Comparing Teaching Versus Nonteaching Hospitals: The Association of Patient Characteristics With Teaching Intensity for Three Common Medical Conditions

Shahian, David M. MD; Liu, Xiu MS; Meyer, Gregg S. MD, MSc; Normand, Sharon-Lise T. PhD

doi: 10.1097/ACM.0000000000000050
Research Reports

Purpose: To quantify the role of teaching hospitals in direct patient care, the authors compared characteristics of patients served by hospitals of varying teaching intensity.

Method: The authors studied Medicare beneficiaries ≥ 66 years old, hospitalized in 2009–2010 for acute myocardial infarction, heart failure, or pneumonia. They categorized hospitals as nonteaching, teaching, or Council of Teaching Hospitals and Health Systems (COTH) members and performed secondary analyses using intern and resident-to-bed ratios. The authors used descriptive statistics, adjusted odds ratios, and linear propensity scores to compare patient characteristics among teaching intensity levels. They supplemented Medicare mortality model variables with race, transfer status, and distance traveled.

Results: Adjusted for comorbidities, black patients had 2.44 (95% confidence interval [CI] 2.36–2.52), 2.56 (95% CI 2.51–2.60), and 2.58 (95% CI 2.51–2.65) times the odds of COTH hospital admission compared with white patients for acute myocardial infarction, heart failure, and pneumonia, respectively. For patients transferred from another hospital’s inpatient setting, the corresponding adjusted odds ratios of COTH hospital admission were 3.99 (95% CI 3.85–4.13), 4.60 (95% CI 4.34–4.88), and 4.62 (95% CI 4.16–5.12). Using national data, distributions of propensity scores (probability of admission to a COTH hospital) varied markedly among teaching intensity levels. Data from Massachusetts and California illustrated between-state heterogeneity in COTH utilization.

Conclusions: Major teaching hospitals are significantly more likely to provide care for minorities and patients requiring transfer from other institutions for advanced care.Both are essential to an equitable and high-quality regional health care system.

Dr. Shahian is professor of surgery, Harvard Medical School, and vice president, Center for Quality and Safety, Massachusetts General Hospital, Boston, Massachusetts.

Ms. Liu is senior research analyst, Center for Quality and Safety, Massachusetts General Hospital, Boston, Massachusetts.

Dr. Meyer is executive vice president for population health and chief clinical officer, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire.

Dr. Normand is professor of health care policy, Harvard Medical School, and professor of biostatistics, Harvard School of Public Health, Boston, Massachusetts.

Funding/Support: None reported.

Other disclosures: Drs. Shahian and Meyer and Ms. Liu are employed by academic medical centers.

Ethical approval: The Partners/Massachusetts General Hospital institutional review board approved this study (2009P001791).

Supplemental digital content for this article is available at

Correspondence should be addressed to Dr. Shahian, Center for Quality and Safety and Department of Surgery, Massachusetts General Hospital, 55 Fruit St., Boston, MA 02114; telephone: (617) 643-4335; e-mail:

Article Outline

Academic medical centers (AMCs) fulfill numerous essential functions in the American health care system.1,2 Their raison d’être is the education of successive generations of new physicians, and they are also the main locus for conducting research and pursuing innovation to improve clinical patient care and health system performance.

Less appreciated, however, are the unique roles that AMCs play in delivering local and regional health care. In particular, they are often the primary health care providers for urban minority populations, and they accept in transfer those complex and seriously ill patients who cannot be cared for in other hospitals. Both of these important functions may contribute to their higher costs of care and, depending on the adequacy of risk adjustment, may negatively impact the clinical quality ratings at these institutions. In an era of aggressive health care reform, intense financial pressures, and performance report cards with inconsistently robust methodology, we must better characterize the unique patient populations of major teaching hospitals.

To quantify these distinctive features of AMCs, we used contemporary national Medicare data to compare the patient populations served by hospitals of varying teaching intensity. We focused on three common medical conditions (acute myocardial infarction, heart failure, and pneumonia), which together accounted for 9.7% of Medicare short-stay discharges in 2009 and 2010.

Back to Top | Article Outline


Study population

We identified Medicare beneficiaries ≥ 66 years of age from the 50 states, the District of Columbia, Puerto Rico, Guam, and Virgin Islands who were hospitalized in short-term, acute care, general hospitals during 2009–2010 with a principal discharge diagnosis of acute myocardial infarction, heart failure, or pneumonia and were enrolled in fee-for-service coverage in the 12 months prior to their index admissions. We obtained our data from the Centers for Medicare and Medicaid Services (CMS) Medicare Provider Analysis and Review (MEDPAR) file.3 To identify our study cases, we used the International Classification of Disease Revision 9–Clinical Modification (ICD-9-CM) codes4 (see Supplemental Digital Table 1,

Back to Top | Article Outline

Hospital teaching intensity

In our primary analyses, we used one common method of categorizing teaching intensity5–8: AMC hospitals that were members of the Council of Teaching Hospitals and Health Systems (COTH); teaching hospitals that were not COTH members but had some residency programs; and nonteaching hospitals that met neither criterion. We identified hospital teaching status (yes/no) using Medicare cost reports and determined COTH status from the COTH membership directory. To link teaching status with patient characteristics from MEDPAR data, we used the hospital Medicare ID codes.

To assess the robustness of our findings, we then repeated these analyses by classifying hospitals on the basis of their number of residents and interns per bed (IRB).9,10 We defined hospitals as nonteaching if they indicated “no” to the teaching hospital question in the Medicare cost reports. We classified all teaching hospitals (“yes” to the teaching hospital question in the Medicare cost reports) as either major or minor teaching hospitals using three different IRB criteria: 0.097 (median IRB of all teaching hospitals), 0.278 (75th percentile IRB of all teaching hospitals), and 0.629 (90th percentile IRB of all teaching hospitals).

We initially identified 2,035,182 discharges with acute myocardial infarction, heart failure, or pneumonia as the principal diagnosis for fee-for-service patients ≥ 66 years of age. We excluded 23,158 cases (acute myocardial infarction: 3,907; heart failure: 9,254; pneumonia: 9,997) from subsequent analyses because we could not match their hospital Medicare ID codes (192 hospitals) with either the COTH membership database or Medicare cost reports. Overall, our analyses included 409,736 acute myocardial infarction, 916,793 heart failure, and 685,495 pneumonia discharges from 272 COTH hospitals, 797 non-COTH teaching, and 2,203 nonteaching hospitals.

Back to Top | Article Outline

Outcome variable and covariates

As originally defined by Rosenbaum and Rubin in 1983,11 the “propensity score is the conditional probability of assignment to a particular treatment given a vector of observed covariates.” Propensity scores are most commonly used to balance, match, or stratify patient populations to reduce the inevitable differences in measured confounders between treatment groups that are always present in nonrandomized studies12 and that lead to bias. In logistic models commonly used to estimate propensity scores, the outcome variable is the treatment group rather than a clinical outcome, such as mortality. For the propensity models used in this study, the primary outcome variable was the type of hospital at which a patient was treated rather than a treatment group, but the concept is analogous.

In our study, propensity scores provided a convenient summary of the available information about particular patients. The distribution of these scores among various types of hospitals served to summarize and illustrate differences in overall patient mix.13 Our propensity model covariates included patient demographic characteristics (age, sex, and race [white, black, Hispanic, Asian/Pacific, and Native American/other]) and admission origin (four mutually exclusive groups: transfer from another hospital’s inpatient setting; transfer from another hospital’s emergency department; transfer from a hospital’s own emergency department; and other). Patient medical history and comorbidities were based on the relevant CMS mortality models.14–16 We used one-year “look-back” data (Medicare 2008–2010 inpatient records) both to determine each patient’s past history and comorbidities and to confirm whether a potential comorbidity on the index admission was previously noted and, thus, unlikely to be a complication of care. Based on CMS mortality models, our models included 10 medical diagnoses and 15 comorbidities for acute myocardial infarction, 8 medical diagnoses and 14 comorbidities for heart failure, and 7 medical diagnoses and 22 comorbidities for pneumonia.

We controlled for distance traveled beyond the nearest hospital (log transformation of the difference in distance between the hospital where a patient was actually admitted and the closest hospital to the patient’s zip code) because proximity is a strong confounder of hospital choice.

Because the main outcome of our logistic propensity model was hospital type and not mortality, we included all discharges for analysis and treated each independently. If a patient was transferred from one hospital to the second hospital, we included both discharges. To avoid classifying complications from an initial hospitalization as comorbidities in a subsequent admission, we derived covariate information for both discharges from the first hospital (except for the location of an acute myocardial infarction, which we obtained from both hospitals).

Back to Top | Article Outline

Statistical analysis

We used bivariate analyses to compare demographic, clinical characteristics, and unadjusted 30-day all-cause mortality status (deaths/number of discharges, measured from the day of admission) for all eligible discharges, across all levels of hospital teaching intensity. We used the chi-square test for categorical variables and ANOVA for continuous variables.

We used multinomial logistic regression to estimate the probabilities of admission to a COTH hospital, a non-COTH teaching hospital, and a nonteaching hospital for each patient, conditional on their individual covariates. Operationally, we performed these analyses using nonteaching hospitals as the reference group for all models, and we referred to the estimates as linear propensity scores. For each level of hospital teaching status, we estimated the distribution of the linear propensity scores of all their patients for admission to a COTH hospital, using a kernel density estimator. We repeated these analyses using various methods for differentiating levels of teaching intensity (e.g., COTH membership or various IRB cut points). After estimating the propensity models, we tabulated the odds ratios (for admission to various types of hospitals) associated with various demographic and clinical characteristics. These ratios represent the odds of admission to each type of hospital when the characteristic is present, divided by the odds when that characteristic is not present, adjusted for other measured confounders.

For each of the three conditions, we constructed bar charts showing the distribution of propensity scores, by deciles, for admission to a COTH hospital. We used patients who were actually admitted to COTH hospitals as the reference group and their propensity for admission to a COTH hospital to determine deciles. We then grouped the propensity scores for COTH hospital admission of patients who were actually admitted to non-COTH teaching and nonteaching hospitals into corresponding deciles. COTH, non-COTH, and nonteaching patients who were in the same decile have similar measured confounders, but the probability of admission to a COTH hospital varies substantially by deciles. To illustrate this variability in measured confounders, we determined the characteristics of patients in the two extreme deciles (1 and 10).

Finally, we determined whether there are state or regional differences in the probabilities of admission to COTH or major teaching hospitals. We contrasted patterns of COTH hospital utilization between two states, Massachusetts and California, which differed substantially in total population, geographic area, and the number and percentage of COTH hospitals. Massachusetts has a smaller population and geographic area but a greater percentage of teaching hospitals (13 [22.4%] COTH hospitals, 20 [34.5%] non-COTH teaching hospitals, and 25 [43.1%] nonteaching hospitals). In contrast, California has a larger population and geographic area but proportionately far fewer teaching hospitals (19 [6.5%] COTH hospitals, 80 [27.2%] non-COTH teaching hospitals, and 195 [66.3%] nonteaching hospitals).

We performed all statistical analyses with SAS software, version 9.3 (SAS Institute, Inc., Cary, North Carolina) and plotted kernel density estimators using R software, version 2.14.0 (publicly available under the GNU General Public License at We used a Gaussian kernel density estimate with bandwidth selection using the normal reference distribution method (nrd)—three times nrd to plot the density of probability for admission to a COTH hospital for national patients, and two times nrd for Massachusetts and California patients.

The Partners/Massachusetts General Hospital institutional review board approved this study.

Back to Top | Article Outline


Descriptive statistics

The distribution of hospitals and discharges by level of teaching intensity for each state are available in Supplemental Digital Table 2, Table 1 presents the overall national demographic and clinical characteristics and unadjusted 30-day all-cause mortality measured from date of admission for COTH, non-COTH teaching, and nonteaching hospitals. COTH and non-COTH teaching hospitals cared for a relatively larger percentage of acute myocardial infarction and heart failure patients than pneumonia patients, who were most often cared for by nonteaching hospitals. COTH hospitals admitted approximately twice the percentage of black patients as nonteaching hospitals for all three conditions. After trimming the highest 3% of distance differences, distance beyond nearest hospital increased monotonically across levels of hospital teaching intensity for all three conditions, although the differences were relatively small.

The percentage of acute myocardial infarction or pneumonia patients transferred into COTH hospitals from another hospital’s inpatient setting or emergency department was two to six times higher than that of nonteaching hospitals. For heart failure, there was a sevenfold higher percentage of patients transferred from another hospital’s inpatient setting to a COTH hospital compared with a nonteaching hospital. Conversely, compared with COTH hospitals, seven times as many acute myocardial infarction patients were transferred out of nonteaching hospitals for the same diagnosis and one to two times as many heart failure and pneumonia patients.

Differences in medical history and comorbidities were all statistically significant, in part because of the large sample sizes. Patients with prior cardiovascular conditions, such as percutaneous coronary intervention, coronary artery bypass grafting surgery, acute myocardial infarction, and unstable angina were generally more likely to be admitted to COTH hospitals, especially for cardiac primary diagnoses. There was a higher percentage of patients with a history of hypertension, renal failure, peripheral vascular disease, metastatic cancer, or chronic liver disease in COTH hospitals, and a higher percentage of patients with chronic obstructive pulmonary disease (COPD), dementia, or pneumonia in nonteaching hospitals.

Back to Top | Article Outline

Adjusted odds ratios for admission to teaching hospitals

Table 2 presents the adjusted odds ratios from the propensity models for admission to COTH or non-COTH teaching hospitals compared with nonteaching hospitals. Compared with white patients, black patients had 2.44 (95% confidence interval [CI] 2.36–2.52) times the odds (for acute myocardial infarction) to 2.58 (95% CI 2.51–2.65) times the odds (for pneumonia) of being admitted to COTH hospitals for the three conditions. The odds ratios of admission to a COTH hospital for patients transferred from another hospital’s inpatient setting (versus patients with another admission source, such as elective or direct) ranged from 3.99 (95% CI 3.85–4.13) for acute myocardial infarction to 4.62 (95% CI 4.16–5.12) for pneumonia. Corresponding odds ratios for transfer to a COTH hospital from another hospital’s emergency department ranged from 1.58 (95% CI 1.51–1.65) for heart failure to 2.29 (95% CI 2.15–2.44) for pneumonia. Patients with some chronic comorbidities, such as COPD, pneumonia, diabetes, and dementia, were less likely to be admitted to COTH hospitals, whereas patients with renal failure, chronic liver disease, or metastatic cancer were more likely.

Back to Top | Article Outline

Propensity score analyses

Probabilities of admission to a COTH hospital for all discharges ranged from 0.031 to 0.786 across the three conditions. The deciles of propensity scores for all study patients for admission to a COTH hospital or major teaching hospital are available in Supplemental Digital Table 3 and Supplemental Digital Figure 1, Because this analysis used patients who were actually admitted to COTH hospitals as the reference, the percentage of patients within each propensity decile is 10%. Thus, summing across all deciles accounts for 100% of all patients actually admitted to COTH hospitals.

Conditional on their measured covariates, patients in lower deciles were less likely to be admitted to a COTH hospital, whereas patients in higher deciles were more likely. Table 3 provides the specific characteristics of patients in the highest and lowest probability deciles (10 and 1). Decile 10, characterized by the highest probability of admission to a COTH hospital, had a high percentage of black patients, ranging from 13.2% for acute myocardial infarction to 74.6% for heart failure; a very high percentage of patients transferred from other hospitals’ inpatient settings (96.5%) or emergency departments (3.4%) for acute myocardial infarction; and a higher percentage of acute myocardial infarction patients with serious comorbidities, such as renal failure and chronic liver disease. Decile 1, representing the highest probability of acute myocardial infarction admission to a nonteaching hospital (and lowest probability of admission to a COTH hospital), had no black patients, no patients transferred from other hospitals, a lower percentage of patients with previous cardiovascular history, and a higher percentage of patients with chronic diseases, such as COPD and dementia. Similar findings were observed for heart failure and pneumonia patients.

Figure 1 shows the estimated density plots for the probability of admission to a COTH hospital for acute myocardial infarction, heart failure, and pneumonia in the national cohort. For each graph, the distribution of linear propensity scores for admission to a COTH hospital shifts to the right with increasing hospital teaching intensity. Although not well seen on these graphs because of the scale, a small number of patients with the highest propensity scores in COTH hospitals have very few comparable patients in nonteaching hospitals. Within the top 1% of propensity scores, the percentage of patients admitted to a COTH hospital was four to seven times higher than the percentage admitted to a nonteaching hospital.

Back to Top | Article Outline

Between-state variation in the use of teaching hospitals

Figure 2 illustrates significant differences across states in patients’ characteristics at hospitals of varying teaching intensity, using Massachusetts and California as examples. For each condition, high-linear-propensity-score patients are much more likely to be cared for at COTH hospitals in Massachusetts, which has a large proportion of such tertiary/quaternary institutions. There were 109 acute myocardial infarction patients in Massachusetts within the top 1% of propensity scores admitted to COTH hospitals, whereas there was only 1 comparable patient at non-COTH teaching or nonteaching hospitals. Conversely, in California, with proportionately fewer teaching hospitals, the distribution of linear propensity scores across the three types of hospitals was quite similar, with few patients uniquely cared for only by COTH hospitals. The findings for heart failure and pneumonia were similar but not as striking.

Back to Top | Article Outline

Alternative hospital teaching intensity classification methods

We observed similar findings to all the preceding analyses when we used various IRB criteria rather than COTH status to categorize hospital teaching intensity (according to the same U.S. national data) (see Supplemental Digital Tables 4–9 and Supplemental Digital Figures 2–4,, for complete data).

Back to Top | Article Outline


Teaching hospitals, especially COTH members, serve a number of important clinical functions in addition to their educational and research missions. Two of these functions—caring for minority populations and serving as a referral center for transfer patients, many of whom are complex and severely ill—are readily apparent in our study of three common conditions. Both illustrate the added value of teaching hospitals in the care of patient cohorts who are generally at higher risk of adverse outcomes.

Back to Top | Article Outline

Care of minority populations

COTH hospitals provide a disproportionate percentage of the care of black patients, as previous studies of minority, underserved, and poor populations have demonstrated.1,2,6,7,17–20 These populations have worse overall health care outcomes, presumably resulting from their generally lower education and socioeconomic status, lack of regular preventive care, delay in accessing acute care, and receipt of less effective, evidence-based care.19,21–37 Disparities in the care of minority and other vulnerable populations led the Institute of Medicine to make the provision of equitable care a major goal of health care reform.29,38

There are several implications of the higher proportion of minority populations cared for at COTH hospitals, including potentially higher-quality and more equitable care for such patients at these institutions.18,19,39 In the study by Kahn and colleagues,18 black or poor Medicare patients received generally inferior processes of care during hospitalizations and were more unstable at discharge compared with other patients. However, the investigators found no overall differences in quality of care by race or poverty status. This was likely because black and poor patients were 1.8 times more likely to receive their care in urban teaching hospitals, which provided higher overall quality of care.

The high concentration of minority populations at teaching hospitals also presents an opportunity for health system improvement.21 Major teaching hospitals can collect and analyze disparities data and train the next generation of physicians to provide just and equitable care regardless of race, ethnicity, gender, or socioeconomic status.

Back to Top | Article Outline

Referral center for complex and severely ill transfer patients

Our study also confirms another unique function of teaching hospitals, especially COTH members—their role as the predominant destination for transfer patients. Most transfer patients are too complex or severely ill to be cared for at their original hospital, or they require specialized services or procedures that are uniquely available at the receiving hospital, most often a teaching hospital (e.g., burn service, cardiac surgery, transplant).40 Our study confirms that COTH hospitals remain the regional “courts of last resort,” and few patients are ever transferred out of COTH teaching hospitals.

Other studies have demonstrated that the necessity to be transferred defines a group of patients at exceptionally high risk for mortality, complications, long length of stay, and high resource consumption.41–50 Gordon and Rosenthal42 estimated that, independent of quality of care, severity-adjusted mortality and length of stay would appear 17% and 8% higher, respectively, for hospitals having 20% interhospital transfer patients compared with hospitals having 2%. Rosenberg and colleagues51 found that even after comprehensive adjustment for case mix and severity of illness, transfer patients had a 38% longer medical intensive care unit stay, a 41% longer hospital stay, and 2.2-fold higher odds of hospital mortality than directly admitted patients. The significantly larger percentage of patients transferred to COTH hospitals will negatively impact their performance results, whereas the converse is true for the hospitals that transfer out these sicker patients.52 That COTH hospitals in our study had both a significantly higher percentage of transfer patients and a significantly lower unadjusted mortality rate compared with nonteaching hospitals is circumstantial evidence that they provide higher-quality care.

Our propensity score decile analyses demonstrate that aggregate patient characteristics vary in direct proportion to hospital teaching intensity. Although overlap does exist, there are certain types of patients at COTH hospitals who are rarely seen at nonteaching hospitals. Our data from Massachusetts and California suggested that the distribution of patients among hospitals of varying teaching intensity varies regionally, presumably because of both the relative availability of AMCs and local referral patterns.

Back to Top | Article Outline


Our analyses were based on contemporary Medicare data and may not be applicable to other age groups and payers. As with any study based on administrative data, there are inherent limitations in the accuracy with which patient clinical status is characterized. Furthermore, as in any observational study, there is the potential for unmeasured confounders.

Back to Top | Article Outline


In addition to their medical education and research missions, teaching hospitals, especially COTH members, provide a disproportionate share of the care to historically underserved minority populations. They also are regional and national referral centers, accepting in transfer the most complex and critically ill patients—those who are most likely to experience adverse outcomes and to consume more resources.

This study focused on a detailed analysis of the characteristics of patients cared for by hospitals of varying teaching intensity. We plan to conduct additional comparative research into the risk-adjusted outcomes at these hospitals. Furthermore, we will expand our studies to include other conditions and procedures beyond the three described in this report.

Back to Top | Article Outline


1. Blumenthal D, Campbell EG, Weissman JS. The social missions of academic health centers. N Engl J Med. 1997;337:1550–1553
2. Commonwealth Fund Task Force on Academic Health Centers. Envisioning the Future of Academic Health Centers. 2003 New York, NY Commonwealth Fund
3. Centers for Medicare and Medicaid Services. . Medicare Provider Analysis and Review File (MEDPAR). Accessed July 29, 2013.
4. Centers for Disease Control and Prevention. . International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Accessed July 29, 2013
5. Ayanian JZ, Weissman JS. Teaching hospitals and quality of care: A review of the literature. Milbank Q. 2002;80:569–593, v
6. Allison JJ, Kiefe CI, Weissman NW, et al. Relationship of hospital teaching status with quality of care and mortality for Medicare patients with acute MI. JAMA. 2000;284:1256–1262
7. Rosenthal GE, Harper DL, Quinn LM, Cooper GS. Severity-adjusted mortality and length of stay in teaching and nonteaching hospitals. Results of a regional study. JAMA. 1997;278:485–490
8. Dimick JB, Cowan JA Jr, Colletti LM, Upchurch GR Jr. Hospital teaching status and outcomes of complex surgical procedures in the United States. Arch Surg. 2004;139:137–141
9. Silber JH, Rosenbaum PR, Romano PS, et al. Hospital teaching intensity, patient race, and surgical outcomes. Arch Surg. 2009;144:113–120
10. Volpp KG, Rosen AK, Rosenbaum PR, et al. Mortality among hospitalized Medicare beneficiaries in the first 2 years following ACGME resident duty hour reform. JAMA. 2007;298:975–983
11. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70:41–55
12. D’Agostino RB Jr. Propensity score methods for bias reduction in the com parison of a treatment to a non-randomized control group. Stat Med. 1998;17:2265–2281
13. Shahian DM, Normand SL. Comparison of “risk-adjusted” hospital outcomes. Circulation. 2008;117:1955–1963
14. 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;113:1683–1692
15. 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;113:1693–1701
16. 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:e17401
17. Ayanian JZ, Weissman JS, Chasan-Taber S, Epstein AM. Quality of care for two common illnesses in teaching and nonteaching hospitals. Health Aff (Millwood). 1998;17:194–205
18. Kahn KL, Pearson ML, Harrison ER, et al. Health care for black and poor hospitalized Medicare patients. JAMA. 1994;271:1169–1174
19. Ayanian JZ, Weissman JS, Chasan-Taber S, Epstein AM. Quality of care by race and gender for congestive heart failure and pneumonia. Med Care. 1999;37:1260–1269
20. Moy E, Valente E Jr, Levin RJ, Griner PF. Academic medical centers and the care of underserved populations. Acad Med. 1996;71:1370–1377
21. Betancourt JR. Eliminating racial and ethnic disparities in health care: What is the role of academic medicine? Acad Med. 2006;81:788–792
22. Andrulis DP. Access to care is the centerpiece in the elimination of socioeconomic disparities in health. Ann Intern Med. 1998;129:412–416
23. Antonovsky A. Social class and the major cardiovascular diseases. J Chronic Dis. 1968;21:65–106
24. Bach PB, Cramer LD, Warren JL, Begg CB. Racial differences in the treatment of early-stage lung cancer. N Engl J Med. 1999;341:1198–1205
25. Gornick ME, Eggers PW, Reilly TW, et al. Effects of race and income on mortality and use of services among Medicare beneficiaries. N Engl J Med. 1996;335:791–799
26. Pincus T, Esther R, DeWalt DA, Callahan LF. Social conditions and self-management are more powerful determinants of health than access to care. Ann Intern Med. 1998;129:406–411
27. Harris DR, Andrews R, Elixhauser A. Racial and gender differences in use of procedures for black and white hospitalized adults. Ethn Dis. 1997;7:91–105
28. Hinkle LE Jr, Whitney LH, Lehman EW, et al. Occupation, education, and coronary heart disease. Risk is influenced more by education and background than by occupational experiences, in the Bell System. Science. 1968;161:238–246
29. Institute of Medicine of the National Academies. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. 2003 Washington, DC National Academies Press
30. Jha AK, Fisher ES, Li Z, Orav EJ, Epstein AM. Racial trends in the use of major procedures among the elderly. N Engl J Med. 2005;353:683–691
31. Johnson PA, Lee TH, Cook EF, Rouan GW, Goldman L. Effect of race on the presentation and management of patients with acute chest pain. Ann Intern Med. 1993;118:593–601
32. Peterson E, Yancy CW. Eliminating racial and ethnic disparities in cardiac care. N Engl J Med. 2009;360:1172–1174
33. Peterson ED, Shaw LK, DeLong ER, Pryor DB, Califf RM, Mark DB. Racial variation in the use of coronary-revascularization procedures. Are the differences real? Do they matter? N Engl J Med. 1997;336:480–486
34. Schulman KA, Berlin JA, Harless W, et al. The effect of race and sex on physicians’ recommendations for cardiac catheterization. N Engl J Med. 1999;340:618–626
35. Todd KH, Samaroo N, Hoffman JR. Ethnicity as a risk factor for inadequate emergency department analgesia. JAMA. 1993;269:1537–1539
36. Vaccarino V, Rathore SS, Wenger NK, et al.National Registry of Myocardial Infarction Investigators. Sex and racial differences in the management of acute myocardial infarction, 1994 through 2002. N Engl J Med. 2005;353:671–682
37. Wisnivesky JP, McGinn T, Henschke C, Hebert P, Iannuzzi MC, Halm EA. Ethnic disparities in the treatment of stage I non-small cell lung cancer. Am J Respir Crit Care Med. 2005;171:1158–1163
38. Institute of Medicine. Crossing the Quality Chasm: A New Health System for the 21st Century. 2001 Washington, DC National Academies Press
39. Goldman LE, Vittinghoff E, Dudley RA. Quality of care in hospitals with a high percent of Medicaid patients. Med Care. 2007;45:579–583
40. Berman RA, Green J, Kwo D, Safian KF, Botnick L. Severity of illness and the teaching hospital. J Med Educ. 1986;61:1–9
41. Bernard AM, Hayward RA, Rosevear J, Chun H, McMahon LF. Comparing the hospitalizations of transfer and non-transfer patients in an academic medical center. Acad Med. 1996;71:262–266
42. Gordon HS, Rosenthal GE. Impact of interhospital transfers on outcomes in an academic medical center. Implications for profiling hospital quality. Med Care. 1996;34:295–309
43. Durairaj L, Will JG, Torner JC, Doebbeling BN. Prognostic factors for mortality following interhospital transfers to the medical intensive care unit of a tertiary referral center. Crit Care Med. 2003;31:1981–1986
44. Escarce JJ, Kelley MA. Admission source to the medical intensive care unit predicts hospital death independent of APACHE II score. JAMA. 1990;264:2389–2394
45. Borlase BC, Baxter JK, Kenney PR, Forse RA, Benotti PN, Blackburn GL. Elective intrahospital admissions versus acute interhospital transfers to a surgical intensive care unit: Cost and outcome prediction. J Trauma. 1991;31:915–918
46. Seferian EG, Afessa B, Gajic O, Keegan MT, Hubmayr RDMayo Epidemiology and Translational Research in Intensive Care. . Comparison of community and referral intensive care unit patients in a tertiary medical center: Evidence for referral bias in the critically ill. Crit Care Med. 2008;36:2779–2786
47. Golestanian E, Scruggs JE, Gangnon RE, Mak RP, Wood KE. Effect of interhospital transfer on resource utilization and outcomes at a tertiary care referral center. Crit Care Med. 2007;35:1470–1476
48. Odetola FO, Clark SJ, Gurney JG, Dechert RE, Shanley TP, Freed GL. Effect of interhospital transfer on resource utilization and outcomes at a tertiary pediatric intensive care unit. J Crit Care. 2009;24:379–386
49. Combes A, Luyt CE, Trouillet JL, Chastre J, Gibert C. Adverse effect on a referral intensive care unit’s performance of accepting patients transferred from another intensive care unit. Crit Care Med. 2005;33:705–710
50. Flabouris A, Hart GK, George C. Outcomes of patients admitted to tertiary intensive care units after interhospital transfer: Comparison with patients admitted from emergency departments. Crit Care Resusc. 2008;10:97–105
51. Rosenberg AL, Hofer TP, Strachan C, Watts CM, Hayward RA. Accepting critically ill transfer patients: Adverse effect on a referral center’s outcome and benchmark measures. Ann Intern Med. 2003;138:882–890
52. Kahn JM, Kramer AA, Rubenfeld GD. Transferring critically ill patients out of hospital improves the standardized mortality ratio: A simulation study. Chest. 2007;131:68–75

Supplemental Digital Content

Back to Top | Article Outline
© 2014 by the Association of American Medical Colleges