Prolonged interruptions during continuous renal replacement therapy (CRRT) reduce delivered therapy and worsen clinical outcomes. The source states that daily downtime exceeding 20% of scheduled treatment time exacerbates acidosis and is associated with increased 28‑day mortality. This study aimed to develop and validate a simple bedside nomogram for early identification of CRRT treatment days at risk of suboptimal downtime control, defined in this analysis as cumulative downtime greater than 2.4 hours per standardized treatment day.
This was a retrospective cohort study conducted in a single intensive care unit in China over a five‑year period (January 2019–December 2023). The analysis used treatment‑day level data to assess predictors of suboptimal CRRT downtime and to construct a prediction model suitable for bedside use by ICU nurses.
The cohort comprised 145 patients contributing a total of 595 CRRT treatment days. Across those days the total recorded treatment time was 14,280 hours and cumulative downtime summed to 1,725.07 hours. Mean daily downtime was 2.90 ± 2.72 hours, corresponding to 12.08% ± 11.35% of scheduled treatment time. Suboptimal downtime control (cumulative downtime > 2.4 h/day) occurred on 232 treatment days, representing 39.0% of the dataset.
The binary outcome analyzed was whether a treatment day had cumulative downtime exceeding 2.4 hours. Candidate predictors included demographic data, laboratory parameters, CRRT circuit characteristics and other treatment factors (for example, filter replacements, catheter problems, extracorporeal procedures and patient factors such as agitation). The full text contains the complete list of candidate variables; the abstract reports the variables that emerged as significant in multivariable analysis.
Generalized estimating equations (GEE) were used to identify independent predictors of suboptimal CRRT downtime while accounting for repeated measures at the patient level. The resulting multivariable model was presented as a nomogram for bedside application. Model performance was assessed by area under the receiver operating characteristic curve (AUC) for discrimination, calibration curves, and decision curve analysis to assess net clinical benefit. The model was evaluated in a training set, an internal validation set and an external validation set, as reported in the abstract.
Multivariable GEE identified five independent risk factors associated with increased odds of a treatment day having cumulative downtime >2.4 h:
These predictors reflect both device/circuit issues (filter changes, catheter dysfunction) and patient or process factors (agitation, plasma exchange, transport), each independently associated with higher downtime risk in this cohort.
The nomogram constructed from the identified predictors demonstrated satisfactory discrimination and validation metrics in the datasets reported. Area under the curve (AUC) values reported in the abstract were:
Calibration curves and decision curve analyses were also performed and are reported as showing satisfactory calibration and net clinical benefit in the training and validation sets. Figures referenced in the source include the nomogram, ROC curves for each set, calibration curves and decision curve plots supporting these assessments.
The authors propose that ICU nurses can apply the nomogram at CRRT initiation to identify treatment days at elevated risk for suboptimal downtime control. Knowing which days are high risk could help prioritise interventions that target the identified factors, for example:
These measures are presented as ways to reduce cumulative downtime and preserve delivered CRRT dose; the nomogram is positioned as a bedside risk assessment tool to guide prioritisation.
In this retrospective single‑center cohort, five independent predictors of suboptimal CRRT downtime control were identified: daily number of filter replacements, agitation, plasma exchange, catheter dysfunction and out‑of‑unit transport. A nomogram based on these variables achieved AUCs of 0.789, 0.813 and 0.835 in the training, internal validation and external validation sets respectively, with reported satisfactory calibration and decision curve results. The authors conclude the nomogram offers ICU nurses a rapid bedside method to identify high‑risk treatment days and to prioritise interventions aimed at reducing CRRT downtime.
Note: the abstract reports study design, sample sizes, predictor variables that were significant and model performance metrics. Detailed model coefficients, operational thresholds for the nomogram and any additional limitations or subgroup analyses are available in the full text; if those specifics are not present in the abstract, they were not reported in the source material provided here.