The packaging of pharmaceutical products is regulated in most countries. Authorities expect that any printed information – such as product, use and storage data on the packaging material artwork or batch-specific data (e.g., expiration date) – is readable by the patient.
Batch-specific data is commonly checked by vision systems using optical character verification. Pharmaceutical companies must be able to document traceability between what patients can read and understand, and how much of single characters can be ‘missing’ before they may be misread.
While the term ‘readability’ in various regulations relates to the contrast between the background color and the text color (or to the font size), it does not consider how much that any missing parts of single characters may impact human interpretation of what is printed. If a patient cannot correctly read the text on a package, there is a risk they may, for example, use out-of-date medicines.
This risk drove BioPhorum to publish its Report for human readability study of single characters – application for biopharmaceutical vision system settings, alongside a Protocol for human readability study of single characters.
“BioPhorum provides a unique opportunity to deep dive into issues like character readability and allows each member to leverage the knowledge base across the different companies to solve common problems.” Brendan Gaffney, Senior Process Development Engineer, Amgen
Good and bad characters
The study supplies the basis for defining acceptable character defects and supports setting up limits for vision systems to detect good/bad characters and minimize labeling errors. For example, the study identified ‘B’ as the most sensitive character for readability if parts of it are erased from the right.
Also, so that a company knows what is sufficient for a patient to understand a printed character, the BioPhorum paper will guide the definition of accept/reject settings so the production line will have a minimum number of false rejects (i.e., good items being rejected even if are deemed good).
The study used a test panel of 48 people of different ages, genders and first languages. It covered all commonly used characters when printing batch-specific data on pharmaceutical sales packages, i.e. A–Z, 0,1–9, ., – and /. The font was OCR-B, in 6pt size as this is the commonly used font and is especially suitable for vision-system checks of printed text.
By erasing parts of each character from the bottom, top, left and right, a ‘limit character’ was identified for each printed character. This is where 90% of the test panel said the character was readable even though part of it had been erased. By correlating this limit character to a complete character, a limit score was identified for each one.
The paper includes:
- A catalog of limit characters with parts of the character erased from the bottom, top, left and right – this provides knowledge about how much of a character can be erased before it becomes unreadable
- A table with corresponding scores for each limit character if parts of the character are erased and the maximum limit score for each character – this provides quantitative guidance for the limiting character
- Test sheets with the identified limit character value marked – this forms the basis for selecting the correct settings in vision systems so they can accept readable characters and reject unreadable ones.
“When our company is questioned about why it selected these vision parameter settings for rejection of poorly printed characters in the dynamic batch data, we can justify the selection based on the study. The study defines the link between what is understandable to the human and the parameter settings.” Steen Howaldt Christiansen Principal scientist – packaging DMD packaging
A webinar is planned for 7 October 2021, when workstream members will discuss how to implement and take value from the study and ending with a Q&A session with its authors. BioPhorum members are encouraged to invite their packaging validation/vision subject matter experts to the event – you can register on this link. How to implement the human readability study of single characters – application for biopharmaceutical vision system settings – BioPhorum