Cardiorespiratory dynamics measured from continuous ECG monitoring improves detection of deterioration in acute care patients: A retrospective cohort study

Abstract

Background: Charted vital signs and laboratory results represent intermittent samples of a patient's dynamic physiologic state and have been used to calculate early warning scores to identify patients at risk of clinical deterioration. We hypothesized that the addition of cardiorespiratory dynamics measured from continuous electrocardiography (ECG) monitoring to intermittently sampled data improves the predictive validity of models trained to detect clinical deterioration prior to intensive care unit (ICU) transfer or unanticipated death.

Methods and findings: We analyzed 63 patient-years of ECG data from 8,105 acute care patient admissions at a tertiary care academic medical center. We developed models to predict deterioration resulting in ICU transfer or unanticipated death within the next 24 hours using either vital signs, laboratory results, or cardiorespiratory dynamics from continuous ECG monitoring and also evaluated models using all available data sources. We calculated the predictive validity (C-statistic), the net reclassification improvement, and the probability of achieving the difference in likelihood ratio χ2 for the additional degrees of freedom. The primary outcome occurred 755 times in 586 admissions (7%). We analyzed 395 clinical deteriorations with continuous ECG data in the 24 hours prior to an event. Using only continuous ECG measures resulted in a C-statistic of 0.65, similar to models using only laboratory results and vital signs (0.63 and 0.69 respectively). Addition of continuous ECG measures to models using conventional measurements improved the C-statistic by 0.01 and 0.07; a model integrating all data sources had a C-statistic of 0.73 with categorical net reclassification improvement of 0.09 for a change of 1 decile in risk. The difference in likelihood ratio χ2 between integrated models with and without cardiorespiratory dynamics was 2158 (p value: <0.001).

Conclusions: Cardiorespiratory dynamics from continuous ECG monitoring detect clinical deterioration in acute care patients and improve performance of conventional models that use only laboratory results and vital signs.

Fig 1. Flowsheet of analysis.

Fig 2. Histograms of the time between measurements, logarithmic scale.

The area of each histogram indicates the quantity of measurements available. R wave to R wave (RR) inter-beat and inter-breath intervals (gray) occur on the order of seconds, 2.6 billion heart beats and 600 million breaths. (Middle Inset) Features from the continuous ECG monitors (green) were aggregated at q15 minute increments. (Upper Right Inset) Standard nursing vital signs (yellow) were most frequently obtained at every 1 to 4 hour intervals, 4.1 million measurements. Automated entries into the EMR are evident at every 1 minute. Laboratory results (red) were most frequently measured every 12 hours or daily.

Fig 3. Distributions of measured heart rate and respiratory rate.

Heart rate (left) and respiratory rate (right) measurement distributions according to source, charted vital signs (VS; blue) vs electrocardiography (ECG; red) where each source had equal numbers of measurements.

Fig 4. Relative statistical significance of component predictors included.

Heatmap depiction of statistical significance of predictors (rows) in each model (columns) with corresponding C-statistics. Saturation or transparency depicts the statistical significance as calculated by the square root of (χ² minus degrees of freedom (df)) and can be likened to the absolute value of a Z score adjusted for the degrees of freedom associated with a predictor due to higher order terms. Hue or color represents the data source: individual data sources (green), pairwise combinations (blue), fully integrated model using all available data (red). Predictors that failed to achieve statistical significance in any of the models are not displayed. VS-HR: charted heart rate, VS-RR: charted respiratory rate, O2.Flow: charted oxygen flow rate, SpO2: charted oxygen saturation, GCS: Glasgow Coma Scale, SBP: systolic blood pressure, WBC: white blood cell count, BUN: blood urea nitrogen, AST: aspartate aminotransferase, Plts: platelets, CO2: carbon dioxide, Glu: glucose, pCO2: partial pressure of carbon dioxide, K: potassium, Na: sodium, sCr: creatinine, ECG-SNR: Signal-to-Noise metric from ECG-derived respirations, ECG-HR: heart rate from continuous ECG, HR-D: trend of HR over prior 24 hours, ECG-RR: ECG-derived respiratory rate, RR-D: trend of RR over prior 24 hours, AF: atrial fibrillation, EMR: electronic medical record.

Fig 5. Schematic of sequential integration of additional data sources.

Each circle represents an individual model with the C-statistic reported below each model abbreviation. Area of each circle is proportionate to the total likelihood ratio test χ² minus two times the degrees of freedom (df) of all non-intercept terms.

Fig 6. Example cases of deterioration.

A: Patient post-procedure day #1 from stenting of left posterior tibial artery for non-healing ulcer deteriorated into mixed cardiogenic and septic shock. Here charted vital signs were sufficient to identify deterioration in the several hours preceding ICU transfer. B: Patient with heart failure undergoing evaluation for coronary artery bypass grafting quickly deteriorated due to acute renal failure. Progressive derangement of laboratory results identified the deterioration without appreciable change in charted vital signs or continuous ECG monitoring. C: Patient post-operative day #3 from left renal vein transposition who developed abdominal pain with associated tachycardia. CT imaging demonstrated intraabdominal hemorrhage and subsequent hemoglobin level had dropped from 9.6 to 6.0. EMR charted vital signs and lab results failed to appreciate a trend that was apparent for several hours by analytics of ECG monitoring. D: Patient admitted with acute hypoxic respiratory failure and acute kidney injury post-procedure day #0 from chest tube insertion to drain a pleural effusion. While charted vital signs and laboratory results were abnormal, integration of all available data sources accentuates her increasing risk in the several hours prior to ICU transfer.