Risk of Frequent Emergency Department Use Among an Ambulatory Care Sensitive Condition Population: A Population-based Cohort Study

Machine learning to improve frequent emergency department use prediction: a retrospective cohort study

Frequent emergency department use is associated with many adverse events, such as increased risk for hospitalization and mortality. Frequent users have complex needs and associated factors are commonly evaluated using logistic regression. However, other machine learning models, especially those exploiting the potential of large databases, have been less explored. This study aims at comparing the performance of logistic regression to four machine learning models for predicting frequent emergency department use in an adult population with chronic diseases, in the province of Quebec (Canada). This is a retrospective population-based study using medical and administrative databases from the Régie de l'assurance maladie du Québec. Two definitions were used for frequent emergency department use (outcome to predict): having at least three and five visits during a year period. Independent variables included sociodemographic characteristics, healthcare service use, and chronic diseases. We compared the performance of logistic regression with gradient boosting machine, naïve Bayes, neural networks, and random forests (binary and continuous outcome) using Area under the ROC curve, sensibility, specificity, positive predictive value, and negative predictive value. Out of 451,775 ED users, 43,151 (9.5%) and 13,676 (3.0%) were frequent users with at least three and five visits per year, respectively. Random forests with a binary outcome had the lowest performances (ROC curve: 53.8 [95% confidence interval 53.5-54.0] and 51.4 [95% confidence interval 51.1-51.8] for frequent users 3 and 5, respectively) while the other models had superior and overall similar performance. The most important variable in prediction was the number of emergency department visits in the previous year. No model outperformed the others. Innovations in algorithms may slightly refine current predictions, but access to other variables may be more helpful in the case of frequent emergency department use prediction.

Association between sustained opioid prescription and frequent emergency department use: a cohort study

BACKGROUND

Chronic non-cancer pain (CNCP) is common among frequent emergency department (ED) users, although factors underlying this association are unclear. This study estimated the association between sustained opioid use and frequent ED use among patients with CNCP.

METHODS

Retrospective cohort study using a Canadian provincial health insurer database (Régie d'Assurance Maladie du Québec). The database included adults with both ≥1 chronic condition and ≥ 1 ED visit in 2012 or 2013. Inclusion in the study further required a CNCP diagnosis, public drug insurance coverage and 1-year survival after the first ED visit in 2012 or 2013 (index visit). Multivariable logistic regression was used to derive ORs of frequent ED use (≥5 visits in the year following the index visit) subsequent to sustained opioid use (≥60 days opioids prescription within 90 days preceding the index visit), adjusting for important covariables.

RESULTS

From 576 688 patients in the database, 58 237 were included in the study. Of these, 4109 (7.1%) had received a sustained opioid prescription and 4735 (8.1%) were frequent ED users in the follow-up year. Sustained opioid use was not associated with frequent ED use in the multivariable model (OR: 1.06, 95% CI 0.94 to 1.19). Novel associated covariables were benzodiazepine prescription (OR: 1.21, 95% CI 1.12 to 1.30) and polypharmacy (OR: 1.23, 95% CI 1.13 to 1.34).

CONCLUSIONS

Due to confounding by social and medical vulnerability, patients with CNCP with sustained opioid use appear to have a higher propensity for frequent ED use in unadjusted models. However, sustained opioid use was not associated with frequent ED use in these patients after adjustment.

Development and internal validation of prediction models for future hospital care utilization by patients with multimorbidity using electronic health record data

Objective

To develop and internally validate prediction models for future hospital care utilization in patients with multiple chronic conditions.

Design

Retrospective cohort study.

Setting

A teaching hospital in the Netherlands (542 beds)

Participants

All adult patients (n = 18.180) who received care at the outpatient clinic in 2017 for two chronic diagnoses or more (including oncological diagnoses) and who returned for hospital care or outpatient clinical care in 2018. Development and validation using a stratified random split-sample (n = 12.120 for development, n = 6.060 for internal validation).

Outcomes

≥2 emergency department visits in 2018, ≥1 hospitalization in 2018 and ≥12 outpatient visits in 2018.

Statistical analysis

Multivariable logistic regression with forward selection.

Results

Evaluation of the models’ performance showed c-statistics of 0.70 (95% CI 0.69–0.72) for the hospitalization model, 0.72 (95% CI 0.70–0.74) for the ED visits model and 0.76 (95% 0.74–0.77) for the outpatient visits model. With regard to calibration, there was agreement between lower predicted and observed probability for all models, but the models overestimated the probability for patients with higher predicted probabilities.

Conclusions

These models showed promising results for further development of prediction models for future healthcare utilization using data from local electronic health records. This could be the first step in developing automated alert systems in electronic health records for identifying patients with multimorbidity with higher risk for high healthcare utilization, who might benefit from a more integrated care approach.

Predicting unplanned hospital visits in older home care recipients: a cross-country external validation study

Background

Accurate identification of older persons at risk of unplanned hospital visits can facilitate preventive interventions. Several risk scores have been developed to identify older adults at risk of unplanned hospital visits. It is unclear whether risk scores developed in one country, perform as well in another. This study validates seven risk scores to predict unplanned hospital admissions and emergency department (ED) visits in older home care recipients from six countries.

Methods

We used the IBenC sample (n = 2446), a cohort of older home care recipients from six countries (Belgium, Finland, Germany, Iceland, Italy and The Netherlands) to validate four specific risk scores (DIVERT, CARS, EARLI and previous acute admissions) and three frailty indicators (CHESS, Fried Frailty Criteria and Frailty Index). Outcome measures were unplanned hospital admissions, ED visits or any unplanned hospital visits after 6 months. Missing data were handled by multiple imputation. Performance was determined by assessing calibration and discrimination (area under receiver operating characteristic curve (AUC)).

Results

Risk score performance varied across countries. In Iceland, for any unplanned hospital visits DIVERT and CARS reached a fair predictive value (AUC 0.74 [0.68–0.80] and AUC 0.74 [0.67–0.80]), respectively). In Finland, DIVERT had fair performance predicting ED visits (AUC 0.72 [0.67–0.77]) and any unplanned hospital visits (AUC 0.73 [0.67–0.77]). In other countries, AUCs did not exceed 0.70.

Conclusions

Geographical validation of risk scores predicting unplanned hospital visits in home care recipients showed substantial variations of poor to fair performance across countries. Unplanned hospital visits seem considerably dependent on healthcare context. Therefore, risk scores should be validated regionally before applied to practice. Future studies should focus on identification of more discriminative predictors in order to develop more accurate risk scores.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-021-02521-2.