A fitting machine learning prediction model for short-term mortality following percutaneous catheterization intervention: a nationwide population-based study

Objective: Patients experiencing signs and symptoms of an acute myocardial infarction (AMI) require prompt evaluation and treatment. There are little contemporary data, however, available on how the extent of delay between the onset of acute coronary symptoms and hospital presentation may impact short-term mortality. The purpose of this population-based study was to examine the relationship between extent of pre-hospital delay with hospital case-fatality rates (HCFRs) and 30-day post-admission mortality rates (PAMRs) among patients hospitalized with validated AMI in all central Massachusetts medical centers, and trends over time therein. Methods: We examined the medical records of residents of the Worcester, MA, metropolitan area hospitalized with a confirmed AMI at all 11 central MA medical centers on a biennial basis between 1999 and 2009 (n = 6,017). Information on patient’s demographic, medical history, clinical characteristics, and time of acute symptom onset and hospital arrival was abstracted. Results: Hospital medical record data on pre-hospital delay were available for 2,913 (48%) subjects of whom their mean age was 68 years, 38% were female, and 90% were Caucasian. The mean and median pre-hospital delay times were 4.0 hours and 2.0 hours, respectively, with little change noted in these times between 1999 and 2009. Patients who reported pre-hospital delay times greater than two hours were more likely to be older, female, and have a history of heart failure or diabetes mellitus as compared with patients who delayed seeking medical care by less than 2 hours. The overall HCFR was 6.6% and 30-day PAMR was 9.4%. The average HCFRs and 30-day PAMRs varied slightly between those with delay times of less than 2 hours (6.5%, 9.2%), 2 to 4 hours (6.3%, 8.6%), and greater than 4 hours (7.0%, 10.6%). No statistically significant changes in HCFRs and 30-day PAMRs were observed as pre-hospital delay times increased. Analyses of our principal study outcomes according to type of AMI (e.g., STEMI and NSTEMI) are ongoing and will be presented subsequently. Conclusions: This population-based study of residents of central MA hospitalized with AMI in all metropolitan Worcester medical centers showed little change in average and median pre-hospital delays between 1999 and 2009. Both the HCFRs and 30-day PAMRs were not significantly increased with greater durations of pre-hospital delay possibly due to potential confounders such as symptom severity. Our preliminary results suggest the need to further investigate trends in pre-hospital delay and short-term mortality, including patients who die in the community before receiving acute medical care.

Download Full-text

Short-Term Mortality After a First Epileptic Seizure: A Population-Based Study

Epilepsia ◽

10.1111/j.1528-1157.1999.tb02010.x ◽

1999 ◽

Vol 40 (10) ◽

pp. 1388-1392 ◽

Cited By ~ 50

Author(s):

Jerome Loiseau ◽

Marie-Christine Picot ◽

Pierre Loiseau

Keyword(s):

Epileptic Seizure ◽

Population Based ◽

Short Term ◽

Population Based Study ◽

Short Term Mortality

Download Full-text

Incidence and Short-term Mortality From Perforated Peptic Ulcer in Korea: A Population-Based Study

Journal of Epidemiology ◽

10.2188/jea.je20120056 ◽

2012 ◽

Vol 22 (6) ◽

pp. 508-516 ◽

Cited By ~ 20

Author(s):

SeungJin Bae ◽

Ki-Nam Shim ◽

Nayoung Kim ◽

Jung Mook Kang ◽

Dong-Sook Kim ◽

...

Keyword(s):

Peptic Ulcer ◽

Perforated Peptic Ulcer ◽

Population Based ◽

Short Term ◽

Population Based Study ◽

Short Term Mortality

Download Full-text

Epidemiology, Characteristics, and Outcomes of ICU-Managed Homeless Patients: A Population-Based Study

BioMed Research International ◽

10.1155/2018/3869652 ◽

2018 ◽

Vol 2018 ◽

pp. 1-11

Author(s):

Lavi Oud

Keyword(s):

Critical Care ◽

Prehospital Care ◽

Population Level ◽

Data File ◽

Population Based ◽

Short Term ◽

Population Based Study ◽

Care Services ◽

Comorbidity Burden ◽

Short Term Mortality

Background. The population-level demand for critical care services among the homeless (H) remains unknown, with only sparse data on the characteristics and outcomes of those managed in the ICU. Methods. The Texas Inpatient Public Use Data File and annual federal reports were used to identify H hospitalizations and annual estimates of the H population between 2007 and 2014. The incidence of ICU admissions in the H population, the characteristics of ICU-managed H, and factors associated with their short-term mortality were examined. Results. Among 52,206 H hospitalizations 15,553 (29.8%) were admitted to ICU. The incidence of ICU admission among state H population rose between 2007 and 2014 from 28.0 to 96.6/1,000 (p<0.0001), respectively. Adults aged ≥ 45 years and minorities accounted for 70.2% and 57.6%, respectively, of the growth in volume of ICU admissions. Short-term mortality was 3.2%, with odds of death increased with age, comorbidity burden, and number of failing organs. Conclusions. The demand for critical care services was increasingly high among the H and was contrasted by low short-term mortality among ICU admissions. These findings, coupled with the persistent health disparities among minority H, underscore the need to effectively address homelessness and reduce barriers to longitudinal appropriate prehospital care among the H.

Download Full-text

Risk factors of short-term mortality after acute nonvariceal upper gastrointestinal bleeding in patients on dialysis: a population-based study

BMC Nephrology ◽

10.1186/1471-2369-14-97 ◽

2013 ◽

Vol 14 (1) ◽

Cited By ~ 6

Author(s):

Ju-Yeh Yang ◽

Tsung-Chun Lee ◽

Maria E Montez-Rath ◽

Glenn M Chertow ◽

Wolfgang C Winkelmayer

Keyword(s):

Risk Factors ◽

Gastrointestinal Bleeding ◽

Upper Gastrointestinal Bleeding ◽

Population Based ◽

Short Term ◽

Population Based Study ◽

Short Term Mortality ◽

Upper Gastrointestinal

Download Full-text

A systematic evaluation of models predicting short-term mortality for cancer patients to highlight uncertainty about model performance and high risk of bias.

Journal of Clinical Oncology ◽

10.1200/jco.2021.39.15_suppl.e13559 ◽

2021 ◽

Vol 39 (15_suppl) ◽

pp. e13559-e13559

Author(s):

Sheng-Chieh Lu ◽

Cai Xu ◽

Chandler Nguyen ◽

Larissa Meyer ◽

Chris Sidey-Gibbons

Keyword(s):

Machine Learning ◽

High Risk ◽

Prediction Model ◽

Cancer Mortality ◽

Model Performance ◽

Risk Of Bias ◽

Short Term ◽

Electronic Health Record Data ◽

Bias Assessment ◽

Short Term Mortality

e13559 Background: Short-term cancer mortality prediction has many implications concerning care planning. An accurate prognosis allows healthcare providers to adjust care plans and take appropriate actions, such as initiating end-of-life conversations. Machine learning (ML) techniques demonstrated promising capability to support clinical decision-making via providing reliable predictions for a variety of clinical outcomes, including cancer mortality. However, the evidence has not yet been systematically synthesized and evaluated. The objective of this review was to examine the performance and risk-of-bias for ML models trained to predict short-term (≤ 12 months) cancer mortality. Methods: We identified relevant literature from five electronic databases: Ovid Medline, Ovid EMBASE, Scopus, Web of Science, and IEEE Xplore. We searched each database with predefined MeSH terms and keywords of oncology, machine learning, and mortality using AND/OR statements. Inclusion criteria included: 1) developed/validated ML models for predicting oncology patient mortality within one year using electronic health record data; 2) reported model performance within a dataset that was not used to train the models; 3) original research; 4) peer-reviewed full paper in English; 5) published before 1/10/2020. We conducted risk of bias assessment using prediction model risk of bias assessment tool (PROBAST). Results: Ten articles were included in this review. Most studies focused on predicting 1-year mortality (n = 6) for multiple types of cancer (n = 5). Most studies (n = 7) used a single metric, the area under the receiver operating characteristic curve (AUROC), to examine their models. The AUROC ranged from .69 to .91, with a median of .85. Information on samples (n = 10), resampling methods (n = 6), model tuning approaches (n = 9), censoring (n = 10), and sample size determinations (n = 10) were incomplete or absent. Six studies have a high risk of bias for the analysis domain in the PROBAST. Conclusions: The performance of ML models for short-term cancer mortality appears promising. However, most studies report only a single performance metric that obfuscates evaluation of a model’s true performance. This is especially problematic when predicting rare events such as short-term mortality. We found little-to-no information on a given model’s ability to correctly identify patients at high risk of mortality. The incomplete reporting of model development poses challenges to risk of bias assessment and reduces the confidence in the results. Our findings suggest that future studies should report comprehensive performance metrics using a standard reporting guideline, such as transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD), to ensure sufficient information for replication, justification, and adoption.

Download Full-text