It's a story as old as medicine. When it comes to treatments, people don't always obey the written word. When it comes to antibiotics, for example, people may stop taking them when they feel better so they can save them for another incident. 

Prescription medication guidelines are written specifically, to help people get the effect. So why do only 50 patient of patients take prescription medication as they should, 160 years after medicine became a proper field?

The problem is deeper than blaming the patients. Even many studies that examine why patients don't take medication have too many flaws in their design to be valid, according to a recent review of 182 trials that were testing different approaches to increasing medication adherence and patient health. 

 Most trials were unreliable casting doubt on the validity of the results instead. Out of 182 trials, only 17 were of high quality, and each of these tested combinations of several different approaches, such as support from family members or pharmacists, education and counselling. Still fewer, only five of these 17 showed improvements in health outcomes for patients, as well as in medication adherence. 

This is not unique to medicine. When environmentalists a decade ago insisted a pesticide was an endocrine disruptor and caused feminization of male frogs, based on a claim made in a journal, the EPA was also only able to use less than 20 studies, and only 1 met their highest standard. The original paper making the claim had to be disqualified because the researcher refused to share his data.

"This review addresses one of the biggest challenges in health care," said Dr. David Tovey, Editor in Chief of the Cochrane Library. "It's a real surprise that the vast amount of research that has been done has not moved us further forward in our understanding of how to address this problem. With the costs of health care across the world increasing, we've never needed evidence to answer this question more than we do now."

The authors have now decided to turn to the research community to help understand the issues. They have created a database of the relevant trials and made this available to other researchers in the field in order to encourage collaboration and more in-depth analyses on smaller groups of trials.

 "The studies varied so much in terms of their design and their results that it would have been misleading to try to come up with general conclusions," said lead researcher Robby Nieuwlaat of the Department of Clinical Epidemiology and Biostatistics of the Michael G. DeGroote School of Medicine at McMaster University. "Based on this evidence, it is uncertain how adherence to medication can be consistently improved. We need to see larger and higher quality trials, which better take in account individual patient's problems with adherence. 

"By making our comprehensive database available for sharing we hope to contribute to the design of better trials and interventions for medication adherence. We need to avoid repeating the painful lessons of adherence research to date and begin with interventions that have shown some promise, or at least have not produced repeatedly negative results."