This article was originally published in the Fall 2012 edition of OnAnalytics, published by the Institute for Business Analytics at Indiana University’s Kelley School of Business.
Contributors to this article include: Anwer Khan, Aditya Sane, Vik Wadhwani, and Mark Zozulia, all working with Deloitte Consulting at the time of original publication.
Total health care spending in the US — by individuals, corporations, and federal, state, and local governments — continues to amount to a significant proportion of the GDP, and is expected to reach 19.5% of GDP by 2017. Individual spending averaged $8,233 per year according to a recent report by the Organization for Economic Co-operation and Development. As such, all sectors of the health care industry are actively engaging in addressing cost and quality challenges using business analytics. Health care providers and related hospital-like settings must improve quality while reducing costs. Managed care and health insurance payers also struggle to reduce costs, but attempt to differentiate themselves by launching and administering disease management and wellness programs more efficiently and effectively than their competitors. Life science organizations, meanwhile, are trying to stay ahead of patent cliffs by searching for derivate products across new therapeutic classes to serve more targeted cohorts.
A number of ways in which business analytics are currently applied across different fields of health care include:
- Health Care Providers: Workflow improvement, re-admission prediction and management, reducing drug interactions, supply chain optimization, and patient load forecasting.
- Health Insurance Payers: Targeted care management and wellness programs; re-admission reduction; targeted marketing; formulary optimization; network design; provider profiling and quality based reimbursements; and fraud, waste, and abuse detection.
- Life Science Organizations: Comparative effectiveness, signal detection, accelerated drug approvals with phase 4 virtual trials, clinical trial recruitment and enrollment, and translational medicine collaboration.
- Public Health Agencies: Disease outbreak prediction; chronic conditions projection; adverse events monitoring; and fraud, waste and abuse detection (for Medicare/Medicaid).
Health care data proliferation
Business analytics goes beyond traditional operational reporting to offer health care organizations the detailed insights into trends and predictors needed to address these and other challenges. Through analytical models and advanced data visualization, applied analytics can provide a range of approaches and solutions from looking backward to evaluate performance to forwardlooking scenario planning and predictive modeling. The rapidly more digital health care ecosystem uses and creates an endless stream of information from a large number of sources, such as medical records, claims systems, clinical registries, financial and payment reimbursement systems, imaging applications, and home health devices, just to name a few. Analytic methods such as classification, logistic regression, decision trees, neural networks, segmentation, time series, sequence models, outlier detection, and natural language processing are some of the tools applied to create actionable recommendations from this vast universe of data. The goal of applied health care analytics, however, is not to analyze everything, but rather to analyze the right things at the right time.
Taking inpatient readmissions as an example, research shows that nearly 20% of inpatient admissions result in readmission. Approximately 90% of readmissions are preventable. These unplanned readmissions have an estimated $42 billion national cost. Moreover, financial penalties and bonuses based on quality outcomes are now being enforced. For instance, Medicare payments for readmissions in select areas will be reduced beginning October 2012. Providers will assume part of the financial risk, which averages $16,000 per readmission.
Recently, our team used business analytics to create a re-admission model using a mathematical algorithm pitted against a set of iterative decision trees that calculated the propensity of re-admission by disease class. A multi-facility hospital system in the western U.S. customized this model to its own data and was able to identify 4,000 cases for intervention. After re-aligning case manager workflows to target the intervention cases, this hospital system projects an incremental net savings of over $1 million per facility.
Business analytics will play a crucial role in improving health care outcomes going forward. Ultimately, such improvement will depend on (a) solid understanding, integration, and management of existing and newer data sources; (b) application of advanced analytical models and techniques; and (c) integration of analytical outputs with existing operational processes (e.g. hospital care management workflows). To maximize effectiveness and reduce problems like preventable readmissions, health care organizations must invest in optimizing their analytics infrastructure and exploring emerging architectures to support the utilization of applied analytics.