Getting to the Root of Quality Problems

The philosopher George Santayana once wrote, “Those who fail to learn from the past are condemned to repeat it.” This is clearly seen in the way that some pharmaceutical manufacturers approach root cause analysis (RCA) and corrective action and preventive action (CAPA).

Both RCA and CAPA are closely intertwined. For example, tracking and trending complaints allows companies to identify recurring problems that may not be caught during inspections on the manufacturing floor, says Kim Jackson, product manager at MasterControl, a vendor of quality management and CAPA software. It also allows them to determine the true severity of specific problems or failure modes, she says. Furthermore, CAPA effectiveness checks ensure that RCA investigations are sufficiently robust, says pharma quality consultant Ajay Pazhayattil.

Both concepts are crucial for establishing continuous improvement and a true culture of quality, and achieving the goals established by International Council for Harmonization (ICH) Q 12. “Efforts always pay off because they prevent supply disruptions and resulting revenue loss,” says Pazhayattil. Addressing them incorrectly, however, can lead to product quality failures as well as noncompliance with current good manufacturing practices (CGMPs). In some cases, particularly for over-the-counter (OTC) drugs, the problems highlighted in warning letters had already been pointed out previously in FDA Form 483s (1), suggesting a need for more senior management involvement and greater investment in CGMP compliance. Inspectors pointed to inadequate validation and conformance to written procedures, as well as deficient RCA and CAPA. In some cases, root causes for batch or product testing failures were either insufficiently investigated or not probed at all (2,3).

Understanding the cost of poor quality

Experts see a number of reasons for this situation, including the need for a more rigorous approach to risk assessment and a better understanding of the cost of poor quality. “If we as an industry could step up and own the cost of poor quality, if we could measure it, we’d be horrified at how much we pay, not just for failing to solve issues but for repeatedly solving the same ones,” says Nuala Calnan, principal of the consulting firm BioPharm Excel and professor at the Technical University of Dublin. “The number of personhours and CAPAs that arise from these investigations is eyewatering. We don’t add them all up and have a trackable number (i.e., ‘that’s how much it cost us not to get to root cause and to keep on failing’),” she says.

But there has also been a lack of good RCA training at many pharmaceutical manufacturers. In October 2019, the International Society for Pharmaceutical Engineers (ISPE) and the Parenteral Drug Association (PDA) launched their joint Quality Culture guidance series with a module devoted to best practices for RCA (4), in an attempt to address this problem. Its goal is to help shift the industry’s focus from compliance to prevention. “Pharma companies typically cover fundamentals (e.g., how to use tools such as Five Whys and Ishikawa fishbone diagrams), but teaching people how to use tools and templates isn’t training them to look at the underlying science behind RCA decision making and why it is so important,” says Calnan, who is also coleader of ISPE’s quality culture initiative and one of the authors of the new RCA module.

Better RCA training needed

Advanced training in the techniques and critical thinking required for investigators, and application of proven RCA tools such as mapping, brainstorming, cause-and-effect, and Five Whys are crucial to improving the state of RCA in the industry, says Pazhayattil. “It’s important not to get stuck on using the same methodology each time, since different problems will call for different solutions,” suggests Marzena Ingram, a senior pharma quality manager who comments as an industry professional rather than on behalf of her company.

Impeding a better approach to RCA at many companies is an ingrained focus on regulatory compliance and being inspection-ready. “Many quality departments must focus on feeding the compliance system rather than considering quality as a broader responsibility. Often, we’re being driven to close out investigations within 30 days because compliance metrics tell us we need to do that, so we’re not addressing root cause,” says Calnan. In some cases, companies may over-respond to smaller quality problems by launching too many full-scale RCA investigations, says Jackson. “When a company goes into full alert, all-hands-on-deck mode for every issue that arises, not only does time get wasted, but employees become jaded and less likely to do due diligence when a real issue presents itself,” she says. “Taking a risk-based approach to escalations and root cause investigations saves efforts for issues that truly pose a risk to patient health and safety,” she says. Jackson suggests that each quality event be investigated first at a lower level to determine the most likely cause, and then evaluated for risk, says Jackson.

Another fatal flaw in many RCA programs is that the approaches typically address only part of the problem, the symptoms of the problem, or its direct cause, rather than the fundamental reason why it happened. For example, operator error, which is often cited as the reason for quality problems, usually shows that the system of controls in place for the process and product has failed, rather than any single individual, Calnan says.  In other cases, the reason behind the direct cause for one batch’s failure (e.g., a product or labeling mix-up) may be solved without considering the series of events that caused the mix-up to occur in the first place.  In all these cases, companies wind up cutting and pasting the same solutions to each new problem, whether or not they have worked in the past. “Because the root cause hasn’t been addressed, the problems will recur,” Calnan says.

Rooting out bias

Success with RCA requires rooting out any potential sources of bias, says Pazhayattil. “It is human nature to pre-empt causes of failures, but theoretical assumptions should never bias an investigator,” he says. For example, an investigator may unconsciously or consciously, focus on the area that he or she is more familiar with (e.g., process engineering or analytical methods). “Generating sound supporting proof is critical to confirming root causes and developing a science- and data-driven investigation method,” he says.

“Coming into an investigation with a biased opinion, even if you are fairly sure you’re right, is a surefire way to miss the opportunity to investigate with an open mind,” says Ingram. “No ideas are bad ideas, so it’s best for teams to brainstorm, throw all possibilities up on the board and then rule out, systematically and with proper justification,” she says.

Devil’s advocates needed

Calnan also sees an overemphasis on consensus-building as impeding the effectiveness of pharma risk management and RCA. With group efforts such as failure modes and effects analysis (FMEA), which require teams to agree on a numerical value to assign to each specific risk, there may be a failure to invite diversity of opinion or different perspectives.

“Consensus [can become] the enemy of good critical thinking and risk management,” she says. When teams are analyzing a quality problem, Calnan suggests that one or more members play devil’s advocate and ask for data or other evidence to support any hypotheses. “In some cases, companies may be pushing for agreement way too soon, before they’ve spent enough time or applied any real rigor to identifying the real issues that are causing the problem,” she says.

“There is very little in pharma’s processes that gets us back out there, onto the lab or manufacturing floor, to shake down the cursory one liner that was given as the reason for the problem. As a result, companies often solve for the wrong problem. When they haven’t even identified the right problem, how can they get to its root cause?” Calnan asks.

Another obstacle to improvement is the fact that the industry doesn’t typically view failures as an opportunity for learning, says Calnan. “All too often, we have to classify failures, include them in an investigation report, and apply a CAPA to them to move them off our desks so that we can get to the next round of firefighting,” she says.

Trending to pinpoint sources of variability

Trending should be done regularly, as suggested by FDA’s revised process validation guidance, to find where sources of variation and potential risk and failure are developing, she says. Market complaints, stability data trends, and multivariate analysis of critical process parameters and critical quality attributes are all sources of data for continuous quality improvement, says Pazhayattil.

In addition to evaluating risk for individual events, monitoring should be put in place to help identify recurring trends that should be investigated, says Richards. If an issue is recurring, or multiple events point to the same recurring cause, the recurrence can be investigated, and a risk assessment performed on the trend. Based on that, a company could continue to monitor, setting a threshold of acceptance, or initiate a full root cause investigation and CAPA, she says.

When a full root cause investigation is warranted, having a standard methodology can help streamline the analysis, she says. For example, the use of Five Whys provides structure to the investigation and can help contain its scope while ensuring thoroughness.

Staffing cross-functional teams offers an opportunity to ensure that the right skill sets are available for RCA. “I am seeing more such teams in action today, but many of them still take an ‘us vs. them’ approach,” says Calnan. This often comes out most clearly in the interpretation of such phrases as ‘the quality department is responsible for closing out the investigation.’  While not meant as an excuse to pass the buck, this phrase often results in quality executives having to develop solutions without enough input and insight from the team, including insights from those much closer to the problem,” says Calnan.

“The belief that the responsibility to assess and address root cause is ‘someone else’s problem’ is a typical pitfall in pharma RCA programs,” says Ingram. “All departments, even those that may not have direct involvement, should play a role in identifying, analyzing, and effectively solving the problem,” she says.

Training and knowledge management can also be an obstacle in some companies, where subject matter experts (SMEs) are almost empowered not to share knowledge. “SMEs need to see themselves as knowledge stewards and trainers for the next generation, and essential to creating a learning organization,” says Calnan.

Need for leading metrics

Beyond training and procedures, pharma’s quality metrics themselves must change if RCA and quality programs are to improve, Calnan says.  She points to a need for leading (rather than lagging) metrics that are aligned with overall patient and business priorities. One example, she says, would be measuring the ratio of preventive actions to corrective actions within CAPA systems, and setting a target to move performance to the next level.

In addition, Calnan says, recurring deviations need to be correctly coded and accounted for. Currently, most companies don’t code errors in a transparent way that would make it easy to see when they recur. “They’re looking for the exact same error to happen on the exact same line or process, rather than asking whether there are common root causes and coding those causes appropriately,” she says. For example, an organization can have one problem in a lab and another on the manufacturing floor that may seem very different yet share the same root causes.

A number of technologies are available that promise to improve the way pharma handles RCA and to make the process easier. More powerful data analytics approaches are already being used to allow data to be drawn from disparate databases to be used for risk analysis and RCA, says Calnan. “Technology can also be leveraged to gather failure modes and occurrence data to better inform risk assessments, preventing unnecessary RCA activities when a simple mitigation would suffice,” says Jackson.

Machine learning and artificial intelligence (AI) tools (e.g., predictive modeling for equipment maintenance, manufacturing process control, and real-time release) are also emerging, says Pazhayattil, who is currently working on research into AI in pharmaceutical manufacturing with CalSouthern. However, it will be a long time before these technologies are found routinely on every pharmaceutical manufacturing plant floor, says Calnan. 

Although modern equipment (e.g., filling equipment that incorporates automation and isolators) can help improve overall operations, Calnan believes that it is not essential to preventing failures. In addition, she says, doing RCA correctly needn’t be expensive. “As an industry, we need to get on with real training. People need to understand the science behind failure, and to understand the differences as well as the connections between risk and failure,” she says.

References

1. ECA GMP, “A Look at FDA’s Warning Letters Over the Last Months,” gmpcompliance.org, Nov. 1, 2019.
2. FDA, “FDA Warning Letter to Torrent Pharmaceuticals,” fda.gov, Oct. 8, 2019.
3. FDA, “FDA Warning Letter to Lupin Ltd.,” fda.gov, Sept. 10, 2019.
4. ISPE, “ISPE and PDA Guide to Improving Quality Culture, Module 1: Root Cause Analysis,” ISPE.org, Oct. 7, 2019.   

Article Details

BioPharm International
Vol. 33, No. 1
January 2020
Pages: 39-41

Citation

When referring to this article, please cite it as A. Shanley, “Getting to the Root of Quality Problems” BioPharm International 33 (1) 2020.