Understanding Medical Systems

The key organizing concept for an effective approach to improvement is the nature of a system. Such was the claim made by Berwick and Nolan in the article introducing this series [1]. Here, I aim to explore the nature of medical systems and how the understanding of systems can properly be applied to medicine and the activities of physicians. Two scenarios are used to ground the discussion and to illustrate some key principles of systems. 1. In an effort to reduce costs, a health maintenance organization places controls on the referral of patients from primary care physicians to specialists. Under this system, when a physician determines that a patient needs a consultation from a specialist, the request is sent to a nurse, who screens the request against criteria developed by the health maintenance organization. Most referrals are approved outright, some are approved after discussion with the physician, and some are rejected or postponed. For approved referrals, the nurse makes the appointment with the specialist and notifies the patient. The delay between the request for the consultation and the patient's appointment with the specialist ranges from 2 to 4 weeks. Patients often complain about the delay and the inconvenient times of their appointments. Physicians are angry that their clinical judgment is questioned by employees of the health maintenance organization who do not have the requisite knowledge. Why is there a 2- to 4-week delay and dissatisfaction on the part of both patients and physicians? 2. Five physicians share a specialty practice. Their patients frequently need diagnostic tests. These tests are sent to one of eight outside laboratories, depending on the patient's health plan. Each laboratory has a slightly different procedure for reporting the results to the practice. Options for reporting include fax, mail, and computer transfer to memory or printer. Each physician has a slightly different way of ordering the tests, requesting them from the laboratories, and communicating the results to patients. Sometimes, the results of the tests are not in the patient's chart when the physician needs them; this happens often enough to be vexing. The physicians are perplexed about the lack of reliability of the staff to perform such a simple task as getting laboratory results in a patient's chart. Why does a procedure as simple as filing laboratory results in patients' charts fail so often? Does the administrative staff understand the problems and delays that it causes for the physicians and patients? A glib answer to the questions posed at the end of these scenarios is that each system is structured to produce the results described, disappointing as they may be. This answer is an invitation to use a system perspective to investigate satisfactory alternatives to the pathologic structures in these and other components of the health care system. Understanding the Structure and Dynamics of Systems A system is defined as a collection of interdependent elements that interact to achieve a common purpose. Examples of systems include the process of treating patients with diabetes; a hospital; the development and testing of new medical procedures; and a medical practice, including locations of care, billing, and collection of fees. The diversity of these examples illustrates the arbitrariness of the description of a system. A description of a system of care for patients with diabetes, for example, is a logical construct that represents a particular point of view. What is included in and excluded from the description is strictly a manifestation of that point of view. Many people in health care organizations identify strongly with their individual professions or departments. Management structures, professional organizations, methods of billing for services, and research studies reinforce these divisions [2-4]. This fragmented environment allows the structure of medical systems to evolve piecemeal. Individual constituents of systems add elements in their spheres of influence to compensate for the perceived weaknesses of other constituents, thereby increasing the complexity of the system and substituting one problem for another. For example, a payer adds an approval step that must be completed before a patient can be referred to a specialist. In reaction, informal processes are created to circumvent the approval step or to resolve conflicts about the appropriateness of a referral. Improvement to a system results from new structures that are purposefully designed. To create functional systems, users must look beyond their professional or organizational identities [5]. Effective changes to the referral system in scenario 1 require that primary care physicians and specialists see the elements of their practices as part of the same system rather than as separate systems at odds with one another. The physicians and office staff in scenario 2 must recognize the common system in which they work to make progress on the problem of the missing laboratory results. Even a rudimentary understanding of the structures and dynamics of systems combined with clinical knowledge can equip a physician to diagnose the faults of a system and design remedies. Scenario 1: The System of Specialty Referrals The system of specialty referrals described contains an intermediate approval step deliberately placed between the specialist and the referring physician to prevent the inappropriate use of specialists. Adding this step is a weak intervention in the system that adds extra steps, causes delays, and devalues the clinical judgment of physicians. Physicians may agree with this assessment of the results and respond that the inspection step was added without their concurrence. They may suggest that administration simply remove the approval step. However, this is a facile answer from one part of the system to the difficult issue of providing good care at an affordable cost from the system as a whole. Although the approval step was probably added to reduce costly, inappropriate referrals, whether real or perceived, such steps are usually reliable markers of weaknesses in the system. However, removing the step would result in a simple return to the old system with its weaknesses intact. A more positive approach would be to redesign the system on the basis of clinical expertise and cooperation among the physicians involved. Consider a system without an approval step but in which 1) primary care and specialty physicians agree on a set of criteria for referrals, 2) specialists provide feedback to the primary care physicians on patients who were inappropriately referred to them according to the criteria, 3) specialists develop diagnostic and administrative aids for primary care physicians aimed at optimizing their referrals, and 4) administrators and physicians monitor the performance of the system by tracking the number and appropriateness of referrals. The structure of the redesigned system differs significantly from the original one. It does not have an approval step; it has a feedback loop from the specialists to the referring physicians to optimize the use of clinical knowledge throughout the system. Action to improve the appropriateness of the referrals is focused on the system as a whole and is done by examining patterns of referrals after the fact rather than by impeding an individual referral. Some organizations are using this improved structure as the basis for service agreements between primary care physicians and the specialists to whom they refer patients. The primary care physicians agree to acquire, with the help of the specialists, a set of core competencies to care for a specific patient population. The specialists agree to provide same-day access for any patient with a condition that the primary care physicians cannot diagnose or manage. Scenario 2: The System for Obtaining and Communicating Laboratory Results The second scenario involves the administrative tasks associated with getting a laboratory result into a patient's chart. It does not include the important clinical decisions of what laboratory tests are appropriate or how the results should be interpreted in the diagnostic process. Because the tasks involved are administrative, it is understandable that physicians may delegate the solution of the problem to administrative personnel. However, this view ignores the interdependence of all of the persons, including physicians, working in this system. Physicians tend to view administrative systems as simple and linear-order the test, send the sample to the laboratory, perform the test, communicate the results, file the results in the chart. Closer scrutiny of this system reveals great complexity caused by needless variation in the system. Physician preferences, styles, or lack of interest are some sources of this variation. Use of multiple laboratories for the same tests is another source. Standardization of the processes by which laboratory results are placed in the chart can reduce this variation. Further examination of the system also reveals that viewing it as linear is an oversimplification. Although the core of a system may be linear, many interacting tasks impinge on that core. For example, laboratory results may arrive while the chart is being used for another purpose. Organizations that have made progress on the issue of missing laboratory results seem to include some or all of the following components in the structure of their system: 1) a standard way for laboratories to report results; 2) synchronization of tasks, such as retrieval of charts from files, to the day before the patient's appointment; 3) a way to minimize the time that charts are out of the central filing area; and 4) a process for recognizing which charts are incomplete. The above scenarios illustrate several important principles for the improvement of systems. Physicians who are interested in expanding their knowledge of medical systems and how to improve the