A study of clinical guidelines.

Laffel and Berwick are leading the conversion of health care from the disjointed, inspection-oriented improvement system into statistical quality control (SQC).[2] The old system's goal of conformance to standards assumes that some rate of poor outcomes is acceptable. There is, therefore, little emphasis on analyzing cases that meet the standard; rather, improvement occurs as a result of analyzing cases that go wrong. Also, the focus is on individual physicians and not the process problems that significantly contribute to substandard quality. SQC theory, on the other hand, has demonstrated that substantial quality improvement can be achieved by eliminating unnecessary variation in the execution of processes through which treatment plans are carried out.

Concurrent measurement of key process or outcome indicators is critical to the statistical control of the various processes put into place. The inherent variation in health care is of two varieties: "common cause" and "special or assignable cause." Common causes of variation are present (designed in) when the system is in "control," i.e., exhibits a binomial or normal distribution. When a system exhibiting predictable variability is acted upon by an event that "comes and goes capriciously," it is no longer predictable and is exhibiting nonsystematic variation as a result of this special cause. The effort of SQC, then, is to discover and eliminate these special causes, at the same time improving the predictability and outcome from the process by reducing the common causes variation.[3]

The common cause variation in health care, broadly called the standard of care, is noticeably broad for most conditions. We accept, for example, a two-fold difference in the incidence of prostatic surgery in two neighboring counties in Vermont, a one-and-one-half-fold difference in cesarian section rates between two hospitals or even two physicians, and a two- to three-fold difference in mortality rates among cardiac surgeons.[4]

Special cause variation is fundamentally different and is not predictable within the system; it is an event that lies well outside the expected range of variation in the system. Berwick calls it the unintended variability in the system.[3] Examples might be where one finds a defibrillator without battery power when using it, selects the wrong tube to draw blood, or misreads the label on a drug dose.

Analysis of Hospital Costs

Consumers Union notes that "Our health care system is so inherently wasteful and inefficient that a complete overhaul is an option worth contemplating."[4] The waste is principally inefficiency. Efforts to reduce administrative costs, schedule operating rooms more efficiently, streamline cafeterias, and root out obvious inefficiency are important, but, until volume and intensity of services are controlled, the expansive growth in costs will hardly slow.[5]

Hospital costs have undergone very little research that explicitly associates costs with the quality of care delivered. It is one thing to evaluate total costs and another to relate them to a given level of quality. Traditional cost accounting can certainly compare total provider salaries and benefits, other salaries, supplies, drugs, interest, and fees among hospitals, but these costs are not typically related to quality.[6] The principal driver of costs in care delivery is the intensity of services provided, which raises the obvious question of the value of the intensity to patient outcome: which level of intensity is most appropriate?[5] A recent study explored this relationship by globally comparing costs among 300 hospitals and relating those costs to mortality rates.[6] Predicted mortality rates were determined and the hospitals' total costs, adjusted for case severity, were matched to peer-group outcomes (mortalities among similar hospitals with similar case mixes). A marginal cost for a "death deferred was derived and regression analysis related costs to this deferred death. A 1 percent increase in the level of quality (reduction in predicted death) was estimated to increase costs by 1.34 percent. There are clearly decreasing returns to scale, and, in larger hospitals (with more complex cases), marginal costs approached $140,000 per death deferred.[6]

Using the Drucker concept of the new manufacturing accounting, health care financing must begin to address variations in service intensity as a specific cost deescalator. Moreover, the cost value of underutilized facilities (empty beds, unused technology) must be accounted for.

Clinical Modular Structures:

Practice Guidelines 

In order for a health care organization to keep up with changes in design, technology, and the market, flexible standardized steps in the process must be developed. The coming health care reform will include benefit packages that will describe general categories of treatments that will be covered, such as inhospital, preventive, and emergency services. There will be within these categories a range of treatments that vary from highly beneficial to unknown to downright harmful.[5] The development of practice guidelines is a business strategy that must be integrated into health care financial accounting. Guidelines, also known as parameters, policies, protocols, and algorithms, are specifically defined by the National Academy of Sciences, Institute of Medicine, as "Statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances."[7] The Institute indicated that guidelines should address several roles[8]:

* Cost control. * Assurance of quality. * Enhancement of access. * Consideration of patient preferences. * Consideration of physician autonomy. * Consideration of variations in utilization.

The randomized clinical trial has been medicine's gold standard, but only 10-15 percent of current therapies have been subjected to such trials because of their cost. The development of clinical protocols on a large scale will provide clinical trials that will answer some of the difficult questions because they will have both effectiveness assessment and cost built into the protocols.[9] Guidelines, as they are developed, must be based on the best evidence and expert judgment available and must project relevant health and cost outcomes of alternative courses of care. They must be flexible enough to adjust to patient preferences and local adaptability.[7] The "production module" design urged by Drucker will put the "hard science" of medicine into a new perspective of both managing variation in outcomes and protecting patients against receiving too little care.[9]

Clinical Systems Design

Drucker indicates that integrating the manufacturing process is the critical effort in converting goods (or services) into economic satisfaction. The delivery of health care lends itself very well to this integrated model. Restructuring of the hospital to optimize integration means breaking down old departmental lines and organizing work around the flow of patient care and treatment, not individual department tasks.[10] Production lines exist in hospitals, but they are now along discipline lines, such as admissions, radiology, or food services. Product lines in the future will be integrated along specific diagnoses or groups of diagnoses and will be organized around the needs of patients. The patient should experience little or no waiting time, reduced time in processes, little movement between stations, and contact with a limited number of employees. The patient should become a partner, rather than an object passed around the hospital at the convenience of the staff.[10]

Such product lines accomplish accountability for costs, patient-focused care, and quality improvement. They require a change in thinking, in that staff members are redeployed throughout the hospital, and some are multifunctional and cross-utilized.[11] Teamwork is essential in these models, as some traditional duties are now shared, e.g., by respiratory therapists (RTs) and nurses (RNs). Although each may now start IVs, administer medication, and perform nebulization and chest physical therapy, each retains the specialist functions of ventilator management for RTs and blood administration for RNs.[10]

Recommendations for the New

Medical Manufacturing

Although Drucker addressed manufacturing, it is apparent that the health care system can be restructured in line with all four of his principles:

* Performance measurements according to standards must be established to effectively reduce common cause variation using methods of statistical quality control.

* Cost analysis of systems must attend to the relationship of costs and quality and the cost-benefit of standardized care.

* Practice guidelines must define minimum standards of quality in terms of patient preference and outcomes in a flexible fashion.

* The entire episode of care must follow carefully designed, systematized processes that emphasize teamwork and integration along horizontal, patient-focused paths.

Development of Guidelines

The guideline is part of the medical record and, in tabular form, outlines appropriate indications for treatment or surgery, its timing and intensity, expected outcomes and criteria to measure outcomes, and a method to track variations. A variation occurs when the patient does not follow the expected pathway. It must be attributed to a key factor: the individuality of the patient, the physician's actions or orders, or other caregiver staff members and their actions or omissions.

At Wilford Hall Air Force Medical Center, the author facilitated the development of a process to redesign clinical treatments through a Clinical Protocol Redesign Guide.[12] It came from a day-long brainstorming and planning session in clinical reengineering. The guide describes a seven-step process (figure 1, below):

1. Select the diagnosis or the procedure to be developed. The diagnosis is chosen on the basis of being high in cost, volume, or risk.

2. Construct the clinical protocol. A team of all caregivers involved in the process should work through the protocol in one or two all-day sessions. The product is a document that identifies the key indicators and criteria that patients must exhibit to be part of the protocol. It describes the "best practice" expectation of the outcome that is expected each day of hospitalization.

3. Implement the protocol. After educating all staff involved in care, the patient is entered onto the protocol and the document becomes part of the medical record. The patient's progress along the expected pathway is documented on the protocol document.

4. Collect data. Variations from the protocol are tracked. Key quality and cost indicators are concurrently trended.

5. Analyze results. Using the tools of SQC, data are analyzed.

6. Improve the protocol. Using the methodology of continuous quality improvement, the protocol is improved.

7. Hold the gains. Formalize the protocol and periodically recheck progress. Report successes to management.

Figure 2, page 33, shows a guideline at Wilford Hall Medical Center that defines the expected treatment and outcome of a patient suffering from a transient cerebral ischemic attack (TIA).[12] Variations are noted in the third, fifth, seventh, and ninth columns of the protocol. Following the patient's discharge from the hospital variations are entered into a data base from which statistical inference can be drawn.

References

[1.] Drucker, P. "The Emerging Theory of Manufacturing." Harvard Business Review 70(3): 94-104, May-June 1990.

[2.] Laffel, G., and Bleumenthal, D. The Case for Using Industrial Quality Management Science in Health Care Organizations." JAMA 262(20):2869-73, Nov. 24, 1989.

[3.] Berwick, D. "Controlling Variation in Health Care." Medical Care 29(12):1212-25, Dec. 1991.

[4.] "Health Care Dollars." Consumer reports, July 1992, pp. 435-48.

[5.] Eddy, D. "Three Battles to Watch in the 1990s." JAMA 270(4):520-6, July 28, 1993

[6.] Morey, R., and others. "The Trade-off between Hospital Cost and Quality of Care." Medical Care 30(8):677-95, Aug. 1992.

[7.] Field, M., and Lohr, K., Eds. Guidelines for Clinical Practice: From Development to Use. Washington, D.C.: National Academy Press, 1992.

[8.] Battista, R., and Hodge, M. "Clinical Practice Guidelines: Between Science and Art." Canadian Medical Association Journal 148(3):385-9, Feb. 1993.

[9.] Cotton, P. "Determining More Good Than Harm Is Not Easy." JAMA 270(2):153-8, July 14, 1993.

[10.] Uttermohlen, D. The Horizontal Hospital: Restructuring for Patient-Focused Care." In Quest for Quality and Productivity in Health Services, 1993 Conference Proceedings, Institute of Industrial Engineers, Norcross Ga., Sept. 1993, pp. 109-116.

[11.] Bell, K., and others. Engineering the Traditional Hospital Organization. In Quest for Quality and Productivity in Health Services, 1993 Conference Proceedings, Institute of Industrial Engineers, Norcross, Ga., Sept. 1993, pp. 149-58.

[12.] Sharp, J., Ed. "Clinical Protocol Redesign Guide, Medical Center Regulation 168, 62, Wilford Hall Medical Center, Oct. 27, 1993.

John R. Sharp, MD, was Director, Professional Services, Wilford Hall Medical Center, Lackland AFB, Tex., when this article was written. He is now Vice President for Systems Integration, Santa Rosa Healthcare Corp., San Antonio, Tex.