Diagnostic Criteria for Refeeding Syndrome Determine Reported Incidence: A Systematic Review and Meta-Analysis

OBJECTIVES: Refeeding syndrome (RFS) is characterized by hypophosphatemia, hypomagnesemia, and hypokalemia, and is prevalent in patients receiving nourishment after a long delay. Clinical symptoms develop when severe abnormalities occur. Despite a sizable literature, definitions lack validation. We aim to understand the impact of diagnostic criteria on reported incidence, as part of a larger effort to create a valid definition. METHODS: This is a systematic review of observational studies following PRISMA guidelines, analyzed using random effects modeling. Studies that reported RFS incidence qualified for analysis. Extracted information included RFS incidence, operationalized definitions, forms of nourishment at feeding resumption, and participant characteristics. RESULTS: In a total of 49 qualifying studies, populations most often studied were patients with restrictive eating disorders, malignancy, and dysphagia. RFS incidence varied across studies, ranging from 0 to 0.80. The overall pooled RFS incidence was 0.23 (0.20, 0.26). Pooled data of the 49 studies showed great heterogeneity (I2 > 0.90; P = 0.00). Large variations in incidence were seen among studies utilizing different RFS definitions. Subgroup analyses by diagnostic criteria revealed lower RFS incidence in studies with definitions requiring clinical symptoms and all electrolyte abnormalities (0.01 [0.00, 0.02]; 5 studies) and clinical symptoms and any electrolyte abnormalities (0.01 [−0.01, 0.02]; 2 studies) compared with studies requiring all electrolyte abnormalities and no clinical symptoms (0.13[0.03, 0.23]; 3 studies) and any electrolyte abnormalities and no clinical symptoms (0.29 [0.24, 0.33]; 39 studies). Heterogeneity remained large in subgroup analyses by diagnostic criteria (I2 > 0.90). CONCLUSIONS: Incidence of RFS varied greatly according to the definition used, supporting the need for a unified definition. Incidence was extremely low in studies with RFS definitions requiring clinical symptoms. This may be due to close monitoring and treatment of low electrolyte levels in the study settings. This supports the recent guideline that recommends the diagnosis of RFS not require the presence of symptoms. Further subgroup analyses are needed to identify additional sources of incidence heterogeneity, and new diagnostic criteria need validation. FUNDING SOURCES: None