Medical Device Link.

NEWS

Cardinal Health’s RFID pilot program tested many different possible reading stations throughout the supply chain. Read-rate data for item- and case-level tags are included in the chart. (Click image to enlarge)

In a normal distribution-process scenario, Cardinal Health Inc. achieved encouraging read rates on Gen2 RFID tags on bottles and cases that prove the technology’s feasibility for item-level drug tracking, the company says.

Cardinal tagged thousands of solid-dose-product bottles in a nine-month end-to-end pilot in which tags were read at 10 points. Bottles were coded in-line, handled at a distribution center, and shipped in totes to a mail-order pharmacy partner.

On-line coding yields for two solid-dose products in 90- and 100-count bottles were 95% and 97%, respectively. Though unit-level read rates subsequently varied widely at different reading stations, Cardinal identified areas for technology and process improvements in its testing of the RFID tags, readers, and software.

The goal of the project was to test a single-frequency Gen2-based system. “We did not do anything to modify our existing product flow process or try to fine-tune the position, placement, and read capabilities of the readers,” says Julie Kuhn, vice president, operations technologies.

“And we over-read purposely in order to understand the capability of the technology at different process stages. Normally, we wouldn’t have five read stations in the distribution center. At stations where we know we want to read the product, we got very good read rates, though the reading requires fine-tuning,” says Kuhn.
The pilot, for example, employed no shielding to consign space around the readers, Kuhn says.

Item-level read rates were 56.4% and 80.8% for each product at the case-to-pallet aggregation stage at the Philadelphia packaging facility. At this point, 10 reads were made as the pallet made 10 rotations in 60 seconds on a shrink-wrapping turntable.

At pallet shipping from Philadelphia and receiving at the Findlay, OH, distribution center, item-level read rates were below 10% for one product and under 14% for the other. Kuhn says low read rates at these stations were expected as pallets had little dwell time at the readers. “People were not asked to move any slower than they would ordinarily.”

“At all points we were reading to the lowest unit level, reading case and item tags. We were not optimizing the reader, tuning it to look for either cases or items,” she says.

High unit-level read rates were found when reading cases one at a time in case receiving at the distribution center. High item-level read rates realized in distribution center totes dropped off considerably when the totes were shrink- wrapped and shipped.

Though tagged bottles were mixed in totes with liquid packages and foil blisters, item-level tote reads at the quality control phase were 99.5%. “Many days we had 100% read rates,” she says. Kuhn says that a metal tote cart employed might be one reason for lower read rates at subsequent tote reader stations and at the pharmacy.

Tags were coded online at normal validated line speeds for the products of 50 to 60 bottles per minute, with testing of up to 120-per-minute speeds. Alien Technology tags were coded with the EPCglobal SNDC 96 format, including the NDC number. Alien readers were loaded with IBM’s Websphere RFID Device Infrastructure (WRDI) software, communicating with its Websphere Premises Servers. In the pilot, Cardinal beta-tested IBM’s EPCIS (EPCglobal Information Systems) data repository where EPCs coded with EPCs read were compared. “We ran metrics on what percentage of each encoded tag was read at each station,” she says.

Case-level read rates were virtually 100% at all stations for one product and generally high for the other. Cardinal reported that case reads can likely be improved to 100%; “however, further tests are needed to prove this hypothesis,” she says.

“The case-level reads were excellent. We believe that in the short term, unit-level inference should be an acceptable practice for normal distribution when unit-level reads are unreadable, but case-level reads approach 100%. Inference will be necessary to help us operationally as we learn this technology. We expect FDA to take a stand as to whether inference is an acceptable practice or not,” Kuhn says.

“Overall, we were very pleased with the performance. We are in the upper 90s at a lot of our critical read points. This is the first pilot to really complete with the Gen2 technology. Everyone had data points on Gen1 technology. I think what we demonstrated is that Gen2 technology is far superior to Gen1 and can yield tremendous results. Now we will tweak the process to push read rates beyond 99.99% to Six Sigma performance,” she says.

“If you are going to leverage RFID for e-pedigree and for inventory management to drive ROI, you need a high level of confidence and 100% reliability,” Kuhn adds.