Customize your JAMA Network experience by selecting one or more topics from the list below.
Bradley EH, Holmboe ES, Mattera JA, Roumanis SA, Radford MJ, Krumholz HM. A Qualitative Study of Increasing β-Blocker Use After Myocardial Infarction: Why Do Some Hospitals Succeed? JAMA. 2001;285(20):2604–2611. doi:10.1001/jama.285.20.2604
Author Affiliations: Departments of Epidemiology and Public Health (Drs Bradley and Krumholz), Medicine (Dr Holmboe), and Section of Cardiovascular Medicine, Department of Medicine (Drs Radford and Krumholz), Yale University School of Medicine; and Yale-New Haven Hospital Center for Outcomes Research and Evaluation (Drs Radford and Krumholz, Mss Mattera and Roumanis), New Haven, Conn.
Context Based on evidence that β-blockers can reduce mortality in patients
with acute myocardial infarction (AMI), many hospitals have initiated performance
improvement efforts to increase prescription of β-blockers at discharge.
Determination of the factors associated with such improvements may provide
guidance to hospitals that have been less successful in increasing β-blocker
Objectives To identify factors that may influence the success of improvement efforts
to increase β-blocker use after AMI and to develop a taxonomy for classifying
Design, Setting, and Participants Qualitative study in which data were gathered from in-depth interviews
conducted in March-June 2000 with 45 key physician, nursing, quality management,
and administrative participants at 8 US hospitals chosen to represent a range
of hospital sizes, geographic regions, and changes in β-blocker use rates
between October 1996 and September 1999.
Main Outcome Measures Initiatives, strategies, and approaches to improve care for patients
Results The interviews revealed 6 broad factors that characterized hospital-based
improvement efforts: goals of the efforts, administrative support, support
among clinicians, design and implementation of improvement initiatives, use
of data, and modifying variables. Hospitals with greater improvements in β-blocker
use over time demonstrated 4 characteristics not found in hospitals with less
or no improvement: shared goals for improvement, substantial administrative
support, strong physician leadership advocating β-blocker use, and use
of credible data feedback.
Conclusions This study provides a context for understanding efforts to improve care
in the hospital setting by describing a taxonomy for classifying and evaluating
such efforts. In addition, the study suggests possible elements of successful
efforts to increase β-blocker use for patients with AMI.
Randomized clinical trials have demonstrated the efficacy of β-blocker
use in reducing mortality and future cardiac events after acute myocardial
infarction (AMI).1 Based on substantial evidence,
clinical practice guidelines for AMI published by the American College of
Cardiology and the American Heart Association2
strongly recommend the use of β-blockers for secondary prevention after
Despite the evidence and the published guidelines, studies have repeatedly
demonstrated wide variation and underuse of β-blockers.3-13
The American Medical Association has reminded physicians of the importance
of β-blocker use after AMI,14 and both
the Health Care Financing Administration and the National Committee for Quality
Assurance have adopted β-blocker use after AMI as a quality-of-care indicator.
Physicians and hospitals thus have considerable motivation to increase the
use of β-blockers after AMI. Yet factors associated with successful improvement
efforts to increase their use over time in the hospital setting are poorly
Identifying factors associated with successful performance improvement
in the use of β-blockers after AMI can provide guidance to hospitals
that have been less successful in increasing their use over time. Also, the
lessons learned from improving the use of β-blockers may be relevant
to other efforts to improve clinical care and patient outcomes. The first
step toward understanding the strategies used by hospitals is to carefully
and systematically collect information about their approaches, with particular
attention to factors associated with success. Despite the substantial attention
given to changing physicians' practices15-19
and improving quality in clinical care,20-25
efforts to classify the essential characteristics of initiatives for improvement
in AMI care are limited, and few studies have systematically explored the
factors that are common to the more successful improvement efforts directed
at increasing β-blocker use after AMI.
To address this topic, we designed a qualitative study intended to provide
an in-depth perspective on the ways that hospitals are improving care for
patients with AMI. The objectives of this study were to develop a taxonomy
that can be used to classify and evaluate hospital-based performance improvement
efforts in the care of patients with AMI with particular focus on β-blocker
use, and to explore how essential factors varied among higher- and lower-performing
We undertook a qualitative study, based on open-ended interviews conducted
with clinical and administrative staff during hospital site visits from March
through June of 2000. The qualitative approach was chosen for 2 reasons. First,
with the exception of a recent randomized controlled trial,24
few studies have investigated factors specific to improvements either in the
care of patients with AMI or in β-blocker use. Qualitative research is
particularly well suited for exploratory studies for which previous literature
is limited.26 Such studies are useful for generating
hypotheses that can later be tested with quantitative data.26,27
Second, we anticipated that some factors, such as administrative philosophy
and physician leadership, may be multifaceted and difficult to measure. Again,
qualitative research provides a method for describing the diverse facets and
dimensions of such factors.26,28
As is standard in qualitative research,26,27
we chose sites using purposeful sampling to ensure that we included a diverse
set of hospitals. Study hospitals were selected to reflect a range of geographical
regions, hospital size, AMI volume, and improvement or decline in β-blocker
use over time. Hospitals were not selected based on knowledge of their approach
to measuring and improving care. Additional hospitals were selected and visited
until no new concepts were identified by the additional interviews, ie, until
the point of theoretical saturation. This occurred after the eighth hospital
site visit and 45 interviews. The research team was blinded to hospital rates
of β-blocker use until the completion of the data collection and coding.
The characteristics of the study hospitals are displayed in Table 1.
Eligible hospitals were those that participated in the National Registry
of Myocardial Infarction (NRMI)12 continuously
for at least 30 months during the 36-month (October 1996-September 1999) observation
period. In addition, eligible hospitals had to report at least 40 patients
with AMI on an annual basis during the study or observation period. The cut
point between the baseline and follow-up periods was chosen as the midpoint
between the time of the first and last patient submission to the NRMI during
the observation period. While NRMI does not include all hospitals in the country,
it does include a broad spectrum of facilities and is the largest ongoing
registry with detailed clinical and medication information. The chosen baseline
and follow-up periods reflect time before and after the dissemination of new
data demonstrating the effectiveness of β-blockers for patients with
AMI.11,14 The mean improvement
in β-blocker use from baseline to follow-up was 6 percentage points.
To ensure that the study hospitals reflected a range of β-blocker
use rates, we arrayed all eligible hospitals into 20 quantiles according to
their change in β-blocker use (mean [SD] β-blocker use rate of all
patients with AMI: 54.2% [14.1%] during baseline vs 60.0% [12.7%] during follow-up).
We randomly selected hospitals from the lowest 2 quantiles (representing declines
in β-blocker use rates ranging from −22 and −6 percentage
points), the middle 2 quantiles (representing increases in use rates ranging
from 5-7 percentage points), and the highest 2 quantiles (representing increases
in use rates from 19-35 percentage points). In 2 cases, randomly selected
hospitals did not meet other selection criteria (ie, they did not reflect
a range of sizes or geographical regions) and thus were replaced with other
randomly selected hospitals from the same quantiles.
In-depth, open-ended interviews27,29
were conducted in person with physician, nursing, quality management, and
administrative staff described by directors of quality assurance or quality
management as key staff involved with improving the care of patients with
AMI. Between 4 and 7 individuals considered to be key respondents were interviewed
at each hospital, for a total number of 45 key respondents interviewed. These
included 8 cardiologists, 4 internists, 2 emergency medicine physicians, 15
members of nursing staff, 11 members of quality management staff, and 5 members
of senior administrative staff. Interviews were generally conducted with a
single respondent, except when sites preferred to have multiple respondents
participating together. Interviews were each 1 to 1½ hours in length,
as is typical with in-depth interviewing.29
Interviews were generally conducted by at least 2 members of the research
team, which included investigators with diverse backgrounds in internal medicine
and cardiology, epidemiology, health administration, and nursing. All researchers
had substantial backgrounds and expertise in quality improvement. At least
2 interviewers were present at nearly all interviews; all interviews were
audiotaped and transcribed by independent professional transcriptionists.
Interviews were conducted using a standard interview guide, with probes
for clarification and additional detail. The interview guide began with the
standard "grand tour"27 question, "In the last
3 years, please describe the major initiatives your hospital has undertaken
to improve care of patients with AMI." Specific probes concerning initiatives
targeted at β-blocker use followed this question. For each initiative,
respondents were asked to describe specific instances of difficulty and of
success in implementing change in their hospital. In addition, respondents
were asked about their experiences related to data monitoring and feedback,
physician leadership, and sustaining change. For all areas of inquiry, respondents
were encouraged to illustrate their experiences with specific stories or vignettes.
Transcribed data were analyzed using common coding techniques for qualitative
data28,30 and the constant comparative
method of qualitative data analysis.27 Coding
of the data was accomplished in a series of iterative steps. An initial code
list was used to organize transcripts of the first 2 site visits and was then
refined during review and analysis of transcripts from subsequent site visits.
During its development, the code structure was reviewed 3 times by the full
research team for logic and breadth. The process of refining the code structure
included adding and reconstructing codes as new insights emerged and identifying
new relationships within units of a given category. A total of 72 specific
codes organized within several broad themes were ultimately developed and
served as the basis for final text review and organization of the transcript
Using this final version of the code structure, members of the research
team (E.H.B., J.A.M., S.A.R.) independently coded all transcripts, then met
as a group to code in several joint sessions, achieving consensus and assigning
codes to observations by a negotiated, group process. Coded data were entered
into a software package designed to handle unstructured qualitative data (NUD*IST
4, Sage Publications Software, Thousand Oaks, Calif) to assist in reporting
recurrent themes, links among the themes, and supporting quotations. Specific
analysis was conducted to identify distinctions in common themes between higher-performing
hospitals (those in which ≥65% vs those in which <65% of all patients
with AMI were prescribed β-blockers at discharge during the follow-up
Several techniques were used to ensure that data analysis was systematic
and verifiable, as commonly recommended by experts in qualitative research.28 These included consistent use of the discussion guide,
audiotaping and independent professional preparation of the transcripts, standardized
coding and analysis of the data, and the creation of an analysis audit trail
to document analytic decisions. The interviewers, the individuals who listened
to and transcribed the interview tapes, and those coding the transcript data
were blinded to the rates of β-blocker use among the study hospitals.
After review of the interview data, respondents' comments were organized
into 6 broad factors that formed the basis of the taxonomy for classifying
efforts to improve AMI care in general and β-blocker use specifically.
These broad factors were: (1) goals for improvement, (2) administrative support
for improvement efforts, (3) support among clinicians for such efforts, (4)
design and implementation of initiatives for performance improvement, (5)
use of data concerning β-blocker use, and (6) a set of contextual variables.
Each broad factor has a set of dimensions that further describe its meaning.
Table 2 summarizes the resulting taxonomy
including the 6 broad factors, their dimensions, and examples of concepts
underlying the dimensions. The
BOX lists direct quotations
to illustrate selected factors and related dimensions
(a longer list of illustrative quotations is available online as an
Respondents identified 4 distinct dimensions of goals, including goal
content, goal specificity, goal challenge, and the degree to which goals were
shared by staff throughout the organization. Improvement of patient care was
the most commonly mentioned goal underlying performance improvement efforts,
although maintaining financial position and improving hospital or professional
reputation were also described in several hospitals. A range of specificity
of goals was also described. Some hospitals stated specific goals, such as
"our target is 80% β-blocker use"; others, less specific, stated that
"the goal was to get β-blocker use within an acceptable range." Goal
challenge varied from what was described as a "zero-defect approach" (the
most challenging) to more lenient goals of having some portion of physicians,
rather than all physicians, use a clinical pathway. Finally, the degree to
which the goals for improvement were shared throughout the organization varied
substantially. For example, in some hospitals staff described widespread buy-in
and agreement with organizational goals for improvement. In contrast, respondents
in other hospitals described having goals for one's own clinical practice
but no overall organizational goals concerning AMI that were widely known
and agreed upon by clinical and administrative staff.
Support from the senior administration was viewed in several hospitals
as the most important factor in the success of performance improvement efforts
for β-blocker use. Administrative support encompassed not only the organizational
philosophy toward performance improvement but also the procurement of needed
resources to sustain improvement in care (eg, care coordinators, chart abstractors,
computer and analytical support, and quality improvement training). In some
hospitals, staff reported that senior administration was focused on and committed
to performance improvement. In other hospitals, administration was reported
to be uninvolved in or less supportive of improvement initiatives, and resources
devoted to quality improvement were inadequate.
Nearly all respondents emphasized that effective support from clinicians
was essential to the success or failure of initiatives to improve the use
of β-blockers or other treatments for AMI. Dimensions of this factor
included the types of clinician (ie, physician, nurse, ancillary support staff),
the degree of clinician engagement in the performance improvement effort,
and clinicians' ability to lead change. Several respondents also described
the importance of support from nursing and other clinical staff, although
physician leadership was perceived as a dominant success factor for enhancing β-blocker
A single physician who was leading AMI improvement efforts, often termed
a physician "champion" by respondents, was identified in several of the hospitals.
Characteristics that enhanced physicians' effectiveness included being highly
respected as busy practitioners and expert clinicians, being committed to
promoting β-blocker use themselves, and having consensus-building skills
to resolve conflicting views of clinicians. Expertise and previous training
in quality improvement techniques generally were not viewed as essential for
physician effectiveness in this capacity.
The dimensions of performance improvement initiatives identified included
the type of initiative and the style of implementation. All hospitals described
a variety of initiatives to improve β-blocker use at discharge. These
included ongoing education (by internal or external experts or both) of physicians
about the benefits of β-blockers, development of pathways and/or protocols
that included β-blocker use, implementation of standing discharge orders
with β-blockers, hiring care coordinators to remind physicians to consider β-blockers
for specific patients or to document why β-blockers were contraindicated,
chart-based reminders to prompt β-blocker use, and providing data feedback
to physicians about their β-blocker use. Nearly all hospitals reported
that changes were slow to be implemented and that continued attention to β-blocker
use through education, reminders, and data feedback to physicians was necessary
to sustain high performance.
The implementation style of performance improvement efforts encompassed
several aspects of managing change. Primary aspects of implementation included
the degree to which participatory teams were used, whether the initiative
was aimed at improvement or fault finding, and methods used to promote or
ensure adherence to new standards.
A central factor described by respondents at all hospitals was the use
of data in improvement efforts. Data were reported to be critical in 2 ways.
First, availability and acceptance of the evidence based on credible research
regarding the benefits of β-blocker use were essential in the early stages
to attain physicians' commitment to change practice, in this case to increase
their use of β-blockers for patients with AMI. Respondents reported that
without credible empirical evidence to support the recommended practice, physician
behavior was unlikely to change. Hence, transmitting such evidence was an
important early step in effective improvement efforts.
Second, data feedback on physicians' use rates of β-blockers was
also reported to be essential for effective improvement efforts. However,
such data feedback efforts were described as successful only if the data were
perceived by physicians to be valid, reflective of current practice, and benchmarked
against a reasonable comparison group. In some hospitals, this comparison
group was performance at similar hospitals; in others the comparison was to
national standards; in still others, physicians within a hospital were compared
with each other. In all cases, the credibility of the data was cited as a
critical ingredient in making data feedback effective.
Respondents described a set of variables that characterized the context
within which quality improvement efforts occurred. These included hospital
size, health system affiliation, ownership type, financial and market constraints,
and organizational turbulence. Staff in smaller hospitals described interdepartmental
communication as easier due to smaller size. Those affiliated with health
systems described the ability to benchmark performance data against other
hospitals in the system as an advantage. Several respondents in markets they
perceived to be highly competitive noted that the competition had encouraged
performance improvement efforts. In contrast, some staff in nonprofit settings
described their nonprofit status as limiting the pressure to improve performance.
Finally, organizational turbulence (eg, turnover of senior administrative
or clinical staff, large financial losses, recent unionization) was described
as slowing quality improvement efforts due to necessary focus on issues that
might jeopardize the organization's stability or survival.
Several of the factors and their related dimensions from the taxonomy
differed between the higher-performing hospitals (those in which ≥65% of
patients with AMI received β-blockers at discharge) and the lower-performing
hospitals (those in which <65% of patients with AMI received β-blockers
at discharge), although statistical associations could not be tested based
on the study design. Nevertheless, findings from these qualitative data may
help frame hypotheses to be formally tested in future quantitative studies.
Four characteristics were apparent in the higher-performing hospitals
and were absent in the lower-performing hospitals: (1) high degree of goal
sharedness, (2) substantial level of administrative support, (3) strong physician
leadership, and (4) high-quality data feedback.
First, higher-performing hospitals described performance goals as shared
and agreed upon by staff throughout the organization. Higher-performing hospitals
did not all have explicitly challenging goals, but all had substantial goal
agreement. In the case of β-blocker use, higher-performing hospitals
described substantial buy-in among clinical staff that use should increase.
In contrast, ambivalence toward use of β-blockers or ambiguity in terms
of whether that goal was a priority was apparent in lower-performing hospitals.
In the lowest-performing hospital, physician leaders described explicit disagreement
that increased β-blocker use improved AMI outcomes.
Second, higher-performing hospitals reported extensive administrative
support. This support was manifest in the chief executive officers and governing
boards requesting performance data, senior administrators participating in
quality improvement teams, and provision of staff and technical resources
to implement improvement initiatives. Respondents in lower-performing hospitals
did not describe such administrative support.
A third distinction was in the area of physician leadership. Higher-performing
hospitals described the presence of a physician leader who was committed to
increasing β-blocker use among his or her peers. In lower-performing
hospitals, physician participants in improvement efforts were described as
weak or nonexistent.
The final distinction was in the ongoing monitoring and feedback of
valid data concerning current practice. Higher-performing hospitals reported
their data as credible to physicians; lower-performing hospitals either did
not routinely report current data on β-blocker use to physician staff
or reported it in ways that were perceived as not helpful, eg, reporting data
on patients from a year previous or reporting data that did not adequately
account for contraindications.
Equally important as the ways in which the higher- and lower-performing
hospitals differed are the ways in which they appeared similar. Somewhat unexpectedly,
the ways in which the higher- and lower-performing hospitals were similar
included the type of β-blocker initiatives and the style of the implementation
of such initiatives.
First, both higher- and lower-performing hospitals described a myriad
of quality-improvement efforts in the care of patients with AMI. Contrary
to our expectations, higher-performing hospitals did not describe innovative
quality initiatives that were not found in the lower-performing hospitals.
The technical and programmatic innovations did not differ substantially among
the higher- and lower-performing hospitals.
Second, nearly all hospitals described an implementation process that
included the use of multidisciplinary teams and participatory management techniques.
Furthermore, both higher- and lower-performing hospitals described change
as taking place slowly. Both higher- and lower-performers reported difficulty
in sustaining improvements already made, regardless of the initial enthusiasm
to improve. A cycle of change, characterized by substantial inertia in the
beginning, a learning curve, accelerated improvement, a plateau, and then
either decline or maintenance, was commonplace. Although the shape and timing
of this trajectory varied across hospitals, it did not distinguish higher-
vs lower-performing hospitals. Even the higher-performing hospitals experienced
inertia and plateaus over time.
These findings suggest a taxonomy of factors and their dimensions that
may influence the success of efforts to increase β-blocker use. Previous
studies have measured improvement efforts by the level of involvement of physicians
and senior administrative staff,22,25,31
the training of staff in quality improvement techniques,23,32
and support for total quality management in general in the organization.21,33 Our taxonomy suggests that these
measures may have omitted key factors that characterize improvement efforts.
Previous evidence that has failed to demonstrate a significant association
between quality improvement and performance may be attributed to imprecise
characterizations of the improvement efforts.
In addition to proposing the taxonomy for classifying improvement efforts,
we propose 4 factors that may distinguish the higher-performing hospitals
in β-blocker use: the presence of shared goals for improvement, administrative
support, physician leadership, and credible data feedback. The first factor
and its related dimensions are consistent with previous research in clinical
medicine34 and organizational theory35 that identifies these as important elements in performance
improvement. However, the taxonomy expands and extends previous research on
total quality management and continuous quality improvement in important ways.
First, the findings highlight physician leadership, rather than merely
physician participation. Although this is consistent with the 1 randomized
controlled trial assessing the role of physician opinion leaders in the care
of patients with AMI,24 most have focused on
rather than physician leadership and may therefore have neglected the more
central aspect of physician involvement.
Second, the types of initiatives undertaken to increase β-blocker
use were notably similar across the study hospitals, and the type of initiatives
did not appear to distinguish higher- vs lower-performing hospitals. This
was unexpected given the considerable literature focusing on systems and process
changes for improving performance in the clinical setting.20-23,34
This finding has several alternative interpretations. Hospital staff
may have reported efforts that were not actually occurring at the hospitals,
due to pressures to appear progressive in this area. Nevertheless, the use
of in-depth, qualitative interviewing of multiple respondents in each setting
limits respondents' tendencies to overstate the truth. Another interpretation
is that the hospitals in this study were limited in the scope of improvement
efforts they conducted, and thus the efforts reported did not distinguish
the more successful hospitals from the less successful ones. Interviewing
more hospitals might reveal innovative systems changes not found in this study;
however, our study did include hospitals that had high β-blocker use
rates for patients with AMI and thus might be expected to have among the more
sophisticated approaches to achieving those rates. Finally, the finding may
suggest that the presence of quality-improvement initiatives and skilled approaches
to implementation is not sufficient to guarantee increased use of β-blockers.
Additional factors identified in our taxonomy, such as shared goals, administrative
support, physician leadership, and effective data feedback, may be necessary
to effect increased β-blocker use. This observation raises the question
of whether the roles of factors in the taxonomy vary depending on the targeted
area of clinical practice. In areas marked by physician judgment and clinical
uncertainty, physician leadership may be especially important. In other areas
marked by complex workflow processes involving multiple departments (eg, time
to primary angioplasty), system redesign initiatives may be more important.
The study has several important limitations. First, the study was conducted
among a limited set of hospitals, purposefully selected to reflect a range
of performance levels, geographical regions, and organizational sizes, and
the findings may differ in other settings. Second, self-reporting of performance
improvement efforts may have been influenced by respondents' knowledge of
their β-blocker use rates. When asked about their rates, however, respondents
commonly were unable to report change in β-blocker rates at their hospital.
Third, a careful look at Table 1
reveals that the 2 hospitals with the highest baseline use of β-blockers
experienced the greatest decline in use, and regression to the mean may have
played a role in this change. Our interviews revealed an absence of the factors
identified as important in higher-performing hospitals, but no other distinguishing
features. The next step in this research is to identify the factors associated
with such improvement or decline in hospital performance using a larger, randomly
selected sample of hospitals.
Finally, this qualitative study was designed to generate, not to test,
hypotheses. Future research based on larger samples of hospitals treating
patients with AMI should be conducted to test these hypotheses. Such research
will require adequate measurement of complex factors such as administrative
support and physician leadership.
The application of rigorous qualitative methods has provided an in-depth
view of a complex area of clinical care that has not been adequately explored
with traditional methods. The interventions to improve care and the context
in which they occur are complex and not easily understood by focusing on a
single facet of the process. Our taxonomy can promote more comprehensive and
sensitive measurement of quality-improvement efforts, so that their role in
clinical practice can be adequately assessed. Moreover, the identification
of essential elements of improvement efforts can help clinicians and researchers
plan effective interventions, focusing on areas that may have been previously
overlooked. This study is an important, though early, step in developing an
evidence-based approach to improving the translation of research into practice.
Create a personal account or sign in to: