The future of health care is value-based care. If Value equals Quality divided by Cost, then a defined, validated way to measure Quality is paramount to that equation. (Fortunately, Cost comes with convenient measurement units called dollars.) Payers now are asking health care providers to shift from a fee-for-service to a value-based reimbursement structure to encourage providers to deliver the best care at the lowest cost. Providers who can embrace this data-driven paradigm will succeed in this new environment.
So how do we define high-quality care? What makes a good quality measure? How do you actually measure what happens in a clinical encounter that impacts health outcomes?
To answer these questions, organizations have constructed standardized clinical quality measures. Clinical quality measures facilitate value-based care by providing a metric on which to measure a patient’s quality of care. They can be used 1) to decrease the overuse, underuse, and misuse of health care services and 2) to measure patient engagement and satisfaction with care.
What are quality measures?
The Academy of Medicine (formerly named the Institute of Medicine) defines health care quality as “the degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge.”1
Clearly defined components and terminology. From a quantitative standpoint, quality measures must have a clearly defined numerator and denominator and appropriate inclusions, exclusions, and exceptions. These components need to be expressed clearly in terms of publicly available terminologies, such as ICD (International Classification of Diseases) codes or SNOMED CT (Systematized Nomenclature of Medicine—Clinical Terms) terms. A measure that asks if “antihypertensive meds” have been given will not nearly be as specific as one that asks if “labetalol IV, or hydralazine IV, or nifedipine SL” has been administered. The decision to tie the data elements in a measure to administrative data, such as ICD codes, or to clinical data, such as SNOMED CT, also affects how these measures can be calculated.
Moving targets. The target of the measure also must carefully be considered. Quality measures can be used to evaluate care across the full range of health care settings—from individual providers, to care teams, to hospitals and hospital systems, to health plans. While some measures easily can be assigned to a specific provider, others are not as straightforward. For example, who gets assigned the cesarean delivery when a midwife turns the case over to an obstetrician?
Timeframe in outcomes measurement. The data infrastructure is currently set up to support measurement of immediate events, 30-day or 90-day episodes, and health insurance plan member years. Longer-term outcomes, such as over 5- and 10- year periods, are out of reach for most measures. To obtain an accurate view of the impact of medical interventions or disease conditions, however, it will be important to follow patients over time. For example, to know the failure rate of intrauterine systems, sterilization, or hormonal contraceptives, it is important to be able to track pregnancy occurrence during use of these methods for longer than 90 days. Failures can occur years after a method is initiated.
Another example is to create a performance measure focused on the overall improvement in quality of life and costs related to different treatments for abnormal uterine bleeding. How does the patient experience vary over time between treatment with hormonal contraception, endometrial ablation, or hysterectomy? Which option is most “valuable” over time when the patient experience and the cost are assessed for more than a 90-day episode? These important questions need to be answered as we maneuver into a value-based health system.
Risk adjustment. Quality measures also may need to be risk adjusted. The “My patients are sicker” refrain must be accounted for with full transparency and based on the best available data. Quality measures can be adjusted using an Observed/Expected factor, which helps to account for complicated cases.2
Clearly, social and behavioral determinants of health also play a role in these adjustments, but it can be more challenging to acquire the data elements needed for those types of adjustments. Including these data enables us to evaluate health disparities between populations, both demographically and socioeconomically.3 This is important for future development of minority inclusive quality measures. Some racial and ethnic minority populations have poorer health outcomes from preventable and treatable diseases. Evidence shows that these groups have differences in access to health care, quality of care, and health measures, including life expectancy and maternal mortality. Access to clinical data through quality measures allows for these health disparities to be brought into quantifiable perspective and assists in the development of future incentive programs to combat health inequalities and provide improved delivery of care.