Pixabay Medication Mysteries

Why healthcare needs to share longitudinal health data now

Sam is a fifty-one-year old gentleman. He has had no previous heart disease. Sam is seeing his doctor today because recent lab work suggests that his lipids are elevated and warrant treatment. This is not new. Sam had similar findings before. Today he is likely to be started on a medication in order to avoid a heart attack or stroke. How sure are we that this scenario will produce the desired health outcomes?

In short, not very sure at all!

One thing that Sam is exceedingly sure of is that he is not alone. He has joined the club of many of his colleagues that are already taking medications to help manage chronic diseases. By 2030, chronic disease will affect about half the adult population in the US [1]. The Centers for Disease Control and Prevention (CDC) consider the majority of this chronic disease burden to be preventable: 80% of Heart disease and Stroke, 80% of Diabetes Type II and 40% of Cancers should be preventable if eliminating smoking, poor diets and low activity levels were possible. This assertion along with others support the general acceptance that 80% of the contributors to health outcomes occur outside of the medical care setting.

This sounds very promising; the majority of poor health outcomes are in fact preventable.

But how does this information translate to Sam’s experience today? He will have no more than fifteen minutes with his physician. During that time, his doctor will have a few short minutes to cover the plan: “continue with lifestyle measures and fill this prescription”. Sam’s doctor will be forced to spend more time paying attention to her screen than to Sam in order to properly document this episode of care. It is important to ensure billing transactions are successful. Furthermore, with a caseload of more than two thousand patients, she can hardly be expected to know much about Sam. To her, Sam exists in the form of periodic visits over a number of years: gastroenteritis secondary to food poisoning, back pain after a construction accident, intermittent chronic dermatitis of unknown etiology. Every day individuals like Sam meet the criterion for prescribing a certain medication. These criteria are confirmed by their physician, a prescription is written and the episode is documented in the electronic medical record (EMR). Revenue-generating transactions are executed. But how much health has been produced?

What are the chances that this medication will be effective? Will Sam prevent death, a heart attack or stroke by taking his prescription? This is not entirely clear.

Healthcare uses a term: Numbers needed to treat (NNT) to help quantify the effectiveness of a given intervention. Given that Sam has had no prior heart health events, his physician would have to treat sixty-one similar patients with medication over five years to prevent ONE heart attack. She would have to treat two hundred and sixty-eight patients over the same period to prevent ONE stroke. During this time, no death would be prevented by the use of this medication. Meanwhile, one in fifty individuals would develop Diabetes and one in ten would develop muscle damage by using this medication [3]. Will Sam be the ONE who benefits in this scenario? Is there any way of knowing more? Clearly Sam deserves to know more. Healthcare could and should do better. Given the technological advances across industries today, we have a responsibility to do much better.

Longitudinal personal health records could go a long way to improving Sam’s chances. Recall that Sam is in control of the 80% of health contributors that happen outside of medical care. These contributors include the CDC cautions regarding smoking, poor nutrition and inactivity mentioned above.

Imagine if Sam did not exist as an event series (gastro, back pain, dermatitis) and instead had been incented to create a longitudinal record of high yield health contributors throughout the decades of his life. High yield health markers are those that provide the greatest lift to predicting future health outcomes. At a minimum they would include lifestyle contributors such as nutrition, activity, sleep as well as social and many other yet to be determined contributors.

At fifty-one, Sam would have accumulated a very informative data trail that when compared with others like him may have predicted today’s visit with his clinician. However, multiple improved scenarios may have been more likely:

#1 Sam’s longitudinal health record includes details such as his diet, step count and cycling history when his lab values were in good control. A simple comparison with his daily habits of today shows a steep curve upwards in the number of lattes consumed often with an additional baked good. His step count is similar but his recent move has left him with few cycling options. These are personally relevant benchmarks that will help provide a reality check for Sam. Given that age carries with it age-related decline, he would likely need to “up his game” in order to produce similar lab results. Historical achievement can provide a solid basis from which to deliberately design future achievement. Sam’s longitudinal health record is a valuable personal resource providing high definition data that his memory alone is ill equipped to produce.

#2 Sam’s longitudinal health record is set up to share basic information with the data commons that organizes the health experience of many others clinically similar to Sam. Thus rather than expending many months on strategies that might not work, Sam can check in with the community data commons to see what seems to work best. His Asian roots drive a significant amount of rice into his diet. Switching from white rice to brown rice improved biomarkers significantly according to others. Given that this is an easy fix, brown rice tops the list of Sam’s strategies. In a similar way, Sam learns that some forms of caloric restriction did not result in health improvements in his group. Great, this is one strategy that Sam hopes to minimize.

#3 Sam’s longitudinal health record makes it easy to enlist the help of various products and services that could help him improve his health. The data commons makes it easy to see which of the hundreds of options have produced the greatest improvements in health. Sam selects a vendor focused on nutritional adjustments (not caloric restrictions) and signs up for services. He is easily able to port his authenticated health information from his longitudinal health record to the vendor in order to personalize and hopefully optimize his health improvement. The vendor is in a risk sharing relationship with Sam and guarantees to produce a certain level of health improvement in order to extract payment. This ensures a strong partnership towards health improvement between Sam and his vendor. Furthermore, the health outcome of this partnership is broadcast to the community in order to inform the next participant regarding their probability of success going this route.

#4 Sam’s longitudinal health record has enabled engagement in the above #1–3 while still in his early thirties. Utilizing this ecosystem of personal health data that can easily be compared to others, Sam has had a view into what is realistically possible in terms of health achievement. Each time he is able to improve or optimize his health expression his longitudinal health record is updated with this fact. With regular incentives to remain close to his best achievable health throughout his middle years, he has successfully delayed lipid abnormalities such that the visit above doesn’t occur at fifty-one years of age, it occurs a decade or more later. Sam saves thousands by delaying the use of expensive medications.

#5 Sam decides to share his longitudinal health record with his grand children. They now benefit from Sam’s experience. They see the risk of poor health associated with college, the rough patches with excess alcohol use, the early onset of chest pain, and more all within the social context of the daily complexities of living. It is sobering for the children to learn that their health may also suffer in similar ways.

#6 Statin medication has adverse side effects including the development of diabetes and muscle damage. The pharmaceutical industry is looking for genetic markers that might help predict which patients will be more likely to suffer these complications. New medications are being developed to block these side effects. Sam’s longitudinal health record includes his genome as well as other information that helps inform his propensity for these complications. Sam permissions the researchers to view his genome and related information and uses this information to adjust his therapy accordingly.

Why does healthcare need longitudinal health data now? If 86% of our national annual healthcare spend of $2.7 trillion is consumed by chronic disease management, and much of this is preventable [4], then new approaches are needed. Putting the patient at the center of both prevention and care strategies depends upon putting the individual at the center of their health data. Recall that 80% of the contributors to health outcomes occur outside of medical care, new and effective ways of expanding the healthcare workforce to engage individuals in prevention and self care will be needed. Owning, managing and benefiting from a longitudinal personal health record may go a long way to impacting health outcomes at the community level.

How will this happen? Clearly it will take more than patient portals and current models of population health. It will take new incentive structures that both reward the healthcare system to produce health (not revenue) as well as reward participants for creating the course-correcting intelligence in the first place. Individuals that digitize their health experiences by contributing to their longitudinal health record provide an asset that will benefit themselves, the healthcare industry and the community at large. Innovate incentive structures will be the key to providing stakeholder sustained engagement at scale. This is a vision I am committed to making a reality.

Endnotes:

1) Wu, Shin-Yi, and Green, Anthony. Projection of Chronic Illness Prevalence and Cost Inflation. RAND Corporation, October 2000.

2) Mensah G. Global and Domestic Health Priorities: Spotlight on Chronic Disease. National Business Group on Health Webinar. May 23, 2006. Available at: http://www.businessgrouphealth.org/opportunities/webinar052306chron- icdiseases.pdf.

3) Ray KK, et al. Statins and all-cause mortality in high-risk primary prevention: a meta-analysis of 11 randomized controlled trials involving 65,229 participants. Arch Intern Med. 2010 Jun 28;170(12):1024–31. Review. PubMed PMID: 20585067.

4) Gerteis J, Izrael D, Deitz D, LeRoy L, Ricciardi R, Miller T, Basu J. Multiple Chronic Conditions Chartbook.[PDF — 10.62 MB] AHRQ Publications No, Q14–0038. Rockville, MD: Agency for Healthcare Research and Quality; 2014.