Translating periodontal data to knowledge in a learning health system

Each of the participating institutions treats a diverse population, including private practices, specialty clinics, and teaching clinics. Institutional Review Board approval from the University of Texas Health Science Center at Houston was obtained to conduct our study. Strengthening the Reporting of Observational Studies in Epidemiology guidelines were followed.

the critical data elements for periodontal diagnoses and new tooth loss due to periodontal disease were itemized. Next, an electronic script was developed to locate and extract each element from the appropriate section of the EHR. The measure summary and respective denominator and numerator specifications are presented in Table 1. Data were extracted from 3 calendar years (2017, 2018, 2019). Data included all qualifying visits in the year and the required data from the prior examination that may have occurred up to 3 years earlier.

they can facilitate the tracking of key dental outcomes.

^,

Increased adoption of EHRs has provided the tools to efficiently extract useful data for performance measures, assess the relationships between these measures and health outcomes, and benchmark population health.

Population health is defined as “the health outcomes of a group of individuals, including the distribution of such outcomes within a group,”

and EHRs already provide some access to public health data to study the population for potential health improvements and act as a safety net for potential health threats.

Until now, the clinical periodontology literature has mostly represented studies that use surrogate end points.

), with more than 95% use of structured diagnoses recorded across each site. Oral health care providers are also required to complete a standardized periodontal assessment form, in which periodontal indexes and risk factors are documented in a structured format. Previous work has assessed the quality of periodontal assessment documentation recorded within the EHR and its ability to identify patients who had received a diagnosis of periodontal disease, complete periodontal charting and periodontal risk factors.

We built on that and reported the development of 2 periodontal outcome measures using a large EHR data set; our approach is scalable and transferable to other important disease markers and oral health outcome measures. The 2 outcome measures we presented can be implemented to assess success in the prevention and treatment of periodontal disease, and to facilitate learning and improvement. Team members have also created checklists, notifications, and other ways to integrate these data into clinicians’ workflow. The aim was to make this information accessible to improve best practices at the point of care. Our research team called this the “Rate [measuring and properly articulating the data], Communicate [presenting the data in a way that is understandable to the appropriate audience], Motivate [implementing reward systems for changed behavior and improved performance], and Iterate [to re-enter the cycle in order to foster continuous improvement]” model.

report affirmed that by 2035 there will be more older adults than youth in the United States. This aging US population is also more dentate. Consequently, there has never been a more important time to pay attention to periodontal health and to strategies that help us learn from clinical data to improve it. The structured data found within the EHR can help promote measure automation, which can ease the implementation process. It is easy to envision these types of practice-level quality measures aiding patients and providers alike.

Clinical information from EHRs can be excessive and is often scattered and hard for clinicians and policy makers to access. Using an interactive dashboard tool for the processing and presentation of large amount of information allows stakeholders to explore the data on different levels. In our study, we depicted a static representation of an interactive dashboard. In actual use, filters can be used to narrow down to a particular clinic in the institution or network. Clicking on a particular quarter or month can allow for other panels to be filtered to show the characteristics for that period only. Likewise, clicking on a risk factor can filter the time series to show comparative performance across the different dimensions. Additional dashboard actions can trigger patient-level detail across time.

These include missing data, erroneous data, uninterpretable data, and inconsistencies in the way data are recorded among providers and over time. Another issue is that patients often receive care from multiple providers using fragmented and often poorly integrated EHR systems, making it difficult to completely track patients across practices

or systems. EHR information exchange, which allows health systems to access and share EHR data across organizational and geographic boundaries, needs continued enhancement and dissemination to increase the value of EHR data. In addition, critical clinical data are often recorded in unstructured, narrative text, complicating its use for learning and improvement. In our work, we relied on structured data. For dental clinics that do not use a standardized diagnosis terminology, innovations in methods such as natural language processing may be used to capture such unstructured data from the clinical narratives in the EHRs.

^,

In addition, effective user interfaces can improve the ease and consistency of data entry, which simultaneously reduces user burden and decreases the amount of unstructured, and potentially uninterpretable, data in the EHR. Finally, there is a major need for rigorous EHR evaluation and data optimization to ensure valid and usable information.

^, 

^, 

is a consortium of 11 dental schools that share EHR data for research and quality improvement and are developing the components of an LHS. We also recognize that EHR data alone may not be inclusive enough to conduct meaningful learning. Rather, the value of EHR data might be realized when linked to other data sources, such as patient-reported behaviors, quantified self-data, and clinical trial data. As value-based care predominates, EHR data will occupy a central role in generating meaningful knowledge in support of LHS for improving oral health.