• Article
  • October 19, 2015

What Big Data Means for Pediatrics

Big data can drive significant improvements in pediatric health care – now and for years to come.

By Megan McDonnell Busenbark

"It's one of those life-changing moments in IT, which you rarely get," says Tod Davis, manager of business intelligence at Children's Healthcare of Atlanta, as he recalls his first real glimpse into the impact data can have on the care of patients. Over a year ago, clinicians at Children's Healthcare of Atlanta sought to learn how NICU patients were tolerating testing for retinopathy of prematurity, an eye condition that, if left untreated, can lead to vision loss and even blindness for premature babies. Typically, charts indicated the infants tolerated this testing well, but it was all very subjective.

So working with six months' of NICU vital signs data, Davis and his team created an application that enabled the clinicians to review the vitals of babies before, during and after this brief exam, which involves putting a clamp in the infant's eye.

The findings shocked the clinicians: When the exam began, the heart rate, respiratory rate, pulse and oxygen saturation for these tiny patients went off the charts—and remained off baseline for up to 30 minutes after the exam. As a result, the NICU team provided additional comfort care to these babies to ease the physiological stress of the exam and send them back to baseline faster, proving the power of vital signs data, according to Davis. "Still, most hospitals that store it, don't store it for very long because they're unsure of the value of that data," he says. "But I always say, 'If you say you don't have big data, you're throwing it away.'"

What constitutes big data?

Generally speaking, big data is defined as data sets—typically consisting of billions or trillions of records—that are too large, complex and dynamic for any conventional data tools to capture, store, manage and analyze. In the world of pediatric health care, though, the definition of big data varies, mostly depending on the size and mission of any given institution.

"Typically, a lot of hospitals are looking internally because they have data that is deeper than it is wide," says Matt Hall, director and principal biostatistician at Children's Hospital Association. "Meaning, they have a lot of data points on a smaller population of children. But when you have a lot of data on a single child, and you put that data in for 100 children, you're talking about a lot of information."

Christopher Forrest, M.D., Ph.D., senior vice president and chief transformation officer at The Children's Hospital of Philadelphia (CHOP), and colleagues set out to define big data in a July 2014 article in Health Affairs: In its most advanced form, big data in pediatrics is a massive combination of electronic health records (EHRs), patient-generated data sources and biospecimens aggregated from multiple institutions and thousands, even millions, of children.

As the field of pediatrics sharpens its focus on evidence-based medicine, aggregating individual data sets into big-data algorithms can give clinicians much more robust evidence on which to base care decisions—much faster.

For Kevin Maher, M.D., director, Pediatric Cardiac Intensive Care, and co-director for the Center of Pediatric Innovation at Children's Healthcare of Atlanta, big data in pediatrics is simply a natural evolution of the EHRs and a significant opportunity to extract value from the data that institutions are already collecting to provide better care and outcomes. The analysis of that data can help care teams identify which patients are at risk for adverse events, he says, and catch the quiet clues that may signal trouble.

"In cardiac intensive care, we see children who have a cardiac arrest, and I can't help but think there was information four, six or eight hours ago showing that they were gradually heading in the wrong direction," Maher says. "As care providers, all of us recognize when patients are in trouble, but there are subtle signs that can lead you to intervene to prevent a decline from occurring in the first place. That's the potential of big data and analytics—you can combine massive amounts of information in milliseconds to provide enhanced patient care."

A (big) data pioneer

When it comes to using data and analytics to transform patient care, Intermountain Healthcare in Utah was among the first at bat. This system of 185 clinics and 22 hospitals, which includes Primary Children's Hospital in Salt Lake City, began its data journey back in the 1950s before Intermountain even got its name. That's when Homer R. Warner, M.D., Ph.D., considered one of the fathers of medical informatics, began bringing computer applications into clinical decision-making. This ultimately led to the nation's first EHR, which Intermountain used until it switched on its new system in February 2015.

In the 1990s, this large, primarily adult health care system created one of the first enterprise data warehouses in the country, making Intermountain a first mover in big data. All clinical programs at Intermountain today—pediatrics included—operate with data at their core. A data manager, data analyst and an operations director working with a medical director, direct all the clinical care activities within each clinical program for the entire system.

"That's how we ended up with our standardization of care—by looking at the best evidence and translating it into this care process model, which we deploy across the system," says Raj Srivastava, M.D., M.P.H., assistant vice president of research, Intermountain Healthcare and hospitalist at Primary Children's Hospital. "We collect the data, feed it back to the providers to influence their behaviors when caring for their patients, and then we track the outcomes."

Today, data-driven decision making at Intermountain has enhanced patient outcomes in cardiovascular medicine, endocrinology, surgery, obstetrics and care processes and has saved millions of dollars in procurement and supply chain processes, according to MIT Sloan Management Review. And Intermountain is extending everything it has done on the adult side over to Primary Children's—leading to thought and behavior change right at the pediatric bedside.

"We're talking about making a difference when you're deciding what to do about the patient right in front of you," says Chris Maloney, M.D., Ph.D., associate chief medical officer at Primary Children's Hospital. "When do you stick a needle into the patient and draw labs? What are the best labs you should get, and when you get those labs, what do you do with the results? What kind of antibiotics do you order, and how long does the patient need to stay in the hospital?"

Looking across institutions

With the exception of a few conditions, pediatric disorders are rare diseases. "For example, one hospital may not treat enough children with rare and complex conditions to generate adequate sample sizes," Forrest says. "But aggregating data from millions of children across many pediatric institutions can accelerate knowledge discovery."

Forrest is the principal investigator for the PEDSnet learning health system, a national network that will support the efficient conduct of clinical trials, observational research and quality improvement across diseases, specialties and institutions. PEDSnet combines big data with family engagement and involves parents as partners in research design to address concerns important to families. PEDSnet is currently composed of eight pediatric academic health centers.

It has data for 4.5 million children dating back to 2009. And it has cracked the code for sharing data across centers—one of the biggest hurdles to big data implementation—by standardizing the definitions and descriptions of clinical observations for pediatric care.

So for example, all medications, lab data and diagnoses are coded in the same way, allowing for rapid analysis. "We like to use the phrase, "data in once, used many times,'" Forrest says. "These data are entered once by clinicians as part of routine clinical care, but because we transform them into common standards, we create this resource that can easily be used for research, quality improvements, population health or benchmarking."

PEDSnet data also can be pulled to identify patients for clinical trials often in a matter of minutes from the network's data coordinating center at CHOP, adds Forrest, which can whittle the typical recruitment process down from months to days—ultimately accelerating clinical trials and research.

Humanizing big data

M. Narendra Kini, M.D., M.H.A., CEO of Miami Children's Health System, is laser focused on building a digital strategy that gives patients and families at Nicklaus Children's Hospital an environment similar to what they experience in everyday life, where they can plan a party or shop for clothes on any mobile device. Here, families can download apps on their smart phones that enable them to access urgent care wait times, order room service to an inpatient room and notify them about care milestones for their children, such as blood tests and results.

Another app transforms the discharge process, making it more of a care handoff from the hospital to the parent than an end process. This app provides a range of data—from reminders on when to take medications to the location of the nearest pharmacy—in a simple format. To date, about 80 percent of patient families have consented to use this app over a paper-based discharge. For Kini, it's yet one more way the system leans on data to create a "human experience" for patients and families.

Today, Nicklaus Children's Hospital is in the advanced stages of building a sophisticated data warehouse that will hold secure and encrypted data from medical records and devices. This essentially takes a holistic picture of every child, not just of his or her disease, and making it available for analysis. "With children, there's a lot more screening, prevention, developmental analysis, counseling and monitoring," Kini says. "That's the difference between pediatrics and the classic adult approach."

Still, children's hospitals can learn a lot about big data from adult systems. "When it comes to understanding data warehouse design, data normalization techniques, standardization of interfaces, extraction techniques and basic analytics, there is really no difference between adult and pediatric systems," he says.

The three big pieces of big data, Kini says, are understanding cohort trends, analyzing those trends and then using the trends to predict variations, clinical severity and disease end points. And the key is in the presentation of data. "Even if organizations have advanced big data capabilities, if they don't know how to present to the data to the clinician at the point of decision, then big data loses some of its potency," Kini says. "You must impact the clinicians as they are about to make a decision. It cannot be in retrospect and it cannot be at the point of care—when it's too late."

An emerging role

So who will present big data in a useful, usable form? "There's this magic job that we call the data scientist," says Denise Zabawski, CIO at Nationwide Children's Hospital in Columbus, Ohio. "There are not very many of them in health care right now, but it's a key role to bridging the gap between the clinician and the IT person. The clinician wants the outcomes, IT is collecting the data, storing it and making it available, but someone in between has to do the science. The data scientist digs into the data, finds the correlations and builds the logic."

Nationwide Children's Hospital, a member of PEDSnet that is currently building its own big data architecture, expects to hire a data scientist within the year. Zabawski adds that The Ohio State University is creating a new degree track for the data scientist, which, according to Harvard Business Review, is "the sexiest job of the 21st Century."

With innovation comes challenges

For all the benefits, clinical and technical teams alike acknowledge that big data come with challenges—from standardization to technology infrastructure and staffing. There are culture hurdles too, as this is a monumental shift for pediatric health care and the issues around sharing and ownership of data across centers requires trust and governance. And that's not all. "One challenge in pediatrics is that patients with medical complexity have very small volumes," says Hall.

"In children versus adults, there is more heterogeneity in the populations, so there are pediatric patients with multiple chronic diseases and only a few hundred of them scattered across the country. But that's the hope of where big data will lead us. If you can integrate data systems from across centers to increase the number of children in your population and look at a lot of data from those children, you might be able to predict and ward off catastrophic events."

Despite the challenges, according to Maher, big data deliver significant benefits, both administratively, including a more streamlined billing process that can lead to more timely reimbursements, and clinically, from higher quality of care and improved patient safety and outcomes to population health management and rapid research.

Forrest agrees there's a bright future in big data. "It will hold the promise of a new science of discovery," he says. It could also lead to big cost savings. According to Bernard Marr, an author, consultant and expert on big data, by better integrating big data analytics into health care, the industry as a whole could save $300 billion a year—that's the equivalent of reducing the health care costs of every man, woman and child by $1,000 a year.

Megan McDonnell Busenbark is a writer and founder and principal of Encore Communications, LLC, in New Fairfield, Conn. Send questions or comments to magazine@childrenshospitals.org.