OUP accepted manuscript

Objective

Health care providers increasingly rely upon predictive algorithms when making important treatment decisions, however, evidence indicates that these tools can lead to inequitable outcomes across racial and socio-economic groups. In this study, we introduce a bias evaluation checklist that allows model developers and health care providers a means to systematically appraise a model’s potential to introduce bias.

Materials and Methods

Our methods include developing a bias evaluation checklist, a scoping literature review to identify 30-day hospital readmission prediction models, and assessing the selected models using the checklist.

Results

We selected 4 models for evaluation: LACE, HOSPITAL, Johns Hopkins ACG, and HATRIX. Our assessment identified critical ways in which these algorithms can perpetuate health care inequalities. We found that LACE and HOSPITAL have the greatest potential for introducing bias, Johns Hopkins ACG has the most areas of uncertainty, and HATRIX has the fewest causes for concern.

Discussion

Our approach gives model developers and health care providers a practical and systematic method for evaluating bias in predictive models. Traditional bias identification methods do not elucidate sources of bias and are thus insufficient for mitigation efforts. With our checklist, bias can be addressed and eliminated before a model is fully developed or deployed.

Conclusion

The potential for algorithms to perpetuate biased outcomes is not isolated to readmission prediction models; rather, we believe our results have implications for predictive models across health care. We offer a systematic method for evaluating potential bias with sufficient flexibility to be utilized across models and applications.

Data Science Methods for Nursing-Relevant Patient Outcomes and Clinical Processes: The 2019 Literature Year in Review

Data science continues to be recognized and used within healthcare due to the increased availability of large data sets and advanced analytics. It can be challenging for nurse leaders to remain apprised of this rapidly changing landscape. In this article, we describe our findings from a scoping literature review of papers published in 2019 that use data science to explore, explain, and/or predict 15 phenomena of interest to nurses. Fourteen of the 15 phenomena were associated with at least one paper published in 2019. We identified the use of many contemporary data science methods (eg, natural language processing, neural networks) for many of the outcomes. We found many studies exploring Readmissions and Pressure Injuries. The topics of Artificial Intelligence/Machine Learning Acceptance, Burnout, Patient Safety, and Unit Culture were poorly represented. We hope that the studies described in this article help readers: (1) understand the breadth and depth of data science's ability to improve clinical processes and patient outcomes that are relevant to nurses and (2) identify gaps in the literature that are in need of exploration.

Evaluation of a multidisciplinary approach to reduce internal medicine readmissions using a readmission prediction index

Purpose

Readmission prediction indices are used to stratify patients by the risk of hospital readmission. We describe the integration of a 30-day hospital readmission prediction index into the electronic medical record (EMR) and its impact on pharmacist interventions during transitions of care (TOC).

Methods

A retrospective cohort study was conducted to compare 30-day readmission rates between adult internal medicine inpatients admitted by a multidisciplinary team providing TOC services (the TOC group) and those who received usual care (the control group). Interventions by a pharmacist serving on the TOC team were guided by an EMR-integrated readmission index, with patients at the highest risk for readmission receiving targeted pharmacist interventions. Inpatient encounters (n = 374) during the 5-month study period were retrospectively identified. Chi-square and Mann-Whitney U tests were performed to analyze differences in nominal and nonparametric continuous variables, respectively. Logistic regression was performed to identify variables associated with 30-day readmissions. The log-rank test was used to analyze hazard ratios for readmission outcomes in the 2 cohorts.

Results

Thirty-day readmission rates did not differ significantly in the TOC group and the control group (20.9% vs 18.3%, P = 0.52). However, patients who received additional direct pharmacist interventions, as guided by use of a hospital readmission index, had a lower 30-day readmission rate than patients who did not (11.4% vs 21.7%, P = 0.04). The readmission index score was significantly associated with the likelihood of 30-day readmission (odds ratio for readmission, 1.25; 95% confidence interval, 1.16-1.34; P < 0.01). The difference in unadjusted log-rank scores at 30 days with and without pharmacist intervention was not significant (P = 0.05). A mean of 4.5 medication changes were identified per medication reconciliation performed by the TOC pharmacist.

Conclusion

A multidisciplinary TOC team approach did not reduce the 30-day readmission rate on an internal medicine service. However, patients who received additional direct pharmacist interventions guided by a readmission prediction index had a reduced readmission rate.

The Impact of Transitions of Care Pharmacist Services and Identification of Risk Predictors in Heart Failure Readmission

BACKGROUND

Heart failure (HF) is a prevalent and costly disease state for adult Americans, with 30-day readmissions rates for patients with HF utilized to limit hospital compensation.

OBJECTIVE

To determine the impact of the transitions of care (TOC) service at our institution on 30-day all-cause and HF readmissions and identify predictive risk factors for 30-day all-cause readmission.

METHODS

Retrospective chart review of patients aged 18 years and older admitted with HF and all subsequent readmissions between October 1, 2015, and September 30, 2017. A weighted logistic regression model was developed to determine risk factors for 30-day all-cause readmission.

RESULTS

There were no significant differences in all-cause or HF readmission rates analyzed by TOC service involvement. Significant risk predictors for 30-day all-cause readmission included discharge to a rehabilitation facility (odds ratio [OR] = 9.3) or home with home health (OR = 1.6) versus home with self-care. Comorbidities associated with an increased risk of 30-day all-cause readmission included diabetes, coronary artery disease, and aortic stenosis. Use of angiotensin-converting enzyme inhibitors, angiotensin-receptor blockers, and spironolactone was associated with decreased risk of 30-day all-cause readmission.

CONCLUSION

Identified predictors in the patient population with HF at our institution may be used to target patients at increased risk of all-cause readmission within 30 days.