Thursday, August 11, 2011

Freakonomics and CKM

Asking the right questions and using the right analytics can lead to major knowledge breakthroughs. However, it is often difficult to know which questions are the right questions.  

Steve D. Levitt and Stephen J. Dubner, the authors of Freakonomics, would argue that you should not focus on asking one specific question based on what is known, but instead ask a variety of questions and use the right analytics to discover hidden connections between seemingly unconnected variables.  They reason that you will hit a number of dead ends and a number of questions won’t have clear answers, but some will be extremely revealing and meaningful.

The analytic approach they write about is based in economics with five major principles:
“Incentives are the cornerstone of modern life.”
“The conventional wisdom is often wrong.”
“Dramatic effects often have distant, even subtle causes.”

““Experts” – from criminologist to real-estate agents – use their informational advantage to serve their own agenda.”

“Knowing what to measure and how to measure it makes a complicated world much less so.”

Their book analyzes the drop in violent crime in the 1990s and cheating in sumo wrestling, among various other topics.  One large subject area that was mostly absent from the book was healthcare.  

I believe that their approach could be very revealing when applied to healthcare quality improvement, or the relative lack thereof.  Deep reliance on expert opinion rather than objective data, misaligned incentives that don’t support quality goals and limited ability to measure and understand the measurements of quality outcomes must play a role in the state of quality improvement. It would be fascinating to see the authors of the book explore this issue. With a robust health information system in place, many questions could be asked and efficiently answered, potentially revealing meaningful hidden connections to quality improvement in the healthcare setting.

Thursday, August 4, 2011

Better Questions, Better Answers, Better Solutions

Often, when we are struggling to make good decisions, the problem is not the quality of the data available… it is the quality of the question.  This is because the threshold for achieving actionable knowledge is dependent on the question being asked. The more simple and specific you make the question, the easier it is to gather enough information to reach the threshold for actionable knowledge. If the question being asked is too broad, the question itself may prevent the creation of actionable knowledge.

A perfect example of this in the medical field is the current discussion about salt intake. The main concern is the sodium component of salt. Diets high in sodium have been associated with increased rates of hypertension, which is clearly linked to an increased chance of heart attack and stroke. Because of these associations, there has been a public health campaign to advise adults in the US to reduce their salt intake. However, the campaign to reduce salt consumption has recently become controversial. This controversy stems from asking a question that is too broad to facilitate the creation of actionable knowledge. The question being asked is “What should we tell the US adult population about salt intake?”

A recent article in the Archives of Internal Medicine demonstrated a reduction in cardiovascular mortality associated with lower sodium diets. The study was statistically adjusted to be a representative sample of the US population. This article gives the appearance of creating sufficient actionable knowledge to tell the US population to eat less salt. However, two other recent studies create serious doubts about that conclusion. The first study, published in the New England Journal of Medicine, was also a study modeling the US and showed that the mortality benefit for a lower sodium diet had a significantly larger impact for blacks than for whites. The second article, published in the Journal of the American Medical Association, was a European study that only looked at a relatively young and healthy white cohort; this study demonstrated a large increase in mortality for the subjects with the lowest sodium diets.

These articles appear to disagree and there have been discussions about the methods of each study. No study is perfect and the conclusions from any of the three may be proven incorrect, but it is also possible that all three may be correct. It is possible that young and healthy whites in Europe may be harmed by a diet that restricts salt intake. Sodium is necessary for several biological functions. However, white, young, and healthy only represent a small group in a study that utilizing a sample representative of the adult US population. The potential harm to the white, young, and healthy cohort may be washed out by the benefits for other groups.

These studies create debate because the question being asked is too broad. If you believe that eating less salt will harm a cohort of people, it is unethical to tell them to do so. If you are then trying to take action based on the question “What should we tell the US adult population about salt intake?” you have quite a dilemma.

I suggest in these cases you don’t try to solve that dilemma with more data, information and knowledge. You solve it by changing the question. If some groups are helped by an action and others harmed, the question should not be “what advice do we give the whole population?” The question should be “how do I advise each cohort in the population?”

With regard to salt intake, we should stop asking “what should we tell the US adult population about salt intake?” and start asking "who in the US do we need to tell to eat less salt?"

Monday, August 1, 2011

Health Information Exchanges and CKM

According to SearchHealthIT, health information exchange is "the transmission of healthcare-related data among facilities, health information organizations (HIO) and government agencies..." Health information exchanges have been suggested as a tool to improve the delivery of care, especially in emergency departments and other acute care settings. However, there has been little evidence that proves they create actionable knowledge.

One study that indicates HIEs do produce actionable knowledge was presented at the Society of General Internal Medicine Annual Conference earlier this year. Dr. Lisa Mabry and colleagues presented a poster entitled "Does Health Information Exchange Use Improve Adherence With Evidence-Based Guidelines for Neuroimaging in the Emergency Evaluation of Headache?" Their study investigated whether ED staff use of the MidSouth e-Heatlh Alliance HIE, a health information exchange in the greater Memphis area, reduced the use of neuroimaging in patients presenting with repeat episodes of headache. Their results showed that although the HIE was only used in 21.8%  of encounters, when it was used, it was associated with a 76% decreased odds of neuroimaging. Additionally, adherence to evidence-based guidelines was improved.

These results indicate that, at least with regard to headache, the HIE does provide the user with actionable knowledge. If the HIE was used more frequently, unnecessary cost, testing, and radiation exposure could be avoided.

As creation and utilization of health information exchanges expands, it is important to design them in a fashion that assists users in reaching the threshold for actionable knowledge. Focusing on overcoming technical challenges to allow for the transmission of data within HIEs rather than insuring their ability to communicate actionable knowledge will limit the effectiveness and relevance of this technology.