----- Original Message ----- From: "mackenzie" <[log in to unmask]> To: <[log in to unmask]> Sent: Monday, April 03, 2006 3:04 AM Subject: Re: DAs & Gambling: The Non-Story, Part III - addendum II [snip] > I think I might hear a collective sigh of relief when I > say that i think I am done debating this - will let you > all know if there are any new developments. As it happens, I have just gotten around to commenting on this issue, so Mackenzie may not be able to keep this promise. (Not that the discussion has stopped in the last few days anyway.) The basic argument made by Mackenzie is that the rates of pathological gambling reported in the various studies are comparable to those observed in the general population, the vast majority of whom do not have PD and are presumably not being treated with dopamine agonists (DAs). Certainly, the rates for patients being treated with DAs in these studies are not statistically significantly greater than rates in the general population. I will argue that this is not an appropriate comparison and does not tell whether or not the purported effects are real. Simplifying somewhat, tests of statistical significance for occurrence rates (technically, prevalence) are used when there are two population groups who may or may not differ on some determinable characteristic. Here, let's assume the characteristic of interest is dichotomous -- you either have it or you don't. For example, we could be studying whether members of two particular ethnic groups are more or less likely to have completed a college degree. The basic procedure is to select random samples from each group and to determine whether each of these individuals has completed a college degree. Now, since we have only a limited number of people in each sample, the fraction with a college degree will differ somewhat from what we would get if we had enough money to ask every member of the two ethnic groups. Further, it could happen that, just by chance, we happened to get a few extra college grads from Group A and a few less from Group B. When we see a higher rate of college grads in Group A than in Group B, we don't know whether it's due to such a random fluctuation or whether there really is a difference in the two populations. A test of statistical significance will tell us whether the difference in rates is so large that it is not likely to be due to chance fluctuations. If the difference is not statistically significant, we don't know whether there really is a difference or not -- we could select larger samples from the two groups and maybe resolve the issue. You can never prove that there is no difference, just that it is too small to be detected in the sample sizes we have available. If there is a statistically significant difference, all we know is that it is not likely to be due random fluctuations in who we happened to get in our two samples. It does not mean that the difference is large enough that we should care, and it does not tell us anything about the reasons for the difference. Now, back to the issue of DAs and pathological gambling by Parkinson's patients. The main problem with the statistical comparisons proposed by Mackenzie is that we do not have two population groups who may differ on some characteristic. Rather, we have two groups, people with Parkinson's and the general population, who have two different characteristics. The studies cited have identified a number of people whose pathalogical gambling behavior began after taking or increasing the dosage of certain drugs and stopped when the drugs were stopped or doses reduced. In many cases, the gambling began with a few months of treatment with DAs (although there are some cases of longer periods), and it stopped typically within a few months of stopping or reducing the drug treatment. By contrast, in the general populaton, pathological gambling usually starts in adolescence or early adulthood and builds up over many years until it reaches the point of being "out of control". There typically does not seem to be any particular event that sets off the behavior, and it continues until treatment is sought and is successful. Treatment is usually similar to that for alcohol addiction. Overcoming it is a very difficult struggle. Thus, it is of no particular interest whether the occurrence rates of these two different behavior patterns in two different populations are about the same or one is larger than the other. It *is* of interest whether this is a real effect -- whether a few percent or so of patients treated with DAs (and perhaps Mirapex in particular) will develop pathological gambling behavior -- regardless of what people do who do not have PD and are not taking dopaminergic drugs. Of course, it is also of interest what the mechanism of causation is and how it is related to the drug-treatment regime and to characteristics of the patients. For example, does it really occur more frequently with treatment with Mirapex, or do the apparently higher occurrence rates with Mirapex reflect more widespread use of that drug? In addition to the difference in the endpoints observed in the PD/DA group and the general population, our situation differs from the model of statistical significance in that it is quite difficult to determine whether a particular individual in either group has the behavior. People with pathological gambling behavior typically lie about it, and it might require an in-depth relationship between the investigator and the patient. This would lead to under-reporting of the number of cases; even worse, the probability of missing cases of the behavior may be different in the two groups. In addition, as Mackenzie points out, the publicity devoted to this issue could lead to greater awareness of the problem and, hence, a jump in the rate at which cases are detected. Another uncertainty is the role, if any, of levodopa in causing this behavior. The recent studies do not seem to have identified cases where patients who were treated with l-dopa alone, without any DA, developed the gambling behavior. On the other hand, most of the cases identified involved treatment with l-dopa *and* a DA; it is quite possible that both play a role in development of the gambling behavior. As far as I know, nobody (except maybe some lawyers) believes that a role for l-dopa has beeen ruled out and that we can be sure that it's DAs alone (or Mirapex in particular) that are the problem. In fact, the article posted on April 6 by Maryse seems to suggest that the problem is more likely if the patient is taking both DAs and l-dopa, but I have not seen the full report. So where does this leave us? I would suggest the following: (1) There is substantial evidence that the behavior pattern we are discussing really does occur, though in a small portion of patients, perhaps one percent to a few percent. The evidence is less convincing than a randomized double-blind trial but goes well beyond isolated anecdotal reports. The similarity of the pattern observed in these studies -- pathological gambling within a few months of starting on or increasing the dosage of DAs and a rapid end to the bahavior when the drug is stopped or dosage reduced -- lends credence to this conclusion. (2) Therefore, doctors, patients, and care givers should be alert to the initial signs of pathological gambling or other impulse-control problems and take appropriate action when they occur. (3) It will be difficult to do a definitive study -- a prospective randomized double-blind trial. The subjects would need to be informed about the purpose of the study, which will influence their actions. I would also question whether it is ethical to withhold DAs from PWPs for long periods. ---------------------------------------------------------------------- To sign-off Parkinsn send a message to: mailto:[log in to unmask] In the body of the message put: signoff parkinsn