Welcome to episode two, framework for measures. Let's take a trip. Let's say you want to travel from San Francisco to Miami? What measures will you use to assess the quality of your trip? Well, it may be tempting to jump to the outcome. Did you get to Miami? Did you make it in time? You allowed yourself to get there and within your budget. In addition to outcomes, there may be important structure and process measures you would want to consider to gain a more comprehensive assessment of your trip. Structural measures might include the type and size of the vehicle you travelled in. Process measures might include the route you chose, was it an easy drive with interesting scenery? How many stops did you make? You might also want to assess how comfortable the seats were in the car, or how interesting your companions were, or the overall experience of your trip. Of course, measuring healthcare is a little bit more complicated than measuring a car trip, but structure, process, outcome remain a useful framework. So let's take a deeper dive now into each type of measure. Structure measures relate to the capabilities in assets of an organization that are presumably related to producing high quality care. An advantage of structural measures is that they're relatively easy to understand, to measure, and often to act upon. They also have some face validity. For example, it might be hard to imagine high quality hospital care being provided in a run down facility with inadequate equipment and poorly trained staff. Most structural measures are used in large organizations and had historically been the backbone of accreditation programs. Some commonly used measures of structure include staff to patient ratio. Is it adequate to meet the needs of patients? Staff competency. Do staff have the right training for their jobs? Well defined policies. Are policies in place to guide care decisions and set expectations? Adequacy of environment. Do operating rooms and patient rooms have specific amounts of space and equipment? Structural measures also get to fine levels of details such as the availability of safe needle disposal devices. To be a useful healthcare measure, there should be strong empirical data linking low levels of performance on a structural measure to poor outcomes of care. If for some reason, it's possible to provide high quality outcomes with lower levels of staff or poorly maintained technology, for example, the measure's not useful to improving care. Unfortunately, in many cases, there is an assumed, rather than empirically proven, link between structure and outcome measures. Which has, in many instances, diminished the perceived value of these measures. However, it should be noted that where there is either very strong face validity or a proven link, structural measures can be foundational. Let's turn to process measures. The US Agency for Health Care Research and Quality, or AHRQ, defines a process as a health care related activity performed for, on behalf of, or by a patient. Process measures provide information about whether or not a certain action has or has not taken place, such as a medication given, a blood pressure taken, an immunization provided, or a test ordered. As with structural measures, the utility of a process measure depends on it being linked to an important outcome. For example, giving an immunization that was not at all effective in preventing influenza would not be a useful measure of quality care for influenza. We also have to be careful to determine if there are other equally effective processes that can create the same outcome. For example, a phone call might be just as effective as a visit for follow up of a change in medication. When there are proven links between process measures and outcomes, process measures may be a very useful adjunct to outcome measures because they are often easier to gather and are usually directly actionable. For example, noting and intervening to raise a low rate and using a checklist to prevent central line infection, is more directly actionable than simply knowing you have a central line infection rate that's high. Measures of process have traditionally been used to look for situations where underuse of interventions or processes are likely to lead to poor outcomes. Examples include underuse of screening for colon cancer, breast or cervical cancer, or diabetes or hypertension, which all lead to failure of early diagnosis and treatment. In addition to gauging underuse, process measures can also be used to determine the potential overuse of an intervention. For example, if a given facility does screening papsmears at a rate of one every six months, the frequency can at least raise questions as to whether the procedure is being over used as compared to national and safety guidelines. Process measures can also show inappropriate use. An example of inappropriate use is the widespread use of antibiotics to treat a viral upper respiratory infection. Okay, let's move into outcome measures. An outcome is simply defined as the health state of a patient resulting from health care. It is simply what happens to a patient as a result of some process or treatment. There are many outcomes that can be measured, including but certainly not limited to mortality, morbidity, length of stay, patient satisfaction, and outcomes associated with specific treatments. There's a great appeal in looking at outcome. We apply a treatment and we simply look at what happened. We want to know that our patients have gotten to the health state that is the goal. However, in many respects, outcome is often quite complex. For example, the outcome of coronary bypass surgery could be defined as the patient being alive at the end of a procedure, at six weeks, at one year or at ten years, or free from angina at one year, or some combination of these outcomes. There are several major challenges to relying on outcome measures. First, there are relatively few situations where intervention is the only factor determining the outcome. For example, success in treating hypertension has been shown to be dependent on many factors including adherence to medication, diet, and other factors not directly under the control of healthcare providers or the healthcare system. In addition, age, sex, and severity of illness are all factors that can also affect outcomes. Statistical methods for taking these factors into account are referred to as risk adjustment. In order to risk adjust, there needs to be a large population to apply the measure to. When developing and applying outcome measures, statisticians are indeed your best friends in determining methods that lead to valid outcome measures. Another challenge in measuring outcomes is that the distance in time between the treatment, such as control of blood pressure, and any definitive outcome, such as heart attack, stroke or death, can be decades. While focusing on what is often termed an intermediate outcome is helpful, there is a less than perfect correlation between blood pressure control and mortality, other than at very high levels of blood pressure. Even for procedures like surgery, there's often a fairly long gap between the treatment and outcome desired that makes gathering out come measures challenging. For example, it would be difficult to measure the distance a patient can walk without pain at six months after knee surgery. In addition, most outcome measures are not directly actionable. For example, a high rate of reoperation after bariatric surgery does not indicate what problem it actually is causing the high rate of reoperation. Another consideration with outcome measures is that patient attributes, such as weight, can influence outcomes. The patient experience of care is another important set of measures that are usually classified as outcome measures. Namely, measures related to patient reported outcomes of care including their experiences in receiving healthcare. These measures gather information directly from patients and can include everything form their opinions about parking to appointment access, whether they felt listened to and respected, as well as physical outcomes such as how far they can walk without pain, or if they can shop, cook, or use public transportation independently. The centers for Medicare and Medicaid in the US requires hospitals to report measures of patient experience of care. These measures are reported through a public website on hospital quality called Hospital Compare. There are also surveys that assess patient experience of care for other settings, including nursing homes, dialysis centers, surgical centers, ambulatory care and more. The Agency for Healthcare Reasearch and Quality maintains a website with information about a variety of tools to assess patient experience of care that are in the public domain and can be downloaded. So additional information is included with the resources for this model. To show how structure, process, and outcomes measures could work together, let's look at an example of an older patient who needs hip replacement surgery. In order for the surgery to be successful, one might start with structural measures that are foundational to high quality hip surgery, including having a surgeon credentialed, at least by the hospital, and an operating room with appropriate equipment that has been reviewed for safety. Process measures might include the use of appropriate surgical procedure as well as the correct delivery of anesthesia. Outcome measures could include no complications during the surgery or hospital stay. That the patient being able to walk one block without pain after a specified period of time, and the patient reporting the experience of care as respectful and responsive to their needs. While these outcome measures alone might suffice to uncover instances per quality, it would still take measures of process or structure to pinpoint the problems. Moreover, as an illustration of the importance of structure and process measures, an unqualified surgeon, or an ill-equipped hospital, would put the patient and health facility at risk for an avoidable harm. There are growing numbers of sets of measures that are being used to measure quality in specific areas. One of the early data sets that continues to evolve is the healthcare effectiveness data and information set, otherwise known as Hedis. Hedis was developed and is maintained by the National Committee for Quality Assurance and Hedis is used by more than 90% of America's health plans to measure performance on important dimensions of care and service. The Hedis performance measures are standardized so that health plans can be compared. The agency for health care research and quality supports a website called The National Quality Measures Clearing House, that has a listing of measure with their full descriptions. The Joint commission has the Oryx measures that are integrated into their accreditation process for hospitals. In addition, many professional organizations have developed measures related to their specific area of practice, such as radiology physician performance measurement set and the nursing database of nursing quality indicators. There are also measures related to specific disease management, such as diabetes, as well as measures related to delivery systems, such as accountable care organizations and health plans as previously noted. In addition payers, such as Medicare, require that certain measures be reported as part of payment requirements. These data sets are specific to settings. Many private payers require the reporting of individual provider data, and that data may be used to determine whether the provider will continue to be part of that payer's network. So in choosing a measure, given the large number of measures that have been developed and which are now required for reporting in the US and many other countries, some providers, as you might understand, are feeling overwhelmed with data collection requirements and with trying to sort out which results are useful and helpful in guiding quality improvement. While some of this confusion and burden is due to the relatively early stage of quality measurement improvement, there's clearly a great need to re-examine what information is really critical and to begin to streamline reporting requirements, and to include only those measures that are important and useful and harmonize the measures where possible. Harmonize means that similar measures, those with the same measure focus or with the same population focus, or definitions applicable to many different measures will have the same specifications so that the same information can be used for all required reporting and will produce results that are comparable. The National Quality Forum has established guidelines for harmonization of measures and continues to work in this very important area. As you can see, there are many different types of measures and measure sets from many different aspects of care. With so many measures to choose from, it's important to have criteria for selecting the right measures for a given setting or situation. In the next episode, we'll examine the attributes of a useful measure. So, see you soon in the next episode.