Measure outcomes for well-defined populations. Outcomes should be measured for all patients within a well-defined population segment (such as those with coronary artery disease) regardless of how their condition is treated (such as with percutaneous coronary intervention, bypass surgery, or pharmaceuticals). In value-based health care, it is fundamental to compare the outcomes of different clinical teams rather than the outcomes of different procedures, although the latter may ultimately provide an important explanation for differences in the results achieved. For example, disease registries that focus on a particular procedure and compare outcomes across centers have played a very important role in improving hip and knee arthroplasty, providing critical insights into the most effective implants and surgical techniques. However, data that compares varying procedures cannot answer the broader question about the optimal treatment for the underlying disease of osteoarthritis. Both perspectives are valuable, but measuring outcomes based on the medical condition will challenge the current treatment paradigm and enable innovative changes in clinical practice.
Measure outcomes across the full cycle of care. Metrics should track every stage of a patient’s journey, including prevention, diagnosis, treatment, recovery, follow-up, and long-term well-being. It is also important to make sure that measurements track the full range of care providers, including primary care, specialized care, and rehabilitation. This ensures that insight is gained into the complete range of outcomes for a patient group and that differences in health outcomes caused by variations in the configuration of health systems or care pathways can be assessed.
Define outcomes based on what matters most to patients. There can be a stark difference between what providers and registries measure and what patients actually care about. For example, measures of the clinical process—such as the number of admissions, length of stay, and interventions—often do not matter as much to patients as do measures of quality of life, functional ability, and emotional well-being. Exhibit 2 demonstrates the extent of this mismatch for breast cancer patients. The primary concerns for breast cancer patients are worrying about the future, being tired, and health insurance or money worries.
Direct reports by patients on the status of their own health can help determine whether treatment culminates in outcomes that patients care about. For this purpose, many providers rely upon patient-reported outcomes measures (PROMs). The Martini-Klinik, in Hamburg, Germany, uses Web surveys to follow up on clinical outcomes for prostate cancer treatment; response rates are typically above 80 percent. Because PROMs capture a patient’s personal, unfiltered assessment, with limited demand on the clinician’s time (and untainted by the clinician’s influence or interpretation), these tools provide a useful complement to clinical measurements and assessments.
Patients’ perspectives are critical when defining outcomes. By including patients, several leading organizations have established targeted, patient-centered outcomes metrics for specific diseases. All 12 outcomes standard sets published by ICHOM, for example, have been developed by working groups that include patient representatives. Some hospitals, too, have conferred with patients when defining outcomes metrics. For instance, in a working group for bipolar disorder at the Sahlgrenska University Hospital, in Gothenburg, Sweden, a patient suggested that self-sufficiency was a significant desired outcome. Bipolar disorder is often linked to damaged relationships, poor job or school performance, and even suicide. But when correctly treated, patients living with this illness can lead full and productive lives. By measuring self-sufficiency as an outcome, physicians in Sweden can evaluate how well patients are coping with the disease and whether they need additional support.
Choose measures that are already standardized and included in registries before working to create new metrics when possible. The more comparable the outcomes data among providers, the more useful they are. Physicians in Australia, Sweden, the U.S., and other countries are using comprehensive outcomes data collected in national disease registries to identify outliers and improve average outcomes. Indeed, ICHOM plans to expand its current 12 metrics to cover more than 50 conditions by 2017, representing approximately 70 percent of the disease burden in industrialized countries. These standard sets can—and should—be leveraged by providers and payers. By selecting standardized outcomes metrics, health care systems can compare outcomes and variations in medical practice within a hospital or across provider networks regionally, nationally, and internationally. With standardized metrics, providers have a broader base for identifying best practices. We therefore recommend that organizations begin by reviewing existing standards and then, if needed, complement them with additional metrics that the providers want to follow (because of, for example, a specific local research or development effort).
Prioritize the most important outcomes—and avoid selecting too many. It is important to prioritize and select the metrics that will have the biggest impact, because requiring clinical teams to track too many of them will burden the teams with an overly complex data-collection process, which could lead to limited compliance and poor-quality reporting. We recommend selecting seven to ten metrics for analysis per patient group. Of course, additional metrics can be tracked to further analyze and understand the results, but they should not be prioritized for steering.
In general, select measures that can be tracked easily, but do not choose convenience over relevance. When starting out, it may make sense to select certain outcomes measures that can be tracked easily to gain traction with systematic monitoring and analysis. But working toward the goal of tracking the most relevant measures is important, even if it requires more effort.
Once outcomes measures have been selected, make sure to avoid any ambiguity in the implementation. The team must define in advance the exact patient group that will be included in the measures, which tools will be used (such as doctor-reported data, electronic medical records, PROMs, and registries), how data will be collected, who will report the data, and when the data will be gathered, analyzed, and reported. Defining all of these factors up front removes ambiguity during implementation and increases the likelihood that comparative analysis will yield high-quality data.