Tuesday, September 1, 2015

Background Information: "Adverse Events" In Clinical Trials

This blog posting applies to all clinical trials, not just those for type-1 diabetes.

I try to avoid using technical jargon when I describe clinical trials.  I think one of the values of this blog is that I turn technical jargon in to plain English.  (I sometimes joke about providing "researcher to patient simultaneous translation", but it's not really a joke.  It's important.)  However, one technical term that I do use is "adverse event".  "Adverse event" generally means "bad side effect".  Serious adverse event generally means "really bad side effect".  But the truth is more complex than this simplified version, and so I thought I would make a post describing exactly what "adverse event" and "serious adverse event" means, because these terms are defined by the US FDA, and their meaning is the same no matter which clinical trial I'm discussing.

The Basics

The basic definition of "adverse event" and "serious adverse event" comes from the US FDA:
for lots more details:
  • An "adverse event" is any bad effect seen in the study.  The FDA uses the term "undesirable experience".  It does not have to be caused by the treatment!
  • A "serious adverse event" is an adverse event that causes death, hospitalization, persistent or significant disability, or a birth defect.  Or one which requires immediate medical attention to prevent one of these outcomes.
For example, a mild allergic reaction would usually be considered an "adverse event", but if the patient had trouble breathing, that would be a "serious adverse event".

The First Complexity: How Bad Is Serious?

No matter how many rules the FDA makes, and how many documents it publishes, there will always be room for researchers to interpret a specific events as serious or not.  Therefore, there will always be controversies about it.

For example, in one Chronic Fatigue Study there was controversy because several adverse events required patients to go to a clinic (but not a hospital).  Now hospitalization clearly means serious, but the company producing the drug felt that going to a clinic was not a sign of a serious adverse event. When the FDA reviewed the study, they disagreed, and the the larger number of serious adverse events the FDA counted were part of the reason the drug was not approved.

The Second Complexity: Treatment Related Or Not?

In general, researchers are required to track all adverse events, no matter what their cause. The FDA terminology is "associated with" not "caused by". After all, these are experimental treatments and some of the bad side effects might be unexpected. I've seen a trial for an immune system drug, which listed one serious adverse event: a broken arm. Now, I don't think that broken arm had anything to do with the immune treatment being tested.  But it is still a serious adverse event, and must be reported.
A more complex situation is adverse events associated with the disease being treated.  For example, if you are testing a drug for depression, and someone in the study commits suicide, that is clearly a serious adverse event, but is it treatment related?  How would anyone ever know?

This whole area is usually resolved by comparing adverse event rates between the control group and the treated group.  If there are statistically significantly more adverse events in the treated group as compared to the control group, that is the important result.  Arguing that some of these adverse events are not "treatment related" much less important. More adverse events is bad, no matter if the researchers think they are not related to treatment.  But what about those phase-I trials that don't have control groups?  For those trials, arguing about "treatment related" can be important.

The Third Complexity:  Who Decides?

Someone goes through every event and decides if it is serious or not.  The same is true for "treatment related" if that is reported separately. Obviously, this is a human activity and the results will be imperfect, but the exact procedure used can minimize (or maximize) risk of bias.  The two things to look for are blinding and reviewers.

The review can be done "blind" or not "blind".   The reviewer looks at the event, and maybe some data about the person who had the event, but does not know if the person is in the control group or the treated group (or what dose the person got, if multiple doses were tested).  But if the reviewer knows that the event occurred in the control group or the treated group, the risk of bias is more pronounced.  If the study doesn't have a control group, then this review will never be blind.

The reviewing is usually done by the same researchers running the trial. However, it can be done by a different group of doctors, recruited especially for that purpose.  Having a different group lowers the risk of bias, and this is done for some particularly controversial or emotional trials.

Some Discussion and Opinions

Overall, I think we are lucky in the world of type-1 diabetes research, in that the reporting of adverse events is generally not complex or controversial.  Type-1 diabetics are generally pretty healthy, and also the bad complications of type-1 diabetes are generally well understood.  Therefore, there is consensus as to the types of adverse events that are likely related to treatment, and those that are not.

Especially in larger clinical trials, serious adverse events will happen.  So the important thing to look at is: Were there more serious adverse events in the treated group than in the control group?  Also, if multiple different doses were given to different groups of people, do the higher dosed groups see more serious adverse events, or are they randomly spread throughout all the dosing groups?

Since all adverse events must be reported, it is important to consider the impact of different adverse events as compared to the disease being treated.  For example, rashes or mild fevers are common adverse events (not serious ones). Compared to curing type-1 diabetes, these might be well worth it.  On the other hand, in a drug which merely treats type-1, the very same adverse event might cause you to use a different drug.

Because the long term outcomes of type-1 diabetes is relatively well known, it's easy for patients to "trade off" the adverse events seen in testing a cure, to the long term complications of having type-1 diabetes.

One problem in the whole approval process is the issue of very rare side effects, especially those which happen rarely in healthy, untreated people. Take the following situation, you treat 300 basically healthy, basically young, people with a drug. One of them has a stroke. That's a serious side effect. But it is something that happens -- although very rarely -- in young people who are not taking the drug. With 300 people you may not have the statistical power to know if it is a statistically significant event. Are you going to delay availability and require a larger (and very expensive) study just to eliminate the statistical chance that the drug causes stroke? Or maybe approve the drug, but require a strong (ie. "black box") warning about stroke? Or just decide that it was random bad luck, and approve the drug?
(The situation is much worse with diseases like type-2 diabetes.  Was that stroke caused by type-2 diabetes, or the drug given to treat the type-2 diabetes?)

In a sense, the FDA cannot win in these cases, because no matter which outcome they choose, some people will want the other one.  So if the drug is delayed, some patients (and the company involved) will scream loudly about delaying needed treatments and creating unnecessary hurdles to drug approval.  On the other hand, if the drug is approved, another group of patients (and consumer advocates) will yell about approving dangerous drugs so big pharma can profit.  If the drug is given to 10,000 and one of them has a stroke (by chance? or because of the drug?) then recriminations will be deafening.

Joshua Levy 
publicjoshualevy at gmail dot com 
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF or JDCA news, views, policies or opinions. My daughter has type-1 diabetes and participates in clinical trials, which might be discussed here. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog.

No comments: