Optimizing Best Practice Advisory alerts in electronic medical records with a multi-pronged strategy at a tertiary care hospital in Singapore.

Objective: Clinical decision support (CDS) alerts can aid in improving patient care. One CDS functionality is the Best Practice Advisory (BPA) alert notification system, wherein BPA alerts are automated alerts embedded in the hospital's electronic medical records (EMR). However, excessive alerts can change clinician behavior; redundant and repetitive alerts can contribute to alert fatigue. Alerts can be optimized through a multipronged strategy. Our study aims to describe these strategies adopted and evaluate the resultant BPA alert optimization outcomes. Materials and Methods: This retrospective single-center study was done at Jurong Health Campus. Aggregated, anonymized data on patient demographics and alert statistics were collected from January 1, 2018 to December 31, 2021. "Preintervention" period was January 1-December 31, 2018, and "postintervention" period was January 1-December 31, 2021. The intervention period was the intervening period. Categorical variables were reported as frequencies and proportions and compared using the chi-square test. Continuous data were reported as median (interquartile range, IQR) and compared using the Wilcoxon rank-sum test. Statistical significance was defined at P < .05. Results: There was a significant reduction of 59.6% in the total number of interruptive BPA alerts, despite an increase in the number of unique BPAs from 54 to 360 from pre- to postintervention. There was a 74% reduction in the number of alerts from the 7 BPAs that were optimized from the pre- to postintervention period. There was a significant increase in percentage of overall interruptive BPA alerts with action taken (8 [IQR 7.7-8.4] to 54.7 [IQR 52.5-58.9], P-value < .05) and optimized BPAs with action taken (32.6 [IQR 32.3-32.9] to 72.6 [IQR 64.3-73.4], P-value < .05). We estimate that the reduction in alerts saved 3600 h of providers' time per year. Conclusions: A significant reduction in interruptive alert volume, and a significant increase in action taken rates despite manifold increase in the number of unique BPAs could be achieved through concentrated efforts focusing on governance, data review, and visualization using a system-embedded tool, combined with the CDS Five Rights framework, to optimize alerts. Improved alert compliance was likely multifactorial-due to decreased repeated alert firing for the same patient; better awareness due to stakeholders' involvement; and less fatigue since unnecessary alerts were removed. Future studies should prospectively focus on patients' clinical chart reviews to assess downstream effects of various actions taken, identify any possibility of harm, and collect end-user feedback regarding the utility of alerts.

Sepsis in a Combined Medical and Surgical High Dependency/Intensive Care Unit in Singapore: A Cohort Study and Survival Analysis.

Background: Sepsis is a common indication for intensive care unit (ICU) admission and is associated with significant mortality and morbidity. The aim of our study was to first assess the incidence, severity, short-term and long-term mortality of sepsis in a combined medical and surgical high dependency/ ICU in Singapore, and to identify factors associated with increasing short-term and long-term mortality. Methods: All admissions from July 1 to December 31, 2017 were retrospectively screened and clinical data were collected. Patients were followed up until 3 years post ICU admission. Results: Of a total 1526 admissions, 281 had infection at ICU admission, and 254 (16.6%) fulfilled sepsis-3 criteria for sepsis. A total of 141 (9.2%) had septic shock. The 30-day, 1-year, 2-year and 3-year mortality of sepsis patients were 19.3%, 25.2%, 30.3% and 32.3%, respectively. Lung was the most common site of infection. Compared with 30-day sepsis survivors, non-survivors were older (median age 70 vs 63, P <0.001), had higher percentage of lung infection (65.3% vs 36.1%, P <0.05), higher admission Sequential Organ Failure Assessment (SOFA) score (median 9 vs 5, P <0.001), and longer ICU stay (median days: 4 vs 3, P = 0.037). In stepwise Cox regression analysis, lung infection was an independent risk factor for both increasing 30-day and 3-year mortality. Immunocompromised host, increasing age and SOFA score were associated with higher 30-day mortality. Diabetes, admission quick Sequential Organ Failure Assessment (qSOFA) score >1 and unplanned ICU re-admission were associated with increasing 3-year mortality in 30-day survivors. Conclusion: Our retrospective cohort single center study first reported sepsis admission incidence of 16.6% in a combined medical and surgical high dependency/ICU in Singapore, with significant short-term and long-term mortality. Lung infection was an independent risk factor for both 30-day and 3-year mortality.

Habit and Automaticity in Medical Alert Override: Cohort Study.

BACKGROUND: Prior literature suggests that alert dismissal could be linked to physicians' habits and automaticity. The evidence for this perspective has been mainly observational data. This study uses log data from an electronic medical records system to empirically validate this perspective. OBJECTIVE: We seek to quantify the association between habit and alert dismissal in physicians. METHODS: We conducted a retrospective analysis using the log data comprising 66,049 alerts generated from hospitalized patients in a hospital from March 2017 to December 2018. We analyzed 1152 physicians exposed to a specific clinical support alert triggered in a hospital's electronic medical record system to estimate the extent to which the physicians' habit strength, which had been developed from habitual learning, impacted their propensity toward alert dismissal. We further examined the association between a physician's habit strength and their subsequent incidences of alert dismissal. Additionally, we recorded the time taken by the physician to respond to the alert and collected data on other clinical and environmental factors related to the alerts as covariates for the analysis. RESULTS: We found that a physician's prior dismissal of alerts leads to their increased habit strength to dismiss alerts. Furthermore, a physician's habit strength to dismiss alerts was found to be positively associated with incidences of subsequent alert dismissals after their initial alert dismissal. Alert dismissal due to habitual learning was also found to be pervasive across all physician ranks, from junior interns to senior attending specialists. Further, the dismissal of alerts had been observed to typically occur after a very short processing time. Our study found that 72.5% of alerts were dismissed in under 3 seconds after the alert appeared, and 13.2% of all alerts were dismissed in under 1 second after the alert appeared. We found empirical support that habitual dismissal is one of the key factors associated with alert dismissal. We also found that habitual dismissal of alerts is self-reinforcing, which suggests significant challenges in disrupting or changing alert dismissal habits once they are formed. CONCLUSIONS: Habitual tendencies are associated with the dismissal of alerts. This relationship is pervasive across all levels of physician rank and experience, and the effect is self-reinforcing.

Prediction of Readmission in Geriatric Patients From Clinical Notes: Retrospective Text Mining Study.

BACKGROUND: Prior literature suggests that psychosocial factors adversely impact health and health care utilization outcomes. However, psychosocial factors are typically not captured by the structured data in electronic medical records (EMRs) but are rather recorded as free text in different types of clinical notes. OBJECTIVE: We here propose a text-mining approach to analyze EMRs to identify older adults with key psychosocial factors that predict adverse health care utilization outcomes, measured by 30-day readmission. The psychological factors were appended to the LACE (Length of stay, Acuity of the admission, Comorbidity of the patient, and Emergency department use) Index for Readmission to improve the prediction of readmission risk. METHODS: We performed a retrospective analysis using EMR notes of 43,216 hospitalization encounters in a hospital from January 1, 2017 to February 28, 2019. The mean age of the cohort was 67.51 years (SD 15.87), the mean length of stay was 5.57 days (SD 10.41), and the mean intensive care unit stay was 5% (SD 22%). We employed text-mining techniques to extract psychosocial topics that are representative of these patients and tested the utility of these topics in predicting 30-day hospital readmission beyond the predictive value of the LACE Index for Readmission. RESULTS: The added text-mined factors improved the area under the receiver operating characteristic curve of the readmission prediction by 8.46% for geriatric patients, 6.99% for the general hospital population, and 6.64% for frequent admitters. Medical social workers and case managers captured more of the psychosocial text topics than physicians. CONCLUSIONS: The results of this study demonstrate the feasibility of extracting psychosocial factors from EMR clinical notes and the value of these notes in improving readmission risk prediction. Psychosocial profiles of patients can be curated and quantified from text mining clinical notes and these profiles can be successfully applied to artificial intelligence models to improve readmission risk prediction.

Artificial intelligence in sepsis early prediction and diagnosis using unstructured data in healthcare.

Sepsis is a leading cause of death in hospitals. Early prediction and diagnosis of sepsis, which is critical in reducing mortality, is challenging as many of its signs and symptoms are similar to other less critical conditions. We develop an artificial intelligence algorithm, SERA algorithm, which uses both structured data and unstructured clinical notes to predict and diagnose sepsis. We test this algorithm with independent, clinical notes and achieve high predictive accuracy 12 hours before the onset of sepsis (AUC 0.94, sensitivity 0.87 and specificity 0.87). We compare the SERA algorithm against physician predictions and show the algorithm's potential to increase the early detection of sepsis by up to 32% and reduce false positives by up to 17%. Mining unstructured clinical notes is shown to improve the algorithm's accuracy compared to using only clinical measures for early warning 12 to 48 hours before the onset of sepsis.

Six-minute walk distance in healthy Singaporean adults cannot be predicted using reference equations derived from Caucasian populations.

OBJECTIVES: The 6-min walk test (6MWT) is commonly used to assess the functional exercise capacity of individuals with cardiopulmonary disease. Recent studies have established regression equations to predict the 6-min walk distance (6MWD) in healthy Caucasian populations; however, regression equations have yet to be established for the Singaporean population. The aim of this study was to determine 6MWD in healthy Singaporeans and identify contributors to 6MWD in this population. We also compared measured 6MWD with predicted 6MWD from two regression equations derived in Caucasian subjects. METHODOLOGY: Thirty-five healthy subjects (32 Chinese, 16 men) aged between 45 and 85 years performed three walking tests using a standardized protocol. 6MWD was defined as the greatest distance achieved from the three tests. Heart rate (HR) was recorded each minute during the 6MWT. Other measurements included age, height, leg length, smoking history and self-reported physical activity. RESULTS: 6MWD was 560 +/- 105 m and was not significantly different between men and women (P = 0.19). 6MWD was related to age (r = -0.36, P = 0.03), height (r = 0.35, P = 0.04), leg length (r = 0.38, P = 0.02) and the maximum HR achieved on the 6MWT when expressed as a percentage of the predicted maximum HR (%predHRmax, r = 0.73, P < 0.001). Stepwise multiple regression analysis showed that age, height, weight and %predHRmax were independent contributors (P < 0.01) to 6MWD, explaining 78% of the variance. Predicted 6MWD using regression equations derived from Caucasian subjects exceeded measured 6MWD by more than 75 m (P < 0.001). CONCLUSIONS: This is the first study to report 6MWD for healthy Singaporeans aged 45-85 years. The regression equation developed in this study explained 78% of the variance in 6MWD. Published equations derived from Caucasian subjects overestimate 6MWD in Singaporean Chinese.