FINAL EXAM: PROVIDER PAYMENT AND MANAGING CARE /PERFORMANCE INDICATORS

Measuring Structures, Processes, and Outcomes: Avedis Donabedian, an influential leader in the study of health care quality, developed a widely used, three-element model of quality measurement in 1966, which included measuring health care structures (the characteristics associated with a health care setting), processes (the activities done in a health care setting), and outcomes (the results achieved for a patient after a given set of interventions).16

• Structural measures include requirements imposed by payers and regulators, such as specifications for the physical plant, management systems, board certification, and staffing ratios.

• Process measures determine whether evidence-based care guidelines were followed, but do not indicate whether a patient’s health actually improved. Process measures, in essence, are used on the assumption that better outcomes should result from evidence-based care processes. Examples of process measures include the rate at which patients experiencing a heart attack are administered aspirin and beta-blockers.

• Outcome measures seek to determine whether the desired results are achieved. Examples of clinical outcome measures are whether a patient was readmitted to the hospital within 30 days of discharge and, for some conditions, whether the patient is alive at 30 days after admission. So-called “intermediate” or “surrogate” outcome measures are those that, while not true outcomes, are assumed to be able to be used as proxies for patient outcomes. For example, hemoglobin A1C blood test results are used both in research and practice as an indicator of whether diabetes is under control, because the results of the test correlate with the likelihood of experiencing diabetes complications. Measuring hemoglobin A1C on a periodic basis is a process measure, whereas achieving desirable hemoglobin A1C blood levels is sometimes labeled an intermediate outcome measure.

• Increasingly, quality experts also include various aspects of patients’ experiences as important outcome measures. Examples of patient experience instruments include the Patient Reported Outcomes Measures Information System, which includes modules that address physical health, mental health, and social health; HealthActCHQ, which has developed pediatric quality of life questionnaires, among others; and the Consumer Assessment of Healthcare Providers and Systems (CAHPS) surveys developed under the auspices of AHRQ.

Issues That Arise from Reliance on Structure and Process Measures: Structural measures, as described earlier, can include metrics such as the volume of a certain type of operation performed by a hospital. Such indicators can sometimes be a predictor of outcomes; for example, there is a literature that shows that for some procedures,

• institutions that do more procedures achieve better health outcomes,45 but the relationship between volume and outcomes is variable—by procedure and provider. Some quality ranking systems, like U.S. News and World Report’s rankings of hospitals, rely at least in part on structural criteria, such as nurse-to-patient bed ratios and availability of new technology. More commonly, structural criteria are included as survey questions to accredit or certify that a provider meets threshold standards to be included as a recipient of program funds. For example, “conditions of participation” establish the structural quality and safety standards that all U.S. hospitals must follow to participate in Medicare and Medicaid.

Meanwhile, process measures—which are the most common type of quality measures—calculate the rate at which a recommended clinical or care process is performed. By one estimate, of 78 HEDIS measures for 2010, all but five were clearly process measures, and none were true outcomes measures.46 Process measures have several theoretical advantages over outcome measures.

• First, calculating process measures is more straightforward because in some cases (for example, when evaluating physician prescribing) there may be less need for risk-adjustment to account for case mix differences that clearly affect outcomes. Yet, for process measures that evaluate patient adherence to treatment recommendations, for example, there may be a need for risk adjustment to account for relevant socioeconomic differences.47

• Second, process measures typically reflect professional standards of care. As a result, they are most often subject to evidence-based, professional standard setting that is readily understood by clinicians. In contrast, the factors that often contribute to different outcomes across institutions include organizational culture, leadership, teamwork, technology, and other factors not part of professional standards of care that clinicians and other individuals can readily control.48 Thus, process measures are “actionable”—that is, the measure itself prescribes the action that the clinician, institution, or health plan needs to take to improve performance.49 Feedback to clinicians is more personally relevant and thus easier to act on.

• Finally, practically, there is often a large research base that provides evidence on which processes reliably improve particular outcomes, although the studies do not always cover all populations of interest, especially the elderly.