New types of data analytics are on track to deliver consistent, provable value in health care. Here are three keys to unlocking today’s data in order to help health systems dramatically improve patient care and operating efciencies in treating chronic conditions. Health care systems around the world are transitioning to new business models in order to improve patient care and operating efciencies. The pressures on providers to fnd new models are signifcant, particularly for patients with chronic diseases— such as diabetes and cardiovascular disease—which now account for 60 percent of deaths globally. The number of people with these conditions is rising at worrying rates. For example, the number of people with diagnosed diabetes in the United States has climbed from about 17 million in 2007 to roughly 24 million now, and the total direct and indirect annual costs of care are estimated at more than $250 billion. New drugs, devices, and personalized therapies, along with innovations in delivery systems, all ofer new approaches for the treatment of a wide range of chronic conditions. These innovations are underpinned by digital tools and technologies that range from traditional electronic health records (EHRs) to data- based integrated diagnostics platforms to cloud-based patient monitoring systems. Wearable sensing devices, for instance, ofer the opportunity to monitor and improve care for patients outside the formal health care delivery environment.

Health systems are applying more powerful analytic technologies—including artifcial intelligence approaches such as machine-learning algorithms. And researchers are not just unlocking medical data—they are also “re-envisioning” how such data should be collected and applied. Across health care today, data-driven analytics is a deep foundation for measuring and improving outcomes, minimizing variations in care, and demonstrating its value. To reach these benefts, health systems should seek to align stakeholders on value, bring predictive and prescriptive analytics to personalized care, and further engage and educate patients in their own care. Done properly, eforts that integrate evidence-based data and sophisticated predictive analytics can identify patients for targeted interventions and improved health behaviors, and allocate resources more efciently and efectively. The results will include a more efective health system, lower costs, and, most importantly, improved patient health.

Solutions must be built on efective data analytics and change management, which in turn live or die based on their solid alignment among all stakeholders.

Align Stakeholders on Value Health care systems around the world struggle to achieve and demonstrate consistent, value-based, and provable services, and this struggle is only intensifying as medical costs soar. Solutions must be built on efective data analytics and change management, which in turn live or die based on their solid alignment among all stakeholders. Today, those stakeholders include clinicians (and their supporting analytics experts and IT staf), patients, medical technology companies, payers, and other players. All must collaborate in gathering, analyzing, acting on, measuring, standardizing, and optimizing the use of appropriate data. This is the foundation for the improved value- based health care delivery of the future (see “The Strategy That Will Fix Health Care,” Harvard Business Review 2013). One key to the foundation is ensuring that physicians and clinical staf participate fully throughout the processes of establishing standard health care outcome measures, creating and validating data analytics, and establishing and ensuring change management steps. Following this practice, each year “we bring our clinical leadership together and ask: What are our key measures of safety and quality?” says Tina Esposito, vice president for information and Health Care in Downers Grove, Illinois “We go through this rigorous process of identifying those measurements and how we will measure them. Using comparative data, we set very challenging goals. We report out every month on our performance.

Leaders and associates all the way to the front line are held accountable for these measures.” In projects to broaden evidence- based analytic capabilities and actions, it’s crucial to embed the perspectives of physicians and other caregivers in multidisciplinary teams, especially since these clinicians otherwise may not pay attention to the resulting guidelines. “Our analytics team incorporates clinicians, who help to provide a boots-on-the-ground perspective,” says Esposito. “We can create the best algorithm possible, if we don’t insert it in the clinical workfow in a way that’s meaningful to a physician, and they’re not looking at it, it doesn’t really matter.” Additionally, experts point out, doctors and other clinical staf may resist change management eforts unless they have actively participated in creating and validating appropriate outcome measurements for those eforts. Also must be aligned on cost—physicians won’t agree to a change in clinical workfow that reduces costs unless they are convinced that treatment quality won’t sufer. To better align stakeholders, medical technology providers and hospital systems need to collaborate to think through the clinical workfow and ensure systems are in place to enable new products and services. “The rush to digitize often leaves out that step of preparing the clinical workfow to accommodate this new fow of information,” notes Kedar Mate, chief innovation and education officer for the Institute for Healthcare.Improvement (IHI) in Cambridge, Massachusetts. “Unfortunately our workfow systems in health care are so fragile that if you don’t pay attention to that step, it will overwhelm the existing systems, and that will quickly make the technologies irrelevant.” (IHI ofers a quality improvement toolkit that can help to assess and improve clinical workfows, with components such as a Failure Modes and Efects Analysis tool.)

Some groups have developed frameworks of strategies tailored specifcally to implementing digital health interventions, which helps align stakeholders on goals and procedures. Among them is the IDEAS (Integrate, Design, Assess, and Share) framework, described in a 2016 Journal of Medical Internet Research article. The IDEAS architecture covers 10 phases in the development and delivery of pilot digital health projects, starting with empathizing with target users and specifying their target behavior and following through to evaluating efficacy in a clinical trial and sharing the fndings. (Other researchers emphasize the growing trend to evaluate efciency by applying suitable analytics to real-world patient data.)

Startup vendors that are rapidly providing innovative digital health devices and services, however, don’t always appreciate the time, attention, and support required to take all of the steps needed for success.

“Silicon Valley entrepreneurs often don’t want to hear these things, because they’re busy disrupting health care and creating incredible technologies,” comments Brennan Spiegel, director of health services research for Cedars-Sinai Health System in Los Angeles and professor of medicine and public health at the University of California, Los Angeles. “But the fact is that digital health is really hard. Being in the clinical trenches, actually doing this stuff is very difficult.” Perhaps unsurprisingly, one of the biggest difculties is reimbursement. Payers in every country are under enormous pressures to hold down costs. Although the path for valuing

costs. Although the path for valuing the potential gains of efciency and cost-efectiveness with new medical devices and procedures can be far from clear, value-based approaches through bundled payments and risk-sharing to guarantee outcomes often are paving the way.Outpatient monitoring and management services ofer one major way to put value into the system and optimize the performance of medical technologies and pharma. Such services, with proper analytics, will reduce overall costs by cutting unnecessary hospital stays, clinic visits, diagnostic tests, and the use of other resources, says Evan Muse, an assistant professor at the Scripps Translational Science Institute and a cardiologist at the Scripps Clinic in La Jolla, California.Right now, however, getting paid for many of these evolving remote monitoring practices can be a tough challenge, according to Muse, who says, “What will be the appropriate reimbursement for that knowledge and time and efort?”“If we can demonstrate that remote monitoring keeps people out of the hospital, away from the doctor, and away from other resource-consuming sites, then we could prove that the model is a cost efective model,” says Spiegel. “But we have a long way to go before that vision is fully realized.” In addition to clinical trial evidence, predictive analytics applied to real-world clinical data can help key stakeholders identify value opportunities, says Sanjeev Mehta, chief medical information ofcer at the Joslin Diabetes Center in Boston, Massachusetts.

For instance, a 2016 Joslin study built a predictive model to estimate the cost of managing a population of 10,000 patients with diabetes. Such frameworks could help better organize care delivery in a value-oriented manner focused less on maximizing volume and more on increasing prevention and other high-value interventions.

Once clinically efective advances in drugs and devices are approved by the Food and Drug Administration, Most people with chronic illness don’t fully observe their recommended behaviors, so doctors seek to increase compliance by tailoring their recommendations to refect the social, environmental, and behavioral factors impacting each patient.

models to address questions such as what support to ofer for new health technologies, such as emerging artifcial pancreas devices for people with type 1 diabetes. These models must incorporate evidence as to whether these technologies will improve patient health and avoid greater costs down the road by minimizing events linked to disease progression such as hospitalization and kidney dialysis treatment.

“We cannot tweak out anticipated value for everything in medicine, but if we want to identify high-value interventions, we have to be much more serious about including the cost of care,” says Mehta. “A medical device may have a signifcant upfront cost, but if its use reduces episodes of life-threatening hypoglycemia or diabetic ketoacidosis, the value to our patients and health care system could be very favorable. One important challenge will be for stakeholders to agree on the time horizon used when defning the value of medical interventions.

Bring Predictive and Prescriptive Analytics Into Personalized CareIn addressing the enormous human and economic costs of chronic diseases, the best hopes for better care lie in more personalized prevention approaches and care delivered to each patient. The best treatments can vary tremendously between individuals with similar-appearing illnesses. Moreover, most people with chronic illness don’t fully observe their recommended behaviors, so doctors seek to increase compliance by tailoring their recommendations to refect the social, environmental, and behavioral factors impacting each patient.

Truly personalized treatment begins with the creation of a complete longitudinal patient record, and then requires health analytics that moves beyond the descriptive capabilities of the past to predictive analytics and then takes the next step into prescriptive capabilities when possible, as noted in a 2016 article on big data analytics in Nature Reviews Cardiology.Cardiovascular conditions, the leading killers worldwide, have long refected this personalization trend, and personalized care can build on rich existing data platforms for the conditions.

The platforms can draw data from advanced clinical instrumentation applied in acute instances, wearable cardiac monitors that help assess patients with less severe symptoms, and telemonitoring systems for pacemakers, internal cardioverter defbrillators, and implantable rhythm monitoring devices.

Moving beyond these well-established health-saving capabilities, “technology developers are busy giving us more sophisticated tools that can help us understand longitudinally, for example, the ups and downs of blood pressure or weight over time,” Mate says. “These tools allow not just the doctor but also the patient to understand the micro experiments they’re making in their lives when they tinker with their diet and their exercise regimens.”

“Many people have a heart attack that no one saw coming,” says Spiegel. “Part of the reason for that is because we deliver care in these tiny punctuated moments within a clinic, where we see people for 15 minutes and make decisions and of they go. People spend 99.9% of their lives far away from the clinic, and that’s where things actually happen.”Spiegel is leading a clinical study aimed at fnding better ways to predict cardiac events, which will measure physical and biological phenomena as well as social and emotional phenomena. More specifcally, participants will use an FDA-approved home biomarker kit to collect a drop of blood and send it in for analysis. The study also will gather patient-reported outcomes, including everything from physical function to emotional status. These data sets will be combined with ftness monitoring (heartbeat variability and a variety of other parameters) and cardiac rhythm monitoring, and all the data will be analyzed by machine-learning algorithms. “The hope is that we can fnd a signal in all that noise that can predict who will have a heart attack,” he says. That’s one example of work that applies the often barely understood power of machine learning to traditional EHR data. Machine learning, which has dramatically altered the structure of online retailing and other industries, is just beginning to make major inroads into medicine, experts note.Back in 2009, the U.S. Agency for Healthcare Research and Quality estimated that almost half of the potentially preventable hospitalizations in the country were due to heart disease and diabetes, with an annual cost approaching $14 billion. In the years since, health systems have made solid progress in lowering preventable hospitalization and emergency room visits for many chronic conditions. But the costs remain huge, and rates of preventable interventions have climbed for some populations—for instance, by a ffth in recent years for people with short-term diabetes complications.


When patients participate more actively in managing their chronic conditions, they enjoy better outcomes. 

these insights will support more proactive patient engagement. Here, the system might go far beyond traditional reporting and analytics to examine thousands of variables, ranging from complex medical and socioeconomic interactions to local weather and trafc patterns—“basically any data we can include will be analyzed without bias to its perceived importance,” says Mehta. He acknowledges the need for data not collected in traditional EHR systems, including patient-reported outcomes, which could further enhance these predictive algorithms.Capitalizing on these opportunities will require health systems to draw on expertise, external or internal as needed, on machine learning, and on other rapidly advancing artifcial intelligence technologies. The payback will be far more powerful individualized understanding of conditions and potential treatments.


Further Engage and Educate Patients in Their Own CareWhen patients participate more actively in managing their chronic conditions, they enjoy better outcomes. A study reported in the Journal of General Internal Medicine in 2012, for instance, found that patients who were more engaged in their own care were signifcantly more likely to have good clinical indicators and to receive preventive care, and less likely to smoke, be obese, or be hospitalized.

Health systems constantly seek to improve outcomes by engaging patients more directly in their own care, and clinicians can draw on fndings from social sciences to help fndings from social sciences to help encourage healthy behaviors. (Selected current approaches are described in a 2016 Athena Health white paper overview, a 2016 Annals of Family Medicine article on patient behavior changes, and a 2016 QJM article on exploiting behavioral economics.)Unfortunately, most people with chronic conditions still don’t fully follow what their doctors advise for diet, exercise, and other behaviors.

Many studies have shown the difculty in truly changing behaviors for extended periods of time, although the need to do so is only intensifying as these diseases rapidly increase worldwide.In cardiovascular disease prevention, “diet and exercise trump almost every medication and test that we can put out there,” comments Scripps’s Muse. “It’s my job as a physician-teacher to put things in patients’ hands.

I say, I’m the copilot here, I’ll come up with a fight plan and help us understand what is important for us to look at, and then put the power in your hands.” “Moving the patient out of the role of simply recipient and into the role of a coproducer of health outcome is actually very important,” says IHI’s Mate. “That role change is pretty fundamental and one that some patients and clinicians are not prepared for.

”The gap among unprepared patients becomes ever more visible with the rapid spread of self-monitoring devices aimed to help with chronic conditions. For example, a recent study by Berg Insight projects that the global total of connected home health monitoring devices will climb from about 7 million in 2016 to more than 50 million in 2021.

Seeking to close the gap, researchers are testing new digital health technologies designed not only to encourage but to also actively measure patient engagement. In one pilot clinical trial reported in 2017, for instance, patients with uncontrolled hypertension and type 2 diabetes volunteered to take “digital medicine” pills that combined medication with the ability to alert a wearable sensor.


Wearables are gaining adoption. Approximately 25% of Americans own a wearable, +12% Y/ Y, 2016 . 

This sensor also measured physical activities and reported all data to a mobile phone app. The app reminded patients to take their medication and allowed them to visualize their data. Their health care providers could also examine the data and ofer guidance. The study found that participants signifcantly lowered their blood pressure, cholesterol, and blood-glucose levels.Other digital health clinical trials that have studied more general-purpose wearable ftness devices, which have seen rapid adoption in the U.S. and some other countries, have amply demonstrated the need to strengthen patient engagement. FIGURE 1 A number of such trials designed to leverage the easy-to-use ftness devices, even those among tech- and health-savvy populations that seem most ready to engage, underline this concern, as noted in a 2015 commentary in Academic Emergency Medicine.

More recently, in a study of the MyHeartCounts mobile health application among almost 49,000 consenting early-adopter participants, reported in 2016, fewer than three percent of participants ended up providing all the information needed to assess their 10-year heart risks. And in another example published in 2016, Cedars-Sinai invited almost 80,000 registrants on its patient portal to report their personal ftness data, but fewer than one percent shared their data. What was missing from this ofering? “There needs to be context, and it’s diferent if a doctor prescribes it and explains exactly how he or she will use the data,” Spiegel says. “Until patients understand why and exactly how their data is being used, and the beneft to them, they won’t necessarily share data just because they can.”

Overall, health systems should seek to align with stakeholders on value, bring predictive and prescriptive analytics to personalized care, and further engage and educate patients in their own care.

Moreover, it’s crucial to avoid over-promoting the benefts.

Although the Cedars-Sinai patients could discuss their uploaded ftness data during visits with their doctors, “we certainly couldn’t promise that they would have a more meaningful discussion, because there’s no evidence that that’s true,” Spiegel says.

Maximizing actual patient benefts will require individualized assessments, education, and delivery, often with a sharp focus on those who either are most willing to change their behaviors or need desperately to do so. “We need to meet people where they are, particularly along their readiness for change,” Spiegel says.

“Some are ready to eat better, or adhere to their medicine, or follow their doctor’s advice, and some are not, and that’s an important distinction.” “The solution is not just personalized analytics,” agrees Muse. “The successes will start coming when we can do a better assessment of how each patient can change.”

One crucial ingredient for change is patient education, which can be personalized not just with advanced interactive learning materials but also with rapidly evolving diagnostic tools, such as handheld ultrasound devices. When his patients with high blood pressure insist they still feel fne, for example, Muse can show them an ultrasound image of the early signs of damage now being inficted on a given region of their heart, which fully gets their attention and helps motivate them to change some of their behaviors.

From Analytics to ActionOverall, health systems should seek to align with stakeholders on value, bring predictive and prescriptive analytics to personalized care, and further engage and educate patients in their own care.Health care data is becoming broad and deep enough that predictive analytics can meaningfully inform fundamental questions such as who will best adhere and respond to a given therapy in more real-world settings, says Mehta. That promise needs to be confrmed in clinical studies, he cautions. “But it’s a really exciting way for us to start thinking about medicine—proactively intervening to support our patients in a way we haven’t done before,” he says.Clinicians must actively integrate such insights into better treatment approaches, whose efcacy and costs can be carefully measured and optimized over time, experts agree. “Digital health is not a computer science or an engineering science—it’s a social and behavioral science,” says Spiegel. We have to create hyper-personalized, contextually appropriate experiences for each individual patient. That’s what human doctors do every day of the week.” do every day of the week.”

Related Posts

Leave A Reply