Abstract: SA-PO0080
Surveillance System for Monitoring Dysnatremias During Saline Shortage
Session Information
- Artificial Intelligence: Kidney Imaging, Pathology, and Precision Phenotyping
October 24, 2026 | Location: Exhibit Hall A, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- Alahmadi, Ziad, Johns Hopkins University, Baltimore, Maryland, United States
- Bitzel, Jack, Johns Hopkins University, Baltimore, Maryland, United States
- Menez, Steven, Johns Hopkins University, Baltimore, Maryland, United States
- Thiessen Philbrook, Heather, Johns Hopkins University, Baltimore, Maryland, United States
- Parikh, Chirag R., Johns Hopkins University, Baltimore, Maryland, United States
Background
Shortages of Intravenous fluids (IVF) may disrupt clinical practices in the United States and impact ordering patterns of IVF. In September 2024, Hurricane Helene created supply chain challenges affecting IVF availability. At Johns Hopkins Health System (JHHS), both lactated ringer's and normal saline supplies were impacted. Weekly supply availability and location-specific were provided to conserve supplies while minimizing care disruption. We examined changes of patterns at JHHS with interrupted time series (ITS) analysis to evaluate whether such disruptions were associated with measurable changes in clinical outcomes.
Methods
Utilizing data from electronic medical records (EMR), we evaluated IVF usage and associated serum sodium values at JHHS =. The analysis was divided into a baseline period (May 1–October 31, 2024) preceding the IVF supply disruption, and a post-intervention period (November 1–December 20, 2024). Sodium measurements were categorized into non-mutually exclusive dysnatremia thresholds. Weekly ITS models were used to estimate immediate level changes and subsequent trend changes following the intervention.
Results
Among 541,568 sodium measurements, (424,899 at baseline and 116,669 post-intervention) we found weekly models demonstrated a significant step decrease in the proportion of sodium values ≤130 mmol/L and ≤135 mmol/L alongside a corresponding increase in values ≥145 mmol/L (Figure 1). Changes in more extreme dysnatremia categories (≤125, ≥150, ≥155 [SM1] mmol/L) were not statistically significant. Although statistically significant changes were identified, clinically meaningful changes were not observed and the risk of moderate to severe dysnatremias did not reach statistical significance.
Conclusion
We were able to set up a dynamic surveillance system for monitoring IV fluid shortage and its effect on electrolytes. Routine monitoring may help ensure electrolyte homeostasis and development of fluid protocols during supply disruptions. Data science methods could be leveraged to integrate this approach into EMR systems to passively identify unintended clinical consequences because of changes in the supply chain.