OR WAIT 15 SECS
The least we can do in our industry to ensure scientific integrity in our pursuit towards effective CAPA is to have our data peer reviewed.
The determination of corrective and preventive actions (CAPA) in biopharmaceutical manufacturing is fundamentally a subjective, experiential activity with a high potential of failure for resolving process issues, which carries far-reaching implications for our industry. CAPA determination can be more successful by understanding the key pitfalls and the application of good science.
As I write this, two Pioneer satellites (10 and 11, launched in 1972 and 1973, respectively), which are hurtling away from Earth in opposite directions, are not where they should be — by about 248,500 miles according to calculations of gravitation. It's back to the drawing board for those involved to determine a cause for the error in the distance calculation. The scientist tracking the two satellites and their distance error had not released his findings for 15 years because the data did not match the theoretical calculations. And maybe we shouldn't blame him because the offset was, after all, just one part in 300,000 (incidentally, this is running just beyond a Six Sigma level of 3.4 parts-per-million). But working with licensed biological products, we don't have the luxury of not submitting aberrant data, and I'd like to think that any of us would feel ethically bound to reveal all the data, for better or for worse. But there is a strong tendency to believe what we perceive to be true. Our perception of the world shapes the things we know, and conversely, what we know skews how we view the world.
As observers, we are also inherently and inextricably active participants in the world around us. Twentieth century physicist Niels Bohr noted that "the experimenter chooses what to observe in a given experiment." This statement has serious implications for solving issues that occur in our biopharmaceutical industry, where empirical data are considered essential for understanding and resolving process deviations and assigning CAPA. But aren't these data susceptible to being assigned different, subjective, levels of importance by us? When faced with process deviations, our understanding of these issues is necessarily tied to how we choose to observe them.
We live in a world of uncertainty, one in which the "path of an object first comes into existence when we observe it" (Heisenberg), and one in which the picture we get is determined by how we choose to observe it (cf. The Double Slit Experiment by Thomas Young). Though this sounds like metaphysics, it is patent truth.
It is essential, then, to consciously strive for objectivity whenever possible. CAPAs are intended to resolve a problem or prevent future recurrences of the same problem. But they are, of course, determined by an analysis of the problem, and that analysis is performed by someone who chooses to observe certain data (which simultaneously disregards other data). Many times when we try to solve a problem we form cause-and-effect hypotheses predicated on ideas that are familiar to us, and this biases our perception of the data we collect (for example, we may select an experiment to test for 'x' preferentially over an experiment to test for 'y'. Though this biased screening of variables may in some cases make sense, it introduces a subjectivity that affects the results of our experiments).
As mentioned, when beginning a large investigation (e.g., bioreactor failure, change in cell culture productivity, contamination), certain variables are selected while others are not – and this selection process comes down largely to intuition (experience) on the part of the investigators. We have, in our industry been a part of myriad process investigations, and many times this experiential intuition makes us want a certain answer to come out of the data, or think that a certain failure mode is at work before the data are there to support the conclusion – rather than letting the data alone guide us to the answer. Should this occur, the very real potential exists for a CAPA to be assigned that doesn't necessarily correct the real issue nor set up effective preventive measures against its recurrence. This can lead to a new CAPA being assigned to the same issue later because the initial assessment was insufficient, wasting both time and money. And should the original (insufficient) CAPA have been assigned to resolve a serious process issue, there is a high level of risk associated with continued operations. In the biopharmaceutical industry, this risk could manifest itself as: increased potential of production failure; an increase in process deviations generating more investigative work, creating an additional investment of time and money; or failure to comply with regulatory requirements, perhaps preventing the manufactured product's release. It is therefore imperative to determine CAPA based on good science and coherent reasoning, and treat this corrective or preventive action as an important process intervention, not merely a requirement for regulatory compliance.
The least we can do in our industry to ensure scientific integrity in our pursuit towards effective CAPA is to have our data peer reviewed. The peer review is part of the systematic approach we employ when analyzing a process issue and determining an appropriate CAPA. The importance of a peer review group cannot be underestimated — it acts, literally, to fulfill the verification aspect of the scientific method. Working in most cases against impressive time pressures, peer review quickly builds or refutes the case for a CAPA's effectiveness by subjecting it to additional rigor which forms the basis of good science – striving towards an unassailable deductive hypothesis for an observed set of data. The CAPA can then be implemented with a very high likelihood for success as a corrective or preventive measure.
Benjamin Locwin is a Process Technologist for Lonza Biologics, Inc., 101 International Drive, Portsmouth, NH 03801, 603.610.4682, firstname.lastname@example.org.
1. Gribbin, John, In Search of SchrÖger's Cat, (Copyright 1984 Bantam Books).
2. Heisenberg, Werner, Physical Principles of the Quantum Theory, (Copyright 1949 Dover Publications).
3. Sloan, M. Daniel and Boyles, Russell A, Profit Signals: How Evidence Based Decisions Power Six Sigma Breakthroughs, (Copyright 2003 Evidence-Based Decisions Inc.).