Association of Variation in Consultant Use Among Hospitalist Physicians With Outcomes Among Medicare Beneficiaries | Health Care Workforce | JAMA Network Open | JAMA Network
[Skip to Navigation]
Sign In
Figure.  Within-Hospital Distribution of Residual Consultations per Admission
Within-Hospital Distribution of Residual Consultations per Admission

Each line spans from a hospital's 25th to 75th percentile in the distribution of residual consults per admission (across hospitalists). The residuals come from the model fit to 2013 admissions, adjusted for diagnosis related group. The solid points mark the hospital medians. The national mean of 1.1 consultations per admission (horizontal dotted line) was added to each value to aid interpretation of the scale.

Table 1.  Patient and Hospital Characteristics Between High-Consulting Hospitalists and Others With Absolute Differences, 2014
Patient and Hospital Characteristics Between High-Consulting Hospitalists and Others With Absolute Differences, 2014
Table 2.  Unadjusted Outcomes of Admission to High-Consulting Hospitalists vs Other Hospitalists, 2014
Unadjusted Outcomes of Admission to High-Consulting Hospitalists vs Other Hospitalists, 2014
Table 3.  Results From Regression Models Comparing Adjusted Outcomes of Admissions to High-Consulting Hospitalists vs Other Hospitalists, 2014
Results From Regression Models Comparing Adjusted Outcomes of Admissions to High-Consulting Hospitalists vs Other Hospitalists, 2014
1.
Mafi  JN, Russell  K, Bortz  BA, Dachary  M, Hazel  WA  Jr, Fendrick  AM.  Low-cost, high-volume health services contribute the most to unnecessary health spending.  Health Aff (Millwood). 2017;36(10):1701-1704. doi:10.1377/hlthaff.2017.0385PubMedGoogle ScholarCrossref
2.
Fendrick  AM, Smith  DG, Chernew  ME.  Applying value-based insurance design to low-value health services.  Health Aff (Millwood). 2010;29(11):2017-2021. doi:10.1377/hlthaff.2010.0878PubMedGoogle ScholarCrossref
3.
Schwartz  AL, Landon  BE, Elshaug  AG, Chernew  ME, McWilliams  JM.  Measuring low-value care in Medicare.  JAMA Intern Med. 2014;174(7):1067-1076. doi:10.1001/jamainternmed.2014.1541PubMedGoogle ScholarCrossref
4.
Beaudin-Seiler  B, Ciarametaro  M, Dubois  RW, Lee  J, Fendrick  AM. Reducing low-value care. Health Affairs. https://www.healthaffairs.org/do/10.1377/hblog20160920.056666/full/. Published September 20, 2016. Accessed March 27, 2019.
5.
Centers for Medicare & Medicaid. National health expenditures 2017 highlights. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/downloads/highlights.pdf. Published 2017. Accessed March 15, 2019.
6.
Kacik  A. Not-for-profit hospital operating margins continue to contract. Modern Healthcare. https://www.modernhealthcare.com/article/20180920/NEWS/180929982/not-for-profit-hospital-operating-margins-continue-to-contract. Published September 20, 2018. Accessed March 15, 2019.
7.
Badgery-Parker  T, Pearson  SA, Dunn  S, Elshaug  AG.  Measuring hospital-acquired complications associated with low-value care.  JAMA Intern Med. 2019;179(4):499-505. doi:10.1001/jamainternmed.2018.7464PubMedGoogle ScholarCrossref
8.
Ward  MJ, Eckman  MH, Schauer  DP, Raja  AS, Collins  S.  Cost-effectiveness of telemetry for hospitalized patients with low-risk chest pain.  Acad Emerg Med. 2011;18(3):279-286. doi:10.1111/j.1553-2712.2011.01008.xPubMedGoogle ScholarCrossref
9.
Durairaj  L, Reilly  B, Das  K,  et al.  Emergency department admissions to inpatient cardiac telemetry beds: a prospective cohort study of risk stratification and outcomes.  Am J Med. 2001;110(1):7-11. doi:10.1016/S0002-9343(00)00640-9PubMedGoogle ScholarCrossref
10.
Wang  TJ, Mort  EA, Nordberg  P,  et al.  A utilization management intervention to reduce unnecessary testing in the coronary care unit.  Arch Intern Med. 2002;162(16):1885-1890. doi:10.1001/archinte.162.16.1885PubMedGoogle ScholarCrossref
11.
Graat  ME, Stoker  J, Vroom  MB, Schultz  MJ.  Can we abandon daily routine chest radiography in intensive care patients?  J Intensive Care Med. 2005;20(4):238-246. doi:10.1177/0885066605277212PubMedGoogle ScholarCrossref
12.
Eaton  KP, Levy  K, Soong  C,  et al.  Evidence-based guidelines to eliminate repetitive laboratory testing.  JAMA Intern Med. 2017;177(12):1833-1839. doi:10.1001/jamainternmed.2017.5152PubMedGoogle ScholarCrossref
13.
Price  MB, Grant  MJ, Welkie  K.  Financial impact of elimination of routine chest radiographs in a pediatric intensive care unit.  Crit Care Med. 1999;27(8):1588-1593. doi:10.1097/00003246-199908000-00033PubMedGoogle ScholarCrossref
14.
Stevens  JP, Nyweide  DJ, Maresh  S, Hatfield  LA, Howell  MD, Landon  BE.  Comparison of hospital resource use and outcomes among hospitalists, primary care physicians, and other generalists.  JAMA Intern Med. 2017;177(12):1781-1787. doi:10.1001/jamainternmed.2017.5824PubMedGoogle ScholarCrossref
15.
Stevens  JP, Nyweide  D, Maresh  S,  et al.  Variation in inpatient consultation among older adults in the United States.  J Gen Intern Med. 2015;30(7):992-999. doi:10.1007/s11606-015-3216-7PubMedGoogle ScholarCrossref
16.
Honda  H, Krauss  MJ, Jones  JC, Olsen  MA, Warren  DK.  The value of infectious diseases consultation in Staphylococcus aureus bacteremia.  Am J Med. 2010;123(7):631-637. doi:10.1016/j.amjmed.2010.01.015PubMedGoogle ScholarCrossref
17.
Ciemins  EL, Blum  L, Nunley  M, Lasher  A, Newman  JM.  The economic and clinical impact of an inpatient palliative care consultation service: a multifaceted approach.  J Palliat Med. 2007;10(6):1347-1355. doi:10.1089/jpm.2007.0065PubMedGoogle ScholarCrossref
18.
Classen  DC, Burke  JP, Wenzel  RP.  Infectious diseases consultation: impact on outcomes for hospitalized patients and results of a preliminary study.  Clin Infect Dis. 1997;24(3):468-470. doi:10.1093/clinids/24.3.468PubMedGoogle ScholarCrossref
19.
Ahmed  A, Allman  RM, Kiefe  CI,  et al.  Association of consultation between generalists and cardiologists with quality and outcomes of heart failure care.  Am Heart J. 2003;145(6):1086-1093. doi:10.1016/S0002-8703(02)94778-2PubMedGoogle ScholarCrossref
20.
Kuo  YF, Sharma  G, Freeman  JL, Goodwin  JS.  Growth in the care of older patients by hospitalists in the United States.  N Engl J Med. 2009;360(11):1102-1112. doi:10.1056/NEJMsa0802381PubMedGoogle ScholarCrossref
21.
Research Data Assistance Center. Claim attending physician NPI number (FFS). https://www.resdac.org/cms-data/variables/claim-attending-physician-npi-number-ffs. Accessed December 20, 2019.
22.
American Hospital Association Linkage Files. https://www.hcup-us.ahrq.gov/db/state/ahalinkage/aha_linkage.jsp. Updated November 2019. Accessed December 20, 2019.
23.
Gade  G, Venohr  I, Conner  D,  et al.  Impact of an inpatient palliative care team: a randomized control trial.  J Palliat Med. 2008;11(2):180-190. doi:10.1089/jpm.2007.0055PubMedGoogle ScholarCrossref
24.
Cohen  HJ, Feussner  JR, Weinberger  M,  et al.  A controlled trial of inpatient and outpatient geriatric evaluation and management.  N Engl J Med. 2002;346(12):905-912. doi:10.1056/NEJMsa010285PubMedGoogle ScholarCrossref
25.
Kunishima  H, Yamamoto  N, Kobayashi  T,  et al.  Methicillin resistant Staphylococcus aureus in a Japanese community hospital: 5-year experience.  J Infect Chemother. 2010;16(6):414-417. doi:10.1007/s10156-010-0076-2PubMedGoogle ScholarCrossref
26.
Lesprit  P, Landelle  C, Brun-Buisson  C.  Clinical impact of unsolicited post-prescription antibiotic review in surgical and medical wards: a randomized controlled trial.  Clin Microbiol Infect. 2013;19(2):E91-E97. doi:10.1111/1469-0691.12062PubMedGoogle ScholarCrossref
27.
Bini  EJ, Weinshel  EH, Generoso  R,  et al.  Impact of gastroenterology consultation on the outcomes of patients admitted to the hospital with decompensated cirrhosis.  Hepatology. 2001;34(6):1089-1095. doi:10.1053/jhep.2001.29204PubMedGoogle ScholarCrossref
28.
Sellier  E, Pavese  P, Gennai  S, Stahl  JP, Labarère  J, François  P.  Factors and outcomes associated with physicians’ adherence to recommendations of infectious disease consultations for inpatients.  J Antimicrob Chemother. 2010;65(1):156-162. doi:10.1093/jac/dkp406PubMedGoogle ScholarCrossref
29.
Jena  AB, Prasad  V, Goldman  DP, Romley  J.  Mortality and treatment patterns among patients hospitalized with acute cardiovascular conditions during dates of national cardiology meetings.  JAMA Intern Med. 2015;175(2):237-244. doi:10.1001/jamainternmed.2014.6781PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    Health Policy
    February 21, 2020

    Association of Variation in Consultant Use Among Hospitalist Physicians With Outcomes Among Medicare Beneficiaries

    Author Affiliations
    • 1Center for Healthcare Delivery Science, Beth Israel Deaconess Medical Center, Boston, Massachusetts
    • 2Division for Pulmonary, Critical Care, and Sleep Medicine, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, Massachusetts
    • 3Department of Health Care Policy, Harvard Medical School, Boston, Massachusetts
    • 4Center for Medicare & Medicaid Innovation, Centers for Medicare & Medicaid Services, Baltimore, Maryland
    • 5Division of General Medicine, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, Massachusetts
    JAMA Netw Open. 2020;3(2):e1921750. doi:10.1001/jamanetworkopen.2019.21750
    Key Points español 中文 (chinese)

    Question  Are hospitalized Medicare beneficiaries who are exposed to higher rates of specialty consultation associated with outcomes such as greater resource use, length of stay, readmissions, and mortality?

    Findings  In this cohort study of 711 654 inpatient medical admissions, hospitalists who used specialty consultation more than their colleagues at the same institution used more resources without a difference in patient mortality. Compared with patients treated by other hospitalists, the patients of high-consulting hospitalists had longer lengths of stay, were less likely to go home, and were more likely to see a specialist within 90 days after discharge, but there was no significant difference in their mortality at 30 days or their likelihood of all-cause readmission.

    Meaning  A decrease in the frequency of specialty consultation may be an opportunity for hospitals to reduce complexity and costs in patient care without adversely affecting patients.

    Abstract

    Importance  Evidence is lacking on the consequences of high rates of inpatient consultation.

    Objective  To examine outcomes and resource use of patients cared for by hospitalists who use more inpatient consultation than their colleagues.

    Design, Setting, and Participants  A retrospective cohort study of medical admissions to hospitalists among fee-for-service Medicare beneficiaries was conducted. Hospitalist consultation tendency was identified from January 1, 2013, to December 31, 2014; admissions were calculated in 2013; and outcomes were measured in 2014. Data were analyzed from January 31, 2017, to May 9, 2019. A total of 711 654 admissions with patients receiving care from 14 584 hospitalists at 737 hospitals were included.

    Exposure  Admission to high-consulting hospitalists, considered to be those who were in the top 25% of the distribution of consulting frequency at their own hospital (adjusted for patient case mix).

    Main Outcomes and Measures  Outcomes included length of stay, Medicare Part B inpatient charges, discharge destination, all-cause 7- and 30-day readmissions, 90-day outpatient specialist visits, and 30-day mortality.

    Results  The 711 654 hospital admissions included 408 489 women (57.4%); mean (SD) age of the population was 80 (8.5) years. Length of stay of patients cared for by high-consulting hospitalists was longer compared with other hospitalists (adjusted incidence rate ratio, 1.04; 95% CI, 1.03-1.05). The admissions resulted in a mean of $137.91 (95% CI, $118.89-$156.93) more in Medicare Part B charges and were less likely to end with the patient going home (adjusted odds ratio [aOR], 0.96; 95% CI, 0.94-0.98) compared with patients cared for by other hospitalists in the cohort. Patients cared for by high-consulting hospitalists also were 7% more likely than patients cared for by other hospitalists to see an outpatient specialist at 90 days (aOR 1.07; 95% CI, 1.05-1.09), with no significant differences in 30-day mortality (aOR 1.01, 95% CI, 0.98-1.03) or readmissions (7-day readmissions: aOR 1.01; 95% CI, 0.98-1.03; 30-day readmissions: aOR, 1.01; 95% CI, 0.99-1.03).

    Conclusions and Relevance  Hospitalists who obtain consultations more than their colleagues at the same institution were associated with greater use of health care resources without apparent mortality benefit. Further investigation should identify whether reducing high rates of consultation can reduce resource use without harming patients.

    Introduction

    As US health care spending continues to increase more quickly than inflation, attention has focused on reducing delivery of health care services that do not benefit patients.1-4 However, because most hospitals in the United States are paid via a fixed diagnosis-adjusted reimbursement per admission, little research has assessed low-value care during hospitalizations. Hospital care represents the largest share of health care spending, despite many hospitals operating with negative or low margins.5,6 Hospitals may need to provide more efficient care to remain financially viable.7 Many hospital services are used without significant benefit to patients, including excessive inpatient telemetry for patients with low risk of acute coronary syndrome,8,9 daily chest radiographs in the intensive care unit,10,11 and routine laboratory testing for clinically stable patients.12 Yet, curtailing use of such services may yield only incremental savings.13

    One opportunity to curb excessive use of services that may add to costs without providing significant benefit to patients, considered low-value care for patients, is the use of specialty care for hospitalized patients. Inpatient specialty consultation, that is, engaging a specialist during the care of a hospitalized patient, is prevalent: Medicare beneficiaries average 1 to 3 consultations per admission.14,15 The substantial variation in physicians’ use of consultation, however, is not fully explained by hospital characteristics or patient need.15 While specialty consultation brings additional clinical expertise and access to procedures,16-19 the use of consultations may increase resource use and have negative clinical consequences. For example, specialist involvement can increase length of stay to accommodate evaluation and testing by the new physician team, and testing or procedures that ultimately are not useful also carry risks for overdiagnosis or complications. Specialists may reflexively recommend ongoing outpatient care after discharge, potentially increasing costs, testing, and travel for patients. Specialty consultation also may lead to low-value radiologic or laboratory studies or risky additional procedures both in the hospital and after discharge. Thus, reducing consultation use beyond clinical need might be an important target for hospitals seeking to become more efficient and reduce low-value care for patients.

    Studying the benefits and risks of consultation is fraught with confounding. Patients with many specialty consults are generally sicker, and the marginal consequences of consultation may be outweighed by these patient differences. Moreover, practice styles and specialty care use vary widely within and between hospitals.15 We sought to identify variation in specialty consultation that was independent of patients’ medical needs, health outcomes, and hospital differences. We examined the nearly random assignment of hospitalized patients to an attending hospitalist as a source of plausibly exogenous variation in patients’ exposure to inpatient consultation. We characterized hospitalists relative to their colleagues in the same hospital and compared the outcomes of admissions to high consulting hospitalists with the outcomes of the remaining hospitalists. We hypothesized that admissions of patients receiving care from high-consulting hospitalists would be associated with greater resource use without appreciable differences in patient outcomes.

    Methods
    Study Population and Data Source

    We used administrative data from Medicare to analyze admissions of fee-for-service Medicare beneficiaries during 2013 and 2014. Using the Medicare Beneficiary Summary File, we included beneficiaries aged 66 years and older during the year of admission who were continuously enrolled in Parts A and B without end-stage renal disease. We excluded those enrolled in Medicare Advantage at any point during the year. We limited our sample to hospitalizations with hospitalist attendings for the 25 most prevalent medical diagnosis related groups (top 25 medical DRGs) (eTable 1 in the Supplement) to capture the most common medical admissions. We restricted institutions to nonfederal (ie, excluding critical access, Indian Health Service, and Veteran Affairs hospitals) acute care hospitals located in the 50 states and the District of Columbia that were large enough to measure variability in consultation behavior across hospitalists (≥250 beds and ≥8 hospitalists with ≥10 admissions). Medical admissions from January 1, 2013, to December 31, 2014, were included. Hospitalist consultation tendency was identified using admissions in 2013, and outcomes were measured in 2014. Data analysis was performed from January 31, 2017, to May 9, 2019. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. Our study was reviewed and deemed exempt with waiver of informed consent as a study conducted using existing publicly available data by the institutional review board at the Beth Israel Deaconess Medical Center, Boston, Massachusetts.

    Identifying Hospitalists and Classifying Hospitalist Consultation Tendency

    We defined hospitalists as generalist physicians with at least 100 claims in 2013, of which at least 90% were for inpatient care.20 Each top 25 medical DRG admission was assigned a single attending hospitalist. Each hospitalist was assigned to the hospital at which the plurality of their 2013 and 2014 admissions were billed. We restricted participation to hospitalists with at least 10 admissions at their assigned hospital in 2013.

    We estimated hospitalist consultation tendency using admissions in 2013. Admissions were assigned to hospitalists based on the attending physician assigned in the Part A claim.21 We computed the number of consultations per admission as the number of unique medical specialties billed per admission, minus 1 to account for the hospitalist. We fit a negative binomial regression model for the number of consultations per admission that included fixed effects for DRGs to account for patient case mix. The residuals from this model represent the relative number of consultations for each admission compared with the mean for admissions with the same DRG.

    We determined the mean of these residuals for each hospital and each hospitalist. Then, for each hospitalist, we subtracted their hospital’s mean residual from their own mean residual to yield a relative measure of the hospitalist’s tendency to use consultation compared with other hospitalists at the same institution. We defined high-consulting hospitalists as those in the top 25% of the distribution. As a sensitivity check, we also compared hospitalist characterizations based on 2013 admissions with how those same hospitalists would have been characterized based on 2014 admissions.

    Because we were interested in the patient outcomes associated with exposure to high consultation use, we sought to capitalize on the pseudorandom assignment of patients to hospitalists by admission date and the relative stability of hospitalists’ consultation tendencies over time. We present the between-hospital variation in consultation tendency as well, as most of the variation occurs between hospitals rather than within hospitals; however, the large degree of between-hospital variation may also be associated with variation in hospital resources, specialty availability, and region of the country.

    Study Variables

    For length of stay and cost, all patient outcomes were measured during admissions in 2014 to further eliminate confounding by indication that may be associated with high consultation use. We defined length of stay as the date of discharge minus the date of admission plus 1 such that each admission had a length of stay of at least 1 day. For cost, we summed Part B–allowed costs between the admission and discharge dates of the hospitalization to quantify the additional costs to the Medicare program beyond the fixed cost per DRG of the hospitalization.

    Using the discharge destination variable in the inpatient file, we dichotomized discharge into the patient’s home vs all other destinations and restricted to patients alive through discharge. We defined outpatient specialist use as any outpatient physician claim from a medical specialist within 90 days of discharge. We measured this outcome only on patients who were alive through 90 days after discharge. We restricted this measure to a patient’s first hospitalization of the year and to those who were alive through discharge to calculate all-cause readmissions at 7 and 30 days after discharge. We identified all-cause mortality at 30 days after admission using the Master Beneficiary Summary File, again restricting to a patient’s first hospitalization.

    Patient and Hospital Control Variables

    Beneficiary characteristics included age (in 5-year increments), race and ethnicity (defined from the Chronic Conditions Warehouse race variable as black, Hispanic, non-Hispanic white, Asian, and other), female sex, dual eligibility for Medicare and Medicaid, Medicare coverage through disability, chronic conditions defined by Chronic Conditions Warehouse chronic condition ever flags, severity of the DRG (defined based on whether the DRG code included complications and comorbidities using the admission DRG severity index), whether the admission occurred on a weekend, and, for some outcomes, inpatient mortality. Hospital characteristics from the 2013 American Hospital Association survey included teaching status (major teaching [Council of Teaching Hospitals members], minor teaching [any hospital with residents or medical students], and nonteaching), rural location, US Census region (Northeast, South, Midwest, and West), percentage of patients admitted to the intensive care unit, and mean hospital hierarchal condition categories score.22 The last 2 measures capture hospital-level resource use and case mix complexity. Scores are determined by the percentage of intensive care unit admissions from 2013 MedPAR files and hospital-level means of merged hierarchal condition categories scores for each admitted beneficiary.

    Statistical Analysis

    We compared the unadjusted outcomes of admissions to high-consulting hospitalists vs the remaining admissions using χ2 tests for categorical outcomes and 2-tailed, paired t tests or Wilcoxon rank sum tests for continuous outcomes. We fit generalized linear models with distributions and link functions appropriate to each outcome. Specifically, for binary outcomes (readmission, discharge home, mortality, and outpatient specialty care), we used binomial distributions with logit links; for the count outcome of length of stay, we used a negative binomial distribution with a log link; and for the continuous outcome of cost, we used a gaussian distribution with an identity link. In these models, we clustered residuals at the hospital level and adjusted for all available beneficiary and hospital characteristics. P < .05 was considered statistically significant. All analyses were performed using SAS Enterprise Guide, version 7.15 (SAS Institute Inc).

    Results

    We analyzed 711 654 admissions between January 1, 2014, and December 31, 2014, with patients cared for by 14 584 hospitalists at 737 hospitals. Patient and hospital characteristics are described in Table 1; in summary, mean (SD) patient age was 80 (8.5) years and 408 489 were women (57.4%). Hospitalists obtained a mean (SD) of 1.2 (1.2) consultations per admission. High-consulting hospitalists used 1.4 (1.3) consults per admission compared with 1.1 (1.2) consults per admission for remaining hospitalists.

    Consultation tendency was relatively stable over time between 2013 and 2014; approximately 7% of previously low-consulting hospitalists in 2013 became top-quartile consulting hospitalists in 2014, and 7% of high-consulting hospitalists in 2013 became low-consulting hospitalists in 2014. Hospitals varied in their mean use of inpatient consultations on the hospitalist services (Figure).

    Characteristics of 2014 admissions to high-consulting hospitalists vs others are presented in Table 1. Patient demographic and admission characteristics for admissions to high-consulting hospitalists vs others were similar; however, high-consulting hospitalists cared for more nonwhite patients (16.9% vs 15.9%), more Medicaid dual-eligible patients (26.4% vs 25.6%), and more patients in the highest severity DRG (31.0% vs 30.3%). There was evidence of differential allocation of admissions to high-consulting hospitalists at some hospitals; although high-consulting hospitalists compose 25% of hospitalists in each hospital (by definition), they may care for more than 25% of admissions.

    Patients cared for by high-consulting hospitalists had 4% longer lengths of stay (unadjusted mean: 6.1 vs 5.8 days; P < .001; adjusted incidence rate ratio, 1.04; 95% CI, 1.03-1.05) and more than 11% higher part B costs relative to patients cared for by their colleagues at the same institution (unadjusted mean: $1427 vs $1265; P < .001; adjusted difference: $137.91; 95% CI, $118.89-$156.93). Patients were 4% less likely to go home (unadjusted risk of going home: 41.5% vs 42.9%; P < .001; adjusted odds ratio [aOR], 0.96; 95% CI, 0.94-0.98) and 7% more likely to see a specialist at 90 days (unadjusted risk of specialist visit at 90 days: 66.5% vs 64.9%; P < .001; aOR, 1.07; 95% CI, 1.05-1.09) if they were cared for by high-consulting hospitalists, with no significant difference in readmission rates at 7 days (aOR, 1.01; 95% CI, 0.98-1.03) or 30 days (aOR, 1.01; 95% CI, 0.99-1.03). In addition, there was no significant difference in 30-day mortality for patients cared for by high-consulting hospitalists (unadjusted mean: 10.3% vs 10.3%; P = .62; aOR, 1.01; 95% CI, 0.98-1.03). Unadjusted analyses are presented in Table 2; adjusted analyses are presented in Table 3. The full models are available in eTables 2-8 in the Supplement.

    Discussion

    Our results suggest an association between high-consulting hospitalists and the use of more resources without clear benefit to patients, which appears to be true even after controlling for consultation tendencies at the hospital level. Patients who were cared for by high-consulting hospitalists stayed longer in the hospital, were less likely to go home, and were more likely to see an outpatient specialist within 90 days of discharge but showed no evidence of differences in readmission rates compared with patients cared for by other hospitalists. There was no significant difference in survival at 30 days for patients between groups after controlling for mean consultation use at the hospital level.

    Prior work examining the patient-level consequences of inpatient specialty use has focused solely on the expertise provided by specific specialties.23-27 For instance, Sellier et al28 noted that adherence to infectious disease consultants’ recommendations was associated with lower inpatient mortality and earlier clinical improvement. However, findings by Jena and colleagues29 raised the concern that patients may not always routinely benefit from specialty consultation; they found that patients admitted with cardiac diagnoses did comparably better during times of major cardiology conferences when only junior cardiologists were available. Although we did not find harm to patients in the form of increased mortality or readmissions with exposure to more consultations, we found increases in resource use, including both costs incurred during the hospitalization and increased subsequent use of outpatient specialty care.

    Our approach did not allow us to distinguish the exact point at which consultations are no longer useful, and this threshold likely is not uniform across different diagnoses and specialties consulted or across different hospitals. Thus, although our results do not highlight a group of patients actively harmed by potentially high rates of inpatient consultation, we have identified a group of physicians who apparently provide relatively more resource-intensive care, which could be a target for education or other interventions by hospitals. Thus, a decrease in frequent consultation, particularly for diagnoses or questions in which there is likely to be little change in management, may be an opportunity for hospitals to reduce complexity and costs in patient care without adversely affecting patients. It is likely that there may be underuse of consultation by some hospitalists, but our research suggests that at least some portion of consultations represent overuse. Further study distinguishing by type of consultation (eg, procedural consultations, specific specialty consultations) may be helpful in clarifying the heterogeneity of associations with various types of inpatient consultation.

    Our study design used the pseudorandom assignment of patients to hospitalists by admission date to identify the benefit or harm from frequent use of specialty consultation. However, we found that some observable hospital-level variables, such as teaching status and geographic location, were not completely balanced. We attribute this imbalance to differential shares of admissions going to high-consulting hospitalists in some hospitals (eg, Northeast hospitals and teaching hospitals). High-consulting hospitalists may use specialty care at their institutions potentially to fill clinical knowledge gaps or, alternatively, compensate for excessive patient caseload and offload clinical work. The lack of balance of some hospital-level variables in our treatment assignment may lend further credence to our hypothesis that the highest-consulting hospitalists at times may be using consultation to accommodate heavy workloads.

    Limitations

    Our study has several limitations. First, we used administrative data, which lack the clinical nuance of medical records. More specifically, we did not explore other patient outcomes beyond mortality, such as patient satisfaction. Second, as with any observational study, our findings may be biased by unmeasured confounders, such as patterns in how certain admissions are assigned to hospitalists, and so cannot claim that hospitalist consultation levels cause certain outcomes. However, we note the relative stability of our findings between the unadjusted and adjusted models and that any unmeasured confounder in the adjusted models would have to be uncorrelated with the patient- and hospital-level variables included, adding further confidence to our results. Third, we sought to study the patient-level outcome of high-consulting hospitalists, which fails to indicate the benefit or harms associated with individual consultations. Fourth, we did not explore additional features of the hospitalists that might be associated with specialty use; these may represent unmeasured confounders with some of the outcomes (eg, length of stay), although less likely with others (eg, outpatient specialty use at 90 days). Fifth, we assigned admissions to hospitalists based on the Part A designation of the attending physician, which may fail to accurately attribute the consultation tendency or patient outcomes to the correct physician. However, the Part A attending is defined as the physician who has “overall responsibility for the beneficiary’s care”21 and is therefore likely to be most responsible for clinical decision-making. Furthermore, any misclassification would likely bias our findings toward the null. Sixth, we did not account from where a patient was admitted (which is not accurately recorded in administrative data), which may influence whether a patient could be discharged home. Seventh, we necessarily limited the sample to hospitals with a high volume of admissions and multiple hospitalists to measure the variability in consultation tendency for the most common medical reasons for hospitalization. Therefore, our findings may not be generalizable to smaller hospitals.

    Conclusions

    This study found that care provided by high-consulting hospitalists may be associated with increased resource use without clear benefit to patients. Our findings suggest that consultation may be one area in which hospitals or health systems seeking to decrease costs, such as under risk-based models of care, could focus their attention. Alternative strategies to minimize overuse of consultation may have the potential to yield additional savings to hospitals and added benefits to patients, particularly among elderly patients for whom attending outpatient visits may be difficult or returning home may be a priority.

    Back to top
    Article Information

    Accepted for Publication: December 26, 2018.

    Published: February 21, 2020. doi:10.1001/jamanetworkopen.2019.21750

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Stevens JP et al. JAMA Network Open.

    Corresponding Author: Jennifer P. Stevens, MD, MS, Center for Healthcare Delivery Science, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Boston, MA 02215 (jpsteven@bidmc.harvard.edu).

    Author Contributions: Dr Nyweide 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.

    Concept and design: All authors.

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

    Drafting of the manuscript: Stevens, Hatfield.

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

    Statistical analysis: Stevens, Hatfield, Nyweide.

    Obtained funding: Stevens.

    Administrative, technical, or material support: Stevens, Landon.

    Supervision: Hatfield, Landon.

    Conflict of Interest Disclosures: Dr Stevens reported receiving grants from the Agency for Healthcare Research and Quality and the Doris Duke Charitable Foundation during the conduct of the study; and receiving royalties from UpToDate as well as for the textbook Understanding Healthcare Delivery Science from McGraw Hill Medical. Dr Hatfield reported receiving personal fees from Beth Israel Deaconess Medical Center during the conduct of the study; personal fees from Cambridge Health Alliance, Mathematica Policy Research, Greylock McKinnon Associates, and RAND outside the submitted work; and grants and personal fees from the Laura and John Arnold Foundation. No other disclosures were reported.

    Funding/Support: This project was supported by grant K08HS024288 from the Agency for Healthcare Research and Quality (Dr Stevens).

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

    Disclaimer: The content is solely the responsibility of the authors and does not represent the official views of the Agency for Healthcare Research and Quality or the Centers for Medicare & Medicaid Services.

    References
    1.
    Mafi  JN, Russell  K, Bortz  BA, Dachary  M, Hazel  WA  Jr, Fendrick  AM.  Low-cost, high-volume health services contribute the most to unnecessary health spending.  Health Aff (Millwood). 2017;36(10):1701-1704. doi:10.1377/hlthaff.2017.0385PubMedGoogle ScholarCrossref
    2.
    Fendrick  AM, Smith  DG, Chernew  ME.  Applying value-based insurance design to low-value health services.  Health Aff (Millwood). 2010;29(11):2017-2021. doi:10.1377/hlthaff.2010.0878PubMedGoogle ScholarCrossref
    3.
    Schwartz  AL, Landon  BE, Elshaug  AG, Chernew  ME, McWilliams  JM.  Measuring low-value care in Medicare.  JAMA Intern Med. 2014;174(7):1067-1076. doi:10.1001/jamainternmed.2014.1541PubMedGoogle ScholarCrossref
    4.
    Beaudin-Seiler  B, Ciarametaro  M, Dubois  RW, Lee  J, Fendrick  AM. Reducing low-value care. Health Affairs. https://www.healthaffairs.org/do/10.1377/hblog20160920.056666/full/. Published September 20, 2016. Accessed March 27, 2019.
    5.
    Centers for Medicare & Medicaid. National health expenditures 2017 highlights. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/downloads/highlights.pdf. Published 2017. Accessed March 15, 2019.
    6.
    Kacik  A. Not-for-profit hospital operating margins continue to contract. Modern Healthcare. https://www.modernhealthcare.com/article/20180920/NEWS/180929982/not-for-profit-hospital-operating-margins-continue-to-contract. Published September 20, 2018. Accessed March 15, 2019.
    7.
    Badgery-Parker  T, Pearson  SA, Dunn  S, Elshaug  AG.  Measuring hospital-acquired complications associated with low-value care.  JAMA Intern Med. 2019;179(4):499-505. doi:10.1001/jamainternmed.2018.7464PubMedGoogle ScholarCrossref
    8.
    Ward  MJ, Eckman  MH, Schauer  DP, Raja  AS, Collins  S.  Cost-effectiveness of telemetry for hospitalized patients with low-risk chest pain.  Acad Emerg Med. 2011;18(3):279-286. doi:10.1111/j.1553-2712.2011.01008.xPubMedGoogle ScholarCrossref
    9.
    Durairaj  L, Reilly  B, Das  K,  et al.  Emergency department admissions to inpatient cardiac telemetry beds: a prospective cohort study of risk stratification and outcomes.  Am J Med. 2001;110(1):7-11. doi:10.1016/S0002-9343(00)00640-9PubMedGoogle ScholarCrossref
    10.
    Wang  TJ, Mort  EA, Nordberg  P,  et al.  A utilization management intervention to reduce unnecessary testing in the coronary care unit.  Arch Intern Med. 2002;162(16):1885-1890. doi:10.1001/archinte.162.16.1885PubMedGoogle ScholarCrossref
    11.
    Graat  ME, Stoker  J, Vroom  MB, Schultz  MJ.  Can we abandon daily routine chest radiography in intensive care patients?  J Intensive Care Med. 2005;20(4):238-246. doi:10.1177/0885066605277212PubMedGoogle ScholarCrossref
    12.
    Eaton  KP, Levy  K, Soong  C,  et al.  Evidence-based guidelines to eliminate repetitive laboratory testing.  JAMA Intern Med. 2017;177(12):1833-1839. doi:10.1001/jamainternmed.2017.5152PubMedGoogle ScholarCrossref
    13.
    Price  MB, Grant  MJ, Welkie  K.  Financial impact of elimination of routine chest radiographs in a pediatric intensive care unit.  Crit Care Med. 1999;27(8):1588-1593. doi:10.1097/00003246-199908000-00033PubMedGoogle ScholarCrossref
    14.
    Stevens  JP, Nyweide  DJ, Maresh  S, Hatfield  LA, Howell  MD, Landon  BE.  Comparison of hospital resource use and outcomes among hospitalists, primary care physicians, and other generalists.  JAMA Intern Med. 2017;177(12):1781-1787. doi:10.1001/jamainternmed.2017.5824PubMedGoogle ScholarCrossref
    15.
    Stevens  JP, Nyweide  D, Maresh  S,  et al.  Variation in inpatient consultation among older adults in the United States.  J Gen Intern Med. 2015;30(7):992-999. doi:10.1007/s11606-015-3216-7PubMedGoogle ScholarCrossref
    16.
    Honda  H, Krauss  MJ, Jones  JC, Olsen  MA, Warren  DK.  The value of infectious diseases consultation in Staphylococcus aureus bacteremia.  Am J Med. 2010;123(7):631-637. doi:10.1016/j.amjmed.2010.01.015PubMedGoogle ScholarCrossref
    17.
    Ciemins  EL, Blum  L, Nunley  M, Lasher  A, Newman  JM.  The economic and clinical impact of an inpatient palliative care consultation service: a multifaceted approach.  J Palliat Med. 2007;10(6):1347-1355. doi:10.1089/jpm.2007.0065PubMedGoogle ScholarCrossref
    18.
    Classen  DC, Burke  JP, Wenzel  RP.  Infectious diseases consultation: impact on outcomes for hospitalized patients and results of a preliminary study.  Clin Infect Dis. 1997;24(3):468-470. doi:10.1093/clinids/24.3.468PubMedGoogle ScholarCrossref
    19.
    Ahmed  A, Allman  RM, Kiefe  CI,  et al.  Association of consultation between generalists and cardiologists with quality and outcomes of heart failure care.  Am Heart J. 2003;145(6):1086-1093. doi:10.1016/S0002-8703(02)94778-2PubMedGoogle ScholarCrossref
    20.
    Kuo  YF, Sharma  G, Freeman  JL, Goodwin  JS.  Growth in the care of older patients by hospitalists in the United States.  N Engl J Med. 2009;360(11):1102-1112. doi:10.1056/NEJMsa0802381PubMedGoogle ScholarCrossref
    21.
    Research Data Assistance Center. Claim attending physician NPI number (FFS). https://www.resdac.org/cms-data/variables/claim-attending-physician-npi-number-ffs. Accessed December 20, 2019.
    22.
    American Hospital Association Linkage Files. https://www.hcup-us.ahrq.gov/db/state/ahalinkage/aha_linkage.jsp. Updated November 2019. Accessed December 20, 2019.
    23.
    Gade  G, Venohr  I, Conner  D,  et al.  Impact of an inpatient palliative care team: a randomized control trial.  J Palliat Med. 2008;11(2):180-190. doi:10.1089/jpm.2007.0055PubMedGoogle ScholarCrossref
    24.
    Cohen  HJ, Feussner  JR, Weinberger  M,  et al.  A controlled trial of inpatient and outpatient geriatric evaluation and management.  N Engl J Med. 2002;346(12):905-912. doi:10.1056/NEJMsa010285PubMedGoogle ScholarCrossref
    25.
    Kunishima  H, Yamamoto  N, Kobayashi  T,  et al.  Methicillin resistant Staphylococcus aureus in a Japanese community hospital: 5-year experience.  J Infect Chemother. 2010;16(6):414-417. doi:10.1007/s10156-010-0076-2PubMedGoogle ScholarCrossref
    26.
    Lesprit  P, Landelle  C, Brun-Buisson  C.  Clinical impact of unsolicited post-prescription antibiotic review in surgical and medical wards: a randomized controlled trial.  Clin Microbiol Infect. 2013;19(2):E91-E97. doi:10.1111/1469-0691.12062PubMedGoogle ScholarCrossref
    27.
    Bini  EJ, Weinshel  EH, Generoso  R,  et al.  Impact of gastroenterology consultation on the outcomes of patients admitted to the hospital with decompensated cirrhosis.  Hepatology. 2001;34(6):1089-1095. doi:10.1053/jhep.2001.29204PubMedGoogle ScholarCrossref
    28.
    Sellier  E, Pavese  P, Gennai  S, Stahl  JP, Labarère  J, François  P.  Factors and outcomes associated with physicians’ adherence to recommendations of infectious disease consultations for inpatients.  J Antimicrob Chemother. 2010;65(1):156-162. doi:10.1093/jac/dkp406PubMedGoogle ScholarCrossref
    29.
    Jena  AB, Prasad  V, Goldman  DP, Romley  J.  Mortality and treatment patterns among patients hospitalized with acute cardiovascular conditions during dates of national cardiology meetings.  JAMA Intern Med. 2015;175(2):237-244. doi:10.1001/jamainternmed.2014.6781PubMedGoogle ScholarCrossref
    ×