Improve Clinical Decision Support with ‘Big Data’ Analytics

Improve Clinical Decision Support with ‘Big Data’ Analytics

As the American health care system continues to move away from traditional fee-for-service payment for physicians and moves towards value-based reimbursement, the ability to use data to inform decision-making becomes more critical. Some say that ‘Big Data’ analytics is the answer.

Big data can be thought of as having three components: a large amount of volume, a large velocity of data, and a large variety of data. Other industries have been using big data analytics for decades to predict behaviors and target populations.

Amazon uses big data to predict and display items one might be interested in purchasing, based on previous purchasing behavior from other customers. Netflix does the same to suggest interesting movies and shows. The department store Target has one of the most sophisticated data analytics systems to predict future purchases; however, Target also utilizes this information in their direct marketing to individualize the message and offers to their customers. Big data analysis is all around us and has been used and refined for years in other markets.

However, healthcare data analytics lags far behind. Several reasons account for this discrepancy. First, the platform for gathering the data is the EHR, which did not achieve universal adoption in the U.S. until recently.

Today, around 90% of physicians collect healthcare data in an EHR. Just 20 years ago, only about 1% of all healthcare data was in a digitized format. Second, and this still applies today, not all EHRs and databases are designed for conducting proper analyses.

Many EHRs still have patient encounter information stored as unstructured data or “freestyle” data. Many EHR companies have solved the difficulty of capturing patient histories and other subjective data by capturing the data in an unstructured format. This data storage format is extremely difficult to properly query for true analytical purposes. For truly effective big data analytics, the data must be ‘granularized’ and stored in a structured format. There is also no universal database for EHRs, or universally accepted data descriptors (i.e., is it First Name or is it FName in your database?). Without EHRs being designed specifically with structured data and research capabilities in mind – things can often be “garbage data in, therefore garbage data out”.

Big data analysis for medical computerized decision support, also known as clinical decision support (CDS), is clearly in its infancy stage. In fact, you probably rolled your eyes when you read that sentence!

The issue is providing and displaying to the physician the right information at the right time – through the EHR – at the point-of-care. Most EHR companies claim to have built-in clinical decision support (CDS). These CDS features are almost always in the form of pop-up windows and reminders. The main problem is that these support parameters are merely reminders of the obvious!

A two-year-old who is behind in their immunizations triggers a pop-up window reminding the physician to vaccinate the child. A pop-up window “reminds” the physician that a patient’s blood pressure is too high, or a lab result lies outside normal values. A patient with the chief complaint of a sore throat triggers an alert to order a rapid strep test.

True clinical decision support is not simply providing information that is geared towards diagnosis and treatment. It should go beyond, to be fundamentally about disease management. Clinical decision support should also be instrumental in influencing the physician’s medical decision-making process. Changing a physician’s treatment behavior at the point-of-care is at the heart of value-based medicine, thereby having the potential to improve clinical outcomes and the health of the population.

Historically, a physician’s behavior and decisions in regards to disease management have mostly been influenced by journal articles, medical conferences, lectures, CDC or NIH recommendations, and even by drug reps pushing a new pharmaceutical agent. These are retrospective means of influencing physician behavior and display remarkably slow universal adoption rates by most physicians.

So it’s not surprising that physician treatment management behavior is largely based on their own treatment successes and failures. Basically, their own experience.

The EHR and real-time data analysis regarding a physician’s treatment and prescribing behaviors are paramount to improving the way physicians practice medicine. Providing analytical information to physicians based on the success or failure of their own treatment plans, at the point-of-care, is the answer.

The timing of the treatment outcome analysis should be displayed to the physician via the EHR after a diagnosis is made, but before a treatment is prescribed or recommended. Data analytics should be performed which displays the real-time effectiveness and outcomes of prescribing various antibiotics or other medications for certain diseases, or the true efficacy of other types of treatment.

Thus, the physician can use their own treatment outcome information to influence point-of-care prescribing behavior. Overall patient treatment outcomes will improve, less effective treatments are more quickly abandoned, and population health improves one doctor-patient treatment relationship at a time.

Xcite Health has the only physician practice platform that lets physicians go home on time! Contact us to find out why — and how we can make this commitment to you – that you will go home on time!  Call (800) 924-8344 or email us at info@xcitehealth.com to book a demo.

No Comments Yet.

Leave a comment