Rights, roles and responsibilities of caregivers, industry and government debated.

Several participants addressed the importance of participation by various health care providers, government agencies and medical products manufacturers in the effectiveness of any future Sentinel Network at the FDA meeting, particularly in ensuring the quality of data on which the network will rely.

Alexander Ruggieri, M.D., of Cerner Galt and a columnist for "AER News" noted that the "gaps and obstacles discussed are all data quality issues," that relate to the relevance and consistency of information provided to the network. Too much data from too many sources could "create the risk of regression to the mean," he said. "Think high-level first," Ruggieri stated, adding that the network "must be mutually beneficial to everyone; don't leave industry out."

Ruggieri cautioned against relying on health care providers to provide information on adverse events, a point that was disputed by Steven A. Goldman, M.D., a consultant in risk management, pharmacovigilance and clinical safety and member of the editorial advisory board of AER News.

Although he agreed that data quality is central to the effectiveness of a safety network, Goldman argued: "Programs to train caregivers in adverse drug event identification and reporting work. The problem is that those programs don't get funded." Goldman, who led the creation of the MedWatch system while at FDA, noted that "the quality and number of reports we received after MedWatch increased significantly." He added that "feedback is critical" as an incentive to caregivers to actively participate in safety reporting. If you make data and feedback clinically relevant, health care providers are very receptive."

In roundtable discussions meeting participants debated the willingness and the ability of clinicians to provide quality data to a safety network. Chute noted that "it's unlikely that clinicians will recognize sentinel events," arguing strongly in favor of a system that relies on data mining of information to detect signals. However, meeting participant Frederic Resnic stated that voluntary reporting and data mining "are complementary processes. We should not dismiss voluntary reporting, especially for unpredicted adverse events. Culling data requires starting with some prediction of the types of adverse events we might expect to see."

Hill suggested that the Sentinel Network be a "tiered system that starts with the unexpected event, then feed backs that information to clinicians for them to monitor. The things you don't expect are what you want the sentinel to see."

Jonathan Selzer of the American Pharmacy Assn. (APhA) stated that surveys have shown that "pharmacists are the most trusted health care providers and have the most contact with patients." As such, pharmacists could play a central role in a system for reporting adverse drug reactions. "Pharmacists are the most dramatically under-utilized professionals," he said, and they could "make a unique contribution" to safety efforts. APhA is conducting a three-year project in 200 sites to utilize pharmacists in improving clinical outcomes for patients with hyperlipidemia and diabetes.

Hugo Stevenson of Quintiles reported that his company's experience has indicated that any reporting system "that relies on physicians to step out of their workflow is inherently flawed." In one Quintiles survey project, Stevenson noted, "only four doctors out of 100 even respond to a request to do a survey on post-market safety. Out of those four, only one will sign up to participate. Within six months, that one doctor has stopped reporting information."

For this reason, Stevenson proposed a system that relies more on patient reporting into a system that provides "timely, relevant information" and at "any hint of adverse events" contacts the patient's caregiver. Quintiles is piloting such a system, iGuard, beginning in April.

Other participants agreed that reporting systems that interfere with the typical operation of a medical practice will not be successful. Many suggested opting for integration of a few simple entries into a patient's electronic medical record that could be done quickly and in the course of daily practice, but still yield clues when databases are searched and analyzed.

The discussion yielded several opinions in favor of a hybrid system that "needs to find common problems as well as rare ones," Platt said. "The public health impact of common events is much greater than the rare events." He suggested that "there is a large upside to looking at data collected in large databases. A system that uses diagnoses and similar information entered into a database to prompt clinicians to suspect adverse events in some cases--as in the case of some vaccines--might take best advantage of both the data and the clinician."

Further, Platt said, participation in a safety reporting system should be seen as an ethical imperative. "We must make clear to everyone that supplying information is part of a social contract."

Participants from pharma agreed that drug and device manufacturers have "as much of an interest as anyone" in the safety of their products and are willing not only to participate but also to help finance the creation and testing of the Sentinel Network. However, industry representatives clearly indicated that the companies "are not comfortable 'stepping back'" from the process of analyzing information gleaned from their own data, but want to be involved in analysis design. Further, industry representatives noted that their data could have "regulatory implications."

But the consensus among industry participants was that the companies "very much want to collaborate" with government and academia and "don't have to take the lead." They believe that "a transparent, collaborative effort will be more credible" to the public and help to rebuild public trust in medical product safety regulation.