Health System Committees Should Carefully Vet Apps Built to Diagnose and Treat Patients
In the absence of rigorous preclinical testing from the Food & Drug Administration, health systems should carefully scrutinize digital tools that interact with electronic health records to recommend patient diagnoses or therapies, say experts at UC San Francisco.
Their commentary, published Tuesday, March 27, 2018, in the Annals of Internal Medicine, comes as the FDA is preparing its new “PreCertification” program that seeks to ensure the quality of digital health software, which evolves too quickly for the traditional testing protocols the FDA uses for drugs and “hard” medical devices. Under the new program, vendors whose standards for software development and attention to continuous improvement meet certain standards would receive a PreCertification stamp of approval, akin to TSA PreCheck.
Because PreCertification does not require evidence that the digital health tools improve outcomes, the UCSF authors say health systems should develop their own internal mechanisms for evaluating whether the tools are a good fit for local care. The authors recommend health systems create a “Digital Diagnostics and Therapeutics Committee,” a parallel to today’s Pharmacy and Therapeutics Committees, to consider whether each new tool is safe and effective. In the digital era, new tools should also be evaluated to ensure they protect patient privacy, improve physician workflow and efficiency, and offer clinical benefits that justify the costs of buying the software.
“Digital tools will arrive with the ‘pre-certified’ label, but we will still have to answer the local question of whether they will work for our patients and providers,” said Andrew Auerbach, MD, MPH, professor of medicine at UCSF. “The upside is huge with digital health innovations. You can do much more, much more efficiently, but how do we make sure we introduce them safely and effectively?”
Digital health vendors are writing software to perform a vast array of tasks, inside and outside the hospital. While tools like fitness trackers may not need extra scrutiny, the authors say more complex tools made to interact with electronic health records need careful oversight. These tools aim to simplify hospital care, provide decision support using genetic or other molecular analyses, or make data from patients’ day-to-day lives accessible to providers.
“It will be important for health systems to learn how to identify the digital tools that could most benefit their provision of care,” said Aaron Neinstein, MD, assistant professor of medicine and director of clinical informatics at the UCSF Center for Digital Health Innovation. “It won’t be easy to pick them out, so it is a skill set that health systems should develop.”
The authors gave three examples of digital health software inventions that hospital committees should consider carefully before adopting:
- A dashboard that creates goals for patients’ recovery in the hospital by taking information from hospital medical records, analyzing and integrating the data, and then re-displaying it on monitors in the patients' rooms and on their caregivers’ computers. This probably has a low risk of causing an error in care, but it could pose difficulties in ensuring patient privacy and data security.
- A service to provide personalized medication doses based on genetic tests that share this highly sensitive data with an outside company. In addition to the privacy risk, the system might overwhelm doctors with alerts about their patients. It also could be expensive to maintain, since it would need to be constantly updated with the latest genetic information.
- A service that recommends lifestyle changes for people with high blood pressure. It would have access to the patients medical record to prompt clinic visits and help guide patients’ treatment, based on their medical history, test results, demographic information, medication refills and home blood pressure readings. While it may represent an ideal digital health application, the researchers said this would be very risky, because it requires the integration of so much complex data. It would probably be based on proprietary algorithms that may rely on scanty or biased data, and its errors would be hard to catch, since its inner workings would be ‘invisible’ to patients and providers.
Even with the approaches being implemented at the federal level, the authors say, health systems will still need to maintain a critical role in evaluating and improving emerging digital technology.
“Hospitals will need to develop plans to evaluate these new tools, either before they’re put into place or by carefully monitoring them as they are implemented,” said Raman Khanna, MD, an associate professor of medicine and Medical Director of Informatics at UCSF Medical Center. “We’re optimistic that these new tools will improve the care that patients receive, help doctors and nurses do their work better and turn out to be worth the expense. But hospitals will have to do the hard work to make sure that happens.”
UC San Francisco (UCSF) is a leading university dedicated to promoting health worldwide through advanced biomedical research, graduate-level education in the life sciences and health professions, and excellence in patient care. It includes top ranked graduate schools of dentistry, medicine, nursing and pharmacy; a graduate division with nationally renowned programs in basic, biomedical, transitional and population sciences; and a preeminent biomedical research enterprise. It also includes UCSF Health, which comprises three top-ranked hospitals, UCSF Medical Center and UCSF Benioff Children's Hospitals in San Francisco and Oakland, and other partner and affiliated hospitals and healthcare providers throughout the Bay Area.