On September 23-24, 2013, the Food Advisory Committee met to address CFSAN’s charge. The task for the Food Advisory Committee was to review and comment on CFSAN’s proposed pilot system and data sources for chemical signal detection. CFSAN was proposing a systematic process that would better enable the Center to recognize and evaluate evidence of emerging chemical hazards or newly recognized risks from known chemical hazards in food, dietary supplements, food and color additives, and cosmetics. This systematic process would also enable CFSAN to be more proactive in identifying and monitoring emerging issues rather than reacting to issues after they occur. For the meeting, CFSAN provided the Food Advisory Committee with a background document describing the proposed pilot system to detect chemical signals and a list of specific questions to address.
Current Activities by CFSAN in Response to the Food Advisory Committee Report
Since the Food Advisory Committee meeting, the CFSAN chemical signal workgroup has continued to meet and address the Food Advisory Committee report. The workgroup also met with CFSAN management to develop a strategy to move forward. In the following section, CFSAN responds to the Food Advisory Committee report which contained both general and specific comments. In addition, CFSAN is working on implementing a two year pilot system with the intention to assess its outcome at the end of the two years. CFSAN is actively developing position descriptions for getting specifically designated personnel for this pilot. The pilot system will be in the Office of Analytics and Outreach which already is familiar with databases, bioinformatics, and statistical/scientific projects. CFSAN is also working on obtaining the suggested data sources through internal and potential outside contract mechanisms. It is anticipated that the pilot process will start in October, 2014.
Written Comments by CFSAN in Response to the Food Advisory Committee
At the end of the meeting, the Food Advisory Committee provided CFSAN with comments on the proposed pilot and responses to the specific questions CFSAN posed to the Committee.
General Comments from the Committee
- Explain how this will improve what FDA has been doing and the ability of FDA to identify new signals (alerts or “blips on the radar”). Give an example. Show how we can learn from history.
- Bring to a higher level sooner; shorten timeframe to action
- Better communication internally among program offices
- Easier to identify what has already been done and what is currently being done
- Establish consistent, meaningful, efficient prioritization process
- Better communication to the public
- Make process more systematic
- Make process more transparent
- Ensure scientific integrity
- Identify potential new hazards for either existing or new chemicals
- Important to define when new information indicates that something is or is not a priority, vs. an absence of information
- Distinguish Anticipatory vs. Responsive (After- the- Fact) Data sources for identifying signals
- Clarify how to sift through information so that priorities can be established (what is low priority vs high priority).
- Redo graphic 1 page 4 (e.g., include additional loop between review and prioritization)
- Given the importance of this project, launch pilot in a short time frame. Revisit to see how it is working after pilot phase. Include matrix to measure success.
- Proposal is under-resourced/understaffed (especially at level of signal manager)
- Need for action (and communicating that) to keep the process proactive and not reactive
- Consider how to ensure the process doesn’t bog down and that more timely action is taken and communicated
- o Signal Management Team and/or CFSAN Management Council should take on this role and involve higher levels in the Agency as appropriate
- Metrics are needed to validate the success of the program
- Ensure that system is transparent to the public and visible
- Make sure that the benefits of this system don’t become “invisible” and undervalued
- Need a process to report successes
- Elicit public input/involvement
- Consider alternative approaches
- team approach instead of two signal managers
- contract out implementation of some/all of the pilot study
- pilot on a subset of CFSAN’s areas of responsibility (e.g., dietary supplements)
- Review Appendix B
- for value-added at each stage
- Establish metrics, guidance on timeframes
- Appendix B should have a box added to indicate that communication should be considered
- Include description under “CFSAN Management Council”, including communication functions
- for value-added at each stage
CFSAN Responses to General Comments from the Committee
CFSAN appreciates the FAC recommendations on the pilot system for chemical signal detection. CFSAN has provided responses to each specific recommendation below. There are several general comments that are not addressed by the individual charge questions; thus, CFSAN comments are listed below.
- CFSAN agrees with the importance of this pilot system and the need to launch the pilot quickly. CFSAN also agrees with establishing timelines and developing a way to evaluate the pilot’s success, e.g. timelines and milestones.
- CFSAN recognizes that the FAC has discussed that the system be transparent, visible, and communicates to the public. However, FAC must recognize that this system is an internal process and because of the sensitive nature of the information, initial discussions and actions must remain within CFSAN. The long term goal of this process is that it will demonstrate to our stakeholders that CFSAN has a systematic and proactive approach to chemical signal detection.
- CFSAN agrees that this pilot system was initiated because CFSAN recognizes the need to be proactive in identifying chemical signals. This process is designed to be more anticipatory, enhance communications within CFSAN, be systematic, and to aid CFSAN in identifying signals more proactively and to act more quickly.
- This pilot is designed to identify new potential hazards more rapidly and more proactively.
- CFSAN recognizes that the pilot is understaffed and under-resourced. CFSAN is discussing how to manage a successful pilot with limited resources, perhaps by relying in part on one or more contracts.
- CFSAN anticipates that developing approaches for prioritization and weighting criteria will evolve during the pilot and from continued discussions.
- CFSAN appreciates FAC’s recommendation of starting the pilot with a subset of CFSAN products. However, the workgroup believes it is important to implement this pilot across the Center. This will help facilitate participation of all the CFSAN Offices and help initiate interest and buy in from all the CFSAN staff.
- CFSAN will revise Graphic 1 to make sure there is proper recognition of a loop between review and prioritization and revise the Appendix B diagram to add a communication box in the CFSAN Management Council functions.
CFSAN Responses to the FAC Answers for the Specific Charge Questions
Question 1: What are the sources of data and information on chemical hazards that might best identify emerging chemical hazards or newly recognized risks from known chemical hazards? Rank these sources of data or information in order of the expected value in identifying new or emerging chemical hazards.
The committee recognizes that different data sources will serve different purposes. Sometimes anticipatory sources will be most useful, for example. It may be useful to distinguish Anticipatory sources (import database, exposure and monitoring data programs (e.g, PDP, Mussel Watch, SAR information) from after-the-fact sources (Literature Reviews, NIEHS grants). Animal feed and animal health information historically have been useful anticipatory sources
- Scientific meeting abstracts
- Work with NLM (e.g, Phil Wexler) to identify sources
- Scan press, use google news and Lexis/Nexis
- Scan Wikipedia pages on relevant chemicals/products/processes for new changes (get alert when a change occurs)
- Canadian Chemical Hot list, EU COSING, and other authoritative lists of chemicals of concern
- PETNET, AAFCO
- TOXCAST data and results from high throughput methods
- Explore eLEXNET
- REACH, NTP, NCTR, Prop 65, IARC, heavy metals database for seafood, MSDS, UN GHS and other relevant databases, occupational data (ACGIH), water monitoring data (from EPA, state, etc.), law enforcement databases, military databases, court case data, RAMS, NEISS (in addition to sources listed in document)
- Use NHANES in conjunction with trends, health outcome data
- Encourage all FDA staff to submit data/information since this is likely to be more targeted than a signal manager can accomplish
- Identify groups/individuals/professional/medical organizations with key expertise from varying perspectives (academic, industry, advocacy) to provide updates to agency on new chemical signal information, especially with expertise on vulnerable populations or scientific areas that are rapidly changing
- Interact with USP, NSF, ACC, Pharma, other professional and certification and scientific organizations and societies to provide new information on chemical signals in their area of expertise
Response by CFSAN
Q1. CFSAN appreciates the suggested new data sources and recognizes that are many databases and data sources that still need to be considered. Some of the suggestions by the FAC were already listed, but CFSAN will continue to identify and evaluate the data sources and databases for the chemical signal detection process. For example, CFSAN will contact the NLM to see what it may have available for literature and database searches. The identification and evaluation of various databases, data sources, and data mining approaches will continue and the final list will evolve through the pilot.
Question 2: Are the signal definition and the categories of signal types clear, well-defined, and inclusive? Are the definition and categories sufficient to detect potential issues related to chemicals in foods, food and color additives, dietary supplements, and cosmetics? Are there other categories that should be included or others that should be deleted?
- Definition 1 might be reworded to say “… that might be considered a risk to public health” as the phrase “might be considered” encompasses perceived risk
- Definition 3 might be broadened/rethought to include issues related to the supply chain (e.g., manufacturing/process changes, or areas of the world where there are no or lax regulatory practices). Ensure that all relevant external factors that might affect exposure or hazard (e.g., climate change, changes in production, changes in use) are considered
- Clarify that definition includes additives (food and color additives), heavy metals, etc.
- Consider reframing the 3 definitions as follows:
- Definition 1: New hazards: …
- Definition 2: New uses: …
- Definition 3: New process: …
- Characterize “increased exposure” (generally and vulnerable populations)
- Every signal that is identified deserves to be evaluated, and the Agency should be prepared to explain how it reached its conclusions
- Committee agrees that it is important to include hazards and reactions that might be considered a risk or a perceived risk.
Response by CFSAN
Q2. CFSAN appreciates the FAC recommendations and discussion about the signal definition and categories. CFSAN appreciates the FAC conversation about risk and perceived risk and agrees with the suggestion that “might be considered” in Definition 1 would encompass perceived risk. We agree that we should drop perceived risk and that we should include food and color additives. CFSAN thinks that the 3 definitions under signal definition already incorporate new hazards, new uses, and new processes. Other than the two identified changes, CFSAN is going to keep the current definition for the pilot. Much discussion and consideration went into the definition, and CFSAN believes that the pilot project will provide further opportunity to revise the definition.
Question 3: Once a potential signal is identified, CFSAN recognizes the need for considering and weighting various factors in the review and prioritization of a signal, and subsequent action. What factors and weighting are most critical in moving an identified signal from the Signal Manager through the process to review? What factors and weighting are most critical in prioritizing a signal into particular categories (“low” versus “high”)? What factors and weighting are most critical in deciding the follow-up and action on a signal?
- Product consumed by a susceptible population (e.g., children, pregnant women, elderly, diabetics, genetic polymorphisms leading to increased susceptibility)
- Product consumed in large quantities or by a large segment of the population
- Potential for contamination of multiple food products
- Acute, teratogenic, mutagenic, clastogenic effects vs. chronic effects
- Chronic effects are still very important but have longer to respond and often have less certainty
- Irreversible effects should be weighted more highly
- New information on adverse effects (a chemical is identified as a carcinogen, reproductive/developmental toxicant, mutagen (e.g., CA, IARC, JECFA, NTP, EFSA, peer-reviewed journal)
- Information issued by authoritative bodies
- E.g., JECFA or another expert body makes a new recommendation (e.g., an additive is no longer considered appropriate for use in a particular product (e.g., infant formula)
- Level of confidence in the data is also a consideration in how to move the signal through the process
Response by CFSAN
Q3. CFSAN recognizes how difficult it is to prioritize action on generated signals, develop weighting criteria, or to determine a threshold of action. CFSAN appreciates the discussion around exposure and hazard and will continue discussions on this topic. CFSAN recognizes that the level of confidence in the data is critical and will work on better defining and/or measuring level of confidence. It is expected that the pilot process will help better define data confidence.
Question 4: How should CFSAN conduct ongoing literature searches to capture new and emerging data on chemical hazards in published literature on foods, food and color additives, dietary supplements, and cosmetics as part of this detection system? What key words would be appropriate to search on? What journals are most valuable for this purpose?
Note that by the time a signal is in the literature, it is likely to have emerged as a signal elsewhere. Nevertheless, literature searches are valuable.
- Consult National Library of Medicine for search terms
- Some key words are generic and some specific/case-by-case
- Consider how done by other organizations (ATSDR’s Tox Profile, IARC monographs, TSCA reporting)
- Neural networks may be useful in the future
- Use available search engines (e.g., journals) with key words that will notify or alert agency to new articles of interest (active system vs. just passive systems)
- Individual scientists/other staff in the agency are conducting their own searches and have their own systems for becoming aware of new information so they should be encouraged/empowered to send in signal forms to the system
Response by CFSAN
Q4. CFSAN appreciates the suggestions by the FAC. CFSAN plans to follow up with the National Library Medicine and plans to continue to work on the neural networks. The pilot process will help refine the process of obtaining more information from CFSAN personnel and other organizations. CFSAN plans to use an existing CFSAN system (Traction) to build in alerts from Google and other sources, including signal entry forms from all CFSAN programs and staff.
Question 5: Are there specific web-based technologies or services which the FAC would recommend for generating effective broad literature searches and monthly reports? How frequently should these searches be done (e.g., monthly or at different frequencies)?
- See answer to question 4
- Using alert systems (see above) will result in the Agency receiving information frequently, most likely daily
- Do Google news daily (alerts or “canned” searches)
Response by CFSAN
Q5. See answer to question 4.
Question 6: Are social media tools available and refined enough to be of use in this area? Should CFSAN take advantage of current contracts with social media or are there other ways to obtain and analyze social media information? Would it be useful for this process?
- Social media tools are available, but we do not know if they are refined enough to be of use in this area. Agency should seek out experts on mining social media (this committee doesn’t have this expertise).
- Piggy back on use of social media by other centers within FDA and by other agencies for adverse event reporting
- This function may require more resources than currently proposed
- Consider using active surveillance in addition to passive surveillance
- Utilize FDA website where public can report effects/signals in addition to social media
Response by CFSAN
Q6. CFSAN appreciates the FAC suggestions and recognizes that social media is an emerging technology. CFSAN already has a web portal for public/stakeholder input, and this proposed pilot is an effort to provide active surveillance. CFSAN is currently evaluating several social media contracts and other data mining systems to see if they will be useful for the pilot system.
Question 7: How should the various data for the chemical signal detection process be stored and managed? The current proposed system enables the data to be collected in one focal point, managed by two designated full-time employees, and stored on a designated server. Future plans involve a CFSAN specific data warehouse and analysis network. Is the current proposed system adequate for a pilot and for designing a long- term system? How should this repository be structured, who would have access, and how often should it be updated?
- This committee lacks the expertise to provide much input
- Is important to collect information in one secure place
- State officials who are commissioned officials should have access to the system as needed
- Should be easily searchable
- Characterize documents by various data fields to improve utility and to easily know what information was used to make a decision vs. new information
- Serves as archive/repository
- System should generate alert to relevant personnel to help facilitate communication/action
- Allow way to communicate with other adverse event reporting systems w/in the Agency (e.g., PETNET)
- Consider back-up system
Response by CFSAN
Q7. CFSAN appreciates the FAC’s comments. The proposed pilot will be a system that is supported by IT at FDA and will be easily searchable, have the ability to track the activities and information around each signal, and provide alerts for CFSAN personnel. Traction will be a common portal for all the data sources around a certain chemical contaminant.
Question 8: What skill set should a signal manager have? Should he or she be an information management specialist, or should he or she have expertise in other scientific disciplines (e.g. chemistry, toxicology, epidemiology)?
- Recommend a team instead of two signal mangers
- Signal manager should be senior-level “big picture,” generalist capable of weighing evidence, making judgment calls and possessing good networking, communication, and management skills
- The team should have expertise in, for example, toxicology, epidemiology, chemistry, and informatics
Response by CFSAN
Q8. CFSAN recognizes that having a signal manager(s) and team with appropriate and outstanding skills are a critical component to the pilot process and appreciates the FAC’s recommendations. CFSAN recognizes that this pilot will be resource intensive and that a team of signal managers and experts would be ideal. The current CFSAN work group also recognizes that budget restraints will affect the Center’s ability to implement an ideal pilot system. The work group is agreeing with the FAC recommendations but also is proposing the limited resource plan to CFSAN management. The plan will, however, include one or more contracts for the duration of the pilot, to enhance resources in the form of CFSAN personnel.
Question 9: Does the signal review team composition make sense? How often should it meet? Should there be any other types of committees considered for decision-making or to facilitate communication? Who should be included as members and what types of scientific disciplines should be included?
- Yes, composition makes sense
- Generally, senior scientists with expertise in, for example, toxicology, epidemiology, chemistry, informatics, physiology, biology, and public health will be needed
- Bring in expertise as needed
- Meet regularly (e.g., monthly, quarterly) and as needed
- Consider how to obtain public input (e.g., call for data, public meeting) and meet need for transparency and visibility
Response by CFSAN
Q9. CFSAN appreciates the FAC recommendations, and agrees that the Signal Manager and Signal Review Team need to establish regular meetings as well as have the ability to meet for critical or emerging issues. CFSAN recognizes that this pilot will help in identifying signals early and that there is a need for visibility. CFSAN communication with the public is handled through a centralized Office and approved by the Office of Foods and Veterinary Medicine, FDA.