Hello again. In the past lessons, I have discussed topics related to creating useful information from administrative health care data. First, I made a few points about how administrative data differ from clinical data. Second, using imperfect administrative data, I reviewed how to use clinical identification algorithms to identify cohorts of patients who have specific medical conditions. Third, I reviewed critical concepts associated with creating quality measures. Now, let's briefly review some quality measures that have been endorsed by the National Quality Forum or NQF and that are commonly used among health care organizations. We will focus on HEDIS and AHRQ or A-H-R-Q measures. By the end of this lesson, you will be able to articulate both the costs and benefits of implementing HEDIS and AHRQ quality measures. Let's get started with HEDIS. The first set of quality measures we'll cover are HEDIS measures. H-E-D-I-S stands for Health care Effectiveness Data and Information Set. HEDIS measures are used by more than 90% of health plans in the United States to measure various aspects of health care. In the current version, there are 92 measures across six domains of care. They are generally process measures and thus most of them are not risk adjusted. The argument in favor of using HEDIS measures is that they allow people to make apples-to-apples comparisons between health plans or provider groups. As a result, benchmarks and ratings can be set that likely help drive improvement. The algorithms are conceptually simple but since they rely on complex administrative or clinical data, they are often complex in their details. HEDIS measures are carefully documented in commercially sold books and software. In my experience, HEDIS measures are imperfect and various stakeholders sometimes complain that they are not perfect. I argue though that perfection is often the enemy of the good and that an imperfect HEDIS measure is better than using no measure at all. Once again, all models are wrong, but some are useful. HEDIS measures the models of complex processes but they can likely help us simplify processes to gain understandings of health care quality. Here's a quick example of a HEDIS measure, adult BMI, this is body mass index. Let's consider the measure description. This is the percentage of members 18 to 74 years of age who had an outpatient visit and whose BMI was documented during the measurement year or the year prior to the measurement year. Now, let's consider the eligibility population or the denominator. This includes product lines, ages, continuous enrollment, allowable gaps, anchor dates, benefits, and diagnoses. The numerator in this example is the BMI event from the HEDIS algorithm as defined by the value set codes. As you can see, the description is quite concrete and easy to understand but the details for the denominator of the numerator are fairly complicated and involve a variety of terms. Next, look at this table showing quality indicators from the Agency for Health care Research and Quality or AHRQ. There's a great deal of clear information about the AHRQ measures. Let's just focus on one here, the prevention quality indicators or the PQIs as shown here in the top row. The PQIs are a set of measures that can be used with hospital inpatient discharge data to identify quality of care for Ambulatory Care Sensitive Conditions. These are conditions for which good outpatient care can potentially prevent the need for hospitalization or for which earlier intervention can prevent complications or more severe diseases. The PQIs are population-based and adjusted for covariates. I encourage you to read about these measures in detail. I think the A-H-R-Q or AHRQ measures are a great introduction to implementing real high value quality metrics. I offer here the following steps that are used to produce the AHRQ measures. First, you need to access the algorithms. This is easy to do from the AHRQ web page, then you need to transform the raw inpatient hospital data into the input data required by the algorithms. Next, assuming all goes well, you can run the algorithms on the data. Of course, the software requires some modification to the computer code that is provided to you. When you get the results, you need to review the validity of your results. Finally, you need to assess the possibility of data quality issues that are associated with your results. This last step might be the most important. The AHRQ measures were originally used with administrative data. More recently how the measures have been implemented using clinical data from EHR systems. Meaningful use has been an important force encouraging this implementation with clinical data. Meaningful use is a government policy that encourages health systems to develop certified electronic health measures to improve quality safety efficiency, engage patients, improve care coordination, and improve public health. There are also areas of maintaining privacy and security of patient health information. Overall, the area of meaningful use is going to result in better clinical outcomes that should improve population health outcomes. This should also increase transparency and efficiency, once again empowering individuals, and this should also lead to more robust research on health data systems. I think all of this data preparation for using data in the clinical domain will lead to greater use and development of clinical quality measures. So, you might ask why is it important to consider clinical quality measures that are implemented using clinical data from EHRs? First, administrative data often lack critical elements. Second, because of meaningful use and other programs, clinical data are starting to be stored and preprocessed in ways that makes it easier to implement clinical quality metrics. Finally, EHR vendors are starting to sell quality modules within their applications. These are so-called eCQMs for electronic clinical quality measures. The eCQMs or clinical measure algorithms that produce the measure outputs within EHRs, and it affect, automate many of the quality metric implementations. Of course, the metrics need to be validated to ensure that complex clinical data are preprocessed correctly. All this said, administrative data continue to be important in the clinical quality measurement revolution is just getting started. Of course, you can ask yourself the question, are clinical data ready to be mined for clinical quality measurement? I think the answer is yes for many organizations. Fortunately, much of the knowledge can be then leveraged from administrative measures. For example, even though the data might be different, a lot of the concepts and the numerators and denominators along with the exclusion rules can be quite similar between the measures. Overall, I think we should look forward to a revolution with the development of clinical quality metrics. I hope you've enjoyed this lesson on reliable health care quality measures. I look forward to our discussions coming up. In our next lesson, we will turn our attention to analytical groupers.
