I am commenting about some of the responses to my previous message.

In one response to my opening remarks, Dr. Gardenier commented that most commenters wanted the guidelines expanded, not narrowed. While it is good to have the guidelines open to public comment, using their reaction may be misleading. It is similar to using the results of a call-in poll conducted on a radio station. Only certain individuals care to respond. I would expect that most ASA members have not even read them. Perhaps there should be some 'field testing' to ask uninterested parties from a broad range of areas to review and comment on the guidelines. I suspect most people don't care because the guidelines have no enforcement. However, in his response to my comments in II.A.8, Dr. Gardenier states that he uses these ethics guidelines when dealing with lawyers to provide expert testimony. If I am called to testify and haven't provided my client with alternative statistical approaches (as an example), I do not want to be considered as unethical. I think there have been some comments about this particular item but my point is that there are differences of opinion about what a valid analysis is. I know you try to say that in other places, I think this is lost in the particular point.

I still believe that the good professional citizenship items (in I.C.5-10) are not ethics and don't believe that they belong in an ethics guideline.

While I understand his response in II.A.2, I believe that the statement is not clear and rambles. There are suggestions sprinkled in the responses that I should provide alternative wording if I don't like the original. While I often try to do this, I simply don't have the the time to do this. At the same time, just because I don't or can't provide an alternative wording, doesn't mean that my comments are not valid.

My revised wording for II.A.e is "Use statistical analyses appropriate for the data." This is a statement that similar to what I used to see in many clinical protocols. It is a hollow statement. It is a matter of competency and it depends who is judging the appropriateness.

In II.A.8, I still question the use of the word 'proud'. I believe that you would satisfied if your testimony was judged by group of your peers.

Although I am not a researcher is statistical methods, I object to your statement that a person wouldn't release a statistical methodology into the public domain is a case of personal greed. Anyone who works to develop a new methodology deserves protection for his/her efforts. As I stated in my comment, the methodology would have to be public to some degree for others to understand how it works and become comfortable with it. Otherwise, no one will use it.

I was interested in reading your response to II.C.3. I read a number of medical journals and am amazed to see rather sophisticated statistical results and find only MDs in the author list. I sent a suggestion to the ASA President a couple of years ago requesting that the President send a letter to the editors of medical journals and ask that the person who performed the statistical analysis be identified. I believe that this would solve the problem about others misinterpreting the results because the person who performed the analysis would have to take responsibility for the analyses. I haven't required that a statistician perform the analysis only that the person who did them be identified. In the long run, it will help produce better analyses because an MD may now feel uncomfortable in taking putting his/her name at that person if they have no training in statistics. Once a statistician is identified, he/she will be much more insistent in ensuring proper interpretation of the statistical results.

After reading his response in II.C.5, I will ask why it is limited to 'statistical' publications. I have analyzed many studies which have been published and have not been an author or even acknowledged. My involvement included the planning of the study as well as the analysis so it seems as if I should have been an author. However, in the pharmaceutical profession there are many who believe that if the publication includes authors from a company, the results are biased. I have seen more examples published by physicians that totally misrepresented the study design or make some exploratory result the primary one. I believe that this statement should be generalized if it is included.

II.C.5. I believe it is clearer to delete the word 'considered'. While I understand your point, this is more related to qualifying the data or providing a disclaimer about the results, e.g., the number of poor may be under reported because of enumeration.

II.C.6. After reading your response I have a larger question. What do you mean by 'assessed adequacy' of the data? I originally thought your were asking how I knew the data were accurate, i.e., steps to ensure quality data. However, adequacy could also mean whether I had enough data. It could also mean whether the data were the appropriate measures to answer the research question. For example, I might want to study 2 drugs to see if they are equally safe. I could use final temperature as the data point to collect. Is that an adequate measure of overall safety of a drug. I think not. By the way, I asked several people in my department what they thought the statement meant and they came to different interpretations.

II.C.9. I completely misread this statement until I read your response. I read it as in 'volunteer' subjects (healthy) not patients. Now I see that you mean 'volunteered data', such as phone calls in response to a survey on radio or TV. I think this needs to be made clearer.

II.D The point is that I don't need to 'know' the guidelines to be ethical. There are procedures set up to ensure that. While I might know the FDA guidelines, some states have special guidelines. I may be aware of them but not 'know' them. Why not revise the statement to be assured that appropriate guidelines for protection of human subjects. Then my company's SOPs requirements and the study protocol saying that they will follow these protections should suffice.

II.D.3. The point is that if you protect the privacy and confidentiality of research subjects then why do you need 'and the data they provide'. As long as the subject can't be identified (directly or indirectly, e.g., what is the salary of all left handed female statisticians who are 63 years old (and you know there was only one person who met that qualification)).

II.D.5. While you didn't like my example, I would remind you that in II.A.2, you want us to avoid biases in the data. In one example, I mentioned, there was a clear expectation about what would happen when. Whether this was because of the investigators bias or the patients expectations, I don't know as I wasn't involved in that trial. Was there another way around this deception, I don't know. Nevertheless, this was useful in this situation.

I will end for now.

• Re: Comments on ASA Ethics Guidelines John Gardenier 16:36:51 10/06/98 (0)