'; if(pArray.length >= 4) { i=2; } else if(pArray.length >= 3) { i=2; inline = ''; } else if(pArray.length >= 2) { i=1; inline = ''; } else if(pArray.length === 1) { i=0; inline = ''; } $('#librarydrawer_story_container script').each(function() { $(this).remove(); }); $(pArray[pArray.length - 1]).after(subscribeBox); $(pArray[i]).after(inline).after($('#librarydrawer_story_container')); $('#text-story').focus(function() { if ($(this).val() == 'EMAIL ADDRESS') { $(this).css({color: '#000000', backgroundColor: '#ffffff'}); $(this).val(''); } }); $('#text-story').blur(function() { var trim = $(this).val().replace(/[\s]/g, ''); if(trim === '') { $(this).val('EMAIL ADDRESS'); $(this).css({color: '#666666', backgroundColor: '#f8f8f8', border: '#666666 1px solid'}); } }); $('.content-subscribe .btn-submit').click(function() { var val = $('.content-subscribe .text').val(); if(val.search(/[a-z0-9!#$%&'*+\/=?^_`{|}~-]+(?:\.[a-z0-9!#$%&'*+\/=?^_`{|}~-]+)*@(?:[a-z0-9](?:[a-z0-9-]*[a-z0-9])?\.)+[a-z0-9](?:[a-z0-9-]*[a-z0-9])?/gi) === -1) { $('.content-subscribe .text').css('border', '#ff6600 1px solid'); $('.content-subscribe .text').focus(); return false; } return true; }); }); //-->

Finding universal truths in healthcare analytics case studies


It's hard to have a broad conversation about either analytics or healthcare. Both topics are quite academic, and neither lends itself to credible generalities, so before very long, the conversation often turns to a narrow set of data or a specific medical specialty.

It's even harder to have a broad conversation about analytics in healthcare. Anyone speaking about the topic will broadly acknowledge the importance of analyzing healthcare data before diving into a detailed use case, complete with charts, graphs, flowcharts and, if you're lucky, mathematical formulae.

What often emerges is a great case study with a clear underlying theme but an otherwise narrow application. To find larger lessons, you need to be able to pull a universal truth, no matter how small, from the conversation.

This was my mindset as I headed to last week's Medical Informatics World conference in Boston. Amid the rapid-fire case study presentations (none of them longer than 30 minutes) I sought some universal truths for healthcare analytics projects, ones that would apply equally to payers, providers, government agencies and other stakeholders.  

After listening to about a dozen speakers, I came up with a few:

  • Separately, the Christiana Care Health System and QualCare Alliance Networks Inc. discussed the benefit of incorporating risk stratification analysis into patient and member engagement efforts. This works best when organizations focus on a specific goal--reducing utilization, say, instead of just cutting costs.
  • Massachusetts General Hospital outlined how it opted to integrate population health IT pillars--data aggregation, analytics/risk stratification, care coordination and patient outreach--to escalate patient interventions.
  • Boston Medical Center described its four-year effort to determine why its mortality observed-to-expected ratio was so high. It pushed causality data to staff in a way that did not assign blame and used the data to design better care interventions. The safety-net hospital went from ranking in the bottom 10 percent for unexpected deaths among University Healthsystem Consortium hospitals in 2009 to the top 25 percent in 2013.
  • The Department of Veterans Affairs, in addition to presenting ways to close common medical device vulnerabilities, described an unintended benefit of data governance: After integrating VA and Department of Defense data into a single interface, data appeared for every single medical center a patient had visited.
  • University of North Carolina Health Care noted the importance of write-once read-many projects that provide definitive answers, once and for all, to questions such as "Who is a diabetes patient?" This ensures that all future analytics projects need not continue to redefine common variables.

However, the simplest, most universal truth of all came from Gowtham Rao, M.D., the chief medical informatics officer with BlueCross BlueShield of South Carolina. Rao discussed how the insurer noticed that trauma center surgeons and gastroenterologists were more likely to seek blood transfusions than other physicians. BlueCross BlueShield investigated why and worked with those specialists to reduce unnecessary transfusions.

The process ultimately boiled down to four simple steps:

  1. Get data
  2. Plot it
  3. Look for outliers 
  4. Manage them

It's fair to say that, for healthcare analytics at least, truths don't get much more universal than that. - Brian (@Brian_Eastwood and @HealthPayer

Related Articles:
Combine member engagement with risk stratification to achieve savings
5 ways to close common medical device vulnerabilities
True patient-centered care needs a personal touch
Using analytics to achieve value-based care in rural areas [Special Report]