Question

3. (a) There is a rare neurological disease (idiopathic hypguesia) that makes food taste bad. It is sometimes treated with zinc sulfate. One group of investigators did two randomized controlled experiments to test this treatment. In the first trial, the subjects did not know whether they were being given the zinc sulfate or a placebo. However, the doctors doing the evaluations did know. In this trial, patients on zinc sulfate improved signifcantly the placebo group showed little improvement The second trial was run double blind: neither the subjects nor the doctors doing the evaluation were told who had been given the drug or the placebo. In the second trial, zinc sulfate had no effect. Should zinc sulfate be given to treat the disease? Answer yes or no, and explain briefly. (b) The second trial used what is called a "crossover design. The subjects were assigned at random to one of four groups: placebo placebo placebo zinc zinc placebo ziI zinc In the first group, the subjects stayed on the placebo through the whole experiment. In the second group, subjects began with the placebo, but halfway through the experiment they were switched to zinc sulfate. Similarly, in the third group, subjects began on zinc sulfate but were switched to placebo. In the last group, they stayed on zinc sulfate. Subjects knew the design of the study but were not told the group to which they were assigned. Some subjects did not improve during the first half of the experiment. In each of the four groups, these subjects showed some improvement (on average during the second half of the experiment. How can this be explained?

Answer 1

Part A:

A good design of an experiment has 3 basic properties. They are

• Randomization (Randomization is the random process of assigning treatments to the experimental units)
• Replication (Replication is the repetition of the basic experiments)
• Local Control(Not must, Local Control means bringing all extraneous source of variation under control)

Replication is achieved here by giving both Zinc Sulfate and Placebo to a group of patients.

Local Control is not achieved because it is a simple experiment.

Now complete Randomization is not achieved in the 1st trial because the doctors had prior knowledge about the drugs and the patients, which may affect their decision of whom to give Placebo and whom to give Zinc Sulfate.

In the 2nd trial, the experiment is completely randomized because both the subjects and the choice of drugs given is totally randomly decided. Thus this design, 2nd trial, should provide more accurate result than the 1st trial.

So, we conclude that there is not sufficient evidence to consider Zinc Sulfate as a possible cure for the disease.

Part B:

Here we have the information that the subjects know the design of the study.

We see that not all of the patients improve in the 1st half of the experiment.

Then this might suggest that Zinc Sulfate works.

Let us say 2 scenarios,

Case 1: The patient shows improvement in the 1st half and 2nd half:

Since the patient is getting better he/she may be given medication(Zinc Sulfate) and that 'Getting better' carryover effect can continue in the 2nd half irrespective of their drug given in 2nd half.

Case 2: The patient does not show improvement in the 1st half and gets better in the 2nd half:

Since the patient is not showing improvement he/she may be given Placebo. Since getting Placebo both times according to a patient is a low probability(1 out of 4 cases) this may trigger a Placebo effect in the patient.

Feel free to comment if you could not understand something.

Please Upvote if the answer solves your problem.