Aaron Berdofe, MHI – Health Informatics

“I’m sorry, Dave. I’m afraid I can’t do that.”

Have you ever tried to write an NPR report?  Have you ever scoured a data definition only to come up with half of the information you need? Or maybe you’ve painstakingly created the perfect NPR report, let it run for an hour and find to your utter dismay that it has only produced three words: “No records found.”

The MEDITECH world (and thus, LSS’s) is structure hierarchically.  Theoretically, this is supposed to allow for faster processing of single node related information (i.e. a single patient’s data).  However, the term “faster” in the 1960s really did mean noticeably faster. Today, in the new millennium, it’s a difference of nano-seconds.  Unfortunately, if you want to report across a variety of nodes, unless there is an index created specifically for the information you want, it’s going to to be slow.  Not new millennium slow, but still 1960s slow.  Maybe it’s a generational thing, but to channel Veruca Salt, “I want it now!”

When working with healthcare organizations that have some local competition, I’m typically get asked to provide information such as where the patients in their physician practices are being referred to.  I could develop an NPR report for this.  You can, again, theoretically, create any kind of report your heart desires in NPR.  So step 1, you print the data definition for RXM.ORD to Preview (don’t get me started on their text viewers).  You do a quick word search for “referral” and find the segment rxm.order.referral.  Great!  That’s what you’ll use!  But wait, it’s a collapsed node so it’s actually split into two child segments: referral.to.groups and referral.to.providers.  Clearly, not much thought was put into designing this for data output (the whole point of an EMR).  Healthcare executives want to know how many referrals are staying in house and how many are going elsewhere.  They don’t care if a provider refers to a single physician or just a group.   If you want both referrals to groups and referrals to providers on the same report you need to either fragment or write yourself up a macro.  Both of which take more time away from you actually getting the information you want.  I run into these situations all the time.  And you get to that point where you realize that it is going to take quite a bit of time not only to run this report, but to write it as well.  The quirky UI/UX of MEDITECH NPR isn’t exactly a pleasant experience after you’ve spent your hours searching for where the data is coming from in the first place.

So what do I do?  Do I lock myself in a room for hours on end to write up these reports?  No way!  Why would I waste my client’s money like that?  I’ll just extract the raw data from the system and process it in something that actually makes sense.  I would go insane if I couldn’t download information out of NPR.  In the above example I create two reports, one for each child segment.  I don’t sort, I don’t filter, I don’t summarize.  I put the important fields on the picture, separate them with the pipe “|” character and then download it as a .txt file.  The pipe character is necessary because some fields in MEDITECH and LSS (*ahem*, Name fields) force the user to put a comma to separate multiple strings within the field.  Therefore, if you use the comma character to separate fields, you’ll find yourself in a messy situation.

After downloading the report I’ll put it into a relational database, like Microsoft Access.  Relational structuring of data branches together like items through their relationships.  For example, instead of pulling out each patient’s chart to find the average blood pressure for your patient base, you would pull out a chart marked “Blood Pressure” that had a list of everyone’s blood pressure written down.  No need to dig.  While downloading the reports out of NPR can be just as tedious as running them in the sense that they’re time hogs, importing them into the database, linking their common denominator and running a query using a very simple UI or writing a SQL statement takes less time than creating an NPR report from scratch.  Not only can you calculate and summarize data more easily, but you can display data more easily and in a much more visually appealing manner.

It gets even better if the information is desired regularly.  Schedule the downloadable report to a spool group which dumps the .txt file to a server location.  Have the .txt file automatically imported into a database which displays all the information on an intranet page.  Time to wait for the report to run when it’s demanded? ~0 seconds.  Oh and did I mention that you can actually graph the data in realtime if you get out of the MEDITECH world?  Who loves charts and graphs?  Everyone!  I can’t stress enough the importance of visualizing vital information relating to the business of running a healthcare organization and the health of your patient population.

MEDITECH and LSS didn’t design their systems to utilize the information captured in to them.  Why fight it?  Do your data processing elsewhere and download those NPR reports out.