Академический Документы
Профессиональный Документы
Культура Документы
Even in organizations with successful quality programs, minimizing human error is an ongoing challenge. Experienced people working on established processes sometimes make costly errors in spite of the most diligent efforts to avoid them. Human error is often the primary contributor to failures or adverse events in the pharmaceutical industry, leading to quality problems that require expensive solutions. Despite the industrys awareness of human errors, companies still frequently fail to substantively and correctly address errors. The typical response to a human error is retraining. But studies have now shown that training or lack thereof is responsible for only about 10 percent of the human errors that occur. While training typically is effective when there is a need for a new skill set or the implementation of a new task, it is not an effective response to a human error that is not caused by a lack of knowledge. So, if human error is the cause but training is not the reason behind it, then what is? This FDAnews White Paper examines how organizations can get to the bottom of events and determine what is behind the errors being committed by operators.
Page 2
Control, from Page 1 human error investigation must be undertaken to comply with the FDA regulations. Thus, once the root cause analysis investigation is complete, a new investigation starts the human error investigation. Human error investigators are people who have specific skills and abilities designed to understand human behavior and understand conditions that could impact behavior in a negative way, leading to human error. In a typical investigation process, the organization gathers and analyzes data, determines the causal factors behind an event, determines the root cause(s), and categorizes and codes those causes. Next, corrective and preventive actions are designed and developed, followed by monitoring and trending of the root causes. Based on the data, the effectiveness of the corrective and preventive actions (CAPA) can be measured, allowing the organization to predict if it has a problem with one of its systems.
For example, an organization discovers it is experiencing human errors related to one of its procedures. The human error investigators identify that the procedure was not being used because it did not exist. Then, if they determine that most of the human error-related events are related to that root cause, the organization can predict that at some point, if it doesnt create a procedure, a mistake will once again be made. Five Whys Using the Five Whys is a simple but powerful technique to help get past the symptoms of a problem and find its root causes. Simply ask the question why several times. For example: 1. Why did the employee make the mistake? He failed to follow the procedure. 2. Why did he fail to follow the procedure? Because there was no procedure. 3. Why was there no procedure? 4. And so on
Page 3
CAPA, from Page 2 The sources of information will vary, but at some point all of the data with the results of the investigation must flow into CAPA. Then, recommended and approved actions can be taken, and implementation and verification of those actions can be documented and tracked. Information for CAPA comes from internal and external feedback sources. Sometimes what is internal and what is external is an arbitrary division. The main thing is to make sure that nothing is missed. Internal sources include:
slightly different way than a normal complaint or problem would be. Procedures must be written to assure that the data captured is complete, accurate and timely. If they are not, investigations will not be complete and any corrective or preventive action could be unnecessarily delayed, making the problem worse. The purpose for bringing the data together is to assist with data trending to detect existing or potential nonconformities. This should be done within or across data sources. For example, comparing electronic repair logs and the pieces of equipment that have been serviced or recalibrated in a given time frame to the in-process test results of product made in those pieces of equipment might show a potential situation for future failure. When Equipment Y, for example, is within 30 days of recalibration, in-process batches of drug product A are usually testing in the lower limits of the range. Although there has not been a failure in the lower limits of the range, it might be wise to change the calibration schedule for Equipment Y to prevent a situation where the calibration might affect the product if it is done late and might make it drop outside the range of acceptable product. This is a preventive action. Also documented in the CAPA system should be actual verification or validation of the actions that were implemented. There must be a mechanism for communicating the now known quality problem or nonconforming product to those responsible for assuring the quality of the product and/or the prevention of such problems. There must be communication capabilities to feed the information to those who can most effectively take advantage of it to prevent future problems, to take corrective action or even to make corrections in some cases. Again, actions must be documented. After the actions have been taken, implemented and verified or validated as effective, management (See CAPA, Page 4)
Inspection test data, in-process data, final inspection test information, quality assurance unit inspection findings, scrap and yield rates and process control data. Internal feedback sources can be anything that is routinely tested as well as environmental control data and data regarding scraps and defect rates. Incoming inspection data collected relative to the part number, a supplier, a batch or a lot number. Although they might be relative to products that come in from the outside, these items also are parts of the internal data sources. Internal audits, equipment maintenance and calibration data also can be valuable in understanding processes. Various reports and records that are generated during manufacturing, packaging, labeling and employee training.
Complaints. The FDA gets complaints from consumers and employees through various systems like MedWatch, medical device reports, drug adverse event reports, and field service reports relative to products, product returns and journal articles. Legal claims, warranty issues and claims, regulatory audits and client audits.
Companies sometimes forget legal claims and warranty claims. These are handled in a
Page 4
CAPA, from Page 3 should again be informed. This is usually done by trending reports during management reviews. It is essential that management is kept informed. The FDA looks favorably on organizations with CAPA programs and outlines its expectations for CAPA in its regulations and guidance documents. One theme throughout the regulations is the need to document. Companies must put into writing actions and decisions including problem identification, the investigations, decisions regarding the problem and solutions, and the actions taken to correct the problem and prevent it from happening again.
An
Composing a CCDS (Company Core Data Sheet) could be the most demanding writing exercise youll ever encounter. This all-new guide spells out how to draft and maintain a CCDS for each of your products a start-to-finish explanation that includes:
Publication
Elements of a CCDS Formatting the CCDS to satisfy international audiences The approval mechanism for a new and a revised CCDS And much more
Price: $377
Page 5
What went wrong? What were the consequences of the incident(s)? What could cause these events? What changes should be made to the proactive analysis process and the management systems to adequately control risk in the future?
the way back to the root cause or merely breaking the chain of events so the problem cannot recur is a decision manufacturers must make on each occasion. At first blush, an adverse event may look as though it was caused by human failure, but investigations frequently reveal that a much deeper issue is to blame. The recommended levels of an investigative analysis, starting at the uppermost level and digging down to the lowest, are:
The RCA has built into it all of the reasons why systems and people fail. Therefore, it doesnt have to be reinvented each time it is used. It identifies the following human error categories:
Procedure-related errors; Human factor engineering related error; Training-related error; Error reduction; Supervision-related error; Communications-related error; and Individual error.
Human error or equipment failure Task control issues Process control issues Management system issues Organizational cultural issues
Neither training nor the individual has been excluded from this list, but it should be noted that they are just two of the six categories. Each of the categories must be further investigated because procedure-related error, for example, does not fully explain what happened or why. Using the root cause chart, the main goal is to have:
If only the top levels of an event (human error or equipment failure) are investigated, only one problem will be solved. The deeper one digs into the event, the more the investigation will learn about organizational problems. Consider the following example: A fuse blows and causes the lights to go off. A manufacturer can identify the immediate causal factor fuse blew and replace the fuse. It can also identify another cause a short and repair the wire that shorted. If it doesnt identify and correct the factor that led to the fuse blowing, however, (e.g., wiring not maintained because there was insufficient maintenance), other systems may experience similar failures. RCA seeks to identify these deeper systemic problems, such as lack of an adequate maintenance budget, and correct them. Unfortunately, some RCA methods fail to dig far enough. Often, organizations will get down to the management systems level and stop at that point primarily because corrective actions and recommendations must be actionable and measurable, and it may be difficult to do that at the organizational cultural level.
Good and accurate administrative management systems that would assure that there are controls for providing clear, accurate procedures, instructions and other aids; Good human factors engineering of control systems, processes and work environment; Job-relevant training and practice; Appropriate supervision; Good communications; and Fit for duty personnel performance.
The primary goals when investigating an incident are to discover what caused it and why with the eventual goal of making sure it does not happen again. Whether that means going all
Page 6
Address the causal factor; Address the specific problem; Fix any similar problems; and Correct the process that created the problems.
If the causal factor is not addressed, the problem is not solved. Most organizations address this level. If the specific problem is not addressed, the problem recurs. This level is sometimes ignored. If similar problems are not fixed, these problems will recur as existing issues continue to surface. If the process that created the problems is not corrected, new problems will occur, even if all of the existing problems are solved. Most organizations fail to address this level without an RCA program. The root causes of an event should be directly connected with associated corrective actions (see story, page 2). If upper management does not see a connection between the cause of the event and the recommendations from an investigation, it is unlikely to want to implement them, and failing to correct problems will come at a high cost to an organization.
Publication
Price: $397
Its certain that FDAs inspectors will examine your complaint management reporting systems on their next visit. Now is the time to start whipping your systems into shape. Dont delay. Order Medical Device Complaint Management: A Guide for Compliance today.
Page 7
Punishment, from Page 6 On the other hand, rewards and discouragements must be a part of a companys normal performance review process. Some industry experts believe that there should be a change in the way organizations view human error. The current way is to identify an employees inaccurate assessments, wrong decisions and bad judgment, assessing blame to the employee. A new vision calls for a new approach to human error as a symptom of a deeper problem inside a system. Human error should be less about the person and more about why that person made the error. Manufacturers should identify how people assess tasks and decisions that make sense at the time given the circumstances that surrounded them, and they will likely find out the real reason for the human error. This can be done with RCA.
For example, complaints often constitute an area that should be regularly addressed, even if some of those are not particularly important, because they deal directly with patients a high-risk area. Statistical Methodologies It is important to use tried and true statistical techniques when evaluating products and assessing risk. Some of the statistical methodologies include Pareto charts, run charts and control charts. Pareto charts summarize and display the relative importance of the differences between groups of data. Control charts focus on the variation in a process. Run charts, which are similar to control charts, focus more on time patterns, displaying process performance over time. The FDA and the International Organization for Standardization (ISO) have several regulations regarding trending, including the following:
21 CFR Part 820.100(a)(1): Analyzing processes, work operations, concessions, quality audit reports, quality records, service records, complaints, returned product, and other sources of quality data to identify existing and potential causes of nonconforming product, and or other quality problems This regulation provides an extensive list of what should be trended. For pharmaceuticals, although the word trending is not explicitly used, as it is in the medical device area, the FDA considers it to be required by this portion of the regulations:
21 CFR Part 211.180(e): Written records required by this part shall be maintained so that data therein can be (See Trending, Page 8)
Page 8
March 2011
Trending, from Page 7 used for evaluating, at least annually, the quality standards of each drug product to determine the need for changes in drug product specifications or manufacturing or control procedures. Trending should be done as often as necessary. For most products and processes, an annual review of trends is not enough, according to experts. Most companies do a good job of trending complaints and monitoring results, but many of the other corrective and preventive actions sources are not trended often enough or are entirely missed.
shall include determination of applicable methods, including statistical techniques, and the extent of their use. Reviewing Trends Management must review all data trends. Trends should be reported so that management hears first about those trends that are occurring most frequently, as well as those that pose the greatest risk to the company, the products and the processes, but first and foremost, to the patients. Management review should be conducted at a level of frequency based on the kinds of problems being detected. As time goes by and the problems are better controlled and the level of risk falls, the reviews may be conducted less often. In other words, manufacturers should look at the array of products being produced, and as products age and new products are introduced, consider the correct frequency. Management must assure that each corrective or preventive action that is taken is implemented and verified or validated. The process should include identifying problems, recommending solutions and verifying that those solutions are successfully fixing the problems. All of these procedures must be properly documented for internal recording purposes and in the event of an FDA inspection.
ISO 9001:2000 8.4: The organization shall determine, collect and analyze appropriate data to demonstrate the suitability and effectiveness of the quality management system and to evaluate where continual improvement of the effectiveness of the quality management system can be made. This shall include data generated as a result of monitoring and measurement and from other relevant sources characteristics and trends of processes and products ISO 9001: 2000 8.1: In regards to measurement, analysis and improvement, this requirement states, The organization shall plan and implement the monitoring, measurement, analysis and improvement processes needed
300 N. Washington St., Suite 200 Falls Church, VA 22046-3431 Phone: (888) 838-5578 +1 (703) 538-7600 Fax: +1 (703) 538-7676 www.fdanews.com
Reporters: David Belian, Virgil Dickson, Wilson Peden, Molly Cohen, Kevin ORourke President: Cynthia Carter; Editorial Director: Pamela Taulbee; Executive Editor: April Hollis
Copyright 2011 by Washington Business Information Inc. All rights reserved. The GMP Letter (ISSN 0196-626X), an executive briefing on FDA enforcement and quality systems requirements for medical devices and diagnostics, is published monthly, 12 issues, for $945. Photocopying or reproducing in any form, including electronic or facsimile transmission, scanning or electronic storage is a violation of federal copyright law and is strictly prohibited without the publishers express written permission. Subscribers registered with the Copyright Clearance Center (CCC) may reproduce articles for internal use only. For more information, contact CCC at www.copyright.com or call (978) 750-8400. For site licenses for multiple users or to purchase multiple copies, contact Andrew McSherry at (703) 538-7642.