I’ve been doing a few packaging validations and my results have been horrible. Where this project fails is when it comes to the count of tablets in the bottle. I had the criteria set at no under counts. We have been unable to reach this result, did I set the criteria to high? We have no true historical data that’s why I’m having trouble setting the criteria. What would you recommend doing and from your experience is a no under count possible or is there an industry standard that I’m not aware of?
First thing, the spec of zero undercount is perfectly correct. We have IMA’s on our site, and the equipment spec is zero underfill. In terms of GMP, the filling operation may overfill but cannot underfill, as the equipment may be used for critical meds (patient-ready packs) where the patient cannot miss any doses. The overfill allowable is somewhere around 2 % on our site. Check the position of the bottles under the filling nozzle, make certain that no tabs are hitting the bottle or PI’s and bouncing out. Also confirm the accuracy of your counter by using specific amounts of tablets, as the problem often lies with the counter sensor settings being too sensitive and registering dust, etc, resulting in under counts. Repeat this exercise a number of times using a range of tablet size/type/bottle combinations. Dusty tablets are usually the worst-case products and a lot of attention needs to be paid to these, but highly polished dosage forms can also cause problems.
As a separate exercise, place specific amounts of tabs, enough to fill say 10 bottles, on the feeders from the filling hopper and use these to adjust the feeder vibration settings, etc until you get the correct fill amount with zero underfills. Then test under production conditions again. Ensure that the exercise covers a range of the products/containers intended for filling on the particular filling line. Once again, no undercounts should be allowed. Often a major problem is encountered with uncoated tablets which tend to leave dust on the counter sensors, causing an erratic beam which doesn’t always register when a tablet falls, or causes it to recognise more than one tablet. Problems also occur when packing small tablets at high feeder speeds where the tablets fall simultaneously, but their spatial alignment is such that the sensor only registers one tablet. Also, tablets which tend to stick tohether can cause count errors. These usually result in overfill errors.
my experience has shown that your problem is in the equipment. Before any validation run is programmed, the fillers must undergo PM and calibration if necessary. A few Dry-Runs before execution are a good way to know if your equipment is ready to go-live. One other thing, if this is a bulk batch for distribution and no historical exists, set your criteria at +/- 1 or some meaningful and realistic targets; also check with the equipment manufacturer, they may have published a criteria of some kind. Also remember that a sampling method is in effect, that no system is perfect and variation exists in everything!
Thanks for all your comments. We did decide to go with a reasonable variation. We do not have the vision system for the slat fillers and are equipment is pretty old (and manufacture states a 98% accuracy) so we factored this into the evaluation of setting up the acceptance criteria. I appreciate all your comment because it lead to decision that senior management was happy with.