Defensive Medicine and Obstetric Practices: Evidence from the Military Health System

• Date: 01 March 2020
• DOI: http://doi.org/10.1111/jels.12241
• Author: Michael Frakes,Jonathan Gruber
• Published date: 01 March 2020

Journal of Empirical Legal Studies

Volume 17, Issue 1, 4–37, March 2020

Defensive Medicine and Obstetric

Practices: Evidence from the Military

Health System

Michael Frakes*and Jonathan Gruber

We estimate the extent of defensive medicine by physicians during labor and delivery,

drawing on a novel and significant source of variation in liability pressure. In particular,

we embrace the no-liability counterfactual made possible by the structure of liability rules

in the Military Heath System. Active-duty patients seeking treatment from military facilities

cannot sue for harms resulting from negligent care, while protections are provided to

dependents treated at military facilities and to all patients—active-duty or not—who

receive care from civilian facilities. Drawing on this variation and addressing endogeneity

in the choice of treatment location by estimating mother fixed effects specifications and

by exploiting exogenous shocks to care location choices stemming from base-hospital clo-

sures, we find suggestive evidence that liability immunity increases cesarean utilization and

treatment intensity during childbirth, with no measurable negative effect on patient

outcomes.

I. Introduction

While the medical liability system is generally prem ised on satisfying the obje ctives of

compensation, corrective jus tice, and/or deterrence, much of the attention paid by

scholars and policymakers to th e medical malpractice system has fo cused on an

unintended, but potentially cri tical, side effect of this system: “defensive medicine.”

Defensive medicine can perhaps best be seen as deterrence gone awry (Mello & Brennan

2002), capturing situations in which fear of medical liabili ty causes physicians to practice

suboptimal care—that is, eith er providing too much care (“posi tive” defensive medicine)

or too little care (“negative” defen sive medicine). Both such response s are alarming

from a policy and welfare point of view. Ne gative defensive medicine impli cates

*Address correspondence to Michael Frakes, Duke University, 210 Science Dr., PO Box 90362, Durham, NC

27708; email: michael.frakes@law.duke.edu. Frakes is at Duke University and National Bureau of Economic

Research; Gruber is at Massachusetts Institute of Technology and National Bureau of Economic Research.

We are extremely grateful to Mike Dinneen and Daryl Crowe at the Military Health System and to Wendy Funk

and Keith Hofmann at Kennell Inc. for making the MHS data available to us. We thank Ronen Avraham for pro-

viding data on damage-cap reforms. We are also grateful to Jonathan Petkun and Chris Behrer for outstanding

research assistance. We acknowledge funding from NIA grant R01AG049898.

4

concerns over patient access to ca re, while positive defensive medic ine implicates

concerns over unnecessary heal th-care spending (to a potential ly substantial degree).

Whether liability causes positiv e or negative defensive medicine or n o defensive medi-

cine at all is likely to depend on a number of context ual factors. Arguably foremost

among those factors is the clinical set ting itself. In this article, we will r evisit the topic of

defensive medicine in a clinical sett ing that has been the most frequent ta rget of the

defensive medicine literatur e and that represents the most fr equent reason for hospitali-

zation in the United States: child birth.

Childbirth provides an especially powerful setting by which to explore the mechanics

of defensive medicine given that obstetrics/gynecology represents one of the higher-risk

specialties from a malpractice perspective. Jena et al. (2011) project that 74 percent of

obstetricians/gynecologists will face a malpractice claim by age 45 compared with 55 per-

cent for physicians practicing in internal medicine and its subspecialties. Though the topic

of defensive medicine in obstetric practices is well-trodden territory in the academic litera-

ture, there remains both theoretical and empirical ambiguity surrounding the question of

whether, to what extent, and in what direction physicians practice defensively during labor

and delivery. In this article, we approach this inquiry while bringing to bear a rich and

novel source of data and a powerful new estimation strategy. For these purposes, we build

on our recent efforts (Frakes & Gruber 2019) and turnour focus to the Military Health Sys-

tem (MHS).

The crux of our methodol ogical approach is th e Feres doctrine, a well-known

and highly controversi al rule that prohibits act ive-duty patients rec eiving medical treat-

ment from military facil ities from having any re course should they suff er harm from

negligent care—that is , they can sue neither the gov ernment nor the treating ph ysi-

cian in such circumstan ces. Importantly, howe ver, liability recou rse is available to

dependents and retirees treated at military facilities and to all patients—active-duty or

not—who receive care from civilian facilities. Variat ion in when Feres does and does

not apply affords us the oppor tunity to construct a pow erful methodological t ool

absent from traditional defensive medicine inve stigations: a treatmen t group of

patients who lack access to the medical liability system. That is, by comparing those

patients over which physicians are not subject to liability pressure to other patients

over which physicians ar e subject to such pressur e, we can identify the ful l impact of

liability forces on trea tment utilization and pat ient outcomes. The exis ting literature,

in contrast, has relied o n more limited forms of var iation—for example, ob serving dif-

ferential behavior in situations whereby patients do and do not have the ability to col-

lect pain and suffering da mages above some specified c ap. Not only may those stud ies

drawing on damages caps an d related tort reforms fai l to inform on the full exten t

of liability’s influence on behavior, but they also raise potential methodological con-

cerns to the extent that there r emains some ambiguity reg arding the extent to which

a physician’s percepti on of liability risk vari es in the presence of such limi ted reforms

(Carrier et al. 2010).

In more specific terms, to capture a liability-immunity treatment effect, we follow

Frakes and Gruber (2019)—which had embraced this approach on the full inpatient

sample—and estimate a difference-in-difference design in which we compare the care

Defensive Medicine and Obstetric Practices 5

received by active-duty versus non-active-duty mothers delivering on the base versus off

the base. To address endogeneity in the choice of on-base versus off-base care while esti-

mating this difference-in-difference design, we draw on exogenous shifts in access to

MHS bases during our sample period due to military hospital closings pursuant to recom-

mendations by the Base Realignment and Closure Commission. Moreover, to further

address endogeneity concerns, we expand on Frakes and Gruber (2019) and take advan-

tage of the fact that many patients in this childbirth subsample have repeated hospitaliza-

tions over the sample period of the same clinical nature. This fact allows us to include

mother fixed effects in many of our empirical specifications.

To execute this design, we rely on a unique set of data on births: the Military

Health System Data Repository (MDR). Covering 2003 to 2013, these data provide

incident-level claim records of inpatient stays for beneficiaries of the MHS, including

stays at both military and civilian hospitals and for active-duty beneficiaries and non-

active-duty beneficiaries. Using these records, we formulate a number of health-care

spending and utilization metrics, in addition to quality metrics.

In our previous work (Frakes & Gruber 2019), we found evidence suggesting that

immunity from liability is associated with a roughly 5 percent reduction in inpatient treat-

ment intensity. These previous findings are consistent with the conventional wisdom that

providers practice “positive” defensive medicine—that is, that liability pressures contrib-

ute to elevated spending. In the present article, however, consistent with the findings of

Currie and MacLeod (2008), we find evidence suggesting that immunity from liability is

associated with a modest increase in treatment intensity in the childbirth setting, including

a 4 percent increase (relative to the mean) in the rate of cesarean delivery. If anything,

these results are contrary to the lay perception that liability fears cause physicians to per-

form additional cesarean deliveries. Moreover, consistent with a defensive—that is,

suboptimal—interpretation of these treatment-intensity findings, we do not find any evi-

dence suggesting that immunity from liability is associated with a change in patient out-

comes. However, we stress that our estimates are imprecise in the case of some of the

health outcome metrics explored.

Our article proceeds as follows . Section II reviews the exist ing literature on

defensive medicine in obstetric pr actices, and describes the variatio n in medical liability

rules pertaining to the militar y population, which forms the b asis for our identification

strategy. Section III discusses ou r data, while Section IV sets forth our em pirical strat-

egy. Section V presents the res ults of our analysis. Finally, Sec tion VI concludes.

II. Background

II.A. Defensive Medicine Background and Literature Review

Analysts frequently attribute defensive medicine to uncertainties in the minds of physi-

cians as to what care standards are expected of them under the law (Frakes 2015;

Craswell & Calfee 1986). Consider a physician deciding whether to perform a particular

procedure in order treat a particular condition, but assume that the physician is uncer-

tain as to whether the courts will expect that she perform the procedure in question. In

6Frakes and Gruber