As the father of a child with severe classic autism--and as a pediatrician--I understand how much parents want to try different treatment approaches to help their child. Many parents ask me about complementary (or alternative or holistic) therapies, such as restrictive diets, secretin, chelation, and omega-3 supplements. Some inquire after hearing claims that a particular diet or supplement or herb can work wonders. Others ask about these therapies after reading studies on the Internet, in magazines, or in newspapers.
Understandably, many parents of a child with autism desperately want something to work for their child. But we do not want to waste our time and money on treatments that will do our child no good. What we need is evidence --objective and unbiased--of therapeutic benefit. Unfortunately, there is little scientific proof that any specific therapy besides intensive behavioral training will improve the autism. There are no large, randomized, double-blind, placebo-controlled, trials (RCTs) available to help us understand what a therapy can or cannot do for a child with autism. RCTs are the gold standard of scientific research, as I describe here.
Whether to try a complementary or an alternative therapy is essentially a personal family decision--one that needs to be made without "irrational exuberance." I hope the information presented here helps you assess the available evidence about a particular therapy before you decide to try it.
Why is it important for a study to be "controlled"?
When investigators want to study the effects of a therapy objectively, they compare it against a "fake" treatment--or placebo--in 2 groups of patients. The researchers are not told which group is getting the active treatment and which is receiving the placebo. If they were told, their interpretation of the results could be biased because they might wish that the therapy would work better than the placebo.
Why is it important for patients to be randomly assigned to a therapy or to a placebo?
The first step in testing a medicine or a particular diet or other therapy is to choose who will get the "real" treatment and who will get the placebo. Patients should be chosen randomly (by chance)--not at the whim of one person. Random selection eliminates that person's unknown or unconscious bias.
Most studies use some type of rigid selection system to eliminate bias. Let's say that all those whose social security number ends in an odd number would receive treatment, and those with an even last number would be given placebo. Because there is an equal chance of having an even or odd last number, then the groups would be divided roughly equally.
How large does a study need to be for results to be "scientific"?
If the total number of patients in a particular study is small, the treatment groups could be divided unevenly. A disproportionate number of people might get the medicine or the placebo. For instance, if there are only 10 people in a study, 9 might have an odd social security and 1 might have an even number.
This imbalance sometimes occurs when you flip a coin 10 times and it comes up "heads" 9 times out of the 10. However, if you flipped that same coin 100 times instead of 10, the "heads"-to-"tails" ratio would even out.
It's the same in a randomized trial. If there are a small number of patients, the balance between the 2 study groups may be skewed. Unless a study has an adequate number of patients, the results may not tell the true story.
Many of the studies of complementary or alternative treatments lack a control group and include fewer than 20 patients. Consequently, their results can't be considered reliable.
How does the placebo effect figure into a scientific study?
The placebo effect is a fascinating concept. If a person is told that a medicine will make him or her feel better, the mere power of that suggestion can create a powerful positive response. This effect was noted during World Wars I and II when morphine supplies ran low: doctors injected saline or sugar water and told the soldiers that they were being given strong pain medicine. The power of the soldier's belief actually diminished their perception of pain.
The same effect applies to therapies for autistic children. For example, you may have heard or read that if your child sticks to a gluten- and casein-free diet, his or her autism will improve. Your child may actually improve with this diet, or it may only seem that the diet is helping because of the placebo effect. It may also be that you are doing several other things in addition to the diet, such as applied behavioral analysis, social skills training, or verbal behavior, that may be helping your child. In other words, it's impossible to tell whether the diet itself is really helping.
Don't patients know they are receiving a placebo and not the real therapy?
In an RCT, patients aren't told whether they are receiving active therapy or placebo. It can be quite difficult to make placebo and active therapy "look alike." In a study of a gluten- and casein-free diet, for example, the control group (the group receiving placebo) wouldn't actually follow the diet. However, patients might be asked to take a pill (a fake pill) that they are told will absorb all the gluten and casein, and that they can eat whatever they want. The patient's parents will believe that they are still maintaining a gluten- and casein-free diet--when in fact they are not.
Placebo-controlled studies are difficult to conduct because parents need to sign a consent acknowledging that they are aware that their child could receive a fake pill. Such a study may actually require 3 control groups:
•Group 1 would follow a strict gluten- and casein-free diet.
•Group 2 would eat a regular diet and would take a "pill" that absorbs gluten and casein.
•Group 3 would eat a regular diet without restrictions.
Only by eliminating variables can the true effect of a therapy be assessed, but it is not always easy to "control" a study. However, it is important to consider these variables before testing a therapy.
Unfortunately, many treatments for autism are not tested with this in mind; this only serves to create or perpetuate false or invalid information.
What does "double-blind" mean?
"Double-blind" does not mean that researchers have to close their eyes (as my daughter suggested when I brought up this topic). "Double-blind" means that neither investigators nor patients know who is receiving therapy or placebo during a study.
How is this possible? Studies of secretin offer an explanation. The doctors were given vials of secretin labeled "A" or "B." Only one of the vials actually contained secretin, but the doctors were not told which one. Parents weren't told either. Only the pharmacist knew; this information was revealed at the end of the study.
In short, patients with vial A or B are "blinded" to what they are actually getting, as are the doctors and study evaluators. It is extremely important to maintain a double blind throughout the study. However, occasionally, when preliminary data suggest that a treatment is potentially harmful or helpful, the blinding can be removed. *