Anticlue

System Monitoring

One of the most overlooked aspects of implementing a system, is monitoring its functionality when it is live.

The reasoning for monitoring is that within healthcare systems, one system depends on many to function properly. To get data to all of the systems that require the data, the data needs to go through many different routes. For example to get your laboratory results in the clinical data repository, your information gets entered into the registration system, then proceeds to the interface engine, which forwards it on to the laboratory system. The laboratory system then send an order status update to the interface engine, which sends the update to the registration system. The registration system then send the confirmation of your information to the interface engine and on to the laboratory system, which sends the actual result to the interface engine and finally to the clinical data repository. There are a lot of steps in this process that are interdependent. I didn't even go into the network connectivity and server functionality. Having the ability to monitor a interconnected system's functionality at the data integrity level (meaning we sent this did you get this where you expected) and at the level of date entry functionality helps to create a reliable functioning system. It also helps to alert the proper individuals quickly when an issue has occurred.

Users need to have faith in the system, and they don't want or need to understand all of the interdependent steps. They want to be assured that what they enter gets to where they expect. Having the same information entered on several test patients in a production system helps this process. After all 1% of an applications downtime can cost a 500-bed hospital 1.4 million a year. (statistic from Healthcare Informatics, April 2002, "The Toll of Downtime") That is 87.6 cumulative hours for the year for application's available 24 hours a day 7 days a week.

Another aspect of this when implementing systems, is to be considerate of all of the interdependent parts, and try to minimize unnecessary functionality until needed.

For example, if you are implementing a routing system for distributed printing for the transcription system. Wait until you print to more than just one printer in the transcriptionist's department.