偏差调查中常见8大问题

By James Blackwell, Ph.D., The Windshire Group, LLC

Deviationinvestigations are one of the most important quality activities in any GMP(good manufacturing practice) organization. They are also perennially near thetop of the list of most frequently cited issues in observations, warningletters, and consent decrees from the U.S. Food and Drug Administration (FDA)and other regulatory authorities.1 (“There is a failure to thoroughly review [any unexplained discrepancy][the failure of a batch or any of its components to meet any of itsspecifications] whether or not the batch has been already distributed.”)

偏差调查是任何GMP组织中最重要的质量活动之一。在FDA和其他监管机构发布的观察、警告信和同意令中，它们也一直处于最常被引用的问题列表的首位。1(“无论批次是否已经分发，都没有彻底审查[任何无法解释的差异][批次或其任何部件没有达到任何规格]。”)

Clearly,many organizations have room to improve in the writing and managing ofdeviation investigations. The following sections identify common misstepscompanies make when conducting deviation investigations — and how you can avoidthem.

显然，许多组织在偏差调查的编写和管理方面仍有改进的余地。以下几节列出了公司在进行偏差调查时所犯的常见错误以及如何避免这些错误。

1.Not leveraging historical data for continuous improvement

Theinformation gathered over time through investigations contains a wealth of datathat can be used for continuous improvement, increasing productivity, andreducing the recurrence of investigations. Unfortunately, many organizationsonly review this data once a year and in a somewhat perfunctory manner.

Agood trending process is an important element in monitoring and proactivelyresponding to developing issues. Tracking investigation data (root cause,functional group, unit operation) will aid in continuous monitoring of thetypes of events and root causes occurring in your facility by product, processarea, and functional group, among others.

Developa list of standard event categories and actionable root causes in order totrend deviation and investigation data. This list can be upwards of 200 or moreand can aid investigators in writing their root cause in actionable terms.
1.不利用历史数据进行持续改进

随着时间的推移，通过调查收集到的信息包含了大量的数据，可用于不断改进、提高生产力和减少调查的再次发生。不幸的是，许多组织每年只审查这一数据一次，而且有些敷衍了事。一个良好的趋势过程是监测和积极应对发展中问题的一个重要因素。跟踪调查数据(根本原因、功能组、单元操作)将有助于持续监控设备中按产品、流程区域和功能组等发生的事件类型和根本原因。制定标准事件类别和可采取行动的根本原因清单，以便趋势偏差和调查数据。这份清单可以超过200份或更多，可以帮助调查人员以可诉的方式写出他们的根本原因。

2.Relying on human error as a root cause

Thisis a common finding that regulatory authorities will cite in theirobservations. Repeatedly stating human error as a root cause is a sign thatyour organization is not interested and/or resourced to find true root causesand to correct the underlying issues behind recurrence.

Humanerror can be a root cause category, but rarely is it the true and actionableroot cause, in and of itself. The true root cause is usually in other areas,such as procedures (“step x.x unclear”), training (“wasn’t assigned training onprocedure since SOP was not on training curriculum”), environment (“distractiondue to false fire alarm”), or machine (“improper equipment design and layout”).

Itis important to find a true, underlying root cause and to describe it inactionable terms to prevent recurrence and reduce the number of future humanerror-related events. Such events cost the industry a staggering amount interms of lost productivity, compliance and labor costs, and the human resourcesneeded to investigate nonconformances. The average cost of a deviation runsinto tens of thousands of dollars for major pharmaceutical companies.Preventing human error recurrence not only saves organizations money, but itreduces the likelihood of compliance issues, including regulatory findings.

Somequality systems will not allow human error to be used as a root cause, in orderto prevent the organization from stopping short of identifying and addressingthe true root cause behind errors (see error #3 below). For example, in many(but definitely not all) human error events, the employee involved could havedetected the error prior to it becoming a deviation. Therefore, “inadequateability to detect the problem” could be the actionable root cause in suchsituations. The resulting CAPA (corrective and preventive action) would becounseling or additional training that focuses on increasing the person’sability to detect and fix an error, or other job aids or improvements in theHMI (human machine interface) that will allow operators to better detectproblems in time to prevent a deviation. Counseling just on “paying attentionto detail” or on “the importance of GMPs” is not specific or adequate as astandalone CAPA. If someone doesn’t understand the importance of GMPs, theyshouldn’t be working in a GMP environment — and they definitely need moretraining.

2.把人为错误作为根本原因

这是监管当局在其意见中引用的共同结论。将人为错误反复声明为根本原因是一个迹象，表明您的组织没有兴趣和/或资源来寻找真正的根本原因，并纠正重复出现的根本问题。人的错误可以是一个根本原因范畴，但它很少是真正的和可行动的根本原因本身。真正的根本原因通常是在其他领域，如程序(“步骤x.x不明确”)、培训(“由于SOP没有列入培训课程而没有被分配关于程序的培训”)、环境或机器(“设备设计和布局不当”)。

重要的是找到一个真正的、潜在的根源，并以可操作的方式描述它，以防止再次发生，并减少今后与人类错误相关的事件的数量。这类事件在生产力损失、合规和劳动力成本以及调查不合格行为所需的人力资源等方面给行业造成了惊人的损失。对于大型制药公司来说，偏差的平均成本高达数万美元。防止人为错误的再次发生不仅节省了组织的资金，而且降低了合规问题的可能性，包括监管结果。一些质量系统将不允许将人为错误用作根本原因，以防止组织无法识别和解决错误背后的真正根源(见下面的错误#3)。例如，在许多(但肯定不是所有)人为错误事件中，所涉及的员工可以在错误变成偏差之前检测到错误。因此，在这种情况下，“发现问题的能力不足”可能是可采取行动的根本原因。由此产生的CAPA(纠正和预防行动)将是咨询或额外培训，重点是提高个人检测和修复错误的能力，或其他工作辅助或改进人机界面(HMI)，使操作者能够及时更好地发现问题，以防止出现偏差。仅仅就“注重细节”或“GMPS的重要性”进行咨询，作为一个独立的CAPA来说，是不具体的，也是不够的。如果有人不理解GMP的重要性，他们就不应该在GMP环境下工作，而且他们肯定需要更多的培训。

3.Not getting to the probable root cause

Thepercentage of investigations resulting in a root cause is a good metric for thehealth of your quality system — the higher the percentage, the better. Thereare many reasons why root causes are not found. Not committing adequate timeand resources is one. However, it is all too common for organizations to put insufficient effort, gather all the necessary facts and information, but stillfail to identify a root cause. Sometimes, this is the direct result of theinvestigator’s skill — they may have been trained insufficiently or lacktechnical command of the issues involved.

However,it is also surprisingly common for an investigation to conclude that a“definitive” root cause could not be identified, despite the fact that all thenecessary information is available and the conclusion is readily discernable. Amisguided interpretation of the facts or an unrealistic notion of definitivecan prevent the investigation from arriving at a most probable root cause.There is no regulatory standard that requires all investigation conclusions bedefinitive. A most probable root cause based on and justified by a thoroughinvestigation and supported by the available data and information issufficient.

Theproper RCA (root cause analysis) tool should be chosen for the problem at hand.For more difficult investigations, a Kepner-Tregoe or IS-IS NOT analysis canoften tease out a challenging most probable root cause from an array ofdiscordant facts.

3.没有找到可能的根本原因

导致根本原因的调查百分比-是衡量质量体系健康的一个很好的指标-这个百分比越高，越好。根本原因是多方面的。没有投入足够的时间和资源是其中之一。然而，各组织作出足够的努力，收集所有必要的事实和信息，但仍然找不到根本原因，这是非常普遍的。有时，这是调查人员技能的直接结果-他们可能没有受过充分的培训，或对所涉问题缺乏技术上的掌握。然而，令人惊讶的是，调查得出的结论也令人惊讶，即无法查明“确定的”根本原因，尽管有所有必要的资料，而且结论是显而易见的。对事实的错误解释或不现实的确定性概念可能会使调查无法找到最可能的根本原因。没有任何监管标准要求所有调查结论都是确定的。一个最可能的根本原因是以彻底调查为基础并以现有数据和资料为依据的，这就足够了。对于手头的问题，应该选择合适的RCA(根原因分析)工具。对于更困难的调查，Kepner-Tregoe或is-不是分析，通常可以从一系列不一致的事实中找出最具挑战性的最可能的根本原因。

4.Not getting to the true root cause

Findingthe true root cause is critically important. There have been many cases whereorganizations failed to identify and correct the true root cause of a problemthat was readily solvable — and they suffered millions of dollars in losses (orworse) as a result. The true root cause is an actionable one that is the mostconsistent with the available facts and information from a thoroughinvestigation and can be the most probable root cause mentioned above.

Atrue root cause is the underlying reason that allowed the event to occur.Understanding the true root cause requires the collection of all relevantfacts. Sometimes, these are clearly understood at the time of the event. Othertimes, it requires in-depth technical assessments that can span several months.

Onegood test for assessing if the true root cause has been found is to see if itcan be stated in terms that are directly actionable, meaning it links clearlyto a corrective action and is within the organization’s control. A simple andeffective way to determine if the true, actionable root cause has been found isto use the 5 Whys tool at the conclusion of an investigation, includingapplication to the outputs from more advanced tools, such as a fishbone diagramor Kepner-Tregoe analysis.

Forexample, failure of the backup power supply is not a true, actionable rootcause. Why did the back-up power supply fail? Hurricane Matthew cannot belisted as an actionable root cause, because the organization cannot preventhurricanes. However, an inadequate procedure for preventative maintenance ofthe backup power supply can be a true root cause, in which case the correctiveaction is to fix the procedure.

4.找不到真正的根源

找到真正的根源是至关重要的。在许多情况下，组织未能找出和纠正一个容易解决的问题的真正根源-结果他们遭受了数百万美元的损失(甚至更糟)。真正的根本原因是一个可采取行动的原因，这是最符合现有的事实和资料，从一个彻底的调查，可以是最可能的根本原因，上述提到。真正的根本原因是允许事件发生的根本原因。了解真正的根本原因需要收集所有相关的事实。有时，这些在事件发生时就被清楚地理解了。其他时候，它需要深入的技术评估，可以跨越几个月.评估是否找到了真正的根本原因的一个很好的测试是，看看是否可以用可以直接采取行动的术语来说明这一问题，这意味着它与纠正行动有着明确的联系，属于组织的控制范围。一个简单而有效的方法，以确定是否找到了真正的，可采取行动的根本原因是使用5 Whys工具在调查结束，包括应用于更先进的工具，如鱼骨图或Kepner-Tregoe分析（KT法，问题解决技巧-方法）。例如，备份电源的故障不是真正的、可采取行动的根本原因。为什么备用电源失灵？飓风马修不能被列为可采取行动的根本原因，因为该组织无法防止飓风。然而，后备电源预防性维护程序不足可能是真正的根本原因，在这种情况下，纠正措施是修复程序。

5.Preparing an unclear or difficult-to-follow investigation report

Manyinvestigation writers forget that their audience is not only internalemployees, but ultimately an external third-party, such as an inspector. Thus,the investigation needs to be readily understandable and clear, with all thenecessary supporting facts and rationale, so that it is comprehensible yearsafter the event. The most difficult challenge in accomplishing this is writinglogically, clearly, and succinctly, without presenting and repeating redundantinformation. Thus, training is an important element of writing effectiveinvestigations. Having mentors available to assist investigators can be animportant element in advancing skills.

5.准备一份不明确或难以跟进的调查报告

许多调查作者忘记了，他们的听众不仅是内部员工，而且最终还是外部第三方，比如检查员。因此，调查必须易于理解和明确，并有所有必要的支持事实和理由，以便在事件发生后几年才能理解。要做到这一点，最困难的挑战是在不显示和重复冗余信息的情况下，以逻辑、清晰和简洁的方式书写。因此，培训是编写有效调查的重要内容。有导师协助调查人员可能是提高技能的一个重要因素。

6. Ignoring contributing factors and associatedCAPAs

Contributing factors are elements that eitherwere necessary — in addition to the root cause — for an event to occur, or thatincreased the event’s impact. Contributing factors also require root causedeterminations and CAPAs. Addressing them limits the likelihood or impact ofsimilar events recurring in the future. Too often, investigations focus only onthe root cause and forget to address contributing factors. Use of the 5-Whystool is a useful mechanism to distinguish contributing factors from rootcauses.

6.忽视影响因素及相关因素

除了根本原因之外，造成事件发生的因素或者是必要的，或者是增加了事件的影响的因素。造成因素也需要根本原因的确定和卡帕。解决这些问题限制了今后再次发生类似事件的可能性或影响。调查往往只关注根本原因，而忘了处理促成因素。使用5-Whys工具是区分成因和根源的有用机制。

7. Prescribing inadequate CAPAs

Too many investigations lead to an appropriateroot cause but never link that cause to a CAPA. Most root causes andcontributing factors should be associated with one or more CAPAs; if not, aclearly justified rationale should be provided. One of the main purposes of aninvestigation is to prevent recurrence of the event, and this can only happenif the investigation determines a root cause and connects it to an appropriate,effective CAPA. Another consideration is whether or not an interim control isneeded while the CAPA(s) is being implemented. A risk assessment should be usedto make this determination.

7.开具不足的Capa

太多的调查导致了一个适当的根本原因，但从来没有将这一原因与CAPA联系起来。大多数根源和促成因素应与一个或多个CAPA相关联；如果没有，则应提供明确的理由。调查的主要目的之一是防止事件再次发生，只有当调查确定了根本原因并将其与适当的、有效的CAPA联系起来时，才能发生这种情况。另一个考虑因素是在实施“CAPA”时是否需要临时控制。应利用风险评估来作出这一决定.

8. Not performing interviews

Too many investigations fail to obtaininformation from the employees with the most relevant insight and informationsurrounding the event, either because the interviews are never conducted orthey occur too long after the event. Memories fade quickly, so interviewsshould be conducted as soon as possible after an event occurs. Someorganizations “swarm” an event immediately after it happens, with a teamconducting interviews. This is called “freezing the scene.” Gatheringhigh-quality information soon after the event will save you future time andeffort, and improves the quality of the investigation.

Talking to key personnel using well thought-outinterview questions prepared in advance, when applicable, will make theinterview more useful and insightful and the investigation much more efficient.The essential details of interviews should be summarized in the investigation.

8.不进行采访

太多的调查无法从员工那里获得与事件相关的洞察力和信息，要么是因为调查从未进行，要么是在事件发生后太久才进行。记忆很快就会消失，所以调查应该在事件发生后尽快进行。一些组织在事件发生后立即“蜂拥而至”，由一个团队进行采访。这就是所谓的“冻结现场”。事后收集高质量的信息将节省你今后的时间和精力，并提高调查的质量。在适当的情况下，用事先准备好的面试问题与关键人员交谈，会使调查更有用、更有洞察力，调查也更有效率。采访的基本细节应在调查中加以总结。

Conclusion

While conducting a thorough deviationinvestigation is hard work, failing to do so will yield inaccurate root causesand misdirected CAPAs, and recurring deviations, along with increasedregulatory and financial risk. The payback from investing in deviationinvestigations comes in the form of improved operational performance, reductionin costs, increased quality, and improved compliance.

结论：

尽管进行彻底的偏差调查是一项艰巨的工作，但如果不这样做，将产生不准确的根本原因和错误的CAPA，以及重复的偏差，并增加监管和金融风险。增加对偏差调研的投入可改进操作性能、降低成本、提高质量和提升法规依从性。

1：References:

FDAFY 2015 Inspectional Observation Summaries: http://www.fda.gov/ICECI/Inspections/ucm481432.htm#Drugs



来源：新药研发质量管理