Data is everywhere and ever expanding beyond the usual sources into real-world data, social media data and big data. As the amount of data continues to grow, it increasingly makes the headlines, most notably in recent months in relation to transparency, or the lack of it, in the pharmaceutical sector.
In January 2014, the UK parliamentary investigation into the Department of Health’s decision to buy £424m (around $700m) of Tamiflu for use in a potential influenza pandemic concluded that this decision was based, in part, on incomplete clinical trial information. It’s just one in a long line of recent complaints from those who think pharmaceutical companies and clinical researchers should share all clinical trial data at their disposal.
However, the answer is not simply to make this data available with no thought for how it will be used. Information derived from data is only valuable if it is unbiased, not sensationalised, and not used to promote a particular agenda. This is especially critical for information concerning drug safety.
To take one example of how data can be misused: how often have you seen the claim that 100,000 or more hospitalised patients die each year in the US from adverse drug reactions (ADRs)? The probable answer is ‘frequently’. This much-cited statistic comes from an analysis that aggregated results from 30 separate studies of hospitalised patients in the US published between 1964 and 1996. Three-quarters of the 16 studies that reported on fatal ADRs were published before 1977 and the rest after 1987. In the pre-1977 studies, the occurrence of fatal ADRs was five times that in the post-1987 studies, but in the statistics released to the public, this was not taken into account. In fact, if only the studies published after 1987 had been used, the statistic would be 13,250 deaths per year, rather than 100,000. This is just one example of analysts inappropriately using outdated information and applying it to the present day.
Misuse of data also occurs when safety problems found in multiple RCTs are aggregated. This is because the means of selecting studies for inclusion or exclusion may lead to biased results. For example, an aggregate analysis of reports of acute myocardial infarction (AMI) in rosiglitazone randomised clinical trials (RCTs) excluded those RCTs that had no reports of AMI (ie those that demonstrated a positive result) and found a statistically significant risk of AMI in rosiglitazone patients compared with users of other oral anti-diabetic drugs. However, when the RCTs with no AMI reports were included and recommended statistical methods used, no significant risk was observed.
In the US and Canada it has been suggested that less-safe drugs have been approved by drug regulators in the last decade than in the previous decade, but this is just one more example of the irresponsible use of data. This claim is based on the assumption that the increase in the probability of a new drug acquiring a serious safety warning or being discontinued for a safety reason means less-safe drugs are being approved. But this simply isn’t the case.
Safety warnings are cautions, not actual health outcomes. They are not the same as studying the effects of the drugs in everyday clinical use. The only hard evidence about serious drug safety risks is the rate of drugs discontinued for a safety reason out of the number approved in the same period, and this rate has decreased. Approximately 2 per cent of drugs approved in both countries between 2002 and 2011 were discontinued for safety reasons, which is lower than the rate of almost 3 per cent for drugs approved in the previous decade. It is inappropriate and misleading to suggest that the increased use of serious warnings means that less-safe drugs are being approved. Drugs are not approved unless their benefits outweigh their risks.
Calls for greater access to trial data have been made for years, but now they have begun to be answered
Clearly the issue of data accessibility has become a priority for industry. Recently the two largest trade associations representing the research-based pharmaceutical put forward their proposals to make data more accessible to bona fide academic researchers.
Calls for greater access to clinical trial data have been made for years, and still persist, but now they have begun to be answered. Of course, this is just the first step, and it will be imperative to ensure that data shared by pharmaceutical companies beginning in January 2014 – particularly data related to drug safety – is used responsibly. Sensational or alarmist statements based on the irresponsible use of data on drug safety raise fears in patients that may lead them to discontinue treatment and negatively impact their health, which benefits no one.
