« Back to Results

Provider Decision-Making and Productivity in Health Care

Paper Session

Friday, Jan. 3, 2020 2:30 PM - 4:30 PM (PDT)

Marriott Marquis, Grand Ballroom 4
Hosted By: American Economic Association
  • Chair: David Chan Jr., Stanford University

Outcomes-Based Payments and Physician Productivity: Evidence from Diabetes Care in Hawaii

Benjamin Handel
,
University of California-Berkeley
Jonathan Kolstad
,
University of California-Berkeley
Michael Whinston
,
Massachusetts Institute of Technology

Abstract

Designing physician payments to incentivize low cost, high quality care is a core component of most health systems. The steady improvement in medical information technology over the past decade has directly led to (i) an increase in the ability of systems to more precisely link payments to health care and health outcomes of interest and (ii) deliver timely information to physicians about their performance on these care and outcomes metrics. We study physician productivity in response to quality-based incentives empirically using novel proprietary data from HMSA, the largest insurer in the state of Hawaii. Over the time period 2006-2015 HMSA implemented significant program changes in its pay-for-performance program including the money at stake and the types of measures being rewarded. We focus on payment incentives that HMSA provides for diabetic HbA1c control. Initially, from 2006-2010, the insurer provided very limited incentives for HbA1c control. From 2011-2013, HMSA provided substantial financial incentives to screen HbA1c while, for 2014-2015, they changed this program to only reward high quality HbA1c medical outcomes for patients with diabetes. We find that the shift to outcomes-based rewards (i) meaningfully increases prescriptions of more intensive interventions (insulin) for patients above the HbA1c quality threshold (ii) leads to additional visits to specialists (iii) leads to additional diagnosis of complications related to diabetes and (iv) improves HbA1c outcomes. We find that increased insulin prescribing is associated with (i) being part of a physician organization, as opposed to being a solo practitioner and (ii) older physicians. Finally, we show that high IT users have relatively higher baseline rates of insulin prescriptions among high A1c patients, while low IT users only alter their insulin prescription rates after the introduction of outcomes-based payments.

Administration Above Administrators: The Changing Technology of Healthcare Management

Abe Dunn
,
U.S. Bureau of Economic Analysis
Joshua Gottlieb
,
University of Chicago
Adam Shapiro
,
Federal Reserve Bank of San Francisco
Pietro Tebaldi
,
University of Chicago

Abstract

This paper measure the costs and types of administrative inputs in health care. We use data on labor and non-labor inputs by industry and categorize them as administrative or not. We find that non-labor inputs are a critical part of administrative spending, over and above labor inputs. Trends in non-labor administrative input spending have differed dramatically from that of labor input spending for hospitals over the last 20 years. Hospitals have substituted away from office workers, and towards externally purchased inputs. The share of managers and technical workers in administration has grown. The technology of healthcare administration is changing.

Triage Judgments in the Emergency Department

David Chan
,
Stanford University
Jonathan Gruber
,
Massachusetts Institute of Technology

Abstract

Efforts to use technology to aid human decisions have typically tried to emulate average human decisions (i.e., "crowdsourcing"), but to improve outcomes, algorithms need to emulate good human decisions. We adopt an approach combining quasi-experimental variation and machine learning to study this question in the setting of emergency department triage. We use rich data from the Veterans Health Administration (VHA) on approximately 11 million emergency visits in 130 VHA health care systems, including triage nurse identities and patient bed assignments, ESI scores, vital signs, comorbidities, and disposition and health outcomes. We show that patients are as good as randomly assigned to triage nurses based on the time of arrival to the emergency department, and we show that triage nurses have a causal effect on patient mortality. Among patients with predicted mortality greater than 0, one standard-deviation increase in triage nurse value added results in a 20% increase in patient mortality. We next characterize the mechanisms by which triages may have a mortality effect, along two key dimensions of triage behavior: First, triage nurses control the amount of time patients spend waiting to receive care. Second, triage nurses communicate the severity of patients' conditions by assigning an Emergency Severity Index (ESI) level to each patient. We find significant discretion and variation in both triage activities across triage nurses. We use machine learning and empirical Bayes methods to project high-dimensional variables describing these behaviors onto smaller-dimensional space. We find that these mechanisms together explain up to 80% of triage nurse effects on mortality. These results suggest that, relative to crowdsourcing, algorithms that link behaviors with outcomes could potentially capture a large amount of the effect of triage on outcomes.

Time Dependency in Physician Decision-Making

Lawrence Jin
,
National University of Singapore
Rui Tang
,
Princeton University
Han Ye
,
National University of Singapore
Junjian Yi
,
National University of Singapore
Songfa Zhong
,
National University of Singapore

Abstract

Using data with over 250,000 emergency department (ED) visits, we study the time-of-day effect on physician decision-making and patient outcomes. After controlling for patient characteristics, physician fixed effects and work hours, we find that cases treated at night have significantly lower probability of inpatient admission, and fewer medical tests. Moreover, these cases also exhibit a higher likelihood of a revisit to the ED in support of the decline of physician performance. While we cannot completely rule out the possibility of limited hospital arrangements, our findings are most consistent with the detrimental effect of disrupted circadian rhythm during night shifts.
Discussant(s)
Robert Gibbons
,
Massachusetts Institute of Technology
Mark Shepard
,
Harvard University
David Silver
,
Princeton University
Alice Chen
,
University of Southern California
JEL Classifications
  • I1 - Health
  • D2 - Production and Organizations