Hospital competition under pay‐for‐performance: Quality, mortality, and readmissions

• DOI: http://doi.org/10.1111/jems.12345
• Author: Odd Rune Straume,Luigi Siciliani,Domenico Lisi
• Published date: 01 April 2020
• Date: 01 April 2020

J Econ Manage Strat. 2020;29:289–314. wileyonlinelibrary.com/journal/jems © 2020 Wiley Periodicals, Inc.

|

289

Received: 13 November 2018

|

Revised: 24 February 2020

|

Accepted: 9 March 2020

DOI: 10.1111/jems.12345

ORIGINAL ARTICLE

Hospital competition under pay‐for‐performance:

Quality, mortality, and readmissions

Domenico Lisi

1

|Luigi Siciliani

2

|Odd Rune Straume

3,4

1

Department of Economics and Business,

University of Catania, Catania, Italy

2

Department of Economics and Related

Studies, University of York, Heslington,

York, UK

3

Department of Economics/NIPE,

University of Minho, Campus de Gualtar,

Braga, Portugal

4

Department of Economics, University of

Bergen, Bergen, Norway

Correspondence

Odd Rune Straume, Department of

Economics/NIPE, University of Minho,

Campus de Gualtar, 4710‐057 Braga,

Portugal; and Department of Economics,

University of Bergen, Bergen, Norway.

Email: o.r.straume@eeg.uminho.pt

Funding information

Fundação para a Ciência e,

Grant/Award Number: POCI‐01‐0145‐

FEDER‐006683; COMPETE,

Grant/Award Number: POCI‐01‐0145‐

FEDER‐006683; ERDF

Abstract

Health outcomes, such as mortality and readmission rates, are commonly used

as indicators of hospital quality and as a basis to design pay‐for‐performance

(P4P) incentive schemes. We propose a model of hospital behavior under P4P

where patients differ in severity and can choose hospital based on quality. We

assume that risk‐adjustment is not fully accounted for and that unobserved

dimensions of severity remain. We show that the introduction of P4P which

rewards lower mortality and/or readmission rates can weaken or strengthen

hospitals' incentive to provide quality. Since patients with higher severity have

a different probability of exercising patient choice when quality varies, this

introduces a selection bias (patient composition effect) which in turn alters

quality incentives. We also show that this composition effect increases with the

degree of competition. Critically, readmission rates suffer from one additional

source of selection bias through mortality rates since quality affects the dis-

tribution of survived patients. This implies that the scope for counter-

productive effects of P4P is larger when financial rewards are linked to

readmission rates rather than mortality rates.

1|INTRODUCTION

The ageing population and the rising prevalence of chronic conditions are putting healthcare systems under pressure.

In light of the recent financial downturn, cost containment and value for money in health spending are common

objectives across a range of health systems. Since the early eighties, many Organisation for Economic Co‐operation and

Development (OECD) countries have introduced prospective payment systems to reimburse providers, and stimulate

cost efficiency in the provision of healthcare services. However, there are concerns that such payment systems could

come at the expense of quality reductions. To address this concern, governments increasingly combine prospective

systems with further regulatory mechanisms, including a variety of pay‐for‐performance (P4P) programs that explicitly

align financial incentives with quality objectives (Busse, Geissler, Quentin, & Wiley, 2011).

The main difficulty in designing P4P programs is the identification of reliable performance indicators which reflect

providers' quality. One appealing option is to measure performance based on health outcomes. Two common indicators are

mortality rates and readmission rates which are measured through routinely collected administrative databases (Cashin, Chi,

Smith, Borowitz, & Thomson, 2014;Milstein&Schreyoegg,2016).

1

One advantage of using mortality rates is that they are

unequivocal. Higher mortality implies, everything else constant, poorer quality of care. But mortality rates are only relevant for

a subset of patient care. Patients receive many treatments for which the mortality rate is zero or negligible (e.g., a cataract

surgery and a hip replacement). For such treatments readmission rate is a valid option, and these have been used both for

treatments with negligible mortality risk and for treatments with a significant mortality risk (e.g., heart attack and hip fracture).

The main limitation of performance measures based on health outcomes is that they may reflect patient case‐mix in

addition to hospital quality. They are reliable measures of quality only if appropriate risk‐adjustment is made to account for

patients' heterogeneity in the risk of mortality and readmission (Laudicella, Li Donni, & Smith, 2013;McClellan&

Staiger, 1999; Papanicolas & McGuire, 2017;Shahian,Wolf,Iezzoni,Kirle,&Normand,2010). It is recognized that unobserved

dimensions of severity remain even after risk‐adjustment (Berenson, Pronovost, & Krumholz, 2013; Mohammed et al., 2009;

Wennberg et al., 2013) and this reduces the reliability of these performance measures. Moreover, readmission rates suffer from

one additional limitation. If some patient characteristics are unobserved (to the regulator or the researcher) and these are

correlated with the mortality risk, variations in readmission rates across hospitals may be confounded by variations in

mortality rates. For example, a hospital with a lower mortality rate may face a larger share of unobservable sicker patients,

resulting in higher readmission rates (Laudicella et al., 2013; McClellan & Staiger, 1999).

Although the empirical literature has highlighted the limitations of developing performance measures based on

health outcomes, such as mortality and readmission rates, little is known about how these limitations affect hospital

incentives to provide quality. This study fills this gap in knowledge. We propose a model of hospital behavior under P4P.

Patients differ in their severity and, thus, in their probability of negative outcomes, that is, in their mortality and

readmission risk. Patients maximize their expected utility, which depends on the mortality and readmission risk, and

choose hospitals based on quality and distance to the hospital. Hospitals operate under a P4P program, which provides a

bonus for a reduction in mortality and/or readmission rates (or, equivalently, a penalty for an increase in mortality and/

or readmission rates), and choose how much to invest in the quality of the treatment offered.

Critically, we assume that risk‐adjustment is not fully accounted for and that unobserved dimensions of severity

remain, so that outcome measures suffer from selection biases. More precisely, unobserved severity affects patient

choices and implies different demand responsiveness to quality across different patient severity types, which in turn

generates patient composition (selection) effects through demand in hospitals' mortality and readmission rates. Conse-

quently, a change in treatment quality of a hospital may not always translate into the expected change in its mortality and

readmission rates. Our model also allows for readmission rates to suffer from a patient composition (selection) effect

through mortality, since quality affects the distribution of survived patients.

Our main research question investigates how hospitals' incentives for quality provision are affected by P4P in this rich

environment. We obtain several policy‐relevant results. First, we show that whether demand responsiveness to quality is

stronger for high‐severity or for low‐severity patients is a priori ambiguous. This implies that the direct effect of P4P on

hospitals' incentives for quality provision can be either counteracted or reinforced by patient composition (selection)

effects through demand. Second, if the demand responsiveness to quality is stronger for high‐severity than for low‐

severity patients, we show that a financial bonus related to reductions in mortality rates will be counterproductive, and

instead lead to lower quality provision (and thus higher mortality rates), if the difference in mortality risks across severity

types is sufficiently large relative to the effect of treatment quality on individual mortality risk.

Third, we show that the relationship between quality and readmission rates is affected by selection bias through mortality,

and this bias increases the scope for counterproductive effects of P4P. Even if the demand responsiveness to quality is stronger

for low‐severity than for high‐severity patients, P4P linked to readmission rates might nevertheless lead to lower quality

provision if the difference in mortality risks across severity types is sufficiently large. Fourth, we show that the presence of this

additional selection bias can make readmission rates an unreliable measure to design P4P programs. In contrast to P4P linked

to mortality, where an observed reduction in mortality rates implies a higher treatment quality, a reduction in readmission

rates cannot, by itself, be taken as evidence of higher treatment quality. To make reliable inferences about treatment quality in

this case, the effects of P4P on readmission and mortality rates must be seen in conjunction.

We also investigate how the degree of competition in the market interacts with P4P incentive schemes. Stronger

competition is usually deemed to have beneficial effects insofar as it stimulates quality provision (Brekke, Cellini,

Siciliani, & Straume, 2010,2011; Gaynor, 2006; Gravelle & Sivey, 2010). However, we show that the presence of P4P can

either dampen or reinforce the positive effect of competition on quality provision, depending on whether the demand

responsiveness to quality is, respectively, stronger for high‐severity or low‐severity patients. This holds regardless of

whether P4P bonuses are linked to mortality or readmission rates. Similarly, the effect of P4P on hospitals' incentives for

quality provision is also affected by the degree of competition. If demand responsiveness to quality is stronger for high‐

severity than that for low‐severity patients, a higher degree of competition will increase the patient composition effect

through demand, thereby increasing the scope for counterproductive effects of P4P. In this case, P4P schemes are more

likely to succeed in markets with less competition. Finally, we investigate the welfare implications of P4P schemes under

different perspectives and scenarios. We discuss these in the concluding section.

290

|

LISI ET AL.