First, there was the original article, published last summer in the Journal of Addictive Behaviors — not publicized by ADAW — which said that there was no link between teen marijuana use and any mental or physical problems in adulthood. Then, last month, in a January 19 press release from the American Psychological Association, which publishes the journal, came the correction: there actually is a link between teen marijuana use and psychosis as an adult, which ADAW did write about (see ADAW, January 25).
That’s quite a reversal. But that reversal is now reversed again. The APA now says it “erred” in correcting the article on January 19. “There were no errors in the analysis or results reported in the original study,” the APA said in the February 2 press release, adding that the authors “voluntarily conducted some additional supplemental analysis at the request of APA.”
The first correction was made after schizophrenia researcher Christine L. Miller, Ph.D., alerted the APA to the errors. The APA then issued the correction and said the original conclusion — that there is no link between teen marijuana use and problems in adulthood — was in error — that there was a link between teen marijuana use and psychosis in adulthood. A clarification was also added to the study itself on the last page. That is what we reported on in the January 25 issue.
The second correction
But on the morning of February 2, we checked the press release about the article again, and that correction had been removed. The headnote to the press release about the original study had been quietly changed from the January 19 version, and no longer stated that there was an error in the study. Citing an unnamed “controversy,” the APA said that the marijuana-psychosis link was statistically insignificant based on a tighter statistical analysis.
ADAW contacted the APA on February 2 asking why the headnote to the press release about the original study had been changed. That’s when we learned that the APA had yielded to the authors’ demand to remove any reference to any errors. And APA press officer Kim Mills told ADAW that she was sorry, that she had made a mistake.
In a press release sent to ADAW on the afternoon of February 2 — but not posted on the APA website at press time — the APA said it “erred in issuing a correction.” In that press release, the APA said even though the supplemental analyses conducted by the authors last month “indicated that teens who engaged in frequent marijuana use had a higher probability of meeting lifetime criteria for a psychotic disorder (5%) than infrequent/nonusers (2%) by their 30s, this difference did not reach statistical significance using a two-tailed test (p=.09).”
The February 2 press release states that “APA regrets its error and would like to apologize to the study authors for misrepresenting the scientific integrity of their research.”
We also contacted Dustin Pardini, Ph.D., a co-author, who told ADAW that the “APA made an egregious and irresponsible error that I have been forced to clear up over the past several weeks to protect my scientific reputation.” Pardini, who is an associate professor of criminology and criminal justice at Arizona State University, said that there “was no error in the original analysis and referring to the study in that manner constitutes a libelous statement.”
So the first “oops” — that marijuana use by teens is linked to psychosis in adulthood — is followed by an “oops again.” The authors are saying that their study doesn’t show any link between teen marijuana use and adult psychosis — but not saying that it doesn’t exist.
“Everyone’s right in this case,” said Kevin A. Sabet, Ph.D., CEO of Smart Approaches to Marijuana (SAM), which encouraged the APA to run the first correction. “The researchers got upset about the word ‘error’” being connected to their work, Sabet told ADAW. But the clarification is still included with the article itself online, he said (see link at end of article). “We need to do more research on this,” he said. “We stand by what we said.”
ADAW also heard from Miller, who provided the original warnings about the study. She wrote a letter to Nancy M. Petry, Ph.D., editor of Psychology of Addictive Behaviors, concerning statistical errors in the study published last August. She said the clarification issued in the article “partially addresses” her concerns. “Whereas the corrected data as originally presented showed a one-to-one ratio of psychosis between the marijuana-use group and the low/non-use group, the uncorrected data presented in the clarification show a 2.5-to-one ratio of psychotic disorders in the marijuana use versus the low/non-use group,” she told ADAW. “It should be pointed out that it is highly unusual for correction factors to completely erase such a large difference, and thus the correction factors themselves deserve scrutiny, particularly when the 2.5-fold marijuana effect was consistent with numerous prior studies on psychosis.”
Small sample size, confounding errors
But her main concern is that the authors corrected for two factors — socioeconomic class and the health insurance status — that are more likely to be a result of psychosis, not a cause of it, and can also be influenced by marijuana use. Correcting for these factors is wrong, she said, because people with psychosis and people with marijuana use disorders have educational or economic problems, and psychosis can influence insurance status, as can marijuana use if it affects employment.
The study was also too small in size to accommodate the large number of correction factors applied by the authors, said Miller.
“Unfortunately, the APA now appears to be backing down from their characterization of the original Bechtold et al. analysis as containing a statistical error,” she said. “What is at stake for the journal is the very real possibility that if they were to impose more strict statistical standards on these authors, they would need to impose such standards on recent publications of a similar nature.”
Ultimately, Miller is concerned that the public will be misled to believe that there is no association between teen marijuana use and psychosis. “Because of the enormous public health implications of research like this, it is imperative that scholars and journals alike present data in the most accurate manner possible,” she said. “The public hears that marijuana has no relationship to psychotic disorders later in life and will change their behaviors accordingly.”
“As a long time schizophrenia researcher, now semi-retired from academia, it is my belief that the authors should publish an erratum in regards to their finding for psychosis,” Miller wrote in her letter to Petry. “The uncorrected data show an approximately 3-fold effect of marijuana and for that effect size to be completely erased by correction for covariates raises some red flags to say the least. I believe I have identified the source of their error.”
Miller’s letter detailed her critique of the statistical methods. But she also said that the “lack of statistical rigor with which the psychosis results were analyzed calls into question the entire paper.”
As a result of her letter, the APA did publish the clarification, which remains (as of press time) on the article.
For the full article, with the clarification, see http://www.apa.org/pubs/journals/releases/adb-adb0000103.pdf.