What does Dr. Eric Klein of the Cleveland Clinic have to hide?
Dr. Klein was the lead researcher of the recent Journal of the American Medical Association study claiming to link vitamin E with increased risk of prostate cancer.
As we pointed out in our coverage of the study,
… it’s not at all clear that data were collected on potential confounding risk factors for prostate cancer. The study, if its omission of potential confounders from mention is any indication, dubiously pretends that vitamin E supplementation is the only risk factor for prostate cancer.
So we followed up by e-mail with Dr. Klein as follows:
JunkScience.com: About your JAMA study on vitamin E/prostate cancer, it does not appear that any data were collected or considered concerning potential confounding risk factors such as family history.
Do you have a comment on that?
Dr. Klein: Yes. We accounted for family history, age, race and frequency of PSA testing and other factors — none of them could account for the findings of an increased risk of prostate cancer in those taking vitamin E.
Surprised that we missed that data in the study, we had the following exchange.
JunkScience.com: Thanks. Where is that in the study?
Dr. Klein: Most of the data is in Tables 1 and 2. We did not have room in the article to include the full multivariable analysis.
Not being able to find any of the data we requested in Tables 1 and 2 and wondering how much space a few relative risks and confidence intervals could possibly take up, we next asked:
JunkScience.com: Can you send me the crude and adjusted [relative risks] and [confidence intervals]?
Dr. Klein: I will ask our statisticians.
After no response for two days, we e-mailed Dr. Klein:
JunkScience.com: I haven’t heard back from anyone.
Dr. Klein: When we adjusted for age at study entry, race (black vs. other), PSA level at study entry and family history of prostate cancer, the hazard ratio of vitamin E vs. placebo was 1.19 for diagnosis of prostate cancer and remained highly significant (compared to unadjusted HR in paper = 1.17).
Getting a nonsensical answer and not what we asked for, we continued:
JunkScience.com: Do you have the confidence intervals and p-values for the 1.19? Also, why then use the unadusted 1.17 when you have an adjusted 1.19?
Ten days later.
JunkScience.com: I never got an asnwer from your statisticians.
On October 25:
Dr. Klein: We discussed this request in a recent call and have decided to let the data stand as is. It’s clear form the JAMA publication how well balanced the groups are.
Frustrated at being stonewalled, we asked:
JunkScience.com: What’s the big secret?
Dr. Klein: None. Not sure what you are trying to prove.
JunkScience.com: I am asking to see the results of the crude and adjusted analyses (RRs and CIs). Your write-up only includes the crude RRs and CIs. I am asking for the adjusted RRs and CIs.
Dr. Klein:I understand. Why? They do not change the interpretation of the results.
If they don’t change the results, then why not share them, Dr. Klein? What’s the big secret?