Trend Analysis in the Pharmaceutical Industry

 Trend Analysis in the Pharmaceutical Industry

By Gilberto Dalmaso, PhD

A lost opportunity in the pharmaceutical industry is most companies do not use data collected from their environmental monitoring systems to perform maximally effective trend analysis. Over 100,000 data points can be generated on an annual basis from a single sampling point of an environmental monitoring system. To find alert and action levels for an entire process, collective data from multiple sampling locations are graphically represented in order to find median and maximal values. Once these levels have been set, however, the large majority of the subsequently collected data is not analyzed in any capacity. As a result, only data causing alerts or actions will have any significance to the reviewer.

When is Alert and Action Level Data Collection Analysis Ineffective?

Typically, monthly alert and action level data collections are compared against each other and annual comparisons are made against previous years, with no impact to current levels. This form of thinking and implementation of data trends, however, is grossly ineffective. Instead, the entire collection of raw data should be reviewed for deviations, and subsequently compared to previous collections in order to fully understand how the process has changed in terms of particle and microbiological contamination.

The graphs below show the data that was collected from three sensors running at 20 lpm at different points of an operating filling line (depyrogenation tunnel, filling point and stopper station).

depyrogenation tunnel, filling point and stopper station

The alert level is 1500 particles for > 0.5 micron and 10 particles for > 5 micron. The action level is 3520 particles for > 0.5 micron and 29 particles for > 5 micron. Comparing data from January and April, an increase in particle counts and frequency in high counts is found in April’s data for both particle sizes.

Conclusive Trend Analysis

When a significant difference between data collections is observed, a root cause analysis should be performed, and a method of improvement attained.

If the trend analysis shows a change in statistical profile (with a known cause, e.g. improved operating procedures and monitoring methods) then alert and action levels must adjust accordingly. Trend analysis should be part of the annual production review, and provide a complete picture that demonstrates the process is under control. The applicable tools and software for statistical profile generation and analysis are widely available in the Pharma and Biotech industry. Continual trend analysis can have a wide-ranging positive effect over production, including personnel activity, cleaning techniques and written procedures.

About the author…
Gilberto Dalmaso, PhD has over 25 years’ experience in pharmaceutical microbiology and sterility assurance, primarily with GlaxoSmithKline (GSK). In 2003, his laboratory gained the distinction of obtaining the world’s first rapid microbial PAT approval from the US FDA. Today Gilberto is the Global Aseptic Processes Development Manager for Particle Measuring Systems, is a reporter to numerous symposia on the microbiology and Pharma in Europe and United States, and is an ISO 9001 and HACCP quality system auditor.

  • <<
  • >>

Comments