Randomized drug studies don't tell full story — (BMJ)

Original Article no longer available

British Medical Journal

By Mark Lemstra, Health Issues

August 18, 2011

A limitation with most randomized trials is that they simply test the efficacy of a drug under ideal conditions with ideal participants.

Although this is a necessary step to determine the efficacy of a drug, it provides little information about the effectiveness of a drug in the real world. For example, in most randomized trials, participants are only included if they do not have social problems, they have only one medical disorder and they are not taking any medications.

In the real world, patients have a range of social issues, they have multiple medical disorders and they are taking an average of 6.5 medications at any one time. As well, participants who volunteer for studies are much younger than those actually seen in clinical practice.

All of that said, one of the biggest concerns with most randomized trials is that they are too short to study adverse events associated with medications.

This coincides with the reality that pharmaceutical companies now fund most drug studies and negative results are normally not published.

So how would you determine negative outcomes from a medication if you wanted to?

A study published in the British Medical Journal took 570 general medical practices and followed 60,746 older patients that were diagnosed with depression for an average of five years. The age of the group ranged from 65 to 100 with an average age of 75.

The authors were able to stratify the group by depression severity and statistically control for other factors such as demographics, the presence of other diseases, other medication use and so on. A comprehensive electronic medical information system was used along with the research database to ensure the data was as thorough as possible.

The authors also used a powerful statistical model called a Cox proportional hazards model to ensure they adjusted for patients that stopped treatment, changed treatment, had different durations of treatment etc.

The three main groups of interest were elderly patients with depression not taking antidepressants, those taking a newer class of antidepressants called selective serotonin reuptake inhibitors (SSRI’s) and those taking an older and much less expensive class called tricyclic antidepressants (TA’s).

The first analysis compared elderly patients with depression not taking antidepressants to seniors with depression taking SSRI’s.

For those taking SSRI’s, in comparison to those not taking antidepressants, the increased risk of mortality was 54 per cent while the risk of attempted suicide increased by 116 per cent, the risk of heart attack increased by 15 per cent, stroke by 17 per cent, falls by 66 per cent, fractures by 58 per cent, upper gastrointestinal bleeding by 22 per cent and seizures by 83 per cent.

The second analysis compared elderly patients with depression taking the newer class of SSRI’s in comparison to the older class of TA’s. When elderly patients took SSRI’s, in comparison to TA’s, the risk of mortality increased by 32 per cent while the risk of attempted suicide increased by 27 per cent, stroke increased by 15 per cent, falls by 27 per cent, fractures by 26 per cent and seizures by 80 per cent.

There were no differences in rates of heart attacks between the two drug classes.

There is a clinical basis for each increased risk observed. For example, the increased risk of heart attacks after taking SSRI’s is believed to be due to the increased risk of bleeding complications because these drugs limit uptake of blood serotonin by platelets, which leads to inhibition of platelet aggregation, which prevents homeostasis.

This mechanism could also explain the increased risk of stroke.

These results need to be put into context with a systematic literature review from the Cochrane Collaboration that found that antidepressants are no more effective than active placebos in treating depression. As well, the number of antidepressant medications filled in Saskatchewan almost doubled in the past ten years to 798,157 in 2010 from 430,692 in 2000. During that time, the prevalence of depression in Saskatchewan increased to 7.6 per cent from 5.7 per cent.

Depression is a serious disorder and should never be minimized. The good news is that a systematic literature review that summarized 132 studies found that psychological interventions improve the chance of recovery by a three to one ratio in comparison to other treatments.

Even if the drug treatment model is retained, more depressed patients should be using much less expensive and safer TA’s instead of SSRI’s, and the TA’s should be prescribed in lower doses. This evidence-based strategy would reduce antidepressant costs from $40 million to $13 million per year and would reduce the negative outcomes associated with these drugs.