A Novel Model for Enhanced Prediction and Understanding of Unplanned 30-Day Pediatric Readmission

Birth Weight and Gestational Age as Modifiers of Rehospitalization after Neonatal Intensive Care Unit Admission

OBJECTIVE

This study aimed to assess interaction effects between gestational age and birth weight on 30-day unplanned hospital readmission following discharge from the neonatal intensive care unit (NICU).

STUDY DESIGN

This is a retrospective study that uses the study site's Children's Hospitals Neonatal Database and electronic health records. Population included patients discharged from a NICU between January 2017 and March 2020. Variables encompassing demographics, gestational age, birth weight, medications, maternal data, and surgical procedures were controlled for. A statistical interaction between gestational age and birth weight was tested for statistical significance.

RESULTS

A total of 2,307 neonates were included, with 7.2% readmitted within 30 days of discharge. Statistical interaction between birth weight and gestational age was statistically significant, indicating that the odds of readmission among low birthweight premature patients increase with increasing gestational age, whereas decrease with increasing gestational age among their normal or high birth weight peers.

CONCLUSION

The effect of gestational age on odds of hospital readmission is dependent on birth weight.

KEY POINTS

· Population included patients discharged from a NICU between January 2017 and March 2020.. · A total of 2,307 neonates were included, with 7.2% readmitted within 30 days of discharge.. · The effect of gestational age on odds of hospital readmission is dependent on birth weight..

Machine learning for hospital readmission prediction in pediatric population

BACKGROUND AND OBJECTIVE

Pediatric readmissions are a burden on patients, families, and the healthcare system. In order to identify patients at higher readmission risk, more accurate techniques, as machine learning (ML), could be a good strategy to expand the knowledge in this area. The aim of this study was to develop predictive models capable of identifying children and adolescents at high risk of potentially avoidable 30-day readmission using ML.

METHODS

Retrospective cohort study was carried out with 9,080 patients under 18 years old admitted to a tertiary university hospital. Demographic, clinical, and biochemical data were collected from electronic databases. We randomly divided the dataset into training (75 %) and testing (25 %), applied downsampling, repeated cross-validation with five folds and ten repetitions, and the hyperparameter was optimized of each technique using a grid search via racing with ANOVA models. We applied six ML classification algorithms to build the predictive models, including classification and regression tree (CART), random forest (RF), gradient boosting machine (GBM), extreme gradient boosting (XGBoost), decision tree and logistic regression (LR). The area under the receiver operating curve (AUC), sensitivity, specificity, Youden's J-index and accuracy were used to evaluate the performance of each model.

RESULTS

The avoidable 30-day hospital readmissions rate was 9.5 %. Some algorithms presented similar AUC, both in the dataset training and in the dataset testing, such as XGBoost, RF, GBM and CART. Considering the Youden's J-index, the algorithm that presented the best index was XGBoost with bagging imputation, with AUC of 0.814 (J-index of 0.484). Cancer diagnosis, age, red blood cells, leukocytes, red cell distribution width and sodium levels, elective admission, and multimorbidity were the most important characteristics to classify between readmission and non-readmission groups.

CONCLUSION

Machine learning approaches, especially XGBoost, can predict potentially avoidable 30-day pediatric hospital readmission into tertiary assistance. If implemented in the computer hospital system, our model can help in the early and more accurate identification of patients at readmission risk, targeting health strategic interventions.

Validation of the HOSPITAL score as predictor of 30-day potentially avoidable readmissions in pediatric hospitalized population: retrospective cohort study

Potentially avoidable pediatric readmissions are a burden to patients and their families. Identifying patients with higher risk of readmission could help minimize hospital costs and facilitate the targeting of care interventions. HOSPITAL score is a tool developed and widely used to predict adult patient's readmissions; however its predictive capacity for pediatric readmissions has not yet been evaluated. The aim of the study was to validate the HOSPITAL score application to predict 30-day potentially avoidable readmissions in a pediatric hospitalized population. This is a retrospective cohort study with patients under 18 years old admitted to a tertiary university hospital (n = 6,344). The HOSPITAL score was estimated for each admission. Subsequently, we classified the patients as low (0-4), intermediate (5-6), and high (7-12) risk groups. In order to estimate the discrimination power, the sensitivity, specificity, and accuracy were determined by the receiver operating characteristics (ROC) and the calibration by the Hosmer-Lemeshow goodness-of-fit. The 30-day hospital readmission was 11.70% (745). The accuracy was 0.80 (CI 95%, 0.77, 0.83), with a sensitivity of 70.96% and specificity of 78.29%, and a good calibration (p = 0.34).    Conclusion: HOSPITAL score showed a good discrimination and can be used to predict 30-day potentially avoidable readmission in a large pediatric population with different medical diagnoses. Our study validates and expands the usefulness of the HOSPITAL score as a tool to predict avoidable hospital readmissions for pediatric population. What is Known: •   Pediatric readmissions burden patients, the family network, and the health system. In addition, it influences negatively child development. •   The HOSPITAL score is one of the tools developed and widely used to identify patients at high risk of hospital readmission, but its predictive capacity for pediatric readmissions has not been yet assessed. What is New: • The HOSPITAL score showed good ability to identify a risk of 30-day potentially avoidable readmission in a pediatric population in different clinical contexts and diagnoses. • Our study expands the usefulness of the HOSPITAL score as a tool for predicting hospital readmissions for children and adolescents.

Development and Validation of an Integrated Suite of Prediction Models for All-Cause 30-Day Readmissions of Children and Adolescents Aged 0 to 18 Years

IMPORTANCE

Readmission is often considered a hospital quality measure, yet no validated risk prediction models exist for children.

OBJECTIVE

To develop and validate a tool identifying patients before hospital discharge who are at risk for subsequent readmission, applicable to all ages.

DESIGN, SETTING, AND PARTICIPANTS

This population-based prognostic analysis used electronic health record-derived data from a freestanding children's hospital from January 1, 2016, to December 31, 2019. All-cause 30-day readmission was modeled using 3 years of discharge data. Data were analyzed from June 1 to November 30, 2021.

MAIN OUTCOMES AND MEASURES

Three models were derived as a complementary suite to include (1) children 6 months or older with 1 or more prior hospitalizations within the last 6 months (recent admission model [RAM]), (2) children 6 months or older with no prior hospitalizations in the last 6 months (new admission model [NAM]), and (3) children younger than 6 months (young infant model [YIM]). Generalized mixed linear models were used for all analyses. Models were validated using an additional year of discharges.

RESULTS

The derivation set contained 29 988 patients with 48 019 hospitalizations; 50.1% of these admissions were for children younger than 5 years and 54.7% were boys. In the derivation set, 4878 of 13 490 admissions (36.2%) in the RAM cohort, 2044 of 27 531 (7.4%) in the NAM cohort, and 855 of 6998 (12.2%) in the YIM cohort were followed within 30 days by a readmission. In the RAM cohort, prior utilization, current or prior procedures indicative of severity of illness (transfusion, ventilation, or central venous catheter), commercial insurance, and prolonged length of stay (LOS) were associated with readmission. In the NAM cohort, procedures, prolonged LOS, and emergency department visit in the past 6 months were associated with readmission. In the YIM cohort, LOS, prior visits, and critical procedures were associated with readmission. The area under the receiver operating characteristics curve was 83.1 (95% CI, 82.4-83.8) for the RAM cohort, 76.1 (95% CI, 75.0-77.2) for the NAM cohort, and 80.3 (95% CI, 78.8-81.9) for the YIM cohort.

CONCLUSIONS AND RELEVANCE

In this prognostic study, the suite of 3 prediction models had acceptable to excellent discrimination for children. These models may allow future improvements in tailored discharge preparedness to prevent high-risk readmissions.

Performance Characteristics of a Machine-Learning Tool to Predict 7-Day Hospital Readmissions

OBJECTIVES

To develop an institutional machine-learning (ML) tool that utilizes demographic, socioeconomic, and medical information to stratify risk for 7-day readmission after hospital discharge; assess the validity and reliability of the tool; and demonstrate its discriminatory capacity to predict readmissions.

PATIENTS AND METHODS

We performed a combined single-center, cross-sectional, and prospective study of pediatric hospitalists assessing the face and content validity of the developed readmission ML tool. The cross-sectional analyses used data from questionnaire Likert scale responses regarding face and content validity. Prospectively, we compared the discriminatory capacity of provider readmission risk versus the ML tool to predict 7-day readmissions assessed via area under the receiver operating characteristic curve analyses.

RESULTS

Overall, 80% (15 of 20) of hospitalists reported being somewhat to very confident with their ability to accurately predict readmission risk; 53% reported that an ML tool would influence clinical decision-making (face validity). The ML tool variable exhibiting the highest content validity was history of previous 7-day readmission. Prospective provider assessment of risk of 413 discharges showed minimal agreement with the ML tool (κ = 0.104 [95% confidence interval 0.028-0.179]). Both provider gestalt and ML calculations poorly predicted 7-day readmissions (area under the receiver operating characteristic curve: 0.67 vs 0.52; P = .11).

CONCLUSIONS

An ML tool for predicting 7-day hospital readmissions after discharge from the general pediatric ward had limited face and content validity among pediatric hospitalists. Both provider and ML-based determinations of readmission risk were of limited discriminatory value. Before incorporating similar tools into real-time discharge planning, model calibration efforts are needed.

Applicability of predictive models for 30-day unplanned hospital readmission risk in paediatrics: a systematic review

OBJECTIVES

To summarise multivariable predictive models for 30-day unplanned hospital readmissions (UHRs) in paediatrics, describe their performance and completeness in reporting, and determine their potential for application in practice.

DESIGN

Systematic review.

DATA SOURCE

CINAHL, Embase and PubMed up to 7 October 2021.

ELIGIBILITY CRITERIA

English or German language studies aiming to develop or validate a multivariable predictive model for 30-day paediatric UHRs related to all-cause, surgical conditions or general medical conditions were included.

DATA EXTRACTION AND SYNTHESIS

Study characteristics, risk factors significant for predicting readmissions and information about performance measures (eg, c-statistic) were extracted. Reporting quality was addressed by the 'Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis' (TRIPOD) adherence form. The study quality was assessed by applying six domains of potential biases. Due to expected heterogeneity among the studies, the data were qualitatively synthesised.

RESULTS

Based on 28 studies, 37 predictive models were identified, which could potentially be used for determining individual 30-day UHR risk in paediatrics. The number of study participants ranged from 190 children to 1.4 million encounters. The two most common significant risk factors were comorbidity and (postoperative) length of stay. 23 models showed a c-statistic above 0.7 and are primarily applicable at discharge. The median TRIPOD adherence of the models was 59% (P₂₅-P₇₅, 55%-69%), ranging from a minimum of 33% to a maximum of 81%. Overall, the quality of many studies was moderate to low in all six domains.

CONCLUSION

Predictive models may be useful in identifying paediatric patients at increased risk of readmission. To support the application of predictive models, more attention should be placed on completeness in reporting, particularly for those items that may be relevant for implementation in practice.

Identifying Children at Readmission Risk: At-Admission versus Traditional At-Discharge Readmission Prediction Model

The timing of 30-day pediatric readmissions is skewed with approximately 40% of the incidents occurring within the first week of hospital discharges. The skewed readmission time distribution coupled with delay in health information exchange among healthcare providers might offer a limited time to devise a comprehensive intervention plan. However, pediatric readmission studies are thus far limited to the development of the prediction model after hospital discharges. In this study, we proposed a novel pediatric readmission prediction model at the time of hospital admission which can improve the high-risk patient selection process. We also compared proposed models with the standard at-discharge readmission prediction model. Using the Hospital Cost and Utilization Project database, this prognostic study included pediatric hospital discharges in Florida from January 2016 through September 2017. Four machine learning algorithms—logistic regression with backward stepwise selection, decision tree, Support Vector machines (SVM) with the polynomial kernel, and Gradient Boosting—were developed for at-admission and at-discharge models using a recursive feature elimination technique with a repeated cross-validation process. The performance of the at-admission and at-discharge model was measured by the area under the curve. The performance of the at-admission model was comparable with the at-discharge model for all four algorithms. SVM with Polynomial Kernel algorithms outperformed all other algorithms for at-admission and at-discharge models. Important features associated with increased readmission risk varied widely across the type of prediction model and were mostly related to patients’ demographics, social determinates, clinical factors, and hospital characteristics. Proposed at-admission readmission risk decision support model could help hospitals and providers with additional time for intervention planning, particularly for those targeting social determinants of children’s overall health.

Targeted Clinical Interventions for Reducing Pediatric Readmissions

BACKGROUND

In this interventional study, we addressed the selection and application of clinical interventions on pediatric patients identified as at risk by a predictive model for readmissions.

METHODS

A predictive model for readmissions was implemented, and a team of providers expanded corresponding clinical interventions for at-risk patients at a freestanding children's hospital. Interventions encompassed social determinants of health, outpatient care, medication reconciliation, inpatient and discharge planning, and postdischarge calls and/or follow-up. Statistical process control charts were used to compare readmission rates for the 3-year period preceding adoption of the model and clinical interventions with those for the 2-year period after adoption of the model and clinical interventions. Potential financial savings were estimated by using national estimates of the cost of pediatric inpatient readmissions.

RESULTS

The 30-day all-cause readmission rates during the periods before and after predictive modeling (and corresponding 95% confidence intervals [CI]) were 12.5% (95% CI: 12.2%-12.8%) and 11.1% (95% CI: 10.8%-11.5%), respectively. More modest but similar improvements were observed for 7-day readmissions. Statistical process control charts indicated nonrandom reductions in readmissions after predictive model adoption. The national estimate of the cost of pediatric readmissions indicates an associated health care savings due to reduced 30-day readmission during the 2-year predictive modeling period at $2 673 264 (95% CI: $2 612 431-$2 735 364).

CONCLUSIONS

A combination of predictive modeling and targeted clinical interventions to improve the management of pediatric patients at high risk for readmission was successful in reducing the rate of readmission and reducing overall health care costs. The continued prioritization of patients with potentially modifiable outcomes is key to improving patient outcomes.

Prolonged hospital length of stay in pediatric trauma: a model for targeted interventions

BACKGROUND

In this study, trauma-specific risk factors of prolonged length of stay (LOS) in pediatric trauma were examined. Statistical and machine learning models were used to proffer ways to improve the quality of care of patients at risk of prolonged length of stay and reduce cost.

METHODS

Data from 27 hospitals were retrieved on 81,929 hospitalizations of pediatric patients with a primary diagnosis of trauma, and for which the LOS was >24 h. Nested mixed effects model was used for simplified statistical inference, while a stochastic gradient boosting model, considering high-order statistical interactions, was built for prediction.

RESULTS

Over 18.7% of the encounters had LOS >1 week. Burns and corrosion and suspected and confirmed child abuse are the strongest drivers of prolonged LOS. Several other trauma-specific and general pediatric clinical variables were also predictors of prolonged LOS. The stochastic gradient model obtained an area under the receiver operator characteristic curve of 0.912 (0.907, 0.917).

CONCLUSIONS

The high performance of the machine learning model coupled with statistical inference from the mixed effects model provide an opportunity for targeted interventions to improve quality of care of trauma patients likely to require long length of stay.

IMPACT

Targeted interventions on high-risk patients would improve the quality of care of pediatric trauma patients and reduce the length of stay. This comprehensive study includes data from multiple hospitals analyzed with advanced statistical and machine learning models. The statistical and machine learning models provide opportunities for targeted interventions and reduction in prolonged length of stay reducing the burden of hospitalization on families.

Criticality index conducted in pediatric emergency department triage

OBJECTIVE

To develop and analyze the performance of a machine learning model capable of predicting the disposition of patients presenting to a pediatric emergency department (ED) based on triage assessment and historical information mined from electronic health records.

METHODS

We retrospectively reviewed data from 585,142 ED visits at a pediatric quaternary care institution between 2013 and 2020. An extreme gradient boosting machine learning model was trained on a randomly selected training data set (50%) to stratify patients into 3 classes: (1) high criticality (patients requiring intensive care unit [ICU] care within 4 h of hospital admission, patients who died within 4 h of admission, and patients who died in the ED); (2) moderate criticality (patients requiring hospitalization without the need for ICU care); and (3) low criticality (patients discharged home). Variables considered during model development included triage vital signs, aspects of triage nursing assessment, demographics, and historical information (diagnoses, medication use, and healthcare utilization). Historical factors were limited to the 6 months preceding the index ED visit. The model was tested on a previously withheld test data set (40%), and its performance analyzed.

RESULTS

The distribution of criticality among high, moderate, and low was 1.5%, 7.1%, and 91.4%, respectively. The one-versus-all area under the receiver operating characteristic (AUROC) curve for high and moderate criticality was 0.982 (95% CI 0.980, 0.983) and 0.968 (0.967, 0.969). The multi-class macro average AUROC and area under the receiver operating characteristic curve were 0.976 and 0.754. The features most integral to model performance included history of intravenous medications, capillary refill, emergency severity index level, history of hospitalization, use of a supplemental oxygen device, age, and history of admission to the ICU.

CONCLUSION

Pediatric ED disposition can be accurately predicted using information available at triage, providing an opportunity to improve quality of care and patient outcomes.

Development and validation of an early warning tool for sepsis and decompensation in children during emergency department triage

This study was designed to develop and validate an early warning system for sepsis based on a predictive model of critical decompensation. Data from the electronic medical records for 537,837 visits to a pediatric Emergency Department (ED) from March 2013 to December 2019 were collected. A multiclass stochastic gradient boosting model was built to identify early warning signs associated with death, severe sepsis, non-severe sepsis, and bacteremia. Model features included triage vital signs, previous diagnoses, medications, and healthcare utilizations within 6 months of the index ED visit. There were 483 patients who had severe sepsis and/or died, 1102 had non-severe sepsis, 1103 had positive bacteremia tests, and the remaining had none of the events. The most important predictors were age, heart rate, length of stay of previous hospitalizations, temperature, systolic blood pressure, and prior sepsis. The one-versus-all area under the receiver operator characteristic curve (AUROC) were 0.979 (0.967, 0.991), 0.990 (0.985, 0.995), 0.976 (0.972, 0.981), and 0.968 (0.962, 0.974) for death, severe sepsis, non-severe sepsis, and bacteremia without sepsis respectively. The multi-class macro average AUROC and area under the precision recall curve were 0.977 and 0.316 respectively. The study findings were used to develop an automated early warning decision tool for sepsis. Implementation of this model in pediatric EDs will allow sepsis-related critical decompensation to be predicted accurately after a few seconds of triage.

Identification of pediatric respiratory diseases using a fine-grained diagnosis system

Respiratory diseases, including asthma, bronchitis, pneumonia, and upper respiratory tract infection (RTI), are among the most common diseases in clinics. The similarities among the symptoms of these diseases precludes prompt diagnosis upon the patients' arrival. In pediatrics, the patients' limited ability in expressing their situation makes precise diagnosis even harder. This becomes worse in primary hospitals, where the lack of medical imaging devices and the doctors' limited experience further increase the difficulty of distinguishing among similar diseases. In this paper, a pediatric fine-grained diagnosis-assistant system is proposed to provide prompt and precise diagnosis using solely clinical notes upon admission, which would assist clinicians without changing the diagnostic process. The proposed system consists of two stages: a test result structuralization stage and a disease identification stage. The first stage structuralizes test results by extracting relevant numerical values from clinical notes, and the disease identification stage provides a diagnosis based on text-form clinical notes and the structured data obtained from the first stage. A novel deep learning algorithm was developed for the disease identification stage, where techniques including adaptive feature infusion and multi-modal attentive fusion were introduced to fuse structured and text data together. Clinical notes from over 12000 patients with respiratory diseases were used to train a deep learning model, and clinical notes from a non-overlapping set of about 1800 patients were used to evaluate the performance of the trained model. The average precisions (AP) for pneumonia, RTI, bronchitis and asthma are 0.878, 0.857, 0.714, and 0.825, respectively, achieving a mean AP (mAP) of 0.819. These results demonstrate that our proposed fine-grained diagnosis-assistant system provides precise identification of the diseases.

Multicenter study of risk factors of unplanned 30-day readmissions in pediatric oncology

Background

Pediatric oncology patients have high rates of hospital readmission but there is a dearth of research into risk factors for unplanned 30‐day readmissions among this high‐risk population.

Aim

In this study, we built a statistical model to provide insight into risk factors of unplanned readmissions in this pediatric oncology.

Methods

We retrieved 32 667 encounters from 10 418 pediatric patients with a neoplastic condition from 16 hospitals in the Cerner Health Facts Database and built a mixed‐effects model with patients nested within hospitals for inference on 75% of the data and reserved the remaining as an independent test dataset.

Results

The mixed‐effects model indicated that patients with acute lymphoid leukemia (in relapse), neuroblastoma, rhabdomyosarcoma, or bone/cartilage cancer have increased odds of readmission. The number of cancer medications taken by the patient and the administration of chemotherapy were associated with increased odds of readmission for all cancer types. Wilms Tumor had a significant interaction with administration of chemotherapy, indicating that the risk due to chemotherapy is exacerbated in patients with Wilms Tumor. A second two‐way interaction between recent history of chemotherapy treatment and infections was associated with increased odds of readmission. The area under the receiver operator characteristic curve (and corresponding 95% confidence interval) of the mixed‐effects model was 0.714 (0.702, 0.725) on the independent test dataset.

Conclusion

Readmission risk in oncology is modified by the specific type of cancer, current and past administration of chemotherapy, and increased health care utilization. Oncology‐specific models can provide decision support where model built on other or mixed population has failed.

Predictors of pediatric readmissions among patients with neurological conditions

Background

Unplanned readmission is one of many measures of the quality of care of pediatric patients with neurological conditions. In this multicenter study, we searched for novel risk factors of readmission of patients with neurological conditions.

Methods

We retrieved hospitalization data of patients less than 18 years with one or more neurological conditions. This resulted in a total of 105,834 encounters from 18 hospitals. We included data on patient demographics, prior healthcare resource utilization, neurological conditions, number of other conditions/diagnoses, number of medications, and number of surgical procedures performed. We developed a random intercept logistic regression model using stepwise minimization of Akaike Information Criteria for variable selection.

Results

The most important neurological conditions associated with unplanned pediatric readmissions include hydrocephalus, inflammatory diseases of the central nervous system, sleep disorders, disease of myoneural junction and muscle, other central nervous system disorder, other spinal cord conditions (such as vascular myelopathies, and cord compression), and nerve, nerve root and plexus disorders. Current and prior healthcare resource utilization variables, number of medications, other diagnoses, and certain inpatient surgical procedures were associated with changes in odds of readmission. The area under the receiver operator characteristic curve (AUROC) on the independent test set is 0.733 (0.722, 0.743).

Conclusions

Pediatric patients with certain neurological conditions are more likely to be readmitted than others. However, current and prior healthcare resource utilization remain some of the strongest indicators of readmission within this population as in the general pediatric population.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12883-020-02028-0.

Validity of the LACE index for identifying frequent early readmissions after hospital discharge in children

The LACE index scoring tool has been designed to predict hospital readmissions in adults. We aimed to evaluate the ability of the LACE index to identify children at risk of frequent readmissions. We analysed data from alive-discharge episodes (1 April 2017 to 31 March 2019) for 6546 males and 5875 females from birth to 18 years. The LACE index predicted frequent all-cause readmissions within 28 days of hospital discharge with high accuracy: the area under the curve = 86.9% (95% confidence interval = 84.3-89.5%, p < 0.001). Two-graph receiver operating characteristic curve analysis revealed the LACE index cutoff to be 4.3, where sensitivity equals specificity, to predict frequent readmissions. Compared with those with a LACE index score = 0-4 (event rates, 0.3%), those with a score > 4 (event rates, 3.7%) were at increased risk of frequent readmissions: age- and sex-adjusted odds ratio = 12.4 (95% confidence interval = 8.0-19.2, p < 0.001) and death within 30 days of discharge: OR = 5.0 (95% CI = 1.5-16.7). The ORs for frequent readmissions were between 6 and 14 for children of different age categories (neonate, infant, young child and adolescent), except for patients in the child category (6-12 years) where odds ratio was 2.8.Conclusion: The LACE index can be used in healthcare services to identify children at risk of frequent readmissions. Focus should be directed at individuals with a LACE index score above 4 to help reduce risk of readmissions. What is Known: • The LACE index scoring tool has been widely used to predict hospital readmissions in adults. What is New: • Compared with children with a LACE index score of 0-4 (event rates, 0.3%), those with a score > 4 are at increased risk of frequent readmissions by 14-fold. • The cutoff of a LACE index of 4 may be a useful level to identify children at increased risk of frequent readmissions.

Race, Ethnicity, and Insurance: the Association with Opioid Use in a Pediatric Hospital Setting

BACKGROUND

This study examined the association between race/ethnicity and health insurance payer type with pediatric opioid and non-opioid ordering in an inpatient hospital setting.

METHODS

Cross-sectional inpatient encounter data from June 2013 to June 2018 was retrieved from a pediatric children's hospital in Southern California (N = 55,944), and statistical analyses were performed to determine associations with opioid ordering.

RESULTS

There was a significant main effect of race/ethnicity on opioid and non-opioid orders. Physicians ordered significantly fewer opioid medications, but a greater number of non-opioid medications, for non-Hispanic African American children than non-Hispanic Asian, Hispanic/Latinx, and non-Hispanic White pediatric patients. There was also a main effect of health insurance payer type on non-opioid orders. Patients with government-sponsored plans (e.g., Medi-Cal, Medicare) received fewer non-opioid prescriptions compared with patients with both HMO and PPO coverage. Additionally, there was a significant race/ethnicity by insurance interaction on opioid orders. Non-Hispanic White patients with "other" insurance coverage received the greatest number of opioid orders. In non-Hispanic African American patients, children with PPO coverage received fewer opioids than those with government-sponsored and HMO insurance. For non-Hispanic Asian patients, children with PPO were prescribed more opioids than those with government-sponsored and HMO coverage.

CONCLUSION

Findings suggest that the relationship between race/ethnicity, insurance type, and physician decisions opioid prescribing is complex and multifaceted. Given that consistency in opioid prescribing should be seen regardless of patient background characteristics, future studies should continue to assess and monitor unequitable differences in care.

A Nested Mixed Effects Multicenter Model Examining the Risk Factors for Pediatric Trauma Return Visits Within 72 Hours

BACKGROUND

Return visits within 72 h are an important metric in evaluating the performance of emergency rooms. This has not been well studied in the pediatric trauma population. We sought to determine novel risk factors for return visits to the emergency department (ED) after trauma that may assist in identifying patients most at risk of revisit.

METHODS

We used the Cerner Health Facts Database to retrieve data from 34 EDs across the United States that care for pediatric trauma patients aged <15 y. The data consist of 610,845 patients and 816,571 ED encounters. We retrieved variables encompassing demographics, payor, current and past health care resource utilization, trauma diagnoses, other diagnoses/comorbidities, medications, and surgical procedures. We built a nested mixed effects logistic regression model to provide statistical inference on the return visits.

RESULTS

Traumas resulting from burns and corrosion, injuries to the shoulder and arms, injuries to the hip and legs, and trauma to the head and neck are all associated with increased odds of returning to the ED. Patients suffering from poisoning relating to drugs and other biological substances and patients with trauma to multiple body regions have reduced odds of returning to the ED. Longer ED length of stay and prior health care utilization (ED or inpatient) are associated with increased odds of a return visit. The sex of the patient and payor had a statistically significant effect on the risk of a return visit to the ED within 72 h of discharge.

CONCLUSIONS

Certain traumas expose patients to an increased risk for return visits to the ED and, as a result, provide opportunity for improved quality of care. Targeted interventions that include education, observation holds, or a decision to hospitalize instead of discharge home may help improve patient outcomes and decrease the rate of ED returns.

LEVEL OF EVIDENCE

III (Prognostic and Epidemiology).

A multicenter mixed-effects model for inference and prediction of 72-h return visits to the emergency department for adult patients with trauma-related diagnoses

Objective

Emergency department (ED) return visits within 72 h may be a sign of poor quality of care and entail unnecessary use of healthcare resources. In this study, we compare the performance of two leading statistical and machine learning classification algorithms, and we use the best performing approach to identify novel risk factors of ED return visits.

Methods

We analyzed 3.2 million ED encounters with at least one diagnosis under “injury, poisoning and certain other consequences of external causes” and “external causes of morbidity.” These encounters included patients 18 years or older from across 128 emergency room facilities in the USA. For each encounter, we calculated the 72-h ED return status and retrieved 57 features from demographics, diagnoses, procedures, and medications administered during the process of administration of medical care. We implemented a mixed-effects model to assess the effects of the covariates while accounting for the hierarchical structure of the data. Additionally, we investigated the predictive accuracy of the extreme gradient boosting tree ensemble approach and compared the performance of the two methods.

Results

The mixed-effects model indicates that certain blunt force and non-blunt trauma inflates the risk of a return visit. Notably, patients with trauma to the head and patients with burns and corrosions have elevated risks. This is in addition to 11 other classes of both blunt force and non-blunt force traumas. In addition, prior healthcare resource utilization, patients who have had one or more prior return visits within the last 6 months, prior ED visits, and the number of hospitalizations within the 6 months are associated with increased risk of returning to the ED after discharge. On the one hand, the area under the receiver characteristic curve (AUROC) of the mixed-effects model was 0.710 (0.707, 0.712). On the other hand, the gradient boosting tree ensemble had a lower AUROC of 0.698 CI (0.696, 0.700) on the independent test model.

Conclusions

The proposed mixed-effects model achieved the highest known AUC and resulted in the identification of novel risk factors. The model outperformed one of the leading machine learning ensemble classifiers, the extreme gradient boosting tree in terms of model performance. The risk factors we identified can assist emergency departments to decrease the number of unplanned return visits within 72 h.

HealtheDataLab - a cloud computing solution for data science and advanced analytics in healthcare with application to predicting multi-center pediatric readmissions

BACKGROUND

There is a shortage of medical informatics and data science platforms using cloud computing on electronic medical record (EMR) data, and with computing capacity for analyzing big data. We implemented, described, and applied a cloud computing solution utilizing the fast health interoperability resources (FHIR) standardization and state-of-the-art parallel distributed computing platform for advanced analytics.

METHODS

We utilized the architecture of the modern predictive analytics platform called Cerner® HealtheDataLab and described the suite of cloud computing services and Apache Projects that it relies on. We validated the platform by replicating and improving on a previous single pediatric institution study/model on readmission and developing a multi-center model of all-cause readmission for pediatric-age patients using the Cerner® Health Facts Deidentified Database (now updated and referred to as the Cerner Real World Data). We retrieved a subset of 1.4 million pediatric encounters consisting of 48 hospitals' data on pediatric encounters in the database based on a priori inclusion criteria. We built and analyzed corresponding random forest and multilayer perceptron (MLP) neural network models using HealtheDataLab.

RESULTS

Using the HealtheDataLab platform, we developed a random forest model and multi-layer perceptron model with AUC of 0.8446 (0.8444, 0.8447) and 0.8451 (0.8449, 0.8453) respectively. We showed the distribution in model performance across hospitals and identified a set of novel variables under previous resource utilization and generic medications that may be used to improve existing readmission models.

CONCLUSION

Our results suggest that high performance, elastic cloud computing infrastructures such as the platform presented here can be used for the development of highly predictive models on EMR data in a secure and robust environment. This in turn can lead to new clinical insights/discoveries.

Prediction of 7-Day Readmission Risk for Pediatric Trauma Patients

BACKGROUND

Pediatric patients admitted for trauma may have unique risk factors of unplanned readmission and require condition-specific models to maximize accuracy of prediction. We used a multicenter data set on trauma admissions to study risk factors and predict unplanned 7-day readmissions with comparison to the 30-day metric.

METHODS

Data from 28 hospitals in the United States consisting of 82,532 patients (95,158 encounters) were retrieved, and 75% of the data were used for building a random intercept, mixed-effects regression model, whereas the remaining were used for evaluating model performance. The variables included were demographics, payer, current and past health care utilization, trauma-related and other diagnoses, medications, and surgical procedures.

RESULTS

Certain conditions such as poisoning and medical/surgical complications during treatment of traumatic injuries are associated with increased odds of unplanned readmission. Conversely, trauma-related conditions, such as trauma to the thorax, knee, lower leg, hip/thigh, elbow/forearm, and shoulder/upper arm, are associated with reduced odds of readmission. Additional predictors include the current and past health care utilization and the number of medications. The corresponding 7-day model achieved an area under the receiver operator characteristic curve of 0.737 (0.716, 0.757) on an independent test set and shared similar risk factors with the 30-day version.

CONCLUSIONS

Patients with trauma-related conditions have risk of readmission modified by the type of trauma. As a result, additional quality of care measures may be required for patients with trauma-related conditions that elevate their risk of readmission.

Gait Speed at Discharge and Risk for Readmission or Death: A Prospective Study of an Emergency Ward Population

Background

There has been a growing interest in measuring gait speed for assessing long-term mortality and risk for hospital readmission in different populations.

Objective

We studied the association between a 10-meter gait speed test at hospital discharge and the risk for 30- and 90-day hospital readmission or death in a mixed population of patients hospitalized for emergency care.

Patients and Methods

Patients were prospectively included from 5 wards at the Karolinska University Hospital. The 10-meter gait speed test was measured on the day of discharge. Statistical analysis was performed using logistic regression.

Results

A total of 344 patients were included. Forty-one patients (n=41) were readmitted to hospital or died within 30 days, and 81 were readmitted or died within 90 days after discharge. Readmitted patients were older and had more comorbidities. A 0.1 m/s reduction in gait speed was associated with a 13% greater odds of readmission or death within 30 days (OR 1.13 [95% CI 1.00–1.26]). The area under the receiver operating characteristic curve (AUC) was 0.59 (95% CI 0.51–0.68). The results were similar for 90-day readmission or death where a 0.1 m/s decrement in gait speed was associated with an OR of 1.13 (95% CI 1.04–1.24). When age, eGFR, hemoglobin concentration, and active cancer, which all were univariate predictors of 30-day readmissions, were added to the model it yielded an AUC of 0.68 (95% CI 0.60 to 0.77).

Conclusion

In a mixed population of patients hospitalized for emergency care, low gait speed at discharge was associated with an increased risk of 30- and 90-day readmission or death. However, the test did not discriminate well between those who were readmitted or died and those who did not; therefore we do not recommend its use as a stand-alone test in this population.

Factors Associated With Return Visits at 7 Days After Hospital Discharge

OBJECTIVE

To identify variables associated with return visits to the hospital within 7 days after discharge.

METHODS

We performed a retrospective study of 7-day revisits and readmissions between October 2012 and September 2015 using the Pediatric Health Information System database supplemented by electronic medical record data from a tertiary-care children's hospital. We examined factors associated with revisits among the top 10 most frequent indications for hospitalization using generalized estimating equations.

RESULTS

There were 736 (4.2%) revisits and 416 (2.3%) readmissions within 7 days. Predictors of 7-day revisits and readmissions included age, length of hospital stay, and presence of a chronic medical condition. In addition, insurance status was associated with risk of revisits and race was associated with risk of readmissions in the bivariate analysis.

CONCLUSIONS

In this study, we identified patient characteristics that may be associated with a higher risk of early return to the emergency department and/or readmissions. Early identification of this at-risk group of patients may provide opportunities for intervention and enhanced care coordination at discharge.

Development and Validation of a Web-Based Pediatric Readmission Risk Assessment Tool

OBJECTIVES

Accurately predicting and reducing risk of unplanned readmissions (URs) in pediatric care remains difficult. We sought to develop a set of accurate algorithms to predict URs within 3, 7, and 30 days of discharge from inpatient admission that can be used before the patient is discharged from a current hospital stay.

METHODS

We used the Children's Hospital Association Pediatric Health Information System to identify a large retrospective cohort of 1 111 323 children with 1 321 376 admissions admitted to inpatient care at least once between January 1, 2016, and December 31, 2017. We used gradient boosting trees (XGBoost) to accommodate complex interactions between these predictors.

RESULTS

In the full cohort, 1.6% of patients had at least 1 UR in 3 days, 2.4% had at least 1 UR in 7 days, and 4.4% had at least 1 UR within 30 days. Prediction model discrimination was strongest for URs within 30 days (area under the curve [AUC] = 0.811; 95% confidence interval [CI]: 0.808-0.814) and was nearly identical for UR risk prediction within 3 days (AUC = 0.771; 95% CI: 0.765-0.777) and 7 days (AUC = 0.778; 95% CI: 0.773-0.782), respectively. Using these prediction models, we developed a publicly available pediatric readmission risk scores prediction tool that can be used before or during discharge planning.

CONCLUSIONS

Risk of pediatric UR can be predicted with information known before the patient's discharge and that is easily extracted in many electronic medical record systems. This information can be used to predict risk of readmission to support hospital-discharge-planning resources.

A Statistical-Learning Model for Unplanned 7-Day Readmission in Pediatrics

OBJECTIVES

The rate of pediatric 7-day unplanned readmissions is often seen as a measure of quality of care, with high rates indicative of the need for improvement of quality of care. In this study, we used machine learning on electronic health records to study predictors of pediatric 7-day readmissions. We ranked predictors by clinical significance, as determined by the magnitude of the least absolute shrinkage and selection operator regression coefficients.

METHODS

Data consisting of 50 241 inpatient and observation encounters at a single tertiary pediatric hospital were retrieved; 50% of these patients' data were used for building a least absolute shrinkage and selection operator regression model, whereas the other half of the data were used for evaluating model performance. The categories of variables included were demographics, social determinants of health, severity of illness and acuity, resource use, diagnoses, medications, psychosocial factors, and other variables such as primary care no show.

RESULTS

Previous hospitalizations and readmissions, medications, multiple comorbidities, longer current and previous lengths of stay, certain diagnoses, and previous emergency department use were the most significant predictors modifying a patient's risk of 7-day pediatric readmission. The model achieved an area under the curve of 0.778 (95% confidence interval 0.763-0.793).

CONCLUSIONS

Predictors such as medications, previous and current health care resource use, history of readmissions, severity of illness and acuity, and certain psychosocial factors modified the risk of unplanned 7-day readmissions. These predictors are mostly unmodifiable, indicating that intervention plans on high-risk patients may be developed through discussions with patients and parents to identify underlying modifiable causal factors of readmissions.

Prediction of 30-day pediatric unplanned hospitalizations using the Johns Hopkins Adjusted Clinical Groups risk adjustment system

Background

The Johns Hopkins ACG System is widely used to predict patient healthcare service use and costs. Most applications have focused on adult populations. In this study, we evaluated the use of the ACG software to predict pediatric unplanned hospital admission in a given month, based on the past year’s clinical information captured by electronic health records (EHRs).

Methods and findings

EHR data from a multi-state pediatric integrated delivery system were obtained for 920,051 patients with at least one physician visit during January 2009 to December 2016. Over this interval an average of 0.36% of patients each month had an unplanned hospitalization. In a 70% training sample, we used the generalized linear mixed model (GLMM) to generate regression coefficients for demographic, clinical predictors derived from the ACG system, and prior year hospitalizations. Applying these coefficients to a 30% test sample to generate risk scores, we found that the area under the receiver operator characteristic curve (AUC) was 0.82. Omitting prior hospitalizations decreased the AUC from 0.82 to 0.80, and increased under-estimation of hospitalizations at the greater risk levels. Patients in the top 5% of risk scores accounted for 43% and the top 1% of risk scores accounted for 20% of all unplanned hospitalizations.

Conclusions

A predictive model based on 12-months of demographic and clinical data using the ACG system has excellent predictive performance for 30-day pediatric unplanned hospitalization. This model may be useful in population health and care management applications targeting patients likely to be hospitalized. External validation at other institutions should be done to confirm our results.