Welcome to episode four, Measurement- Why and What. The choice of what area or areas to measure is clearly a critical step. Even if the measures are being imposed from the outside, there may be some leeway in choosing which parameters or which areas to focus most of your effort and time on. Besides situations where measures are an imposed requirement from outsider practice, or reporting measures provides an opportunity for extra payment. Starting a program of measurement is usually directed towards gathering information. That might guide future quality improvement efforts or at least shed some light in a concern or a problem area that you've identified. Ideas we put into measurement areas may come to light through many different channels, including your own observation and the potential area of concern. From comments or concern raised by patients, by family members or your staff. If you were planning a measurement effort and you have a full latitude in choosing measures, it's usually best to resist the initial temptation to create a new measure specifically for use in your setting. While this sometimes may be necessary, given the lessons we've learned in episode three about what constitutes a useful measure. It's not easy to develop a measure that will actually get you the answer you need or intend. The good news is it that there are number of good sources for potential measures. One very useful and very easily searchable site is the national quality measures clearing house but you all call the clearing house. The clearing house is a publicly accessible resource reported by the US agency for healthcare research in quality or ARQ. While the clearing house is the most readily accessible site, there are other resources available such as the European Commission's healthcare indicator site. Also the World Healthcare Organization periodically produces monographs and other publications of indicators that are helpful in measurement of diseases like tuberculosis or other global health issues. The listing on a clearing house includes a searchable function that allows searching by disease or problem. Includes a full description of a measure, the type of measure, the organization that developed and manages the measure, the evidence based for the measure and other information critical to potential use of measure in your site. It cannot be overstated that for most situations, using measures that were well constructed and fully documented is a critical early step in successful measurement. Performance Measures listed on the clearing house, have clearly stated specifications. So that if applied and very careful and precise manner, there is a reasonable probability that the same data will be collected in the same way. Which in turn increases the chances that the information will in fact be useful and comfortable. There are additional issues to consider as you move forward within a grading measuring into your own practice. Most of these considerations are looking at how some of the desirable attributes or measures apply directly to your own site and interest in measurement. And how the measures you are contemplating are likely to actually perform in your practice. Is the data that you will need likely to be accurate, accessible, and complete in your own medical records? Will there be a sufficient number of cases in the practice to draw reliable and valid conclusions? Making valid comparisons from data collected at a prior date or to data from bench marking outside your practice is very dependent on having a sufficient sample size. In other words, having enough patients or instances of a problem that's actually being studied. For example, if you are in an outpatient practice and are concerned about the care of patients with rheumatoid arthritis. You will need to ask yourself whether or not you see a sufficient number of patients with RA to have an adequate sample size to answer the question that you wish to answer. While there are power tables available that indicate the number of patients needed to have a certain probability of finding a significant difference between two groups of patients. It is rare that valid comparisons can be done with fewer than 30 patients. In order to directly compare one practice to another, it may take several hundred patients to find significant differences using most measures. In the example we used, the prevalence of rheumatoid arthritis is estimated at about 1%. So that in a family practice of 2000 patients, one might expect about 20 patients with RA, likely not enough for valid measurement. You will also need to think about whether you will be able to attribute the variation that you find in the measure, to the care you actually provide. In other words can variation that you find be associated with the healthcare practice, that is actually controllable within your practice. In the example of rheumatoid arthritis, interventions made by consultants may have a much larger impact than any intervention that you might do in your own practice. In addition to sample size and variation, there are a number of issues related to the data itself that must be considered. First do you have access to the data you'll need to actually collect? Is the data readily available in existing medical records, either paper or electronic? Or are you going to need to create and enter new data, or even enter existing information in new ways to the patient records? How expensive will it be to add data versus just using the data that's already available. Second, how will the data actually be collected? What's the source of the data? Hard copy charts, electronic health records or other data sources? Will the data also be collected through observation and interview? While electronic heath records or EHRs can provide a useful and efficient way of collecting data for measures, there are major challenges. To mention just two, the completeness of data and the accuracy of the algorithms setup to extract and organize the data. There may be inconsistencies across systems in how EHRs are organized that may affect the data extraction. Potentially, these sort of details will ensure that the data obtain as relevant to what you are trying to measure. Another factor to consider is the quality of the data, are the data entered in the same way by all providers, are there many omissions? Who will collect and organize data? If hospital based you need to get a hospital quality improvement group together to do this? Or ambulatory care based, do you have someone with the skills to collect and analyze the data? What steps will need to be taken to collect the data? Minimizing the number of steps will reduce the likelihood of problem analyze. You also will need to consider how much will the data collection and analysis cost. Are there potential unintended consequences of using these measures? As you can see there is a lot to think about to not only select the right measures but to apply the measures and actual practice in the right way. The questions we have posed will hopefully help you participate in deciding or at least influencing, which measures to use in monitoring care, in any practice setting, discipline or geographic location. While the questions may seem extensive, they often can be addressed fairly easily, especially if you are only gathering data for a project within your own practice. Note that it almost always gets better and easier after the initial project. Among your resources you will find a measured selection checklist that you may find helpful. Being thoughtful in your approach will help you avoid the pitfalls of choosing the wrong measure, or applying the wrong measure in the wrong way. Remember that the immediate goal is choosing the right measure and applying it to the right issue in the right way. Note finally, that measures and measurement is only one critical step along the path to actually improving the care provided. Final observation in this regard measuring quality doesn't improve anything, just as weighing a steer does not add any beef.