A machine learning algorithm for detecting abnormal patterns in continuous capnography and pulse oximetry monitoring.

Continuous capnography monitors patient ventilation but can be susceptible to artifact, resulting in alarm fatigue. Development of smart algorithms may facilitate accurate detection of abnormal ventilation, allowing intervention before patient deterioration. The objective of this analysis was to use machine learning (ML) to classify combined waveforms of continuous capnography and pulse oximetry as normal or abnormal. We used data collected during the observational, prospective PRODIGY trial, in which patients receiving parenteral opioids underwent continuous capnography and pulse oximetry monitoring while on the general care floor [1]. Abnormal ventilation segments in the data stream were reviewed by nine experts and inter-rater agreement was assessed. Abnormal segments were defined as the time series 60s before and 30s after an abnormal pattern was detected. Normal segments (90s continuous monitoring) were randomly sampled and filtered to discard sequences with missing values. Five ML models were trained on extracted features and optimized towards an Fß score with ß = 2. The results show a high inter-rater agreement (> 87%), allowing 7,858 sequences (2,944 abnormal) to be used for model development. Data were divided into 80% training and 20% test sequences. The XGBoost model had the highest Fß score of 0.94 (with ß = 2), showcasing an impressive recall of 0.98 against a precision of 0.83. This study presents a promising advancement in respiratory monitoring, focusing on reducing false alarms and enhancing accuracy of alarm systems. Our algorithm reliably distinguishes normal from abnormal waveforms. More research is needed to define patterns to distinguish abnormal ventilation from artifacts.

Cost savings through continuous vital sign monitoring in the medical-surgical unit.

OBJECTIVE: This study aimed to determine the potential cost-savings for implementing continuous vital sign monitoring in a hospital's medical-surgical units. METHODS: A cost-savings analysis was designed to calculate potential cost-savings for an average-sized U.S. community hospital (153 total beds) over a 1-year time horizon. Analysis parameters were extracted from national databases and previous studies that compared outcomes for patients receiving continuous vital sign monitoring (SpO2, HR, and RR) or standard of care (intermittent vital sign measurements) in medical-surgical units based on a targeted literature review. Clinical parameters and associated costs served as analysis inputs. The analysis outputs were costs and potential cost-savings using a 50% and 100% adoption rate of continuous monitoring technologies across the medical-surgical unit. RESULTS: Potential annual cost-savings for in-hospital medical-surgical stays were estimated at $3,414,709 (2022 USD) and $6,829,418 for a 50% and 100% adoption rate, respectively. The cost-savings for an adoption rate of 100% equated to a ∼14% reduction in the overall annual cost of medical-surgical unit stays for an average-sized hospital. The largest contribution to potential cost-savings came from patients that avoided serious adverse events that require transfer to the intensive care unit; this resulted in annual cost-savings from reduced average length of stay between $1,756,613 and $3,513,226 (50% and 100% adoption rate, respectively). Additional cost-savings can be attained from reductions in in-hospital cardiac arrest-associated hospitalizations and decreased rapid response team activation. CONCLUSIONS: Our findings demonstrate that there is the potential for cost-savings of over $6.8 million dollars per year in an average-sized US community hospital by improving patient outcomes through implementation of continuous monitoring technologies in medical-surgical units. Continuous vital sign monitoring technologies that increase patient mobility and facilitate recovery may further contribute to cost-savings and should be considered for economic analyses. Future research is needed to explore these health-related outcomes.

Frequency and Temporal Distribution of Postoperative Respiratory Depressive Events.

BACKGROUND: The frequency and temporal distribution of postoperative respiratory depression (RD) events are not completely understood. This study determined the temporal distribution and frequency of RD episodes in postsurgical patients continuously monitored by bedside capnography and pulse oximetry. METHODS: This was a post hoc study of a subset of postsurgical patients enrolled in The PRediction of Opioid-induced respiratory Depression In patients monitored by capnoGraphY (PRODIGY) trial from 2 sites in the United States. These patients had undergone continuous bedside monitoring on general care wards. These data were adjudicated for potential RD episodes. The number of RD episodes per patient and the time of each RD episode were determined. The first RD episode experienced by a patient was classified as an "initial" episode, and the initial and all subsequent RD episodes experienced by a patient were classified as "all" episodes. A PRODIGY risk score was calculated. RESULTS: Data analyzed from 250 patients contained 2539 RD episodes in 155 (62.0%, 95% confidence interval, 55.7-68.0) patients with median 2 [0-8], range of 0-545 RD episodes per patient, with a PRODIGY risk score distribution of 100 (40.0%) low, 79 (31.6%) intermediate, 70 (28.0%) high (missing data from 1 patient). Median time to the initial RD episode was 8.8 [5.1-18.0] hours postoperatively. There was a peak occurrence of initial RD events between 14:00 and 20:00 on the day of surgery, and these were associated with a large number of subsequent events in the same timeframe. The peak time of all RD episodes occurred from 02:00 to 06:00. Patients with high PRODIGY risk scores had higher incidence and greater number of RD episodes per patient (P < .001, overall comparisons between groups for both incidence [χ2] and number of episodes [Kruskal-Wallis test]). CONCLUSIONS: Continuous monitoring of surgical patients demonstrates that RD episodes are common, and risk increases with higher PRODIGY scores. In this patient cohort, the rate of initial RD episodes peaked in the afternoon to early evening, while peak rate of all RD episodes occurred in early morning. Further, among patients with RD episodes, the number of episodes increased with higher PRODIGY scores.

Identification of Sleep Medicine and Anesthesia Core Topics for Anesthesia Residency: A Modified Delphi Technique Survey.

BACKGROUND: Sleep disorders affect up to 25% of the general population and are associated with increased risk of adverse perioperative events. The key sleep medicine topics that are most important for the practice of anesthesiology have not been well-defined. The objective of this study was to determine the high-priority sleep medicine topics that should be included in the education of anesthesia residents based on the insight of experts in the fields of anesthesia and sleep medicine. METHODS: We conducted a prospective cross-sectional survey of experts in the fields of sleep medicine and anesthesia based on the Delphi technique to establish consensus on the sleep medicine topics that should be incorporated into anesthesia residency curricula. Consensus for inclusion of a topic was defined as >80% of all experts selecting "agree" or "strongly agree" on a 5-point Likert scale. Responses to the survey questions were analyzed with descriptive statistical methods and presented as percentages or weighted mean values with standard deviations (SD) for Likert scale data. RESULTS: The topics that were found to have 100% agreement among experts were the influence of opioids and anesthetics on control of breathing and upper airway obstruction; potential interactions of wake-promoting/hypnotic medications with anesthetic agents; effects of sleep and anesthesia on upper airway patency; and anesthetic management of sleep apnea. Less than 80% agreement was found for topics on the anesthetic implications of other sleep disorders and future pathways in sleep medicine and anesthesia. CONCLUSIONS: We identify key topics of sleep medicine that can be included in the future design of anesthesia residency training curricula.

Postoperative opioid administration characteristics associated with opioid-induced respiratory depression: Results from the PRODIGY trial.

STUDY OBJECTIVE: Opioid administration for pain in general care floor patients remains common, and can lead to adverse outcomes, including respiratory compromise. The PRODIGY trial found that among ward patients receiving parenteral opioids, 46% experienced ≥1 respiratory depression episode. The objective of this analysis was to evaluate the geographic differences of opioid administration and examine the association between opioid administration characteristics and the occurrence of respiratory depression. DESIGN: Prospective observational trial. SETTING: 16 general care medical and surgical wards in Asia, Europe, and the United States. PATIENTS: 1335 patients receiving parenteral opioids. INTERVENTIONS: Blinded, alarm-silenced continuous capnography and pulse oximetry monitoring. MEASUREMENTS: Opioid-induced respiratory depression, defined as respiratory rate ≤ 5 bpm, SpO2 ≤ 85%, or ETCO2 ≤ 15 or ≥ 60 mmHg for ≥3 min; apnea episode lasting >30 s; or any respiratory opioid-related adverse event. RESULTS: Across all patients, 58% received only long-acting opioids, 16% received only short-acting (<3 h) opioids, and 21% received a combination of short- and long-acting (≥3 h) opioids. The type and median total morphine milligram equivalent (MME) of opioid administered varied significantly by region, with 31.5 (12.5-76.7) MME, 31.0 (6.2-99.0) MME, and 7.2 (1.7-18.7) MME in the United States, Europe, and Asia, respectively (p < 0.001). Considering only postoperative opioids, 54% (N = 119/220) and 45% (N = 347/779) of patients receiving only short-acting opioids or only long-acting opioids experienced ≥1 episode of opioid-induced respiratory depression, respectively. Multivariable analysis identified post-procedure tramadol (OR 0.62, 95% CI 0.424-0.905, p = 0.0133) and post-procedure epidural opioids (OR 0.485, 95% CI 0.322-0.731, p = 0.0005) being associated with a significant reduction in opioid-induced respiratory depression. CONCLUSIONS: Despite varying opioid administration characteristics between Asia, Europe, and the United States, opioid-induced respiratory depression remains a common global problem on general care medical and surgical wards. While the use of post-procedure tramadol or post-procedure epidural opioids may reduce the incidence of respiratory depression, continuous monitoring is also necessary to ensure patient safety when receiving postoperative opioids. REGISTRATION NUMBER: www.clinicaltrials.gov, ID: NCT02811302.

Prediction of Opioid-Induced Respiratory Depression on Inpatient Wards Using Continuous Capnography and Oximetry: An International Prospective, Observational Trial.

BACKGROUND: Opioid-related adverse events are a serious problem in hospitalized patients. Little is known about patients who are likely to experience opioid-induced respiratory depression events on the general care floor and may benefit from improved monitoring and early intervention. The trial objective was to derive and validate a risk prediction tool for respiratory depression in patients receiving opioids, as detected by continuous pulse oximetry and capnography monitoring. METHODS: PRediction of Opioid-induced respiratory Depression In patients monitored by capnoGraphY (PRODIGY) was a prospective, observational trial of blinded continuous capnography and oximetry conducted at 16 sites in the United States, Europe, and Asia. Vital signs were intermittently monitored per standard of care. A total of 1335 patients receiving parenteral opioids and continuously monitored on the general care floor were included in the analysis. A respiratory depression episode was defined as respiratory rate ≤5 breaths/min (bpm), oxygen saturation ≤85%, or end-tidal carbon dioxide ≤15 or ≥60 mm Hg for ≥3 minutes; apnea episode lasting >30 seconds; or any respiratory opioid-related adverse event. A risk prediction tool was derived using a multivariable logistic regression model of 46 a priori defined risk factors with stepwise selection and was internally validated by bootstrapping. RESULTS: One or more respiratory depression episodes were detected in 614 (46%) of 1335 general care floor patients (43% male; mean age, 58 ± 14 years) continuously monitored for a median of 24 hours (interquartile range [IQR], 17-26). A multivariable respiratory depression prediction model with area under the curve of 0.740 was developed using 5 independent variables: age ≥60 (in decades), sex, opioid naivety, sleep disorders, and chronic heart failure. The PRODIGY risk prediction tool showed significant separation between patients with and without respiratory depression (P < .001) and an odds ratio of 6.07 (95% confidence interval [CI], 4.44-8.30; P < .001) between the high- and low-risk groups. Compared to patients without respiratory depression episodes, mean hospital length of stay was 3 days longer in patients with ≥1 respiratory depression episode (10.5 ± 10.8 vs 7.7 ± 7.8 days; P < .0001) identified using continuous oximetry and capnography monitoring. CONCLUSIONS: A PRODIGY risk prediction model, derived from continuous oximetry and capnography, accurately predicts respiratory depression episodes in patients receiving opioids on the general care floor. Implementation of the PRODIGY score to determine the need for continuous monitoring may be a first step to reduce the incidence and consequences of respiratory compromise in patients receiving opioids on the general care floor.

New-Onset Atrial Fibrillation Detected by Continuous Capnography Monitoring: A Case Report.

BACKGROUND Asymptomatic postoperative atrial fibrillation (AF) may go undetected. As part of a multicenter observational trial designed to develop a risk prediction score for respiratory depression, the respiratory patterns of patients admitted to standard wards were continuously assessed with capnography and pulse oximetry. The monitor measured end-tidal carbon dioxide, respiratory rate, heart rate (HR), and oxyhemoglobin saturation. CASE REPORT Two men ages 75 and 72 experienced abrupt and variable postoperative changes in HR consistent with AF with rapid ventricular response, coinciding with an abnormal breathing pattern with apneic episodes. In both cases, the changes were not detected by routine clinical monitoring. CONCLUSIONS Continuous capnography identified respiratory distress in 2 patients who experienced symptoms of AF. Continuous monitoring devices can help health care providers minimize the risk of morbidity and mortality for patients at risk of respiratory depression.

Respiratory depression in low acuity hospital settings-Seeking answers from the PRODIGY trial.

Predicting episodes or severity of cardiorespiratory decompensation has proved to be challenging in patients with stable surgical or medical conditions, recovering on the general care floor (ward). Critical cardiorespiratory events on hospital floors may be prevented by early detection of deterioration using continuous, electronic cardiorespiratory monitoring (CEM). The PRediction of Opioid-induced Respiratory Depression In Patients Monitored by capnoGraphY (PRODIGY) trial investigates CEM using pulse oximetry and capnography in 1650 patients at 16 centers in North America, Europe, and Asia (ClinicalTrials.gov Identifier: NCT02811302). The primary goal of the study is to derive a risk prediction score for respiratory depression (RD) on the ward. The validation-derivation cohort design will derive this score from RD detected by continuous, blinded, multiparameter cardiorespiratory (heart rate, respiratory rate, end tidal carbon dioxide, and pulse oximetry) monitoring of patients on the ward receiving parenteral (including epidural) opioids for primary analgesia. This review provides a comprehensive synopsis on respiratory compromise in lower acuity hospital settings (ward) and describes the protocol of the PRODIGY trial as a means to enable prediction and early response to these events.

Risk factors for cardiopulmonary and respiratory arrest in medical and surgical hospital patients on opioid analgesics and sedatives.

BACKGROUND: Opioid induced respiratory depression is a known cause of preventable death in hospitals. Medications with sedative properties additionally potentiate opioid-induced respiratory and sedative effects, thereby elevating the risk for adverse events. The goal of this study was to determine what specific factors increase the risk of in-hospital cardiopulmonary and respiratory arrest (CPRA) in medical and surgical patients on opioid and sedative therapy. METHODS: The present study analyzed 14,504,809 medical inpatient and 6,771,882 surgical inpatient discharges reported into the Premier database from 2008 to 2012. Patients were divided in four categories: on opioids; on sedatives; on both opioids and sedatives; and on neither opioids nor sedatives. RESULTS: During hospital admission, 57% of all medical patients and 90% of all surgical patients were prescribed opioids, sedatives, or both. Surgical patients had a higher incidence of CPRA than medical patients (6.17 vs. 3.77 events per 1000 admissions; Relative Risk: 1.64 [95%CI: 1.62-1.66; p<0.0001). Opioids and sedatives were found to be independent predictors of CPRA (adjusted OR of 2.24 [95%CI: 2.18-2.29] for opioids and adjusted OR 1.80 [95%CI: 1.75-1.85] for sedatives in medical patients, and adjusted OR of 1.12 [95%CI: 1.07-1.16] for opioids and adjusted OR of 1.58 [95%CI: 1.51-1.66] for sedatives in surgical patients), with the highest risk in groups who received both types of medications (adjusted OR of 3.83 [95% CI: 3.74-3.92] in medical patients, and adjusted OR of 2.34 [95% CI: 2.25-2.42] in surgical patients) compared with groups that received neither type of medication. The common risk factors of CPRA in medical and surgical patients receiving both opioids and sedatives were Hispanic origin, mild liver disease, obesity, and COPD. Additionally, medical and surgical groups had their own unique risk factors for CPRA when placed on opioid and sedative therapy. CONCLUSIONS: Opioids and sedatives are independent and additive predictors of CPRA in both medical and surgical patients. Receiving both classes of medications further exacerbates the risk of CPRA for these patients. By identifying groups at risk among medical and surgical in-hospital patients, this study provides a step towards improving our understanding of how to use opioid and sedative medications safely, which may influence our treatment strategies and outcomes. More precise monitoring of selected high-risk patients may help prevent catastrophic cardiorespiratory complications from these medications. As a retrospective administrative database analysis, this study does not establish the causality or the temporality of the events but rather draws statistically significant associations between the clinical factors and outcomes.

Improving detection of patient deterioration in the general hospital ward environment.

: Patient monitoring on low acuity general hospital wards is currently based largely on intermittent observations and measurements of simple variables, such as blood pressure and temperature, by nursing staff. Often several hours can pass between such measurements and patient deterioration can go unnoticed. Moreover, the integration and interpretation of the information gleaned through these measurements remains highly dependent on clinical judgement. More intensive monitoring, which is commonly used in peri-operative and intensive care settings, is more likely to lead to the early identification of patients who are developing complications than is intermittent monitoring. Early identification can trigger appropriate management, thereby reducing the need for higher acuity care, reducing hospital lengths of stay and admission costs and even, at times, improving survival. However, this degree of monitoring has thus far been considered largely inappropriate for general hospital ward settings due to device costs and the need for staff expertise in data interpretation. In this review, we discuss some developing options to improve patient monitoring and thus detection of deterioration in low acuity general hospital wards.

Association of Opioids and Sedatives with Increased Risk of In-Hospital Cardiopulmonary Arrest from an Administrative Database.

BACKGROUND: While opioid use confers a known risk for respiratory depression, the incremental risk of in-hospital cardiopulmonary arrest, respiratory arrest, or cardiopulmonary resuscitation (CPRA) has not been studied. Our aim was to investigate the prevalence, outcomes, and risk profile of in-hospital CPRA for patients receiving opioids and medications with central nervous system sedating side effects (sedatives). METHODS: A retrospective analysis of adult inpatient discharges from 2008-2012 reported in the Premier Database. Patients were grouped into four mutually exclusive categories: (1) opioids and sedatives, (2) opioids only, (3) sedatives only, and (4) neither opioids nor sedatives. RESULTS: Among 21,276,691 inpatient discharges, 53% received opioids with or without sedatives. A total of 96,554 patients suffered CPRA (0.92 per 1000 hospital bed-days). Patients who received opioids and sedatives had an adjusted odds ratio for CPRA of 3.47 (95% CI: 3.40-3.54; p<0.0001) compared with patients not receiving opioids or sedatives. Opioids alone and sedatives alone were associated with a 1.81-fold and a 1.82-fold (p<0.0001 for both) increase in the odds of CPRA, respectively. In opioid patients, locations of CPRA were intensive care (54%), general care floor (25%), and stepdown units (15%). Only 42% of patients survived CPRA and only 22% were discharged home. Opioid patients with CPRA had mean increased hospital lengths of stay of 7.57 days and mean increased total hospital costs of $27,569. CONCLUSIONS: Opioids and sedatives are independent and additive risk factors for in-hospital CPRA. The impact of opioid sparing analgesia, reduced sedative use, and better monitoring on CPRA incidence deserves further study.

Remote video auditing with real-time feedback in an academic surgical suite improves safety and efficiency metrics: a cluster randomised study.

IMPORTANCE: Compliance with the surgical safety checklist during operative procedures has been shown to reduce inhospital mortality and complications but proper execution by the surgical team remains elusive. OBJECTIVE: We evaluated the impact of remote video auditing with real-time provider feedback on checklist compliance during sign-in, time-out and sign-out and case turnover times. DESIGN, SETTING: Prospective, cluster randomised study in a 23-operating room (OR) suite. PARTICIPANTS: Surgeons, anaesthesia providers, nurses and support staff. EXPOSURE: ORs were randomised to receive, or not receive, real-time feedback on safety checklist compliance and efficiency metrics via display boards and text messages, followed by a period during which all ORs received feedback. MAIN OUTCOMES AND MEASURES: Checklist compliance (Pass/Fail) during sign-in, time-out and sign-out demonstrated by (1) use of checklist, (2) team attentiveness, (3) required duration, (4) proper sequence and duration of case turnover times. RESULTS: Sign-in, time-out and sign-out PASS rates increased from 25%, 16% and 32% during baseline phase (n=1886) to 64%, 84% and 68% for feedback ORs versus 40%, 77% and 51% for no-feedback ORs (p<0.004) during the intervention phase (n=2693). Pass rates were 91%, 95% and 84% during the all-feedback phase (n=2001). For scheduled cases (n=1406, 71%), feedback reduced mean turnover times by 14% (41.4 min vs 48.1 min, p<0.004), and the improvement was sustained during the all-feedback period. Feedback had no effect on turnover time for unscheduled cases (n=587, 29%). CONCLUSIONS AND RELEVANCE: Our data indicate that remote video auditing with feedback improves surgical safety checklist compliance for all cases, and turnover time for scheduled cases, but not for unscheduled cases.