KP Flexing EMR Data for Public Health

Don’t vaccinate your child for whooping cough and put them at a 23x greater risk of contracting this nasty disease.  That is the finding of a research report released today by Kaiser-Permanente (KP).  Last week Chilmark Research had the opportunity to speak with the lead researcher, Dr. Jason Glanz of KP on this research project and lessons learned.  Our interests lied in better understanding how the research leveraged the KP EMR.  Here is what we learned.

First some quick background:

Whooping cough is actually on the rise having hit a low of ~1,000 reported cases in 1976 and in 2004, over 25,000 cases were reported.  The rise is believed to be caused from two factors: teenagers contracting the disease as vaccine loses effectiveness and secondly, lack of full set of vaccines to infants (requires three successive shots).  While whooping cough is not that dangerous for teenagers and adults, it can be life-threatening for infants and small children, however, is preventable.

Goal & Methodology:

One of the primary goals of the research was to determine if parents who refuse to give their children the full set of vaccinations put their children at greater risk.  While there is the assumption that this is the case, Dr. Glanz wanted to determine what is the level of that risk in quantifiable terms.  In doing so, the hope is that pediatricians will have better, evidence-based research at their disposal to assist in educating their customers (parents) the risk they put upon their child when refusing a vaccination.

Digging into the data of KP’s EMR in Colorado, the researchers extracted the records of children over an 11 year period (1996-2007) to look for cases of whooping cough and correlation to vaccine refusal.  Over 100K records were reviewed. Once the data was extracted and reviewed, SAS statistical tools were applied for analysis.

Why this is important:

As we march down the EHR adoption road with a $36B wind at our back, compliments of Uncle Sam, a critical issue is: What do we, as tax payers who are ultimately paying for this, get in return?

Part of the answer may lie in research such as this that leverages massive amounts of data in an EHR (or in the future a network of EHRs) to determine with a high degree of confidence such issues as behavioral risk, adverse drug events, effectiveness of various treatments to demographic subsets of the population, etc.  This could have a massive impact on future healthcare practices resulting in better, more effective and potentially personalized care.

For example, in this whooping cough example researchers found that the demographic most likely to refuse a vaccination is Caucasian, upper middle-class and well-educated.  Knowing that information in advance will assist educators in crafting educational content targeting that demographic.  If, on the contrary, it were primarily a Hispanic population, content could be generated in both Spanish and English.

By no means a slam dunk:

While it is great to hypothesize on the potential that prudent, secure and safe use of health record data to perform such studies will create a new revolution in healthcare research, it is far easier said than done.  Today, most EMRs can not readily share data in support of such research efforts.  One can only hope that the future criteria for meaningful use (information sharing) will reflect the need to support this type of research.

But even within most hospitals and IDNs, including KP, there is insufficient attention paid to how to structure the EMR to support research projects such as this example.  In the whooping cough case example, Dr. Glanz stated that one of the most time consuming processes in the research was the examination of records by researchers to determine if a parent exercised the exemption clause to refuse the vaccination for their child.  An outcome of this research, is a new field in the pediatrician’s EMR that they will check if the exemption clause is exercised by a parent.  This will automate future analysis of data sets for this critical variable.

Extending this example further, what do hospital and IDN CIOs need to be thinking about today in their EMR implementations to insure that data for such research is more readily collected and analyzed at some future point in time?  While it may be impossible to predict all potential use cases of EMR data, CIOs would be wise to start in areas where they have develop, or will develop strong competencies of care (e.g., cardiology, urology, oncology, pediatrics, etc.). They would also be wise to actively solicit the involvement/representation of their internal research group in defining attributes for the EHR prior to install.