quire several steps. Objective and reproducible outcomes must be defined, these outcomes must be adjusted for patient-specific risk factors, and the role of specific processes of care in producing these outcomes must be evaluated. Generally, outcomes should be related to specific patient conditions, diagnoses, and problems. Because the knowledge, skills, and systems are not yet available to put this program in place for a broad set of conditions and care settings, it will necessarily have to develop incrementally.
Operationally, we picture a local or regional unit, not unlike the current PROs, that would be responsible for collecting systematic information on patient outcomes and care. We begin with hospitals and inpatient care, because such facilities address some of the larger problems in medicine. We would immediately include other forms of care that substitute for inpatient hospital care, such as ambulatory surgery. We would then extend this approach to other forms and settings of care—nursing home, home care, and ambulatory care—as quickly as feasible, allowing for more technical development in these areas.
Systematic information on patients’ outcomes would be collected across a number of dimensions. The MPAQ and MQROs must choose outcomes that are easily and reproducibly defined, are feasible to obtain, and are important to Medicare beneficiaries. These outcomes could include mortality and medical complications, relevant physiologic measures, functional outcomes (such as patients’ mental and emotional status), physical capabilities (such as the ability to walk or climb stairs), activities of daily living (such as bathing, dressing, feeding, and toileting), placement of the patient at home or in a long-term-care facility, and patients’ and families’ satisfaction with care.
A very difficult aspect of outcome-directed quality assurance efforts will be to adjust outcomes for the risk factors present in the population being studied. This will be necessary to ensure that comparisons of the outcomes of patients who are treated by different physicians, groups, and hospitals or who are covered by different plans are appropriate.
The choice of the initial conditions to be studied must reflect the availability of information about known risk factors. For example, few data are available that predict the mortality, morbidity, loss of function, development of confusion, or discharge site of elderly patients admitted to a hospital with pneumonia. For patients with a hip fracture, by contrast, numerous authorities agree that mental status, functional status before the fracture, associated medical conditions, age, sex, and race affect both the mortality and recovery of function. Thus, a study of patients with hip fractures can adjust for risk factors more effectively than a study of elderly patients hospitalized for pneumonia.
Effective outcome studies must ensure that information on all relevant risk factors is identified and collected at the time that care is provided (e.g.,