The Big Data Healthcare Is Ignoring

By Kurt Waltenbaugh, founder and CEO, Carrot Health

In the years since the mandated adoption of electronic health records (EHRs), healthcare providers and technology developers have sought out solutions to leverage the resultant newfound abundance of patient data to improve outcomes. There’s just one problem. The value of that data is often limited to immediate treatment at a hospital or clinic—treatment that, even if successful, does not guarantee the patient will remain healthy.

Too often patients walk out of a healthcare facility with little more than a Band-Aid concealing a more serious condition, through no fault of the medical staff. At some point in time, every doctor and scientist trying to solve a problem has asked themselves, “What am I missing?”

Health informaticists, analysts and other data experts continue diving deeper into the data contained within their own clinical systems in the hopes of solving today’s medical puzzles. Their focus, however, is too narrow, as the key to solving healthcare’s greatest mysteries lies outside the hospital walls.

Expanding The Focus

Roughly 88 percent of U.S. healthcare spending goes toward patient care, yet medical needs and encounters determine just 10 percent of an individual’s overall health. Conversely, socioeconomic and behavioral determinants of health impact approximately 60 percent of overall health, but make up a mere 4 percent of the national health budget.

This disparity in funding in relation to their impacts reveals a new challenge for the healthcare industry—how to leverage socioeconomic data to more effectively target healthcare resources. The solution lies in the answers to two questions: 1) what do we need to know and 2) how can we harvest that information to create an actionable solution?

Different variables have different weights, depending on the condition and population in question. For example, social isolation compounds any diagnosis. A senior living alone and with limited access to transportation has a much harder time making it to medical appointments and picking up medications from the pharmacy. This severely limits physicians’ ability to treat these patients effectively and, in turn, puts the patient at high risk for serious complications.

But how can socially isolated patients be identified?

The answer won’t be found in the EHR. It is contained within data that reveals such information as how often someone shops online, commutes to work alone, and/or regularly votes in federal, state and local elections. Participation—or lack of participation—in these activities reveals the extent of a person’s social isolation.

Beyond the patient’s clinical information are thousands of variables that medical staff cannot reasonably be expected to collect, yet provide valuable clues to lifestyle and quality of life, such as the value of a person’s home or cost of rent, the number of credit cards and vehicle or boat ownership. Information on environmental factors such as air quality and potential radon and/or lead exposure can help determine if a patient is at risk for developing serious conditions later in life. By analyzing these data alongside clinical information, the overall impact of social/environmental circumstances and health behaviors on a patient’s long-term care needs can be validated.

The challenge is finding ways to capture and correlate that data with non-obvious behaviors that indicate medical risks, such as the above-mentioned determinants of social isolation, so that payers can better-allocate financial and clinical resources and clinicians can make more informed care decisions, resulting in healthier patients and patient populations.

The Cost Of Care—Predicting Risk, Improving Coordination

For the payer community, the key is to compile the data necessary for a predictive vision into a population’s overall health and the resultant risk-of-spend. When payer organizations first take responsibility for an individual, they have no information on that person’s health and the risk they pose. Nor do they understand how best to intervene to avoid a costly illness or condition—or if they will even need to intervene. Unfortunately, it can take months until this information is available through clinical documentation and claims data, significantly delaying the ability to identify at-risk patients.

A better approach is to identify potential health risks at the time of enrollment through more thorough assessments. This enables the payer’s medical team to properly evaluate and match the patient to the appropriate interventions, thereby avoiding the higher costs associated with treating more serious conditions.

This is the approach taken by Medica, a non-profit health services company that provides healthcare coverage in the employer, individual, Medicaid and Medicare markets. When new members enroll in its Medicaid plan, the company has limited clinical history with which to determine the prevalence and severity of conditions such as diabetes, cardiovascular or pulmonary disease. This makes it difficult, if not impossible, to properly stratify risk and prioritize deployment of health management resources.

To close this gap, Medica partnered with a data analytics vendor to gain access to a robust database of social, environmental and behavioral health determinants, which enabled development of a predictive model for ranking new members by level of forecasted risk. For the cohort identified as the top decile of risk for undocumented chronic conditions, the results demonstrated an 89 percent diagnosis rate, more than three times higher than the lowest decile at 24 percent.

Predictive analyses like those undertaken by Medica give payer organizations the ability to connect with at-risk patients on day one and work with them over time to change the behaviors that adversely affect their health—a far more efficient strategy than waiting for the moment when the patient’s chronic condition exacerbates.

More Data Mitigates Risk, Improves Patient Engagement

Whether it’s a provider-sponsored health plan or an accountable care organization (ACO), the more risk a provider organization takes on, the more they stand to benefit from detailed analyses of socioeconomic and environmental data. Approximately 40 percent of a population sees a medical professional less than once a year, severely limiting providers’ and payers’ insight into those patients’ medical histories.

Having a clear view of a population’s medical, social and environmental data allows healthcare organizations to focus resources where they are most needed, such as patients with uncontrolled diabetes or asthma, or at high risk for stroke or cardiac events. This improves outcomes, which in turn influences the population’s overall health. It also drives collection of an even higher volume and wider variety of data, enabling more effective population health management.

However, data alone cannot create the behavioral change necessary for improved outcomes and lower costs. That requires determining the most effective way to engage patients in their care. Luckily, the key to patient engagement can often be found by examining behavioral data to identify which channels of contact—email, landline, mobile phone, text, home visit, or direct mail—will yield the greatest response. Once identified, payers and providers can then group individuals into subcategories based on their preferred method of contact and how likely they are to change their behavior, then customize outreach to maximize engagement.

This was the process undertaken by North Memorial Hospital (NMH) when it sought to improve the rate at which the community used mammography for cancer screening. By first identifying women likely to need mammograms and ranking them to identify those most likely to respond to offers at one of its four clinics.

Leveraging a comprehensive market analytics tool, the hospital ascertained that direct-mail postcards were most likely to generate the highest response rate among its target audiences. Using this method to identify the top prospects, NMH increased its geographic service area by 704 percent which would result in a 117 percent increase over previous attempts to reach the target audience.

Engaging target populations based on preferred methods of contact has been used by retail companies for years. Through the aggregation and analysis of environmental and social data and consumer buying habits, retailers customize the messaging they send to their customers, reducing marketing costs and increasing ROI. The time has come for healthcare to follow suit.

By combining clinical and non-clinical data, healthcare organizations can more effectively connect patients with essential services. For example, when NMH opened a new urgent care center, the hospital mined its EHR patient data to identify users and non-users of its other centers. This was then analyzed against a database of social, environmental and behavioral determinants to create an advanced predictive analytical model to determine who within the patient population was most likely to use the new urgent care center.

This predictive model enabled identification of 71.5 percent of urgent care users by targeting just 30 percent of the total market. Using those targets, NMH demonstrated the potential to reduce marketing costs by $68.57 per acquired patient compared to more general, non-predictive outreach.

The ROI Of Social Interventions

While socioeconomic and behavioral data allows payers and providers to more effectively connect patients to the right care, the greatest outcome improvements derive from social intervention. A 2011 study in BMJ Quality & Safety found that better outcomes in infant mortality and life expectancy are directly affected by greater spending on social services over health services. Yet when compared to 10 other nations with similar healthcare systems, the U.S. is the only one to spend more on health than social services. Indeed, social and behavioral intervention benefits 82 percent of the country. The challenge lies in connecting individuals with the services that will have the greatest effect on health and financial outcomes for both the industry and society.

Analytics At A Crossroads

Payers and providers are at a crossroads. As pay-for-performance becomes the dominate model in healthcare, both groups can play an equal role in improving both financials and patient care—if they are willing to take a lesson from the retail sector and expand their data horizons.

Leveraging solutions that allow doctors to mitigate environmental and behavioral risks can help patients live healthier lives, reduce healthcare costs and revolutionize population health across the nation.