| Abstrakt: |
Signals detected by measuring disproportionality of drug-event combinations are only statistical indicators of possible real safety issues, and are not per se necessarily medically important. Nevertheless, once a signal is observed, sponsors are obligated by regulations and ethical considerations to determine whether it represents a new product-associated risk by additional analysis, validation and evaluation of its clinical relevance. Signal strength does not necessarily correlate with medical significance. Strong signals most often represent known, expected and/or medically trivial adverse reactions or confounding by treatment indication, common co-morbidities or other common concomitant treatments. Conversely, any product with reasonably extensive clinical use and reporting of suspected adverse reactions is likely to manifest many weak but clinically unimportant signals, creating significant background 'noise'. Since relatively rare, medically important adverse drug reactions are often likely to manifest as weak signals, sponsors face a potentially onerous burden of evaluating multiple signals in order to distinguish true, clinically important events of concern from spurious signals. This paper discusses the regulatory, clinical and potential legal liability issues that confront industry as a consequence of signal identification activities, including: current and anticipated regulatory requirements for detection, assessment and reporting; the reliability of the data used for signal generation; assessment of clinical relevance; organisational approaches and responses to observed signals; targeted clinical and scientific responses to observed signals; and potential regulatory, legal and commercial impact. |
Full text from SpringerLink