Secondary Logo

Comparison of Direct Patient Care Costs and Quality Outcomes of the Teaching and Nonteaching Hospitalist Services at a Large Academic Medical Center

Perez, Jose, A., Jr, MD, MSEd, MBA; Awar, Melina, MD; Nezamabadi, Aryan, MD; Ogunti, Richard, MD, MPH; Puppala, Mamta, MS; Colton, Lara, MD; Clewing, Johanna, M., MD; Ketkar, Sayali, MPH; Wong, Stephen T., C., PhD; Robbins, Richard, J., MD

doi: 10.1097/ACM.0000000000002026
Research Reports
Free

Purpose To compare costs of care and quality outcomes between teaching and nonteaching hospitalist services, while testing the assumption that resident-driven care is more expensive.

Method Records of inpatients with the top 20 Medicare Severity Diagnosis-Related Groups admitted to the University Teaching Service (UTS) and nonteaching hospitalist service (NTHS) at Houston Methodist Hospital from 2014–2015 were analyzed retrospectively. Direct costs of care, length of stay (LOS), in-hospital mortality (IHM), 30-day readmission rate (30DRR), and consultant utilization were compared between the UTS and NTHS. Propensity score matching and case mix index (CMI) were used to mitigate differences in baseline characteristics. To compare outcomes between matched groups, the Wilcoxon rank sum test and chi-square test were used. A sensitivity analysis was conducted using multivariable regression analysis.

Results From the overall study population of 8,457 patients, 1,041 UTS and 3,123 NTHS patients were matched. CMI was 1.07 for each group. The UTS had lower direct costs of care per case ($5,028 vs. $5,502, P = .006), lower LOS (4.7 vs. 5.2 days, P = .0002), and lower consultant utilization (1.0 vs. 1.6, P ≤ .0001) versus the NTHS. The UTS and NTHS 30DRR (17.2% vs. 19.3%, P = .110) and IHM (2.9% vs. 3.7%, P = .206) were comparable. The multivariable regression analysis validated the matched data and identified an incremental cost savings of $333/UTS patient.

Conclusions Patients of an academic hospitalist service had significantly shorter LOS, fewer consultants, and lower direct care costs than comparable patients of a nonteaching service.

J.A. Perez Jr is professor of clinical medicine, Department of Medicine, Houston Methodist Institute for Academic Medicine and Weill Cornell Medicine, Houston Methodist Hospital, Houston, Texas.

M. Awar is assistant professor of clinical medicine, Department of Medicine, Houston Methodist Institute for Academic Medicine and Weill Cornell Medicine, Houston Methodist Hospital, Houston, Texas.

A. Nezamabadi is a hospitalist, Salinas Valley Healthcare System, Salinas, California. At the time of the study, the author was a third-year internal medicine resident, Houston Methodist Hospital, Houston, Texas.

R. Ogunti is a first-year resident, Internal Medicine, Howard University, Washington, DC. At the time of the study, the author was a project specialist, Department of Systems Medicine and Bioengineering, Houston Methodist Hospital, Houston, Texas.

M. Puppala is a senior applications analyst, Department of Systems Medicine and Bioengineering, Houston Methodist Hospital, Houston, Texas.

L. Colton is assistant professor of clinical medicine, Department of Medicine, Houston Methodist Institute for Academic Medicine and Weill Cornell Medicine, Houston Methodist Hospital, Houston, Texas.

J.M. Clewing is assistant professor of clinical medicine, Department of Medicine, Houston Methodist Institute for Academic Medicine and Weill Cornell Medicine, Houston Methodist Hospital, Houston, Texas.

S. Ketkar is clinical quality analytics/integration manager, Houston Methodist Hospital, Houston, Texas.

S.T.C. Wong is professor and chair, Department of Systems Medicine and Bioengineering, Houston Methodist Institute for Academic Medicine and Weill Cornell Medicine, Houston Methodist Hospital, Houston, Texas.

R.J. Robbins is professor and chair, Department of Medicine, Houston Methodist Institute for Academic Medicine and Weill Cornell Medicine, Houston Methodist Hospital, Houston, Texas.

Funding/Support: John S. Dunn Research Foundation (S.T.C.W.).

Other disclosures: None reported.

Ethical approval: This study obtained an exemption by the institutional review board of Houston Methodist Hospital, May 12, 2015.

Previous presentations: This research was presented in poster format at the Alliance for Academic Internal Medicine Meeting, Baltimore, Maryland, March 20, 2017.

* The HMH hospitalist program’s governance structure was established in 2013, after which all hospitalists were registered and began to participate in a quality incentive (bonus) program. Before then, it was difficult to determine which physicians were hospitalists. Therefore, in this study, we were not able to compare the UTS and NTHS groups using data for patients admitted prior to January 2014.

Correspondence should be addressed to Jose A. Perez Jr, Houston Methodist Hospital, 6550 Fannin SM1001, Houston, TX 77030; telephone: (713) 441-6729; e-mail: JAPerez@houstonmethodist.org.

Since the inception of Medicare in 1965, health care costs in the United States have continued to rise, with hospital care constituting the largest proportion (32%) and consuming approximately $1 trillion per year as of 2014.1 Strategies to reduce hospital costs are at the center of efforts to preserve Medicare. Care coordinated by hospitalists can reduce length of stay (LOS) and costs for inpatient hospital admissions.2 This may be beneficial at academic medical centers (AMCs), which have traditionally been considered inefficient and expensive as a result of treating higher-acuity patients and training inexperienced residents. With the rise of hospitalist-run teaching services at AMCs, however, this view may be changing.

Several studies have attempted to determine whether inpatient medical care costs more when provided on a teaching hospitalist service compared with a nonteaching hospitalist service (NTHS) at the same AMC. Kulaga et al3 and Everett et al4 found that teaching services had significantly lower LOS and costs compared with private nonteaching services. Other studies, however, have found variable effects on clinical outcomes and no differences in overall patient care costs between the two models.5,6 Recently, Iannuzzi et al7 reported that hospitalist–resident teams had lower LOS and costs compared with hospitalist–midlevel practitioner teams, but they found no differences in the in-hospital mortality rate or the 30-day readmission rate (30DRR).

In response to a growing national effort to improve the value of medical care, several initiatives have provided practical guidance.8 The High Value Care (HVC) Curriculum was jointly developed by the Alliance for Academic Internal Medicine (AAIM) and the American College of Physicians (ACP) and released in July 2012, with the goal of training physicians to help eliminate health care waste and over-ordering of tests.9 The Choosing Wisely educational campaign of the American Board of Internal Medicine Foundation was created to encourage medical specialty societies to identify practices in their field that do not add value and to discuss strategies aimed at reducing health care costs.10 These educational tools have been incorporated into clinical training programs as evidenced by the VALUE framework, which was designed to help residency programs educate physicians-in-training about providing value-based care.11

Since early 2013, the internal medicine residency program at the Houston Methodist Hospital (HMH) has emphasized HVC training for our medical residents and core teaching faculty, including academic hospitalists. This began with the incorporation of the AAIM–ACP HVC Curriculum into the didactic conference schedule and was followed in 2015 by the establishment of monthly HVC morning reports, which in 2017 evolved into HVC bimonthly conferences along with journal clubs that review clinical guidelines. Weekly Choosing Wisely updates have been provided at Medicine Grand Rounds, along with dedicated faculty development sessions geared toward the practice of HVC, since 2015.

In contrast, there are no formal HVC programs for the NTHS physician group at the HMH. These hospitalists are private practitioners employed by local medical groups and are not affiliated with the academic enterprise. Although there is an expectation that they will keep up with the medical literature, and they may be aware of HVC, their participation in related educational activities is not required and cannot be quantified.

In this study, we sought to compare the direct costs of patient care and quality outcomes between the inpatient academic service and the NTHS at our large AMC while testing the assumption that resident-driven care is more expensive.

Back to Top | Article Outline

Method

Setting

The HMH is an 830-bed, tertiary AMC with more than 250 residents in training, including 51 approved positions in the internal medicine residency program. A large medical student program, comprising three major medical school affiliations (Weill Cornell Medicine, Texas A&M College of Medicine, and the University of Texas Medical Branch School of Medicine), is also part of the academic environment.

In 2008, one year after the opening of the internal medicine residency program, the University Teaching Service (UTS) was established as the academic inpatient general medicine service. The HMH employs academic hospitalists, who serve as the teaching faculty for the UTS. At the time of this study in 2014–2015, the UTS consisted of four provider teams, each with one or two medical students, two postgraduate year 1 (PGY-1) residents, and one PGY-2/PGY-3 resident. The UTS maintains a resident night team in-house for patient care. The academic hospitalists do not have a night team, nor do they spend nights in the hospital.

The UTS admits patients from the Emergency Department who are assigned as part of the “no-doc” call—that is, patients without an identifiable HMH attending physician. No-doc admissions are rotated between the UTS and NTHS within the hospital. The UTS also admits all patients being transferred to the HMH for a higher level of care unless the NTHS physician has a prior referral relationship with the outside physician.

Twelve UTS attending physicians were identified in this study’s retrospective analysis of patient data (described below). All were board certified in internal medicine, and 2 (17%) had an additional certification in pulmonary and critical care. Forty-eight NTHS physicians were identified. Almost all (n = 46; 96%) were board certified in internal medicine, and 8 (17%) had an additional certification (nephrology = 3, geriatrics = 2, critical care = 1, sleep medicine = 1, and hospice and palliative medicine = 1). The NTHS physicians were not involved in teaching activities and did not host residents or medical students on their services. Some of these NTHS physicians occasionally worked with midlevel providers. Patients admitted to the UTS were neither geographically localized nor segregated from NTHS patients. Thus, the same nurses and support staff cared for both UTS and NTHS patients within nursing units.

As described above, HVC training was emphasized for the residents and UTS faculty during the study period. The NTHS physicians did not participate in this training.

Back to Top | Article Outline

Data

Patient data were obtained from the Methodist Environment for Translational Enhancement and Outcomes Research (METEOR), an enterprise-wide clinical data warehouse and analytics environment that integrates existing business data warehouse and patient records across the HMH system to support clinical research and outcome studies.12 The METEOR data warehouse contains records dating back to January 1, 2006, with more than 3 million unique patients and over 10 million unique patient encounters that encompass a range of diverse, heterogeneous data. An individual record usually consists of multiple binary (e.g., gender), categorical (e.g., race), and continuous (e.g., age) attributes at a given time point; time-lapse information (e.g., multiple blood pressure reads during the patient’s stay in the hospital); and nonalphanumeric data (e.g., free text and images).

Using METEOR, we retrospectively analyzed records of all inpatients admitted to the UTS comprising the top 20 Medicare Severity Diagnosis-Related Groups (MS-DRGs) in 2014 and the same MS-DRGs in 2015 (see Table 1). We also analyzed the records of all inpatients admitted to the NTHS in 2014 and 2015 comprising the same 20 MS-DRGs.*

Table 1

Table 1

Within METEOR, the 3M Grouper Plus System (3M Company, Salt Lake City, Utah), a commonly employed proprietary tool, was used to determine severity of illness (SOI) and risk of mortality (ROM).13 These two parameters were calculated both at admission and at discharge once charges were generated; in this analysis we used only admission SOI and ROM. Discharge type was tabulated independently and had no connection to either SOI or ROM.

We compared patients with the top 20 MS-DRGs admitted to the UTS versus those admitted to the NTHS during January 2014–December 2015. We eliminated from the study data any patient with an intensive care unit (ICU) stay, as dedicated intensivists assumed care when a patient entered an ICU and because all patients were admitted to an ICU under the primary hospitalist’s name with mandatory intensivist consultation, whether the patients came from the Emergency Department or transferred from the floor. Although ICU costs should be considered in the overall cost of care, because neither the UTS residents nor the NTHS physicians actually managed ICU patients, we felt including ICU costs would not be an accurate reflection of practice patterns for either residents or private hospitalists.

Direct patient care costs were calculated in accordance with the HMH activity costing guidelines and methodology using the Allscripts Sunrise Enterprise Performance Systems Incorporated (EPSi) Costing Module (version 7.5 SP2, EPSi, San Ramon, California). Direct costs were defined as any hospital resources used to provide patient services, exclusive of professional fees. Indirect costs, such general support for operating the facility, were not included. Because one of the goals of this study was to compare practice patterns and outcomes and because activity costing for resident and faculty services would be difficult to calculate, neither professional fees paid to NTHS physicians or UTS attending physicians, nor resident salary costs, both paid by the Centers for Medicare and Medicaid Services (CMS), were included.

Back to Top | Article Outline

Statistical analysis

To ensure similar baseline characteristics of patients between the two groups (UTS and NTHS), propensity score matching was performed to reduce confounders and improve the homogeneity of the case mix. Propensity is the probability of a patient being assigned to either the UTS or NTHS on the basis of the respective patient characteristics. Patients with the same propensity score would have similar characteristics. Using nearest neighbor matching without replacement, we constructed a cohort matching each UTS patient to the three closest NTHS patients and discarding those NTHS patients who were not selected as matches.

Propensity score matching was performed based on all covariates listed in Table 1, and a standardized mean difference (SMD) was generated. The matched data, also shown in Table 1, contained only patients who were similar in baseline characteristics and therefore could provide an unbiased assessment of each group on the measured outcomes. To further ensure homogeneity between the two patient groups, a case mix index (CMI) was calculated for each group. The ability of matching to balance the two groups was assessed with the SMD for each of the covariates. The SMD is the ratio of the difference in main outcomes between groups divided by the difference in standard deviations among participants. An SMD < 10%, for a given covariate, was considered to be an inconsequential imbalance.14

Outcome measures were compared between the two groups, post match, using the Wilcoxon rank sum test for continuous variables and the chi-square test for categorical outcomes variables. Data were analyzed for LOS, in-hospital mortality, 30DRR, utilization of specialist consultants, and direct patient care costs. A sensitivity analysis was conducted to validate the results obtained in the propensity-score-matched cohort. We included all variables for this model-based adjustment. All statistical tests were two sided, and tests with P < .05 were considered significant. Propensity score matching was performed using the MatchIt package14 for R software (R for Windows 3.2.4; R Foundation for Statistical Computing, Vienna, Austria). Further statistical analysis was performed with STATA statistical software version 13 for Windows (StataCorp LP, College Station, Texas).

This study was determined to be exempt by the institutional review board of the HMH.

Back to Top | Article Outline

Results

The total number of admissions during January 2014–December 2015 was 3,627 for the UTS and 33,556 for the NTHS. On the basis of the top 20 MS-DRGs, 8,457 admissions were reviewed, including 1,041 on the UTS and 7,416 on the NTHS. Propensity score matching, in a ratio of 1 UTS admission to 3 NTHS admissions, decreased the latter number to 3,123. The demographics and baseline characteristics of the two patient groups before and after propensity score matching are shown in Table 1.

After propensity score matching, there were no statistically significant differences between the two groups in patient characteristics, including gender, race, discharge type, MS-DRG, SOI, ROM, or payer (Table 1). The CMI was 1.07 for both groups. The UTS group admitted more of its patients from outside hospitals via the transfer center than did the NTHS (UTS 20.5% vs. NTHS 10.8%) and less directly from the Emergency Department (UTS 76.2% vs. NTHS 85.5%). This was associated with high SMDs of 24 and 21.7, respectively.

Table 2 shows the data for LOS, 30DRR, direct patient care costs, in-hospital mortality, and consultant utilization. Care on the UTS was associated with lower direct costs per case (UTS $5,028 vs. NTHS $5,502, P = .006), lower LOS (UTS 4.7 vs. NTHS 5.2 days, P = .0002), and lower consultant utilization (UTS 1.0 vs. NTHS 1.6, P < .0001). The 30DRR (UTS 17.2% vs. NTHS 19.3%, P = .110) and in-hospital mortality (UTS 2.9% vs. NTHS 3.7%, P = .206) were comparable between the two groups.

Table 2

Table 2

Table 3 shows that, using a multivariate linear regression model, the incremental direct cost savings for every patient managed by the UTS group were $333 (P = .016), which was statistically significant. Similar results were noted for the UTS for LOS (0.37 days [8.8 hours] less than NTHS; P = .001) and consultant utilization (60% of equivalent use by NTHS; P < .001). Table 3 also shows that the sensitivity analyses done using multivariable regression analysis validated the propensity-score-matched data.

Table 3

Table 3

Back to Top | Article Outline

Discussion

In this study, we demonstrated that an academic inpatient service (UTS) achieved statistically significantly lower LOS and direct patient care costs compared with a private hospitalist service (NTHS) at the same large, tertiary AMC for a matched set of patients. No significant difference in 30DRR or in-hospital mortality between groups was identified after propensity score matching and multivariate regression analysis. If the UTS cost savings of $333 per patient were extrapolated only to the 3,123 NTHS patients used in this study for propensity score matching, a savings of more than $1.4 million over the two-year period could have been achieved. The significantly shorter LOS (mean of 8.8 hours less) on the UTS also likely contributed to the lower direct costs of care.

Our results are qualitatively similar to those of Everett et al4 and Chin et al,15 who both noted lower LOS and costs for an academic service compared with private hospitalists. However, those two studies found a higher 30DRR for the teaching service. In a smaller study, using methods similar to ours (including propensity score matching to mitigate bias), Khaliq et al6 compared 1,637 patients on an academic service with 522 patients on a nonteaching service, and found no difference in overall patient care costs, LOS, or in-hospital mortality.

In our study, the patients in both groups were well matched for all variables analyzed. Consultant utilization on the UTS was significantly lower than on the NTHS and likely contributed to the lower direct costs of care. The higher consultant utilization on the NTHS cannot be explained by differences in SOI given our patient groups. It is known that patients who need consultants generally have longer LOS and higher SOI.16 To our knowledge, this is the first study to specifically compare the utilization of consultants on teaching versus nonteaching services. The phenomenon of hospitalist–consultant interactions may be rooted in behaviors that have been socialized into practice rather than on patient complexity, but that was not formally evaluated in our study.

While our residency program has implemented the AAIM–ACP HVC Curriculum and other educational activities to imprint HVC as a core competency, the data collection period for this study preceded the full implementation of our curriculum. We cannot demonstrate that the expansion of our curriculum has led to results, but we do host many required educational activities that teaching faculty frequently participate in or facilitate. These activities include a bimonthly evidence-based medicine conference and a monthly HVC morning report which has, since the study period, evolved into bimonthly HVC conferences and monthly journal club conferences where updated clinical guidelines are presented and expert group recommendations are taught. Monthly morbidity and mortality conferences are consistently attended by the teaching faculty, and the selected cases often identify lapses in the standard of care and/or highlight practices that result in wastefulness (i.e., unnecessary testing and procedures). The online AAIM–ACP HVC Curriculum9 covers a range of similar topics, including avoiding unnecessary testing and identifying barriers to HVC. These topics have been presented as part of our mandatory noon conferences. Finally, faculty development sessions aimed at teaching faculty, but inclusive of all trainees, have provided teaching tips and resources to promote HVC utilization and the practice of evidence-based medicine.

It is known that residents are not often aware of the actual costs of care, and few internal medicine or family medicine residency programs have a formal curriculum in cost-conscious care.17,18 However, there are data suggesting that spending behaviors learned in residency persist and are a result of the environment of practice.19,20 To better educate residents, faculty must be better prepared to teach, promote, and role model cost-conscious care. Unfortunately, data suggest that this preparation may currently be lacking.21 If we are to intervene with the goal of educating future physicians about the costs of care to have an impact on their behavior, our interventions likely should take place during residency training.

This study had some limitations. We did not include resident or faculty salaries or professional fees paid to UTS or NTHS physicians. This allowed the focus to be on practice behaviors that could be analyzed. Also, these are costs paid by the CMS. It is known that many hospitals independently provide subsidies to hospitalists. That is not the case at the HMH, although a quality incentive was introduced in 2013. In addition, patient volume is different for the UTS and NTHS groups at our institution. Each PGY-1 resident is limited to a maximum of 10 patients, and the UTS team census is limited to 20, while the NTHS physicians may carry many more patients than this. It is known that hospitalist workload may negatively impact outcome measures including LOS and cost, but the outcomes for in-hospital mortality and 30DRR appear to be unaffected by workload.22 Also, it has been noted that the team approach seen on the UTS, with resident physicians in-house 24 hours per day, may better attend to the needs of patients and result in decreased LOS,4 but this was not formally studied. Achieving lower LOS may involve many factors. One that was perhaps active may have been a resident’s natural eagerness to reduce his or her caseload.15 In our study, no apparently negative outcomes were seen as a result of this potential inclination.

Further, several potential confounders may exist in our study. Although Emergency Department physicians chose which no-doc patients were admitted to the UTS, we cannot ensure that there was a true random distribution, which may have led to bias. We attempted to mitigate this by using the propensity-score-matching method. Additionally, the patients admitted to the UTS included more direct hospital transfers and fewer patients admitted directly from the ED compared with the patients admitted to the NTHS. This difference persisted after matching and could have played a role in the results. Finally, given that this was a single-institution study, generalizing these outcomes beyond our AMC should be done cautiously.

In summary, patients cared for by an academic service had significantly shorter LOS, used significantly fewer consultants, and had significantly lower direct care costs than comparable patients cared for by private hospitalists. In-hospital mortality and 30DRR were no different for the two services. Many AMCs have one large employed hospitalist service for all admissions, making head-to-head comparisons of teaching versus nonteaching services difficult. In this retrospective study, however, we had the opportunity to compare two different hospitalist models in the same AMC and on the same nursing units. We conclude that the quality and costs of care of a traditional academic resident–faculty hospitalist team at our medical center were as good as—or better than—those of a group of private hospitalists.

Acknowledgments: The authors thank Nan Chi, director of budget and compliance, Houston Methodist Hospital, Houston, Texas.

Back to Top | Article Outline

References

1. Centers for Disease Control and Prevention, National Center for Health Statistics. Healthcare expenditures. 2015. http://www.cdc.gov/nchs/fastats/health-expenditures.htm. Updated May 3, 2017. Accessed September 29, 2017.
2. Rachoin JS, Skaf J, Cerceo E, et al. The impact of hospitalists on length of stay and costs: Systematic review and meta-analysis. Am J Manag Care. 2012;18:e23–e30.
3. Kulaga ME, Charney P, O’Mahony SP, et al. The positive impact of initiation of hospitalist clinician educators. J Gen Intern Med. 2004;19:293–301.
4. Everett G, Uddin N, Rudloff B. Comparison of hospital costs and length of stay for community internists, hospitalists, and academicians. J Gen Intern Med. 2007;22:662–667.
5. Shine D, Beg S, Jaeger J, Pencak D, Panush R. Association of resident coverage with cost, length of stay, and profitability at a community hospital. J Gen Intern Med. 2001;16:1–8.
6. Khaliq AA, Huang CY, Ganti AK, Invie K, Smego RA Jr. Comparison of resource utilization and clinical outcomes between teaching and nonteaching medical services. J Hosp Med. 2007;2:150–157.
7. Iannuzzi MC, Iannuzzi JC, Holtsbery A, Wright SM, Knohl SJ. Comparing hospitalist-resident to hospitalist-midlevel practitioner team performance on length of stay and direct patient care cost. J Grad Med Educ. 2015;7:65–69.
8. Porter ME. What is value in health care? N Engl J Med. 2010;363:2477–2481.
9. Smith CD; Alliance for Academic Internal Medicine–American College of Physicians High Value; Cost-Conscious Care Curriculum Development Committee. Teaching high-value, cost-conscious care to residents: The Alliance for Academic Internal Medicine–American College of Physicians Curriculum. Ann Intern Med. 2012;157:284–286.
10. Cassel CK, Guest JA. Choosing wisely: Helping physicians and patients make smart decisions about their care. JAMA. 2012;307:1801–1802.
11. Patel MS, Davis MM, Lypson ML. The VALUE framework: Training residents to provide value-based care for their patients. J Gen Intern Med. 2012;27:1210–1214.
12. Puppala M, He T, Chen S, et al. METEOR: An enterprise health informatics environment to support evidence-based medicine. IEEE Trans Biomed Eng. 2015;62:2776–2786.
13. 3M Health Information Systems. 3M Grouper Plus System fact sheet. http://multimedia.3m.com/mws/media/217377O/3m-gps-fact-sheet.pdf. Accessed September 29, 2017.
14. Ho D, Imai K, King G, Stuart E. MatchIt: Nonparametric preprocessing for parametric causal inference. J Stat Softw. 2011;42(8)
15. Chin DL, Wilson MH, Bang H, Romano PS. Comparing patient outcomes of academician-preceptors, hospitalist-preceptors, and hospitalists on internal medicine services in an academic medical center. J Gen Intern Med. 2014;29:1672–1678.
16. Jordan MR, Conley J, Ghali WA. Consultation patterns and clinical correlates of consultation in a tertiary care setting. BMC Res Notes. 2008;1:96.
17. Patel MS, Reed DA, Loertscher L, McDonald FS, Arora VM. Teaching residents to provide cost-conscious care: A national survey of residency program directors. JAMA Intern Med. 2014;174:470–472.
18. Carlson J, Dachs RJ. Family medicine residents remain unaware of hospital charges for diagnostic testing. Fam Med. 2015;47:466–469.
19. Chen C, Petterson S, Phillips R, Bazemore A, Mullan F. Spending patterns in region of residency training and subsequent expenditures for care provided by practicing physicians for Medicare beneficiaries. JAMA. 2014;312:2385–2393.
20. Sirovich BE, Lipner RS, Johnston M, Holmboe ES. The association between residency training and internists’ ability to practice conservatively. JAMA Intern Med. 2014;174:1640–1648.
21. Patel MS, Reed DA, Smith C, Arora VM. Role-modeling cost-conscious care—A national evaluation of perceptions of faculty at teaching hospitals in the United States. J Gen Intern Med. 2015;30:1294–1298.
22. Elliott DJ, Young RS, Brice J, Aguiar R, Kolm P. Effect of hospitalist workload on the quality and efficiency of care. JAMA Intern Med. 2014;174:786–793.
Back to Top | Article Outline

References cited in Table 1 only

23. Centers for Medicare and Medicaid Services. Appendix A: List of MS-DRGs version 28. Draft ICD-10-CM/PCS MS-DRGv28 definitions manual. https://www.cms.gov/icd10manual/fullcode_cms/P0029.html. Accessed October 6, 2017.
    24. Centers for Medicare and Medicaid Services. Appendix C: Complications or comorbidities exclusion list. Draft ICD-10-CM/PCS MS-DRGv28 definitions manual. https://www.cms.gov/icd10manual/fullcode_cms/p0031.html. Accessed October 6, 2017.
      © 2018 by the Association of American Medical Colleges