Big Data in Clinical Decision Support

Executive Summary

The need for robust financial management is the main driver for the use of use of any analytics or decision support system in a healthcare organization. (Prestigiacomo, 2012) If the healthcare organization can spot a patient at risk of readmission days in advance, it can take steps to intervene and prevent readmission and thus avoid financial penalties.

Accountable Care Organizations (ACO) will be the main driver for analytics solutions due to their participation in the Medicare Shared Savings Program (MSSP). The Shared Savings Program will reward ACOs that lower their growth in health care costs while meeting performance standards on quality of care and putting patients first. (Centers for Medicare and Medicaid, 2014) There are various quality measures that an ACO must meet to be able to share the cost savings. The meaningful use guidelines drive the requirements for these measures. The quality measures are divided into the following domains:

  • Patient/Caregiver Experience
  • Care Coordination/Patient Safety
  • Preventive Health
  • At-Risk Population

The Accountable Care Organizations are dependent on software and analytics to help meet these requirements. Just in the analytics arena in healthcare, there are many different types of solutions. In this paper, I am going to focus on some of the key drivers for big data solution in the area of Clinical Decision Support.

Introduction

Clinical Decision Support (CDS) refers broadly to providing clinicians or patients with clinical knowledge and patient-related information, intelligently filtered, or presented at appropriate times, to enhance patient care. (Osheroff, Pifer, Teich, Sittig, & Jenders, 2005) A Clinical Decision Support System (CDSS) is defined as software that integrates information on the characteristics of individual patients with computerized knowledge base for the purpose of generating patient-specific assessments or recommendations designed to aid clinicians and/or patients in making clinical decisions. (Institute of Medicine, 2001)

Key Drivers

If the ACO business model has to succeed, these organizations need to transform themselves into data savvy business operators. Now, more than ever, they are dependent on sophisticated big data analytics to provide them with key insights to keep their business profitable while providing better care.

The key drivers for a Clinical Decision Support solution are:

  • Impact of meaningful use
  • Rise of Accountable Care Organizations
  • Need to reduce financial risk
  • Resurgence of data warehouses
  • Centered on privacy protection

Impact of Meaningful Use

Meaningful use is the biggest driver for data storage and usage in the healthcare industry. It has contributed directly to the rise of ACOs and it aligns financial incentives for physicians and hospitals with that of positive healthcare outcomes for patients. For more on meaningful use, please read my blog post.

Rise of Accountable Care Organizations (ACO)

ACOs are a way to measure the impact of meaningful use on the cost of care. The Centers for Medicare and Medicaid has approved the formation of ACOs across the country. Each ACO has to agree to manage a certain number of patients (5,000 or more) for at least 3 years. Depending on the costs savings, these ACOs will be allowed get a part of the cost savings as incentive for keep the population healthy.

The ACOs will have to measure their performance on a daily basis and they are dependent on software to help them with their cause. The CIO for each of these ACOs, now plays a central role in the healthcare organization.

Need to Reduce Financial Risk

When physicians and hospitals were operating in a fee-for-service (FFS), they faced very little financial risk and accountability. But, the ACO model raises both penalties and incentives.

Screen Shot 2014-11-14 at 6.25.26 AM

Source: Frakt, A., How ACOs are not like 1990s-style capitation, (2012), AcademyHealth Blog

Fee For Service (FFS)

FFS is a payment model by which most of the healthcare industry in the U.S. operated until now. The physician is paid for each time he/she sees the patient, irrespective of the outcome for the patient. There was very little risk or incentive for the physician to ensure that the patient was healthy.

Capitation

Capitation is a payment model where the payer pays the healthcare provider with a fixed amount for each person assigned to him or her, per period of time, whether or not they seek care.

Resurgence of Data Warehouses

Data warehouses built on relational database technologies have been around for a very long time. But the use of data warehouses in healthcare organizations has been pretty limited until now. The healthcare professionals have been using standardized code sets, controlled vocabulary, and language for many years to facilitate communication and billing. Due to the structure that standards such as ICD-10-CM, SNOMED, and LOINC bring to the data; analytics can be performed using traditional relational databases. There is a role for NoSQL or other newer database technologies in this arena to store and analyze clinical notes, medical images, etc.

Centered on Privacy Protection

Irrespective of whether CDS systems are deployed on-premise or uses a third-party SaaS based product, privacy of patient data is a major concern. So, before sharing data into vast data warehouses, any personally identifiable information is completely removed. So, this leads to duplication of data across various systems. Also, data warehouses usually have different schema definition compared to transactional systems, so duplication of data may be a necessary to conduct fast analytics. If a healthcare organization is using a SaaS vendor for their analytics, then data needs to be transferred to the vendor for them to conduct the analysis and produce results. Finally, there are different systems that a CDS system need to tap into to get the data for analytics. When all this data is brought into one system, it makes it easy to do the analytics. So, the main drivers for storage for CDS are:

  • Need to protect privacy.
  • Database schema differences between Transactional and Data Warehouse.
  • Bring data from multiple healthcare applications into a single system.
  • Need to share data with a SaaS vendor.

Here are some of the inputs to a CDS system

Inputs into CDS:

A typical CDS takes many inputs such as:

  • Electronic Medical Records
  • Patient Billing
  • Costing
  • Medications
  • Problem List
  • Lab Results
  • Pathology Reports
  • Operative Reports
  • Patient History
  • Family History
  • Genetics
  • Clinical Notes

Any CDS system can be expected to operate in two modes:

  • Batch Mode.
  • Near Real-time.

The system, performance, and service level agreement (SLA) requirements vary for both these operation types. For example, to operate at near real-time, a system may require flash or in-memory hardware to perform analytics.

Also, more of the CDS systems are being deployed as a cloud service, which leads to having totally different requirements for hardware and storage compared to systems deployed on-premise.

Companies in the CDS Space

Here are some companies operating in this space:

Accountable Care Organizations

Here are some of the ACOs currently in operation:

For more information on ACOs, please visit this page on CMS.gov.

Works Cited

Centers for Medicare and Medicaid. (2014, March 11). Shared Savings Program. Retrieved November 25, 2014, from Centers for Medicare & Medicaid services: http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/index.html

Institute of Medicine. (2001). Committee on Quality of Health care in America. Institute of Medicine. Washington D.C.: National Academy Press.

Osheroff, J. A., Pifer, E. A., Teich, J. M., Sittig, D. F., & Jenders, R. A. (2005). Improving Outcomes with Clinical Decision Support. An Implementer’s Guide. Chicago, IL: Himss.

Peters, G. S. (2013). Advancing the Electronic Health Record.

Prestigiacomo, J. (2012, September 4). Survey: 77% of Healthcare Organizations Use Analytics Software. Retrieved November 25, 2014, from Healthcare Informatics: http://www.healthcare-informatics.com/article/survey-77-healthcare-organizations-use-analytics-software

Shea, S., DuMouchel, W., & Bahamonde, L. A meta-analysis of 16 randomized controlled trials to evaluate computer-based clinical reminder systems for preventive care in ambulatory setting. Journal of American Medical Information Association.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s