The majority of healthcare providers are still struggling to aggregate their disparate enterprise data in order to drive clinical decision support. While very valuable, this is generally used for retrospective analysis. But what if that data could be used near real time, and give you the ability to effect change while the capability to do so still exists? This sounds great, so why isn’t everyone just jumping ahead? Well of course it’s not that simple.
There are three core types of data analysis, all of which should and can play a key role in supporting clinical decision making and driving us all closer to true evidence-based decision making; retrospective, real-time monitoring, and predictive analysis.
Retrospective analysis is valuable as it allows us to comb through vast amounts of previously acquired data to determine trends and identify valuable data points. This is classic clinical decision support. This should be utilized to drive best practices, standard order sets, and key performance metrics. Real-time monitoring is valuable as the data is current and when properly captured should affect clinical, financial, and operational decision making in near real time. This is closer to business process monitoring and alerts. Predictive analytics looks to identify trends and or events in the future based on trends and related data points identified through the analysis of a large amount of previously acquired data.
All of these practices have their place and should be used appropriately in some form within the clinical, financial, and operational environments. To me though the most valuable of all three above mentioned data analysis varieties is real-time monitoring and alerts. While retrospective analysis can allow you to identify what happened, there isn’t anything that you can do about it now as the event has long since occurred. Predictive analysis can in some cases identify events before they happen, but this is based on the application of models and probability to the data so it has a lower success rate and provides a wider variant of accuracy.
The true gold lies within real-time monitoring and alerts. If applied properly these processes will allow you to affect change on a clinical interaction while the ability to do so still exists. A quick example; a patient is seen in the ER and is given the primary diagnosis of Pneumonia. The Core Measure states that upon a Pneumonia diagnosis an antibiotic must be administered within four hours. If this doesn’t occur, the patient is then at risk as it may be more difficult to treat their condition (and may now cost more to do so), and the providers core measures have been impacted as they have failed to meet the metric.
What if there was a system in place that would alert the assigned nurse at hour one that they still hadn’t given the medication, and then alerted the charge nurse at hour two, and as a last ditch effort alerted the CNO and CMO at hour three before the four hours elapsed?
This is only a very simple, yet powerful example of the power of shifting from the use of clinical data for clinical decision support to clinical decision assist. Next week we will take a look at further examples of the power of clinical decision assist within the provider environment, so stay tuned!
