• Hypertrophic Cardiomyopathy (HCM)
  • Vaccines: 2023 Year in Review
  • Eyecare
  • Urothelial Carcinoma
  • Women's Health
  • Hemophilia
  • Heart Failure
  • Vaccines
  • Neonatal Care
  • NSCLC
  • Type II Inflammation
  • Substance Use Disorder
  • Gene Therapy
  • Lung Cancer
  • Spinal Muscular Atrophy
  • HIV
  • Post-Acute Care
  • Liver Disease
  • Biologics
  • Asthma
  • Atrial Fibrillation
  • Type I Diabetes
  • RSV
  • COVID-19
  • Cardiovascular Diseases
  • Prescription Digital Therapeutics
  • Reproductive Health
  • The Improving Patient Access Podcast
  • Blood Cancer
  • Ulcerative Colitis
  • Respiratory Conditions
  • Multiple Sclerosis
  • Digital Health
  • Population Health
  • Sleep Disorders
  • Biosimilars
  • Plaque Psoriasis
  • Leukemia and Lymphoma
  • Oncology
  • Pediatrics
  • Urology
  • Obstetrics-Gynecology & Women's Health
  • Opioids
  • Solid Tumors
  • Autoimmune Diseases
  • Dermatology
  • Diabetes
  • Mental Health

Analytics puts power in the hands of patients, payers, physicians

Article

There is so much data available at all levels of healthcare, and technology and public support is on the cusp of a breakthrough in using this information for widespread improvements.

Have you ever heard of a zettabyte or a yottabyte? A zettabyte is the equivalent of one sextillion bytes; a yottabyte, one septillion bytes; and both, a few trillion gigabytes.

They also represent the amount of complex health data that can likely be found in the U.S. healthcare system today, according to a report published in the journal Health Information Science and Systems.

This data could be used to potentially save hundreds of billions of dollars in clinical operations, and research and development, but the challenge is taking that data and extracting useful, predictive information from it.

Optimal use of data is still a ways off, but analysts are hopeful that innovations over the next several years could lead to improved outcomes, more effective population-based healthcare, improved operations, and lower healthcare costs.

Increasing the use of data analytics is a top priority among most healthcare organizations, even though a KPMG survey last year of more than 270 healthcare professionals found that only 10% were using analytics with predictive capabilities. But, a 2015 CDW survey of 150 healthcare executives, found that 65% said their healthcare organizations planned to increase spending on developing better analytics, with an average expected expenditure of $1.9 million for each organization.

Next: Too much data, too little time

 

Too much data, too little time

Electronic health records (EHRs) have taken the nuts and bolts of every healthcare encounter-from a phone call with a physician to complex surgical notes-and put them at the fingertips of payers and providers. Yet, contrary to the original vision for EHRs, that data is not always accessible across health systems. And, in layers upon layers of information, it can be difficult for payers or providers to get a clear picture of one patient’s history, let alone gain population-based information that can lead to greater predictive medicine and more efficient care.

Peele“Everybody’s talking about data, and the proliferation of electronic [health] records has produced a mountain of new data sources,” says Pamela Peele, PhD, chief analytics officer, UPMC Health Plan. That mountain of sources, she says, however, has led to challenges. “If you’re just raining dots of information down on people, you can paralyze them,” says Peele. “We can use data in many ways, but before you just start pouring data on people, [you need to] understand what the data needs to illuminate and bring that data together.”

Ideally, she says, data analytics should serve the clinical management of the patient and organize it from a clinical perspective. “This is really the foundation of developing personalized care,” Peele says. “Our protocols are built to an average patient, but what works best for whom? That’s really the question that needs to be answered.”

 

Next: In an ideal world

 

 

In an ideal world

Peele says analytics could be used to mesh data together with clinical results to try and understand what elements and individual characteristics contribute to outcomes. That requires huge amounts of data in an unstructured study, she says. In the past, medical research has approached its questions from a hypothesis standpoint, but in order to have a hypothesis you have to know what the problem is. In the future, Peele says, research will involve coming to the data and letting it speak for itself.

“We’re going to walk away from what we think we know,” Peele says. “It doesn’t begin with an individual. You’re going to come to hundreds of thousands of medical records and you have to have a target in mind. You need to know what you’re looking for, but you don’t need to know the association.”

For example, in examining positive clinical outcomes, plans and providers will need to look at what patients have the best outcomes, and what elements contribute to those outcomes. Then, take into consideration the protocols, sequences, and timing of services.

Finding the answers to these questions will lead to better population-based medicine, with patients divided into subpopulations based on their data and anticipated outcomes. Patients could then work with their physicians to select the care paths that best fit their medical history, predicted progression of disease, and even socioeconomic status. For instance, if two courses of treatment are shown through analytics to have equal outcomes, patients could choose a less costly option and still be confident they are receiving the care they need.

Next: Harnessing the power

 

 

Harnessing the power

The first step to achieving better data organization and analytics is to develop the computation power, Peele says. The ability to extract data varies by EHR system, and she says it’s clear from recent policy updates that the Centers for Medicare and Medicaid Services (CMS) is carefully watching what vendors are doing to promote or prevent extraction of data out of their systems. “That is a huge challenge we have. We have many different vendors that don’t talk to each other and data is not imported back and forth,” Peele says.

True interoperability won’t come, Peel believes, until it’s legislated. “It would have to be the government that will come down behind this,” she says. “The government got us started on meaningful use, and they have accelerated the process. But they are not getting value because records are isolated from each other.”

Even if interoperability progresses, plans and providers face other data analytics struggles. “You have to have sophisticated individuals to work on it, time, and resources,” Peele says, adding computer can’t do everything and it takes a human touch to pull it all together. “We’re never going to get rid of the huge mountains of information. The piece that’s going to bring clarity and transform healthcare is the storytelling narrative behind it. The [EHR], at this point, can’t build a good narrative.”

The Senate Committee on Health, Education, Labor & Pensions (HELP) recently released draft legislation related to EHRs, including efforts to improve interoperability and eliminate information blocking.

Next: The payer's role

 

 

The payer’s role

Sherri Zink, vice president of medical informatics, BlueCross BlueShield of Tennessee, says data analytics is growing increasingly complex for payers, and not just because of interoperability issues. “It’s much bigger than just the claims coming through doors. A lot of customers have wellness programs in place with incentives,” Zink says. “There’s a lot of expectations from clients that they can take in all that data and track incentives.”

Payers, she says, need to utilize numerous teams to help sort through all of the information. At BlueCross, client reporting groups help employers understand their plans and opportunities to mitigate costs; another team focuses on internal analytics and works with operations; predictive modeling teams look at how to drive better consumer engagement; and other teams looks at accreditation analytics.

Payers also need to use analytics to help improve care coordination, Zink says. “You can have an endocrinologist that’s managing your diabetic care, but your primary care physician really needs to know and understand that data,” Zink says. “Once we get data in, we can showcase and share with participating providers.”

Payers can also use data analytics to help identify patients who are noncompliant in their medications, identify the reasons for the noncompliance, and help the patient find programs to help with costs, manage side effects, or find alternative treatments, she says. “That’s where you get the full benefit of the payer, provider collaboration.”

Next: The provider's role

 

 

The provider’s role

As the market is moving more toward pay-for-performance, providers are looking to payers to help identify performance metrics like outcomes and continuity. “[EHRs] only show them what’s going on inside their office. So we provide tools and help them understand opportunities to work with patients,” Zink says. We might even define when some tests should be done and send an alert to the physician.”

ShroffAnand Shroff, cofounder and chief technology officer of the data analysis firm Health Fidelity, says data analytics can also improve risk management, which is key for providers in risk-sharing reimbursement models such as accountable care organizations (ACOs). ACO providers need to identify chronically ill cohorts; identify their risks; predict outcomes; and model risks for readmission, behavioral health problems, and more. Then, they can reach out to patients at risk with early interventions.

Mike Myers, executive vice president of solution delivery at AxisPoint Health, says providers participating in risk-based models also need to predict who their high-cost members will be in advance, because that’s where providers have the most ability to impact care quality and costs. Still, only 30% of high-cost members stay the same year-to-year, he says, and today’s EHRs and analytics platforms aren’t good at predicting who the other 70% will be the following year. “They acquire the information, they provide the ability to look at data, but they weren’t designed to anticipate and predict whose likely going to be your high-cost members,” Myers says. “I think the next phase is going to be about accurately predicting [which patients] in this population that … they need to focus on.”

MyersMyers adds that data analytics platforms can help providers meet quality indicators set by CMS and demonstrate progress on a continually expanded set of metrics each year. “If you look at the shift of Medicaid to managed models, it’s going to force health systems to think very differently about the populations and the communities they service and the risk they’ve inherited,” he says.

Having good patient histories helps develop predictive models, but Myers says he thinks more accessible, genomic profiles-costing maybe less than $600 per person-could offer the most benefit in predicting the probability of patients developing costly medical conditions.

In five years, he predicts, patients will be able to have a complete genomic profile done for less than $250, and that information will be fed into predictive models that-when combined with family and personal health histories, lifestyle, and socioeconomic factors-can generate predictive models rather than the historic models used today.

“We’ll be able to screen for 20 or so cancers, five to seven cardiovascular diseases, and a variety of neurological problems,” Myers says. “We’ll have an ability to look at people and make more effective predictions.”

Next: Improving patient outcomes

 

 

Improving patient outcomes

Michael Dulin, MD, PhD, chief clinical officer for analytics and outcomes research for Carolinas HealthCare System (CHS) and director of research at the Carolinas Medical Center Department of Family Medicine, says his organization uses four predictive models. One predicts readmissions and is used to identify patients that need additional resources to prevent readmissions. Others predict length of stay, emergency department wait times, and the risk of initial hospitalization.

DulinThrough segmentation, Dulin says CHS uses statistical analysis on the 2.5 million patients in its system, looking at clinical and claims data, consumer data, geographical information and more, and uses those to assign multiple risk scores and identify overall patient risks. “It’s been really nice for us because it’s based on our data,” Dulin says. “It flips the whole model of care delivery upside down.”

The system analyzes the patient’s risk of moving from one segment (low risk of disease) to another (high risk of disease) and identifies health changes patterns. It also reveals information about which patients are the most impactable.

CMS determined that 10,000 patients can be managed in 2016 through care management.

“For somebody that’s going to the emergency department [ED] for primary care needs, we can look at when primary care services are delivered in ED, then at those people who are looking for another way to get care. If we reach out to them proactively, they’re more likely to get impacted. They’re looking for care, they just don’t know the right path to get there,” Dulin says. By connecting the dots and enlisting the help of a case manager, everybody wins. “Patient gets better services, more often higher quality service for prevention, and the ED can focus on true emergencies. It’s one of those really nice win-wins where the patient gets better care, the cost is less, the physicians are happier, and we all get to focus on what we do best.”

Dulin says CHS has seen readmission rates decline across the system, but a number of other interventions were implemented around the same time as those resulting from data analytics, and a review of which interventions were of most benefit is not yet complete.

Next: Future interventions

 

 

Future interventions

CHS is also working on a tool that randomizes interventions based on risk to evaluate the efficacy of those interventions, and results are expected by mid-year.

CHS is also working on multiple models to predict length of hospital stay, driven in part by the medical needs of the patient and part by hospital operations. Dulin is hopeful the models will identify ways to expedite various services and operations on a broad level throughout the hospital system.

In the future, Dulin says he would like to see better tools that assist with shared decision making between providers and patients. “Patients can have access to their data and really understand their own risk and their own journey through the healthcare system,” Dulin says.

Another way patients could benefit from data analytics is to see more information about records and physicians’ notes, and have access to models that predict how individual behavior or lifestyle changes can affect outcomes. Dulin envisions tools similar to apps on smartphones were patients could input various changes such as weight loss or smoking cessation and see a prediction of a personalized improvement in outcomes based on their medical and family history, metrics, genetics, and socioeconomic status. “We could use data to allow patients to see real-time predictions of changes they make and how it will affect their outcomes,” he says.

Myers agrees, noting that, within the next three years, he believes analytics will enable healthcare teams to develop population health models that can inform patients based on their own data and create specific plans for care.

“If you help people understand why they should be afraid of something, compliance goes up,” he says. “Within five years or so, consumers will interact with these advanced systems, and these advanced systems will tell them the likely impact of their healthcare decisions in the next 10, 20, or 30 years.”

Rachael Zimlich is a writer in Columbia Station, Ohio.

 

 

 

 

 

 

 

 

 

 

 

 

 

Related Videos
Related Content
© 2024 MJH Life Sciences

All rights reserved.