The Process Flow Model
Figure 2. A process flow model for testing laboratories
|
Understanding how a laboratory's workflow takes place, where data comes from, how information is processed and stored, and
how data is entered into the LIMS, will give you a basis for deciding how the results of testing from an outsourced testing
services can be integrated into your laboratory's systems. Figure 2 shows the workflow in a testing laboratory along with
the entry points for test results and new analytical protocols. The ovals represent databases and the arrows are processes
used to carry out a stage in the test protocol. Starting at the top, a document management system contains test method protocols
and supporting documents (hazardous materials, etc.). Test methods are implemented in the measurements and experiments process
and would include sample preparation and chromatographic work in Figure 1. All of the steps in Figure 1's chromatographic
data station are shown in the grey box in Figure 2 as the detector signal is measured and stored in the instrument data set.
The data stations analysis software processes the chromatographic data and the result is entered into the LIMS. Data and results
from external sources (ES) can enter the system in three places, allowing outsourced laboratory work and method development
to be integrated into the laboratory's operations.
A comparison of Figure 1 and Figure 2 shows:
1. Analysis protocols would be contained in the knowledge database.
2. The sample preparation, chromatographic separation, and data acquisition steps are part of the measurements and experiments
process.
3. The acquired data are stored in the data station's database structure, which would also contain any processed data (peak
retention time, size, etc.).
4. The results of the assay are entered into the LIMS.
Other analytical protocols would have their own measurement and experiments processes. Each instrument type would have its
own database structure, so that the laboratory's overall instrument data system would comprise several databases, one for
each type of instrument.
This chart represents only a basic model workflow. Because the specific steps and flow at each laboratory may vary, it is
important for the two laboratories to resolve ahead of time any differences that could pose a problem.
Outsourcing Method Development
The considerations for outsourcing method development work are similar to what we've just discussed. All of the validation
and regulatory issues apply to the outsourced company just as they would if the work were done in your own facility. In addition,
the following need to be kept in mind:
-
Who owns the test method? Don't assume that simply because you paid for the work you have complete control over the rights to work. Ownership should
be defined in the contract.
-
Incorporating the method into your workflow will require training. If those developing the method use equipment that is similar to what you have in the laboratory, but from a different vendor,
determine if any systematic differences can affect the results. Does a particular instrument have a difference in sensitivity
or control that is critical to the successful implementation of the procedure?
-
Consider how to integrate the testing protocol. Are there any problems integrating the new protocol into your laboratory's workflow, including updates to the LIMS, for keeping
track of tests?
-
Hardware and Software: If the new method required the development or modifications of hardware or software, are the rights transferable to you?
Is the actual equipment yours or do you have to purchase a duplicate set and re-implement the modifications? Does the company
that developed the modifications have plans to convert them into a product? What considerations does your company get for
its investment?
Outsourcing Systems Development
When outsourcing systems development, it is essential to know whether the outsourced project is a completely new development
or a modification to an existing software. A successful automation project should include the following:
|
