Defining episodes of care. Measurements of the process of care hinge on the ability to define and measure an episode of care.6 Such definition is occasionally possible, for example, with a series of vaccinations or the diagnosis and complete healing of a fracture. Typically, reviewers must make arbitrary decisions about the beginning and end of an episode and about the relevance of intervening office visits, emergency room visits, or telephone calls involving one or more (known) practitioners. Such arbitrary decision making reflects our lack of understanding of the natural history of most ambulatory conditions. It is difficult to judge whether a series of visits by the same patient for similar problems represents (1) continued appropriate treatment (one episode of care), (2) an independent recurrence (two episodes), or (3) a failure of treatment (one episode). Many efforts at assessment avoid the issue of an episode of care and base the assessment on some other unit of analysis, such as a single-patient contact, completion of a health screening protocol, or a patient’s summary assessment of satisfaction with care.
Incomplete and fragmented data systems. Case-by-case review of the process of care using ambulatory records is often seriously compromised by poor organization of data and lack of documentation of key aspects of care in the medical record. The degree of documentation in the medical record is said to vary inversely with the experience of the clinician—the most extensive notes reportedly are written by medical students. Where used the problem-oriented medical record (POMRs) might help a reviewer to evaluate the medical logic of care (such as the relationship between a set of patient complaints and their diagnostic evaluation), but the use of the POMR in its complete format is rare.
In general, clinical data systems in ambulatory care are poorly developed, and billing systems rarely include clinical information. The Uniform Ambulatory Care Data Set, approved in June 1989 by the National Committee on Vital and Health Statistics, includes certain clinical data (DHHS, 1981; Felts, 1989). In addition to patient and provider identifiers, patient demographic descriptors, the date and place of encounter, and itemized charges, recommended encounter information would include: problem; diagnosis or assessment; laboratory, radiology, and surgical services; and the patient’s reason for encounter (optional). If this minimal data set is widely adopted, claims data might be suitable for initial screening in ambulatory care for those who have access to and the expertise to analyze these data.
Even beyond the difficulties in assessing care from the review of one chart, patients may independently see multiple providers, with little or no communication or record linkage among them. In the hospital setting, practitioners, patients, and their records are gathered in one place; in the ambu-