Friedman (2011) USFDA presented at the Pharmaceutical Quality System (ICH Q10) Conference October/November 2011.  On slide 21, Chowdhury (2001) as cited by Friedman provides a comparison of common day events when operating at 3.8 Sigma meaning that you are getting it right 99% of the time.  This sounds really good—BUT  Chowdhury (2001) identified that getting it right 99% of the time still leaves a 1% error rate and this 1% error is the equivalent of the postal service losing 20,000 articles of mail every hour, or 5,000 botched operations each week, or 4 accidents each day at a major airport.

99% right equates to operating at 3.8 Sigma, but most companies seek to operate at a level of Six Sigma.  Gygi, Williams, and DeCarlo (2012) provided a real life comparison of these two operating conditions.

Attribute99% Good (3.8 Sigma)99.99966% Good (6 Sigma)
Lost mail20K per hour7 per hour
Botched Operations5K per week1.7 per week
Incorrect Prescriptions200,000 per year68 per year
Incorrectly Traded NYSE Shares11.8M per day4,021 per day

How many FDA warning letters have you read where the common response from the company for the underlying root cause was human error and the correction is a new procedure, more training or ultimately more checks on items that were already subjected to rigorous inspection!  However, not only do humans make errors when performing their jobs, in most cases there are humans that are responsible for receipt, in-process and final product acceptance.  Juran and Godfrey (1999) summarized human inspection as achieving an approximate 80% accuracy in finding deficiencies; ultimately, the inspectors may miss 20% of defects!

Friedman (2011) identified that human error is a cause of substantial variation across our industry but that this can be prevented by analyzing process for failure modes and increasing automation.  For example taking the human element out of the equation. Migliaccio, Ricciardi, Scott, and Winskill (2010), as cited by Friedman identifies that Human Error Analysis – HE training allows for “deeper insights into the underlying cases of human error in order to identify and avoid its sources”.

Do you analyze for human errors in your risk assessments?  Do you include error proofing such as Poka-Yoke to reduce or minimize human variation?  As described by Friedman (2011) getting it right 99% of the time is not good enough.  Even at Six Sigma with a defect rate of 3.4 defects per million, is this good enough when we are manufacturing hundreds of millions of doses?  What steps are you taking to improve your manufacturing practices and ultimately improving product quality?


Chowdhury, S. (2001). The power of six sigma. Chicago, IL: Dearborn Trade.
Friedman. (2011). Pharmaceutical quality systems: US perspective. Paper presented at the Pharmaceutical Quality (ICH Q10) Conference, Arlington, VA.
Gygi, C., Williams, B., & DeCarlo, N. (2012). Six sigma for dummies (2nd ed.). Hoboken, NJ: John Wiley & Sons.
Juran, J. M., & Godfrey, A. B. (1999). Juran’s quality handbook (5th ed.). London: McGraw-Hill.
Migliaccio, G., Ricciardi, N., Scott, C., & Winskill, N. (2010). Building a continuous improvement culture: Pfizer moves beyond “right first time”. In T. Friedli, P. K. Basu, T. Gronauer & J. Werani (Eds.), The Pathway to Operational Excellence. Germany: Editio Cantor Verlag Aulendorf.

Previous Paper Based Data Integrity
Next Intended Use