Abstract
Keywords
Drugs approved and used for treatment of stroke vary from country to country. This is, at least, partly because of variation in the standards of evidence that is required for approval of drugs across countries. For example, some countries, usually low and middle income ones, approve a drug on the basis of a small positive study or post hoc subgroup analysis in a overall negative study, while others, usually high income countries, require confirmatory or replication studies before approving the same drug. Needless to say that such confirmatory or replication studies need to be published and adequately disseminated as soon as possible so that approval decisions can be revised in the light of additional evidence. However, this does not always happen. The new evidence often remains hidden in the form of nonpublication or abbreviated publication. Sometimes, it is published but with undue delay. What is worrisome is that these problems occur more often when the new evidence does not favor the new drugs than when it favors them, a manifestation of what is called ‘publication bias.’ This bias is well documented in the stroke research literature (1, 2).
Effects of publication bias
Publication bias involving stroke trials may perpetuate the use of ineffective drugs and compromise the decisions made on a daily basis for thousands of patients with stroke around the world. Often these decisions affect health, income, and even survival of the patients. Clearly, publication bias is of public health concern. Identifying its root causes and eliminating it, or at least controlling it, should increase our confidence in making unbiased decisions that are based on the totality of existing evidence, not just some of it, and protect people from avoidable losses of health and income.
Factors underlying publication bias
There are many factors that underlie publication bias: lack of statistically significant findings, disinterest on the part of the authors, lack of time, unimportant results, and co-investigator or other operational problems. One reason of particular interest is pharmaceutical sponsorship. While there is mixed evidence linking nonpublication or delayed publication to pharmaceutical sponsorship, one study (3) found that among pharmaceutically sponsored stroke trials, the interval between enrollment completion to submission tended to be shorter for studies with beneficial trends in outcome than for those with harmful or neutral trends (1·6 vs. 2–3 years, P = 0·07), although the sample size for this analysis was only 13. If this is true, then patients with stroke may receive ineffective or harmful treatments before publication of the studies showing that the treatment is ineffective. There are many examples but two may suffice to illustrate the point: piracetam and fraxiparine in acute ischemic stroke.
Examples
Piracetam has been approved for use in acute ischemic stroke in some countries, mostly developing countries, like India. This was based on a positive post hoc subgroup analysis in the piracetam in acute stroke study (PASS). Another randomized trial, PASS II, began in 1998 to confirm this subgroup effect, and the study plan envisaged enrollment completion in 2000. Eight years later, we are still waiting for the results of this study. A Cochrane review on the topic states: ‘this study was interrupted by the sponsor after a futility analysis; no results were made available to the scientific community. Requests for data, both to the main investigator and the sponsor, were unsuccessful’ (4). This suggests that the results may be neutral or possibly harmful, but thousands of patients with acute ischemic stroke continue to receive piracetam. This means that many patients with stroke are receiving an ineffective and potentially harmful treatment. This situation is even more disturbing because in many countries where piracetam is in use, the patients have no insurance cover and have to pay for it themselves. Often these patients are poor and slide deeper into poverty after every illness.
The Fraxiparine in Ischemic Stroke (FISS) trial involved 312 patients with acute ischemic stroke, the results suggested that fraxiparine improved neurological outcome, and it was published in the New England Journal of Medicine (5). A larger trial, FISS-bis, failed to confirm the findings, but unfortunately has been published only in an abstract form in 1998; not surprisingly it has received little publicity, certainly very little compared with the publication of FISS in a prestigious medical journal. Hence, thousands of patients with acute ischemic stroke continue to incur the risks and costs associated with the use of this drug without any certainty at all that it improves their neurological outcome.
Suggested actions
It is evident that publication bias affecting stroke trials harms patients. To avoid this, several considerations seem necessary. First, it is ethically imperative that the involved physician investigators publish the results of clinical trials fully and promptly to avoid unnecessary risks and cost to patients (6, 7). To facilitate this, clinical trialists should discuss ownership of data with industry sponsor, and create a publication plan ahead of time, and publish the results irrespective of whether they are positive, neutral, or negative. Second, it is obligatory for pharmaceutical companies to dispel any concern that they might have a role in delaying the full publication of the results of clinical trials in stroke. Third, local scientific and professional organizations should be responsible for disseminating evidence-based guidelines and practices of medicine. Needless to mention that full and timely publication of trial results provide the evidence on which these guidelines and practices ought to be based.