Wednesday, March 12, 2014

Seize the Data! Will Big Data Save Us from Ourselves?

Today, Partnerships welcomes guest blogger Paul Ivsin VP, Consulting Director at CAHG Clinical Trials. He specializes in study enrollment strategies and patient engagement. He writes on clinical trial issues at companion blogs Placebo Lead-In and Placebo Control.

For me, one of the more intriguing panel discussions slated for this year’s Partnerships in Clinical Trials conference is “Seize the Healthcare Data: Moving the Industry Forward in an Era of Overabundance of Information, Data & Analytics”. An impressively high-level group of speakers will be discussing how “Big Data” can overcome obstacles associated with patient enrollment and retention.

This is a hot button topic for me because a) I love data, but b) I've been extremely frustrated with much of the Big Data hype that’s entered the patient recruitment arena in the past few years. (In fact I've gone so far as to call much of it “snake oil” and “boiler room tactics”.)

Here are three questions I’ll be keeping in mind while listening to this panel:
1.) How can we avoid having ever-increasing restrictions on personal medical data throttle any growth in this area?
2.) Besides certain (limited) cases, is healthcare data really even “Big” data?
3.) Will we be able to get the data we need, or will be have to be content with analyzing whatever we can get?

Carpe Datum! Can Big Data bring these rings together?

Part of the problem, of course, is inherent in the nature of our business. Healthcare data is both a lot more private and a lot smaller than many other areas where Big Data appears to be a viable solution.

When big retailers start trawling through their masses of purchase data, it certainly makes some people nervous. However, these kinds of consumer big data activities don’t usually trigger calls for tighter regulations or enhanced purchaser privacy. Most of us seem to be OK with these analyses, as they seem fundamentally unlikely to lead to anything threatening. If my grocery wants to pore over my historical produce selections to glean deep insights about my personality, then more power to them. (At the very least, I've noticed it’s made them a bit better at giving me coupons I might actually consider using.)

However, clearly we cannot say the same thing about health and medication data. HIPAA is close to celebrating its 20th birthday, and its effects only appear to be getting more powerful over time – both formally with recent legal extensions like the HITECH Act, and informally as institutions adopt increasingly restrictive healthcare information policies.

So, if we find it hard – and getting harder – to aggregate personal health data from multiple sources (payers, physician, pharmacies, etc.), how will this not severely hamper our ability to use this data to match patients to clinical trials?

This lack of aggregation, combined with the generally fragmented and local nature of most healthcare provision, also may limit the “3 V’s” of healthcare data from ever becoming truly Big:
  • * Not aggregating sources limits Variety – hospital data stays with hospitals, and pharmacy data stays with pharmacies.
  • * Local provision restricts Volume. Consider: Healthcare giant Kaiser Permanente is often highlighted as being on the cutting edge of Big Health Data. But Kaiser’s databases comprise only 9 million total people. Compare that to Walmart, which collects 35 million customer interactions every single day.
  • * And healthcare in general limits Velocity. The overwhelming majority of people only interact with their physicians once, or possibly twice, per year. Thus, the rate of change in the data, except for the small fraction of patients who are seriously ill, is not particularly high.

So, do many of the latest and greatest Big Data techniques and technologies really work for healthcare? There are perhaps some exceptions – the massive growth in genetics data comes to mind – but these may still be fragmented and isolated from the rest of our health data.

Both of these concerns lead to the biggest question of them all: will we be able to aggregate enough patient data to really answer the questions we need answered (such as: “who exactly are the patients who will be eligible for this trial?”)? Or will we be forced to continue down our current path: basing decision on the data we can actually get, rather than the data we really need?

I look forward to some robust discussion on the topic on April 2. Details of the panel are:

Seize the Healthcare Data: Moving the Industry Forward in an Era of Overabundance of Information, Data & Analytics

According to the National Institutes of Health, over 80% of clinical trials fail due to complications with enrollment timelines and troubles with patient retention. These problems lead to insufficient data sets, stalled drug development, and ultimately billions of lost dollars spent by pharmaceutical companies. What are the possibilities of leveraging big data in the drug development world and where do the opportunities and obstacles lie? Experts discuss how you can turn data into insights.

  • Moderator:
    • Graham Hughes, M.D., Chief Medical Officer, SAS Institute
  • Panelists:
    • Eric Perakslis, PhD,  former CIO and Chief Scientist (Informatics), U.S. FDA, Executive Director, Center for Biomedical Informatics, Harvard Medical School; Senior Advisor, Precision for Medicine
    • Bryan R. Luce, PhD, MBA, Chief Science Officer, Patient-Centered Outcomes Research Institute
    • Brian Hagen, PhD, Leader in Decision and Risk Management, Managing Director, Decision Empowerment Institute

No comments: