When customers ask their contract manufacturers to share data in an electronic format, they are not simply asking them to scan all the documents into a PDF format and e-mail them.
Why This Is Not Enough
Customers are usually looking for data they can read into their own systems to run statistics and do a variety of other tasks. Manually scanning documents does not put them in the right format to meet customers' needs. While they often receive batch records in this manner, they cannot necessarily use batch records for data-sharing purposes.
As we all know, batch records are big and complicated. The amount of data a customer might need from the batch record for its electronic data purposes might be quite small. Also, the data format needs to be more easily and reliably readable via electronic means than traversing through the entire batch record. Data should be limited and should be sent in the same format and order each time.
A Simple Solution?
Companies need concise data in a reliable format - just the necessary data - in addition to the batch record. That does sound easy and just about anyone with any technical skills can figure out how to set up some reliable format; one might think the problem is solved, but unfortunately, it is not.
A format and a group of data cannot be selected without input from the customer. Whatever might seem practical to the sender is not necessarily so to the receiver. Additionally, the customer might not want the data merely sent in some format, but might want a true interface between their own systems.
Trying To Standardize
While there are a few standards in the industry, there is no overall standard for this type of data interchange. This means each interface or data transfer is basically a custom one. When the data interchange created is customized for each client, it becomes expensive to create and to maintain. Additionally, creating an interface with each customer's system can be even more expensive than creating a standard format for the data, because of all the technical issues involved.
When one considers that each CMO has many customers and that those customers might have multiple CMOs , multiplied by the number of different systems each one is using to manage the data they want to share, you can start to imagine that trying to standardize this isn't a trivial issue, which is why we still don't have a set of standards.
For example, data sharing would probably occur at the point when the batch record is sent to the customer. At that point, there is usually a COA (Certificate of Analysis), and some might think that the system that creates the COA could easily share the data with the next system. However, consider that the COA is created by some companies using their ERPs (Enterprise Resource Planning, such as SAP or Oracle ERP), while other companies use their LIMS (Laboratory Information Management System), and still others type them out in something like MS Word. Even if two companies use the same type and brand of system, their implementations will be different, and the data fields and format will often be significantly different.
Legal and Cost Issues
When a customer asks its CMO for this type of data sharing, it sometimes occurs after the contract has been signed and the work has begun. Contracts contain specific terms and deliverables; asking for data to be sent in a specific format may mean altering the contract terms - usually, this is not a trivial matter.
But customers sometimes think this is a trivial matter and do not realize that they are asking for special formats that would incur additional cost. They do not realize they might have to pay extra for what they are asking, nor do they realize that the CMO might not be able to free up its IT staff to take on the task, even if the customer has the extra money to spend to achieve it.
Industry Standardization Failures
One reason that industry-wide standards have failed is that they end up including too much data, and the strategies are too complicated. Keeping things as straightforward as possible is a key factor. Additionally, even with a straightforward approach regarding the actual formats, the issue of mapping the data from one company to another must be addressed, too.
For a single example, consider that many, many fields of data are being shared from one company to another. Looking just at a condition for a stability study, consider that one company has a system that calls their 20 C/40 RH condition 20C40RH, while another calls it 20C_40RH, and yet another may refer to it as 20 C / 40 RH. There are tools that can address this issue. But obviously, there are quite a number of variations and this is just an example of a single field! Use your imagination, and you can quickly see how very complicated this issue is.
If only we could start system implementation from scratch all together today. We would come up with standard fields for companies to use. However, with their huge investment in the systems they're already using, along with their proprietary, internally developed naming conventions, companies are not likely to realistically embrace any standardization that requires them to change very much, if at all.
Some companies believe that the biggest player wins. So, if the customer is a huge company with many CMOs, they believe that they can dictate a format for sharing data. In other cases, when the CMO is large and serves many smaller customers, they believe they are the controlling factor. The message from the larger to the smaller company is, "If you want to work with us, a huge company, you will comply with what we want you to do."
This strategy probably works if the large company takes the initiative and creates some standards for everyone to follow. These standards need to be economical for all involved, such that even the smallest companies can afford to comply. Additionally, mapping issues must be addressed along with formatting ones.
Still, if every large player forces the companies working with it to standardize formats, smaller players that work with a number of larger companies will still end up supporting multiple formats - but consider that they havethe fewest Resources and are least able to support multiple formats. Thus, the smallest companies would end up shouldering the largest burden.
If the largest players were to get together to create a shared format, this effort could result in the establishment of a standard. In the pharmaceutical industry, the Pistoia Alliance (www.pistoiaalliance.org/), a consortium of large pharmaceutical firms, is hoping to do just that.
Understand that the issue of sharing data is not trivial. Before agreeing to do it, some amount of in-depth discussion needs to occur regarding requirements, expectations and deliverables.
Gloria Metrick is the owner of GeoMetrick Enterprises,a provider of consulting services for laboratory informaticsprojects. She can be reached at firstname.lastname@example.org.