Section 2. Aims, Existing Constructs, and Organizing Framework

2.1. Aims

There were two primary aims for our work:

1. Estimate the amount of waste in current American hospital operations.
2. Provide a set of tools that hospital Chief Financial Officers (CFOs) can use to identify and eliminate said waste in their operations.

In considering these aims, it was important to set boundaries on our work. We limited our evaluation to hospital operations, without direct comment on outpatient care, long-term care, etc. One useful extension would be to examine episodes of care across these settings.

Return to Contents

2.2. Overview of Existing Quality Constructs for Waste/Poor Quality

Chassin coined the useful terms "overuse" (providing a treatment when its risk of harm exceeds its potential benefit), "underuse" (failing to provide an effective treatment when it would have produced favorable outcomes), and "misuse" (avoidable complications of appropriate care) to describe common classes of quality failures in health care delivery (Chassin, 1991; Chassin et al., 1998). He recognized that all three represent process defects, and he linked them to quality improvement (QI) methods that have significantly reduced defect rates in settings outside of health care (Chassin, 1998). He also considered the relationship of defect reduction to health care costs, arguing that (1) most health care delivery is not positively cost-effective; (2) in a perfect system, any cost savings generated from reducing overuse and misuse likely would be consumed, correcting underuse; and (3) therefore, QI activities, broadly applied, will require active, long-term investment with expected returns primarily coming in the form of improved health, rather than better access to care through reduced costs.

Poor quality can produce waste. Industrial quality theory adds a complementary dimension to Chassin's three categories, under which improvements in the quality of processes, products, and services can significantly decrease production costs (Crosby, 1979; Deming, 1986). When mapped into clinical practice, two specific mechanisms drive quality-based cost reduction. Both rely upon managing processes to reduce or eliminate process failures, or defects.

Quality waste. When a step in a clinical process fails, some proportion of those process failures will lead to outcome failures. There are only two choices for dealing with the resulting outcome failures, and both raise costs:

1. Invest additional resources to repair the initial failure (rework). For example, treating a medical complication consumes more health care resources than if the complication had never occurred.
2. Discard the defective output and start again (scrap). For example, repeating an x ray when the initial image is unreadable or making multiple attempts to track down a missing laboratory result, both produce scrap (i.e., wasted images and unnecessary telephone calls).

Quality waste is often a cost to the system, not a penalty to the service provider, under current payment policies. Quality waste also includes the often significant cost of detecting outcome failures (inspection costs), because such vigilance would not be necessary if the process produced no failures. A good example of this is the efforts hospitals across the country are dedicating to medication reconciliation.

Currently, the QI infrastructure in health care should not be considered waste but rather as a necessary conduit for high reliability service delivery. However, it is technically still quality waste.

Recognizing the forms and magnitude of quality waste provides a way to identify opportunities for improvement (i.e., watch for any instances of rework or scrap) and to respond to them (i.e., build and manage a process that does not fail in the first place). Quality waste strategies center on prevention, applied in a very broad way: for any defect, move upstream in the process, find the root causes of the failure, then do it right the first time for future cases. Under this theory, better clinical process design and management can produce better medical outcomes, eliminate quality waste, and reduce health care costs.

Inefficiency waste. Two processes accept identical inputs and produce identical outputs, but one process consumes more resources to do so (is less efficient). Which one should care providers use? With limited health care resources, an inefficient process wastes resources that could otherwise generate health benefits and thus reduces the total health benefit achieved for a population of patients.

Roberts and Zangwill (1993) define inefficiency waste as "any non-value adding work" (p. 2). They catalog an impressive list of specific mechanisms by which inefficiency wastes resources, including unnecessary redundancy (e.g., repetitive collection of patient histories); downtime and delays (e.g., time during which a piece of expensive equipment is not in active use); unnecessary complexity; failure to use all available resources, including knowledge, people, and equipment, to add value when there is good economic opportunity to do so; efforts spent improving processes that were useless to begin with; and consuming resources to produce products that see no use (e.g., some bureaucratic reports or mandatory quality reporting systems that produce no change).

One particularly interesting entry in Roberts and Zangwill's hierarchy of inefficiencies is care design waste. Over time, many processes improve through thoughtful experimentation and refinement. Steps initially thought to be essential shorten or disappear, producing efficiencies. In retrospect, those eliminated steps and their associated resources were waste.

For example, over the past several years, many U.S. cardiac surgery programs have reduced median postsurgical intubation times from more than 25 hours to less than 8 hours, producing major cost savings while maintaining excellent clinical outcomes (Cheng, 1998; Silbert et al., 1998). In retrospect, more than two-thirds of that initial thoracic intensive care unit (ICU) care process represented inefficiency waste, compared with what was eventually discovered. As care delivery teams change, refine, and eliminate process steps, new opportunities for further waste removal become obvious in continuing cycles of improvement. According to this bold framing, failure to innovate and failure to use proven innovations are both forms of waste.

Better clinical process design and management can eliminate inefficiency waste, causing health care costs to fall while medical outcomes remain stable. If health systems use those savings to extend effective care to other patients, costs hold stable while other medical outcomes, at a population level, can improve.

Return to Contents

2.3. Planned Approach for Identifying a Typology of Waste/Inefficiency

This task initially intended to identify a typology for waste/inefficiency in health care, drawing on the literature and abstraction of a set of projects completed at Intermountain Healthcare (i.e., Advanced Training Program [ATP] projects) and Providence Health System (PHS) (i.e., Six Sigma projects). When executed, this approach did not provide the expected, emergent typology. A primary limitation was that these projects were largely done for training purposes, there was variability over time, and they did not provide a representative cross-section of waste/inefficiency in health care. Nevertheless, we present a summary of findings from this assessment below to document our effort.

We reviewed 10 years of Intermountain Healthcare ATP projects and 3 years of PHS projects for these purposes. The ATP projects represent work by representatives of many U.S. health systems to target areas of waste and poor quality. The ATP projects span a number of years; 58 projects for which summary reports were available were used in the review. The Six Sigma projects were conducted over 3 years at PHS. Some initial projects were selected as training projects for the initial Six Sigma rollout, but there is reason to believe that the projects are representative of the kinds of waste reduction efforts found nationally. The ATP and Six Sigma projects are summarized in Exhibits 1 and 2, respectively.

Even within the QI approach, differences were observed between the ATP targets and Six Sigma targets. The ATP projects tended to focus on overuse, whereas the Six Sigma projects concentrated on throughput. These differences reflect different internal organizational agendas and external, industrywide forces and did not appear to be method related.

The three areas of problems that emerged from our abstractions that we believed would be most productive for drill down and waste elimination in hospitals were

1. Throughput/unnecessary length of stay (LOS).
2. Bottlenecks, such as those caused by delayed laboratory results.
3. Substitutions in role/function.

Throughput was determined to be the most productive focus for inquiry because of interest among financial analysts at partner health systems and as evidenced by reports on this topic in relevant trade journals. Bottlenecks and substitutions are causal elements and are observed at the point of service.

In addition to this review of independently produced analyses (i.e., ATP and Six Sigma projects), we conducted several directed analyses at Intermountain Healthcare and PHS:

• An incident analysis (Appendix B) that demonstrates extended LOS for patients where any type of voluntary incident report is submitted at PHS; and
• TPS/Lean observations (Appendix C) that were completed as part of an operational initiative at Intermountain Healthcare and were validated with reliability testing and training at UNC Health System, depicting waste and inefficiency at the point of care.

Several more themes emerged from our QI, TPS, Lean, and incident analyses:

• Throughput/delay.
• Errors/defects.
• Using resources that are more expensive than necessary (treatment, setting, provider, equipment) whereby initial choices determine subsequent ones, preventing cheaper alternatives (i.e., for the same choice of setting, there can be more or less expensive provider levels and more or less expensive supplies).
• Reducing variation.
• Improving diagnosis.

While these exploratory analyses produced a set of focal areas for further investigation of variances in health care delivery, they did not yield a constructive typology to guide our work. This initial work did, however, contribute to our ability to provide specific examples and demonstrate how to capture waste and inefficiency in health care. Drawing on available constructs from the literature, we identified an organizing framework to guide our subsequent efforts, which is presented in Section 2.4.

Exhibit 1. Major categories targeted by ATP projects

Description	Number
Address underuse or overuse of treatment (often medications)	17
Devise better scheduling, information flow, and other types of coordination	15
Reduce delays	8
Reduce variation in how a condition is clinically treated (e.g., via guideline adherence, etc.)	7
Streamline administrative processes such as billing, Institutional Review Board (IRB), incident 5 reporting system, etc. Create methods for better diagnosis of patients' conditions	3
Reduce supply waste	3
Total	58

Exhibit 2. Major categories targeted by Six Sigma projects

Description	Number
Reduce delays	14
Devise better scheduling, information flow, and other types of coordination	4
Address underuse or overuse of treatment	2
Streamline administrative processes	2
Reduce supply waste	1
Create methods for better diagnosis of patients' conditions	1
Reduce variation in how a condition is clinically treated	0
Total	24

Return to Contents

2.4. Selected Framework for Considering Cost of Waste/ Poor Quality

The Institute of Medicine defined quality of care as the extent to which health services increase the likelihood of desired health outcomes and are consistent with current professional knowledge (IOM, 1990). In a more recent report, Crossing the Quality Chasm, the Institute of Medicine (IOM) alleges that the quality of health care is inadequate and has called for improvements in six areas—safety, effectiveness, patient-centeredness, timeliness, efficiency, and equity—concluding that QI cannot be achieved given the constraints of the current system and processes of care (IOM, 2001; Detmer et al., 2001).

"Perfect care may be a long way off, but much better care is within our grasp. The committee envisions a system that uses the best knowledge, that is focused intensely on patients, and that works across healthcare providers and settings. Taking advantage of new information technologies will be an important catalyst to moving us beyond where we are today. The committee believes that achieving such a system is both possible and necessary" (IOM 2001, p. 21).

In 2000, while serving on IOM's Committee on Quality of Healthcare in America (which produced To Err Is Human and Crossing the Quality Chasm), Dr. Donald Berwick introduced a useful hierarchy of health care delivery structures that he termed the Chain of Effect for Quality. The Chain of Effect for Quality provided a high-order structure that helped us organize examples of waste at a functional level and make sense of other waste classification systems (i.e., Chassin's overuse/underuse/misuse model and Deming's quality waste and inefficiency waste; Berwick, 2002) (go to Exhibit 3).

Exhibit 3. The chain of effect for quality applied to waste analyses

Waste Analysis Level	The Chain of Effect for Quality
	Structural Care System Element	Organization/People	Activity
1. Population Level	Environmental context	Federal and state government; employers and insurers	Health policy and oversight; health finance
2. Episode Level	Organizational context	Health care delivery organizations	Health system management
	Microsystems	Clinical teams (e.g., physicians, nurses)	Process management
3. Patient Level	Patient and community	Patients and families	Seeking help and care

From a financial management perspective, the Chain of Effect for Quality corresponds, roughly, to a method for building up health care costs as follows (go to Exhibit 4):

• Cost per unit (Patient Level): a "unit of care" is any single granular service that a hospital provides to a patient; for example, a single specific lab test (e.g., a CBC, an SMA-6, a c. difficile assay); a single dose of a particular drug, including route of delivery (1 gram of IV Ancef); a single imaging examination (a two-view chest radiograph); an acuity-adjusted minute in a procedure room; an acuity-adjusted hour of nursing services; or a particular disposable product (e.g., an emesis bowl, a meal, an incentive spirometer, the components of an artificial hip implant).
• Number of units per case (unit frequency; Episode Level): the number of "units of care" that are provided to make up a case.
• Number of cases per population (case frequency; Population Level).

Exhibit 4. The cost per unit versus number of units (frequency) triangle for understanding total care delivery costs

Source: Courtesy of Dr. David A. Burton, Intermountain Healthcare.

This method of thinking about the cost of care derives, in large part, from activity-based cost (ABC) accounting systems, which record individual units of care in a "transaction file" as they are provided across traditional cost/revenue silos (i.e., departments) (Emmett & Forget, 2005). ABC systems divide all care into detailed services and then track each service delivered to a patient by recording utilization in a transaction file. Such systems typically organize individual units of care into seven large categories:

1. Professional/room services (e.g., an acuity-adjusted hour of nursing service, a level 4 evaluation by a primary care physician, a 1-hour use of a private room).
2. Pharmacy (e.g., individual medications by amount and route, such as a 1-gram intravenous dose of cephtriaxone).
3. Laboratory tests (e.g., a complete blood count with differential [CBC with diff] or a basic metabolic panel [BMP]).
4. Imaging/radiology (e.g., a two-view chest x ray).
5. Operating and procedure rooms (e.g., acuity-adjusted minutes in an operating room with laminar airflow, a catheterization laboratory, or a labor and delivery suite).
6. Central supply (e.g., physical devices, such as an incentive spirometer or the components of a total hip arthroplasty implant).
7. Other.

An ABC system centers around a cost master file (sometimes called a charger master) that contains an entry for every billable service. Ideally, individual unit of care entries within an ABC cost master file provide detailed breakouts of fixed and variable and direct and indirect costs, which are updated on a rotating basis by management engineers, using direct measurement. Some hospitals carefully maintain their cost masters to reflect true unit costs, applying a standard markup across all units of care. Others modify individual unit costs without a systematic approach, which can result in major cost disparities within their cost master (e.g., the $900 toilet seat), even though total costs superficially appear to be accurate. A typical inpatient cost master file might contain more than 20,000 unit of care entries.

The Patient Level roughly corresponds to the ABC systems used in some health care delivery organizations, in both inpatient (e.g., Intermountain Healthcare) and outpatient (e.g., the Marshfield Clinic) care settings. Fee-for-service payment schemes usually function at the Patient Level, identifying and paying for individual units of care on a cost-plus basis.

The Microsystem or Episode Level bundles together the individual units of care associated with a clinic visit or a hospitalization. In those circumstances where a particular type of episode uses a consistent set of units of care, the bundled services may themselves, as a group, be treated as a unit. For example, the Diagnosis-Related Group (DRG) Prospective Payment System identifies 477 patient treatment categories, then bundles payment for all underlying unit services into a single transaction. Similarly, per diem payment systems bundle all services contained within a hospital day. Within the Chain of Effect for Quality, the Episode Level also corresponds to care delivery process management, the major focus of clinical QI (note: we discussed labeling this the Process Level, but because process management theory applies at the other two levels as well, we decided that it might be confusing).

The Population Level extends care delivery episodes across the continuum of a patient's life, addressing longitudinal disease entities from their prevention to their detection, treatment, and eventual conclusion. It corresponds to total lifetime costs for a particular disease or across a particular patient's life. Most capitation payment systems function at the Population Level, at least to the limits of a single contracting cycle (e.g., an insurance company may renegotiate insurance premiums and care delivery payment contracts annually, but in most circumstances patients can continue their coverage over time, from year to year, even if they develop significant chronic diseases).

To identify and capture specific aspects of costs that could be reduced/eliminated at the Patient Level, several tool sets are now being embraced by the health care system and were integrally involved in our work. Most notable among these are

• Application of Lean Thinking and Toyota Production System (TPS).
• Six Sigma (Pande et al., 2000; http://www.isixsigma.com).

Overall, we found Six Sigma and Lean to be complementary approaches. Six Sigma, as applied at PHS, looks at the entire process, focusing on all delays. The speed of the process is controlled by (1) the way the work is organized, (2) the efficiency of individuals in implementing their role, and (3) defects that require extra processing (rework). Factor 1 is process centered, whereas Factor 2 is staff person centered. PHS Six Sigma targets Factor 1, whereas Lean targets Factors 2 and 3. Obviously, there is overlap between the two methods.

For our purposes, we relied heavily on TPS, which offers a problem-solving approach for process improvement and operational excellence. Growing interest in TPS and lean production thinking in health care, coupled with the relevance of its systematic and granular focus on waste/inefficiency, led us to explore this approach within our study.

We next introduce key concepts that depict how financial reimbursement influences our ability to capture waste/inefficiency in health care (go to Section 2.5). In Section 3, we conclude our report with specific depictions of our examination of quality waste and inefficiency in health care.

Return to Contents

2.5. Financial Reimbursement Influences on Capturing Waste/ Inefficiency

In examining the U.S. health care system, we classified expenditures as waste if they demonstrably consumed health care resources, without producing a health benefit or contributing value to health care operations, at the Population Level. This viewpoint highlights two forms of suboptimization. Many factors that represent waste at the population level represent business profit at lower levels in the system. In counterpoint, changes that generate cost savings for the system as a whole may financially punish those who must implement the changes that drive the savings.

Return to Contents

2.5.1. Increases in Waste that Enhance Health Care Providers' Profits (Unit Costs versus Frequency)

For many disease-treatment entities, the number of units of care necessary to achieve a good result for a patient is not empirically known. Faced with rapidly escalating health care costs, governmental and private payers usually try to control their expenditures by reducing unit costs. Health care providers often respond by relaxing indications criteria and thus increasing the number of units of care used to treat a patient (for example, imaging examinations, clinic visits, or hospitalizations). As a result, even while unit payments drop, total costs (and income for health care providers) rise (Exhibit 4). Providers may also "unbundle" services by breaking a bundled package of services into more granular underlying units of care and then move those units of care outside of the negotiated payments for the bundled service, exposing higher unit prices than were hidden within the negotiated bundled service. Care providers may also unilaterally redefine the content of a bundled service. For example, faced with relative reductions in payment rates within DRG categories for per case (bundled) payment, acute care hospitals now often quickly transfer those patients to long-term care facilities. That has the effect of reducing the acute care hospital's LOS and associated frequency of underlying units of care, so that the hospital has lower costs while receiving the same fixed payment. Meanwhile, the governmental or private insurer is faced with additional payments to a second entity (the long-term care facility), greatly increasing its total payment for the same care episode.

The frequency effect is exacerbated by the introduction of new technology, which introduces completely new units of care for consumption and payment. In addition, even though the unit cost of a particular technology drops over time, as economies of scale and other efficiencies associated with the technology come into play, many care providers may supplement, rather than replace, the old technology with later technologies for a single case. For example, instead of moving directly to a nuclear stress test heart scan, a cardiologist may first order a plain stress test (relatively inexpensive today), followed by an ultrasound stress test (moderately expensive), and finally a nuclear stress test (much more expensive). Again, frequency increases, driving up total costs even though cost per unit drops.

Woolhandler, Campbell, and Himmelstein (2003) compared the administrative costs of health care delivery in the United States and Canada. They concluded that administrative costs in the United States ($1,059 per capita) were much higher than costs in Canada ($307 per capita) and that costs were increasing more rapidly in the United States than in Canada. They also showed that administrative costs were not evenly distributed among the various elements of the care delivery system. In particular, administrative personnel and costs for health care providers, as opposed to insurers, were much higher in the United States than in Canada.

Arguably, much of the higher administrative costs in the United States arose directly from mechanisms implemented to prevent increases in unit frequency by care providers, such as preauthorization systems. Related administrative mechanisms focused on fraud and abuse—another name for particularly egregious cases of frequency increases. Canada's national health care system achieves the same end more efficiently. Canada limits hospital beds and specialty physician practices through national health policy. Canada also uses provincial budgets to match primary care resource consumption to available resources, by automatically changing unit payments. Both of these mechanisms provide strong disincentives against case frequency. Many economists argue that administrative expenses undertaken to prevent frequency overuse still represent a net gain for the U.S. system. They also note that the Woolhandler et al. analysis categorized for-profit insurance companies' return on investment profit margins as administrative expense. As with inspection costs, this once again illustrates how systems designed to control costs or quality can themselves introduce inefficiency waste compared with more efficient alternatives.

Return to Contents
Proceed to Next Section