Methods to identify saline-contaminated electrolyte profiles

Selection of Single-Analyte Delta Check Rules with Logistic Regression for Detection of Intravenous Fluid Contamination in a Clinical Chemistry Laboratory

BACKGROUND

The conventional single-analyte delta check, utilized for identifying intravenous fluid contamination and other preanalytical errors, is known to flag many specimens reflecting true patient status changes. This study aimed to derive delta check rules that more accurately identify contamination.

METHODS

Results for calcium, creatinine, glucose, sodium, and potassium were retrieved from 326 103 basic or comprehensive metabolic panels tested between February 2021 and January 2022. In total, 7934 specimens showed substantial result changes, of which 1489 were labeled as either contaminated or non-contaminated based on chart review. These labeled specimens were used to derive logistic regression models and to select the most predictive single-analyte delta checks for 4 common contaminants. Their collective performance was evaluated using a test data set from October 2023 comprising 14 717 specimens.

RESULTS

The most predictive single-analyte delta checks included a calcium change by ≤-24% for both saline and Plasma-Lyte A contamination, a potassium increase by ≥3.0 mmol/L for potassium contamination, and a glucose increase by ≥400 mg/dL (22.2 mmol/L) for dextrose contamination. In the training data sets, multi-analyte logistic regression models performed better than single-analyte delta checks. In the test data set, logistic regression models and single-analyte delta checks demonstrated collective alert rates of 0.58% (95% CI, 0.46%-0.71%) and 0.60% (95% CI, 0.49%-0.74%), respectively, along with collective positive predictive values of 79% (95% CI, 70%-89%) and 77% (95% CI, 68%-87%).

CONCLUSIONS

Single-analyte delta checks selected by logistic regression demonstrated a low false alert rate.

New solutions to old problems: A practical approach to identify samples with intravenous fluid contamination in clinical laboratories

OBJECTIVES

Contamination with intravenous (IV) fluids is a common cause of specimen rejection or erroneous results in hospitalized patients. Identification of contaminated samples can be difficult. Common measures such as failed delta checks may not be adequately sensitive nor specific. This study aimed to determine detection criteria using commonly ordered tests to identify IV fluid contamination and validate the use of these criteria.

METHODS

Confirmed contaminated and non-contaminated samples were used to identify patterns in laboratory results to develop criteria to detect IV fluid contamination. The proposed criteria were implemented at a tertiary care hospital laboratory to assess performance prospectively for 6 months, and applied to retrospective chemistry results from 3 hospitals and 1 community lab to determine feasibility and flagging rates. The algorithm was also tested at an external institution for transferability.

RESULTS

The proposed algorithm had a positive predictive value of 92 %, negative predictive value of 91 % and overall agreement of 92 % when two or more criteria are met (n = 214). The flagging rates were 0.03 % to 0.07 % for hospital and 0.003 % for community laboratories.

CONCLUSIONS

The proposed algorithm identified true contamination with low false flagging rates in tertiary care urban hospital laboratories. Retrospective and prospective analysis suggest the algorithm is suitable for implementation in clinical laboratories to identify samples with possible IV fluid contamination for further investigation.

Automating the Detection of IV Fluid Contamination Using Unsupervised Machine Learning

BACKGROUND

Intravenous (IV) fluid contamination is a common cause of preanalytical error that can delay or misguide treatment decisions, leading to patient harm. Current approaches for detecting contamination rely on delta checks, which require a prior result, or manual technologist intervention, which is inefficient and vulnerable to human error. Supervised machine learning may provide a means to detect contamination, but its implementation is hindered by its reliance on expert-labeled training data. An automated approach that is accurate, reproducible, and practical is needed.

METHODS

A total of 25 747 291 basic metabolic panel (BMP) results from 312 721 patients were obtained from the laboratory information system (LIS). A Uniform Manifold Approximation and Projection (UMAP) model was trained and tested using a combination of real patient data and simulated IV fluid contamination. To provide an objective metric for classification, an "enrichment score" was derived and its performance assessed. Our current workflow was compared to UMAP predictions using expert chart review.

RESULTS

UMAP embeddings from real patient results demonstrated outliers suspicious for IV fluid contamination when compared with the simulated contamination's embeddings. At a flag rate of 3 per 1000 results, the positive predictive value (PPV) was adjudicated to be 0.78 from 100 consecutive positive predictions. Of these, 58 were previously undetected by our current clinical workflows, with 49 BMPs displaying a total of 56 critical results.

CONCLUSIONS

Accurate and automatable detection of IV fluid contamination in BMP results is achievable without curating expertly labeled training data.

GPT-4 Underperforms Experts in Detecting IV Fluid Contamination

BACKGROUND

Specimens contaminated with intravenous (IV) fluids are common in clinical laboratories. Current methods for detecting contamination rely on insensitive and workflow-disrupting delta checks or manual technologist review. Herein, we assessed the utility of large language models for detecting contamination by IV crystalloids and compared its performance to multiple, but variably trained healthcare personnel (HCP).

METHODS

Contamination of basic metabolic panels was simulated using 0.9% normal saline (NS), with (n = 30) and without (n = 30) 5% dextrose (D5NS), at mixture ratios of 0.10 and 0.25. A multimodal language model (GPT-4) and a diverse panel of 8 HCP were asked to adjudicate between real and contaminated results. Classification performance, mixture quantification, and confidence was compared by Wilcoxon rank sum.

RESULTS

The 95% CIs for accuracy were 0.57-0.71 vs 0.73-0.80 for GPT-4 and HCP, respectively, on the NS set and 0.57-0.57 vs 0.73-0.80 on the D5NS set. HCP overestimated severity of contamination in the 0.10 mixture group (95% CI of estimate error, 0.05-0.20) for both fluids, while GPT-4 markedly overestimated the D5NS mixture at both ratios (0.16-0.33 for NS, 0.11-0.35 for D5NS). There was no correlation between reported confidence and likelihood of a correct classification.

CONCLUSIONS

GPT-4 is less accurate than trained HCP for detecting IV fluid contamination of basic metabolic panel results. However, trained individuals were imperfect at identifying contaminated specimens implying the need for novel, automated tools for its detection.