Why a registry?
What is a registry? Many practices will be asking themselves this question as they move toward becoming patient-centered medical homes or just work to improve the care they deliver. In its simplest form, a registry is a list of patients who share certain characteristics – all patients with diabetes, for example – with information on their status. To continue the example, a practice with a diabetes registry should be able to consult it to see quickly which patients are up-to-date on eye exams or have reached their LDL goals. Most registries today are capable of tracking multiple clinical data points across many patients. The AAFP offers a number of resources concerning registries, and Googling “chronic disease registry” will give you a plethora of additional examples and advice.
Why would a practice want to start a registry? Reasons vary. One practice may decide it wants to send reminder letters to all patients with diabetes who have not had a documented eye exam in the past year. Another may want to incorporate reminders for eye exams into quarterly diabetes office visits. One practice in the National Demonstration Project (NDP) used the CINA product to print a complete sheet of services due at each visit for each patient. They used the sheet as a template to guide the visit and then gave the sheet to the patient to take home as a visit summary. The practice was able to customize their registry to pull only the clinical information they were interested in knowing.
Registries come in different types with different functionalities. The simplest ones, but by no means the easiest to use, are freestanding databases that require manual entry of all data; the work of keeping the registry up-to-date can make these challenging to implement. Probably the easiest ones to use are parts of EHRs and mine data directly from the EHR. More complicated (and usually more functional) ones interface with the EMR, scheduling software, billing software and/or other databases and mine data from them. The data identifies patients or groups who are due services related to their diagnosis codes.
Registries are only as good as the data entered or imported into them. Most will import only structured data – data stored with each separate data point (e.g., A1C level) in the same "box" in each record. Free text data, which many physicians are more comfortable entering, is not stored as discrete data points and therefore will not import easily into a tool like a registry. Before you implement any sort of registry, make sure as much of the relevant data as possible is entered into your EHR as structured data. Several of the NDP practices delayed starting registries because the technology was not really ready. Many systems required the physician or staff to “double enter” data in the EMR and then again in the registry. To read more about the implementation process in the NDP, see the two final TransforMed reports from the project evaluators: "Preliminary Answers to Policy-Relevant Questions" and the "More Preliminary Answers" follow-up report.
What about practices that don't have electronic records? Registries can be done in Excel or even on paper. Practices that saw the benefit of managing a disease state through their patient population jumped into doing this before computerizing. Typically, the practice chooses what clinical data it wants to track, then writes it down in an easily identifiable location in each patient’s chart. Data points are collected into another file that is searched and acted upon periodically. For example, say a practice wants to track influenza immunization rates in patients with diabetes. Each patient with diabetes could have a yellow sticker on their chart. When the nurse pulls a chart with a yellow sticker, she knows to go to the immunization flow sheet (the same in each chart) and look for the immunization. (Examples of flow sheets are available in the FPM Toolbox). If the influenza immunization has not been given, she gives it by standing order for the clinic. If it has not been given, is contraindicated, or the patient refuses, she notes that. Then, an index card with the patient’s name and yes or no to influenza and the date is filed in a box. At the end of the flu season, the practice can calculate compliance rates and decide how they want to handle immunizations the next year. They could take the registry one step further and pull all the charts that have yellow stickers but do not have a card in the file and recall those patients.
Building a registry, whether electronically or manually, takes time. If you're interested in starting, first decide on what data you will track and what your practice will do with the data. Start simple as in the examples above. Try to track too much data, and you'll be at it forever. It is nearly impossible to track all parameters available to a primary care practice. Then devise a way to structure the data you need at the point of data entry. When enough time has passed for you to have a reasonably complete data set, put the data to use. Generally, the output of a registry is one or more reports that require some work to be useful. The practice decides who will work the list and how (e.g., by recalling patients overdue for services). Finally, don’t forget to take time to review the process and software for problems and ways to improve the next time.