qual. 2014;29(4):335-343.PubMedGoogle ScholarCrossref
26.
Trivedi  AN, Grebla  RC.  Quality and equity of care in the veterans affairs health-care system and in medicare advantage health plans.  Med Care. 2011;49(6):560-568.PubMedGoogle ScholarCrossref
27.
Trivedi  AN, Matula  S, Miake-Lye  I, Glassman  PA, Shekelle  P, Asch  S.  Systematic review: comparison of the quality of medical care in Veterans Affairs and non-Veterans Affairs settings.  Med Care. 2011;49(1):76-88.PubMedGoogle ScholarCrossref
28.
Donabedian  A.  The Definition of Quality and Approaches to its Assessment. Ann Arbor, MI: Health Administration Press; 1980.
29.
DesHarnais  SI. The outcome model of quality. In: Sollecito  WA, Johnson  JK, eds.  McLaughlin and Kaluzny’s Continuous Quality Improvement in Health Care. 4th ed. Burlington, MA: Jones & Bartlett; 2013:155-180.
30.
Health  MP. The role of the hospital. https://www.aha.org/ahahret-guides/2012-04-01-managing-population-health-role-hospital. Published 2016. Accessed March 8, 2018.
31.
Stiefel  M, Nolan  K.  A Guide to Managing the Triple Aim: Population Health, Experience of Care, and Per Capita Cost. Cambridge, MA: Institute for Healthcare Improvement; 2012.
32.
Alter  DA, Naylor  CD, Austin  P, Tu  JV.  Effects of socioeconomic status on access to invasive cardiac procedures and on mortality after acute myocardial infarction.  N Engl J Med. 1999;341(18):1359-1367.PubMedGoogle ScholarCrossref
Original Investigation
July 2018

Outcomes of Care for Ischemic Heart Disease and Chronic Heart Failure in the Veterans Health Administration

Author Affiliations
  • 1Department of Veterans Affairs’ Center for Health Equity Research and Promotion, Corporal Michael J. Crescenz Veterans Affairs Medical Center, Philadelphia, Pennsylvania
  • 2Division of General Internal Medicine, Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia
  • 3Leonard Davis Institute of Health Economics, University of Pennsylvania, Philadelphia
  • 4Cardiovascular Outcomes, Quality, and Evaluative Research Center, University of Pennsylvania, Philadelphia
  • 5Jazz Pharmaceuticals Inc, Philadelphia, Pennsylvania
  • 6Medicus Economics, LLC, Milton, Massachusetts
JAMA Cardiol. 2018;3(7):563-571. doi:10.1001/jamacardio.2018.1115
Key Points

Question  Are there meaningful differences in the mortality rates across Department of Veterans Affairs medical centers for patients receiving care for ischemic heart disease and chronic heart failure?

Findings  In this cohort study, among the 930 079 veterans with ischemic heart disease and the 348 015 veterans with chronic heart failure receiving care at 138 Veterans Affairs medical centers from 2010 to 2014, annual risk-standardized mortality rates varied from 5.5% to 9.4% for ischemic heart disease and 11.1% to 18.9% for chronic heart failure; both differences were statistically significant.

Meaning  There is substantial variation in risk-standardized cardiovascular mortality rates across Department of Veterans Affairs medical centers, potentially signaling differences in the quality of cardiovascular health care provided by these hospitals.

Abstract

Importance  The Department of Veterans Affairs (VA) operates a nationwide system of hospitals and hospital-affiliated clinics, providing health care to more than 2 million veterans with cardiovascular disease. While data permitting hospital comparisons of the outcomes of acute cardiovascular care (eg, myocardial infarction) are publicly available, little is known about variation across VA medical centers (VAMCs) in outcomes of care for populations of patients with chronic, high-risk cardiovascular conditions.

Objective  To determine whether there are substantial differences in cardiovascular outcomes across VAMCs.

Design, Setting, and Participants  Retrospective cohort study comprising 138 VA hospitals and each hospital’s affiliated outpatient clinics. Patients were identified who received VA inpatient or outpatient care between 2010 and 2014. Separate cohorts were constructed for patients diagnosed as having either ischemic heart disease (IHD) or chronic heart failure (CHF). The data were analyzed between June 24, 2015, and November 21, 2017.

Exposures  Hierarchical linear models with VAMC-level random effects were estimated to compare risk-standardized mortality rates for IHD and for CHF across 138 VAMCs. Mortality estimates were risk standardized using a wide array of patient-level covariates derived from both VA and Medicare health care encounters.

Main Outcomes and Measures  All-cause mortality.

Results  The cohorts comprised 930 079 veterans with IHD and 348 015 veterans with CHF; both cohorts had a mean age of 77 years and were predominantly white (IHD, n = 822 665 [89%] and CHF, n = 287 871 [83%]) and male (IHD, n = 916 684 [99%] and CHF n = 341 352 [98%]). The VA-wide crude annual mortality rate was 7.4% for IHD and 14.5% for CHF. For IHD, VAMCs’ risk-standardized mortality varied from 5.5% (95% CI, 5.2%-5.7%) to 9.4% (95% CI, 9.0%-9.9%) (P < .001 for the difference). For CHF, VAMCs’ risk-standardized mortality varied from 11.1% (95% CI, 10.3%-12.1%) to 18.9% (95% CI, 18.3%-19.5%) (P < .001 for the difference). Twenty-nine VAMCs had IHD mortality rates that significantly exceeded the national mean, while 35 VAMCs had CHF mortality rates that significantly exceeded the national mean. Veterans Affairs medical centers’ mortality rates among their IHD and CHF populations were not associated with 30-day mortality rates for myocardial infarction (R2 = 0.01; P = .35) and weakly associated with hospitalized heart failure 30-day mortality (R2 = 0.16; P < .001) and the VA’s star rating system (R2 = 0.06; P = .005).

Conclusions and Relevance  Risk-standardized mortality rates for IHD and CHF varied widely across the VA health system, and this variation was not well explained by differences in demographics or comorbidities. This variation may signal substantial differences in the quality of cardiovascular care between VAMCs.

Introduction

Ischemic heart disease (IHD) and chronic heart failure (CHF) are highly prevalent in the Veterans Affairs (VA) health care system, and mortality is high for both conditions.1 For more than 20 years, the VA has attempted to measure and improve health care quality for veterans with these conditions,2 yet most of these efforts have focused on either process measures of quality (eg, β-blockers for heart failure) or surrogate clinical outcomes (eg, hypertension treatment targets).3 Very limited data are available on clinical outcomes, and many of these outcome measures are restricted to patients hospitalized with acute cardiovascular events such as myocardial infarction or CHF exacerbations.4 It is not clear whether short-term hospital outcomes among selected subsets of acutely ill patients are predictive of longer-term outcomes among the much larger IHD and CHF patient populations receiving VA continuity care.

The VA currently comprises 144 VA medical centers (VAMCs), with at least 1 VAMC located in every state and the District of Columbia. Each VAMC manages a local network of outpatient clinics, located at the main hospital and/or in surrounding communities. The VA’s health care system is highly integrated; VAMCs share a common electronic medical record and have uniform organizational characteristics such as governance, a national formulary, and national practice guidelines.2

While a highly integrated system would be expected to have little systemwide variation in health outcomes, prior studies have suggested such variation occurs.5 Therefore, the goal of our study was to examine risk-standardized mortality rates among patients with IHD and/or CHF across the VA from 2010 to 2014 and to discern whether there are meaningful differences in VAMCs’ mortality rates, which might imply underlying differences in the quality of health care delivered by VAMCs.

Methods
Data Sources

We obtained administrative health care data from October 1, 2009, through September 30, 2014, from the VA’s corporate data warehouse, which contains detailed information on all inpatient, outpatient, laboratory, and pharmacy encounters throughout the VA health care system. For veterans coenrolled in the Medicare program, we also obtained Medicare enrollment information and inpatient/outpatient claims submitted by (non-VA) health care clinicians from 2009 to 2014.

Cohort Selection

Using VA data, we identified veterans as having IHD if they had at least 1 inpatient or outpatient administrative record with International Classification of Diseases, Ninth Revision (ICD-9) codes 410.x, 411.x, 414.00-414.07, 414.12, 414.2-414.4, or 414.8-414.9. We likewise identified veterans as having CHF if they had at least 1 administrative record with ICD-9 codes 398.91, 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.13, 404.91, 404.93, 425.4, 428.0, 428.1, 428.20-428.23, 428.30-428.33, 428.40-428.43, or 428.9. Patients whose earliest qualifying diagnosis occurred between October 1, 2009, and March 31, 2010, and who were alive on April 1, 2010, were entered into the relevant cohort (ie, IHD or CHF) in the first calendar quarter of our 4.5-year outcomes observation window (ie, April 1, 2010, to September 30, 2014). This ensured that all patients in the cohort had a minimum of 6 months of prior health care data to permit accurate coding of comorbidities and antecedent health events. Patients whose earliest qualifying diagnosis occurred on or after April 1, 2010, but before April 1, 2014, entered the cohort during the calendar quarter of that “index” diagnosis. Cohort entry ceased on April 1, 2014. The IHD and CHF cohorts were constructed independently; the same patient could appear in both cohorts.

Outcomes

We assessed the VA’s vital status master file to ascertain deaths occurring from April 1, 2010, through September 30, 2014. This data set incorporates death records from the VA’s Beneficiary Identification and Records Locator Subsystem database as well as the Social Security Death Master File; it is considered a highly complete record of deaths among VA-enrolled veterans.6

Demographics, Non-VA Care, Comorbidities, Prior Events, and Socioeconomic Status

We obtained each veteran’s age, race/ethnicity, and sex from the VA’s enrollment database. When VA race/ethnicity data were absent, Medicare’s enrollment race/ethnicity data were used. For veterans who were Medicare beneficiaries, Medicare claims data were used to estimate the proportion of health care that was delivered by non-VA clinicians. Comorbidities were assessed using diagnosis codes in either VA or Medicare administrative claims, including the Elixhauser comorbidities7 as well as additional Hierarchical Condition Classifications that are typically incorporated in hospital comparison models of cardiovascular disease care,8,9 specifically cerebrovascular disease, dementia, dialysis, and (for the CHF cohort only) IHD. We also assessed each patient’s VA and Medicare data for the occurrence of major medical/surgical events, including myocardial infarction/acute coronary syndrome, coronary revascularization, pneumonia, and stroke. Veterans’ zip codes were used to estimate community-level per capita income using data from the 2011 to 2015 American Community Survey.10

Outcome Variable Coding

Each patient’s health care data from each of the 18 calendar quarters from April 2010 through September 2014 were aggregated quarterly, and these aggregated data were analyzed as a patient-quarter–level discrete-time survival analysis, with the occurrence of death during a calendar quarter as the primary outcome variable.

Predictor Variable Coding

Each patient’s comorbidity vector of binary predictor variables was updated quarterly based on the VA and Medicare health care encounters occurring during that quarter. Each patient’s chronic comorbid conditions (eg, diabetes or paralysis) were assumed to be present perpetually after their onset, unless the condition and a related condition were mutually exclusive (eg, tumor without metastases vs metastatic cancer). Because recent prior medical/surgical events are strong predictors of clinical outcomes,11 the occurrence of selected major medical/surgical events in the prior 2 calendar quarters was assessed for each patient and updated quarterly.

VAMC Assignment and Censoring

For conceptual simplicity, we attributed to each VAMC all the health care provided by either the main hospital or its affiliated outpatient clinics. Within each calendar quarter, a patient was assigned to his/her primary VAMC where he/she had received the plurality of his/her VA health care, quantified as inpatient hospital days plus outpatient clinic visits. In cases where a patient had more than 1 VAMC with the largest number of encounters, he/she was assigned to the VAMC that provided care closest in time to the calendar quarter’s end. When a patient had no VA encounters during a quarter, the veteran’s VAMC assignment was carried over from the previous calendar quarter. Patients were censored from the cohort after 2 consecutive calendar quarters (ie, >6 months) of no VA encounters. Censored patients who subsequently reinitiated care at a VAMC were readded to the study cohorts during the quarter when their VA health care resumed.

Statistical Analysis

We estimated hierarchical generalized linear models with VAMC random effects, analogous to the methods used by the Centers for Medicare and Medicaid Services (CMS) in their Hospital Compare methods.11,12 These models included annual fixed effects to control for temporal changes in outcomes independent of the location of care as well as an indicator of being in the cohort at inception (the first quarter necessarily included patients who had longstanding CHF or IHD, while patients joining the cohort in subsequent quarters were more likely to have incipient disease). Models used a complementary log-log linking function, as is typical for interval-censored survival models.13 All statistical tests were 2-sided, with P less than .05 considered significant.

The VAMC-level random effects estimated by the hierarchical model were used to calculate a risk-standardized mortality ratio for each VAMC, using CMS’s Hospital Compare formula (ie, the VAMC’s predicted mortality as calculated from the observed outcome rate, with Bayesian shrinkage applied to the estimate to account for instability of rate estimators in small samples, divided by the expected mortality as derived from the observed characteristics of each VAMC’s patients).11 Confidence intervals for each VAMC’s risk-standardized mortality estimate were derived using bootstrap replications (n = 380 for CHF and n = 384 for IHD, thus assuring each individual VAMC was included in at least 200 replications) of the patient-level data set with random selection at the VAMC level, following the methods described by Ash et al.12

We then compared our IHD population outcomes with published 30-day myocardial infarction survival for VAMCs from CMS’s Hospital Compare program (2012-2014),4 and we similarly compared our CHF population outcomes with Hospital Compare’s published 30-day survival after CHF hospitalizations. We next calculated a single measure of each VAMC’s cardiovascular population outcomes by normalizing each VAMC’s survival rate for IHD (ie, subtracting the national mean survival rate and dividing by the national standard deviation) as well as for CHF and taking the mean of these 2 measures to produce a composite standardized survival rate. This composite measure then was compared with each VAMC’s star rating as reported on the VA’s Strategic Analytics for Improvement and Learning system,14 a summary quality measure similar to the 5-star system used by CMS.

The institutional review board of the Corporal Michael J. Crescenz VAMC (Philadelphia, Pennsylvania) approved the study protocol. Because it was infeasible to obtain informed consent from more than 1 million veterans, a waiver was granted by the institutional review board. All analyses were conducted using SAS, version 9.4 (SAS Institute Inc) and Stata, version 15.0 (StataCorp).

Results

We excluded veterans at 6 VAMCs with fewer than 200 qualifying patients because low numbers made accurate mortality estimation infeasible. We identified 930 079 veterans with IHD and 348 015 veterans with CHF (Table 1) receiving care at 138 VAMCs. Veterans with IHD were predominantly white and men, with sizeable percentages having hypertension (n = 762 149 [82%]), diabetes (n = 454 372 [49%]), chronic pulmonary disease (n = 213 064 [23%]), and heart failure (n = 156 915 [17%]). Veterans with CHF were slightly more likely to be women and/or not white than IHD veterans, with high prevalence of hypertension (n = 293 752 [84%]), diabetes (n = 211 709 [61%]), chronic pulmonary disease (n = 124 904 [36%]), and chronic kidney disease (n = 91 548 [26%]). There were 266 406 veterans who were members of both disease cohorts. Both the IHD and CHF survival models had mortality hazard ratios for chronic diseases that were similar to prior studies using similar predictors (eTable in the Supplement).15

Numbers of Patients per VAMC

The number of unique patients with IHD per VAMC ranged from 1060 at the lowest-volume VAMC to 19 955 at the highest-volume VAMC. The number of unique patients with CHF per VAMC ranged from 335 at the lowest-volume VAMC to 7917 at the highest-volume VAMC. The ratio of patients with CHF to patients with IHD at the 138 VAMCs varied from 18% to 52%. Although Bayesian shrinkage methods can obscure associations between case volume and outcomes,16 the sizeable case volume at each VAMC in our cohorts, even at low-volume centers, minimized the effect of shrinkage on mortality estimates at low-volume centers.

Ischemic Heart Disease Mortality Across VAMCs

The overall annual mortality rate in the IHD cohort was 7.4%. Risk-standardized IHD mortality varied across the 138 VAMCs from 5.5% (95% CI, 5.2%-5.7%) at the lowest-mortality VAMC to 9.4% (95% CI, 9.0%-9.9%) at the highest-mortality VAMC (Figure 1; P < .001 for the difference). For 29 high-mortality VAMCs, the lower 95% confidence limit exceeded 7.4%, suggesting these VAMCs’ IHD mortality rates were statistically significantly higher than the national mean. Conversely, for 36 low-mortality VAMCs, the upper 95% confidence limit was less than 7.4%, similarly suggesting these VAMCs’ IHD mortality rates were statistically significantly lower than average.

Chronic Heart Failure Mortality Across VAMCs

The overall annual mortality rate in the CHF cohort was 14.5%. Risk-standardized CHF mortality varied across the 138 VAMCs from 11.1% (95% CI, 10.3%-12.1%) at the lowest-mortality VAMC to 18.9% (95% CI, 18.3%-19.5%) at the highest-mortality VAMC (Figure 2; P < .001 for the difference). The lower 95% confidence limit for 35 high-mortality VAMCs exceeded 14.5%, suggesting these VAMCs’ CHF mortality rates were statistically significantly higher than the national mean. Conversely, the 95% upper confidence limit for 37 low-mortality VAMCs also was less than 14.5%, similarly suggesting these VAMCs’ CHF mortality rates were statistically significantly lower than average.

Association of Patient Attribution With Mortality

Risk-standardized mortality between the IHD and CHF cohorts was highly correlated (R2 = 0.70; P < .001), as expected because of the overlapping membership between cohorts. Our method of attributing veterans to VAMCs conceivably could have penalized “hub” VAMCs that received referrals of veterans with advanced disease at the end of life from smaller VAMCs. However, only 2.4% of IHD deaths (n = 5284) and only 2.3% of CHF deaths (n = 3008) occurred during a calendar quarter in which the veteran’s VAMC had been reassigned. These low percentages suggest that our patient-attribution scheme had minimal effect on our results.

VAMC Characteristics and Composite Outcomes

Mean composite IHD-CHF standardized survival scores (ie, z scores) were slightly higher in the East (z = 0.51) and Midwest (z = 0.53) Census regions compared to the West (z = −0.29) and South (z = −0.58) regions (P < .001) (Table 2).3 Conversely, we found no association between a VAMC’s complexity (a VAMC categorization system based on patient volume, patient risk, the number and breadth of physician specialists, and the presence of academic programs3) and the composite IHD/CHF standardized survival rate.

Comparisons With Other VA Hospital Quality Measures

A VAMC’s risk-standardized IHD mortality was not associated with 30-day acute myocardial infarction posthospitalization mortality (R2 = 0.01; P = .35) (Figure 3). There was weak association between VAMCs’ risk-standardized CHF mortality and 30-day CHF posthospitalization mortality (R2 = 0.16; P < .001). We also observed a weak (R2 = 0.06) but statistically significant (P = .005) association between our composite cardiovascular outcome measure and the VA’s Strategic Analytics for Improvement and Learning star system for quality measurement.

Discussion

We observed marked variation in mortality rates across 138 VAMCs for both IHD and CHF risk-standardized mortality. Ischemic heart disease annual death rates at the VAMC with the highest mortality were 3.9 percentage points larger than at the VAMC with the lowest mortality (ie, 1 excess death per year on average among every 26 patients with IHD at the highest-mortality VAMC compared with patients with IHD at the lowest-mortality hospital), and similarly, CHF annual death rates were 7.8 percentage points larger (ie, 1 excess death per year among every 13 patients with CHF at the highest-mortality VAMC compared with patients with CHF at the lowest-mortality VAMC). We also found only modest association between our mortality measures and several published measures of VAMC quality, suggesting that our population-level outcomes measures may be capturing dimensions of chronic disease care that are not well measured by either short-term posthospitalization mortality measures or by aggregate measures such as the VA’s Strategic Analytics for Improvement and Learning star system.

Prior studies of VA hospital variation in cardiovascular outcomes have generally focused on the outcomes of acute hospitalization for severe illness17-20 or for major surgery.21-23 While these outcomes are important for assessing hospitals’ quality of care for the most acutely ill patients, most patients receiving care from the VA each year are not hospitalized nor do they undergo major surgery.24 Mortality rates among broader populations of chronic cardiovascular disease patients receiving care by health care systems therefore can provide important insights into health system performance, including how health systems care for their most medically and socially vulnerable patients. Furthermore, wide variation across VAMCs in risk-adjusted mortality among these populations may be an indicator of variation in the cardiovascular care quality provided by these hospitals.

Since the 1990s, the VA’s leadership has intently focused on quality of care,2 and the VA has often demonstrated impressive quality performance, particularly in comparison with non-VA health systems.25-27 However, much of VA’s quality measurement efforts and reporting have focused on either process measures of quality or on surrogate clinical outcomes.3,14 While process measures are undeniably important in the assessment of quality of care,28 these are only important to the extent that they are associated with the clinical outcomes that matter to patients.29 Measurement of risk-adjusted mortality among chronic disease populations therefore presents an important dimension of quality measurement that might be missed if process measures or acute-care outcomes were the sole metrics.

Differences in mortality rates among VA chronic cardiovascular disease populations may reflect differences across medical centers in the quality of care for these conditions, including clinician adherence to evidence-based treatment and screening guidelines, access to care for urgent medical conditions, posthospitalization care protocols, chronic disease management programs, and access to specialty care, social work services, and behavioral health care, as well as the integration of cardiovascular disease care with the treatment of other common concurrent chronic diseases. Our finding that cardiovascular mortality was not associated with the complexity of the VAMC providing health care provides some reassurance that VAMCs with fewer specialty services are able to deliver comparable care to more complex VAMCs, suggesting that the VA’s internal system of referral for complex cardiovascular care is functioning adequately. However, this general finding does not eliminate the possibility that programmatic, organizational, and/or structural differences between VAMCs may have contributed to mortality differences.

Our findings may be placed in the context of increasing emphasis on the importance of hospital/health system quality performance in the care of populations of patients with chronic disease.30,31 While hospital outcomes for the treatment of acute medical episodes, such as myocardial infarction, remains an important quality measure, mortality rates among the broader population of patients receiving care by a hospital and/or its outpatient clinics provides important insight into the effectiveness of the hospital/health system in optimizing the health of its patient population.

Limitations

There are several limitations to this study. Administrative data imperfectly measure comorbidity and do not measure disease severity; thus, it is possible that the differences in mortality rates across VAMCs that we observed were partially caused by unobserved differences across VAMCs in comorbidity, disease severity, and/or social determinants of health outside of the domain of health care, although differences in population characteristics would seem unlikely to produce the high degree of mortality variation we observed. While VA laboratory and vital sign data were available for many patients in our cohorts, we chose not to incorporate these data owing to their inconsistent frequency, unexplained variability, and frequent missingness. Also, variation in coding practices for IHD/CHF across VAMCs may have influenced the size of the patient denominators at each VAMC, ie, if patients with mild disease were more frequently recorded as having IHD and/or CHF in some VAMCs, those VAMCs’ mortality rates would have necessarily been lower than VAMCs that were less aggressive about coding. Additionally, there is substantial geographic variation in veterans’ socioeconomic status, and low socioeconomic status is associated with mortality independently of health status.32 Our measure of veterans’ socioeconomic status (ie, local per capita income) may not have fully accounted for this variation. Finally, the study population was almost entirely men and predominantly white, consistent with the demographics of VA-enrolled veterans older than 50 years.

Conclusions

There were marked variations in risk-standardized mortality across the nation’s VAMCs among veterans with IHD and among veterans with CHF. This variation could signal important differences in quality of care across the VA health system. Cardiovascular mortality in VAMCs’ chronic cardiovascular disease populations was only modestly associated with hospitals’ posthospitalization 30-day outcomes or with the VA’s 5-star quality ratings system.

Back to top
Article Information

Corresponding Author: Peter W. Groeneveld, MD, MS, 1204 Blockley Hall, 423 Service Dr, Philadelphia, PA 19104 (peter.groeneveld@va.gov).

Accepted for Publication: March 26, 2018.

Published Online: May 16, 2018. doi:10.1001/jamacardio.2018.1115

Author Contributions: Dr Groeneveld had full access to all of the data in the study and takes responsibility for the integrity of the data analyses.

Concept and design: Groeneveld, Richardson Menno, Epstein.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Groeneveld, Medvedeva.

Critical revision of the manuscript for important intellectual content: Groeneveld, Walker, Segal, Richardson, Epstein.

Statistical analysis: Groeneveld, Medvedeva, Richardson, Epstein.

Obtained funding: Groeneveld.

Administrative, technical, or material support: Groeneveld, Walker, Segal.

Supervision: Groeneveld.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: This research was sponsored by grant IIR 14-077 from the Veterans Affairs’ Health Services Research and Development Service.

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The opinions expressed in this article are the authors’ own and do not represent the official views of the US Department of Veterans Affairs.

References
1.
Kochanek  KD, Murphy  SL, Xu  J, Tejada-Vera  B. Deaths: final data for 2014. Centers for Disease Control and Prevention. https://www.cdc.gov/nchs/data/nvsr/nvsr65/nvsr65_04.pdf. Published June 30, 2016. Accessed December 14, 2017.
2.
Jha  AK, Perlin  JB, Kizer  KW, Dudley  RA.  Effect of the transformation of the Veterans Affairs Health Care System on the quality of care.  N Engl J Med. 2003;348(22):2218-2227.PubMedGoogle ScholarCrossref
3.
VA Office of Quality, Safety, and Value. 2012 VHA facility quality and safety report. https://www.va.gov/HEALTH/docs/2012_VHA_Facility_Quality_and_Safety_Report_FINAL508.pdf. Published 2012. Accessed December 4, 2017.
4.
Centers for Medicare and Medicaid Services. Hospital Compare overview. https://www.medicare.gov/hospitalcompare/about/what-is-HOS.html. Published 2017. Accessed December 14, 2017.
5.
Ashton  CM, Petersen  NJ, Souchek  J,  et al.  Geographic variations in utilization rates in Veterans Affairs hospitals and clinics.  N Engl J Med. 1999;340(1):32-39.PubMedGoogle ScholarCrossref
6.
Sohn  MW, Arnold  N, Maynard  C, Hynes  DM.  Accuracy and completeness of mortality data in the Department of Veterans Affairs.  Popul Health Metr. 2006;4:2.PubMedGoogle ScholarCrossref
7.
Elixhauser  A, Steiner  C, Harris  DR, Coffey  RM.  Comorbidity measures for use with administrative data.  Med Care. 1998;36(1):8-27.PubMedGoogle ScholarCrossref
8.
Keenan  PS, Normand  SL, Lin  Z,  et al.  An administrative claims measure suitable for profiling hospital performance on the basis of 30-day all-cause readmission rates among patients with heart failure.  Circ Cardiovasc Qual Outcomes. 2008;1(1):29-37.PubMedGoogle ScholarCrossref
9.
Krumholz  HM, Lin  Z, Drye  EE,  et al.  An administrative claims measure suitable for profiling hospital performance based on 30-day all-cause readmission rates among patients with acute myocardial infarction.  Circ Cardiovasc Qual Outcomes. 2011;4(2):243-252.PubMedGoogle ScholarCrossref
10.
American Community Survery (ACS). https://www.census.gov/programs-surveys/acs/. Published 2018. Accessed March 5, 2018.
11.
Krumholz  HM, Wang  Y, Mattera  JA,  et al.  An administrative claims model suitable for profiling hospital performance based on 30-day mortality rates among patients with an acute myocardial infarction.  Circulation. 2006;113(13):1683-1692.PubMedGoogle ScholarCrossref
12.
Ash  AS, Fienberg  SE, Louis  TA, Normand  SL, Stukel  TA, Utts  J. Statistical issues in assessing hospital performance. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/Downloads/Statistical-Issues-in-Assessing-Hospital-Performance.pdf. Published 2012. Accessed November 20, 2017.
13.
Finkelstein  DM.  A proportional hazards model for interval-censored failure time data.  Biometrics. 1986;42(4):845-854.PubMedGoogle ScholarCrossref
14.
U.S. Department of Veteran Affairs. End of year hospital star rating (FY2016). https://www.va.gov/qualityofcare/measure-up/End_of_Year_Hospital_Star_Rating.asp. Published 2016. Accessed December 8, 2017.
15.
Menendez  ME, Neuhaus  V, van Dijk  CN, Ring  D.  The Elixhauser comorbidity method outperforms the Charlson index in predicting inpatient death after orthopaedic surgery.  Clin Orthop Relat Res. 2014;472(9):2878-2886.PubMedGoogle ScholarCrossref
16.
Silber  JH, Rosenbaum  PR, Brachet  TJ,  et al.  The Hospital Compare mortality model and the volume-outcome relationship.  Health Serv Res. 2010;45(5, pt 1):1148-1167.PubMedGoogle ScholarCrossref
17.
Borzecki  AM, Christiansen  CL, Loveland  S, Chew  P, Rosen  AK.  Trends in the inpatient quality indicators: the Veterans Health Administration experience.  Med Care. 2010;48(8):694-702.PubMedGoogle ScholarCrossref
18.
Ho  PM, Luther  SA, Masoudi  FA,  et al.  Inpatient and follow-up cardiology care and mortality for acute coronary syndrome patients in the Veterans Health Administration.  Am Heart J. 2007;154(3):489-494.PubMedGoogle ScholarCrossref
19.
Keyhani  S, Cheng  E, Arling  G,  et al.  Does the inclusion of stroke severity in a 30-day mortality model change standardized mortality rates at Veterans Affairs hospitals?  Circ Cardiovasc Qual Outcomes. 2012;5(4):508-513.PubMedGoogle ScholarCrossref
20.
Wheeler  S, Bowen  JD, Maynard  C,  et al.  Women veterans and outcomes after acute myocardial infarction.  J Womens Health (Larchmt). 2009;18(5):613-618.PubMedGoogle ScholarCrossref
21.
Bush  RL, DePalma  RG, Itani  KM, Henderson  WG, Smith  TS, Gunnar  WP.  Outcomes of care of abdominal aortic aneurysm in Veterans Health Administration facilities: results from the National Surgical Quality Improvement Program.  Am J Surg. 2009;198(5)(suppl):S41-S48.PubMedGoogle ScholarCrossref
22.
Khuri  SF, Daley  J, Henderson  WG.  The comparative assessment and improvement of quality of surgical care in the Department of Veterans Affairs.  Arch Surg. 2002;137(1):20-27.PubMedGoogle ScholarCrossref
23.
Smith  T, Li  X, Nylander  W, Gunnar  W.  Thirty-Day postoperative mortality risk estimates and 1-year survival in veterans health administration surgery patients.  JAMA Surg. 2016;151(5):417-422.PubMedGoogle ScholarCrossref
24.
Groeneveld  PW, Kruse  GB, Chen  Z, Asch  DA.  Variation in cardiac procedure use and racial disparity among Veterans Affairs Hospitals.  Am Heart J. 2007;153(2):320-327.PubMedGoogle ScholarCrossref
25.
Chen  Q, Hanchate  A, Shwartz  M,  et al.  Comparison of the Agency for Healthcare Research and Quality patient safety indicator rates among veteran dual users.  Am J Med Qual. 2014;29(4):335-343.PubMedGoogle ScholarCrossref
26.
Trivedi  AN, Grebla  RC.  Quality and equity of care in the veterans affairs health-care system and in medicare advantage health plans.  Med Care. 2011;49(6):560-568.PubMedGoogle ScholarCrossref
27.
Trivedi  AN, Matula  S, Miake-Lye  I, Glassman  PA, Shekelle  P, Asch  S.  Systematic review: comparison of the quality of medical care in Veterans Affairs and non-Veterans Affairs settings.  Med Care. 2011;49(1):76-88.PubMedGoogle ScholarCrossref
28.
Donabedian  A.  The Definition of Quality and Approaches to its Assessment. Ann Arbor, MI: Health Administration Press; 1980.
29.
DesHarnais  SI. The outcome model of quality. In: Sollecito  WA, Johnson  JK, eds.  McLaughlin and Kaluzny’s Continuous Quality Improvement in Health Care. 4th ed. Burlington, MA: Jones & Bartlett; 2013:155-180.
30.
Health  MP. The role of the hospital. https://www.aha.org/ahahret-guides/2012-04-01-managing-population-health-role-hospital. Published 2016. Accessed March 8, 2018.
31.
Stiefel  M, Nolan  K.  A Guide to Managing the Triple Aim: Population Health, Experience of Care, and Per Capita Cost. Cambridge, MA: Institute for Healthcare Improvement; 2012.
32.
Alter  DA, Naylor  CD, Austin  P, Tu  JV.  Effects of socioeconomic status on access to invasive cardiac procedures and on mortality after acute myocardial infarction.  N Engl J Med. 1999;341(18):1359-1367.PubMedGoogle ScholarCrossref
×