From Potassium Snapshots to Time in Range During Diabetic Ketoacidosis: A Three-Cohort Study

Background Potassium management during diabetic ketoacidosis (DKA) is inherently dynamic, yet observational studies commonly characterize potassium exposure using isolated values or discrete episodes of dyskalemia. Whether time spent within the guideline-recommended potassium range captures reproducible clinical information during early DKA treatment is uncertain. Methods We conducted a multicohort observational study using MIMIC-IV, eICU, and NWICU. Adults with DKA were eligible if they had at least three serum potassium measurements during the first 24 hours of ICU care and at least 12 hours between the first and last eligible measurement. Potassium time in target range (K-TIR) was the proportion of the observed monitoring interval in which linearly interpolated serum potassium was within the prespecified guideline-anchored range of 4.0–5.0 mmol/L. The primary outcome was ICU length of stay of at least 72 hours. The same prespecified multivariable model was fitted independently in each database, and adjusted effects per 10-percentage-point increase in K-TIR were synthesized by random-effects meta-analysis. Restricted cubic splines assessed nonlinearity. Results The final cohorts comprised 1,672 ICU stays; primary complete-case models included 1,639 stays and 381 prolonged-stay events. Adjusted odds ratios (ORs) per 10-percentage-point increase in K-TIR were below 1.0 in all three databases: 0.977 (95% CI, 0.932–1.025) in MIMIC-IV, 0.935 (95% CI, 0.811–1.079) in eICU, and 0.929 (95% CI, 0.853–1.010) in NWICU. The pooled OR was 0.963 (95% CI, 0.925–1.002; P = 0.062; I2 = 0.0%), indicating directional concordance across cohorts but statistical uncertainty around the magnitude of association. Restricted cubic splines showed no evidence of nonlinearity. For invasive mechanical ventilation, the pooled OR was 0.956 (95% CI, 0.912–1.002; P = 0.062; I2 = 0.0%). A cross-database adjusted mortality estimate was not estimable because of sparse events and incomplete outcome availability. Conclusions Guideline-defined K-TIR can be operationalized as a longitudinal phenotype of early potassium management in DKA. Its association with prolonged ICU stay was directionally consistent across three independent critical-care databases, but the pooled estimate remained compatible with no association. These findings support further evaluation of K-TIR as a measurement framework while leaving unresolved whether it is a marker of recovery or a modifiable treatment target.

Authors

Publication Details

Journal
Research Square
Published
2026-11-19
DOI
https://doi.org/10.21203/rs.3.rs-11285580/v1
Primary Topic
Potassium and Related Disorders
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

From Potassium Snapshots to Time in Range During Diabetic Ketoacidosis: A Three-Cohort Study

Zheng Yao, Shijia Zhu, Zhaotao Fang
Research Square
Potassium and Related Disorders
preprint

From Potassium Snapshots to Time in Range During Diabetic Ketoacidosis: A Three-Cohort Study

Zheng Yao, Shijia Zhu, Zhaotao Fang
preprint en

Abstract

Background Potassium management during diabetic ketoacidosis (DKA) is inherently dynamic, yet observational studies commonly characterize potassium exposure using isolated values or discrete episodes of dyskalemia. Whether time spent within the guideline-recommended potassium range captures reproducible clinical information during early DKA treatment is uncertain. Methods We conducted a multicohort observational study using MIMIC-IV, eICU, and NWICU. Adults with DKA were eligible if they had at least three serum potassium measurements during the first 24 hours of ICU care and at least 12 hours between the first and last eligible measurement. Potassium time in target range (K-TIR) was the proportion of the observed monitoring interval in which linearly interpolated serum potassium was within the prespecified guideline-anchored range of 4.0–5.0 mmol/L. The primary outcome was ICU length of stay of at least 72 hours. The same prespecified multivariable model was fitted independently in each database, and adjusted effects per 10-percentage-point increase in K-TIR were synthesized by random-effects meta-analysis. Restricted cubic splines assessed nonlinearity. Results The final cohorts comprised 1,672 ICU stays; primary complete-case models included 1,639 stays and 381 prolonged-stay events. Adjusted odds ratios (ORs) per 10-percentage-point increase in K-TIR were below 1.0 in all three databases: 0.977 (95% CI, 0.932–1.025) in MIMIC-IV, 0.935 (95% CI, 0.811–1.079) in eICU, and 0.929 (95% CI, 0.853–1.010) in NWICU. The pooled OR was 0.963 (95% CI, 0.925–1.002; P = 0.062; I2 = 0.0%), indicating directional concordance across cohorts but statistical uncertainty around the magnitude of association. Restricted cubic splines showed no evidence of nonlinearity. For invasive mechanical ventilation, the pooled OR was 0.956 (95% CI, 0.912–1.002; P = 0.062; I2 = 0.0%). A cross-database adjusted mortality estimate was not estimable because of sparse events and incomplete outcome availability. Conclusions Guideline-defined K-TIR can be operationalized as a longitudinal phenotype of early potassium management in DKA. Its association with prolonged ICU stay was directionally consistent across three independent critical-care databases, but the pooled estimate remained compatible with no association. These findings support further evaluation of K-TIR as a measurement framework while leaving unresolved whether it is a marker of recovery or a modifiable treatment target.

Research Square
Potassium and Related Disorders
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon