Clinical registries are important in monitoring healthcare processes and outcomes, and for establishing treatment benchmarks [1]. Data quality in orthopaedic registries however, remains an ongoing challenge for evidence-based practice change. A key barrier is the complexity of clinical registries across multiple users and systems, and as the complexity of the registry increases, data quality is reduced by transcription issues, logical inconsistencies, missing information, duplicate records and measurement errors [2]. Monitoring patient outcomes in the clinic alone can prove difficult, and is even more challenging for larger registries with specific eligibility criteria, defined cohorts, a high patient volume and small windows in time to follow-up patients.
So what can be done to improve registry capture rates, and ensure that patients and their data are not lost in the archives of multiple, not-quite-compatible systems? With lessons learned from implementing five clinical orthopaedic registries in a variety of clinical environments, we developed a solution to streamline the registry workflow, in the hopes of minimising administrative load and associated human error.
How did we perform? Read on to learn about the processes we put in place for an orthopaedic department in a medium-size metropolitan hospital, established to help one surgeon track six observational cohorts of shoulder and knee pathologies.
The process
The process starts with weekly patient lists. These include a waiting list with over 6000 patients and a surgical list of over 1300 patients, that are generated by the hospital. Of course, these lists are flooded with duplicate records and irrelevant information, so the first step is to clean the lists, which leads to a reduction of ~20%, and makes it easier to filter by surgeon and date (for the coming week) to arrive at a list of ~100 patients. In the end, only 1% of more than 7000 entries are relevant - imagine performing that task manually!
Parallel to this process, an audit framework is also applied to the patients already in the registry, to identify if they are due for a postoperative follow-up. The two lists are matched to compile the final consult list, from which new patients are either added to the registry, and returning patients have their follow-up data collected.
That's three inputs from two systems with 7000+ entries, condensed to final list of 100 with a few button clicks that take <10 minutes of user effort.
Of course, the exports and code running in the back end take a little longer, but we don't recommend that anyone sit and watch the wheels turn.
The maintennance
No process is viable without appropriate quality control measures, and so we put in a quality control framework to run checks on the data. The archived consult lists are routinely compared to the research registry to cross-check patient unit record numbers (URN), protocol/cohort labels and affected side, and any inconsistencies are flagged for manual review.
This not only ensures that all patients are captured within the registry, but also verifies that the information stored for them is accurate. Discrepancies are routinely rectified, and thus errors do not accumulate.
The result
So what does registry capture look like, for a quality-controlled registry supported by our custom solution to enhance its processes? We've been managing this registry for two years, and so can confidently report a first-pass accuracy of 83-100% across the six cohorts. After routine quality checks in the last reporting period, this accuracy increased by up to 17%, ensuring 100% capture rates across all cohorts.
This means that every patient eligible for research has been consented, correctly identified and added to the registry in the appropriate cohort.
Given that this registry is for a department that performs 1000+ elective surgeries each year, we think the numbers speak for themselves.
Of course, the effort and iterations that have gone into developing these processes are challenging and time consuming. Navigating multiple software packages and different IT environments is never easy, and is often slowed by regulatory frameworks. The good news is that you may never have to worry about establishing these systems yourself - let us know whether you are looking to customise your registry or streamline your existing processes, and we can help get the wheels turning for you too!
References
Evans S.M., et al., Med J Aust . 194 (7), 360-363, 2011.
Hickey G.L., et al. European Journal of Cardio-Thoracic Surgery. 44 (2013) 605-614.
Comments