It says something about the overheated nature of some biopharma stocks that even the hint of an adverse event can cause investor panic and a precipitous share price drop.
Witness Friday’s sharp dips in Sarepta, Ionis and Biogen, which were entirely due to unfounded fears over deaths revealed in the US FDA’s Adverse Event Reporting System. In hindsight the agency’s move to open public access to the database looks ill-conceived, and investors would do well to heed the lessons of looking at data in the absence of vital additional context.
The problem lies in the fact that overnight the public was given access to all adverse event data on the US system, in raw format. Ostensibly this suggested that drugs like Sarepta’s Exondys 51 and Ionis/Biogen’s Spinraza were associated with an unexpected number of side effects, including deaths.
The move was part of the US government’s initiative to improve transparency and give open access to the FAERS database, announced by the FDA commissioner, Scott Gottlieb, on September 28.
What had not been spelled out clearly enough, however, was that the adverse events contained in the FAERS were purely as reported to the FDA by clinical centres. The data were not verified, and might well have contained duplicates if, for instance, a side effect had been reported both by a doctor and company sponsor.
Perhaps most importantly there was no effort made to establish causation; an adverse even might be due to the drug in question, or it might be completely unrelated, simply reflecting a patient’s underlying disease. Investors seemed to have ignored the FDA’s caveat that “FAERS data cannot be used to calculate the incidence of an adverse event or medication error”.
As such Sarepta, Ionis and Biogen initially fell as much as 7%, 8% and 2% respectively, before recovering as reason resumed its seat. The data stated baldly that, for instance, Exondys 51 was associated with three deaths and two cases of cardiac arrest, and Spinraza with nine deaths.
There is little doubt that large numbers of people started accessing the suddenly available data, as at one point on Friday the FAERS website, overloaded with requests, went down.
Analysts at Leerink and Mizuho quickly cautioned that the data had serious limitations and required investigation before a full interpretation could be made. Biogen released a statement to the sellside that none of the deaths contained in the FAERS entry for Spinraza “appeared to be related to drug”.
Freedom of information
Until now FAERS data had been available only via freedom of information requests; the company Advera Health Analytics, for instance, submits these regularly, and then curates the results using various algorithms before publishing them to subscribers in a format that attempts to draw firm conclusions about the risk and incidence of each event.
Brian Overstreet, Advera’s president, said there were many issues with opening access to the FAERS, which he said had limitations at best, and that inaccurate conclusions were inevitable.
There is an interesting parallel here with a move 30 years ago by the US Health Care Financing Administration to release hospital mortality data. This drew criticism since the figures were unadjusted for case mix, and made no allowance for hospitals treating very sick patients.
Both cases show that not all datasets are as clear-cut as they might appear at first sight.
Advera Health Analytics is partly financed through venture capital provided by Evaluate Ltd, the parent company of EP Vantage.