This is the accessible text file for GAO report number GAO-11-445 
entitled 'Value in Health Care: Key Information for Policymakers to 
Assess Efforts to Improve Quality While Reducing Costs' which was 
released on July 26, 2011. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as 
part of a longer term project to improve GAO products' accessibility. 
Every attempt has been made to maintain the structural and data 
integrity of the original printed product. Accessibility features, 
such as text descriptions of tables, consecutively numbered footnotes 
placed at the end of the file, and the text of agency comment letters, 
are provided but may not exactly duplicate the presentation or format 
of the printed version. The portable document format (PDF) file is an 
exact electronic replica of the printed version. We welcome your 
feedback. Please E-mail your comments regarding the contents or 
accessibility features of this document to Webmaster@gao.gov. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

United States Government Accountability Office: 
GAO: 

Report to Congressional Requesters: 

July 2011: 

Value in Health Care: 

Key Information for Policymakers to Assess Efforts to Improve Quality 
While Reducing Costs: 

GAO-11-445: 

GAO Highlights: 

Highlights of GAO-11-445, a report to congressional requesters. 

Why GAO Did This Study: 

The U.S. has devoted an increasing proportion of its economy and 
federal budget to the provision of health care services, but high 
levels of spending do not guarantee good care. Policymakers, health 
practitioners, and others have implemented numerous health care 
interventions that make discrete changes in the organization of health 
care services in order to enhance the value of health care—that is, 
improve the quality of care while reducing costs. Examples include 
programs to reduce bloodstream infections and to coordinate patient 
care following hospital discharges. 

This report (1) examines the availability of evidence on the effect of 
selected interventions on quality of care and costs; (2) identifies 
key dimensions for assessing the strength of such evidence; and (3) 
examines factors that can facilitate the implementation and 
replication of health care interventions. GAO identified a broad and 
diverse set of health care interventions using published and 
unpublished sources. For 127 of those interventions, GAO analyzed 
responses to a questionnaire that it sent to persons knowledgeable 
about available information on the effect of that particular 
intervention on quality of care and costs. GAO’s questionnaire also 
asked respondents to assess the relative importance of seven factors 
in the implementation and potential replication of the health care 
intervention. In addition, GAO consulted the methodological literature 
and experts on assessing evidence on the effects of health care 
interventions. 

What GAO Found: 

About half of the respondents to our questionnaire reported some 
information on the effect of an intervention on both quality of care 
and costs—the two types of data needed to determine whether or to what 
extent a particular intervention enhanced the value of health care. 
Overall, the vast majority of our respondents reported at least some 
information on the observed effect of the intervention on quality of 
care. Relatively fewer—though still over half—of our respondents 
reported at least some information on the effect of the intervention 
on costs. 

Whether or not policymakers can rely on information that indicates an 
intervention enhances value depends on the strength of the underlying 
evidence about quality and cost effects. From studies on the effect of 
health care interventions on quality of care and costs, policymakers 
and others can assess the strength and limitations of available 
evidence along three dimensions. One, the credibility of evidence on 
the effect of health care interventions on quality of care and costs 
depends primarily on whether those studies apply rigorous study 
designs. Two, the applicability of the results of studies to a broader 
population depends on the extent to which the study population is 
representative of that larger population. Finally, the capacity of 
health care interventions for widespread replication can be examined 
in terms of the consistency of results obtained by each intervention 
across diverse health care organizational contexts. 

Respondents reported, generally by large margins, that leadership 
support as well as other factors, such as organizational culture and 
staff resources, significantly facilitated implementation. However, 
respondents were more divided when asked about the reported effect 
that health IT had on implementation, and most respondents reported 
that financial incentives were not a factor in the implementation of 
the intervention. A majority of respondents reported that each of 
these factors, with the exception of financial incentives, would be 
either very or somewhat important if one were to attempt to replicate 
the intervention as widely as possible. 

Progress in achieving greater value in the U.S. health care system 
will depend, in part, on the availability of information regarding the 
effect of interventions on quality of care and costs and on how 
policymakers and others assess and use that information. Information 
can guide the choices of policymakers among multiple interventions 
vying for support, but those decisions will have a sounder basis if 
the information meets certain criteria regarding its content and 
strength of evidence. At least some information on both cost and 
quality effects was available for about half of the interventions GAO 
examined. However, for many interventions the credibility of this 
information was put into question by widespread reliance on studies 
that did not incorporate rigorous designs that could isolate the 
effect of an intervention from other factors. 

We requested comments from the Department of Health and Human 
Services, but none were provided. 

View [hyperlink, http://www.gao.gov/products/GAO-11-445] or key 
components. For more information, contact James Cosgrove at (202) 512-
7114 or cosgrovej@gao.gov. 

[End of section] 

Contents: 

Letter: 

Background: 

Respondents Reported Basic Information Needed to Assess Value 
Available for About Half of Selected Interventions: 

The Strength of Evidence on the Effect of Interventions Can Be 
Assessed along Three Dimensions: 

Leadership Support and Other Factors Reported As Important for Both 
Implementation and Replication of Interventions: 

Concluding Observations: 

Agency Comments: 

Appendix I: Scope and Methodology: 

Appendix II: Types of Health Care Interventions That Seek to Improve 
the Value of Health Care: 

Appendix III: What Makes Some Study Designs More Rigorous Than Others: 

Appendix IV: Key Questions for Assessing Evidence from Studies of 
Interventions That Seek to Enhance Value: 

Appendix V: GAO Contact and Staff Acknowledgments: 

Tables: 

Table 1: Type of Information Reported on Quality of Care and Cost 
Savings for Selected Interventions: 

Table 2: Frequency of Types of Quality Measures Used to Assess 
Selected Health Care Interventions: 

Table 3: Characteristics Distinguishing Rigorous and Weak Study Design 
Types: 

Table 4: Reported Effect of Identified Factors on Implementation of 
Selected Health Care Interventions: 

Table 5: Expected Degree of Importance of Identified Factors for 
Widespread Replication of Selected Health Care Interventions: 

Table 6: To Assess the Credibility of Attributing Observed Changes in 
Quality of Care and Costs to the Intervention: 

Table 7: To Assess the Applicability of Study Results for Broader 
Populations of Interest: 

Table 8: To Assess an Intervention's Capacity for Widespread 
Replication: 

Figure: 

Figure 1: Number of Different Types of Quality Measures Used to Assess 
the Effects of Selected Interventions: 

Abbreviations: 

AHRQ: Agency for Healthcare Research and Quality: 

CEA: Cost-Effectiveness Analysis: 

EHC: Effective Health Care Program: 

EPOC: Effective Practice and Organisation of Care Group: 

GRADE: Grading of Recommendations Assessment, Development and 
Evaluation: 

GDP: gross domestic product: 

HCIE: Health Care Innovation Exchange: 

IT: information technology: 

PPACA: Patient Protection and Affordable Care Act: 

RCT: randomized controlled trial: 

[End of section] 

United States Government Accountability Office: 
Washington, DC 20548: 

July 26, 2011: 

The Honorable Kent Conrad: 
Chairman: 
Committee on the Budget: 
United States Senate: 

The Honorable Sheldon Whitehouse: 
United States Senate: 

For many years the United States has devoted an increasing proportion 
of its gross domestic product (GDP) and federal budget to the 
provision of health care services. National health expenditures rose 
from 7 percent of GDP--$308 billion--in 1970 to 16 percent--$2.2 
trillion--in 2008. Over this same period, the proportion of federal 
budget outlays devoted to health care increased even more rapidly from 
10.5 to 32.6 percent.[Footnote 1] Unless this trend is reversed, 
spending on health care will consume an escalating share of federal 
resources, leaving fewer and fewer dollars for other national 
priorities. 

Furthermore, high levels of spending do not guarantee good quality of 
care. While resources are needed to provide quality care, spending 
more to increase the number or technical complexity of treatments 
provided does not always lead to a corresponding increase in the 
quality of care.[Footnote 2] At the same time, studies have documented 
that many U.S. patients receive care of inconsistent quality as 
measured in terms of adherence to recognized standards of practice and 
in terms of clinical outcomes.[Footnote 3] 

Faced with these challenges, policymakers, health practitioners, and 
others have looked for ways to enhance the value of our health care by 
improving the quality of that care and at the same time reducing 
costs.[Footnote 4] To promote greater value across a range of health 
care settings, they have implemented numerous interventions that make 
discrete changes in who delivers health care services, how care is 
organized, or where care is delivered for a specified population. Some 
of these interventions are designed to restructure the process of 
health care in ways that guide the behavior of clinicians in fairly 
defined ways; for example, hospitals have implemented checklists in 
their intensive care units to help reduce the incidence of bloodstream 
infections by ensuring more consistent compliance with recommended 
procedures. Other interventions focus on restructuring how providers 
are paid to encourage them to produce greater value without trying to 
specify what steps they should take to achieve that objective. For 
example, payments have been structured so that providers earn more if 
their patients experience high quality care while overall costs are 
held in check. Still other interventions are designed to motivate or 
assist patients to take actions that will enhance their own health and 
thereby reduce their need for health care services, ranging from self- 
management of chronic conditions such as diabetes or heart failure to 
maintenance of recommended drug regimens.[Footnote 5] The 2010 Patient 
Protection and Affordable Care Act (PPACA) includes multiple 
provisions that support one or more of these approaches to enhance the 
value of health care.[Footnote 6] 

The wide array of different interventions gives policymakers the 
opportunity to help improve the value of U.S health care by supporting 
those interventions for which there is good evidence that they improve 
quality and reduce costs. To identify these interventions, 
policymakers need information on the effects of interventions on both 
quality of care and costs, and they need that information to be 
credible. It is therefore important for policymakers to be able to 
weigh the strengths and limitations of the evidence that an 
intervention, on net, has led to positive changes in quality of care 
and costs. Finally, interventions may vary in their potential for 
replication; that is, their effects on quality of care and costs may 
differ substantially among the organizations--such as hospitals and 
physician practices--that attempt to implement them. Therefore, 
assessments of the evidence for different health care interventions 
will be more complete if policymakers consider the interventions' 
effects on quality of care and costs across different contexts. 
Policymakers may also find it useful to know what factors may inhibit 
or facilitate the implementation and replication of interventions 
across varied organizational contexts. 

To assist policymakers in their efforts to enhance the value of health 
care, you requested that we provide you with information about health 
care interventions. This report (1) examines the availability of 
information on the effect of selected health care interventions on 
quality of care and costs; (2) identifies key dimensions for assessing 
the strengths and limitations of available evidence on the effect of 
interventions on quality of care and costs; and (3) examines factors 
that can facilitate the implementation and replication of health care 
interventions. 

To examine the availability of information on the effect of selected 
health care interventions on quality of care and costs, we drew on 
multiple sources to identify a broad and diverse set of interventions 
that related to value in health care. In addition to an extensive 
literature review, we conducted a comparable examination of other, 
nonbibliographic sources including a database on quality improvement 
initiatives maintained by the Agency for Healthcare Research and 
Quality (AHRQ) and materials from presentations at research 
conferences. The latter two sources allowed us to include 
interventions that had not yet been described in academic or 
professional literature. From these various sources we selected for 
review a set of 239 interventions that appeared to meet all of the 
following criteria: (1) the intervention had implemented a discrete 
change in the organizational structure or process of health care 
delivery; (2) available descriptions of the intervention suggested 
that it both improved quality and reduced health care costs (or held 
one constant while improving the other); and (3) the intervention 
addressed issues relevant to the U.S. health care system.[Footnote 7] 
(See appendix I for a more complete description of the sources and 
methods we used to identify interventions.) 

Because the documentation that we obtained on these interventions 
varied widely in focus, substantive content, and date of issue, we 
chose to collect detailed information from individuals with expert 
knowledge of each intervention using a standardized data collection 
instrument. Thus, we developed a Web-based questionnaire that we sent 
to researchers who were primary authors of articles identified in our 
literature search or prepared other materials describing these 
interventions and their results. The questionnaire included a mix of 
open-ended and closed-ended questions that examined what information 
was available on the effect of the intervention on quality of care and 
costs as well as what factors had facilitated or impeded 
implementation and replication of the intervention. With respect to 
quality of care, we asked respondents to describe up to five key 
measures that were used to assess the effect of their intervention on 
quality of care and the magnitude of change observed in those measures 
relative to a baseline or a control group. Regarding costs, we asked 
respondents to report the amount of any savings attributable to the 
intervention as well as the methods and information used to calculate 
those savings. 

To identify key dimensions for assessing the strengths and limitations 
of available evidence on the effect of interventions on quality of 
care and costs, we reviewed the relevant methodological literature on 
conducting systematic reviews and evaluations of health care 
interventions. We also consulted with several subject matter experts 
and obtained their reaction to the set of criteria that we identified 
through that review to help policymakers critically assess the 
information presented to them on value-enhancing interventions. It was 
beyond the scope of this engagement to apply this set of criteria to 
individual interventions. However, drawing on the documentation 
collected in identifying interventions related to value, we were able 
to categorize the types of study designs employed by studies that 
examined the interventions for which we received responses to our 
questionnaire. 

With respect to factors that facilitated or impeded implementation and 
replication of the intervention, we identified from relevant 
literature seven factors that have been found to facilitate or impede 
efforts to change the organizational structure and process of health 
care delivery: leadership support, organizational culture, staff 
resources, health information technology (IT), availability of tools 
and activities to standardize care, financial resources, and financial 
incentives. We asked our questionnaire respondents to assess the 
extent to which each of these seven factors facilitated or impeded the 
implementation of their intervention and the expected degree of 
importance that each of these factors could have on attempts to 
replicate the intervention to the widest scale possible.[Footnote 8] 
To gain contextual understanding of how these factors affect 
implementation and replication, we asked respondents to provide a 
narrative description of how each factor facilitated or impeded 
implementation of the intervention and why these factors would be 
important for widespread replication, respectively. (See appendix I 
for a more extensive description of the information collected through 
our questionnaire.) 

We received usable responses for 127 of the 239 selected 
interventions. These 127 interventions applied a broad range of value-
enhancing strategies in different health care settings and among 
diverse patient populations, including enhanced management of patients 
with chronic conditions and coordination of care across multiple 
providers (see appendix II for a complete list of the different types 
of interventions included). Although our efforts to identify relevant 
interventions were extensive, we could not ensure that every 
intervention meeting our selection criteria had been identified. As a 
result, our findings are limited in scope to the interventions for 
which we received completed questionnaires and cannot be generalized 
to all value-enhancing health care interventions. 

We conducted this performance audit from April 2009 to July 2011, in 
accordance with generally accepted government auditing standards. 
Those standards require that we plan and perform the audit to obtain 
sufficient, appropriate evidence to provide a reasonable basis for our 
findings and conclusions based on our audit objectives. We believe 
that the evidence obtained provides a reasonable basis for our 
findings and conclusions based on our audit objectives. 

Background: 

The methodological literature provides insight into conducting 
systematic assessments of evidence for health care interventions that 
change the delivery or structure of care. Furthermore, the literature 
on organizational change is pertinent to understanding the key factors 
that can facilitate or impede implementation and replication of such 
health care interventions. 

Systematic Assessments: 

Applied social science research has developed a core set of 
methodological questions and approaches for assessing the effect of 
programs or other interventions on a wide range of organized 
activities.[Footnote 9] They address two key issues: how best to 
determine the independent effect of a program or intervention and how 
best to generalize from the results obtained from one or more studies 
to broader populations of interest.[Footnote 10] A number of 
organizations have developed more specialized guidance for applying 
these general methodological principles to health care interventions. 
For example, the Effective Practice and Organisation of Care Group 
(EPOC) is a component of the Cochrane Collaboration--an international 
network of individuals who analyze the effect of health care 
interventions--which focuses on interventions that change the practice 
of care and the delivery of health care services. EPOC provides 
guidance to researchers on how best to prepare systematic reviews of 
such interventions in order to synthesize the information available in 
multiple studies.[Footnote 11] AHRQ's Effective Health Care Program 
(EHC) and the Grading of Recommendations Assessment, Development and 
Evaluation (GRADE) working group have developed similar guidance, 
though both of these efforts focus more on assessing alternative 
medical treatments rather than alternative approaches for organizing 
health care services. 

Key Factors That Contribute to Organizational Change: 

Research on organizational change has identified certain factors as 
key contributors to successful implementation of health care 
interventions for quality improvement.[Footnote 12] For example, the 
literature consistently cites leadership support as essential for 
successful implementation of quality-of-care interventions within 
health care settings. Further, it describes the role of leaders in 
promoting the adoption of interventions by their organizations, in 
winning acceptance among affected staff members for the changes those 
interventions entail, and in marshaling sufficient resources for the 
intervention to succeed. In addition, this literature has shown that 
organizations vary in their attitudes, beliefs, and values, and that 
this "organizational culture" can either promote or inhibit change. 
Organizations tend to achieve quality of care improvement more readily 
if they have a culture with such characteristics as receptiveness to 
change, placing high value on ensuring the quality of care provided, 
and prizing innovation with a willingness to take risks. The 
literature also cites the role of infrastructure factors, such as the 
sufficiency and appropriateness of staff resources and the adequacy of 
existing health information technology (health IT) systems, in the 
successful implementation of quality improvement interventions. 
Another factor cited in the literature is the availability of 
previously developed tools and procedures for standardizing health 
care processes--such as checklists or guidelines--as well as other 
types of technical assistance that can facilitate the implementation 
of a given intervention. Additionally, the literature has pointed to 
financial factors that affect the implementation of interventions for 
quality improvement, including both the level of financial resources 
needed to sustain an intervention and the use of financial incentives 
to promote quality enhancement activities. Financial incentives 
represent a particular application of financial resources that involve 
the contractual or other provisions that determine how much health 
care providers are paid and for what. Such financial incentives affect 
who benefits from and who pays for the cost of an intervention. This 
in turn can facilitate or impede the implementation and replication of 
interventions. 

Respondents Reported Basic Information Needed to Assess Value 
Available for About Half of Selected Interventions: 

About half of the respondents to our questionnaire reported basic 
information on the effect of their intervention on both quality of 
care and costs--the two types of data needed to determine whether or 
to what extent a particular intervention enhanced the value of health 
care. Overall, the vast majority of our respondents reported at least 
some information on the observed effect of their intervention on 
quality of care. Relatively fewer--though still over half--of our 
respondents reported at least some information on the effect of their 
intervention on costs. 

About Half of Respondents Reported Basic Information on the Effect of 
Selected Interventions on Both Quality of Care and Costs: 

The ability of policymakers to identify interventions that 
substantially improve quality and reduce costs depends on the 
availability of basic information on the size of the effect of an 
intervention on both quality of care and costs. These are the two 
types of data needed to determine whether or to what extent a 
particular intervention enhanced the value of health care. Just over 
half of the respondents to our questionnaire reported such basic 
information on their interventions. Sixty-four of 127 respondents 
reported information on both improvements observed in at least one 
quality measure and a specific amount of cost savings (see table 1). 
For the remaining interventions, the missing information most often 
concerned the effect of the intervention on costs. Furthermore, even 
fewer respondents, 45, reported improvements observed in at least one 
quality measure and a specific amount of cost savings that accounted 
for the costs of implementing their intervention--net cost savings. 

Table 1: Type of Information Reported on Quality of Care and Cost 
Savings for Selected Interventions: 

Type of information reported: Reported observed improvements in at 
least one quality measure; 
Number of Interventions: 114. 

Type of information reported: Reported a specific amount of cost 
savings; 
Number of Interventions: 72. 

Type of information reported: Reported observed improvements in at 
least one quality measure and a specific amount of cost savings; 
Number of Interventions: 64. 

Type of information reported: Reported observed improvements in at 
least one quality measure and a specific amount of net cost savings; 
Number of Interventions: 45. 

Source: GAO analysis of responses to GAO questionnaire. 

Note: Based on responses for all 127 interventions in our analysis. 
Respondents who reported a specific amount of net cost savings stated 
that this amount of reported savings took account of the costs of 
implementing their intervention. 

[End of table] 

Information on the Effect of Selected Interventions on Quality of Care 
Was Frequently Reported: 

Compared to information on both quality of care and costs, information 
on the effect of selected interventions on quality of care alone was 
more frequently reported. The vast majority of respondents to our 
questionnaire reported at least some information on the observed 
effect of their intervention on quality of care. Specifically, 114 of 
127 respondents reported improvements in one or more measures used to 
assess the effect of their intervention on various aspects of care 
quality.[Footnote 13] Of these, 112 respondents reported a specific 
magnitude of improvement observed in at least one quality measure in 
terms of a percentage change or other quantitative measurement. 
[Footnote 14] Additionally, 2 respondents reported improvement in at 
least one quality measure, but did not report a specific magnitude of 
improvement. In contrast, the remaining 13 respondents did not report 
sufficient information to determine whether their intervention had any 
effect on quality of care. Six of 127 respondents described one or 
more measures used to assess the effect of their intervention on 
different aspects of care quality, but did not report a magnitude of 
improvement observed in these measures. Seven respondents did not 
report any information on the measures used to assess the effect of 
their intervention on aspects of care quality. 

Respondents reported that the effect of their intervention on quality 
of care was assessed using a range of measures that generally fell 
into five broad types reflecting different aspects of care quality 
(see table 2).[Footnote 15] Respondents most frequently described one 
or more quality measures that were used to assess the effect of their 
intervention on outcomes resulting from care. Specifically, 82 
respondents reported that the effect of their intervention on quality 
of care was assessed using outcome measures such as patient mortality, 
the overall physical and emotional health of a patient, or the level 
of stress reported by a patient caregiver.[Footnote 16] In addition, 
56 respondents described one or more measures that assessed the effect 
of their intervention on the amount of health care services consumed. 
[Footnote 17] These measures included the length of hospital stay, the 
number of emergency department visits, and the number of hospital 
readmissions for a specified population. Forty-four respondents 
described measures that assessed the effect of their intervention on 
processes of care. Process-of-care measures assess the extent to which 
the care provided to a patient was appropriate based on current 
professional knowledge and the particular circumstances.[Footnote 18] 
For example, process-of-care measures could examine whether diabetes 
patients had received foot exams, eye exams, and regular glucose 
monitoring at specified intervals. Fewer respondents described 
measures that assessed quality in terms of the experience of a patient 
or caregiver or the structure of care.[Footnote 19] 

Table 2: Frequency of Types of Quality Measures Used to Assess 
Selected Health Care Interventions: 

Type of measure: Outcome; 
Examples of specific quality measures reported: 
* Mortality; 
* Quality of life; 
* Prevalence of hospital-acquired pressure ulcers; 
* Caregiver strain index or stress level; 
Number of interventions: 82. 

Type of measure: Use of services[A]; 
Examples of specific quality measures reported: 
* Proportion of patients (participating in the intervention) admitted 
to the hospital; 
* Length of hospital or ICU stay; 
* Duration of mechanical ventilation use; 
Number of interventions: 56. 

Type of measure: Process; 
Examples of specific quality measures reported: 
* Diabetes quality indicators such as number of patients who received 
a foot exam, eye exam, and glucose monitoring; 
* Patient adherence to prescribed medications; 
Number of interventions: 44. 

Type of measure: Experience; 
Examples of specific quality measures reported: 
* Patient satisfaction with care; 
* Patient and caregiver feeling of support or connection with provider; 
Number of interventions: 31. 

Type of measure: Structure; 
Examples of specific quality measures reported: 
* Physician, nurse, or other clinician satisfaction; 
* Physician knowledge of local guidelines of antibiotic prescribing 
practices; 
* Staff injury; 
Number of interventions: 27. 

Source: GAO analysis of responses to GAO questionnaire. 

Note: One hundred twenty of 127 respondents to our questionnaire 
reported one or more measures used to assess the effect of their 
intervention on different aspects of care quality. We categorized 
these quality measures into five broad types based on measure domains 
in AHRQ's National Quality Measures Clearinghouse. We excluded from 
this analysis eight measures reported by six respondents who did not 
clearly specify what aspect of care quality was assessed by those 
measures. Most respondents reported that the effect of their 
intervention was assessed using more than one type of quality measure. 

[A] AHRQ does not consider use of services measures to be direct 
measures of the quality of clinical care. 

[End of table] 

Although the information provided by any one type of quality measure 
is limited, most of our respondents reported that the effect of their 
intervention on quality of care was assessed using more than one type 
of quality measure. Each type of quality measure offers insight into a 
particular domain of quality such as outcomes of care, processes of 
care, or experience of care. Just 41 respondents reported that only 
one type of measure was used to assess the effect of their 
intervention on quality of care (see fig. 1). For most--79--of the 
interventions in our review, respondents reported that the effect of 
their intervention on quality of care was assessed using measures 
belonging to two or more different types of quality measures, thereby 
providing a broader perspective on the effect of the intervention on 
quality of care. 

Figure 1: Number of Different Types of Quality Measures Used to Assess 
the Effects of Selected Interventions: 

[Refer to PDF for image: illustration] 

One type of quality measure: 41 respondents; 
Two types of quality measures: 43 respondents; 
Three types of quality measures: 31 respondents; 
Four types of quality measures: 5 respondents; 
79 respondents reported two or more types of quality measures. 

Source: GAO analysis of responses to GAO questionnaire. 

Note: One hundred twenty of 127 respondents to our questionnaire 
reported that the effect of their intervention on quality of care was 
assessed using one or more of five different types of quality 
measures: outcome, use of services, process, experience, and 
structure. Another 7 respondents did not describe any measures used to 
assess the effect of their intervention on quality of care. 

[End of figure] 

Information on the Effect of Selected Interventions on Costs Was Less 
Frequently Reported: 

Somewhat fewer respondents to our questionnaire reported information 
on the effect of their interventions on costs than quality of care. 
Specifically, 72 of 127 respondents reported a specific amount of 
change in costs--cost savings.[Footnote 20] Respondents most 
frequently reported that costs were assessed by calculating the total 
dollars saved or the average dollars saved per person annually. 
Respondents less frequently reported that costs were assessed by 
calculating the financial return on investment, percentage change in 
total health care costs per patient, or an alternative cost metric 
such as dollars saved per member per month for patients participating 
in a certain health care plan.[Footnote 21] In contrast, the remaining 
55 respondents did not report sufficient information to determine 
whether their intervention had any effect on costs. Nine of 127 
respondents reported that costs were assessed, but did not report a 
specific amount of cost savings. Forty-five respondents reported that 
cost savings were not assessed, and one respondent did not report any 
information on whether cost savings were assessed. 

Most, but not all, of the respondents who reported a specific amount 
of cost savings stated that these cost savings accounted for the costs 
associated with implementing the intervention. Among the 72 
respondents who reported a specific amount of cost savings, 51 
respondents reported net cost savings that took account of 
implementation costs; another 20 respondents reported gross cost 
savings that did not take implementation costs into account.[Footnote 
22] When asked to provide additional detail on their implementation 
cost calculations, 35 respondents reported that the cost savings took 
account of both start-up costs associated with developing and 
initially implementing the intervention as well as ongoing costs 
associated with operating and maintaining the intervention over time. 
Two respondents reported that cost savings took account of start-up 
costs but not ongoing costs to maintain the intervention, and 19 
reported that cost savings took account of ongoing costs but not start-
up costs.[Footnote 23] 

The interventions we reviewed also varied in the extent to which the 
reported cost savings attributed to them were based on information 
directly related to the intervention. Forty-nine respondents reported 
that cost savings were calculated using only data that were collected 
specifically to assess the effect of their intervention on costs. In 
contrast, 26 respondents reported that cost savings were calculated 
using a mix of data that were collected specifically to assess the 
intervention and data from a secondary source such as published 
literature or a national database. For example, one respondent 
reported cost savings attributable to an intervention designed to 
improve patient self-management of asthma based on data that were 
collected on changes over time in the actual number of health care 
encounters for patients enrolled in the program and the estimated 
costs for those encounters derived from national averages for several 
types of health care services such as hospital days or emergency 
department visits. While data from secondary data sources may provide 
otherwise missing information needed to estimate the cost savings 
achieved by an intervention, the relevance of such secondary data to 
that particular intervention may be open to question, which makes the 
accuracy of the cost savings estimate more uncertain. 

The Strength of Evidence on the Effect of Interventions Can Be 
Assessed along Three Dimensions: 

Policymakers and others can assess the strength and limitations of 
available evidence from studies on the effect of health care 
interventions on quality of care and costs along three dimensions. 
One, the credibility of evidence on the effect of health care 
interventions on quality of care and costs depends primarily on 
whether those studies apply rigorous study designs. Two, the 
applicability of the results of studies to a broader population 
depends on the extent to which the study population is representative 
of that larger population. Finally, the capacity of health care 
interventions for widespread replication can be examined in terms of 
the consistency of results obtained by each intervention across 
diverse organizations. Appendix III provides a more detailed 
explanation of what makes some study-design types more rigorous than 
others and appendix IV presents a list of key questions that describe 
the information that policymakers can look for to assess the evidence 
provided by particular studies along these three dimensions. 

Examining the Credibility of Evidence on the Effects of Health Care 
Interventions: 

For policymakers and others, the benefit obtained from basic 
information on the effect of interventions on quality of care and 
costs depends in large part on the strength of that evidence. 
Information based on weak evidence can provide policymakers a 
misleading indication of an intervention's potential to enhance value. 
For example, the direction and magnitude of the changes in quality of 
care and cost reported for the 127 interventions examined through our 
questionnaire could deviate substantially from the actual impact of 
those interventions, depending on the characteristics of the studies 
that generated that reported information. To determine what 
information has the kind of evidentiary support that they can rely on, 
policymakers can assess the strengths and limitations of studies that 
examine health care interventions of interest along three broad 
dimensions. 

The first of these dimensions is the credibility of evidence that 
attributes any changes in quality of care and costs to those 
interventions. The methodological experts we consulted uniformly 
emphasized the primacy of study design in determining the credibility 
of evidence on the effect of health care interventions on quality of 
care and costs. Observed changes in quality of care and costs that one 
might attribute to a health care intervention may in fact be due in 
large part to the effect of a wide variety of other factors. The 
choice of study design type is critical because rigorous designs have 
the capacity to isolate the effects of a health care intervention from 
other factors that may affect changes in quality of care and costs. 

[Side bar: 
Credible Evidence Depends on Rigorous Study Designs: 

* A 2009 study assessing a surgical safety checklist using a simple 
pre/post design found the rate of surgical complications declined 36 
percent after the checklist was implemented. 

* Critics noted that the pre/post design did not control for potential 
confounding factors, such as other quality improvement initiatives 
occurring at the same time. 

* A 2010 study of a different surgical safety checklist used a 
controlled before and after design that compared a group of hospitals
that implemented the checklist to a separate control group of similar 
hospitals that did not implement the checklist. 

* Surgical complication rates in the control group hospitals did not 
change significantly, while they decreased 39 percent in the hospitals 
that implemented the checklist. That contrast provides a substantial 
basis for crediting the checklist for the decrease in complications as
opposed to any other factors. End of side bar] 

The methodological literature we reviewed identifies several different 
study design types that have sufficient rigor to isolate the effect of 
interventions on quality of care and costs. They include randomized 
controlled trials (RCTs), interrupted time series studies, and 
controlled before and after studies.[Footnote 24] RCTs and controlled 
before and after studies both use control groups--consisting of study 
participants who are not exposed to the intervention--to adjust for 
the effect of other factors besides the intervention.[Footnote 25] 
Interrupted times series studies do not use control groups; instead 
they rely on analyzing data collected at multiple time points both 
before and after an intervention is implemented to adjust for other 
factors. (See appendix III for more information on how these study 
design types isolate the effect of an intervention.) 

In contrast, according to the methodological literature we reviewed, 
some other types of study designs lack the capacity to isolate the 
effect of a health care intervention from that of other factors. For 
example, a simple pre/post study that assesses quality of care and 
costs once before an intervention is implemented and a second time 
after implementation of the intervention has no mechanism analogous to 
a control group to take account of the effect of other factors. The 
same is true for post-only studies that rely entirely on data 
collected after an intervention was implemented. With studies using 
these types of designs, there is no way to determine how much of the 
difference observed between the pre and post measurements, or among 
any groups following an intervention, was due to the intervention and 
not to other factors. Consequently, such studies will not provide 
policymakers credible information about the extent to which the 
intervention itself affected both quality of care and costs. 

Table 3 describes key distinguishing characteristics to help 
policymakers identify the type of study design employed in a study of 
an intervention.[Footnote 26] 

Table 3: Characteristics Distinguishing Rigorous and Weak Study Design 
Types: 

Relative design strength: Rigorous; 
Design type: Randomized controlled trial; 
Distinguishing characteristics: Outcomes compared for study 
participants randomly allocated to intervention and control groups. 

Relative design strength: Rigorous; 
Design type: Controlled before and after study; 
Distinguishing characteristics: Outcomes measured before and after 
implementation of intervention in nonrandomly selected intervention 
and control groups. 

Relative design strength: Rigorous; 
Design type: Interrupted time series study; 
Distinguishing characteristics: Trends in measured outcomes examined 
over many time points both before and after intervention implemented. 

Relative design strength: Weak; 
Design type: Pre/post study; 
Distinguishing characteristics: Outcomes measured only for those 
exposed to an intervention, once (or a few times) before and once (or 
a few times) after implementation of the intervention. 

Relative design strength: Weak; 
Design type: Post-only study; 
Distinguishing characteristics: Outcomes measured only after the 
intervention was implemented, with or without a control group. 

Source: GAO synthesis of methodological literature: 

Note: The three rigorous design types can be used in a wide variety of 
circumstances; there are other rigorous design types that apply in 
more specialized circumstances. 

[End of table] 

Among studies addressing the effect of health care interventions on 
quality of care and costs, a range of rigorous to weak design types 
are used. For example, among the 127 interventions for which we 
received responses to our questionnaire, we found 22 interventions 
with studies involving RCTs and another 11 interventions assessed 
using controlled before and after studies. However, for a 
substantially larger number of the 127 interventions, the studies we 
identified employed the types of study designs that do not isolate the 
effect of the intervention from other factors. Specifically, the 
results for 67 interventions were based on pre/post studies, and 
another 19 were based on post-only studies of one kind or another. In 
this one, diverse set of interventions that we reviewed, policymakers 
could find credible evidence based on rigorous study designs 
concerning the effects of certain interventions on quality of care and 
costs; however, for many other interventions such studies were lacking. 

In addition to study design, the methodological literature we reviewed 
emphasized the importance of how a study is conducted. Even rigorous 
study designs can lose their capacity to isolate the effect of an 
intervention on quality of care and costs if researchers do not adhere 
to the requirements of those designs. Thus, assessments of the 
strengths of study results should consider how well the study design 
was implemented. 

One component of a study's implementation that policymakers can 
examine involves the selection and management of control groups used 
in the study. In order to isolate the effects of an intervention, the 
control group has to be equivalent to the treatment group--except for 
the latter's exposure to the intervention. According to the 
methodological literature we reviewed, that equivalence can be 
compromised in a number of ways. In the case of RCTs, for example, 
allocation to treatment and control groups may not be truly random if 
there are flaws in the process for assigning study subjects to those 
groups. Moreover, for both RCTs and controlled before and after 
studies, losing a disproportionate number of study participants from 
either treatment or control groups can also undermine their 
equivalence.[Footnote 27] 

Another component of a study's implementation that policymakers can 
examine concerns the measures and procedures adopted for data 
collection. According to the methodological literature we reviewed, a 
study will produce stronger evidence when it employs measures that are 
recognized as valid and reliable.[Footnote 28] For example, central 
line-associated bloodstream infections can be tracked using a 
surveillance measure developed by the Centers for Disease Control 
(CDC) or with less labor-intensive measures that draw on 
administrative data. Clinicians consider the CDC measure to be the 
most valid and reliable measure for this type of infection because it 
calls for laboratory confirmation of identified infections and it 
accounts for varying risks of infection based on the number of days 
that a central line catheter is in place.[Footnote 29] In addition, 
the data for those measures should be collected at the same time and 
in the same way from all groups in the study. Any systematic 
inconsistencies in how data are collected for a study can skew the 
results.[Footnote 30] 

Examining the Applicability of Study Results to Broader Populations: 

If a study produces credible evidence that a health care intervention 
has a positive effect on both quality of care and costs within the 
population it examined, a second dimension that policymakers and 
others can assess concerns the scope of that effect--for what broader 
populations or groups are the results applicable? Applicability 
depends on the representativeness of the study population for a 
broader population of interest. The methodological literature 
identifies two different approaches for establishing 
representativeness: (1) randomly selecting the study population from a 
known universe, or (2) examining the degree to which a study 
population matches a given broader population on characteristics 
relevant to the intervention. The first approach, random selection, 
intrinsically makes the study population representative of the 
particular universe from which it was selected and the study results 
applicable to that population.[Footnote 31] 

[Side bar: 
Applicability of a Study Population with Nontypical Characteristics: 

* The Medicare Physician Group Practice (MPGP) Demonstration provided 
selected group practices bonus payments for meeting quality and cost 
constraint targets. 

* Physician practices were selected for the demonstration based in 
part on their having sophisticated information systems to track 
patient services and outcomes. 

* All 10 MPGP practices were very large, with more than 200 
physicians. However, fewer than 5 percent of all U.S. physicians
report practicing in groups of 50 or more. 

* An evaluation of the MPGP demonstration noted that its modestly 
positive outcomes were applicable to physicians with similar 
characteristics, i.e., those who belong to very large practices. 
Whether or not physicians belonging to smaller group practices would 
respond similarly to the same set of incentives cannot be determined 
from the MPGP demonstration results. End of side bar] 

The second approach for establishing representativeness--examining the 
extent of similarity between the study population and a broader 
population of interest--can be used by policymakers whenever the study 
population was not chosen randomly or the broader population of 
interest to policymakers is not the universe from which the study 
population was selected. Policymakers can assess the degree of 
similarity between the study population and a broader population 
through an examination that focuses on two issues: (1) identifying 
characteristics where the study population and broader population of 
interest differ and (2) assessing whether any differences found could 
influence the effect of the intervention on quality of care and costs 
(see appendix IV ).[Footnote 32] 

Major differences between a nonrandomly selected study population and 
a broader population of interest to policymakers should raise 
questions about the applicability of the study's results for that 
broader population. For example, an intervention to improve care 
coordination for patients with diabetes might be implemented and 
assessed in a few academic medical centers. In that situation, the 
representativeness of the study population for all patients with 
diabetes could come into question on at least two counts--the kind of 
care provided in an academic medical center might well differ from 
that usually provided by community-based providers and the patients 
treated by academic medical centers might have a higher level of 
severity than diabetics treated elsewhere. If patients in the study 
received a different overall set of services, that could affect the 
impact of the intervention on those patients even if the intervention 
itself were implemented the same way for the two populations. 
Similarly, an intervention could have a more pronounced effect on 
patients with a higher level of severity, or the intervention might 
work less well for such patients. Thus, to establish the applicability 
of the study results to a broader population of diabetic patients, 
studies of the intervention would need to provide evidence that the 
differences between the study population and the broader population of 
diabetics would not affect the performance of the intervention. 

Examining the Capacity of Interventions for Widespread Replication: 

A third dimension on which policymakers and others can assess the 
strength of evidence for health care interventions concerns the 
capacity of an intervention for replication across diverse 
organizations. Because organizations vary across the factors that 
affect the implementation of health care interventions, including 
leadership, organizational culture, and staff and financial resources, 
a particular intervention may work more or less well depending on the 
organizational environment in which it operates. As a result, some 
organizations may be more receptive to a particular value-enhancing 
intervention than others. That, in turn, can make it more difficult to 
take an intervention that proved successful in a small number of 
organizations and replicate it widely across many others. However, 
some interventions have produced positive results on quality of care 
and costs in a range of different organizations, which suggests that 
they may be less sensitive to varying circumstances across 
organizations. 

[Side bar: 
Inconsistent Results across Diverse Sites Indicates Restricted 
Capacity for Replication: 

* An intervention to facilitate prevention of pressure ulcers enrolled 
21 nursing homes in 4 states during 2006/2007. 

* All 21 nursing homes implemented the intervention’s core elements—
streamlining patient documentation forms with integration into an IT 
reporting system to promote accurate and timely monitoring of patient 
condition. 

* In aggregate, participating nursing homes achieved a 9 percent 
improvement in CMS’s quality measure for preventing pressure ulcers 
among high risk patients. 

* A subset of “high implementing” nursing homes experienced a 31 percent
improvement in the CMS quality measure while the remaining “non-high
implementing” nursing homes experienced a 12 percent decline in the 
same measure. 

* The intervention’s effectiveness was limited to the subset of 
nursing homes that researchers found had most fully involved staff at 
all levels in adopting the new workflow and technology. End of side 
bar] 

According to the methodological literature and experts that we 
consulted, certain information can provide the basis for an assessment 
of the consistency in an intervention's effects on quality of care and 
costs in different organizations. Specifically, this information 
concerns the number of different organizations where the intervention 
has been implemented, the degree of diversity exhibited by those 
organizations, and the consistency in observed changes in quality of 
care and costs across those organizations. However, such information 
would not be available for assessing the consistency of results across 
diverse organizations if an intervention has been implemented in only 
a few different organizations, or in multiple organizations that are 
generally quite similar. That is also the case if studies only analyze 
and report changes in quality of care and costs attributed to an 
intervention in the aggregate, rather than separately for the 
different organizations that implemented it. 

On the other hand, for interventions that have been implemented in 
multiple, diverse organizations, and their results analyzed separately 
at the different organizations, it is possible for policymakers to 
compare the results of the intervention across those organizations to 
examine the consistency of the intervention's effect. To the extent 
that those interventions consistently produce positive effects on 
quality of care and costs among diverse organizations, that provides 
evidence of their capacity for widespread replication. For other 
interventions, if data on the changes in quality of care and costs 
across the different organizations indicate a lack of consistency in 
outcomes, that provides evidence of a more restricted capacity for 
replication. 

Leadership Support and Other Factors Reported As Important for Both 
Implementation and Replication of Interventions: 

Respondents to our questionnaire reported, generally by large margins, 
that leadership support as well as other factors, such as 
organizational culture and staff resources, significantly facilitated 
implementation. However, respondents were more divided when asked 
about the reported effect that health IT had on implementation, and 
most respondents reported that financial incentives were not a factor 
in the implementation of their intervention. A majority of respondents 
reported that each of these factors, with the exception of financial 
incentives, was expected to be either very or somewhat important if 
one were to attempt to replicate their intervention as widely as 
possible. 

Most Respondents Reported Leadership Support and Other Factors 
Significantly Facilitated Implementation of Interventions: 

Taking account of factors that prior research has shown tend to 
facilitate or impede the implementation and replication of 
interventions may enhance efforts by policymakers and others to 
promote the adoption of interventions across varied organizational 
contexts. In examining the relative impact of seven factors identified 
in our literature review, we found that respondents to our 
questionnaire reported, generally by large margins, that five of the 
seven factors significantly facilitated implementation of their 
intervention. Health IT and financial incentives were the exceptions. 
Leadership support was the factor that the largest number of 
respondents reported as having significantly facilitated 
implementation of their intervention (see table 4). When asked to 
describe how leadership support facilitated implementation, 
respondents frequently explained that a leader who visibly prioritized 
and endorsed the intervention, allocated necessary resources, and 
championed the development and implementation of the intervention and 
drove necessary organizational or behavioral changes facilitated the 
implementation of the intervention. Respondents also explained that 
having champions, specifically clinicians, was a key factor in 
encouraging cooperation and participation in the intervention by 
staff, especially fellow clinicians. The prominent role attributed to 
leadership in implementing the many different types of interventions 
in our sample suggests that policymakers will have greater success in 
implementing and replicating interventions to the extent that they can 
take steps to ensure that strong leadership is in place before 
interventions are initiated. 

Table 4: Reported Effect of Identified Factors on Implementation of 
Selected Health Care Interventions: 

Number of respondents: 

Factors identified from relevant literature: Leadership support; 
Significantly facilitated: 92; 
Somewhat facilitated: 22; 
Somewhat impeded: 2; 
Significantly impeded: 3; 
Not a factor: 2; 
Don't know/No basis to judge: 4; 
Total: 125. 

Factors identified from relevant literature: Availability of tools and 
activities to standardize care; 
Significantly facilitated: 72; 
Somewhat facilitated: 33; 
Somewhat impeded: 3; 
Significantly impeded: 0; 
Not a factor: 11; 
Don't know/No basis to judge: 6; 
Total: 125. 

Factors identified from relevant literature: Staff resources; 
Significantly facilitated: 68; 
Somewhat facilitated: 29; 
Somewhat impeded: 14; 
Significantly impeded: 2; 
Not a factor: 10; 
Don't know/No basis to judge: 2; 
Total: 125. 

Factors identified from relevant literature: Organizational culture; 
Significantly facilitated: 60; 
Somewhat facilitated: 35; 
Somewhat impeded: 13; 
Significantly impeded: 6; 
Not a factor: 8; 
Don't know/No basis to judge: 3; 
Total: 125. 

Factors identified from relevant literature: Financial resources; 
Significantly facilitated: 50; 
Somewhat facilitated: 27; 
Somewhat impeded: 15; 
Significantly impeded: 5; 
Not a factor: 23; 
Don't know/No basis to judge: 5; 
Total: 125. 

Factors identified from relevant literature: Health IT; 
Significantly facilitated: 29; 
Somewhat facilitated: 30; 
Somewhat impeded: 14; 
Significantly impeded: 10; 
Not a factor: 36; 
Don't know/No basis to judge: 5; 
Total: 124. 

Factors identified from relevant literature: Financial incentives; 
Significantly facilitated: 20; 
Somewhat facilitated: 18; 
Somewhat impeded: 1; 
Significantly impeded: 2; 
Not a factor: 68; 
Don't know/No basis to judge: 15; 
Total: 124. 

Source: GAO analysis of responses to GAO questionnaire. 

Note: Table excludes missing responses. 

[End of table] 

Respondents typically reported that a combination of additional 
factors along with leadership support significantly facilitated 
implementation of their intervention. The 92 respondents who reported 
leadership support as having significantly facilitated implementation, 
reported, on average, another three factors as having significantly 
facilitated implementation. Of the 86 respondents who reported at 
least one factor in addition to leadership support as significantly 
facilitating implementation, more than half reported staff resources 
(60), organizational culture (55), and the availability of other tools 
(50), respectively, as having significantly facilitated 
implementation. Nearly half (42) reported that financial resources, in 
addition to leadership, significantly facilitated implementation. Just 
six respondents reported leadership support and no other factor as 
having significantly facilitated implementation. 

In contrast to the five factors that a clear majority of respondents 
reported having facilitated implementation of their intervention, 
respondents were more divided on how health IT affected 
implementation, as shown in table 4. Health IT had the highest number 
of respondents, compared to the other factors, that reported the 
factor impeded implementation of their intervention.[Footnote 33] 
Further, a substantial group of respondents reported that health IT 
was not a factor. On the other hand, nearly half of respondents 
reported that health IT either significantly or somewhat facilitated 
implementation of their intervention. Respondents frequently explained 
that health IT facilitated implementation of their intervention by 
enhancing the exchange of information and communication across 
providers or organizations, facilitating the collection of data or the 
evaluation of the intervention and improving the efficiency and 
productivity of staff. Of those who reported that health IT impeded 
implementation, respondents commonly cited the limited functional 
capacity of existing systems or the lack of interoperability across 
settings as impediments to successful implementation.[Footnote 34] 
Other respondents explained that the general lack of health IT 
altogether acted as a barrier that impeded implementation. Variation 
in the role of health IT across different types of interventions does 
not appear to explain the mixed assessment of this factor; as 
respondents for each of the intervention types included in our sample--
with two exceptions--were similarly divided on how health IT affected 
implementation. However, proportionately more respondents for care 
coordination or transitions of care interventions as well as care-
process-improvement interventions reported health IT as having 
facilitated implementation compared to respondents for other types of 
interventions.[Footnote 35] This result suggests that as policymakers 
consider different health care interventions, implementation of some 
of their options will depend more heavily than others on having 
appropriately configured health IT in place. 

Financial incentives were most often reported as not a factor. 
Slightly more than half of our 127 respondents reported financial 
incentives--as distinct from the related, but broader, financial 
resources factor--as having not been a factor in implementation of 
their intervention. The exception was for the two types of 
interventions for which financial incentives were an integral 
component--provider payment restructuring and insurance redesign--
where respondents most often reported financial incentives as having 
significantly facilitated implementation. When asked to explain how 
financial incentives facilitated or impeded implementation most 
respondents simply provided a description of the incentives they used 
to implement the intervention, such as payments to providers or 
patients to participate in the intervention. However, a few 
respondents explained that the expected cost savings generated from 
the intervention was an indirect incentive to implement the 
intervention while other respondents explained that incentives within 
existing payment systems, or the lack thereof, affected 
implementation. While the implementation of many interventions 
included in our sample may not have been affected by financial 
incentives, current means of paying for health care, such as fee-for-
service payment structures, may have hindered the successful 
implementation of other interventions. 

Most Respondents Expected Leadership Support and Other Factors Would 
Be Very Important for Widespread Replication: 

Much as they had reported regarding the implementation of their 
intervention, nearly all respondents consistently expected that 
leadership support would be very important if one were to attempt to 
replicate their intervention as widely as possible (see table 5). 
Leadership support was reported nearly unanimously by respondents as 
being very important for widespread replication of their intervention, 
paralleling respondents' relatively consistent assessment of the 
effect of leadership on implementation. In addition, a clear majority 
of respondents expected that each of the other factors--except for 
financial incentives--would be either very or somewhat important for 
replication. 

Table 5: Expected Degree of Importance of Identified Factors for 
Widespread Replication of Selected Health Care Interventions: 

Number of respondents: 

Factors identified from relevant literature: Leadership support; 
Very important: 116; 
Somewhat important: 7; 
Not important: 0; 
Don't know/No basis to judge: 0; 
Total: 123. 

Factors identified from relevant literature: Staff resources; 
Very important: 79; 
Somewhat important: 41; 
Not important: 2; 
Don't know/No basis to judge: 1; 
Total: 123. 

Factors identified from relevant literature: Organizational culture; 
Very important: 79; 
Somewhat important: 39; 
Not important: 1; 
Don't know/No basis to judge: 2; 
Total: 121. 

Factors identified from relevant literature: Availability of tools and 
activities to standardize care; 
Very important: 71; 
Somewhat important: 41; 
Not important: 9; 
Don't know/No basis to judge: 2; 
Total: 123. 

Factors identified from relevant literature: Financial resources; 
Very important: 66; 
Somewhat important: 44; 
Not important: 11; 
Don't know/No basis to judge: 2; 
Total: 123. 

Factors identified from relevant literature: Health IT; 
Very important: 48; 
Somewhat important: 48; 
Not important: 23; 
Don't know/No basis to judge: 4; 
Total: 123. 

Factors identified from relevant literature: Financial incentives; 
Very important: 30; 
Somewhat important: 23; 
Not important: 53; 
Don't know/No basis to judge: 15; 
Total: 121. 

Source: GAO analysis responses to GAO questionnaire. 

Note: Table excludes missing responses. 

[End of table] 

In contrast to the highly divided views health IT evoked from 
respondents regarding its role in the implementation of their 
interventions, it was reported by a substantial majority of 
respondents as either very (48) or somewhat (48) important for 
widespread replication. This could be an indication that, if health-IT-
related impediments experienced when implementing the intervention, 
such as the lack of interoperability across settings, were 
ameliorated, health IT could be important to the successful 
replication of some interventions. Similar to views expressed about 
the implementation of care coordination or transitions of care 
interventions, respondents for these types of interventions commonly 
reported that health IT would be very important for widespread 
replication more so than respondents for other types of interventions. 

Financial incentives was the factor that drew the most mixed 
assessment from respondents with regards to its expected importance 
for the widespread replication of interventions. Nearly half of 
respondents indicated that financial incentives were not important for 
widespread replication, which is similar to the view of most 
respondents regarding the role of such incentives in the 
implementation of their interventions. Another substantial group of 
respondents (30) indicated that financial incentives would be very 
important for replication. When respondents were asked to explain why 
factors would be important for widespread replication, respondents 
discussed financial factors more frequently than any other factor. 
Respondents' explanations about these financial factors often 
concerned a misalignment of financial incentives within existing 
payment systems that limited the attractiveness of replicating 
interventions that seek to enhance value. For example, some 
respondents noted that it would be difficult to replicate 
interventions that involved providing additional services, such as 
care coordination, under existing payment systems that typically do 
not compensate providers for those services. 

Concluding Observations: 

Our work suggests that progress in achieving greater value in health 
care in the U.S. will depend, in part, on the availability of 
information regarding the effect of different interventions on quality 
of care and costs and on how policymakers and others assess and use 
that information. Such information can guide the choices of 
policymakers among multiple interventions vying for support, but those 
decisions will have a sounder basis if the information meets certain 
criteria regarding its content and strength of evidence. With respect 
to content, information on the magnitude of an intervention's effect 
on both quality of care and costs is needed to determine if an 
intervention has enhanced value. In the case of the responses to our 
questionnaire on 127 diverse interventions, we found that this basic 
level of information was reported as available about half the time. 
With respect to the strength of evidence, the most critical indication 
comes from the types of study designs used to produce that 
information. There are a range of rigorous study designs which can 
provide credible support for the attribution of observed changes in 
quality of care and costs to a particular intervention. Our review of 
studies associated with the 127 interventions examined by our 
questionnaire found that while a number of studies employed rigorous 
study designs, a substantially larger number employed weaker designs 
that could not isolate the effect of an intervention from other 
factors. To the extent that policymakers find and use information on 
health care interventions that provides sufficient credible evidence 
on the effects of those interventions on both quality of care and 
costs, they will be better equipped to determine which interventions 
produce greater value in health care. Our work also suggests that 
successful efforts to encourage the widespread adoption of value-
enhancing interventions will need to take into account a complex mix 
of factors, including leadership support, organizational culture, and 
staff resources, that facilitate the implementation of health care 
interventions across a wide range of organizational contexts. 

Agency Comments: 

We requested comments from the Department of Health and Human 
Services, but none were provided. 

We are sending copies of this report to the Secretary of Health and 
Human Services and other interested parties. In addition, the report 
will be available at no charge on the GAO Web site at [hyperlink, 
http://www.gao.gov]. 

If you or your staff have any questions regarding this report, please 
contact me at (202) 512-7114 or cosgrovej@gao.gov. Contact points for 
our Offices of Congressional Relations and Public Affairs may be found 
on the last page of this report. GAO staff who made major 
contributions to this report are listed in appendix V. 

Signed by: 

James Cosgrove: 
Director, Health Care: 

[End of section] 

Appendix I: Scope and Methodology: 

To examine the availability of information on the effect of selected 
health care interventions on quality of care and costs as well as 
factors that can facilitate the implementation and replication of 
these interventions, we studied a diverse set of specific 
interventions that seek to enhance the value of health care through 
making changes in the way care is delivered. Specifically, these 
interventions make changes in who delivers health care services, how 
care is organized, or where care is delivered for a specified 
population. To identify interventions for our study, we drew upon six 
distinct sources to select a broad and diverse, though not exhaustive, 
set of interventions that have been implemented in one or more 
locations in the U.S. or abroad. These sources allowed us to identify 
a wide range of value-enhancing strategies implemented in different of 
health care settings, such as hospitals, integrated delivery systems, 
and physician practices, over more than a 10-year time span, including 
interventions that have not been described in academic or professional 
literature. We identified 828 interventions of potential relevance to 
our study through the following six sources: 

* A review of relevant literature on health care interventions that 
make changes in who delivers care services, how it is delivered, or 
where it is delivered. We conducted several searches using online 
databases, including Medline and ProQuest Health, to identify articles 
on interventions that were published from 1999 to 2009. 

* A review of interventions contained in the Agency for Healthcare 
Research and Quality's (AHRQ) Health Care Innovations Exchange (HCIE) 
as of August 20, 2009. The HCIE is a Web site that acts as a 
repository for information on quality improvement interventions and 
other innovative strategies to improve health care submitted by their 
implementers. Many of the interventions contained in the HCIE make 
changes in the way health care is delivered and include information on 
cost as well as quality.[Footnote 36] 

* A review of the relevant articles contained in the Tufts Medical 
Center's Cost-Effectiveness Analysis (CEA) Registry that were 
published from 1999 to 2009. The CEA Registry is a comprehensive 
database of health care cost-utility analyses that examine the health 
benefits and costs of strategies to improve health care. The CEA 
Registry contains articles from 45 peer-reviewed publications. 
[Footnote 37] 

* Interviews with experts on health care interventions associated with 
organizations such as state governments, integrated delivery systems, 
employer groups, and other countries. 

* Information on interventions that we identified from press reports, 
select journal articles published after 2009, and presentations at 
conferences. 

* Information on interventions submitted by their innovators or 
evaluators to either the Senate Budget Committee or GAO. 

To select interventions for inclusion in our study, we reviewed source 
documents for each of the potentially relevant interventions that we 
identified through our six sources. We selected 239 interventions that 
met the following seven criteria: 

* The intervention made a discrete change in who delivers health care 
services, how care is organized, or where care is delivered. 

* The intervention targeted a population or problem that was relevant 
to the U.S. health care system.[Footnote 38] 

* The intervention may have included health information technology 
(health IT) as one of its components of change, but health IT was not 
the intervention's only component of change. 

* The primary goal of the intervention was not focused on increasing 
access to care. 

* The intervention activities must fall within the health care system. 
[Footnote 39] 

* The source document or documents for the intervention either 
contained information or indicated that information is available on 
the effect of the intervention on quality of care and its effect on 
costs. Moreover, the source documents indicated that the intervention 
enhanced the value of health care by meeting one of the following 
three conditions: (1) increases quality of care and reduces costs; (2) 
maintains quality of care and reduces costs; or (3) increases quality 
of care and maintains costs. 

* The intervention was implemented in at least one health care 
setting. Interventions that were studied by examining their potential 
costs and benefits based on simulated outcomes rather than analyzing 
data from their actual implementation were excluded. 

To collect information on the 239 health care interventions that we 
selected for our study, we developed a Web-based questionnaire that 
contained 22 open-and closed-ended questions on interventions, their 
effect on quality of care and costs, and factors that may affect their 
implementation and replication. We sent our questionnaire to 235 
individuals who participated in developing, implementing, or 
evaluating each intervention.[Footnote 40] We identified these 
individuals through the source documents that we used to select 
interventions for our study. We received usable responses--responses 
that contained relevant information on the effect of the intervention 
on quality of care, the effect of the intervention on costs, or key 
factors that may affect implementation--for 127 interventions. We 
developed protocols for cleaning and analyzing data that we received 
from questionnaire respondents.[Footnote 41] These protocols included: 
identifying usable responses; reviewing source documents to clarify 
responses; and, if necessary, contacting respondents directly to 
obtain additional information on their intervention. 

To determine the availability of information on the effect of selected 
health care interventions on quality of care, we analyzed data that we 
collected from respondents through our questionnaire. We asked 
respondents to describe up to five key measures used to assess the 
effect of their intervention on quality of care and the magnitude--a 
percentage change or other quantitative assessment--of change observed 
in each measure described relative to a control group that did not 
experience the intervention or a baseline assessment made prior to 
implementing the intervention. We conducted a content analysis on 
questionnaire responses to determine the number of respondents who 
described one or more key measures used to assess the effect of their 
intervention on quality of care and the number of respondents who 
reported improvements in those measures attributable to their 
intervention. 

As part of our analysis on the availability of information on the 
effect of selected health care interventions on quality of care, we 
examined the types of quality measures respondents reported. We 
conducted a content analysis on questionnaire responses to determine 
what aspect of care quality--such as patient mortality, hospital 
readmissions, or patient satisfaction with care--each measure 
examined. We categorized each measure that respondents described by 
type based on the aspect of care quality it examined; for example, we 
categorized a measure that assessed the effect of an intervention on 
patient mortality as an outcome measure. We categorized quality 
measures into types that are largely based on the measure domains laid 
out by AHRQ in its National Quality Measure Clearing House.[Footnote 
42] We did not include all measure domains laid out by AHRQ in our 
analysis, because some domains, such as access to care, fell outside 
of the scope of our engagement. Moreover, measures that did not 
clearly specify which aspect of care quality was assessed were 
categorized as unspecified measures. We analyzed this information to 
determine the types of quality measures used to assess the effect of 
each intervention on quality of care. 

To determine the availability of information on the effect of selected 
health care interventions on costs, we analyzed data that we collected 
from respondents through our questionnaire. We asked respondents to 
report the type of cost savings, such as total dollars saved or 
dollars saved per patient, calculated to assess the effect of their 
intervention on costs and the specific amount saved for type of cost 
savings calculated. We also asked respondents if their reported 
savings accounted for costs associated with implementing the 
intervention and what information was used to calculate those savings. 
We determined the number of respondents who reported calculating each 
type of savings and a specific amount saved for those savings. 
Furthermore, we analyzed responses by finding the number of 
respondents who reported accounting for costs associated with 
implementing the intervention--net cost savings--and the type of 
information used to calculate those savings. Additionally, we used 
this information along with information we obtained through our 
analysis of quality measures to determine (1) the number of 
respondents who reported a magnitude of improvement in quality 
measures and a specific amount saved attributable to their 
intervention and (2) the number of respondents who also reported net 
cost savings rather than gross cost savings. 

To identify key criteria that can be used to assess the strength of 
available evidence on the capacity of interventions to enhance the 
value of health care we interviewed methodological experts and 
conducted a literature review to identify relevant systems for 
assessing the strength of evidence. We reviewed methodological 
literature published by entities that have well-established systems 
for evaluating health care interventions, including the Cochrane 
Collaboration; AHRQ's Effective Health Care Program, which includes 
the Evidence-based Practice Centers; and in the United Kingdom, the 
National Institute for Health and Clinical Excellence and the Centre 
for Reviews and Dissemination. We focused on those entities with 
systems for evaluating organizational interventions that change the 
structure or delivery of health care. This led us to pay particular 
attention to the guidance developed by Cochrane's Effective Practice 
and Organisation of Care (EPOC) Group, a collaborative review group 
that specializes in conducting systematic reviews of organizational 
interventions.[Footnote 43] 

Our review of this methodological literature and guidance together 
with our expert interviews led us to develop a set of questions to 
help decision makers and policy analysts who support them to 
critically examine the strengths and limitations of evidence about 
health care interventions that seek to enhance value. These questions 
target three broad areas: (1) assessing the true effect of the 
intervention on quality of care and costs, (2) assessing the scope of 
study results, and (3) assessing an intervention's capacity for 
replication. We submitted our initial draft questions to several 
different experts in assessing the comparative effectiveness of health 
care interventions and received their feedback on the content and 
clarity of those questions. Based on that feedback, we made revisions, 
resulting in the criteria described in our report and the set of 
questions listed in appendix IV. 

As part of our efforts to identify key criteria for assessing the 
strengths and limitations of available evidence on the capacity of 
interventions to enhance the value of health care, we examined the 
choice of study design used by evaluators to study the interventions 
for which we received usable responses to our questionnaire. To 
determine the type of study design used to assess the effect of 
interventions, we reviewed source documents and questionnaire 
responses. (See appendix III for more information on study designs.) 
Some interventions reported results from multiple studies. In these 
cases, we identified each type of study design used to assess the 
intervention. We used this information to find the number of 
interventions that were assessed using more rigorous study designs 
such as randomized controlled trials and the number of interventions 
that were assessed using less rigorous study designs such as pre/post 
or cross sectional studies. 

Our approach is designed to assist decision makers and policy analysts 
in assessing the strengths and limitations of evidence provided to 
them about the effects of health care interventions on quality of care 
and costs. Our approach does not involve the performance of systematic 
reviews that could synthesize information about those effects from 
multiple studies. Nor does it attempt to describe a process for 
producing a numerical or qualitative rating of the methodological 
strength of a study along one or more specified dimensions. Rather, 
our approach emphasizes the questions that decision makers and policy 
analysts should ask and leaves open the format and content of the 
answers to those questions. 

To examine factors that can facilitate the implementation and 
replication of health care interventions that seek to enhance value, 
we analyzed data collected from respondents through our questionnaire. 
We reviewed key literature sources and interviewed experts to identify 
seven factors that may affect implementation including leadership 
support, organizational culture, and resources.[Footnote 44] 
Respondents were asked to indicate, from the list of close-ended 
categorical options, to what degree each of the seven factors 
facilitated or impeded implementation and to provide an open-ended 
explanation of how the factors facilitated or impeded implementation. 
We asked respondents who were familiar with the replication of their 
intervention to explain if and how the factors differed from site to 
site. Respondents were also asked to indicate the expected degree of 
importance that each factor could have in attempting to replicate the 
intervention as widely as possible and to explain why these factors 
were expected to be important. In addition to the factors identified 
through our literature review, we asked respondents to identify and 
describe up to three additional factors that facilitated or impeded 
implementation of their intervention or that would be important for 
wide-scale replication. All close-ended responses were analyzed by 
assessing the frequency distribution of responses for each factor. We 
conducted a content analysis on open-ended responses to identify 
common explanations of how these factors affected implementation and 
why these factors would be important for widespread replication of the 
intervention. 

As part of our analysis of factors that may affect implementation and 
replication, we examined differences in questionnaire responses by the 
intervention type. To determine the types of interventions for which 
we received usable questionnaire responses, we reviewed source 
documents and questionnaire responses for each intervention and 
assigned them to one of eight categories (see appendix II for more 
information about intervention type). To categorize interventions by 
type we assessed key intervention characteristics, including the 
population targeted for behavior change and levers or activities used 
to change the way health care services are delivered. For example, a 
hospital surgical team that implemented a checklist was categorized as 
a patient safety improvement intervention. Some interventions 
exhibited key characteristics of more than one type of intervention. 
For example, a primary care practice that implemented a nurse case 
manager to facilitate care transitions and employ disease management 
strategies exhibits key characteristics of both care coordination or 
transition of care programs and chronic condition management 
interventions. Interventions that exhibited key characteristics of 
more than one type of intervention were categorized in all appropriate 
types. To determine if the effect or expected degree of importance of 
the factors differed by the type of intervention, we assessed the 
frequency distribution of responses for each factor across 
intervention type. 

Although our efforts to identify relevant interventions for our study 
were extensive, we could not ensure that every intervention meeting 
our selection criteria had been identified. Therefore the results from 
our questionnaire are limited in scope to the 127 interventions for 
which we received usable responses, and cannot be generalized to all 
value-enhancing health care interventions. 

[End of section] 

Appendix II: Types of Health Care Interventions That Seek to Improve 
the Value of Health Care: 

Intervention type (number in study): Provider payment restructuring 
(3); 
Description: 
* Interventions that seek to alter provider behavior by systematically 
changing the basis for provider payments; 
Examples: 
* Providing a single payment, or bundled payment, for all health care 
services that are delivered for a defined episode of care or a 
specified period of time; 
* Providing physician group practices performance payments if the 
practice meets or exceeds performance targets. 

Intervention type (number in study): Insurance redesign (6); 
Description: 
* Interventions that seek to alter patient behavior by restructuring 
health insurance plan provisions or related health care benefits; 
Examples: 
* Insurers offer enrollees a tiered network of providers. Enrollees 
who choose a provider in the higher cost tier pay higher premiums or 
cost sharing than enrollees who choose a provider in a lower cost tier; 
* Enrollees are charged a lower or no copay for specific drugs that 
are part of a recommended medical regimen for a medical condition. 

Intervention type (number in study): Chronic condition management (38); 
Description: 
* Interventions that seek to improve care for patients with chronic 
conditions; 
* Can be implemented in either inpatient or outpatient settings; 
* Can focus on patient or clinician activities, or both; 
Examples: 
* A nurse-social worker team is introduced into a primary care 
practice to provide education, help patients improve self management 
skills, and develop care plans with patients; 
* A multidisciplinary team holds classes for children with severe 
asthma and their parents to address physical needs and group, 
individual and family therapy for psychological needs. 

Intervention type (number in study): Patient safety improvement (26); 
Description: 
* Interventions that seek to prevent or reduce adverse events caused 
by medical care; 
* Adverse events include improper prescriptions or administration of 
medications, health-care associated infections, and pressure sores; 
Examples: 
* A surgical team implements a check list that enhances team 
communication and situational awareness among clinicians to prevent 
wrong-site surgeries; 
* A program of patient risk assessments, specialist consultations, and 
new equipment is designed to minimize pressure sores. 

Intervention type (number in study): Care coordination or transition 
of care programs (24); 
Description: 
* Interventions that facilitate patient transfers from one setting to 
another; 
* Some focus on coordination of patient care provided by multiple 
providers; 
Examples: 
* An advanced practice nurse and a trained elder peer provide support 
to older adults who are discharged home after a heart attack or 
undergoing bypass surgery to encourage compliance with medications and 
lifestyle changes; 
* A team of nurses and social workers work with patients with multiple 
chronic conditions to coordinate care from multiple providers and to 
provide ongoing monitoring and referrals. 

Intervention type (number in study): Continuous system improvement (3); 
Description: 
* Interventions that seek to change health care organization as a 
whole through ongoing and iterative reassessment of health care 
practices; 
* Such interventions seek to both reduce inefficiency or waste and 
improve patient outcomes; 
Examples: 
* A hospital created teams trained in "lean" principles, based on 
Toyota's manufacturing approach, to identify where changes in routine 
procedures could reduce waste and increase efficiency. 

Intervention type (number in study): Prevention; 
(4); 
Description: 
* The primary goal is to improve health by forestalling the 
development of illness in the first place; 
Examples: 
* Programs to promote wellness activities and health screenings or to 
prevent falls; 
* These interventions do not include programs to prevent adverse 
events. 

Intervention type (number in study): Care process improvement; 
(31); 
Description: 
* Interventions that seek to ensure that clinical staff adhere to 
specified treatment protocols or other forms of standardized practices; 
* These interventions seek to modify care processes by changing where 
care is delivered, how care is organized or structured, or who 
delivers care; 
Examples: 
* Multi-site intensive care unit telemedicine program; 
* A team of clinicians use a four-step mobility protocol to regularly 
assess the functional and clinical status of intensive care unit 
patients with respiratory failure. 

Source: GAO. 

[End of table] 

[End of section] 

Appendix III: What Makes Some Study Designs More Rigorous Than Others: 

The methodological literature on assessing the effect of interventions 
places a major emphasis on study design for identifying those studies 
that have the capacity to assess an intervention's effect on an 
outcome.[Footnote 45] The key strength of rigorous study designs is 
that they can take account of other factors that could affect the 
outcome of interest, and thereby isolate the effect of the 
intervention itself. 

Randomized controlled trials (RCTs) are widely considered to be among 
the most rigorous types of study designs because their basic structure 
inherently minimizes the potential impact of confounding factors on 
their results. RCTs accomplish this by randomly allocating study 
participants to groups that either receive the intervention--generally 
referred to as intervention or treatment groups--or do not receive the 
intervention--the control groups. The consequence of random allocation 
is that the only systematic difference between study participants in 
the two groups is exposure to the intervention.[Footnote 46] Thus, the 
effect of all other factors is the same on the two groups and 
therefore neutralized in making comparisons between the intervention 
and control groups. 

A second design type, known as the controlled before and after study, 
can be used in situations where the random allocation of study 
participants between intervention and control groups required for an 
RCT is not feasible. Controlled before and after studies use data 
collected from separate treatment and control groups, both before and 
after the intervention's implementation, to help to separate the 
effect of the intervention from that of other factors at work over 
that time period. In this design type, the control group is generally 
chosen in a way that is likely to produce a group that is broadly 
similar to the treatment group prior to the implementation of the 
intervention. However, methodologists generally recommend an explicit 
analysis to compare the intervention and control groups used in 
controlled before and after studies in order to demonstrate that they 
were in fact similar before the intervention took place. 

A third design type, an interrupted time series study, is not based on 
a comparison of intervention and control groups.[Footnote 47] Instead, 
it tracks an outcome of interest over time with measurements taken at 
many different time points both before and after the intervention. The 
multiple data points from before the implementation of the 
intervention enable analysts to take account of the impact of other 
factors on the outcome and thereby isolate the intervention's effect 
on that outcome. The interrupted time series design works best when 
there are data from a substantial number of different time points, 
both before and after implementation of the intervention.[Footnote 48] 

Other types of study designs cannot isolate the effect of an 
intervention from that of other factors because they provide no 
separate information on what would have happened without the 
intervention. For example, in a simple pre/post study all one has is a 
measurement of the outcome before implementation of the intervention 
and a measurement of the outcome after the intervention. The observed 
difference reflects all the factors (including the intervention) 
affecting the outcome over that time period. Because confounding 
factors could potentially affect the outcome in either the same or the 
opposite direction as the intervention, the actual effect of the 
intervention itself could be either greater or smaller than the simple 
pre/post difference. Even the direction of the intervention's effect, 
to increase or decrease the outcome, could be the opposite of the 
overall change from pre to post. That is why the results of a pre/post 
study generally cannot be relied on to provide even an approximation 
of what the likely effect of a health care intervention is on quality 
of care and costs.[Footnote 49] 

[End of section] 

Appendix IV: Key Questions for Assessing Evidence from Studies of 
Interventions That Seek to Enhance Value: 

The following three tables provide a set of questions that are 
intended to help policymakers and others find the information needed 
to assess the strengths and limitations of evidence drawn from studies 
of health care interventions that seek to enhance value relating to 
their impact on quality of care and costs.[Footnote 50] The three 
tables focus on the three broad dimensions described in the body of 
this report: (1) the credibility of evidence that attributes changes 
in quality of care and costs to the intervention, (2) the 
applicability of study results for broader populations of interest, 
and (3) the intervention's capacity for widespread replication. 

Each table lists a series of questions that highlight key information 
for assessing the evidence produced by relevant studies along with 
guidance on how to look for that information in published reports. 
Answers to most of these questions may be found in relevant sections 
of those reports;[Footnote 51] if not, one can ask the investigators 
who conducted the studies. While this set of questions is selective 
and does not cover every potential methodological issue, the 
information it calls for should provide policymakers a basis for 
making an informed assessment of the overall credibility and scope of 
the available evidence regarding the apparent impact of these 
interventions on quality of care and costs, as well as the 
demonstrated capacity of those interventions for widespread 
replication. 

Table 7: To Assess the Credibility of Attributing Observed Changes in 
Quality of Care and Costs to the Intervention: 

Key question: 1. Did the study use a rigorous design type? 
Guidance on finding and interpreting information needed: 
* Determine from study descriptions of methodology if the study used a 
rigorous study design type, such as: 
- randomized controlled trials; 
- interrupted times series; 
- controlled before and after study; 
* Or if it used a non-rigorous design type, such as: 
- pre/post study; 
- post-only study. 

Key question: 2. Were the intervention and control groups similar? 
Guidance on finding and interpreting information needed: 
* The two groups were similar if a study: 
- used random allocation to create the treatment and control groups, 
or; 
- analyzed how well the treatment and control groups matched on major 
characteristics prior to implementation of the intervention and found 
no major differences. 

Key question: 3. Were changes in quality assessed using appropriate 
measures? 
Guidance on finding and interpreting information needed: 
* Determine if the study cites other research documenting that the 
validity and reliability of the quality of care measures used in the 
study had previously been tested or to demonstrate that the measures 
had frequently been used in related research. 

Key question: 4. Were changes in costs assessed in a way that took 
account of implementation costs? 
Guidance on finding and interpreting information needed: 
* In calculating the cost effect of the intervention, studies should 
subtract all the costs of implementing the intervention, including 
both start-up costs and ongoing costs; 
* Return on investment is another way of presenting information on the 
cost effects of an intervention.[A] 

Key question: 5. Were data collected consistently? 
Guidance on finding and interpreting information needed: 
* Examine study methodology for descriptions of procedures to ensure 
that the same data were collected from different groups in the study--
such as intervention and control groups--at the same time and in the 
same way; 
* Determine if the study considered potential vulnerabilities to 
consistent data collection and took steps to mitigate them, such as 
blinded assessment of outcomes.[B] 

Key question: 6. Were the data collected sufficiently complete? 
Guidance on finding and interpreting information needed: 
* Studies should report the proportion of participants that dropped 
out of the study and therefore provided either no or partial data; 
* Concerns about data completeness are raised if: 
- the overall proportion of study participants who dropped out of the 
study exceeds 20 percent; 
- the proportion of study participants who dropped out of the 
treatment and control groups is not broadly similar. 

Source: GAO. 

[A] Return on investment (ROI) for an intervention represents the 
ratio of the change in overall costs of care attributed to the 
intervention divided by the cost of implementing the intervention. A 
positive ROI would have a value greater than one. An ROI between 0 and 
1 means that savings attributed to the intervention were less than the 
cost of implementing the intervention. An ROI of less than 0 indicates 
that the intervention led to increased costs. 

[B] Blinded assessment of outcomes means that those who collect data 
on quality outcomes do not know which study participants were assigned 
to the treatment and control groups. 

[End of table] 

Table 8: To Assess the Applicability of Study Results for Broader 
Populations of Interest: 

Key Question: 1. From what larger group or groups are study subjects, 
also known as the study population, chosen? 
Guidance on finding and interpreting information needed: 
* Examine description of study methodology to identify the group or 
groups from which study participants were selected (also known as a 
universe); 
* Identify the key characteristics that defined the universe and 
thereby determined who or what was eligible for inclusion in the 
study, such as type of provider. 

Key Question: 2. What method or mechanisms are used to select study 
participants? 
Guidance on finding and interpreting information needed: 
* If study descriptions indicate that the participants were selected 
randomly from a larger defined group, then this random selection makes 
the study population representative of the universe from which it was 
selected. 

Key Question: 3. Do analyses show that the study population and any 
broader populations of interest are broadly similar on key 
characteristics?[A] 
Guidance on finding and interpreting information needed: 
* Discussion or comment sections may analyze key similarities and 
differences between the study population and broader populations of 
interest; 
* To the extent that analyses of the study population compared with 
that of a population of interest establish that the two populations 
are broadly similar in terms of their key characteristics, those 
analyses support the representativeness of the study population for 
that broader population; 
* To the extent that analyses of the salient characteristics of the 
study population compared with that of a population of interest find 
that the two populations differ on one or more key characteristics, 
those analyses diminish support for the representativeness of the 
study population for that broader population; 
* If a study population differs from a broader population of interest 
on a given characteristic, it may still be representative of that 
broader population if persuasive evidence is presented to show that 
the characteristic on which the two differ does not affect performance 
of the intervention. 

Source: GAO. 

[A] Key characteristics are those that are likely to affect the 
performance of the intervention. Determining which characteristics are 
key may involve a mix of judgment, reference to past research, and 
analysis of study data. 

[End of table] 

Table 9: To Assess an Intervention's Capacity for Widespread 
Replication: 

Key Question: 1. In how many different organizations has the 
intervention been tested? 
Guidance on finding and interpreting information needed: 
* Determine if study reports indicate the number of different 
locations or sites at which the intervention was tested; 
* Different physical locations or sites represent different 
organizations to the extent that the success of an intervention in one 
site is independent of its success in another. 

Key Question: 2. How diverse were the organizations in which the 
intervention has been tested? 
Guidance on finding and interpreting information needed: 
* Determine if study reports describe the characteristics of the 
locations or sites at which the intervention was tested; 
* The more varied the organizations in which the intervention is 
tested, the stronger the test of its ability to be replicated 
successfully in different organizational contexts; 
* The more detailed the description of the sites, the more complete is 
the information on the extent of diversity among them. 

Key Question: 3. How uniform has implementation of the intervention 
been in different organizations? 
Guidance on finding and interpreting information needed: 
* Determine from study descriptions the extent to which the 
intervention accommodates variation in how it is implemented in 
different organizations; 
* If variation across organizations is substantial, identify, if 
possible, a subset of organizations that implemented essentially 
similar versions of the intervention. 

Key Question: 4. How consistent are the intervention's results across 
the different organizations where the intervention was implemented? 
Guidance on finding and interpreting information needed: 
* Examine study results for analyses of the extent of variation across 
different organizations included in the study that implemented similar 
versions of the intervention[A]; 
* An intervention could have consistent results on quality measures 
but not costs, or vice versa. 

Source: GAO. 

[A] If not provided in published studies, disaggregated results for 
individual organizations can be requested from the researchers who 
conducted assessments of the intervention. 

[End of table] 

[End of section] 

Appendix V: GAO Contact and Staff Acknowledgments: 

GAO Contact: 

James Cosgrove (202)-512-7114 or cosgrovej@gao.gov: 

Acknowledgments: 

In addition to the individual named above, Jessica Farb, Assistant 
Director; Kristin Ekelund; Krister Friday; Katie Mack; and Eric 
Peterson made key contributions to this report. 

[End of section] 

Footnotes: 

[1] Office of the Actuary, Centers for Medicare & Medicaid Services, 
National Health Expenditure Tables, table 1, accessed October 1, 2010, 
[hyperlink, 
https://www.cms.gov/NationalHealthExpendData/downloads/tables.pdf]. 
Total expenditures were adjusted for inflation in 2005 constant 
dollars. See M Hartman et al, "Health Spending Growth at A Historic 
Low in 2008," Health Affairs, 29:1, (Jan 2010) 148. Total federal 
budget outlays obtained from U.S. Budget for Fiscal Year 2011, Table 
3.2--Outlays By Function and Subfunction: 1962-2015, accessed January 
25, 2011, [hyperlink, 
http://www.gpoaccess.gov/usbudget/fy11/sheets/hist03z2.xls]. 

[2] E.S. Fisher and H.G. Welch, "Avoiding the Unintended Consequences 
of Growth in Medical Care: How Might More Be Worse?" Journal of the 
American Medical Association, vol. 281, no. 5 (1999): 446-453; E.S. 
Fisher, et al., "The Implications of Regional Variations in Medicare 
Spending. Part 1: The Content, Quality, and Accessibility of Care," 
Annals of Internal Medicine, vol. 138, no. 4 (2003): 273-287; E.S. 
Fisher, et al, "The Implications of Regional Variations in Medicare 
Spending. Part 2: Health Outcomes and Satisfaction with Care," Annals 
of Internal Medicine, vol. 138, no. 4 (2003): 288-298; and Joseph P. 
Newhouse and the Insurance Experiment Group, Free for All? Lessons 
from the RAND Health Insurance Experiment (Cambridge, Mass.: Harvard 
University Press, 1993). 

[3] For example, a seminal study found that on average Americans 
receive about half of recommended medical care processes, and the 
Institute of Medicine reported that at least 1.5 million preventable 
adverse drug events injure patients in the U.S. each year. See 
Elizabeth A. McGlynn, et al, "The Quality of Health Care Delivered to 
Adults in the United States," The New England Journal of Medicine, 
348:26 (June 26, 2003), 2643, and Institute of Medicine, Preventing 
Medication Errors, (2006), 3. 

[4] In this report we define value in health care as lowering or 
holding costs constant while sustaining or increasing quality of care. 

[5] See appendix II for a more complete description of different types 
of health care interventions that are intended to enhance value. 

[6] Pub. L. No. 111-148, 124 Stat. 119 (2010). Prominent examples 
include the establishment of a Center for Medicare and Medicaid 
Innovation within the Centers for Medicare & Medicaid Services 
(sections 3021 and 10306), the Medicare Shared Savings Program 
(sections 3022 and 10307), and the Hospital Readmissions Reduction 
Program (sections 3025 and 10309). 

[7] This criterion excluded interventions that focused on diseases or 
conditions that rarely occur in the U.S. or that addressed the 
particular challenges of delivering health care services in lesser- 
developed countries. 

[8] Respondents had the opportunity to provide the same information 
for any additional factors, beyond the seven listed on the 
questionnaire, that they had found also affected implementation or 
replication of the intervention. However, only a few respondents 
identified any additional factors. 

[9] See WR Shadish, TD Cook, and DT Campbell, Experimental and Quasi- 
Experimental Designs for Generalized Causal Inference, (Boston: 
Houghton Mifflin Co., 2002) for an authoritative explanation of these 
questions and approaches. 

[10] In the methodological literature, these issues are usually 
referred to as internal validity and external validity. 

[11] This guidance is available online at [hyperlink, 
http://epoc.cochrane.org/epoc-resources-review-authors]. 

[12] See, for example, J. Øvretveit. Does improving quality save 
money? A review of evidence of which improvements to quality reduce 
costs to health service providers. (London: The Health Foundation, 
September 2009); C. Homer and R. Baron. "How to Scale Up Primary Care 
Transformation: What We Know and What We Need to Know?" Journal of 
General Internal Medicine, vol. 25, no. 6 (2010); and E. Ferlie & S. 
Shortell. "Improving the Quality of Health Care in the United Kingdom 
and the United States: A Framework for Change." The Milbank Quarterly 
vol. 79, no. 2, 2001. For further description of research studies that 
identify factors that can affect the implementation and replication of 
health care interventions, see appendix I. 

[13] Respondents listed up to five measures used to assess the effect 
of their intervention on different aspects of care quality; we did not 
attempt to verify the appropriateness of the quality measures selected. 

[14] Respondents reported a magnitude of change observed in quality 
measures attributable to their intervention relative to a baseline 
assessment prior to implementation of the intervention or a control 
group that did not experience the intervention. We did not assess the 
extent to which observed changes in quality measures were due to the 
intervention and not to other factors. 

[15] We categorized quality measures into types that are largely based 
on the measure domains laid out by the Agency of Healthcare Research 
and Quality (AHRQ) in its National Quality Measures Clearinghouse. See 
appendix I for more information. 

[16] A caregiver is a family member, friend, or other individual who 
is responsible for meeting the physical and psychological needs of a 
patient. Caregivers are distinct from clinical providers such as 
physicians or nurses. Caregivers are the targeted population of some 
interventions. 

[17] AHRQ includes use of health care services measures as one of its 
measure domains, but it specifies that these measures should be used 
to monitor trends and are not direct measures of the quality of 
clinical care provided by health care professionals or organizations. 

[18] Many process of care measures are based on scientific evidence or 
established guidelines developed by professional organizations. 

[19] Experience of care measures assess quality of care in terms of 
the perspective of patients and caregivers toward the care that they 
participate in. These measures include patient and caregiver 
satisfaction. Structure of care measures assess a feature of a health 
care organization or clinician relevant to their capacity to provide 
health care. These measures include physician satisfaction and 
physician knowledge. 

[20] Respondents who reported that cost savings attributable to their 
intervention were assessed also reported information on the specific 
amount of costs saving annually and how these cost savings were 
calculated. We did not assess the extent to which these cost savings 
were due to the intervention and not to other factors. 

[21] The financial return on investment is the amount of money gained 
or lost as a result of an intervention relative to the money invested 
in the intervention. 

[22] One respondent who reported specific cost savings did not respond 
when asked if the savings were net or gross. 

[23] Some respondents who reported net cost savings also took account 
of other implementation costs such as costs associated with developing 
and implementing a study of the intervention. One respondent who 
reported net cost savings did not specify what costs were included in 
the costs of implementing the intervention. 

[24] These three design types can be used in a wide variety of 
circumstances. There are other rigorous design types that apply in 
more specialized circumstances, such as regression discontinuity 
designs. 

[25] In an RCT the study population is allocated randomly between 
control and treatment groups while in controlled before and after 
studies the treatment and control groups are formed on some nonrandom 
basis such as self selection or judgments made by clinicians or 
program administrators. 

[26] The methodology section of a study report may name the type of 
study design used or provide the information needed to identify the 
design type from its distinguishing characteristics. 

[27] If, for example, a substantially larger proportion of control 
group members drop out of a study than do members of the corresponding 
treatment group, the results of the study can be biased by any 
differences that distinguish those study participants that are lost 
from those that remain in the study and provide the data that generate 
the study's results. 

[28] Study reports may cite previous research to demonstrate that the 
study used measures with established validity and reliability. 

[29] P.J. Pronovost et al, "Preventing Bloodstream Infections: A 
Measurable National Success Story in Quality Improvement," Health 
Affairs, 30:4 (April 2011), pp. 629-30. 

[30] For example, if those collecting data on patient outcomes know 
which study participants are in the treatment group and which are in 
the control group, that knowledge can lead them to assess members of 
the treatment group differently. One technique for addressing this 
issue is blinded assessment of outcomes, which conceals from those 
making these assessments knowledge of which study participants are in 
the treatment and control groups. 

[31] Study results based on a random sample will have some uncertainty 
due to sampling error. However, because sampling error is random and 
therefore unbiased, it can be quantified in terms of a confidence 
interval and the study results are representative of the universe from 
which the sample was selected. 

[32] References to such analyses, if they are conducted, may appear in 
the discussion or comment sections of study reports. 

[33] Respondents who reported a factor as having impeded 
implementation of their intervention--either somewhat or 
significantly--commonly explained that the lack or insufficiency of 
the factor acted as a barrier that impeded implementation. 

[34] Health IT is interoperable when systems are able to exchange data 
accurately, effectively, securely, and consistently with different IT 
systems, software applications, and networks in such a way that the 
clinical or operational purposes and meaning of the data are preserved 
and unaltered. 

[35] Care coordination or transitions of care interventions set out to 
facilitate patient transfers from one setting to another. Some focus 
on coordination of patient care delivered by multiple providers. 

[36] See [hyperlink, http://www.innovations.ahrq.gov] for more 
information on the HCIE, last accessed on February 23, 2011. 

[37] See [hyperlink, http://www.cearegistry.org] for more information 
on the CEA Registry, last accessed on June 2, 2011. 

[38] For example, a treatment program to improve care for children and 
adults with malaria in sub-Saharan Africa was excluded from our set of 
interventions. Although this intervention was instituted to improve 
and standardize the delivery of health care services, malaria is not a 
disease commonly identified in the U.S. population. 

[39] For example, a community-based prevention program to limit 
alcohol consumption at licensed premises was excluded from our set of 
interventions. This intervention made changes in the community rather 
than within the health care system to reduce alcohol-related health 
problems. 

[40] Contacts for four of our selected interventions declined to 
participate in our questionnaire. 

[41] Our analysis of factors that facilitate or impede implementation 
and replication were based on 125 responses. Two out of 127 
interventions for which we received usable responses did not answer 
any questions on key factors for implementation. 

[42] The National Quality Measure Clearinghouse identifies seven 
measure domains to classify the focus of a quality measure. AHRQ's 
measure domains: access of care, outcome of care, patient experience 
of care, population health, process of care, structure of care, and 
use of services. See [hyperlink, http://qualitymeasures.ahrq.gov/] to 
access AHRQ's National Quality Measure Clearinghouse. Last accessed on 
June 22, 2011. 

[43] Other assessment approaches that we examined, but found less 
relevant for organizational interventions, were the Grading of 
Recommendations Assessment, Development and Evaluation (GRADE) system 
and the U.S. Preventive Services Task Force. See G.H. Guyatt et al, 
"GRADE: an emerging consensus on rating quality of evidence and 
strength of recommendations," BMJ, 336 (April 2008), 924-926; Agency 
for Healthcare Research and Quality, U.S. Preventive Services Task 
Force Procedure Manual, AHRQ Pub. No. 08-05118-EF, July 2008. 

[44] J. Ovretveit. Does improving quality save money? A review of 
evidence of which improvements to quality reduce costs to health 
services providers, (London: The Health Foundation, September 2009); 
C. VanDeusen Lukas et al, "Transformational change in health care 
systems: An organizational model," Health Care Management Review, vol. 
32, no. 4 (2007); C. Homer and R. Baron, "How to Scale Up Primary Care 
Transformation: What We Know and What We Need to Know?" Journal of 
General Internal Medicine, vol. 25, no. 6 (2010); M. Wang et al, 
"Redesigning Health Systems for Quality: Lessons from Emerging 
Practices," Joint Commission Journal on Quality and Patient Safety, 
vol. 32, no. 11 (2006); T. Greenhalgh, G. Robert, F. MacFarlane, P. 
Bate, and O. Kyriakidou, "Diffusion of Innovations in Service 
Organizations: Systematic Review and Recommendations," The Milbank 
Quarterly, vol. 82, no. 4 (2004); E. Ferlie and S. Shortell, 
"Improving the Quality of Health Care in the United Kingdom and the 
United States: A framework for Change," The Milbank Quarterly, vol. 
79, no. 2 (2001); S. Shortell et al, "Assessing the Impact of 
Continuous Quality Improvement/Total Quality Management: Concept 
versus Implementation," HSR: Health Services Research, vol. 30, no. 2 
(1995); T. Bodenheimer, "The American Health Care System: The Movement 
for Improved Quality in Health Care," New England Journal of Medicine, 
vol. 340, no. 6 (1999); and E. Bradley et al, "The Roles of Senior 
Management in Quality Improvement Efforts: What are the Key 
Components?," Journal of Healthcare Management, vol. 48, no. 1 (2003). 

[45] See for example WR Shadish, TD Cook, and DT Campbell, 
Experimental and Quasi-Experimental Designs for Generalized Causal 
Inference, (Boston: Houghton Mifflin Co., 2002). 

[46] Systematic differences involve patterns or trends. Where there 
are no systematic differences, all that is left is random variation. 

[47] However, it is possible to add a control group to an interrupted 
times series design to enhance its capacity to isolate the effect of 
an intervention. 

[48] There are other rigorous study design types that can be used in 
certain situations. One example is the regression discontinuity design 
that allocates study participants to either an intervention or control 
group based on whether they have a score above or below a specified 
cut-off point on an assignment variable. 

[49] An exception would be the unusual situation where it was 
established that no other factors besides the intervention could have 
affected the outcome over that time period. 

[50] Health care interventions enhance value when they lower or hold 
costs constant while sustaining or increasing quality of care. 

[51] These could include sections that describe study methods and 
results (e.g., characteristics of study participants), as well as 
sections describing the implications of study results including their 
limitations. 

[End of section] 

GAO's Mission: 

The Government Accountability Office, the audit, evaluation and 
investigative arm of Congress, exists to support Congress in meeting 
its constitutional responsibilities and to help improve the performance 
and accountability of the federal government for the American people. 
GAO examines the use of public funds; evaluates federal programs and 
policies; and provides analyses, recommendations, and other assistance 
to help Congress make informed oversight, policy, and funding 
decisions. GAO's commitment to good government is reflected in its core 
values of accountability, integrity, and reliability. 

Obtaining Copies of GAO Reports and Testimony: 

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through GAO's Web site [hyperlink, http://www.gao.gov]. Each 
weekday, GAO posts newly released reports, testimony, and 
correspondence on its Web site. To have GAO e-mail you a list of newly 
posted products every afternoon, go to [hyperlink, http://www.gao.gov] 
and select "E-mail Updates." 

Order by Phone: 

The price of each GAO publication reflects GAO’s actual cost of
production and distribution and depends on the number of pages in the
publication and whether the publication is printed in color or black and
white. Pricing and ordering information is posted on GAO’s Web site, 
[hyperlink, http://www.gao.gov/ordering.htm]. 

Place orders by calling (202) 512-6000, toll free (866) 801-7077, or
TDD (202) 512-2537. 

Orders may be paid for using American Express, Discover Card,
MasterCard, Visa, check, or money order. Call for additional 
information. 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 

Web site: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]: 
E-mail: fraudnet@gao.gov: 
Automated answering system: (800) 424-5454 or (202) 512-7470: 

Congressional Relations: 

Ralph Dawn, Managing Director, dawnr@gao.gov: 
(202) 512-4400: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7125: 
Washington, D.C. 20548: 

Public Affairs: 

Chuck Young, Managing Director, youngc1@gao.gov: 
(202) 512-4800: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7149: 
Washington, D.C. 20548: