Small Data is as Valuable as Big Data in Healthcare!

The good news is that the use of big data to improve healthcare outcome is quickly approaching. The tricky news is that the use of big data for improving healthcare outcome is quickly approaching.

It is clear that healthcare information can provide insight into a patient’s health which allows the provider to personalize communications and provide more feedback for the patients benefit. It is also clear that analyzing a “lot” of data, particularly when originating from different sources takes on a corresponding level of challenges that encompass cost, security, interoperability, distribution and manageability.

The temptation with “big data” is to approach improving patient outcome goals by examining an ever increasing set of data which may extend across disparate sources like electronic health records, genome research, and behavior data. With each of those sources increasing in content on an ever shortening interval.

Another challenge to the approach to improve healthcare outcome through big data is supporting patients residing in underserved communities. What happens to those patients with limited communications with their provider, or the insured who cannot afford their current health care bills, let alone additional complex data analysis? Those patient’s most vulnerable may not be able to take advantage of the analysis that would arguably benefit the most.

Let’s take an example to show how big data may get in the way of solving a healthcare dilemma. The Centers for Disease Control and Prevention reports that over 30 million people, or more than nine percent of the total United States population, had Diabetes in 2016. It was the seventh leading cause of death in 2015 and 1.5 million new cases are identified each year. Diabetes is a chronic illness in which the level of sugar in someone’s blood (measured as a Hemoglobin A1C) is too high. It can lead to visual impairment, peripheral vascular disease, kidney disease and early death. Diabetes also may result in lost productivity and disability. “Diabetes and its related complications account for $245 billion in total medical costs to include lost work and wages in 2012.” In 2017 this cost was estimated at $327B. In the United States, the number of emergency department visits with diabetes as part of the any-listed diagnosis increased from 9,464,000 in 2006 to 11,492,000 in 2009.

Direct medical costs of patients with diabetes were significantly higher than those of patients without diabetes, with higher costs stemming from avoidable hospitalizations and renal issues.

Unfortunately, a key step to improving and maintaining a better level of health for diabetes patients is based on the patients willingness to “take their medication” as prescribed. Poor medication adherence for patients with this chronic ailment (and others as well) range from 30% to 50% depending on certain variables.

A temptation with using big data to resolve the issue of improving medication adherence for chronically ill patients is to examine and review the entirety of the patient’s data, including electronic medical records, socio economic research, behavior patterns encompassing caloric intake, quality of diet and level of exercise! Yikes: synthesizing that amount of data and developing a patient engagement program to improve their total behavior is overwhelming. And most likely won’t be implemented equally across all communities.

The use of small data to help patients improve their medication adherence behavior, versus big data may be an option. 

What the heck is small data? Small data is using a smaller and more precise amount of patient information to remedy a more defined issue. In the case of improving medication adherence we examine a narrow collection of patient behavior data to determine if they are taking their meds on time, and why not. Notwithstanding the patients reported adherence is the success of their actions against certain metrics. In the case of our diabetes patient- it is their self-reported adherence resulting in improvement in the overall level of their hemoglobin A1C that is the ultimate measure of success, for example.

This single measure is certainly not the only variable when managing a diabetes patients overall health outcome, yet there are many advantages to doing so:

1.   Engaging the patient to educate them of the importance of med adherence and tracking their behavior is straight forward. There are many simple solutions like text messaging, phone calls, family “kick in the shins”, etc. that are inexpensive and easy to implement.

2.   Electronic medical records have API’s so the patient data collected to monitor med compliance can be integrated with the emr system. Or managed independent of the system.

3.   Patient’s that meet their targeted goals can receive recognition from their doctor and provider and feel they are part of a team that is concerned about their well-being.

4.   Family members can provide encouragement to the patient to meet their goals with special attention when those goals are meet.

5.   Such solutions can be more easily implemented in a HIPAA compliant manner.

6.   Additional incentives can be formulated for patients when they meet their goals, like discounts for various things.

7.   Certain results, like improving a patients A1C measure, can be used to accelerate reimbursement to the provider for meeting certain Quality of Payment Program measures.

Using small data requires a more targeted approach to help patients. While big data requires constant collection and analysis of more and more data, small data requires a specification of outcome to define what data is to be collected and studied in the first place. In this example, collecting med adherence data to help patients improve their A1C value, is based on the research that an improvement in a patients A1C value from 8% to 7% can have a dramatic impact to reducing avoidable costs, reducing long term care investment and dramatically improving the quality of life for the patient. Even though many areas of behavior impact the diabetes patient’s health, medication adherence is the first step in the engagement process with other measures to follow.

Using these levels of data as incremental steps toward fixing bigger health issues may also be a strategy that truly vectors healthcare to a more patient centric focus. Small data, being more manageable is also more understood by healthcare stakeholders. In the case of patient medication adherence behavior data- understanding the reasons why a patient is or is not taking their medication is valuable information for patient, care provider, pharma and insurer. Identifying the reason for lack of proper adherence gives each of these stakeholders data from which to develop solutions that ultimately contribute to improved quality and lower costs.

Incremental levels of patient data also enable applications such as blockchain. From a business perspective it is easier to determine the value of a succinct data set, and to more readily identify the entities that appreciate its value over very large and complex data sets. From a technology perspective, one argument is that less quantities of data is easier to manage, as well.

Author is Terry Wolters with www.NotifiUs.com. Contact at terry@ notifius.com. He is CEO of NotifiUs, LLC which focuses on helping care providers help their patients to improve medication adherence behavior through the NotifiUs Patient Engagement Platform and templates such as the Diabetes Engagement Program. Terry is a technologist, entrepreneur, veteran and very happy granddad to Rhett and Savannah.