The ‘Goldilocks’ retraction? Revealing differences in how several neurology journals handled related problems

Four neurology journals have retracted articles by a Japanese researcher who admitted to having made “mistakes” in his handling of data. Although the cases are related, the way the journals have handled the  notices is startlingly different. One chose to say nothing, one chose to say little, while two went for full — or at least, approximately that — disclosure.

Guess which ones we like the most?

Here’s one of the good notices, which appeared in the Journal of the Neurological Sciences (Note: the timing of all this is ambiguous. It seems the retractions came in 2008; however, the JNS notice appears in the October 2011 issue, perhaps because it was posted online initially and only now is making it into print. We’re trying to sort it all out). It starts with a letter from the researchers, Hisayoshi Oka, of Jikei University School of Medicine in Tokyo (links added), to Robert Lisak, who edits the journal.

Dear Prof. Lisak:

I am writing concerning the retraction of our papers entitled “Cardiovascular dysautonomia in de novo Parkinson’s disease” J Neurol Sci 2006; 241:59–65 and “Cardiovascular autonomic dysfunction in dementia with Lewy bodies and Parkinson’s disease” J Neurol Sci 2007; 254:72–77.”

The investigative committee founded in the Jikei University School of Medicine was organized in February 2008. This committee found some serious errors as follows. In control groups, some subjects who underwent the Valsalva maneuver and head-up tilt table test mistakes correspond to those who underwent MIBG scintigraphy. I also made mistakes in the calculation and copying of the data. These errors were made by me, without the knowledge of the other authors. These problems are considered serious errors. The investigative committee concluded that these errors in dealing with the data did not occur intentionally.

Therefore, we wish to retract our article entitled “Cardiovascular dysautonomia in de novo Parkinson’s disease” J Neurol Sci 2006; 241:59–65 and “Cardiovascular autonomic dysfunction in dementia with Lewy bodies and Parkinson’s disease” J Neurol Sci 2007; 254:72–77.” We sincerely apologize to you and your readers for the trouble.

The 2006 paper has been cited 31 times, while the 2007 one has been cited 17, according to Thomson Scientific’s Web of Knowledge. What follows is an editorial from Dr. Lisak:

A reader of the Journal of the Neurological Sciences wrote to me and pointed out some issues in the two articles published in the Journal by Dr Oka and colleagues. At around the same time I became aware of similar issues raised about papers by Dr Oka published in several other journals. Dr Oka was contacted about the issues by me as well as by the editors of the other journals. Dr Oka’s response related to the two articles published in the Journal of Neurological Sciences is provided above and these two articles are retracted. Apparently there have been retractions of articles from some of the other journals.

We regret any inconvenience to our readers. We attempted to maintain the highest quality and ethics in the Journal, as do the other journals in question, and thank our readership when they detect problems that were missed by our reviewers during the confidential peer review process.

A similarly enlightening notice appeared in Movement Disorders in October 2008 for “Impaired cardiovascular autonomic function in Parkinson’s disease with visual hallucinations,” which has been cited 8 times:

The investigative committee found some serious errors as follow. In control groups, some subjects who underwent the Valsalva maneuver and head-up tilt table test did no [sic] correspond to those who underwent MIBG scintigraphy.

The age of the controls (65.7±7.0 years) that was stated in PATIENTS AND METHODS is for MIGB scintigraphy. The age of controls for the Valsalva maneuver and head up tilt table test also should have been mentioned in the article.

I also made mistakes in calculation and copying of the data. These errors were made by myself, without the knowledge of the other authors. These problems are considered serious errors. The investigative committee concluded that these errors in dealing with the data did not occur intentionally.

Therefore, we wish retract our articles entitled ‘Impaired cardiovascular autonomic function in Parkinson’s disease with visual hallucinations’ 2007; 22:1510–1514 in Movement Disorders. We sincerely apologize to you and your readers for the trouble.

Yours sincerely,

Hisayoshi Oka, MD, PhD

Masayuki Yoshioka, MD, PhD

Kenji Onouchi, MD, PhD

Masayo Morita, MD

That followed an exchange in print between Oka and Christopher Goetz, then editor of Movement Disorders, earlier that year:

I am writing about your recent article published in Movement Disorders on cardiovascular autonomic function in Parkinson’s disease subjects.1 I have been alerted that data in this article and one published by you in Neurology (2007) have identical MIBG Delay Heart to Mediastinum ratios in the control groups, though the numbers and ages of these control subjects are quite different in the two articles. Further scrutiny of your articles in Acta Neurological Scandinavia (2006), and Journal of Neurological Sciences (2006, 2007) documents a similar pattern of results, with different numbers of control subjects and different mean ages, but each time, the MIBG data for early and delayed phases are exactly the same among these three latter articles.

I am not an expert in your technology, but from many years of dealing with scientific data, I consider the likelihood of data at the hundredth decimal repeating themselves in two or even three studies quite surprising. In this light, my co-editor, Guenther Deuschl, and I would appreciate your response to these observations and a specific comment on the accuracy of the information printed in our journal. Please check your data set again and respond to me without delay. It is not my business to formally request that you check the data sets from the other journals, but, in best scientific interest, it might be a good idea. I look forward to hearing from you.

Christopher G. Goetz, MD

To which Oka replied:

Thank you for your e-mail and your valuable advice in January 30, 2008.

On the re-analyses of individual data sets previously reported, I had realized that the age and the MIBG values for the control groups were incorrectly reported in Acta  Neurol Scand, (2006), J Neurol Sci (2007) and Neurology. Therefore, on January 9, 2008, I had already informed those journals about our mistakes and have requested them to correct the data. The document I sent to those journal offices is attached with this e-mail.

The number of the patients served as control group has been gradually accumulated and was appropriately selected. In your edited journal the number of control group was 33. Anyway all of the actual MIBG values in Movement Disorders as control are shown in the table attached with this e-mail.

Please make sure that the age and mean value of Early and Delay H/M were correctly described. Even though those values were the same of previous data coincidentally, one mistake I have to state following the re-analyses, were the standard deviation (SD) of Early and Delay H/M, so should be revised to 0.22 and 0.27 from 0.21 and 0.26, respectively.

I must apologize for this error. I do not think this revision alter the conclusions of the study. I have already reported these mistakes to co-authors. I am sorry for causing so much trouble.

Yours sincerely

Hisayoshi Oka, MD

Tokyo, Japan

The result was the following erratum (which was obviated by the subsequent retraction — there will be an open-book quiz at the end of all this):

Oka H, Yoshioka M, Onouchi K, Morita M, Mochio S, Suzuki M, Hirai T, Urashima M, Inoue K. Impaired cardiovascular autonomic function in Parkinson’s disease with visual hallucinations. Mov Disord. 2007;22:1510–1514.

The control group data were incorrectly reported in the above named article. (see Letter section). The first-named author of the article considers that in spite of the error, the conclusions of the article remain unchanged.

We praise Goetz and Lisak for airing their communications with Oka. Not only does it provide an accounting of what went wrong — and an opportunity for readers to judge for themselves the plausibility of the author’s explanation — but we’re guessing it goes a long way toward building trust between the journals and their readers.

We can’t say as much for the other journals involved in this affair, Neurology and Acta Neurologica Scandinavica. Unfortunately, while Lisak might be right that those publications “maintain the highest quality and ethics,” they don’t appear to share his commitment to transparency.

Here’s the retraction notice from Neurology, of a paper that has been cited 13 times:

Authors’ voluntary retraction: The authors of the following article recently published in Neurology® wish to retract the article because the data contain serious errors: Oka H, Yoshioka M, Morita M, Onouchi K, Suzuki M, Ito Y, Hirai T, Mochio S, Inoue K. Reduced cardiac ¹²³I-MIBG uptake reflects cardiac sympathetic dysfunction in Lewy body disease. Neurology 2007;69:1460–1465.

Serious errors? Well, in that case, surely readers will want to know if any of the authors’ other papers are affected, right? Guess not.

But we reserve the most scorn for the folks at Acta Neurologica Scandinavica, who deigned only to print the following:

Cardiovascular dysautonomia in Parkinson’s disease and multiple system atrophy [This article has been retracted]

That paper has been cited 16 times.

It’s hard to imagine Oka didn’t offer each journal the same kind of letter, or at least the same information about the investigation. And yet their approaches were completely different.

We’ve done this sort of report card before, with the Bulfone-Paus retractions, and will continue to look for such comparisons.