Automated stocking and ordering systems raise the issue of compliance with the system's removal, restock, and counting requirements. These quite expensive (but often well worth the cost) systems do not work as intended unless users are reasonably compliant — mainly that they conscientiously record item removals and restocks.
So, in some hospitals, you end up seeing compliance charts all over the walls in patient care areas. Compliance becomes a routine topic in staff meetings.
The reality, at least so far as we have observed it, is that compliance is generally pretty good but that it varies greatly by item. Some items have steady 100% compliance, while others struggle to get above 60%.
If you look at transaction data from these supply stations, you can see the compliance behavior in detail. Some items exhibit occasionally weak compliance periods, with other periods of very high compliance. Some items show just a single period of weak compliance over a 3-month sample.
Compliance behavior appears in the form of the difference between manual stock counts, which are keyed into the stations, and system-tracked counts, which show only compliant removals and restocks. The manual-to-system count difference is a direct measure of compliance.
Where available, system transaction data makes it feasible to adjust ordering automatically for each item's current compliance behavior. We do this in our simulation models where point-of-use transaction data can be obtained. If you track the differences between counts (actual) and system stock levels, you can adjust inventory to reflect the non-compliant demand. We call this the "shadow inventory".
An order management system should include, as our simulation models do, a shadow inventory count that includes the adjustment for compliance. You can, for example, spread the non-compliant demand evenly across actual (tracked) demand, so that each actual demand transaction quantity has a increment to account for non-tracked demand.
You can also adjust for non-compliant (untracked) demand only over periods where non-compliance is detected. This leaves actual demand unchanged during compliant periods.
An order management application that tracks end-use demand for a hundred or more automated supply stations has to handle a great volume of data. There are a number of tricks for doing this efficiently that are built into our models.
This approach means that 100% compliance is not necessary if you are able to adjust ordering to accommodate a "natural" level of compliance. What is a "natural" level of compliance?
Based on our observations, it is the best that conscientious staff members are able to do in each point of use location. Operating rooms cannot do as well as less crisis-prone locations. Some nursing units achieve very high levels of compliance while others struggle to meet minimal levels.
The point here is that compliance level goals should be experience-based, set after some period of reasonable encouragement to improve. This approach assumes that staff members are by then "doing their best" under their local circumstances.
Storage space for hospital supplies is nearly always tightly constrained, especially at and near point of use locations such as nursing, ICU and operating room areas. Using available space efficiently is critical ... Space Utilization ...
To return to the main topic list ...

These systems appear to be either wonderful or way too costly, depending upon whether you are talking with a user or someone in finance. We believe that both are correct.
Users value the protected and highly organized storage. They also like automated ordering that prevents the stockout frequency typical of many manual systems.
Finance folks may find the systems too costly because they do not see their value clearly. They may also realize that the stations, if not well-disciplined, can be major cost generators.
When you compare manual point-of-use stocking and ordering process costs with automated system costs, the savings are normally large. Manual purchasing is a very costly process.
One of the outputs of our models could be comparison of automated vs. manual storage and ordering costs.