Gleaning Good Practices from Warning Letters

An FDA field investigator leaves your site after spending five days scrutinizing your device clinical trial records, interviewing study staff, asking pointed and open-ended questions, and reviewing your procedures. About four months later, an overnight carrier delivers an envelope to you; it is from CDRH. You carefully open the envelope. It is a multipage letter with the following words bolded across the top of the first page: “warning letter.”

But all is not lost. This article examines warning letter data from FY 2007 and discusses how device companies can learn from them.

Behind the Warning Letter

According to FDA's Office of Enforcement, the agency issued 471 warning letters to regulated industry in FY 2007 (October 2006–September 2007).¹ Surprisingly, the CDRH Bioresearch Monitoring program (BIMO) issued 29 or 6% of those letters in 2007 (in FY 2004, those numbers reached record highs of 44 and 9% respectively). These BIMO warning letter percentages are significant considering that the CDRH BIMO program accounts for approximately 2% of the inspections conducted by FDA annually.

FDA's Regulatory Procedures Manual indicates that a warning letter is informal and advisory in nature. It communicates the agency's position on a matter, but it does not commit FDA to taking enforcement action (e.g., civil money penalties). FDA's practice is to give individuals and firms an opportunity to take voluntary and prompt corrective action before it initiates an enforcement action. The use of warning letters is based on the expectation that most individuals and firms will voluntarily comply with the law.²

CDRH BIMO looked at its FY 2007 warning letter data and wondered whether additional insights could be gleaned from mining and analyzing these data. The BIMO unit examined all 2007 warning letters to device clinical investigators (CI), sponsor/monitors (S/M), and institutional review boards (IRBs) for information that could be useful to device companies as they develop risk assessment strategies for conducting clinical trials. BIMO rounded the percentages used in this article up or down to whole numbers.

CDRH BIMO issued 52% of its warning letters to clinical investigators (see Figure 1). This is not surprising considering that CDRH allocates almost 60% of its BIMO inspection resources to the oversight of clinical investigators. When we looked closer at the data, 60% of those investigators were located in academic medical centers or local community hospitals and the remainder were in private practice. On the other hand, 31% of the CDRH BIMO warning letters went to device sponsors, to which 12% of BIMO inspectional resources are assigned. The majority of the sponsors were noncommercial or small businesses. Five warning letters (17%) also went to hospital IRBs, two of which were repeat offenders. CDRH BIMO inspected 323 regulated research entities in FY 2007, with the breakdown as follows: 183 clinical investigators, 92 IRBs, 40 sponsors, and 8 nonclinical laboratories, which must follow good laboratory practices (GLP). See Figure 2.

We also looked at the distribution of device good clinical practice (GCP) warning letters across the FDA district offices that conducted device GCP inspections. Interestingly, six firms under the regulatory umbrella of the Denver district office received warning letters, which accounted for almost 20% of the total number issued by CDRH—even though the Denver district receives only about 3% of the device BIMO inspection requests annually. Moreover, five of the 19 FDA district offices (Cincinnati, Denver, Dallas, New Orleans, and Seattle) accounted for 62% of all the warning letters issued by CDRH BIMO. These data are again surprising because those five districts generally receive about 25% of all BIMO inspection assignments in a given fiscal year.

One explanation may be that CDRH BIMO does a better job of focusing its resources on high-risk device research areas, which may account for the apparent lack of geographical warning letter distribution. Figure 3 shows the distribution of warning letters across the associated FDA district offices. The district office names are shortened and most are readily apparent. The exceptions may be New England (NEW), New Orleans (NOL), and San Francisco (SAN).

Internally, CDRH BIMO's Special Investigations Branch (SIB) conducted 69% of the inspections that resulted in a warning letter issuance (see Figure 4). SIB's primary role is to follow up on complaints of research misconduct and previously violative inspections. The majority of SIB-related warning letters were the result of research misconduct complaints. Other warning letters issued were related to repeat offenders and other surveillance inspections.

BIMO's Program Enforcement Branch issued nine warning letters due primarily to CDRH's early intervention and marketing application activities. Early intervention activities are risk based and focus on active IDE studies of novel technologies, vulnerable populations, or significant public health issues. It is interesting to note that CDRH BIMO inspections conducted pursuant to marketing applications (i.e., PMA, 510(k), etc.) resulted in the issuance of just two warning letters in FY 2007, none of which generated data reliability concerns. One explanation is that BIMO's early intervention program sensitizes device firms to the importance of well-managed clinical trials during the active research phase, leading to marketing applications with higher-quality studies.

CDRH BIMO issued 23% of the warning letters to sites that had a previous BIMO inspection. Of the seven sites that had a previous BIMO inspection, nearly three-quarters of those sites had a previously violative inspection (i.e., had received at least a warning letter). Based on this information, research entities with previously violative inspections tend to be repeat offenders. On the other hand, CDRH issued 77% of the warning letters to sites that FDA inspected for the first time (see Figure 5). In most cases, systemic issues such as lack of GCP training, lack of management or supervisory oversight, and ineffective corrective and preventive actions were the primary causes of warning letter issuance.

Warning Letter Response

When a regulated entity receives a warning letter, the recipient must provide a response within 15 days of receipt. In some cases, CDRH BIMO grants extensions in order for the regulated entity to provide its clear and comprehensive plans to comply with the law. The warning letter response is a critical component of the regulated research entity's continuous quality improvement process. It also provides the foundation upon which the entity will make corrections and propose preventive actions to minimize or eliminate recurrence. Regulators look closely at these responses to ensure that appropriate remedial measures are planned and implemented within reasonable timeframes. Some device research entities prepare adequate responses for themselves and their clinical sites by using a corrective and preventive action (CAPA) approach that is similar to the one used for their finished-product quality system. CAPA strategies include the following:

Our data suggest that regulated entities need to improve their warning letter responses. For example, CDRH BIMO received inadequate first responses to warning letters from more than half of those regulated entities that received letters in FY 2007 (see Figure 6). When we looked further into our data, 66% of the sponsors that received warning letters had an inadequate first response, and 60% of the clinical investigators had an inadequate first response. In most cases, the responses lacked proper remedial actions or documentation to support such actions. Inadequate responses may waste valuable time and money as companies pursue the advancement of their technologies to the marketplace.

Lastly, we looked at specific device categories to assess whether certain device clinical studies tended to generate warning letters. In FY 2007, six categories of devices accounted for all warning letters issued by CDRH BIMO with cardiovascular, orthopedic, general hospital, and plastic surgery device studies topping the list (see Figure 7). When CDRH stratified the data further, they indicated that two-thirds of the clinical investigator warning letters involved cardiovascular or orthopedic studies (see Figure 8). FDA expected the data to be skewed in that direction because the vast majority of active IDEs fall within the cardiovascular and orthopedic categories. In contrast, more than half of the device sponsors that received warning letters were associated with general hospital and plastic surgery devices.

Conclusion

What can we learn from a device GCP warning letter? Many device research entities view a warning letter as an opportunity for improvement. They incorporate the regulatory departures outlined in the warning letter into their CAPA program as part of their in-house quality system. This action promotes continuous quality improvement of research activities, which leads to higher-quality research data and enhanced human subject protection.

These outcomes may also lead to immediate and long-term dividends such as minimal clinical trial delays and unplanned costs, as well as reduced time to market—a win-win scenario for all parties involved. Additionally, CDRH has seen some device firms use risk analysis principles to determine the likelihood of an unfortunate event occurring in their clinical trial processes. Firms then apply risk mitigation strategies to minimize its occurrence. Although this warning letter sample size is small, the information within this article may be helpful toward that end.

Acknowledgments

Special thanks go to CDRH's Special Investigations Branch (Doreen Kezer, Catherine Parker, et al.) for their work in generating the CDRH BIMO data that formed the basis for this article.

References

1. J Favole, “Drop in FDA Warning Letters Signals Enforcement Shift,” Wall Street Journal [online], [cited 20 July 2008]; available from Internet: http://online.wsj.com/article/BT-CO-20080606-711156.html.

2. FDA Regulatory Procedures Manual [cited 20 July 2008]; available from Internet: www.fda.gov/ora/compliance_ref/rpm/chapter4/ch4-1.html.

Copyright ©2008 Medical Device & Diagnostic Industry