ArticlePDF Available

Abstract and Figures

To investigate the opportunity for hospitals to achieve better care at lower cost, we examine two key process quality measures – conformance quality and experiential quality – and two measures of performance – readmission rate and cost per discharge. Conformance quality represents hospital’s level of adherence to evidence-based standards of care, while experiential quality represents the level of interaction between hospital’s caregivers and patients. Analyzing six years of data from 3,474 U.S. acute care hospitals, we find that combining conformance and experiential quality results in lower readmission rates. However, conformance quality and experiential quality each independently increase cost per discharge which suggests that a readmissions-costs tradeoff is unavoidable. To investigate this further, we conduct post-hoc analyses by distinguishing between the granular elements of experiential quality (EQ) based on task type: responsefocused EQ and communication-focused EQ. Response-focused EQ measures caregivers’ ability to respond to patient’s explicit needs, while communication-focused EQ measures caregivers’ ability to engage in meaningful conversations with the patient. We find that combining communication-focused EQ with conformance quality reduces readmission rates. Moreover, as conformance quality increases, the cost of improving communication-focused EQ decreases, indicating complementarity. Response-focused EQ in combination with conformance quality also results in reduced readmission rates. However, as conformance quality increases, the cost of improving response-focused EQ also increases, suggesting that these dimensions might compete for resources. Taken together, our results suggest that hospital administrators can mitigate the tradeoff between reducing readmissions and controlling costs by prioritizing communication-focused EQ over response-focused EQ.
Content may be subject to copyright.
The Impact of Combining Conformance and Experiential Quality
on Hospitals’ Readmissions and Cost Performance
Claire Senot
A.B. Freeman School of Business
Tulane University
7 McAlister Dr.
New Orleans, LA 70118
csenot@tulane.edu
Phone: (504) 314-2460
Fax: (504) 865-6751
Aravind Chandrasekaran
Fisher College of Business
The Ohio State University
chandrasekaran.24@osu.edu
Peter T. Ward
Fisher College of Business
The Ohio State University
ward.1@osu.edu
Anita L. Tucker
Brandeis International Business School
Brandeis University
atucker@brandeis.edu
Susan D. Moffatt-Bruce
Wexner Medical Center
The Ohio State University
moffatt-bruce.1@osu.edu
Forthcoming in Management Science
2
The Impact of Combining Conformance and Experiential Quality
on Hospitals’ Readmissions and Cost Performance
To investigate the opportunity for hospitals to achieve better care at lower cost, we examine two key process
quality measures – conformance quality and experiential quality – and two measures of performance –
readmission rate and cost per discharge. Conformance quality represents hospital’s level of adherence to
evidence-based standards of care, while experiential quality represents the level of interaction between
hospital’s caregivers and patients. Analyzing six years of data from 3,474 U.S. acute care hospitals, we find
that combining conformance and experiential quality results in lower readmission rates. However,
conformance quality and experiential quality each independently increase cost per discharge which suggests
that a readmissions-costs tradeoff is unavoidable. To investigate this further, we conduct post-hoc analyses
by distinguishing between the granular elements of experiential quality (EQ) based on task type: response-
focused EQ and communication-focused EQ. Response-focused EQ measures caregivers’ ability to respond
to patient’s explicit needs, while communication-focused EQ measures caregivers’ ability to engage in
meaningful conversations with the patient. We find that combining communication-focused EQ with
conformance quality reduces readmission rates. Moreover, as conformance quality increases, the cost of
improving communication-focused EQ decreases, indicating complementarity. Response-focused EQ in
combination with conformance quality also results in reduced readmission rates. However, as conformance
quality increases, the cost of improving response-focused EQ also increases, suggesting that these
dimensions might compete for resources. Taken together, our results suggest that hospital administrators
can mitigate the tradeoff between reducing readmissions and controlling costs by prioritizing
communication-focused EQ over response-focused EQ.
1. Introduction
In their latest report, the Institute of Medicine argues that delivering the “Best Care at Lower Cost” is the
fundamental path to reviving America’s healthcare system (IOM 2012). However, research suggests that
this goal might entail a tradeoff between care outcomes and cost during health care delivery (Pauly 2014).
We investigate this issue in U.S. hospitals by looking at two key process quality measures: conformance
quality and experiential quality and two measures of performance – readmission rate and cost per discharge.
Conformance quality represents hospital’s level of adherence to evidence-based standards of care during
health care delivery, as documented on patients’ medical records. In particular, for specific medical
conditions (e.g., heart attack, heart failure, and pneumonia), the U.S. government has published standards
of care that have been shown to improve patient’s health (Chassin et al. 2010). Experiential quality, on the
other hand, represents the level of interaction between hospital’s caregivers and patients during health care
delivery, as experienced by the patient (Chandrasekaran et al. 2012).
To encourage hospitals to focus on both these process quality dimensions, the new health care
reimbursement policy, implemented in October 2012 by the Centers for Medicare and Medicaid Services
(CMS), evaluates hospitals based on their scores on both conformance and experiential quality (HHS 2011).
Hospitals initially risk losing 1% of their reimbursements for Medicare patients if they do not demonstrate
a focus on both conformance and experiential quality. By 2017, that penalty will increase to 2%. To reflect
3
the hospital’s emphasis on both these dimensions, we define combined quality as the extent to which the
hospital jointly pursues conformance and experiential quality.
Combined quality can result in reduced readmission rates because patients not only receive evidence-
based care but also have that care tailored to their individual needs. However, achieving combined quality
at a systemic level can also be challenging for hospitals. The health care industry has historically favored
evidence-based practices (Levinson 2010). Thus, creating a culture that emphasizes patient-centered care
without compromising the existing focus on conformance quality could involve significant training costs.
Operationalizing combined quality could also result in additional staffing costs. Indeed, because of the
different requirements for conformance and experiential quality, hospitals may need to not only allow their
caregivers to spend more time with each patient but also hire additional peripheral staff to support these
quality initiatives. These challenges make it difficult for hospitals, which are tasked with achieving
combined quality, to evaluate the related benefits and costs. The purpose of this research is to investigate
the following research question: How does a hospital’s joint pursuit of conformance and experiential
quality (i.e., combined quality) affect its readmissions and cost performance?
Prior research addresses certain elements of the process quality-performance relationship. For example,
Boulding et al. (2011) investigate the link between experiential quality and readmission rates. Jha et al.
(2009) study the cost consequences of conformance quality, while Bechel et al. (2000) investigate the cost
consequences of experiential quality. However, these studies are limited by their small sample size or a
mismatch of timeframes between process quality and performance. They also fail to ask how process quality
affects multiple aspects of performance. Finally, to our knowledge, no studies investigate the benefits and
the costs associated with combined quality – a significant gap, particularly in light of the policy changes in
hospital reimbursements.
Our research addresses these limitations and examines the relationships between combined quality and
performance in terms of readmission rates and cost per discharge. To do this, we analyze six years of
secondary data from the 3,474 U.S. acute care hospitals included in the CMS database as of June 2012. Our
results indicate synergies between conformance and experiential quality as shown by the negative effect of
combined quality on readmission rates, a key measure of hospital performance (Boulding et al. 2011). Thus,
hospitals that seek to reduce their readmission rates benefit from pursuing both conformance and
experiential quality. We also find that combined quality does not increase costs, suggesting that hospitals
do not incur an additional financial burden for jointly pursuing conformance and experiential quality.
However, we do find that improving the individual process quality dimensions independently (i.e.,
conformance and experiential quality) increases cost. Together, these results suggest that hospitals face a
tradeoff between readmissions and costs when improving their health care delivery.
4
To better delineatethis tradeoff, we conduct post-hoc analyses, looking into the granular elements of
experiential quality. Using insights from the task effectiveness literature (Stewart and Barrick 2000), we
disaggregate experiential quality (EQ) into two distinct dimensions based on the type of tasks performed
by caregivers: (i) response-focused EQ that measures caregivers’ ability to respond to patient’s explicit
needs, and (ii) communication-focused EQ that measures caregivers’ ability to engage in meaningful
conversations with the patient. We then look at the interactions between these dimensions and conformance
quality with respect to readmission rate and cost per discharge. Our results indicate that combining either
dimension of experiential quality with conformance quality reduces readmission rates. From a cost
standpoint, our results suggest that as conformance quality increases, the cost of improving communication-
focused EQ decreases while the cost of improving response-focused EQ increases. This finding suggests a
complementarity in resources between communication-focused EQ and conformance quality but not
between response-focused EQ and conformance quality. Taken together, these findings suggest that
hospital administrators can mitigate the tradeoff between readmissions and costs by initially favoring
investments that can help develop communication-focused EQ in conjunction with conformance quality
among their caregivers.
2. Prior Research and Hypotheses Development
2.1. Conformance Quality
Conformance quality represents the degree to which a product meets established standards (Garvin, 1987).
Generally, improving conformance quality has been shown to reduce internal and external failures (Deming
1982, Hendricks and Singhal 2001). In our context, conformance quality represents the level of adherence
to disease-specific evidence-based standards of care (Donabedian 1988). One manifestation of conformance
quality is the set of core process measures for common and serious conditions developed by the Joint
Commission and CMS. Studies show that following these standards improves patient’s health (Chassin et
al. 2010). For instance, when a heart attack patient is admitted to a hospital, CMS specifies a set of six
essential standard steps that must be followed, providing the patient is eligible (see Online Appendix C1
for more details). Following these six steps is likely to facilitate the patient’s recovery and help maintain
better health upon discharge (The Joint Commission 2010).
We propose that improving conformance quality will incur substantial costs for hospitals. Processes
must be restructured and employees trained on these new processes, which involves considerable expense
(Ittner et al. 2001). In addition, medical experts point to the resource-intensive nature of documenting and
monitoring conformance quality (Fonarow and Peterson 2009, Boulding et al. 2011). Despite these initial
investments, studies in the manufacturing context suggest that the reduction in costs of internal and external
failures will ultimately outweigh the increase in appraisal and prevention costs, once processes mature
5
(Juran 1988). In the healthcare environment, a patient’s condition that worsens while in the hospital because
the correct treatment was not administered on time would represent an internal failure. An unplanned
readmission because a recommended vaccine was forgotten during the initial hospital stay, which led to the
patient acquiring an infection, could be considered an external failure. However, the health care context is
characterized by rapidly evolving underlying knowledge of what is considered best practice (Bohmer and
Lee 2009), and many hospitals still have much room for improvement in developing and adhering to
standardized processes (Jewell and McGiffert 2009). Thus, we expect most hospitals to incur significant
initial and recurring costs when pursuing conformance quality.
2.2. Experiential Quality
Researchers acknowledge the consumer’s perception of the level of interaction with service providers as a
measure of experiential quality (Kellogg and Chase 1995, Parasuraman et al. 1998). Similarly, in healthcare,
experiential quality relates to the level of interaction between caregivers and individual patients, as
experienced by the patient, and is an important dimension of process quality. In 2006, CMS and the U.S.
Agency for Healthcare Research and Quality jointly developed the Hospital Consumer Assessment of
Healthcare Providers and Systems (HCAHPS) survey to measure patients’ perceptions of the level of
interaction with their caregivers during their hospital stay. Online Appendix C2 contains the list of
composites and their underlying survey items from the HCHAPS survey. Perception measures include
elements such as communication with caregivers (COMP1, COMP2, COMP5 & COMP6) and caregivers’
responsiveness to patients’ requests (COMP3 & COMP4). Researchers find that experiential quality results
in lower readmission rates (Boulding et al. 2011). Increasing the level of interaction between patients and
caregivers during health care delivery can encourage patients to not only bring to light important
information that enables caregivers to more efficiently diagnose and care for them (Groopman 2008), but
also to better adhere to discharge instructions (Blackwell 1973, Cameron 1996).
Despite these potential benefits, improving experiential quality can, on average, increase costs for
hospitals for several reasons. Elements of experiential quality – such as being responsive to patients when
they request assistance – may require hospitals to substantially invest in resources such as dedicated nurses
per floor (Neighmond 2012) and advanced information technology systems (Myers and Reed 2008), all of
which can increase operating costs. In addition, improving other elements of experiential quality – such as
communication with physicians and nurses – may require hospitals to support initiatives that require
significant added time and effort during delivery of care, such as including patients in the rounding
discussions (Nair et al. 1998). Also, although medical schools have been required to include in their
curriculum the teaching and assessment of interpersonal skills since 2002 (acgme.org), this education,
which relates to experiential quality, has been found to vary greatly across academic programs (Hojat et al.
6
2002) and is often neglected (Levinson et al. 2010). Thus, hospitals that seek to improve experiential quality
are likely to incur heavy costs to train their caregivers (Merlino and Raman 2013).
2.3. Combined Quality
Research recognizes the multi-dimensional nature of process quality and the importance of combining these
dimensions (Garvin 1987, Krishnan et al. 2000, Oliva and Sterman 2001, Voss et al. 2008). For example,
quality systems such as lean management and Six Sigma deliver high performance by integrating
conformance to standards (e.g., conformance quality) with a focus on interactions with the consumer (e.g.,
experiential quality) (Ittner and Larcker 1997, Kaynak 2003). In fact, organizational learning scholars offer
insights on this complementarity. For instance, Levinthal and Rerup (2006) argue for strong synergies
between an organization’s ability to follow routines and its ability to adapt interactions to consumers’
unique needs – skills that map onto our concepts of conformance quality and experiential quality,
respectively. Similarly, March (1994) describes how the enactment of rules that underlies conformance
quality can free-up resources needed for interacting with the consumer (i.e., experiential quality). However,
empirical evaluation of the potential synergies offered by combined quality and the associated cost in the
health care delivery context is still lacking, a gap that this study seeks to address.
Impact on Readmission Rate. Consistent with lessons from non-medical domains, the new CMS payment
program requires hospitals to simultaneously focus on both conformance and experiential quality when
delivering care. Because it emphasizes explicit standards of care, conformance quality is based on a
repository of existing medical knowledge (Swensen et al. 2010). Hospitals that have such stable knowledge
can create a more targeted interaction between caregivers and patients, which facilitates effective care.
Furthermore, experiential quality can result in better and faster identification of conditions to which
conformance quality standards can be applied, information that helps hospitals to avoid potential
complications and assist patient’s full recovery. As an illustration of the importance of experiential quality
in enhancing the effect of conformance quality, consider CMS’ standards of care. These standards dictate
that a pneumonia patient should receive an influenza vaccination to reduce the chances of re-acquiring
pneumonia as a complication of the flu (see PN7 in Online Appendix C1). However, in the absence of
experiential quality, important information (e.g., allergies precluding the patient from receiving the vaccine)
can be missed during the delivery of care, which can result in the patient’s readmission to the hospital.
Therefore, hospitals that improve patient interactions in the area of conformance quality standards can better
identify the treatments for which the patient is truly eligible and avoid unnecessary or conflicting
medications and procedures, reducing chances of readmission (Goold and Lipkin 1999). Thus, overall, we
expect combined quality to result in a healthier patient upon discharge.
7
In addition, patients from hospitals that have achieved high combined quality are likely to have higher
compliance rates with discharge instructions once they leave the hospital setting. This greater compliance
results from a combination of practices: The hospital demonstrates the importance of these guidelines by
setting an example through its own conformance quality, while at the same time it ensures that the delivery
of these instructions varies according to patient needs and preferences (e.g., visual tools, teach-back
methods), through its experiential quality. As a result of this higher compliance rate, the risk of readmission
may decrease. Hence the following hypothesis:
Hypothesis 1 (H1): Combined quality will be associated with lower readmission rates.
Impact on Cost per Discharge. Despite potential benefits of emphasizing both conformance and
experiential quality, the health care industry has historically favored conformance quality over experiential
quality (Levinson et al. 2010). Most clinicians still consider experiential quality to be a mere “bonus” for
the patient or even a burden to clinicians (Groopman 2008). To create a culture that values patient
experience without compromising evidence-based care will require caregivers to change their own mindsets
and behaviors so as to use interactions with patients to shape a delivery of care that is both standardized
and personalized. This task could be daunting for hospitals, and may involve significant training costs.
Moreover, beyond these training expenses, hospitals may also need to make significant investments in
staffing to make combined quality operationally feasible. Indeed, organizational learning theorists
recognize the challenges for individuals to simultaneously undertake activities that draw on different
learning mechanisms (Gavetti and Levinthal 2000, Gupta et al. 2006). In health care delivery, conformance
and experiential quality represent such contrasting activities. As Donabedian (1988) emphasizes,
conformance quality requires close adherence to standard guidelines, while experiential quality requires
adaptation to countless variations in patient needs and preferences. Thus, hospitals that pursue combined
quality may need to allow their caregivers to spend more time with each patient as well as to hire additional
peripheral staff to perform certain specialized tasks.
These additional training and staffing costs are well documented. For instance, Merlino and Raman
(2013) report that in 2009, all 42,000 employees at the Cleveland Clinic received training in the combined
approach to quality, a process that incurred substantial costs.Related to staffing costs, Massachusetts
hospitals increased their workforce by 11.4% (11,800 additional FTEs) between 2004 and 2008 to support
combined quality initiatives (Massachusetts Hospital Association 2010). This additional workforce was
hired to, among other tasks, collect and report measurement data, attend to patients’ needs, and advance
electronic medical reports, electronic health records, and physician order entry systems. Hospitals that
implement quality initiatives are also likely to experience a shift to more qualified and certified workers,
which can significantly increase wages (Massachusetts Hospital Association 2010). Thus, we hypothesize
the following:
8
Hypothesis 2 (H2): Combined quality will be associated with higher cost per discharge.
3. Research Design and Data
The unit of analysis in this study is the U.S. acute care hospital. We collected secondary data from multiple
sources for six years from July 2006 to June 2012 for the 3,474 U.S. acute care hospitals included in the
CMS database as of June 2012. Our study begins with fiscal year July 2006-June 2007 (year t), the first
year for which the data on experiential quality are available.
Online Appendix A lists the following seven sources of secondary data used to investigate our research
question: CMS process of care measures (conformance quality), CMS Hospital Consumer Assessment of
Healthcare Providers & Systems (HCAHPS) surveys filled out by patients (experiential quality), Medicare
cost reports (cost per discharge), CMS outcomes files (30-day readmission rate), CMS Impact files
(controls), and two websites that track state legislation and are maintained respectively by the Committee
to Reduce Infection Deaths and by the National Association on State Health Policy (instruments for
endogeneity checks for readmission rate analyses).
Online Appendix B shows the number of observations collected for the key variables for the six years
considered. Based on data availability, the final sample contains 12,538 hospital years across 2,983
hospitals for Cost per Discharge analyses, and 5,872 three-year observations across 2,936 hospitals for
Readmission Rate analyses. Hospitals in our sample are located in all 50 U.S. states and the District of
Columbia.
3.1. Performance Outcomes
Readmission Rate is reported by CMS as a three-year rolling average (at the hospital level) for three
conditions: Heart Attack (AMI), Heart Failure (HF), and Pneumonia (PN). It reflects the proportion of
patients, within each condition, who were readmitted for the same diagnosis within 30 days of discharge.
For each hospital, this percentage is adjusted by CMS for patients’ age, gender, past medical history, and
co-morbidities using hierarchical logistic regression models based on Medicare claims data
(www.medicare.gov). Low readmission rates occur when hospitals deliver the most effective care when
first admitting patients and provide helpful instructions about care plans to ensure that complications do not
arise upon discharge (Boulding et al. 2010). Following CMS guidelines, only measures that are based on a
sample of at least 25 patients for a given condition are included in the study. Relative to our study
timeframe, CMS reports hospital readmission rates for the July 2006-June 2009 and July 2009- June 2012
time periods. We thus compute, for each hospital, the weighted average of the three conditions’ readmission
rates for both of these time periods. Thus the final Readmission Rateit* value for the three-year time period
t* for hospital i with respective readmission rates AMIit*, HFit*, and PNit* and number of patients nAMIit*,
nHFit*, and nPNit* is given as follows:
9
Readmission Rateit*
***
****** )(
ititit
itititititit nPNnHFnAMI nPNPNnHFHFnAMIAMI
Cost per Discharge is estimated using the approach promoted by the Agency for Healthcare Research and
Quality and adopted by many healthcare scholars (e.g., Every et al. 1996, Chen et al. 2010, Marks et al.
2014) and state agencies (e.g., Ohio Bureau of Workers Compensation, Wisconsin ForwardHealth). That
is, we convert each hospital’s total inpatient operating charges for their fiscal years beginning in time
periods t to t+5 to 2012 U.S. dollars using the consumer price index for inpatient hospital services. We then
divide these inflation-adjusted inpatient charges by the total number of inpatient discharges and exclude the
top and bottom 1% to prevent outliers from unduly affecting the results (Every et al. 1996). Finally, we
multiply these charges by the hospital-specific Medicare inpatient operating cost-to-charge ratio to estimate
inpatient operating costs per discharge. Both inpatient operating charges and discharges are extracted from
CMS cost reports. Medicare inpatient operating cost-to-charge ratio is derived from these cost reports and
reported on CMS Impact Files with a three-year lag, which we accounted for when collecting the data. To
satisfy normality and homoscedasticity requirements, we apply the natural logarithm transformation to the
resulting ratio. The final Cost per Dischargeit value for hospital i in year t with inflation-adjusted inpatient
operating charges Oit, number of discharges Dit, and inpatient operating cost-to-charge ratio CCRit is:
Cost per Dischargeit
it
it
it CCR
D
O
ln
3.2. Process Quality
Conformance Quality (CQ) corresponds to the level of systematic adherence to evidence-based standards
achieved by hospitals when delivering care to the patient. We evaluate this construct using CMS process of
care measures that report the percentage of eligible hospitalized patients who received care in accordance
with the evidence-based guidelines in time periods t to t+5. These measures were developed in 2003 by
CMS and the Joint Commission; results are reported on the CMS Hospital Compare website
(hospitalcompare.hhs.gov).
Specifically, consistent with our readmission measure, we consider process of care measures for three
conditions: AMI, HF, and PN. Given the definition of Conformance Quality – level of systematic adherence
to evidence-based standards – we focus our attention on the 11 measures that have been deemed to
“accurately capture whether the evidence-based care has been delivered (Chassin et al. 2010: p. 685).” For
each hospital, the measure reports the percentage of eligible patients who actually receive the treatment. A
complete list of the conformance quality measures used in this study, along with sample averages and
standard deviations over the six years considered, appears in Online Appendix C1.
10
Following CMS guidelines, only measures that are based on a sample of at least 25 eligible patients are
included in the study. We compute hospitals’ weighted average percentage across all selected measures,
based on the number of patients eligible for each measure (Theokary and Ren 2011, Andritsos and Tsang
2014). Then, in accordance with statistical theory (Collett 2003) and previous research (Chandrasekaran et
al. 2012), we transform this percentage into its normally distributed logit form to satisfy the distributional
assumptions such as normality and homoscedasticity required for regression.
Conformance quality (CQit) for hospital i in year t with weighted average percentage across process of
care measures Pit is hence given by1:
it
it
it P1 P
CQ ln
Experiential Quality (EQ) measures the level of interaction between caregivers and patients, as
experienced by the patient (Chandrasekaran et al. 2012). In the context of health care delivery, this construct
is evaluated using patients’ responses to the HCAHPS survey obtained in time periods t to t+5. These
measures were developed by CMS and the U.S. Agency for Healthcare Research and Quality in 2006; the
results are also reported, at a composite level (i.e., set of two to three questions related to a common topic),
on the CMS Hospital Compare website. Our Experiential Quality construct incorporates the six composites
from this survey that measure hospitals’ emphasis on the interactions between caregivers and individual
patients (Boulding et al. 2011). Questions included in these composites ask patients to rate the extent to
which their individual care needs were considered during these interactions. These composites address
general communication (COMP 1 and COMP 2) and targeted communication (COMP 5 and COMP 6)
between caregivers and patients, as well as the level of responsiveness of caregivers to patients’ more
explicit needs (COMP 3 and COMP 4). Full text of items for each composite, along with sample averages
and standard deviations for the six years considered, appear in Online Appendix C2. Cronbach’s alpha for
these items is 0.93, which indicates excellent internal consistency (Hair et al. 2010).
Based on CMS guidelines, only data from hospitals that received survey responses from at least 100
patients are included in the study. To address potential bias from the mode of survey administration (e.g.,
phone, letter) and patient characteristics that may differ across hospitals, CMS adjusts the score for each
survey item for each hospital using patient-mix adjustments (i.e., education, self-rated health, non-English
primary language, age, and service line) and survey mode adjustments. CMS also adjusts for impact of the
time lag between discharge and completion of the survey (www.hcahpsonline.org). After making these

1 We dropped 250 observations--1% of the total sample distributed among 198 hospitals, showing conformance quality score of
100%. However, these hospitals did not have measures for all of the process conformance items and therefore had incomplete data,
which produced artificially high scores. Moreover, dropping these observations reduced our overall sample size by only six
hospitals, and including these hospitals using linear extrapolation for CQ scores did not change results. None of the hospitals in our
sample had P=0%.
11
patient-level adjustments, CMS aggregates the data to the hospital level for public reporting. Thus, although
the data in our analysis are at the hospital level, they have ex-ante been adjusted for patient-level
characteristics. For each question COMP 1 through COMP 5, CMS reports the adjusted percentage of
patients at the hospital who answered the question using the response categories “Never/Sometimes,”
“Usually,” or “Always.” We designate the percentage of patients who answered “Always” as the measure
for the items’ individual scores. COMP 6’s response categories are only “Yes” or “No,” so the percentage
score for that item is the percentage of respondents who answered the question with “Yes.” Finally, an
overall score for each hospital is calculated as the average of the percentage scores for the six items. Similar
to the Conformance Quality measure, this percentage score is then transformed into its normally distributed
logit form.
Experiential Quality (EQit) for hospital i in year t with composite percentage score is given by2:
it
it
it E1 E
EQ ln
Combined Quality (CBQ). Consistent with other studies that measure the ability of organizations to pursue
two distinct dimensions (Gibson and Birkinshaw 2004, Jansen et al. 2009), we measure Combined Quality
as the product of Conformance Quality and Experiential Quality scores. This approach best reflects the
potential synergies between the two dimensions. To minimize multicollinearity issues, we center the quality
measures before computing the product term (Aiken and West 1991). The Combined Quality (CBQit) score
for hospital i in year t is given by:
ititit EQCQCBQ
(with CQit and EQit centered)
3.3. Control variables
Previous studies identify several variables as potential sources of heterogeneity in performance across acute
care hospitals. Hence, we control for their effects in this study to minimize concerns related to differences
in service offerings (e.g., the ability to treat more severe cases). Our analysis includes six time-varying
controls: Teaching Intensity, calculated from residents-to-bed ratio (Sloan et al. 2001); Bed Size,
represented as ln(number of beds); Case Mix Index and Wage Index (Shwartz et al. 2011), both calculated
after we control for the effect of teaching intensity, because teaching hospitals tend to treat a more complex
case mix and pay higher wages than non-teaching hospitals (Nath and Sudharshan 2006, Koenig et al.
2003); OPDSH Adjustment Factor, or CMS Operating Disproportionate Share hospital payment
adjustment factor, which reflects the hospital’s propensity to treat uninsured and Medicaid patients who
often require more resources (Coughlin and Liska 1998); and Outlier Adjustment Factor, or CMS

2 None of the hospitals in the sample had E=0% or E=100%.
12
Operating Outlier adjustment factor, which reflects unusually costly cases treated by the focal hospital.
Both OPDSH and Outlier Adjustment Factors are calculated and reported by CMS. We also include year
dummies to control for unobserved factors causing overall population change in hospital performance.
Finally, through panel-data modeling, we control for hospital-level fixed effects, which include all time-
invariant hospital characteristics (e.g., corporate goals, ownership, and location).
4. Analyses and Results
The 3,474 acute care U.S. hospitals demonstrate sufficient variation in process quality (CQ, EQ), and
performance (Readmission Rate, Cost per Discharge). Online Appendix A shows the summary statistics
for all variables in this study. Table 1 presents the correlations among these variables, averaged for each
hospital across all six time periods considered. The negative and significant correlation between CQ and
EQ (r= -0.06, p <0.01) underlines the inherent tension that exists between those two dimensions (Gupta et
al. 2006).
--------------------------------- Insert Table 1 about here ------------------------------------
4.1. Endogeneity Checks
Conformance quality and experiential quality are only proxies for a hospital’s process quality initiatives,
and may raise endogeneity concerns with respect to cost. That is, a hospital’s past cost performance cannot
be linked only to its current cost performance but can also influence its current levels of conformance and
experiential quality by freeing or constraining available resources. To account for this endogeneity issue
with respect to cost, we apply a system generalized method-of-moments (System GMM) estimation
approach that uses previous lags of endogenous variables as instruments. We discuss this approach in more
detail in §4.3. A Durbin-Wu-Hausman test comparing results between the instrumented and the non-
instrumented system GMM estimations offers support for our endogeneity concerns in the prediction of
Cost per Discharge (χ2(24)=161.59, p<0.01) (Davidson and MacKinnon 1993).
Models predicting Readmission Rate may not suffer from the same endogeneity issues, because during
the time period studied (July 2006 -June 2012), hospitals were not penalized for excess readmissions
(legislation on reimbursements changed after 2012). As a result, hospitals with lower readmission rates did
not receive additional revenues, when compared to hospitals with higher readmission rates, to invest in
improving process quality. Hence, compared to cost models, theoretical arguments for endogeneity of
process quality measures in the analysis of readmissions are rather weak.
Nevertheless, we also empirically examine the endogeneity concerns for Readmission Rate models. In
order to include a sufficient number of eligible admissions, CMS only provides hospitals’ readmission rates
as a three-year rolling average. We thus have Readmission Rate observations for only two three-year time
periods per hospital (as opposed to six one-year time periods for Cost per Discharge). This arrangement
13
prevents the use of the system GMM approach – which requires at least three observations per unit of
analysis – to generate instruments based on lagged variables and test for endogeneity (Blundell and Bond
1998). Thus, we rely on prior literature to identify potential instruments for conformance quality and
experiential quality. Specifically, we use number of years since the first state-level initiative was enacted
for 1) healthcare-associated infections (HAI) and 2) patient-centered medical homes (PCMH) as
instruments for conformance and experiential quality respectively. Chandrasekaran et al. (2012) find that
hospitals located in states with longer duration of HAI laws tend to do well on conformance quality.
Similarly, legislation on PCMH emphasizes interactions between patients and caregivers in the care
delivery process (www.pcmh.ahrq.gov) and hence can be used as an instrument for experiential quality.
After centering both state legislative measures, we also compute the interaction term between them. We
use this interaction as an instrument to predict combined quality. First-stage regressions, the Angrist-
Pischke (2008) F-test for weak instruments, and Anderson’s (1984) canonical correlation test all support
the quality of these instruments. The Durbin-Wu-Hausman test (Durbin 1954, Wu 1973, Hausman 1978),
which compares results between the instrumented regression and the regular OLS regression (χ2(10)=9.19,
p=0.51), supports the lack of endogeneity concerns with respect to Readmission Rate (Davidson and
MacKinnon 1993). Because of the lack of theoretical and empirical reasons to be concerned about
endogeneity with Readmission Rate, and to achieve more efficient estimation, we do not use instrumental
variables to predict Readmission Rate in our main analyses (Wooldridge 2008).
4.2. Modeling Readmission Rate
To control for hospital-level effects, we model Readmission Rate using both fixed-effects and random-
effects estimators. A Durbin-Wu-Hausman test result indicates that modeling hospital-level effects as fixed
rather than random is most appropriate in our analyses (χ2(9) =318.38, p<0.01)3. Hence, we report the results
from the fixed-effects regressions which, given the two three-year time periods available, effectively
corresponds to a first-differences model. Specifically, the following model represents the Readmission Rate
for hospital i in the 3-year time period t*:
Readmission Rateit* *iti*it vuX
where Xit* is a vector of the independent and control variables averaged over the three years considered, ui
represents the fixed hospital-level effect, and vit* represent the idiosyncratic error term.
4.3. Modeling Cost per Discharge
We adopt the system generalized method-of-moments (System GMM) estimation approach (Arrelano and
Bover 1995, Blundell and Bond 1998, Angelini and Generale 2008, Kuhnen and Niessen 2012, Rego et al.

3 We also repeated our analyses treating hospital-level effects as random, which provided very similar results.
14
2013) to model Cost per Discharge using the xtabond2 command in STATA12. We chose this approach
based on the characteristics of our sample, namely 1) a “small T (6 time periods), large N (3000 hospitals)”
panel, 2) a linear functional relationship between our predictors and outcome variables, 3) dynamic outcome
variables (e.g., cost) whose current values depend on past realizations, 4) predictors of interest (i.e., process
quality variables) that are likely endogenous and hard to find, 5) a need to control for fixed hospital-level
effects, and 6) heteroskedasticity and autocorrelation within hospitals but not across them (Roodman 2009,
Rego et al. 2013).
The System GMM estimator uses a system of two equations, one based on the first differences in
regressors (Arellano and Bond 1991) and the other based on regressors’ previous levels. Specifically, the
two general equations estimated simultaneously for hospital i at time t are4:
ititit1itit vW2X1CostCost
(first-differences equation)
itiitit1itit vuW2X1CostCost
(levels equation)
where Xit is a vector of endogenous predictors (CQ, EQ and CBQ), Wit is a vector of exogenous predictors
(controls and time dummies), and the error term includes a hospital-specific fixed effect ui (which
disappears in the first-differences equation – control for fixed effects) and an observation-specific error vit.
Instruments for the pre-determined (i.e., Costit-1) and endogenous (i.e., Xit) variables are generated using
their lags. Specifically, in the first-differences equation, past levels of these variables are used as
instruments for their differences while, in the levels equation, past differences of these variables are used
as instruments for their levels. Arellano-Bond tests allow researchers to determine valid lags to use as
instruments, as the next section describes. See Arrelano and Bover (1995) and Blundell and Bond (1998)
for more details on this approach.
4.4. Estimation Results
For each dependent variable, we first run a regression with only the main effects of CQ and EQ, and then
include CBQ. Models 1 & 2 in Table 2 summarize the results for the Readmission Rate analyses, which use
hospital-level fixed-effects regressions; Models 3 & 4 in Table 3 show results for the Cost per Discharge
analyses, which use system GMM estimations. Multiple statistics support our Cost per Discharge models’
specification. First, for all models, α (coefficient for lagged dependent variable) has an absolute value below
unity (i.e., |α|<1), which ensures that the process converges (Blundell and Bond 1998)5. Second, while first-
order serial correlation is expected (significance of AR(1)), the Arellano-Bond test for AR(2) in first

4We also included first lagged regressors when predicting Cost per Discharge to account for longer term effects (Anderson and
Hsiao 1982, Arelanno and Bond 1991). For clarity purposes, these are not shown in the equations because they are not the focus of
this study.
5 Additional checks on the upper and lower bounds for this coefficient are described in the Robustness Checks section.
15
differences fails to reject the null hypothesis (p>0.10) that there is no second-order serial correlation in
residuals in differences (i.e., no first-order serial correlation in residuals in levels), thus supporting the
validity of using lags 2 and longer for the differences equation and lags 1 and longer for the levels equation
as GMM instruments (Arellano and Bond 1991). Third, the Hansen test of overidentifying restrictions fails
to reject (p>0.10) the null hypothesis of joint validity of the instruments, thus offering further support for
our model specification.
--------------------------------- Insert Tables 2 & 3 about here -----------------------------------
Effect of Combined Quality on Readmission Rate. H1 posits that combined quality reduces readmission
rates. Model 2 shows a strong significant negative relationship between CBQ and Readmission Rate (βCBQ=
-0.43, p<0.01), providing support for H1. It is worth noting that CQ and EQ appear to directly affect
Readmission Rate in Model 1, but the model fit is improved when CBQ is entered into the model (Δχ(1)=38,
p<0.01). This result indicates that the effect of CQ on Readmission Rate depends on the level of EQ and
vice-versa. All these observations further support the importance of considering combined quality when
studying hospitals’ performance.
Figure 1 represents the interactions plot between CQ and EQ (i.e., CBQ) with regard to Readmission
Rate. The total effect shown corresponds to both the interaction and the main conditional effects (Aiken
and West 1991). The importance of combined quality in reducing readmission rates is reflected in this plot.
Consider a hospital that is in the 75th percentile of CQ. In this case, a 1.00 percentage point increase in raw
EQ scores would correspond to a 4.95 percentage point decrease in Readmission Rate, which roughly means
avoiding one readmission for every 20 patients discharged. In contrast, for hospitals with relatively low
levels of CQ (25th percentile), a 1.00 percentage point increase in EQ would result in a 2.08 percentage
point decrease in Readmission or one readmission avoided for every 48 patients discharged.
--------------------------------- Insert Figure 1 about here ------------------------------------
Effect of Combined Quality on Cost per Discharge. H2 posits that combined quality increases cost per
discharge. Model 4 shows no significant association between CBQ and Cost per Discharge (βCBQ=0.01,
p=0.97). This result indicates that hospitals that jointly pursue both conformance and experiential quality
(i.e., combined quality) do not incur an additional cost. Thus, H2 is not supported. However, Models 3 & 4
show a significant positive main effect of both CQ (βCQ=0.09; p<0.01 for both models) and EQ (βEQ=0.32,
p<0.05 for Model 3; βEQ=0.28; p<0.01 for Model 4) on Cost per Discharge. This finding suggests that each
process quality dimension independently increases cost.
Summary of Results. Overall, our results indicate that hospitals face a tradeoff between reducing
readmissions and controlling their costs. Combined quality reduces readmission rates and thus makes
16
improvement along both conformance quality and experiential quality an imperative for hospitals.
However, costs increase independently with conformance quality and with experiential quality.
Surprisingly, no additional cost is incurred by jointly pursuing conformance and experiential quality.
To unpack the reasons for this result, we conduct post-hoc analyses to examine the granular elements of
experiential quality. To split conformance quality would require us to divide it based on patient conditions
(i.e., Heart Attack, Heart Failure, and Pneumonia), which would prohibit analyses at the hospital level.
Therefore, we do not pursue this option. We examine experiential quality in greater detail because it is
common across all patients admitted to the hospital, and hence splitting it allows us to replicate the hospital-
level analyses. Furthermore, conformance quality has been a longstanding priority for health care
practitioners. As changes to reimbursements signal a new recognition and promotion of experiential quality,
additional insights on how different elements of experiential quality interact with conformance quality
become managerially relevant. For instance, some of the elements of experiential quality may be harder –
and thus more costly – for hospitals to implement with conformance quality, while other elements may
complement and therefore be less costly for hospitals to implement in conjunction with conformance
quality. Thus, the two may cancel each other’s effect when aggregated in the main cost analysis.
4.5. Post-hoc: Granular Investigation of Experiential Quality
Two dimensions of Experiential Quality: Response-focused EQ and Communication-focused EQ.
Online Appendix C2 presents the full text of items that constitute experiential quality (COMP1- COMP6).
A closer look at these items suggests that they map onto a variety of task routines among the caregivers
during their interactions with patients. According to the task effectiveness literature, tasks can be subdivided
into behavioral and conceptual tasks depending on the type of work performed by the individuals and the
resources required to execute them. Behavioral tasks are more standardized with clear specifications of
means and ends, while conceptual tasks are less standardized with no clear specifications of means and
ends (Stewart and Barrick 2000). Researchers have shown that promoting both types of task among
individuals requires different organizational resources. For instance, organizations invest in technologies
and automation to facilitate behavioral tasks (Goodman 1986), but invest in training and educational
systems to facilitate conceptual tasks (Herold 1978).
Based on these distinctions, consider the individual items that constitute experiential quality. Scoring
high on items such as COMP3 (Staff Responsiveness) and COMP4 (Pain Management) means that
caregivers immediately responded to patients’ requests. We refer to this dimension as “Response-Focused
EQ.” From a task effectiveness standpoint, this dimension reflects behavioral task routines (McGrath 1984)
performed by the caregivers that primarily rely on motor skills–that is, a caregiver’s ability to detect and
respond to explicit patient requests. An example of this routine is the nurse’s ability to detect a patient’s
request in the nurse call system (e.g., light turning on) and immediately travel to the patient room to assist
17
with toileting or pain medication. In terms of hospital resources, response-focused EQ benefits from
investments in technologies such as visual monitoring systems (Myers and Reed 2008), RFID location
systems (Yao et al. 2012), and advanced communication systems (Wu et al. 2013) that help caregivers
quickly identify and respond to patients’ requests.
In contrast, high scores on COMP1, COMP2, COMP5, and COMP6 means that caregivers were able
to effectively communicate with patients on various topics such as general information (COMP1 &
COMP2), new medications (COMP5), and discharge instructions (COMP6). We refer to this dimension as
“Communication-focused EQ.” From a task effectiveness standpoint, this dimension reflects conceptual
task routines (McGrath 1984). These routines primarily rely on the caregiver’s ability to assimilate a
patient’s request and alter his or her response according to the patient’s implicit needs and preferences. An
example of this routine is that the physician or the nurse carefully listens to a patient’s question about
medication, answers her question in a manner that she understands, and addresses any other questions or
concerns clearly and respectfully. In terms of hospital resources, training programs that teach interpersonal
skills to the caregivers can increase communication-focused EQ. An example would be Cleveland Clinic’s
teaching of empathy and patient-centeredness to all of its caregivers (Cosgrove 2014).
Given the differences in both the type of tasks and the hospital’s investments for response-focused EQ
versus communication-focused EQ, we replicate our main analyses to examine the impact on readmission
rate and cost per discharge of combining each of these experiential quality dimensions with conformance
quality.
Post-hoc Analyses and Results. We create the Communication-Focused EQ and Response-Focused EQ
constructs in the following manner. We measure Response-focused EQ for a given hospital using the
average percentage score across COMP3 and COMP4. We computed the normally distributed logit of this
average to obtain our final measure. Similarly, we started by computing the logit of a hospital’s average
percentage score across COMP1, COMP2, COMP5, and COMP6 to assess Communication-focused EQ.
However, given that both Communication-focused EQ and Response-focused EQ are sub-constructs of
experiential quality, the degree of multicollinearity between these variables is very high (r=0.89, p<0.01),
which can be problematic if we include both variables in the same regression model. Under such conditions,
scholars recommend differentiating these variables by creating orthogonal constructs through sequential
regression (Ridker and Henning 1967, Sine et al. 2003, Nagar and Rajan 2005, Hastie et al. 2009). This
approach requires selecting one variable to remain as is and regressing the other variable against it. The
residuals of this regression are then used to represent the second construct. Sequential regression allows
assigning the common variance between two variables to one construct only – the variable selected to
remain as is – when performing multiple variable regression. For instance, Sine et al. (2003) use this
approach to disentangle prestige (evaluated through rankings) and past licensing performance (which could
18
also influence rankings) when investigating these variables’ individual effects on university inventions’
current licensing performance. Along these lines, we select Response-focused EQ to remain as is because
this construct necessarily involves some communication, which is typically standardized, to respond to
patients’ explicit needs (e.g., caregivers’ inquiring about patient’s level of pain before delivering pain
medication). Thus, although it primarily relies on the behavioral tasks identified in COMP3 and COMP4,
Response-focused EQ is also partly reflected in the other items of experiential quality, which measure all
communication. On the other hand, Communication-focused EQ does not involve any of the behavioral
tasks measured in COMP3 and COMP4. We therefore regress Communication-focused EQ (dependent
variable) against Response-focused EQ (independent variable) and use the residuals from this regression as
our final measure of Communication-focused EQ. This approach allows us to distinguish between
communication directly related to the execution of behavioral tasks and the assignment of the related
variance solely to Response-focused EQ, and the strictly conceptual communication that uniquely defines
Communication-focused EQ. Similar to the combined quality construct, Response-focused CBQ and
Communication-focused CBQ constructs were measured using the product term of CQ and the
corresponding Response-focused EQ and Communication-focused EQ dimensions, respectively. Table 4
summarizes the results of our post-hoc analyses that use analytical approaches consistent with the main
analyses (i.e., hospital-level fixed-effects regressions for Readmission Rate and system GMM estimations
for Cost per Discharge). Also mirroring the main analyses, for each dependent variable we first show results
with only the main effects of CQ, Response-focused EQ, and Communication-focused EQ (Models 5 & 7).
We then show results when Response-focused CBQ and Communication-focused CBQ are included as
additional regressors (Models 6 & 8).
--------------------------------- Insert Table 4 about here ------------------------------------
Readmission Rate. Model 6 reveals that both Response-focused CBQ (βR-CBQ= -0.26, p<0.01) and
Communication-focused CBQ (βC-CBQ= -0.92, p<0.01) have a strong negative association with Readmission
Rate. As for the main conditional effects (Aiken and West 1991) in Model 6, we find that both
Communication-focused EQ and Response-focused EQ are negatively associated with Readmission Rate
(βR-EQ= -0.47, p<0.01; βC-EQ= -1.46, p<0.01). These results suggest that both behavioral and conceptual
dimensions of experiential quality have a strong direct and synergistic effect (with conformance quality)
on readmission rate.
Figures 2 & 3 represent the interactions plots (Aiken and West 1991) between CQ and Response-
focused EQ (Figure 2) and CQ and Communication-focused EQ (Figure 3) with regard to Readmission Rate
based on Model 6. While we see that each experiential quality dimension reduces readmission rate,
irrespective of CQ, we also observe the synergies at play through the difference in slopes. For instance, a
1.00 percentage point increase in Response-focused EQ raw average score corresponds to a 1.32 percentage
19
point average decrease in Readmission Rate under low CQ versus a 2.98 percentage point average decrease
under high CQ. Similarly, a 1.00 percentage point increase in Communication-focused EQ raw average
score would decrease Readmission Rate by 2.00 percentage points under low CQ, versus 5.12 percentage
points under high CQ.
--------------------------------- Insert Figures 2 & 3 about here ------------------------------------
Cost per Discharge. Model 8 shows the effect of Response-focused CBQ and Communication-focused CBQ
on Cost per Discharge. Results reveal that Response-focused CBQ has a significant positive effect on Cost
per Discharge (βR-CBQ= 0.16, p<0.01), while Communication-focused CBQ has a significant negative effect
on Cost (βC-CBQ = -0.19, p<0.05). This result can perhaps explain the lack of interaction between
conformance quality and overall experiential quality (i.e., Combined Quality) on Cost per Discharge
(Model 4), such that these effects cancel each other out. When looking at the main conditional effects in
Model 8, we find that Response-focused EQ is not associated with Cost per Discharge (βR-EQ= 0.09, p=
0.32) while Communication-focused EQ has a weak, positive association with Cost per Discharge (βC-EQ=
0.41, p<0.10).
Figures 4 and 5, which represent the interactions plots (Aiken and West 1991) based on Model 8, help
interpret these results. Specifically, we see two observations of interest. First, under high levels of CQ,
increasing either Response-focused EQ or Communication-focused EQ increases Cost per Discharge.
Second, for each experiential quality dimension, the slopes between high and low levels of CQ add an
important insight: As CQ increases, the impact of improving Response-focused EQ on Cost per Discharge
increases (-$7 vs. $62), while the impact of improving Communication-focused EQ on Cost per Discharge
decreases ($80 vs. $48).
--------------------------------- Insert Figures 4 & 5 about here ------------------------------------
Overall, by distinguishing between the behavioral and conceptual dimensions of experiential quality,
our post-hoc analyses provide additional insights on the readmissions-costs tradeoff. Specifically, we find
that a complementarity exists between communication-focused EQ and conformance quality with respect
to both readmission rate and cost per discharge. In contrast, response-focused EQ complements
conformance quality with respect to readmission rate, but not with respect to cost per discharge. These
results offer some preliminary insights for hospital administrators on how to combine process quality
dimensions.
4.6. Robustness Checks
We performed several additional sets of analyses to check the robustness of our results to alternative model
specifications. First, to assess the validity of the system GMM results derived for costs in this study, Bond
(2002) recommends using coefficients on the lagged dependent variable from both the simple OLS and the
20
within-hospital fixed effects regressions as bounds for the true parameter. Specifically, in the OLS
regression, the lagged dependent variable would be positively correlated with the error, which would bias
its coefficient estimate upward. On the other hand, in the fixed effects regression, the lagged dependent
variable would be negatively related with the error, biasing its coefficient downward (Roodman 2006).
Thus, if our model’s validity is supported, the coefficient on the lagged Cost per Discharge variable should
be between the coefficient from the fixed-effects regression (lower bound) and the coefficient from the OLS
regression (upper bound). Fixed effects and OLS regressions consistently determine the credible range for
the true parameter of the lagged variable as [0.10 – 0.83]. Thus, the coefficient found in our system GMM
analyses (between 0.37 and 0.40) belongs to this credible range. This finding further supports the validity
of our system GMM models.
Second, to ensure that our results are robust to patient-level variations, we collected patient-level data
from one of the largest hospitals in the United States. This hospital is a public hospital located in the
Midwest with close to 1,000 beds. It is nationally ranked in the 2013-2014 top 50 best hospitals by U.S.
News for many adult specialties including cardiology, heart surgery, and pulmonology. A total of 9,910
patients were admitted between September 2008 and June 2011 for the three conditions considered in this
study: AMI, HF, and PN. From this population, data from 2,645 patients (26.69%) on core measures were
reported to the CMS and made available to the research team. The sampling criterion was based on the Joint
Commission report6 and the sampling was done by the University Healthcare Consortium--the core measure
vendor for this hospital. For these 2,645 patients, process measure data were matched with their
corresponding completed HCAHPS survey, to produce a reduced sample of 444 patients. Because all
patients were treated by the same hospital, they all shared the same overall indirect costs (e.g., training of
staff, hiring). Thus, we were unable to use these data to study the effect of process quality on cost per
discharge. Instead, we focused on the relationships between process quality and readmission rates. Scores
for conformance quality were 100% for all except nine patients. This low variation in conformance quality
would have prevented us from deriving any meaningful results if this construct had been included as a
predictor. Therefore, we limited patient-level analyses to those 435 patients with 100% conformance quality
score. Despite this limitation, these patient-level analyses allow us to investigate the effect of EQ, Response-
focused EQ, and Communication-focused EQ at high levels of CQ and thus align with our investigation of
combined quality. Data availability for the other variables considered led to a final sample of 374 patients.
In accordance with CMS procedure, we adjusted raw patients’ answers to the HCAHPS survey before
computing scores related to experiential quality according to the following patient-level factors: age group,
self-assessed health, education, and non-English primary language (www.hcahpsonline.org). This

6 For more details on the sampling process please refer to
https://manual.jointcommission.org/releases/TJC2013A/SamplingChapterTJC.html#Sample_Size_Requirements
21
adjustment was achieved through OLS regressions of each HCAHPS raw composite score (COMP1-
COMP6) on the patient-level predictors. Adjusted composite scores correspond to the residuals of these
regressions and were used to compute EQ, Response-focused EQ, and Communication-focused EQ
constructs. We adopt a condition-level fixed-effect logistic regression model to control for different
intercepts across conditions when predicting readmission (binary outcome). We also added illness severity
index and gender as controls in our analyses. Finally, to account for the possibility that a patient might elect
to get re-admitted to a different hospital and thus not be recorded by the focal hospital as a readmission
(Gonzales 2013)7 , we controlled for patient’s intention to recommend the hospital (source: HCAHPS
survey) for readmission. As demonstrated by the results shown in Table 5, using patient-level data we were
able to derive support for the strong negative association between EQ and Readmission under high levels
of CQ (i.e., Model a shows that at high levels of CQ, EQ is associated with a strong decrease in
readmissions) and for the importance of Communication-focused EQ in reducing readmissions under high
levels of CQ (i.e., Model b shows that at high levels of CQ, Communication-focused EQ is associated with
a strong decrease in readmissions). This analysis at the patient level further supports our results regarding
the effects of combined quality on readmission rate and highlights the importance of Communication-
focused EQ.
--------------------------------- Insert Table 5 about here ------------------------------------
Third, in addition to readmission rates we also collected mortality rates for hospitals. Similar to the
readmission rate measure, each 30-day mortality rate is given as an average over a 3-year period by CMS.
We therefore collected two observations per hospital to match the timeframe in our study (July 2006-June
2009 and July 2009-June 2012). We re-ran analyses with Mortality Rate as a dependent variable. Results
show no effect of Combined Quality on Mortality Rate (βCBQ= 0.11, p=0.17) and a positive effect of
Conformance Quality on Mortality Rate (βCQ= 0.16, p<0.01) (see Online Appendix D). This would suggest
not only that combined quality (as well as experiential quality) does not significantly affect mortality rates
but, even more surprisingly, that adherence to best technical practices significantly increases risk of dying
for the patient. Further investigation revealed an important flaw in this Mortality Rate measure based on
the data provided by CMS. In particular, Mortality Rate was positively and significantly correlated with
hospital’s raw Case-Mix Index for the first period considered (July 2006-June 2009: r = 0.053, p<0.01).
However, there was no longer a significant correlation between Mortality Rate and hospital’s raw Case-
Mix Index for the second period considered (July 2009-June 2012: r = 0.022, p=0.21). The positive
correlation in the first period suggests some statistical flaw in the initial risk adjustment done by CMS. We

7Hospital-level data used in our main analyses are provided by CMS and, as such, include readmissions to other hospitals.
22
urge future researchers to assemble mortality data from other sources to conduct a more thorough
investigation of this relationship.
5. Discussion and Conclusion
This study examines the relationship between a hospital’s joint pursuit of conformance and experiential
dimensions of quality, which we define as combined quality, and its impact on readmission rate and cost
per discharge. These are important relationships to investigate because hospitals may face a tension between
improving care outcomes and maintaining their financial bottom line (Berwick et al. 2008). The changes
made by CMS to its reimbursement policy, rewarding hospitals based on their performance on both
conformance and experiential quality, can only increase this tension. However, little is known about the
joint impact of these quality dimensions on multiple aspects of hospital’s performance. Our study offers
insights into these relationships.
Results show that combined quality reduces readmission rates in hospitals. However, conformance and
experiential quality each independently increases cost, even though no additional cost is associated with
combining these dimensions. The absence of a significant effect of combined quality on cost called for a
more granular investigation. We therefore conducted post-hoc analyses, looking at the different elements
of experiential quality and their interaction with conformance quality. We used insights from the task
effectiveness literature and split experiential quality into response-focused EQ and communication-focused
EQ, depending on the type of tasks.
From a hospital’s accountability standpoint, we find that combining either response-focused EQ or
communication-focused EQ with conformance quality reduces readmission rates. This finding indicates
that either dimension of experiential quality decreases the likelihood of readmission for a patient. From a
cost standpoint, post-hoc analyses reveal two important insights. First, we find that, under high levels of
conformance quality, improving either experiential quality dimension is costly for hospitals, which
underlines a tradeoff between reducing readmissions and controlling costs. However, we also find that as
conformance quality increases, the cost of improving communication-focused EQ decreases, indicating
complementarity between the two constructs. Conversely, the cost of improving response-focused EQ
increases as conformance quality increases, suggesting tension between the two dimensions.
One possible explanation for these observations is that conformance quality and response-focused EQ
require hospitals to invest in different resources. While conformance quality may benefit from dedicated
staff to gather and extract process compliance data, response-focused EQ requires hospitals to invest in
nurse call management systems and patient monitoring systems, or in adding more peripheral staff to
respond quickly to patients’ explicit needs. Therefore, response-focused EQ might compete with
conformance quality for scarce investment funds. However, hospitals’ investments to promote
communication-focused EQ, such as to educate caregivers on the importance of communication or on how
23
to interpret and address individual patients’ concerns, are also useful to promote conformance quality: They
facilitate quick identification and adherence to best technical practices and help to avoid unnecessary
procedures or tests (Wen and Kosowsky 2013). For example, chest pain might indicate a heart attack but is
also a symptom for a variety of other conditions ranging from pneumonia to a simple indigestion. Several
tests exist to identify or exclude specific conditions, such as a blood test for markers that would show
damage to the heart in the case of a heart attack. However, these markers would take time to form, thus
delaying identification in the case of a heart attack. Also, running every possible test for every possible
diagnosis would further delay care and increase cost. Simply talking with the patient can reveal whether
the pain feels like tightness or like a knife, comes in spikes or lasts several minutes at a time, and so on.
Such information is likely to allow a much faster, yet accurate, diagnosis (Harvard Heart Letter, May 2010).
Thus, where conformance quality and communication-focused EQ are concerned, investments in one
domain benefit performance in the other.
Overall, these results suggest that the readmissions-costs tradeoff can be partially mitigated through
investments that enable meaningful communication between patients and caregivers – that is, promote
communication-focused EQ. This underlying synergy may partially explain why leading healthcare
organizations such as Cleveland Clinic are spending millions of dollars and implementing mandatory
training on communication skills for their staff (Merlino and Raman 2013, Cosgrove 2014). Given the
penalties associated with readmission rates, this investment offers a potential quality improvement avenue
for hospitals.
5.1. Contributions to Theory
Our research offers three important theoretical contributions. First, we empirically demonstrate synergies
between two process quality dimensions, conformance quality and experiential quality, with regard to
readmission rates. Indeed, results indicate that combined quality reduces readmission rates. We also find
that the pursuit of either conformance or experiential quality is associated with an increased cost per
discharge. These findings emphasize the importance of including both conformance and experiential quality
measures as well as their interaction in the study of hospital performance, and offer important insights for
healthcare management scholars. For instance, Jha et al. (2009), who do not control for experiential quality,
report a weak to non-existent relationship between conformance quality and hospital costs. Thus, we
recommend that healthcare researchers investigate both process quality dimensions.
Second, quality management researchers call for more empirical research that treats quality as a
multidimensional rather than unidimensional construct (Sousa and Voss 2002), and for research that
identifies industry-specific process quality dimensions (Roth and Menor 2003). We make such
contributions to this literature by assessing two process quality dimensions, conformance and experiential
quality, that are specific to the context of healthcare delivery. We also look at more granular elements of
24
experiential quality based on the type of task – behavioral versus conceptual – performed by the caregivers.
With regard to cost performance, the different nature of interplay between conformance quality and
response-focused EQ versus communication-focused EQ suggests the need to adopt a more nuanced
approach to the study of process quality’s impact on hospital performance in the health care industry.
Finally, we find that research on health care delivery often reports mixed findings on the importance of
process quality dimensions. These results can be attributed to limitations such as studying performance
dimensions individually (e.g., Boulding et al. 2011), deriving inferences based on small sample (e.g.,
Betchel et al. 2000), or mismatching timeframes between process quality dimensions and performance (e.g.,
Jha et al. 2009). Our study largely overcomes the above limitations and represents the first large-scale
empirical test of the readmissions-costs tradeoff in a health care setting. Specifically, our results show that
despite important synergies, both dimensions of experiential quality remain expensive to improve. Thus, it
does appear difficult for hospitals to simultaneously reduce readmissions and control costs. These findings
reaffirm the importance for healthcare scholars to adopt a more encompassing view of health care delivery
by systematically considering multiple aspects of performance.
5.2. Contributions to Practice
Our study also offers important implications for hospital administrators and clinical care providers. First,
our results provide evidence that a dual focus, represented by combined quality, does reduce readmission
rates. This result underscores the importance not only of changing caregivers’ mindsets so that they can
deliver high levels of care, but also of creating organizational structures that can integrate both process
quality dimensions. For example, in hospitals, conformance and experiential quality dimensions are
typically handled by different departments. Even hospitals acknowledged as leaders in quality – such as the
Cleveland Clinic – put distinct entities in charge of conformance quality (i.e., Office of Patient Safety) and
experiential quality (i.e., Office of Patient Experience). The strong synergy with respect to readmission
rates found in our study suggests that hospitals may benefit from organizational structures that facilitate
management of the interdependencies between these two quality dimensions. One possible solution would
be to integrate both departments within the same entity.
Second, from a financial standpoint, hospitals that aim to achieve high levels of both conformance and
experiential quality should anticipate increased spending because of the independent costs associated with
each dimension. However, there appear to be some synergies between communication-focused EQ and
conformance quality with respect to cost. Therefore, given hospitals’ financial constraints, initial
investment in training and enriching communication with patients rather than in IT systems and the hiring
of additional staff, may be an ideal starting point for hospitals. Such a strategy would allow them to
simultaneously support part of experiential quality and conformance quality, and, very possibly, to avoid a
significant loss in revenues given the important and strong synergy between communication-focused EQ
25
and conformance quality in the reduction of readmission rates. Indeed, as of October 2012, CMS will levy
a significant penalty on hospitals that show excess readmissions. This result suggests that the Cleveland
Clinic’s investment in improving interpersonal skills among its caregivers is a better strategy, under
financial constraints, when compared to the hospitals from Massachusetts that focused on increasing their
staffing levels.
5.3. Policy Implications
This study also offers important implications for policy makers. The consideration of both conformance
and experiential quality included in Medicare’s Value-Based Purchasing program (beginning in October
2012) appears well targeted at reducing readmission rates. However, by weighing conformance quality and
experiential quality separately, the current legislation appears to highlight the duality rather than the
complementarity between these dimensions. Conversations with caregivers and administrators during our
patient-level data collection revealed that this legislative view makes it challenging for hospitals to
understand and promote such complementarity and can ultimately affect the delivery of care. Thus, policy
makers may want to consider a reimbursement scheme for hospitals whereby scores on conformance and
experiential quality would be combined such as through an interaction term – before being linked to
reimbursement decisions.
Furthermore, operationalizing this dual focus requires not just the implementation of control and
feedback mechanisms, but a change in longstanding mindsets – including the willingness to spend more
time with each patient – and caregivers’ training. These investments can be costly for hospitals. The
readmissions-costs tradeoff we find implies that, for the policy to work, it is important that the benefits of
achieving this dual focus outweigh its costs. Thus, this tradeoff should be properly managed not only at the
hospital but also at a policy level. Under the new payment program, hospitals that do not perform well along
both conformance and experiential quality dimensions are financially penalized and can hence suffer
millions of dollars in yearly revenue losses. Given the heavy cost of achieving combined quality, reducing
hospitals’ reimbursement for low-performing hospitals is likely to reduce their opportunity to improve their
process quality in subsequent years. Hence, the current method adopted by CMS of using a “stick over a
carrot” may increase the gap between high and low performers rather than lead to homogenously better
care. Instead, our results suggest that CMS may want to consider providing assistance, such as free training,
to low-performing hospitals at the beginning of the evaluation period instead of simply penalizing them at
the end. Such assistance could be used initially to improve the interpersonal skills, which our study shows
strongly complement conformance quality in terms of both readmission rates and cost per discharge. If such
assistance is successful, the cost incurred could subsequently be deducted from the end-of-period
reimbursement for these hospitals.
26
5.4. Limitations and Conclusion
We acknowledge that our study has several limitations that suggest avenues for further research. First, our
analyses are conducted at the hospital level. Thus, we do not control for physician-level or patient-level
characteristics that have been shown to influence performance (Hannan et al. 1989, Jollis et al. 1997, Pisano
et al. 2001, Gawande 2012). However, considering only hospital-level data also has several benefits. Most
important, recent studies in the field of medicine find benefit from hospital-wide initiatives aimed at
improving hospital performance, irrespective of patient-mix variations (Kansagara et al. 2011, Glass et al.
2012, Joynt and Jha 2013, Cosgrove 2014). We were also able to validate our results through patient-level
data from a large major teaching hospital. Nevertheless, we urge scholars to investigate multiple levels of
analyses to deepen our understanding of these trade-offs.
Second, this study shows that combined quality is a worthy endeavor for hospitals in terms of
readmission rates. However, we do not shed light on the approaches used by hospitals that have achieved
combined quality. We encourage future research to investigate the specific organizational mechanisms that
allow hospitals to achieve combined quality.
Third, we chose to study readmission rates because recent policy changes in the form of the
Readmissions Reduction Program enacted in October 2012, highlight their importance not only to patients
but also to hospitals, which face pressure to develop actionable plans to improve their readmission rates.
However, readmission rate is just one indicator of hospital performance and, as such, is limited (Press et al.
2013). For instance, Press et al. (2013) found readmission rates to be weakly correlated with quality
indicators, such as risk-adjusted mortality. They also found that hospitals’ readmission rates suffer from a
regression to the mean with high performing hospitals getting slightly worse and low performing hospitals
getting slightly better overtime. We encourage future research to further investigate the robustness of our
findings while controlling for these limitations.
Overall, this research demonstrates that the pursuit of combined quality, as promoted by the value-
based purchasing program, carries a readmissions-costs performance tradeoff for hospitals. However, given
the strong synergies between conformance and experiential quality with regard to readmission rates, this
tradeoff must be faced and managed rather than avoided. We believe that identifying the approaches to
combat this tradeoff will be of continuing interest to researchers, administrators, and policy makers.
Acknowledgments
The authors thank the department editor Serguei Netessine, the anonymous associate editor, the reviewers,
and Enno Siemsen, for their constructive feedback.
27
Table 1 Pairwise Correlations
Variable 1 2 3 4 5 6 7 8 9 10 11
1. Cost per Discharge 1.00
2. Readmission Rate -0.15 1.00
3. Conformance Qualit
y
0.29 -0.08 1.00
4. Experiential Qualit
y
-0.04 -0.27 -0.06 1.00
5. Combined Qualit
0.13 0.01 0.02 -0.27 1.00
6. Teaching Intensit
y
0.27 0.26 0.07 -0.22 0.03 1.00
7. Case Mix Index 0.58 -0.09 0.43 0.03 0.13 0.27 1.00
8. Wage Index 0.49 0.03 0.17 -0.34 0.12 0.19 0.16 1.00
9. OPDSH Adj. Factor -0.06 0.28 -0.12 -0.37 0.14 0.35 -0.03 0.16 1.00
10. Outlier Adj. Factor 0.40 -0.05 0.14 0.01 0.06 0.16 0.25 0.11 0.03 1.00
11. Bed Size 0.21 0.17 0.37 -0.56 0.09 0.39 0.35 0.21 0.29 0.07 1.00
Note. Significance levels: p 0.01 if |r| > 0.02.
Table 2 Effect of Process Quality on Readmission
Rate: Hospital-level fixed-effects regressions
Readmission Rate
Model 1 Model 2
Conformance Quality -0.48**
(0.03)
-0.46**
(0.03)
Experiential Quality -0.48**
(0.20)
-0.73**
(0.20)
Combined Quality -0.43**
(0.10)
Time-varying controls
Teaching Intensity
-2.73**
(
0.78
)
-2.92**
(
0.79
)
Bed Size
0.14
(
0.22
)
0.13
(
0.22
)
OPDSH Adj. Factor
4.80**
(
0.43
)
5.08**
(
0.43
)
Case Mix Index
(after adjusting for
-3.45**
(0.34)
-3.35**
(0.34)
Wage Index
(after adjusting for
1.89**
(0.70)
1.59*
(0.69)
Outlier Adj.Factor
-0.65
(0.98)
-0.78
(0.97)
Observations 5858 5858
Hospitals 2929 2930
Chi-square χ2(8)=1784** χ2(9)=1822**
R-square (adjusted) 67.74% 67.97%
ΔAIC (base: Model 1) - - 43.81
ΔBIC (base: Model 1) - - 37.12
Note. *p 0.05; **p 0.01. Panel data over 6 years: two 3-year time period
observations per hospital. Standard errors are clustered at the hospital-level.
Table 3 Effect of Process Quality on Cost per
Discharge: Hospital-level system GMM estimations
Cost per Discharge
Model 3 Model 4
Conformance Quality 0.09**
(0.03)
0.09**
(0.04)
Experiential Quality 0.32*
(0.16)
0.28**
(0.12)
Combined Quality 0.01
(0.06)
Time-varying controls
Teaching Intensity
0.46**
(
0.13
)
0.43**
(
0.12
)
Bed Size
-0.12**
(
0.03
)
-0.12**
(
0.03
)
OPDSH Adj. Factor
-0.01
(
0.05
)
-0.02
(
0.04
)
Case Mix Index
(after adjusting for
0.59**
(0.18)
0.54**
(0.18)
Wage Index
(after adjusting for
0.23
(0.04)
0.23
(0.04)
Outlier Adj. Factor
1.13**
(0.19)
1.16**
(0.19)
Yt-1 (lagged dependent
variable)
0.37**
(0.05)
0.37**
(0.05)
Year dummies yes yes
Lagged regressors yes yes
Observations 12538 12538
Hospitals 2983 2983
Chi-square χ2(9)=369** χ2(10)=379**
AR(1) test (p-value) (0.00) (0.00)
AR(2) test (p-value) (0.36) (0.41)
Hansen test of
overidentification (p-value) (0.55) (0.56)
Note..^p .10; *p .05; **p .01. Panel data over 6 years: six 1-year time
period observations per hospital. Standard errors are corrected for
heteroskedasticity and clustered at the hospital level. AR(1) and AR(2) test
for first- and second-order serial correlation in the first-differenced residuals,
under the null of no serial correlation. AR(2) not significant indicates that lags
2 and longer for the differences equation and lags 1 and longer for the levels
equation are valid GMM instruments. Hansen test of overidentifying
restrictions not significant supports joint validity of the instruments used.
28
Table 4 Post-Hoc: Granular Investigation of Dimensions of Experiential Quality
Readmission Rate
(fixed-effects) Cost per Discharge
(system GMM)
Model 5 Model 6 Model 7 Model 8
Conformance Quality (CQ) -0.47**
(0.03)
-0.45**
(0.03)
0.12**
(0.03)
0.10**
(0.03)
Experiential Quality (EQ)
Response-focused EQ -0.27
(0.18)
-0.47**
(0.18)
-0.03
(0.10)
0.09
(0.09)
Communication-focused EQ -0.90**
(0.34)
-1.46**
(0.33)
0.84^
(0.45)
0.41^
(0.24)
Combined Quality (CBQ)
Response-focused CBQ -0.26**
(0.09)
0.16**
(0.05)
Communication-focused CBQ -0.92**
(0.20)
-0.19*
(0.10)
Time-varying controls
Teaching Intensity
-2.71**
(
0.77
)
-2.93**
(
0.79
)
0.10
(
0.09
)
0.14
(
0.09
)
Bed Size
0.14
(
0.22
)
0.13
(
0.22
)
-0.07**
(
0.02
)
-0.06**
(
0.01
)
OPDSH Adj. Factor
4.77**
(
0.43
)
4.96**
(
0.43
)
-0.11*
(
0.05
)
-0.07^
(
0.04
)
Case Mix Index
-3.40**
(
0.34
)
-3.22**
(
0.34
)
0.04
(
0.03
)
0.05^
(
0.03
)
Wage Index
1.89**
(
0.69
)
1.50*
(
0.69
)
0.19**
(
0.05
)
0.23**
(
0.04
)
Outlier Adj. Factor
-0.65
(0.98)
-0.83
(0.98)
1.51**
(0.17)
1.53**
(0.16)
Yt-1 (lagged dependent variable) - - 0.40**
(0.05)
0.39**
(0.05)
Year dummies - - yes yes
Lagged regressors - - yes yes
Observations 5858 5858 12538 12538
Hospitals 2929 2929 2983 2983
Chi-square 1795** 1868** χ2(10)=293** χ2(12)=343**
R-square (adjusted) 67.78% 68.18% 66.66% 66.66%
ΔAIC (base: Model 1) - 8.50 - 83.00 - -
ΔBIC (base: Model 1) - 1.80 - 62.91 - -
AR(1) test (p-value) - - (0.00) (0.00)
AR(2) test (p-value) - - (0.80) (0.99)
Hansen test of overid. (p-value) - - (0.13) (0.32)
Note. ^p 0.10; *p 0.05; **p 0.01. Readmission Rate: see notes for Table 2; Cost per Discharge: see notes for Table 3.
29
Table 5 Robustness Check: Condition-level fixed-effects logistic models for patient-level data
Note. **p 0.01. Conformance Quality is 100% for all patients included. Patient-level variables included when computing independent variables: education level, self-
assessed health, age group, non-English primary language; Additional patient-level variables used as controls in the analyses as described in section 4.5: illness severity
index, gender, intention to recommend. Bootstrap standard errors.
Figure 1 Effect of Combined Quality on Readmission Rate
Note. 25th-75th percentile ranges are represented for Conformance Quality (89.9% - 97.3%).
Figure 2 Post-hoc: Effect of Response-focused CBQ on Readmission Rate
Note. 25th-75th percentile ranges are represented for Conformance Quality (89.9% - 97.3%).
Readmission Rate
Low Conformance Quality
High Conformance Quality
Low Experiential Quality High Experiential Quality
E% E+1%
-4.95%
-2.08%
Readmission Rate
Low Conformance Quality
High Conformance Quality
Low Response-focused EQ High Response-focused EQ
R-EQ% R-EQ+1%
-2.98%
-1.32%
Readmission (n=374)
Model a
Model b
Experiential Quality
-1.31**
(0.49)
Response
-
focused EQ
-0.73
(0.46)
Communication-focused EQ
-1.01***
(0.41)
+ Patient-level controls included but not shown for brevity
McFadden’s pseudo R-square 3.32%
3.43%
30
Figure 3 Post-hoc: Effect of Communication-focused CBQ on Readmission Rate
Note. 25th-75th percentile ranges are represented for Conformance Quality (89.9% - 97.3%).
Figure 4 Post-hoc: Effect of Response-focused CBQ on Cost per Discharge
Note. 25th-75th percentile ranges are represented for Conformance Quality (89.9% - 97.3%).
Figure 5 Post-hoc: Effect of Communication-focused CBQ on Cost per Discharge
Note. 25th-75th percentile ranges are represented for Conformance Quality (89.9% - 97.3%).
Readmission Rate
Low Conformance Quality
High Conformance Quality
Low Communication-focused EQ High Communication-focused EQ
C-EQ% C-EQ+1%
-5.12%
-2.00%
Cost per Discharge
Low Conformance Quality
High Conformance Quality
Low Response-focused EQ High Response-focused EQ
R-EQ% R-EQ+1%
-$7
+$62
Cost per Discharge
Low Conformance Quality
High Conformance Quality
Low Communication-focused EQ High Communication-focused EQ
C-EQ% C-EQ+1%
+$80
+$48
31
References:
Aiken, L.S., S.G. West. 1991. Multiple regression: Testing and interpreting interactions. Sage Publ., Inc.
Anderson, T.W. 1984. Estimating linear statistical relationships. Ann. Statist. 12(1):1-45.
Andritsos, D.A., C.S. Tang. 2014. Linking Process Quality and Resource Usage: An Empirical Analysis. Forthcoming in
Production Oper. Management. doi 10.1111/poms.12249
Angelini, P., A. Generale. 2008. On the evolution of firm size distributions. Amer. Econ. Rev. 98(1):426-438.
Angrist, J., J. Pischke. 2008. Mostly harmless econometrics: An empiricist's companion. Princeton University Press.
Arellano, M., S. Bond. 1991. Some tests of specification for panel data: Monte Carlo evidence and an application to
employment equations. Rev. Econ. Stud. 58(2):277-297.
Arellano, M., O. Bover. 1995. Another look at the instrumental variable estimation of error-components models. J.
Econometrics 68(1):29-51.
Bechel, D., W.A. Myers, D.G. Smith. 2000. Does patient-centered care pay off? Joint Commission J. Quality Patient Safety
26(7):400-409.
Berwick, D.M., T.W. Nolan, J. Whittington. 2008. The Triple Aim: Care, health, and cost. Health Affairs 27(3):759-769.
Blackwell, B. 1973. Patient compliance. New England J. Medicine 289(5):249-252.
Blundell, R., S. Bond. 1998. Initial conditions and moment restrictions in dynamic panel data models. J. Econometrics
87(1):115-143.
Bohmer, R.M., T.H. Lee. 2009. The shifting mission of health care delivery organizations. New England J. Medicine
361(6): 551-553.
Boulding, W., S.W. Glickman, M.P. Manary, K.A. Schulman, R. Staelin. 2011. Relationship between patient satisfaction
with inpatient care and hospital readmission within 30 days.Amer. J Managed Care 17(1):41-48.
Cameron, C. 1996. Patient compliance: recognition of factors involved and suggestions for promoting compliance with
therapeutic regimens. J. Adv. Nursing 24(2):244-250.
Chandrasekaran, A., C. Senot, K.K. Boyer. 2012. Process management impact on clinical and experiential quality: Managing
tensions between safe and patient-centered healthcare. Manufacturing & Service Oper. Management 14(4): 548-
566.
Chassin, M.R., J.M. Loeb, S.P. Schmaltz, R.M. Wachter. 2010. Accountability measures – Using measurement to promote
quality improvement. New England J. Medicine 363(7):683-688.
Chen, L.M., A.K. Jha, S. Guterman, A.B. Ridgway, E.J. Orav, A.M. Epstein. 2010. Hospital cost of care, quality of care,
and readmission rates: Penny wise and pound foolish? Arch. Internal Medicine 170(4):340-346.
Collett, D. 2003. Modelling binary data. CRC Press.
Coughlin, T.A., D. Liska. 1998. Changing state and federal payment policies for Medicaid disproportionate-share hospitals.
Health Affairs 17(3):118-136.
Cosgrove, T. 2014. The Cleveland Clinic Way: Lessons in Excellence from One of the World's Leading Healthcare
Organizations.
Davidson, R., J.G. MacKinnon. 1993. Estimation and Inference in Econometrics. Oxford University Press, NY.
Deming, W.E., Massachusetts Institute of Technology. 1982. Quality, Productivity, and Competitive Position. Vol. 183:
Massachusetts Institute of Technology, Center for Advanced Engineering. Cambridge, MA.
Donabedian, A. 1988. The Quality of Care. JAMA: J. Amer. Medical Association 260(12):1743-1748.
Durbin, J. 1954. Errors in variables. Rev. Internat. Statist. Inst. 22(1/3):23-32.
Fonarow, G.C, E.D. Peterson. 2009. Heart Failure Performance Measures and Outcomes. JAMA: J. Amer. Medical
Association 302(7):792-794.
Garvin, D.A. 1987. Competing on the eight dimensions of quality. Harvard Bus. Rev. 65 (6):101-109.
Gavetti, G., D. Levinthal. 2000. Looking forward and looking backward: Cognitive and experiential search. Admin. Sci.
Quart. 45(1):113-137.
Gawande, A. 2012. Big med. The New Yorker 20:53-63.
Gibson, C.B., J. Birkinshaw. 2004. The antecedents, consequences, and mediating role of organizational ambidexterity.
Acad. Management J. 47(2):209-226.
Glass, D., C. Lisk, J. Stensland. 2012. Refining the Hospital Readmissions Reduction Program. Washington, DC: Medicare
Payment Advisory Commission.
Gonzalez, A., T. Shih, J.B. Dimick, A.A. Ghaferi. 2013. Does same-hospital readmission rate correlate with all-hospital
readmission rate? J. Amer. College Surgeons 217(3):S105.
Goodman, P.S. 1986. Impact of task and technology on group performance. In P.S. Goodman (Ed.), Designing Effective
Work Groups: 120-167. San Francisco: Jossey-Bass.
Goold, S.D., M. Lipkin. 1999. The doctor-patient relationship. J. General Internal Medicine 14:26-33.
Groopman, J.E. 2008. How Doctors Think. Boston: Houghton Mifflin.
Gupta, A.K., K.G. Smith, C.E. Shalley. 2006. The interplay between exploration and exploitation. Acad. Management J.
49(4):693-706.
32
Hastie, T., R. Tibshirani, J. Friedman. 2009. Linear Methods for Regression. Springer.
Hausman, J.A. 1978. Specification tests in econometrics. Econometrica: J. Econometric Soc. 46(6):1251-1271.
Hair, J.F., W. C. Black, B. J. Babin, R. E. Anderson. 2010. Multivariate Data Analysis. Prentice-Hall, Upper Saddle River,
NJ.
Hannan, E.L, J.F. O'Donnell, H. Kilburn Jr, H.R. Bernard, A. Yazici. 1989. Investigation of the relationship between volume
and mortality for surgical procedures performed in New York State hospitals. JAMA: J. Amer. Medical Association
262(4):503-510.
Harvard Health Letter. May 2010. Chest Pain: A Heart Attack or Something Else? Retrieved from
http://www.health.harvard.edu/newsletters/Harvard_Heart_Letter/2010/May/chest-pain-a-heart-attack-or-
something-else.
Hendricks, K.B., V.R. Singhal. 2001. The long-run stock price performance of firms with effective TQM programs.
Management Sci. 47(3):359-368.
Herold, D.M. 1978. Improving the performance effectiveness of groups through a task-contingent selection of intervention
strategies. Acad. Management Rev. 3(2): 315-325.
HHS (U.S. Department of Health & Human Services). 2011. News release: Administration Implements Affordable Care Act
Provision to Improve Care, Lower Costs. Retrieved from
http://www.hhs.gov/news/press/2011pres/04/20110429a.html.
Hojat, M., J.S. Gonnella, S. Mangione, T.J. Nasca, J.J. Veloski, J.B. Erdmann, C.A. Callahan, M. Magee. 2002. Empathy in
medical students as related to academic performance, clinical competence and gender. Medical Ed. 36(6):522-527.
IOM (Institute of Medicine). 2012. Best Care at Lower Cost: The Path to Continuously Learning Health Care in America.
Washington, DC: The National Academies Press.
Ittner, C.D., D.F. Larcker. 1997. The performance effects of process management techniques. Management Sci. 43(4):522-
534.
Ittner, C.D., V. Nagar, M.V. Rajan. 2001. An empirical examination of dynamic quality-based learning models. Management
Sci. 47(4):563-578.
James, J.T. 2013. A new, evidence-based estimate of patient harms associated with hospital care. J. Patient Safety 9(3):122-
128.
Jansen, J.J.P., M.P. Tempelaar, F.A.J. Van den Bosch, H.W. Volberda. 2009. Structural differentiation and ambidexterity:
The mediating role of integration mechanisms. Organ. Sci. 20(4):797-811.
Jewell, K., L. McGiffert. 2009. To Err is Human—To Delay is Deadly: Ten years later, a million lives lost, billions of
dollars wasted. Austin (TX): Consumers Union.
Jha, A.K., E.J. Orav, A. Dobson, R.A. Book, A.M. Epstein. 2009. Measuring efficiency: The association of hospital costs
and quality of care. Health Affairs 28(3):897-906.
Jollis, J.G., E.D. Peterson, C.L. Nelson, J.A. Stafford, E.R. DeLong, L.H. Muhlbaier, D.B. Mark. 1997. Relationship between
physician and hospital coronary angioplasty volume and outcome in elderly patients. Circulation 95(11):2485-
2491.
Joynt, K.E., A.K.Jha. 2013.A path forward on medicare readmissions.New England J. Medicine 368(13):1175-1177.
Juran, J.M., F.M. Gryna. 1988. Juran's quality control handbook (4th ed.). New York: McGraw-Hill.
Kansagara, D., H. Englander, A. Salanitro, D. Kagen, C. Theobald, M. Freeman, S. Kripalani. 2011. Risk prediction models
for hospital readmission: A systematic review. JAMA: J. Amer. Medical Association 306(15):1688-1698.
Kaynak, H. 2003. The relationship between total quality management practices and their effects on firm performance. J.
Oper. Management 21(4): 405-435.
Kellogg, D.L., R.B. Chase. 1995. Constructing an empirically derived measure for customer contact. Management Sci.
41(11):1734-1749.
Krishnan, M.S, C.H. Kriebel, S. Kekre, T. Mukhopadhyay. 2000. An empirical analysis of productivity and quality in
software products. Management Sci. 46(6):745-759.
Kuhnen, C.M., A. Niessen. 2012. Public opinion and executive compensation. Management Sci. 58(7):1249-1272.
Levinson, W., C.S. Lesser, R.M. Epstein. 2010. Developing physician communication skills for patient-centered care. Health
Affairs 29(7):1310-1318.
Levinthal, D., C. Rerup. 2006. Crossing an apparent chasm: Bridging mindful and less-mindful perspectives on
organizational learning. Organ. Sci.17(4):502-513.
March, J.G. 1994. Primer on decision making: How decisions happen. New York: Free Press.
Massachussetts Hospital Association. 2010. Hospital Costs in Context: A Transparent View of the Cost of Care.
McGrath, J.E. 1984. Groups: Interaction and Performance. Vol. 14: Prentice-Hall Englewood Cliffs, NJ.
Merlino, J.I., A. Raman. 2013. Health care's service fanatics. Harvard Bus. Rev. 91(5):108-16, 150.
Myers, M.A, Kevin D.R. 2008. The virtual ICU: A new dimension for critical care nursing practice. Critical Care Nursing
Clinics North America 20(4):435-439.
Nagar V., M.V. Rajan. 2005. Measuring customer relationships: the case of the retail banking industry. Management Sci.
51(6) 904-919.
33
Nair, B., J. Coughlan, M. Hensley. 1998. Impediments to bed-side teaching. Medical Ed. 32(2):159-162.
Nath, D., D. Sudharshan. 2006. Measuring strategy coherence through patterns of strategic choices. Strategic Management
J. 15(1):43-61.
Neighmond P. 2012. Need a Nurse? You May Have to Wait. NPR News last accessed on 11/06/2013 at
http://www.wbur.org/npr/153634317/need-a-nurse-you-may-have-to-wait
Oliva, R., J.D. Sterman. 2001. Cutting corners and working overtime: Quality erosion in the service industry. Management
Sci. 47(7):894-914.
Parasuraman, A., V.A. Zeithaml, L.L. Berry. 1988. SERVQUAL: A multiple-item scale for measuring consumer perceptions
of service quality. J. Retailing. 64(1):12-40.
Pauly, M. 2014. Health care outcomes: When the more effective choice costs more. Knowledge@Wharton. Retrieved from
http://knowledge.wharton.upenn.edu/article/health-care-outcomes
Pisano, G.P., R.M.J. Bohmer, A.C. Edmondson. 2001. Organizational differences in rates of learning: Evidence from the
adoption of minimally invasive cardiac surgery. Management Sci. 47(6):752-768.
Press, M.J., D.P. Scanlon, A. M. Ryan, J. Zhu, A.S. Navathe, J.N. Mittler, K.G. Volpp. 2013. Limits of readmission rates in
measuring hospital quality suggest the need for added metrics. Health Affairs 32(6): 10831091.
Rego, L.L., N.A. Morgan, C. Fornell. 2013. Reexamining the market share-customer satisfaction relationship. J. Marketing
77(5):1-20.
Ridker R.G., J.A. Henning. 1967. The determinants of residential property values with special reference to air pollution.
Rev. Econom. Statist. 246-257.
Roodman, D. 2009. How to do xtabond2:An introduction to difference and system GMM in Stata.Stata J.9(1):86-136.
Roth, A.V., L.J. Menor. 2003. Insights into service operations management: A research agenda. Production Oper.
Management 12(2):145-164.
Shwartz, M., A.B. Cohen, J.D. Restuccia, Z.J. Ren, A. Labonte, C. Theokary, R. Kang, J. Horwitt. 2011. How Well Can We
Identify the High-Performing Hospital? Medical Care Res. Rev. 68(3):290-310.
Sine W.D., S. Shane, D.D. Gregorio. 2003. The halo effect and technology licensing: The influence of institutional
prestige on the licensing of university inventions. Management Sci. 49(4) 478-496.
Sloan, F. A., G. A. Picone, D. H. Taylor, S. Y. Chou. 2001. Hospital ownership and cost and quality of care: is there a dime's
worth of difference? J. Health Econom. 20(1):1-21.
Sousa, R., C.A .Voss. 2002. Quality management re-visited: a reflective review and agenda for future research. J. Oper.
Management 20(1):91-109.
Stewart, G.L., M.R. Barrick. 2000. Team structure and performance: Assessing the mediating role of intrateam process and
the moderating role of task type. Acad. Management J. 43(2):135-148.
Swensen, S.J., G.S. Meyer, E.C. Nelson, G.C. Hunt Jr, D.B. Pryor, J.I. Weissberg, G.S. Kaplan, J. Daley, G.R. Yates, and
M.R. Chassin. 2010. Cottage industry to postindustrial care – The revolution in health care delivery. New England
J. Medicine 362(5):e12(1-4).
Theokary, C., Z.J. Ren. 2011. An empirical study of the relations between hospital volume, teaching status, and service
quality. Production Oper. Management 20(3):303-318.
Toussaint, J. S., L.L.Berry. 2013. The Promise of Lean in Health Care. Mayo Clinic Proceedings.
Voss, C., A.V. Roth, R.B. Chase. 2008. Experience, service operations strategy, and services as destinations: foundations
and exploratory investigation. Production Oper. Management 17(3):247-266.
Wen, L., and J. Kosowsky. 2013. When Doctors Don't Listen: How to Avoid Misdiagnoses and Unnecessary Tests. Thomas
Dunne Books.
Wooldridge, J. 2008. Introductory Econometrics: A Modern Approach. Mason: South-Western College Publications.
Wu, D.-M. 1973. Alternative tests of independence between stochastic regressors and disturbances. Econometrica: J.
Econometric Soc. 41(4):733-750.
Yao, W., C.-H. Chu, Z. Li, Z. 2012. The adoption and implementation of RFID technologies in healthcare: a literature
review. J. Medical Systems. 36(6): 3507-3525.
34 ONLINE APPENDIX
Online Appendix A Summary statistics and sources
Variables Description Mean Std.
dev.
Min. Max. Source
Performance Measures
Cost per Discharge
Ln(Total Inpatient Operating
Charges x CCR / Number of
Discharges)
8.75 0.44 5.31 11.22 CMS Cost Reports
& Impact Files
Readmission Rate Weighted average readmission rate
across AMI, HF and PN 0.21 0.02 0.16 0.29 CMS Outcomes
Files
Process Quality dimensions
Conformance Quality
(prior to centering) Logit of hospital’s weighted average
scores on AMI, HF and PN process
of care measures
2.89 1.11 -3.90 7.95 CMS Process of
Care measures
Experiential Quality
(prior to centering) Logit of hospital’s average scores on
COMP 1-COMP 6 HCAHPS survey 0.90 0.28 -0.71 3.42 CMS HCAHPS
survey
Combined Quality Conformance Quality x Experiential
Quality (centered variables) 0.02 0.31 -3.71 5.60 -
Time-Varying Controls
Teaching Intensity Residents-to-beds ratio 0.06 0.15 0.00 1.75
CMS Impact Files
Case Mix Index Case mix index residuals after
regression on Teaching Intensity 0.00 0.32 -0.98 2.63
Wage Index Wage Index residuals after
regression on Teaching Intensity 0.00 0.19 -0.65 0.97
OPDSH Adjustment
Factor Operating disproportionate share
payment reflects low income patient
population
0.09 0.11 0.00 0.88
Outlier Adjustment
Factor Operating outlier payment reflects
exceptionally costly cases 0.04 0.09 0.00 3.52
Bed Size Ln(number of beds) 4.85 0.96 0.69 7.68 CMS Cost Reports
State Legislations
HAI Legislation
(prior to centering)
Number of years, as of year t, since
the first HAI initiative was enacted by
the focal state
2.10 1.73 0 6 CRID database
PCMH Legislation
(prior to centering) Number of years, as of year t, since
the first PCMH initiative was enacted
by the focal state
0.91 1.96 0 12 NASHP database
Ambidextrous
Legislation HAI Legislation x PCMH Legislation
(centered variables) -0.27 1.97 -8.70 4.89 -
35 ONLINE APPENDIX
Online Appendix B Time periods and number of observations available for key variables
Variable Time Period Number of observations (N=3474)
Process Quality Dimensions
Conformance Quality July 2006- June 2007 (t)
July 2007- June 2008 (t+1)
July 2008- June 2009 (t+2)
July 2009- June 2010 (t+3)
July 2010- June 2011 (t+4)
July 2011- June 2012 (t+5)
3075
3074
3054
3058
3072
2912
Experiential Quality July 2006- June 2007
July 2007- June 2008
July 2008- June 2009
July 2009- June 2010
July 2010- June 2011
July 2011- June 2012
2110
3115
3114
3174
3158
3148
Combined Quality July 2006- June 2007
July 2007- June 2008
July 2008- June 2009
July 2009- June 2010
July 2010- June 2011
July 2011- June 2012
2077
2979
2941
2967
2945
2793
Performance Measures
Readmission Rate
July 2006- June 2009 3118
July 2009- June 2012 3136
Cost per Discharge Fiscal Year beginning July 2006- June 2007
Fiscal Year beginning July 2007- June 2008
Fiscal Year beginning July 2008- June 2009
Fiscal Year beginning July 2009- June 2010
Fiscal Year beginning July 2010- June 2011
Fiscal Year beginning July 2011- June 2012
3161
3188
3201
3213
3160
2677
36 ONLINE APPENDIX
Online Appendix C1 Conformance Quality measurement items and descriptive statistics
Online Appendix C2 Experiential Quality measurement items and descriptive statistics
Timeframe: July 2006-June 2012 Measure Grand Mean Avg. St. Dev. Min. Mean Max. Mean
Process of care measures 92.08% 7.59% 86.79% 96.34%
HEART ATTACK (AMI)
Patients given aspirin at arrival AMI 1 97.09% 3.27% 95.46% 98.62%
Patients given aspirin at discharge AMI 2 97.30% 4.80% 95.63% 98.69%
Patients given ACE inhibitor or ARB for LVSD AMI 3 93.98% 6.17% 88.95% 96.72%
Patients given beta blocker at discharge AMI 5 97.18% 4.69% 95.70% 98.18%
Patients given fibrinolytic medication within 30 minutes of
arrival AMI 7a 73.07% 13.83% 66.29% 81.50%
Patients given PCI within 90 minutes of arrival AMI 8a 84.06% 11.21% 64.29% 94.88%
HEART FAILURE (HF)
Patients given ACE inhibitor or ARB for LVSD HF 3 92.85% 7.18% 86.85% 96.44%
PNEUMONIA (PN)
Patients assessed and given pneumococcal vaccination PN 2 87.45% 13.81% 77.88% 93.60%
Patients whose initial ER blood culture was performed prior
to admin. of the 1st hospital dose of antibiotics PN 3b 93.90% 5.74% 90.00% 96.94%
Patients given the most appropriate initial antibiotic(s) PN 6 90.84% 7.16% 87.39% 94.79%
Pneumonia patients assessed and given influenza
vaccination PN 7 85.87% 14.04% 77.96% 92.69%
Timeframe: July 2006-June 2012 Measure Grand Mean Avg. St. Dev. Min. Mean Max.
Mean
HCAHPS survey measures 70.72% 5.35% 68.81% 73.06%
NURSE COMMUNICATION COMP 1 74.45% 6.12% 72.21% 77.19%
Nurses treated patients with courtesy and respect
Nurses listened carefully to patients
Nurses explain things to patients in a way they could
understand
DOCTOR COMMUNICATION COMP 2 79.43% 5.32% 78.69% 80.45%
Doctors treated patients with courtesy and respect
Doctors listened carefully to patients
Doctors explain things to patients in a way they could
understand
STAFF RESPONSIVENESS COMP 3 61.75% 8.49% 58.98% 64.82%
Patients got help as soon as wanted after pressing the call
button
Patients got help as soon as wanted to use the restroom
PAIN MANAGEMENT COMP 4 68.36% 5.59% 66.92% 70.02%
Patients’ pain was well controlled
Hospital staff did everything they could to help patients manage
their pain
COMMUNICATION ABOUT MEDICATION COMP 5 59.21% 6.37% 57.05% 62.16%
The purpose for new medications was explained to patients
Side effects of new medications were clearly described
COMMUNICATION ABOUT DISCHARGE COMP 6 81.14% 4.97% 79.01% 83.75%
Staff verified that patient will have the help needed after leaving
the hospital
Patients received written instructions regarding symptoms or
health problems to monitor after leaving the hospital
37 ONLINE APPENDIX
Online Appendix D Hospital-level fixed-effects regressions predicting mortality rate
Mortality Rate
Model 1 Model 2
Conformance Quality 0.17**
(0.03)
0.16**
(0.03)
Experiential Quality -0.13
(0.16)
-0.06
(0.16)
Combined Quality 0.11
(0.08)
Time-varying controls yes yes
Observations 5844 5844
Hospitals 2922 2922
Chi-square 72** 72**
R-square (adjusted) 60.95% 60.96%
Note. **p 0.01. Panel data over 6 years: two 3-year time period observations per hospital. Standard errors are clustered at the hospital-level.
... Consistent with past scholarship (Senot et al., 2016;Sharma et al., 2016), we measure performance in terms of conformance quality (CQ), which represents a caregiver's adherence to evidence-based standards of care during health care delivery, as documented on the patient's EMR. It has been shown that if caregivers follow these guidelines, patient health is improved (Chassin et al., 2010). ...
... Conformance quality 11 represents a caregiver's adherence to evidence-based standards of care during health care delivery (Senot et al., 2016;Sharma et al., 2016). More specifically, it is a collection of core process of care measures that report the proportion of eligible patients who receive care in accordance with these guidelines or standards. ...
... The measures were developed in 2003 by CMS for common and serious health conditions. Consistent with past scholarship (Senot et al., 2016), we consider process of care measures for three health conditions: acute myocardial infarction (AMI), heart failure (HF), and pneumonia (PN), as these measures "accurately capture whether the evidence-based care has been delivered" (Chassin et al., 2010, p. 685). Conformance quality data was collected from CMS's HospitalCompare website (see Appendix A for a complete list of the 11 individual items representing CQ that were used in this study). ...
Article
The significance of information technology (IT) to organizational performance is increasingly acknowledged by senior executives and scholars alike. However, an important aspect that is underdeveloped in the extant literature is whether and how an organization's IT sourcing strategy affects performance. We examine this dimension in the context of hospitals’ sourcing strategy for electronic medical record (EMR) systems. Leveraging the organization‐IT alignment principle grounded in contingency theory, we argue that the sourcing strategy chosen by hospitals impacts the quality of patient care rendered and that the nature of this relationship has changed over time. Using data on US hospitals that operated continuously from 2006 to 2013 and employing a multilevel longitudinal modeling framework to analyze change over time, we examine how a key dimension of the sourcing strategy for EMR systems—change in closeness to single‐sourcing—impacts conformance quality, a critical measure of hospital performance that assesses how frequently hospitals comply with evidence‐based practices. We find that a change closer to single‐sourcing is positively associated with conformance quality, but this positive effect is reduced over time, meaning that the benefits realized by hospitals using a single‐sourcing approach diminish toward the end of our observation period. As explicated in our discussion of results, findings from our research reveal critical insights about how sourcing decisions impact performance, and how these effects evolve over time due to changes in regulations, technology, and competition.
... In our GMM analysis, we used the lagged value of violation experiences and the interaction terms as instrumental variables (Senot et al., 2016). Appendix Table F presents the GMM analysis results. ...
Article
Full-text available
Scholars, the government, and the public expect that legal enforcement should be an effective means to prevent a firm's future safety violation behavior in daily productions. However, the literature provides limited insight into whether safety regulation enforcement is effective in helping firms reduce future safety violations. Therefore, this study examined the relationship between occupational health and safety enforcement and a firm's future violation behavior. We carried out a regression analysis based on a panel sample of 2965 listed United States manufacturers with 4474 violation records issued by the Occupational Safety and Health Administration. We find limited effectiveness of enforcement. Our results indicate that although past violation experiences are negatively correlated with subsequent repeat violations (violations of the same clause), they are positively correlated with non-repeat violations (violations of a different clause) and the number of overall violations (the sum of both). The factors that can reinforce effectiveness were also explored in this study. Our analyses show that the effectiveness of safety regulation enforcement is significantly enhanced when reinforced by stringent penalties and Occupational Health and Safety Assessment Series 18001 certifications. This study contributes to the literature on the operational safety and policy–operations interaction. We challenge conventional wisdom by proposing that regulatory pressures do not necessarily improve firms' overall social responsibility practices in terms of worker health and safety. We also discuss the implications of this for the occupational health and safety (OHS) management practice.
Purpose The purpose of this paper is to suggest how decision-makers may work towards a broader perspective on value than that expressed in financial economics-based accounting terms to enhance value in healthcare. Design/methodology/approach The authors review published academic research and reports on practice across a range of disciplines. Findings The authors find that while value is a multidimensional concept, which is open to perceptions that differ across stakeholders in healthcare, financial economics-based accounting is essentially mono-disciplinary and dominates decisions. Enhancing value in health is a wicked problem, and a trans-disciplinary approach has the potential to enable decision-makers to enhance value. Practical implications The suggest that a trans-disciplinary approach, which dissolves disciplinary boundaries, is capable of enabling decision-makers to work towards understanding and enhancing value by fostering awareness of stakeholders' perceptions of value. A critical caveat is that a trans-disciplinary approach does not guarantee ready-made or immediate solutions; it does, however, offer the means to struggle towards a destination which may be continually shifting. Originality/value This study highlights the importance of a broader understanding of the concept of value than that implied by financial economics-based accounting and recognises the perceptions of stakeholders. It explores the inter-relationship among “the view from nowhere”, wicked problems and trans-disciplinarity and recommends a trans-disciplinary approach with a view to enhancing value in that broader sense. In this way, it contributes to the accounting literature, which has previously paid little attention to some of these aspects.
Article
Problem definition: We investigate how the degree of standardization (DoS) across service chain-belonging units impacts units’ performance outcomes. We study this question in the nursing home industry, focusing on the DoS in three operational dimensions—customer mix, service offering, and service delivery—and its impact on three nursing home outcomes: financial performance, clinical outcome, and resident welfare. Academic/practical relevance: All service chains must decide how unique or standardized each of their units should be, but it is unclear how this decision affects units’ performance outcomes. We assess standardization strategies in multiple operational dimensions in chain-belonging nursing homes and their impact on important operational outcomes, and generalize our findings to other service industries. Methodology: We use a panel data set covering 11 years (2005–2015) of comprehensive measurements of all nursing homes in the United States. We estimate the relationship between the DoS and important nursing home outcomes using fixed-effect models with instrumental variables and clustered standard errors. Results: We find that chain-belonging nursing homes that (1) customize service delivery and standardize customer mix relative to the chain norm tend to experience improved financial performance; (2) standardize customer mix tend to experience improved clinical outcome; and (3) customize service delivery tend to experience enhanced resident welfare. Managerial implications: To our knowledge, this study is the first systematic multidimensional assessment of multiunit service firms’ standardization strategies from a chain perspective. Our results guide nursing home chains in their (1) strategies of within-chain standardization, and (2) investment decisions in acquiring new properties. Our results also provide broader managerial insights for service chains’ standardization strategies and merging and acquisition activities.
Article
Reduction in hospital readmissions has long been identified as a target area for healthcare public policy reform by the U.S. government. In October 2012, the Affordable Care Act (ACA) established the Hospital Readmissions Reduction Program (HRRP) program, which requires the Centers for Medicare and Medicaid Services (CMS) to reduce payments to hospitals with excess readmissions. Even with recent changes in ACA, the HRRP program is still in place. In this study, we empirically examine the effectiveness of the introduction of the HRRP on hospital readmission and mortality rates. We observe that, in general, the introduction of the HRRP has significantly reduced the readmission rates. However, the introduction of the HRRP does not necessarily decrease the mortality rates, which highlights the unintended consequences of public policy. What is more interesting is that the impact of the HRRP is heterogeneous in hospital size and racial groups. First, after the HRRP introduction, large hospitals have experienced a greater reduction in readmission rates than small hospitals. Second, after the introduction of the HRRP, the zip code regions with a higher percentage of Hispanic and African‐American populations have experienced a larger reduction in readmission rates. These results contribute to both theory and practice in public policy and provide important and nuanced policy implications for evaluating the effectiveness of the HRRP. Policy‐makers also need to pay close attention to these results for future implementations of policies similar to the HRRP. This article is protected by copyright. All rights reserved
Article
Problem definition: Diabetes is a highly prevalent and expensive chronic disease that affects millions of Americans and is associated with multiple comorbidities. Clinical research has found long-term variation in a patient’s glycated hemoglobin levels to be linked with adverse health outcomes, such as increased hospitalizations. Consequently, there is a need for innovative approaches to reduce long-term glycemic variability and efficient ways to implement them. Academic/practical relevance: Although the operations literature has extensively explored ways to manage variability across patients, relatively little attention has been paid to within-patient variability. We draw on the management and healthcare literatures to hypothesize and then show that a key operational lever—continuity of care (CoC)—can be used to reduce glycemic variability, which in turn, improves patient health. In the process, we explore the moderating role of a key demographic characteristic: patient’s marital status. We also shed light on an important mechanism through which CoC reduces variability—adherence of patients to prescribed medications—thereby advancing the compliance literature. Academically, our study adds to the understanding of the importance of managing variability (via continuity in service) in settings where customers repeatedly interact with service providers. Methodology: We use a detailed and comprehensive data set from the Veterans Health Administration, the largest integrated healthcare delivery system in the United States. This permits us to control for potential sources of heterogeneity. We analyze more than 300,000 patients—over an 11-year period—with diabetes, a chronic disease whose successful management requires managing glycemic variability. We use an empirical approach to, first, quantify the relationship between CoC and glycemic variability and second, show how this relationship differs based on patient’s marital status. Third, we estimate the mediation effect of patients’ adherence to medications. Finally, we quantify how glycemic variability mediates the relationship between CoC and three important outcomes. Our findings are validated by extensive robustness checks and sensitivity analyses. Results: We find that CoC is related to reductions in glycemic variability, more so for patients who are not married. However, this reduction is not linear in continuity; we find evidence of curvilinearity but with a sufficiently high stationary point so that benefits almost always accrue, albeit at a diminishing rate. Additionally, we find that one mechanism through which CoC may reduce variability is through patients’ adherence to medications. We also find evidence of partial mediation for glycemic variability in the CoC outcomes process chain. Our counterfactual analysis reveals the extent of improvement that enhanced continuity can bring, depending on where it is targeted. Managerial implications: Identifying the process measures through which continuity of care reduces variability is of interest to practitioners and policy makers as it can help design appropriate policies and pathways in terms of both processes and staffing/work allocation.
Article
Purpose In the face of growing demand for care and tightening resource constraints, hospitals need to ensure access to care that is affordable and effective. Yet, the multiplicity of objectives is a key challenge in this industry. An understanding of the interrelationships (tradeoffs) between the multiple outcome objectives of care (throughput, experiential and financial performance) and returns to operational inputs (diversification of care) is fundamental to improving access to care that is effective and affordable. This study serves to address this need. Design/methodology/approach The empirical analysis in the study builds on an output-oriented distance function model and uses a longitudinal panel dataset from 153 hospitals in California. Findings This study results point to key insights related to output–output tradeoffs along the production frontier. Specifically, the authors find that higher throughput rates may lead to significantly lower levels of experiential quality, and net revenue from operations, accounting for the clinical quality of care. Similarly, the authors’ findings highlight the resource intensity and operational challenges of improving experiential quality of care. In regards to input–output relationships, this study finds diversification of care is associated with increased throughput, improvements in service satisfaction and a corresponding increase in the net revenue from operations. Originality/value Highlighting the tradeoffs along the production frontier among the various outcomes of interest (throughput, experiential quality and net revenue from operations), and highlighting the link between diversification of care and care delivery outcomes at the hospital level are key contributions of this study. An understanding of the tradeoffs and returns in healthcare delivery serves to inform policy-making with key managerial implications in the delivery of care.
Article
Hospitals are under tremendous pressure to improve multiple performance dimensions that may be at odds, including cost containment, clinical quality, and patient satisfaction. The need to improve performance on multiple fronts is compounded by sources of uncertainty that include upstream diversity of supply chain (SC) partners, internal diversity of clinical specialties provided, and downstream diversity of patient conditions. Information processing theory (IPT) suggests two strategies for dealing with uncertainty: reduce the amount of information elevated to higher levels for processing and accommodate the amount of information by increasing information processing at other levels. We apply IPT to investigate (i) the relationship between the two information processing strategies in a hospital's SC and its cost containment, clinical quality, and patient satisfaction performance, and (ii) how these relationships may be moderated by diversity in the hospital's SC partners, clinical specialties, and patients, the common sources of uncertainty in hospital SCs. These questions were examined using regression analysis of primary data from acute care hospitals matched with secondary data from the Centers for Medicare and Medicaid Services, American Hospital Association and the Agency for Healthcare Research and Quality measuring hospital operating characteristics and performance. The analysis yielded several important findings: (i) while a slack resources strategy (uncertainty reduction) is effective for improving patient satisfaction performance, it is ineffective for reducing cost, (ii) a lateral relations strategy (uncertainty accommodation) is effective in improving all three performance dimensions, and (iii) these relationships are moderated differentially by the three diversity dimensions only for lateral relations. These findings offer important managerial insights about hospitals’ efforts to improve potentially conflicting performance outcomes and suggest interesting opportunities for future research applying IPT to this important topic.
Article
Problem definition: This study explores the performance implications of collaboration structures in an integrated healthcare delivery system, namely, an accountable care organization (ACO). ACOs consist of providers from various stages of the care continuum (preacute, acute, and postacute) that voluntarily assume collective responsibility for the quality and cost of care for a defined patient population. Academic/practical relevance: ACOs’ unsatisfied outcomes are largely due to a lack of provider collaboration. There is a dearth of empirical studies on how to develop collaboration structures. Studies in the healthcare operations management primarily have focused on collaboration within a single organization, shedding little light on this problem. We address this issue by exploring two distinct dimensions of collaboration: partnership scope and scale. Partnership scope measures the presence of providers from the preacute, acute, and/or postacute care-continuum stages, whereas partnership scale measures the presence of providers within a single care-continuum stage. Methodology: We assemble a unique data set of provider types, collaboration structures, and system-level performance for 528 Medicare Shared Savings Program (MSSP) ACOs from 2013–2016. To investigate the impact of partnership scope and scale on ACO performance, we then use econometric estimation approaches that account for endogeneity in collaboration structure decisions. ACO performance is measured by experiential quality and 30-day readmission rates. As additional tests, we also investigate our research questions by assembling data sets at both the hospital level (20,975 hospital-year panel data spanning 2009 to 2015) and patient level (859,145 Medicare patients admitted to 39 California hospitals over a four-year period from 2012 to 2015). Results: We find that synergies exist between partnership scope and scale with respect to ACO performance. Specifically, an average-sized ACO can realize 3.2% more improvement in experiential quality and a 6.6% greater reduction in 30-day readmission rates through partnership scope and scale synergies in the preacute care stage. We also show that the benefits of increasing partnership scope are consistent across providers and patient-level analysis. Further, we find that these benefits come at some cost, suggesting an initial cost-quality trade-off when developing collaboration structures. Managerial implications: Our results offer important insights into the healthcare operations management literature on designing effective healthcare delivery systems extending beyond a single organization.
Article
Quality management (QM) has become an all‐pervasive management philosophy, finding its way into most sectors of today’s business society. After the initial hype and enthusiasm, it is time to take stock of the knowledge accumulated in what is now a mature field of study and look for directions to take the field further forward. This article reflects on the mass of literature in the field, synthesizing, organizing and structuring knowledge and offering suggestions for future research. It reviews QM research organized along five main themes: the definition of QM, the definition of product quality, the impact of QM on firm performance, QM in the context of management theory and the implementation of QM. The article draws on these themes to reflect on three questions which are fundamental to re‐visit and re‐appraise QM: (i) What is QM? (ii) Is the set of practices associated with QM valid as a whole? (iii) How to implement QM in a real business setting?
Article
Recent research on total quality management (TQM) has examined the relationships between the practices of quality management and various levels of organizational performance. These studies have produced mixed results, probably because of the nature of the research designs used such as measuring TQM or performance as a single construct. Based on a comprehensive literature review, this study identifies the relationships among TQM practices and examines the direct and indirect effects of these practices on various performance levels. A proposed research model and hypotheses are tested by using cross‐sectional mail survey data collected from firms operating in the US. The test of the structural model supports the proposed hypotheses. The implications of the findings for researchers and practitioners are discussed and further research directions are offered.
Article
We used data from 45 production teams (626 individuals) and their supervisors to test hypotheses related to team structure. For teams engaged primarily in conceptual tasks, interdependence exhibited a boolean OR-shaped relationship with team performance, whereas team self-leadership exhibited a positive, linear relationship with performance. For teams engaged primarily in behavioral tasks, we found a boolean AND-shaped relationship between interdependence and performance and a negative, linear relationship between team self-leadership and performance. Intrateam process mediation was found for relationships with interdependence but not for relationships with team self-leadership. Overall, findings support a model of team structure and illustrate how relationships between structural characteristics and a team's performance can be moderated by its tasks.