Support Services Centralization and Automation Assessment at Atrium Health Pharmacy System
Atrium Health’s Pharmacy network had invested in two centralized robots to pick medications for distribution to nine different acute care sites. These medications were mostly prescribed by hospitalists on inpatient floors. The primary justification of the automation project was to improve efficiency.
Over time it became apparent that the efficiency gains by centralizing this process were offset by a number of quality and delivery issues. While some time was saved by not having to pick each med at its final location this, was offset by the following:
- Leadtime: In order to get the meds picked and delivered to the floor the robot picking process for the first of the nine locations was started more than 24 hours before delivery. This means that med dosage and type may change. This led to over 7,400 meds being returned to the robots in one year. These returned meds all were handled by RNs on the floors, techs in the local Pharmacy, couriers, and techs in the robot area having to sort and restock meds.
- Meds had to be individually packaged, leading to a slight increase in cost per unit.
- The pharmacy had to have a high level of technical expertise and they would need to update and or change the robot program and inventory as meds changed and new ideas were thought of.
- Downtime: there was a lot of downtime and the internal staff could not fix the issues thus forcing the pharmacy (who, before this, could do their own picking) had to wait for outside technical support.
- There were techs that staffed the robot 24/7 in order to prepare and load the robot racks, collect the picked meds and organize them for delivery to each site.
BAScience Director of Operational Improvement Michael Oswald, then CHS Lean Master lead a team effort to analyze current state and develop a future state process with a much shorter lead time—from 30 hours to less-than-four hours and reduce the number of returned medicines.
The lean team’s solution centered around two main ideas:
- Implementing multiple four-hour delivery cycles per day, instead of the 15-hour, one-batch-a-day process. This significantly addressed the issues with missing medication, reduced the return rate, and volume of med stat requests. It also expedited returns to the pharmacy as they were picked up on the four-hour cycle as well.
- Implementing a local-picking process, where local pharmacy staff custom picked each patient order as it arrived. Not only did this prove more efficient, it could also be done in a space that took up only 20% of the footprint of the pharmacy robot. The robots were discontinued, and each pharmacy was able to pick and deliver meds within four hours of orders being received.
In order to make each location more efficient a couple of things needed to be improved:
- A Kanban ordering system for meds was implemented at each site. This eliminated manual counting, visual checks of inventory and manual order entry of orders. It also eliminated the need to rely on electronic inventory, which was very difficult to keep accurate.
- Most sites also went away from vertical storage units as this was very inefficient from a picking a standpoint and these units relied on electronic inventory and this was not accurate and caused shortages. A Kanban picking process was implemented and this led to very efficient and accurate picks.
Not only did the lean solutions solve the initial issues the pharmacy faced, these operational improvements allowed the facility to remove the existing robot, which saved significant maintenance costs. This also freed up 800 square feet of space for future growth. We freed up an FTE which allowed the facility to activate a number of programs related to medication education and counseling that were previously not able to be staffed. $100,000 was saved in upgrade cost avoidance and $20,000 in maintenance costs.
In general, automation should be used primarily for improvement in safety (ergonomics, direct safety improvement etc.) and quality—improving the accuracy and reliability of a process. Attempts at just improving efficiency can lead to increased cost as the support of the automation now is beyond the health care staffs ability to support. It leads to a higher level of skillset for maintenance and adds a lot of time on the local team for PMs and troubleshooting. Automation can be effective if it is applied properly and with a deep understanding of all the additional technical labor and potential process downtime it can cause.