[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 54.161.130.145. Please contact the publisher to request reinstatement.
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Download PDF
Figure 1.
Risk-Standardized Mortality Rates for Acute Myocardial Infarction, Heart Failure, or Pneumonia, 2010-2013
Risk-Standardized Mortality Rates for Acute Myocardial Infarction, Heart Failure, or Pneumonia, 2010-2013

Bin width is equal to 0.5%, and bars are centered over the midpoint of each bin.

Figure 2.
Risk-Standardized Readmission Rates for Acute Myocardial Infarction, Heart Failure, or Pneumonia, 2010-2013
Risk-Standardized Readmission Rates for Acute Myocardial Infarction, Heart Failure, or Pneumonia, 2010-2013

Bin width is equal to 0.5%, and bars are centered over the midpoint of each bin.

Table 1.  
Medicare Beneficiaries Hospitalized for Acute Myocardial Infarction, Heart Failure, or Pneumonia, 2010-2013
Medicare Beneficiaries Hospitalized for Acute Myocardial Infarction, Heart Failure, or Pneumonia, 2010-2013
Table 2.  
Hospital Characteristics for Veterans Affairs and Non–Veterans Affairs Hospitals, 2010-2013
Hospital Characteristics for Veterans Affairs and Non–Veterans Affairs Hospitals, 2010-2013
Table 3.  
Thirty-Day Mortality and Readmission Rates for Medicare Beneficiaries in Veterans Affairs and non-Veterans Affairs Hospitals, 2010-2013
Thirty-Day Mortality and Readmission Rates for Medicare Beneficiaries in Veterans Affairs and non-Veterans Affairs Hospitals, 2010-2013
1.
Selim  AJ, Kazis  LE, Rogers  W,  et al.  Risk-adjusted mortality as an indicator of outcomes: comparison of the Medicare Advantage Program with the Veterans’ Health Administration. Med Care. 2006;44(4):359-365.
PubMedArticle
2.
Landrum  MB, Guadagnoli  E, Zummo  R, Chin  D, McNeil  BJ.  Care following acute myocardial infarction in the Veterans Administration Medical Centers: a comparison with Medicare. Health Serv Res. 2004;39(6 Pt 1):1773-1792.
PubMedArticle
3.
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.
PubMedArticle
4.
Fihn  SD, Vaughan-Sarrazin  M, Lowy  E,  et al.  Declining mortality following acute myocardial infarction in the Department of Veterans Affairs Health Care System. BMC Cardiovasc Disord. 2009;9:44-56.
PubMedArticle
5.
Weeks  WB, West  AN, Wallace  AE, Fisher  ES.  Comparing the characteristics, utilization, efficiency, and outcomes of VA and non-VA inpatient care provided to VA enrollees: a case study in New York. Med Care. 2008;46(8):863-871.
PubMedArticle
6.
National Center for Veterans Analysis and Statistics. Utilization. 2014; http://www.va.gov/vetdata/utilization.asp. Accessed January 4, 2016.
7.
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.
PubMedArticle
8.
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.
PubMedArticle
9.
Lindenauer  PK, Normand  SL, Drye  EE,  et al.  Development, validation, and results of a measure of 30-day readmission following hospitalization for pneumonia. J Hosp Med. 2011;6(3):142-150.
PubMedArticle
10.
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 heart failure. Circulation. 2006;113(13):1693-1701.
PubMedArticle
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.
PubMedArticle
12.
Bratzler  DW, Normand  SL, Wang  Y,  et al.  An administrative claims model for profiling hospital 30-day mortality rates for pneumonia patients. PLoS One. 2011;6(4):e17401.
PubMedArticle
13.
Ross  JS, Bernheim  SM, Lin  Z,  et al.  Based on key measures, care quality for Medicare enrollees at safety-net and non-safety-net hospitals was almost equal. Health Aff (Millwood). 2012;31(8):1739-1748.
PubMedArticle
14.
US Department of Labor, Office of Workers’ Compensation Programs. Geographic practice cost index values by zip code. http://www.dol.gov/owcp/regs/feeschedule/fee/fee11/fs11_gpci_by_msa-ZIP.xls. Published July 10, 2011. Accessed January 4, 2016.
15.
Weeks  WB, Yano  EM, Rubenstein  LV.  Primary care practice management in rural and urban Veterans Health Administration settings. J Rural Health. 2002;18(2):298-303.
PubMedArticle
16.
Institute of Medicine. Quality Through Collaboration: The Future of Rural Health Care. Washington, DC: National Academies Press; 2005.
17.
Borrero  S, Kwoh  CK, Sartorius  J, Ibrahim  SA.  Brief report: gender and total knee/hip arthroplasty utilization rate in the VA system. J Gen Intern Med. 2006;21(S3)(suppl 3):S54-S57.
PubMedArticle
18.
QualityNet. Measure methodology reports: readmission measures. http://www.qualitynet.org/dcs/ContentServer?c=Page&pagename=QnetPublic/Page/QnetTier4&cid=1219069855841. Accessed January 4, 2016.
19.
American Hospital Association. AHA data and directories. http://www.aha.org/research/rc/stat-studies/data-and-directories.shtml. 2014. Accessed January 4, 2016.
20.
US Census Bureau. Population estimates: metropolitan and micropolitan statistical areas. https://www.census.gov/popest/data/metro/totals/2012/. 2015. Accessed January 4, 2016.
21.
Centers for Medicare & Medicaid Services. Denominator file. https://www.cms.gov/Research-Statistics-Data-and-Systems/Files-for-Order/IdentifiableDataFiles/DenominatorFile.html. Accessed January 4, 2016.
22.
Asch  SM, McGlynn  EA, Hogan  MM,  et al.  Comparison of quality of care for patients in the Veterans Health Administration and patients in a national sample. Ann Intern Med. 2004;141(12):938-945.
PubMedArticle
23.
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.
PubMedArticle
24.
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.
PubMedArticle
25.
Werner  RM, Bradlow  ET.  Relationship between Medicare’s hospital compare performance measures and mortality rates. JAMA. 2006;296(22):2694-2702.
PubMedArticle
26.
US Department of Veterans Affairs. Heart failure provider network. http://www.queri.research.va.gov/chf/networks. Accessed January 4, 2016.
27.
Weinberger  M, Oddone  EZ, Henderson  WG.  Does increased access to primary care reduce hospital readmissions? N Engl J Med. 1996;334(22):1441-1447.
PubMedArticle
28.
Holloway  JJ, Medendorp  SV, Bromberg  J.  Risk factors for early readmission among veterans. Health Serv Res. 1990;25(1 Pt 2):213-237.
PubMed
29.
US Department of Veterans Affairs. VA hospital to home (H2H) initiative. http://www.queri.research.va.gov/chf/products/h2h/default.cfm. Accessed January 4, 2016.
30.
Chronic heart failure goals and activities. http://www.queri.research.va.gov/chf/goals.cfm. Accessed January 4, 2016.
31.
Gerhardt  G, Yemane  A, Hickman  P, Oelschlaeger  A, Rollins  E, Brennan  N.  Medicare hospital readmission rates showed meaningful decline in 2012. Medicare Medicaid Res Rev. 2013;3(2):E1-E11.
PubMedArticle
32.
Krumholz  HM, Nuti  SV, Downing  NS, Normand  SL, Wang  Y.  Mortality, hospitalizations, and expenditures for the Medicare population aged 65 years or older, 1999-2013. JAMA. 2015;314(4):355-365.
PubMedArticle
33.
Veterans Access, Choice, and Accountability Act of 2014, HR 3230, 113th Cong, 2nd session (2014). https://www.govtrack.us/congress/bills/113/hr3230. Accessed January 4, 2016.
34.
Ross  JS, Maynard  C, Krumholz  HM,  et al.  Use of administrative claims models to assess 30-day mortality among Veterans Health Administration hospitals. Med Care. 2010;48(7):652-658.
PubMedArticle
35.
Jones  LG, Sin  MK, Hage  FG,  et al.  Characteristics and outcomes of patients with advanced chronic systolic heart failure receiving care at the Veterans Affairs versus other hospitals. Circ Heart Fail. 2015;8(1):17-24.
PubMedArticle
36.
Agha  Z, Lofgren  RP, VanRuiswyk  JV, Layde  PM.  Are patients at Veterans Affairs medical centers sicker? Arch Intern Med. 2000;160(21):3252-3257.
PubMedArticle
37.
Rogers  WH, Kazis  LE, Miller  DR,  et al.  Comparing the health status of VA and non-VA ambulatory patients. J Ambul Care Manage. 2004;27(3):249-262.
PubMedArticle
38.
Kazis  LE, Ren  XS, Lee  A,  et al.  Health status in VA patients. Am J Med Qual. 1999;14(1):28-38.
PubMedArticle
39.
Payne  SM, Lee  A, Clark  JA,  et al.  Utilization of medical services by Veterans Health Study (VHS) respondents. J Ambul Care Manage. 2005;28(2):125-140.
PubMedArticle
40.
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.
PubMedArticle
Original Investigation
February 9, 2016

Association of Admission to Veterans Affairs Hospitals vs Non–Veterans Affairs Hospitals With Mortality and Readmission Rates Among Older Men Hospitalized With Acute Myocardial Infarction, Heart Failure, or Pneumonia

Author Affiliations
  • 1Center for Outcomes Research and Evaluation, Yale–New Haven Hospital, New Haven, Connecticut
  • 2Section of Cardiovascular Medicine, the Robert Wood Johnson Foundation Clinical Scholars Program, the Section of General Internal Medicine, and Section of Rheumatology, Department of Internal Medicine, Yale School of Medicine, New Haven, Connecticut
  • 3The VA Eastern Colorado Healthcare System, Denver
  • 4Department of Health Policy and Management, Yale School of Public Health, New Haven, Connecticut
  • 5Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora
  • 6Department of Health Care Policy, Harvard Medical School, Boston, Massachusetts
  • 7Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
  • 8Veterans Affairs Medical Center, West Haven, Connecticut
JAMA. 2016;315(6):582-592. doi:10.1001/jama.2016.0278
Abstract

Importance  Little contemporary information is available about comparative performance between Veterans Affairs (VA) and non-VA hospitals, particularly related to mortality and readmission rates, 2 important outcomes of care.

Objective  To assess and compare mortality and readmission rates among men in VA and non-VA hospitals.

Design, Setting, and Participants  Cross-sectional analysis involving male Medicare fee-for-service beneficiaries aged 65 years or older hospitalized between 2010 and 2013 in VA and non-VA acute care hospitals for acute myocardial infarction (AMI), heart failure (HF), or pneumonia using the Medicare Standard Analytic Files and Enrollment Database together with VA administrative claims data. To avoid confounding geographic effects with health care system effects, we studied VA and non-VA hospitals within the same metropolitan statistical area (MSA).

Exposures  Hospitalization in a VA or non-VA hospital in MSAs that contained at least 1 VA and non-VA hospital.

Main Outcomes and Measures  For each condition, 30-day risk-standardized mortality rates and risk-standardized readmission rates for VA and non-VA hospitals. Mean aggregated within-MSA differences in mortality and readmission rates were also assessed.

Results  We studied 104 VA and 1513 non-VA hospitals, with each condition-outcome analysis cohort for VA and non-VA hospitals containing at least 7900 patients (men; ≥65 years), in 92 MSAs. Mortality rates were lower in VA hospitals than non-VA hospitals for AMI (13.5% vs 13.7%, P = .02; −0.2 percentage-point difference) and HF (11.4% vs 11.9%, P = .008; −0.5 percentage-point difference), but higher for pneumonia (12.6% vs 12.2%, P = .045; 0.4 percentage-point difference). In contrast, readmission rates were higher in VA hospitals for all 3 conditions (AMI, 17.8% vs 17.2%, 0.6 percentage-point difference; HF, 24.7% vs 23.5%, 1.2 percentage-point difference; pneumonia, 19.4% vs 18.7%, 0.7 percentage-point difference, all P < .001). In within-MSA comparisons, VA hospitals had lower mortality rates for AMI (percentage-point difference, −0.22; 95% CI, −0.40 to −0.04) and HF (−0.63; 95% CI, −0.95 to −0.31), and mortality rates for pneumonia were not significantly different (−0.03; 95% CI, −0.46 to 0.40); however, VA hospitals had higher readmission rates for AMI (0.62; 95% CI, 0.48 to 0.75), HF (0.97; 95% CI, 0.59 to 1.34), or pneumonia (0.66; 95% CI, 0.41 to 0.91).

Conclusions and Relevance  Among older men with AMI, HF, or pneumonia, hospitalization at VA hospitals, compared with hospitalization at non-VA hospitals, was associated with lower 30-day risk-standardized all-cause mortality rates for AMI and HF, and higher 30-day risk-standardized all-cause readmission rates for all 3 conditions, both nationally and within similar geographic areas, although absolute differences between these outcomes at VA and non-VA hospitals were small.

Introduction

The Department of Veterans Affairs (VA) and the Centers for Medicare & Medicaid Services (CMS) are collaborating in calculating and reporting hospital performance. The alignment of reporting systems provides an opportunity to compare performance of VA and non-VA hospitals directly with respect to risk-adjusted outcomes for key conditions, including acute myocardial infarction (AMI), heart failure (HF), or pneumonia. Prior studies, which are a decade or more old or are not nationally representative, and occurred before validated hospital-level outcome measures existed, found that patients who received care at VA hospitals had either similar or lower mortality rates than patients who received care at non-VA hospitals14; with respect to readmission rates, a study in New York found that elderly patients had higher rates of 30-day readmission in VA hospitals than in non-VA hospitals.5 Contemporary knowledge about hospital-level outcomes would help to better understand quality of care for the approximately 9 million veterans enrolled in the nation’s largest integrated health care system compared with those in Medicare.6

Accordingly, we assessed and compared 30-day risk-standardized all-cause mortality rates and 30-day risk-standardized all-cause readmission rates in VA hospitals and non-VA hospitals for patients hospitalized with AMI, HF, or pneumonia. We applied the measure method that is used to calculate publicly reported mortality and readmission rates. To avoid confounding geographic effects with health care system effects, we studied VA and non-VA hospitals within the same metropolitan statistical area (MSA).

Methods
Study Population

We used a combined data set of the CMS Standard Analytic Files and Enrollment Database together with VA administrative claims data to identify Medicare fee-for-service beneficiaries aged 65 years or older hospitalized from July 1, 2010, through June 30, 2013, with a principal discharge diagnosis of AMI, HF, or pneumonia in VA or non-VA acute care hospitals in the United States using International Classification of Diseases, Ninth Revision, Clinical Modification codes identical to those used in the publicly reported mortality and readmission measures.712 For VA beneficiaries, we studied only those enrolled in the Medicare fee-for-service program. These databases include enrollment status, admissions, patient demographic information, inpatient and outpatient diagnosis codes, mortality, and readmissions. They also include unique patient identifiers that enable the identification of patients across VA and non-VA systems. We excluded beneficiaries for whom a complete claims history for the 12-month period preceding the claim was unavailable because of the use of data in the year before hospitalization in risk adjustment. We included patients admitted to hospitals in urban MSAs that contained at least 1 VA and non-VA hospital to reduce confounding geographic effects from health system effects (see eTable 1 in the Supplement for VA hospitals included in the study).13,14 This also helps minimize potential confounding from the inclusion of nonurban hospitals, given that the challenges faced by rural and urban hospitals differ.15,16 Because more than 90% of patients in VA hospitals are male,17 we restricted our analyses to men.

The Yale University Human Investigation Committee reviewed the study and waived the requirement for informed consent.

Risk-Standardized Mortality and Readmission Rates

To calculate 30-day risk-standardized all-cause mortality rates and 30-day risk-standardized all-cause readmission rates for each condition, we used the specifications used in the measures developed for CMS and endorsed by the National Quality Forum.711 Briefly, the technique of risk-adjustment employs a hierarchical logistic regression model to create a hospital-level 30-day risk-standardized mortality or readmission rate. This modeling appropriately accounts for the structure of the data (patients clustered within hospitals) and the underlying risk due to patients’ age and comorbidities when estimating hospital mortality or readmission rates. This approach simultaneously models 2 levels (patient and hospital) to account for the variance in patient outcomes within and between hospitals. At the patient level, the model uses the log-odds of mortality or readmission within 30 days of admission or discharge as the dependent variable and age and selected clinical covariates as the independent variables (or predictors). The second level models the hospital-specific intercepts as arising from a normal distribution. The hospital-specific intercept, or hospital-specific effect, represents the hospital contribution to the risk of mortality or readmission after accounting for patient risk and can be inferred as a measure of quality. The hospital-specific intercepts are given a distribution to account for the clustering (nonindependence) of patients within the same hospital.

We defined 30-day mortality as death from any cause within 30 days of the admission date for an index hospitalization for AMI, HF, or pneumonia. Readmission was defined as 1 or more rehospitalization for any unplanned cause to any acute care hospital within 30 days of the discharge date for an index hospitalization for AMI, HF, or pneumonia among patients surviving the index hospitalization. Planned readmissions were identified and excluded using planned readmission algorithms (version 2.1).711,18 We used the top 25 diagnosis and procedure codes for risk adjustment across all outcomes and conditions in both VA and non-VA hospitals.

For the mortality cohort for each condition, we randomly selected 1 admission per year for patients with multiple admissions for the same diagnosis during the study period.13 For individuals admitted to both a VA and non-VA hospital (<2.1% of patients for each condition-outcome analysis in our cohort; Table 1), we randomly selected 1 admission from all admissions regardless of location hospitalized. For transfers between hospitals, relevant hospitalizations were linked into a single episode of care, with mortality attributed to the hospital to which the patient was first admitted; transfers between VA and non-VA hospitals were treated similarly to transfers between VA and VA or between non-VA and non-VA hospitals.

For the readmission cohort for each condition, we included all admissions for patients with multiple admissions for the same diagnosis during any study year. Hospitalizations within 30 days of discharge from an index hospitalization were considered readmissions and not additional index hospitalizations. Readmission to another system (VA to non-VA or non-VA to VA) was counted as a readmission attributed to the index hospital. Transfers were linked into a single episode of care, with readmissions attributed to the hospital that discharged the patient to a nonacute setting; transfers between VA and non-VA hospitals were treated as described above. We also calculated the number of patients who died after discharge, but before 30-day readmission, within each health care system for each condition.

Hospital Characteristics

Hospital characteristics were derived from the American Hospital Association Annual Survey in the corresponding year, other than Medicare condition-specific 3-year volumes, which were derived from the Standard Analytic Files and Enrollment Database.19 These variables include total staffed beds, ownership type, region, teaching status, Joint Commission accreditation, and hospital capacity to provide coronary artery bypass graft surgery or percutaneous coronary intervention. Metropolitan statistical area population size data, from April 1, 2010, to July 1, 2012, were obtained from the US Census Bureau.20

Patient Characteristics

Demographic information included age and race (white, black, other); there are no measures of socioeconomic status available in these data. Race (white, black, other) for these categories is derived from the Social Security Administration21 and was assessed as a descriptive characteristic of the populations studied. We defined comorbidities according to the specifications used by CMS to profile hospitals, using principal and secondary diagnosis codes from hospitalizations in the 12 months before the index hospitalization (See eTable 2 in the Supplement for medical history characteristics).711

Statistical Analysis

We compared VA and non-VA hospitals for differences in both patient factors (demographic and clinical characteristics) and hospital factors. For the 3 conditions, within the full 3-year pooled sample, we estimated mortality and readmission rates for each VA and non-VA hospital using hierarchical logistic regression models. Each model included age and selected clinical covariates (eTable 2 in the Supplement) as well as a hospital-specific random effects intercept. Risk-standardized mortality and readmission rates are calculated as the ratio of the number of predicted outcomes to the number of expected outcomes (death or readmission), multiplied by the unadjusted rate of the given outcome within our overall sample. We calculated aggregate mortality and readmission for each condition-outcome group for VA and non-VA hospitals by weighting hospital-specific mortality or readmission rate estimates by a hospital’s condition-specific volume.

In addition, for each condition, we calculated the mean-aggregated MSA-specific mortality and readmission rates stratified by VA hospital status to reduce confounding geographic effects from health system effects and minimize potential confounding from the inclusion of nonurban hospitals, and determined the average within-MSA difference between VA and non-VA hospitals. As above, within MSAs, we weighted hospital-specific mortality or readmission rates by their condition-specific volume. After aggregating outcome rates for the VA hospitals and non-VA hospitals within an MSA, we calculated the difference between the rates of VA and non-VA hospitals in each MSA, and then calculated the mean and 95% confidence interval of the differences among all MSAs in our sample. Differences were presented as the rate in VA hospitals minus the rate in non-VA hospitals; values of less than 0 indicate a lower mortality or readmission rate for VA hospitals.

Analyses were performed separately for AMI, HF, and pneumonia. For all 3 conditions, we calculated the mean (95% CI) mortality or readmission rate grouped by VA and non-VA hospitals. The corresponding P values were obtained using t tests. In the above calculations, weights were added by the condition-specific number of hospitalizations to account for the variability in hospital sample sizes. We considered P values <.05 (2-sided) to be statistically significant. Analyses were conducted using SAS software, version 9.3 (SAS Institute Inc).

Results
Study Sample

Among 388 MSAs in the United States, 92 had at least 1 VA and 1 non-VA hospital; 7 MSAs had more than 1 VA hospital. We studied 104 VA and 1513 non-VA hospitals, with each condition-outcome analysis cohort for VA and non-VA hospitals containing at least 7900 patients. For analyses focused on AMI hospitalizations, there were 1542 hospitals (103 VA hospitals with 8012 patients and 1439 non-VA hospitals with 124 220 patients) for mortality analyses and 1523 hospitals (103 VA hospitals with 7929 patients and 1420 non-VA hospitals with 132 276 patients) for readmission analyses; for HF, 1604 hospitals (104 VA hospitals with 22 882 patients and 1500 non-VA hospitals with 215 312 patients) for mortality and 1604 hospitals (104 VA hospitals with 26 231 patients and 1500 non-VA hospitals with 269 856 patients) for readmission; and for pneumonia, 1609 hospitals (104 VA hospitals with 21 092 patients and 1505 non-VA hospitals with 190 798 patients) for mortality and 1610 hospitals (104 VA hospitals with 21 229 patients and 1506 non-VA hospitals with 205 060 patients) for readmission (Table 1). There were 124 VA medical centers in our data set (eTable 1).

Patient Characteristics

For the mortality and readmission analyses for all 3 conditions, patients who received care in VA hospitals were younger and generally less likely to be white than those who received care in non-VA hospitals (Table 1). Among patients hospitalized for AMI, VA patients were more likely to have a history of HF or diabetes. Among patients hospitalized for HF, VA patients were also more likely to have a history of HF or diabetes, but less likely to have a documented history of coronary atherosclerosis or valvular heart disease. Among patients hospitalized for pneumonia, VA patients were more likely to have a history of chronic obstructive pulmonary disease and less likely to have a history of pneumonia, protein-calorie malnutrition, or fibrosis of the lung or other chronic lung disorders.

Hospital Characteristics

Veterans Affairs hospitals were more likely to be teaching hospitals and were larger, with a greater number of hospital beds (Table 2). For HF and pneumonia hospitalizations, but not for AMI, a greater proportion of VA hospitals had 3-year condition-specific volumes of more than 150 patients.

30-Day Mortality Rates

Before risk-adjustment, VA hospitals had lower mortality rates than non-VA hospitals across all 3 conditions (Table 3). After risk-adjustment, VA hospitals had lower mortality rates than non-VA hospitals for AMI (13.52%; 95% CI, 13.38% to 13.66% vs 13.69%; 95% CI, 13.64% to 13.74%; −0.17 percentage-point difference) and HF (11.43%; 95% CI, 11.11% to 11.75% vs 11.87%; 95% CI, 11.80% to 11.93%; −0.44 percentage-point difference) (P < .03 for both; Table 3, Figure 1). However, for pneumonia, VA hospitals had higher mortality rates than non-VA hospitals (12.63%; 95% CI, 12.19% to 13.07% vs 12.17%; 95% CI, 12.08% to 12.26%, P = .045; 0.46 percentage-point difference). The distribution of mortality rate performance was not markedly different between VA and non-VA hospitals for all 3 conditions (Figure 1; eFigure, eTables 3 and 4 in the Supplement).

In within-MSA comparisons, VA hospitals had lower mean-aggregated within-MSA mortality rates for AMI (−0.22 percentage points; 95% CI, −0.40 to −0.04; P = .02) and HF (−0.63 percentage points, 95% CI, −0.95 to −0.31; P < .001) than did non-VA hosptials. However, in contrast with the overall comparison, VA hospitals had mean-aggregated within-MSA mortality rates that were not significantly different for pneumonia (−0.03 percentage points; 95% CI, −0.46 to 0.40; P = .90).

30-Day Readmission Rates

Before risk-adjustment, VA hospitals had higher readmission rates than non-VA hospitals across all 3 conditions (Table 3). After risk-adjustment, VA hospitals had higher readmission rates than non-VA hospitals for all 3 conditions (Table 3; Figure 2). For AMI, the readmission rates were 17.84% (95% CI, 17.71% to 17.96%) for VA hospitals vs 17.21% (95% CI, 17.17% to 17.25%) for non-VA hospitals (0.63 percentage-point difference); for HF, 24.66% (95% CI, 24.31% to 25.02%) vs 23.46% (95% CI, 23.39% to 23.53%) (1.2 percentage-point difference); and for pneumonia, 19.44% (95% CI, 19.19% to 19.69%) vs 18.68% (95% CI, 18.63% to 18.73%) (0.76 percentage-point difference) (P < .001 for all). The distribution of readmission rate performance differed between VA and non-VA hospitals (Figure 2; eFigure, eTables 5 and 6 in the Supplement): for all 3 conditions, VA hospitals had a greater proportion of hospitals with higher readmission rates compared with non-VA hospitals.

In within-MSA comparisons, VA hospitals had higher mean-aggregated within-MSA readmission rates than non-VA hospitals for all 3 conditions: 0.62 percentage points (95% CI, 0.48 to 0.75) for AMI, 0.97 percentage points (95% CI, 0.59 to 1.34) for HF, and 0.66 percentage points (95% CI, 0.41 to 0.91) for pneumonia (P < .001 for all; Table 3).

Across conditions, among those who were initially hospitalized in VA hospitals, approximately 12% were readmitted to non-VA hospitals; among those initially hospitalized in non-VA hospitals, less than 1% were readmitted to VA hospitals (Table 1).

Discussion

In this national study comparing CMS 30-day mortality and readmission measures for 3 major conditions, we found that VA hospitals have lower mortality rates for AMI and HF compared with non-VA hospitals. However, VA hospitals had higher readmission rates for AMI, HF, and pneumonia than non-VA hospitals. These differences persisted after accounting for geographic variation in hospital location by limiting comparisons of VA and non-VA hospitals to those within the same MSA. In general, however, the magnitudes of differences were small for both measures across all 3 conditions.

We were unable to find another national study using CMS hospital mortality and readmission measures to compare the quality of care between VA hospitals, which comprise the largest integrated health care system in the United States, and non-VA hospitals. This study is also distinct because it compares outcomes between VA and non-VA hospitals within similar geographic areas. The finding that risk-standardized mortality rates for cardiovascular conditions were lower, albeit with small absolute differences, in VA hospitals may reflect higher quality of care in VA hospitals as represented by adherence to process measures.3,2225 The lower mortality rates also may be due to the quality improvement efforts that can be implemented across the VA’s integrated delivery system. For example, upon evidence of possible higher cardiovascular mortality in VA hospitals compared with non-VA hospitals, in 2003 the VA undertook a comprehensive cardiac care improvement initiative across all its hospitals, after which AMI mortality declined.4 Similar initiatives and programs, such as the VA Heart Failure Provider Network,26 may have helped lower mortality rates. However, the lower mortality result for VA hospitals was not consistent for pneumonia, with no difference in mortality rates between VA and non-VA hospitals after accounting for geographic location, and the differences in cardiovascular condition mortality outcomes were modest. Another possibility is that there are unmeasured case-mix differences, which could have produced the small differences.

There are several possible reasons for the finding of higher readmission rates in VA hospitals. First, VA hospitals may have a higher propensity to admit. For example, a study that sought to reduce readmissions in VA hospitals by increased access to primary care resulted in an increase in readmissions in the intervention group.27 Second, the distance VA patients need to travel to the hospital may be greater than that for non-VA patients, which has been associated with higher readmission rates among veterans28; however, greater distance could also produce lower readmission rates because patients may be less likely to go to the hospital when they need care. Third, the CMS Hospital Readmission Reduction Program introduced financial penalties for non-VA hospitals in 2012, and national public reporting of, and incentives for reducing, readmissions in non-VA hospitals may have contributed to lower readmission rates for non-VA hospitals during our study period. Nevertheless, the VA has undertaken efforts to prevent readmissions in recent years,29,30 and we are unable to assert whether or how the combination of efforts in VA hospitals and public reporting for non-VA hospitals contributed to relatively lower readmission rates among non-VA hospitals. The results of this study may be helpful to the VA system by highlighting the variation in performance across VA and non-VA hospitals even after case-mix adjustment, so that resources could be targeted to reduce unnecessary readmissions.

Of note, the range in performance among both VA and non-VA hospitals for both outcomes was as little as approximately 3.5 percentage points, although in some cases it was as much as approximately 13 percentage points. Some readers may consider the difference modest and an indication of little opportunity for improvement. However, it may also be that, for some conditions, hospitals are aggregating around a similar performance but have opportunities for marked improvement, as has been demonstrated with mortality over the last 15 years and more recently by readmission.31,32 In any case, this study sought to compare 2 health systems, and whether they have greater opportunities to improve does not negate the fact that their current performance is separated by only a few percentage points.

Although VA outcome data for these 3 key conditions are currently available on the Hospital Comparewebsite, this initial integration of measures is going to be followed by an expectation that a broader set of quality measures be reported across both VA and non-VA hospitals as part of the Veterans Access, Choice, and Accountability Act of 2014.33 Within this context, our study has multiple implications. First, the availability of this information, with equal standards of reporting, will allow for the assessment of comparative quality, which can inform targeted improvements in VA and non-VA hospitals alike. It also affords a unique opportunity for the VA to partner with public and private entities to test and implement strategies to improve care. The current study serves as an example of national performance comparison for VA and non-VA hospital care, which sets the stage for future performance and quality improvement studies. Moreover, the results of our study and other benchmarking efforts could inform efforts to improve quality in the VA, particularly our findings of variation in performance, by identifying and learning from high-performing hospitals and disseminating best practices to lower-performing hospitals to elevate the entire performance curve. Finally, given the new Veterans Choice Program for the funding of private care for some veterans, VA policy makers should recognize that there is also variation in quality among non-VA hospitals and take steps to encourage veterans who choose to use non-VA hospitals to use the highest quality ones to ensure optimal outcomes and high-value care.

Our study has several limitations. First, we only studied elderly, male VA patients and Medicare patients, and are unable to assert whether our findings are generalizable to younger or female populations. Second, we studied administrative claims data, which lack information about clinical disease severity, but these are the same data used for the publicly reported outcomes measures and, at the hospital level, studies have shown that administrative data and clinical data perform similarly for both VA and non-VA hospitalizations.79,34 Third, while prior literature suggests that VA and non-VA populations are different, with VA populations generally having worse health status but similar mortality rates,3537 these reports did not study older populations. Other studies suggest that VA beneficiaries aged 65 years or older, who comprise our cohort, have similar or better physical health status and better mental health status than younger VA beneficiaries,3840 which is in contrast with non-VA populations in which older populations are less healthy. Fourth, we could not study all VA hospitals in the United States because some were in rural areas or were not within an urban MSA with another non-VA hospital; nevertheless, studying hospitals within the same urban MSA allowed us to make more direct comparisons between VA and non-VA hospitals by accounting for geographic differences. Fourth, there were hospitals for which hospital characteristics were not available; however, these hospitals represented only approximately 4% of VA or non-VA hospitals in our cohort, so it is unlikely that this would influence the results. Fifth, we only studied 3 common conditions among elderly patients, which may not be representative of care for all patients in either VA or non-VA hospitals; however, these outcome metrics are CMS core condition measures.

Conclusions

Among older men with AMI, HF, or pneumonia, hospitalization at VA hospitals, compared with hospitalization at non-VA hospitals, was associated with lower 30-day risk-standardized all-cause mortality rates for AMI and HF, and higher 30-day risk-standardized all-cause readmission rates for all 3 conditions, both nationally and within similar geographic areas, although absolute differences between these outcomes at VA and non-VA hospitals were small.

Back to top
Article Information

Corresponding Author: Harlan M. Krumholz, MD, SM, Section of Cardiovascular Medicine, Department of Internal Medicine, Yale School of Medicine, One Church St, Ste 200, New Haven, CT 06510 (harlan.krumholz@yale.edu).

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

Study concept and design: Nuti, Rumsfeld, Murugiah, Bernheim, Krumholz.

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

Drafting of the manuscript: Nuti, Qin, Murugiah.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Qin, Normand.

Obtained funding: Krumholz.

Administrative, technical, or material support: Bernheim, Suter.

Study supervision: Krumholz.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Drs Ross and Krumholz reported that they are the recipients of research agreements from Medtronic and from Johnson & Johnson (Janssen), through Yale University, to develop methods of clinical trial data sharing. Dr Krumholz reported that he chairs a cardiac scientific advisory board for UnitedHealth. Drs Krumholz, Ross, Normand, Bernheim, and Suter reported that they work under contract to the Centers for Medicare & Medicaid Services to develop and maintain performance measures. No other disclosures were reported.

Funding/Support: Dr Ross is supported by grant K08 AG032886 from the National Institute on Aging and by the American Federation for Aging Research through the Paul B. Beeson Career Development Award Program. Dr Krumholz is supported by grant U01 HL105270-05 (Center for Cardiovascular Outcomes Research at Yale University) from the National Heart, Lung, and Blood Institute. Dr Suter is partially supported by the VA Connecticut Healthcare System.

Role of the Funder/Sponsor: The funders 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.

References
1.
Selim  AJ, Kazis  LE, Rogers  W,  et al.  Risk-adjusted mortality as an indicator of outcomes: comparison of the Medicare Advantage Program with the Veterans’ Health Administration. Med Care. 2006;44(4):359-365.
PubMedArticle
2.
Landrum  MB, Guadagnoli  E, Zummo  R, Chin  D, McNeil  BJ.  Care following acute myocardial infarction in the Veterans Administration Medical Centers: a comparison with Medicare. Health Serv Res. 2004;39(6 Pt 1):1773-1792.
PubMedArticle
3.
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.
PubMedArticle
4.
Fihn  SD, Vaughan-Sarrazin  M, Lowy  E,  et al.  Declining mortality following acute myocardial infarction in the Department of Veterans Affairs Health Care System. BMC Cardiovasc Disord. 2009;9:44-56.
PubMedArticle
5.
Weeks  WB, West  AN, Wallace  AE, Fisher  ES.  Comparing the characteristics, utilization, efficiency, and outcomes of VA and non-VA inpatient care provided to VA enrollees: a case study in New York. Med Care. 2008;46(8):863-871.
PubMedArticle
6.
National Center for Veterans Analysis and Statistics. Utilization. 2014; http://www.va.gov/vetdata/utilization.asp. Accessed January 4, 2016.
7.
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.
PubMedArticle
8.
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.
PubMedArticle
9.
Lindenauer  PK, Normand  SL, Drye  EE,  et al.  Development, validation, and results of a measure of 30-day readmission following hospitalization for pneumonia. J Hosp Med. 2011;6(3):142-150.
PubMedArticle
10.
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 heart failure. Circulation. 2006;113(13):1693-1701.
PubMedArticle
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.
PubMedArticle
12.
Bratzler  DW, Normand  SL, Wang  Y,  et al.  An administrative claims model for profiling hospital 30-day mortality rates for pneumonia patients. PLoS One. 2011;6(4):e17401.
PubMedArticle
13.
Ross  JS, Bernheim  SM, Lin  Z,  et al.  Based on key measures, care quality for Medicare enrollees at safety-net and non-safety-net hospitals was almost equal. Health Aff (Millwood). 2012;31(8):1739-1748.
PubMedArticle
14.
US Department of Labor, Office of Workers’ Compensation Programs. Geographic practice cost index values by zip code. http://www.dol.gov/owcp/regs/feeschedule/fee/fee11/fs11_gpci_by_msa-ZIP.xls. Published July 10, 2011. Accessed January 4, 2016.
15.
Weeks  WB, Yano  EM, Rubenstein  LV.  Primary care practice management in rural and urban Veterans Health Administration settings. J Rural Health. 2002;18(2):298-303.
PubMedArticle
16.
Institute of Medicine. Quality Through Collaboration: The Future of Rural Health Care. Washington, DC: National Academies Press; 2005.
17.
Borrero  S, Kwoh  CK, Sartorius  J, Ibrahim  SA.  Brief report: gender and total knee/hip arthroplasty utilization rate in the VA system. J Gen Intern Med. 2006;21(S3)(suppl 3):S54-S57.
PubMedArticle
18.
QualityNet. Measure methodology reports: readmission measures. http://www.qualitynet.org/dcs/ContentServer?c=Page&pagename=QnetPublic/Page/QnetTier4&cid=1219069855841. Accessed January 4, 2016.
19.
American Hospital Association. AHA data and directories. http://www.aha.org/research/rc/stat-studies/data-and-directories.shtml. 2014. Accessed January 4, 2016.
20.
US Census Bureau. Population estimates: metropolitan and micropolitan statistical areas. https://www.census.gov/popest/data/metro/totals/2012/. 2015. Accessed January 4, 2016.
21.
Centers for Medicare & Medicaid Services. Denominator file. https://www.cms.gov/Research-Statistics-Data-and-Systems/Files-for-Order/IdentifiableDataFiles/DenominatorFile.html. Accessed January 4, 2016.
22.
Asch  SM, McGlynn  EA, Hogan  MM,  et al.  Comparison of quality of care for patients in the Veterans Health Administration and patients in a national sample. Ann Intern Med. 2004;141(12):938-945.
PubMedArticle
23.
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.
PubMedArticle
24.
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.
PubMedArticle
25.
Werner  RM, Bradlow  ET.  Relationship between Medicare’s hospital compare performance measures and mortality rates. JAMA. 2006;296(22):2694-2702.
PubMedArticle
26.
US Department of Veterans Affairs. Heart failure provider network. http://www.queri.research.va.gov/chf/networks. Accessed January 4, 2016.
27.
Weinberger  M, Oddone  EZ, Henderson  WG.  Does increased access to primary care reduce hospital readmissions? N Engl J Med. 1996;334(22):1441-1447.
PubMedArticle
28.
Holloway  JJ, Medendorp  SV, Bromberg  J.  Risk factors for early readmission among veterans. Health Serv Res. 1990;25(1 Pt 2):213-237.
PubMed
29.
US Department of Veterans Affairs. VA hospital to home (H2H) initiative. http://www.queri.research.va.gov/chf/products/h2h/default.cfm. Accessed January 4, 2016.
30.
Chronic heart failure goals and activities. http://www.queri.research.va.gov/chf/goals.cfm. Accessed January 4, 2016.
31.
Gerhardt  G, Yemane  A, Hickman  P, Oelschlaeger  A, Rollins  E, Brennan  N.  Medicare hospital readmission rates showed meaningful decline in 2012. Medicare Medicaid Res Rev. 2013;3(2):E1-E11.
PubMedArticle
32.
Krumholz  HM, Nuti  SV, Downing  NS, Normand  SL, Wang  Y.  Mortality, hospitalizations, and expenditures for the Medicare population aged 65 years or older, 1999-2013. JAMA. 2015;314(4):355-365.
PubMedArticle
33.
Veterans Access, Choice, and Accountability Act of 2014, HR 3230, 113th Cong, 2nd session (2014). https://www.govtrack.us/congress/bills/113/hr3230. Accessed January 4, 2016.
34.
Ross  JS, Maynard  C, Krumholz  HM,  et al.  Use of administrative claims models to assess 30-day mortality among Veterans Health Administration hospitals. Med Care. 2010;48(7):652-658.
PubMedArticle
35.
Jones  LG, Sin  MK, Hage  FG,  et al.  Characteristics and outcomes of patients with advanced chronic systolic heart failure receiving care at the Veterans Affairs versus other hospitals. Circ Heart Fail. 2015;8(1):17-24.
PubMedArticle
36.
Agha  Z, Lofgren  RP, VanRuiswyk  JV, Layde  PM.  Are patients at Veterans Affairs medical centers sicker? Arch Intern Med. 2000;160(21):3252-3257.
PubMedArticle
37.
Rogers  WH, Kazis  LE, Miller  DR,  et al.  Comparing the health status of VA and non-VA ambulatory patients. J Ambul Care Manage. 2004;27(3):249-262.
PubMedArticle
38.
Kazis  LE, Ren  XS, Lee  A,  et al.  Health status in VA patients. Am J Med Qual. 1999;14(1):28-38.
PubMedArticle
39.
Payne  SM, Lee  A, Clark  JA,  et al.  Utilization of medical services by Veterans Health Study (VHS) respondents. J Ambul Care Manage. 2005;28(2):125-140.
PubMedArticle
40.
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.
PubMedArticle
×