SOUNDING BOARD cepted that monitoring should focus on those those involved in the conduct of clinical trials aspects of trials that are of most relevance to are being confronted by an overwhelming vol the rights and safety of participants (e.g, the ume of reports on suspected unexpected serious consent procedures and ascertainment of serious adverse reactions25 and, despite a substantial ex- adverse events) and on the reliability of the study penditure of money and effort, this approach to results(e.g, the integrity of the randomization pharmacovigilance has rarely led to useful in process and completeness of follow-up). 20, 21,23 sights or improved safety Monitoring should be designed to detect impor- By contrast, moderate adverse effects of treat tant problems early in a trial so that they can be ment on common outcomes, such as an increase Dressed, as opposed to discovering problems in cardiovascular events with coxib treatment, 26 retrospectively by auditing, when there is no may well be of much greater relevance to public longer an opportunity to rectify them. Conse- health, but their reliable detection requires evi- quently, the use of centralized statistical moni- dence from studies that are both randomized toring as part of a risk-based approach to ensur- and controlled. 1 Consequently, a more effective ing trial quality has been advocated by the strategy for safety monitor n randomize Clinical Trials Transformation Initiative(www controlled trials would be based on the regular ctti-clinicaltrials. org)24 and endorsed by the review of the emerging safety data that is cus Food and Drug Administration(FDA). 6 European tomarily conducted by independent data and regulators hav ly recommended risk- safety monitoring committees, with the study based quality management, 17 18 and the proposed treatment assignments revealed and considered Union regulation allows the extent and in the context of the efficacy results. The degree nature of monitoring to be modified depending to which such safety information is shared with on the characteristics of the trial. However, be- regulatory authorities could be agreed on before cause the current draft of the regulation refers the start of the trial and could depend or to the ICH-GCP guidelines as a quality standard ber of factors, such as the amount of (despite their inherent problems), the successful experience with the treatment being implementation of a risk-based approach in Eu- The cogency of these concerns about the present rope will require either a fundamental shift approach to safety monitoring has been recog away from the current rigid interpretation of the nized by the FDA, which has issued revised guidelines or a substantial modification of the guidance for reporting requirements. 25,27 In par- guidelines themselves ticular, to reduce current levels of overreporting a risk-based approach should also be applied of alleged serious drug reactions, the FDA seeks safety monitoring during trials. However, the rapid reporting of such an event only when there ICH-GCP guidelines and related regulations cur- is considered to be a reasonable possibility that it ently require rapid reporting of all serious ad- was caused by the study drug. The FDa guidance verse events thought to be related to the study also distinguishes between the rare circumstanc- treatment and not previously documented with es in which it is appropriate to submit individual that treatment to all relevant regulatory authori- case reports and the more common circumstanc- ties,ethics committees, and site investigators. es in which cases should be aggregated and com- In accordance with the regulatory require- pared with those in a control group. However, ments, these suspected unexpected serious ad- the European Commission guidance and the pro- verse reactions"are typically reported on a case- posed regulation from the European Union have by-case basis (rather than as grouped reports not yet made this important distinction. with a meaningful denominator) only for the In conclusion, some regulatory agencies have participants in the active-treatment group rath- responded positively to the need for impro er than with the corresponding event rates for ments in the regulatory environment for clinical the control group. The use of this type of un- trials. However, the effect of those responses controlled data can reasonably be expected to may be limited by the complex system of regula- detect only large effects of drug exposure on tion and related bureaucracy that applies to clin- rare outcomes, such as the Stevens-Johnson ical trials at local, national, and international syndrome, hepatic failure, or angioedema. Cur- levels. Those responsible for the ICH-GCP guide- rently, however, regulatory authorities and all lines have been resistant to engaging adequately N ENGLJMED 369: 11 NEJMORC SEPTEMBER 12, 2013 106n engl j med 369;11 nejm.org september 12, 2013 1063 sounding board cepted that monitoring should focus on those aspects of trials that are of most relevance to the rights and safety of participants (e.g., the consent procedures and ascertainment of serious adverse events) and on the reliability of the study results (e.g., the integrity of the randomization process and completeness of follow-up).20,21,23 Monitoring should be designed to detect impor￾tant problems early in a trial so that they can be addressed, as opposed to discovering problems retrospectively by auditing, when there is no longer an opportunity to rectify them. Conse￾quently, the use of centralized statistical moni￾toring as part of a risk-based approach to ensur￾ing trial quality has been advocated by the Clinical Trials Transformation Initiative (www .ctti-clinicaltrials.org)24 and endorsed by the Food and Drug Administration (FDA).16 European regulators have also recently recommended risk￾based quality management,17,18 and the proposed European Union regulation allows the extent and nature of monitoring to be modified depending on the characteristics of the trial. However, be￾cause the current draft of the regulation refers to the ICH-GCP guidelines as a quality standard (despite their inherent problems), the successful implementation of a risk-based approach in Eu￾rope will require either a fundamental shift away from the current rigid interpretation of the guidelines or a substantial modification of the guidelines themselves. A risk-based approach should also be applied to safety monitoring during trials. However, the ICH-GCP guidelines and related regulations cur￾rently require rapid reporting of all serious ad￾verse events thought to be related to the study treatment and not previously documented with that treatment to all relevant regulatory authori￾ties, ethics committees, and site investigators. In accordance with the regulatory require￾ments, these “suspected unexpected serious ad￾verse reactions” are typically reported on a case￾by-case basis (rather than as grouped reports with a meaningful denominator) only for the participants in the active-treatment group, rath￾er than with the corresponding event rates for the control group. The use of this type of un￾controlled data can reasonably be expected to detect only large effects of drug exposure on rare outcomes, such as the Stevens–Johnson syndrome, hepatic failure, or angioedema.1 Cur￾rently, however, regulatory authorities and all those involved in the conduct of clinical trials are being confronted by an overwhelming vol￾ume of reports on suspected unexpected serious adverse reactions25 and, despite a substantial ex￾penditure of money and effort, this approach to pharmacovigilance has rarely led to useful in￾sights or improved safety. By contrast, moderate adverse effects of treat￾ment on common outcomes, such as an increase in cardiovascular events with coxib treatment,26 may well be of much greater relevance to public health, but their reliable detection requires evi￾dence from studies that are both randomized and controlled.1 Consequently, a more effective strategy for safety monitoring in randomized, controlled trials would be based on the regular review of the emerging safety data that is cus￾tomarily conducted by independent data and safety monitoring committees, with the study￾treatment assignments revealed and considered in the context of the efficacy results. The degree to which such safety information is shared with regulatory authorities could be agreed on before the start of the trial and could depend on a num￾ber of factors, such as the amount of previous experience with the treatment being studied. The cogency of these concerns about the present approach to safety monitoring has been recog￾nized by the FDA, which has issued revised guidance for reporting requirements.25,27 In par￾ticular, to reduce current levels of overreporting of alleged serious drug reactions, the FDA seeks rapid reporting of such an event only when there is considered to be a reasonable possibility that it was caused by the study drug. The FDA guidance also distinguishes between the rare circumstanc￾es in which it is appropriate to submit individual case reports and the more common circumstanc￾es in which cases should be aggregated and com￾pared with those in a control group. However, the European Commission guidance and the pro￾posed regulation from the European Union have not yet made this important distinction. In conclusion, some regulatory agencies have responded positively to the need for improve￾ments in the regulatory environment for clinical trials. However, the effect of those responses may be limited by the complex system of regula￾tion and related bureaucracy that applies to clin￾ical trials at local, national, and international levels. Those responsible for the ICH-GCP guide￾lines have been resistant to engaging adequately