The impact of Health Information Technology bundles on Hospital performance: An econometric study

• DOI: http://doi.org/10.1016/j.jom.2015.10.001
• Author: Luv Sharma,Christopher M. McDermott,Kenneth K. Boyer,Aravind Chandrasekaran
• Published date: 01 January 2016
• Date: 01 January 2016

The impact of Health Information Technology bundles on Hospital

performance: An econometric study

Luv Sharma

a

, Aravind Chandrasekaran

b

,

*

, Kenneth K. Boyer

c

,

Christopher M. McDermott

d

a

251A Fisher Hall, Fisher College of Business, 2100Neil Avenue, Columbus, OH 43210, USA

b

650 Fisher Hall, Fisher College of Business, 2100 Neil Avenue, Columbus, OH 43210, USA

c

644 Fisher Hall, Fisher College of Business, 2100 Neil Avenue, Columbus, OH 43210,USA

d

The Lally School of Management and Technology, Rensselaer Polytechnic Institute, Troy,NY 12180e3590, USA

article info

Article history:

Received 4 June 2015

Received in revised form

13 October 2015

Accepted 19 October 2015

Available online 1 December 2015

Accepted by Mikko Ketokivi

Keywords:

HIT

Healthcare delivery

Cost

Quality of care

Econometric study

abstract

Hospitals are characterized by high levels of technical expertise as well as patient interactions. In an

attempt to improve their performance along these dimensions, hospitals are making signi ficant in-

vestments in health information technologies (HIT). However, the performance benefits from these in-

vestments are largely unknown. This study employs a portfolio approach to study HIT adoption using a

large longitudinal panel data for 3615 US hospitals from 2007 to 2012. Insights from the Advanced

Manufacturing Technology (AMT) and existing HIT literature are used to categorize 76HITs into 3 distinct

bundles based on their extent of patient centered integration, and the extentof caregiver interaction. We

then examine how two key HIT bundles: Clinical HIT (defined as HIT systems primarily used for patient

data collection, diagnosis and treatment) and Augmented Clinical HIT (defined as HIT systems primarily

used for integrating patient information and augmenting decision making capability of caregivers)

jointly impact cost and process quality outcomes. Cost is measured in terms of total hospital operating

expenses per bed while process quality is assessed along two dimensions: conformance quality or the

ability to adhere to technical standards and experiential quality or the ability to cater to preferences of

the patient. Results suggest complementarities between Clinical and Augmented Clinical HIT with

respect to process quality but not cost outcomes. A follow-up post-hoc analysis which divides

Augmented Clinical HIT into Electronic Medical Record (EMR) and Non-EMR technologies offers addi-

tional explanation to the lack of association with cost. We discuss these implications to both theory and

practice of HIT adoption.

©2015 Elsevier B.V. All rights reserved.

1. Introduction

Advances in information technology (IT) have altered the

interface between customers and service providers in numerous

professional services settings such as banks, hotels and legal ser-

vices (Froehle and Roth, 2004). A study by Lewis and Brown (2012)

sheds light on the challenges associated with IT implementation,

noting that while the firm studied had invested heavily in recent

years in distinct IT systems, “the benefit of this type of automation

was clear to some but questioned by others”(p. 7). There is a strong

concern in professional service firms regarding the trade-off be-

tween the benefits of “distilling knowledge in a reproducible form”

versus “treating human beings like bits on a conveyor belt”(Lewis

and Brown, 2012: p.12). While Lewis and Brown examine a legal

firm, this same sentiment is very common in professional service

settings such as hospitals characterized by high levels of technical

standards (that lend themselves to standardization) as well as

many patient interactions (which are much more heterogeneous).

Studies show that hospitals struggle to simultaneously improve on

conformance quality focused on technical standards as well as

experiential quality that is focused on interactions with the pa-

tients (Chandrasekaran et al., 2012). Recent changes in re-

imbursements by the Centers for Medicare and Medicaid (CMS)

penalizes hospitals if they do not show improvement on both these

process quality outcomes.

*Corresponding author.

E-mail addresses: sharma.154@fisher.osu.edu (L. Sharma), chandrasekaran.24@

fisher.osu.edu (A. Chandrasekaran), boyer.9@fisher.osu.edu (K.K. Boyer), mcderc@

rpi.edu (C.M. McDermott).

Contents lists available at ScienceDirect

Journal of Operations Management

journal homepage: www.elsevier.com/locate/jom

http://dx.doi.org/10.1016/j.jom.2015.10.001

0272-6963/©2015 Elsevier B.V. All rights reserved.

Journal of Operations Management 41 (2016) 25e41

Health Information Technology (HIT) offers one potential

avenue to successfully improve on both conformance and experi-

ential quality. To promote HIT adoptions, the US government

passed the Health Information Technology for Economic and Clin-

ical Health (HITECH) Act in 2009 offering stimulus payments of

approximately $27 Billion over the next 10years to eligible hospital

systems (HITECH, 2009). Government policy initiatives such as the

HITECH Act have had a huge impact on hospital operations, with

the average capital expenditure per bed on HIT increasing by 62%

from 2010 to 2011, while total capital expenditure in the same

hospitals increased by only 2.6% (Cerrato, 2013). In dollar terms, HIT

spending is estimated to be $34.5 billion in 2014 while maintaining

a steady growth to reach $56.7 billion in 2017 (Manos, 2013;

Cerrato, 2013).

Given this emphasis on HIT, numerous scholars have called for

empirical evidence on the relationship between HIT and hospital

performance (Agarwal et al., 2010). Following these calls, studies

have examined the impact of HIT on hospital performance. These

studies however yield mixed results and report either a positive

impact (Devaraj and Kohli, 2003; Aron et al., 2011), marginal im-

provements (McCullough et al., 2010) or negative impact (Koppel

et al., 2005) on hospital performance. Potential explanations for

these conflicting results include limitations such as focusing on a

single technology (Kohli and Devaraj, 2004; Wang et al., 2003;

Koppel et al., 2005), a lack of consideration for the user/systems

interface, and looking at hospital performance measures such as

readmissions and mortality that are subjected to several patient

characteristics (DesHarnais et al., 1990).

Our study overcomes the above limitations by taking a portfolio

approach to study HIT adoption and builds on existing HIT and

advanced manufacturing technologies (AMT) literature (Meredith,

1987; Boyer, 1999). Recent studies on HIT have recognized the

need to study these technologies in bundles (e.g. Angst et al., 2012).

We extend this idea by using insights from AMT literature and

categorizing HIT based on patient-centered integration and care-

giver interactions. We define patient centered integration as the

degree to which various HITs allow exchanging, coordinating and

effectively utilizing patient health records (excluding administra-

tive data such as billing, insurance, payroll, etc.) to enhance the

delivery of care. The second dimension, caregiver interaction, en-

compasses the degree to which a given HIT is intended to facilitate

the work of caregivers such as physicians and nurses (excluding

administrative support staff). Based on these dimensions, we

categorize HIT into three distinct bundles: (1) Administrative HIT

which constitutes technologies that have minimum levels of

patient-centered integration and almost no caregiver interaction,

(2) Clinical HIT which constitutes technologies that have moderate

levels of patient-centered integration (primarily responsible for

collection of patient data and helping with diagnosis and treat-

ment) and are used infrequently by caregivers. Finally, (3)

Augmented Clinical HIT which constitutes technologies that have a

high degree of patient-centered integration and also requires

extensive caregiver interaction. Given the minimal to no caregiver

interactions with Administrative HIT, our study primarily in-

vestigates the relationships between Clinical HIT and Augmented

Clinical HIT, after controlling for Administrative HIT, on cost and

process quality outcomes. Specifically, the following research

question is addressed in our study: How do Clinical HIT and

Augmented Clinical HIT jointly affect cost and process quality

outcomes?

We collect longitudinal data on 76 HIT and their adoption status

from 3615 U.S. hospitals during the period 2007e2012 to examine

our research question. Cost performance is measured in terms of

hospitals' operating cost per bed, while process quality is measured

in terms of conformance quality ethe level of caregivers'

adherence to evidence-based standards of care (Boyer et al., 2012),

and experiential quality ethe caregivers' ability to adapt in-

teractions to patients' specific needs (Chandrasekaran et al., 2012).

The rationale to investigate the effects of HIT adoption on the

process quality outcomes (e.g., conformance and experiential

quality) rather than the final quality of care outcomes (e.g., mor-

tality and readmissions) is supported by the following facts. First,

studies have shown that final quality of care outcomes such as

mortality and readmissions are strongly associated with several

diagnosis-related group (DRG) characteristics (DesHarnais et al.,

1990) and process quality outcomes (Senot et al., 2015) and

hence may not be ideal to study HIT adoption. Second, studying the

effects of HIT on final clinical outcomes requires a well-established

technology infrastructure which is certainly not the case for a vast

majority of U.S. hospitals (Jha et al., 2009). Finally, studies have

shown that IT adoption in professional service settings must bal-

ance the standardization of procedures with the ability to

customize customer care (Lewis and Brown, 2012). A similar

reduction in process quality can be detrimental for hospitals that

are now being reimbursed by CMS based on their conformance

quality and experiential quality scores. Hospitals are at risk losing

as much as 2% of their Medicare reimbursements if they do not

show improvements in both their conformance and experiential

quality scores beginning fiscal year 2013.

Results from our analyses indicate complementarities between

Clinical and Augmented Clinical HIT with respect to process quality

outcomes but not with respect to cost outcomes. To understand the

lack of complementarities with cost, we conducted a post-hoc

analysis by looking within the Augmented Clinical HIT bundle.

Specifically, we divided Augmented Clinical HIT into Electronic

Medical Record (EMR) and Non-EMR technologies. EMR HITs form

the basic set of technologies that are required for linking patient

records (Furukawa et al., 2010). These technologies have been the

primary focus of adoption following the HITECH Act regulations in

2009 (HITECH, 2009). The post-hoc analysis shows that EMR HIT e

Clinical HIT interaction is positively associated with cost while the

non-EMR eClinical HIT interaction is negatively associated with

cost thereby canceling each other in our main analyses. We also

find both EMR and non-EMR HITs benefit process quality outcomes.

Taken together these results offers important insights to hospital

administrators on cost-quality tradeoffs when implementing HITs.

In addition, we also highlight synergies between HIT bundles

which can be instrumental in achieving simultaneous improve-

ments in conformance and experiential quality outcomes.

2. Theoretical background

2.1. Technology adoption in healthcare professional service setting

Numerous studies in service settings have argued for a positive

association between IT investments and firm performance

(Brynjolfsson and Hitt, 1996; Boyer,1999). However, findings from

these studies may not be directly transferable to the hospital pro-

fessional service settings due to several reasons. First at the orga-

nizational level, Harris (1977) described hospitals as a non-

cooperative oligopoly with caregivers and administrators focusing

on competing objectives eeffective care vs. efficient operations.

Second, at the individual caregiver level, there is considerable

tension between physicians in different specialties as well as be-

tween physicians and nurses with respect to the healthcare quality

outcomes (Pronovost and Vohr, 2010). For instance, physicians are

more focused on the technical aspects of care delivery (i.e.,

conformance quality) while nurses are considered to be experts in

engaging with patients and families and hence are considered to be

chief architects to improve experiential quality. These differences

L. Sharma et al. / Journal of Operations Management 41 (2016)25e4126