2

Investigations, root cause analyses and CAPAs

Abstract:

This chapter considers the organizational response to observations, particularly those occurring in manufacturing and quality lab areas. Observations are either triaged or they are escalated. A number of observations that will likely result in escalation are listed, as well as a number that will likely be triaged because they are non-GMP issues. Next, this chapter addresses the process of investigation and root cause analysis (RCA). Finally, it discusses the corrective action and preventive action (CAPA) that are the outcome of the investigation.

Key words

corrective action

deviation investigation

escalation

human error

Ishikawa diagram

Notice of Event

preventive action

program logic model

remediation

root cause

triage

2.1 Triage and escalation

Resources are limited - resources such as staff who are experienced in conducting investigations, resources such as time to conduct investigations. It is necessary to ensure that those scarce resources are being expended on critical issues. A non-GMP issue such as a burned out ceiling light bulb is usually less critical than a GMP issue such as failure to follow a standard operating procedure (SOP). Triage is the process of ranking events or things in terms of importance or priority; resources are allocated accordingly. For our purposes, triage is binary - some observations will be set aside, while others will be escalated further.

In the manufacturing and lab areas, an observation will typically result in a Notice of Event (NoE), whereby employees notify management that something has been observed, an exceptional event has occurred. Should the observation be triaged, it will be set aside, it will remain at the level of "NoE Only." However, if the observation and associated NoE is escalated, then an investigation may occur.

The decision to escalate is not a trivial one. As Margaret Hambleton has expressed it:

A formal, well-documented RCA should be conducted when you are investigating a significant compliance failure. It should be used in those situations where reoccurrence would have a significant impact. You typically would not use an RCA for simple mistakes, inadvertent slips, or one-time events. You would consider using an RCA when those simple mistakes turn into a trend.1

Table 2.1 shows some of the kinds of observations in the manufacturing area that will likely be escalated.

Table 2.1

Illustrative production observations that will be escalated

1. Deviation from an SOP 2. Foreign matter 3. Processing issues that result in segregating and destruction of product 4. Non-adherence to specified time limits for a processing step 5. Temperature, humidity, and/or pressure excursions during batch processing 6. Product attribute defect issues 7. Raw materials found damaged during inspection 8. Contaminated product 9. Out-of-calibration results for GxP instruments 10. Use of non-calibrated GMP equipment 11. Use of non-qualified GMP equipment 12. Missing data 13. Rejected materials not properly identified and controlled 14. Use of an incorrect form to record GMP activity

Table 2.2 shows some of the kinds of observations in the quality labs that will likely be escalated.

Table 2.2

Illustrative lab observations that will be escalated

1. Deviation from an SOP 2. Out-of-Spec (OOS) test results 3. Microbiological cleaning failures 4. Incorrectly labeled samples 5. Error in stability samples for testing 6. Identification of an adverse stability trend 7. Use of non-calibrated GMP equipment 8. Use of non-qualified GMP equipment 9. Missing data 10. Use of an incorrect form to record GMP activity

There are a number of observations that will most likely be triaged, typically because they are not GMP issues, as displayed in Table 2.3.

Table 2.3

Illustrative observations that will be triaged

1. Minor documentation errors that are addressed in the record 2. Typographical errors (e.g. misspelled words, grammatical errors, punctuation errors) 3. Missed periodic maintenance, out of calibration status of non-GMP equipment 4. Equipment challenge failures that occur during set-up 5. Shutdown of the redundant utility system 6. Utility shutdown that does not disrupt operations 7. Routine service calls 8. Materials spilled during weighing, measuring, or dispensing operations

In any case, the observations (events) that are escalated will lead to investigations and root cause analyses.

This section has discussed the kinds of observations that are likely to be escalated, and those that are likely to be triaged. The next section will consider the process of investigation that follows from an escalation.

2.2 The process of investigation

Faced with a problem (call it P) that is referenced in a NoE - a manufacturing deviation or out-of-spec lab result, say - the organization will conduct an investigation to find the root cause.

Management assigns an employee to lead the investigation, typically a subject matter expert (SME), and establishes the due date for the completion of the investigation report. The lead investigator will review the NoE and also review related observations (events), trend data, training records, etc. Of particular interest is determining whether this deviation is a recurrence, which means that earlier remediations have failed. In such a case, the prior corrective action and preventive action (CAPA) must be reconsidered to determine why it failed.

It is usually recommended that a team be assembled by the project lead, since this permits triangulation in assessing the deviation and its causes. This also improves the likelihood that the effects of personal biases in assigning causes will be lessened.

Many times the investigation of a deviation involves a number of departments. This requires a cross-functional investigation involving employees from various departments to assess the deviation. The participants should consist of SMEs from departments that have been impacted by the deviation, as well as personnel from Quality Assurance (QA).

The investigation will identify a number of elements of the manufacturing or quality lab system, call them En, where the variation in Ei causes variation (deviation) in P. Consider the following Ishikawa diagram.2 Six major elements are included (Figure 2.1).

Figure 2.1 Template of Ishikawa diagram

Elements that are identified as part of En - that is, potential causes of the deviation - can be considered as elements of the FDA Quality Systems Approach, including equipment, production, quality, materials, laboratories, packaging, etc.3 Consider the Ishikawa diagram displayed in Figure 2.2. Each of these elements has its own constituents. The investigation proceeds by eliminating the various elements and constituents of the system that might have been the cause of the deviation. Each of the elements and constituents is reviewed in turn (Figures 2.2 and 2.3).

Figure 2.2 All candidates for root cause

Figure 2.3 Disaggregation of a major element

The disaggregation of a major element, say equipment, into its constituents is illustrated in Figure 2.3. For example, what is the ease of operation of this piece of equipment? Was the equipment maintained (or not)? Was it calibrated? Was this piece of equipment appropriate to the tasks? Were there adequate units of equipment?

By a process of elimination, elements and constituents are considered and eliminated once it is determined...