In order to better understand latent tuberculosis infection (LTBI) screening practices in the U.S., the CDC National Center for HIV/AIDS, Viral Hepatitis, STD and TB Prevention funded this project using longitudinal data from a large national research network (OCHIN) to characterize the incidence of LTBI and TB, and also analyze the adherence to and relevance of current TB screening best practices in the primary care setting. The primary goals of this project were to: (1) evaluate the utility and feasibility of data generated through routine healthcare delivery as a surveillance tool for latent tuberculosis infection (LTBI); (2) provide accurate LTBI screening metrics by identifying patients at risk for LTBI using the PCORnet Common Data Model (CDM); and (3) identify appropriate partnerships and institutional mechanisms needed to transform data into improved action.
This study has been expanded to include efforts to define and validate EHR data related to TB diagnostic tests, TB and LTBI diagnoses, and treatment regimens. The primary method explored will apply machine learning (ML) algorithms to available EHR data to predict TB disease outcome ML offers an opportunity to improve prediction accuracy by discovering complex interactions between risk factors. ML models applied to EHR data have been used successfully to predict a patient’s risk of cardiovascular disease, diabetes, and mortality. The ML models in these studies identified who would benefit from preventative treatment; they identified unnecessary interventions among low risk patients. In addition, a simplified categorization of risk using predefined risk factors will be evaluated (e.g., non- US-born, immunocompromised, or a close contact to a TB case). This approach offers a simplified algorithm that is easier to communicate to providers and for providers and clinics to implement. Both the ML model and simplified risk guidelines offer opportunities to minimize disruption to clinic practice and to streamline clinical decision making for targeted TB/LTBI testing. This study offers an exploratory evaluation of possible interventions to improve targeted testing in community clinics that care for populations traditionally at higher risk for TB infection and subsequent development of TB disease.
In response to the need to identify latent tuberculosis infections in the US prior to their progression to tuberculosis cases, NNPHI is collaborating with CDC and OCHIN to work on the following project objectives: (1) develop informatics algorithms to accurately identify patients at high risk for developing TB who would benefit from an active LTBI screening program; (2) develop informatics algorithms to accurately identify patients who completed an LTBI treatment regimen; and (3) improve targeted TB testing and reporting of treatment outcomes through the development and implementation of an EMR module.
This project is supported by the Centers for Disease Control and Prevention of the U.S. Department of Health and Human Services (HHS) as part of a financial assistance award (NOFO OT18-1802, titled Strengthening Public Health Systems and Services through National Partnerships to Improve and Protect the Nation’s Health) totaling $300,000 with 100 percent funded by CDC/HHS. The contents are those of the author(s) and do not necessarily represent the official views of, nor an endorsement, by CDC/HHS, or the U.S. Government.