New tool speeds up diagnosis and treatment of dehydration.
A medical provider observes a patient’s sunken eyes. They rest their fingers on the patient’s wrist, noting the rush of their pulse and the rhythmic rise and fall of their chest. They pinch the skin, assessing elasticity. Finally, they examine the urine for a darkened hue.
In less than a minute, health care providers using the most basic version of the FluidCalc clinical decision support tool (CDST), developed in part by Brown researchers, can assess a patient’s diarrhea-induced dehydration status and determine what treatment they should receive. In a recent study, the tool outperformed a current diagnostic standard—and has potential to improve recognition and treatment of this medical emergency.
Diarrheal diseases are the eighth leading cause of death worldwide and the second leading cause of death in children under 5, according to the World Health Organization. The study authors specifically assessed the CDST for efficacy among populations in low-income countries, which are most vulnerable.
“The vast majority of almost all diarrheal deaths occur in low-resource settings in low- and middle-income countries, so having a tool to improve management is much more important in those settings,” says first author Adam C. Levine, MD, MPH, a professor of emergency medicine and of health services, policy, and practice at Brown.
Treatment includes giving fluids to patients to rehydrate them. Often they can simply drink more fluids, but in more extreme cases survival is contingent on injecting fluids into the veins—though this process requires more time and resources and poses potentially fatal side effects, Levine says. In a setting with scarce resources, the goal is to most effectively treat those who need it.
The current method to evaluate dehydration from diarrhea was developed by the WHO but has not been validated, says Levine, who is also Brown’s associate dean of global health equity. Other validated assessments have proved effective only in very young children.
Levine set out to develop a tool for patients over 5, including adults, back in 2018; the findings on the efficacy of the CDST in a medical setting were published in The Lancet Global Health in September. The team included researchers from Brown, the University of Florida, and the International Centre for Diarrhoeal Disease Research in Bangladesh (icddr,b).
The tool uses a machine learning, logistic regression model, which was developed by studying more than 2,000 patients at the icddr,b to determine correlations between clinical characteristics and patient status. The team created two versions of the tool: a full model that uses medical devices and a simplified version that does not, Levine says.
The scoring tools were validated in Bangladesh by comparing the CDST’s accuracy with that of the existing WHO algorithm and the more precise method of directly assessing changes in patients’ weights. Both versions of the CDST were comparably accurate and produced better results than the WHO algorithm.
Since publishing their findings, the team has been working to make the tool available on their website, fluidcalc.org, and as a mobile app. Levine hopes to validate the model in other low-resource settings. Co-authors Nur Haque Alam, MD, and Mohsena Bint-E Sharif, MBBS, add that further study is needed using modified conditions and additional settings before deploying the CDST clinically in Bangladesh.
The researchers predict the tool will “prevent hundreds of millions of patients from being overtreated … which will save countries a lot of money and resources,” Levine says. “If we can get this tool to be used on a large scale, we estimate that it will save hundreds of thousands of lives each year.”
