Eating-Disorders Experts Challenge Economists' Conclusions About Bulimia

Janis Kelly
April 1, 2009

Two prominent bulimia researchers are unconvinced by results of a new analysis by economists from University of Southern California (USC), in Los Angeles, which conclude that bulimia nervosa (BN) is an addiction rather than an eating disorder and is more common among black females than their white counterparts.

Michelle Goeree, PhD, and colleagues applied various economic analyses to data from the National Heart, Lung, and Blood Institute (NHLBI) Growth and Health Study and concluded that BN meets an economist's definition of addiction, that African American females are 50% more likely to be bulimic than white females, and that "bulimic behavior" is less likely among wealthier, better-educated families.

"These results are in contrast to the popular conceptions of who is most likely to experience BN, and we argue that this is a result of differences in diagnosis of BN across racial and income classes," Dr. Goeree and colleagues write.

The Institute of Economic Policy at USC recently published the working paper.

Addiction Definition "Incomplete"

Maybe not, say eating disorders specialists James L. Hudson, MD, and Linda Smolak, PhD. Dr. Hudson is director of the Psychiatric Epidemiology Research Program at McLean Hospital, in Belmont, Massachusetts and professor of psychiatry at Harvard Medical School, in Boston, Massachusetts. He was lead investigator of the National Comorbidity Survey Replication (NCS-R), a survey of more than 9000 people from across the United States, which was the first national survey of individuals with eating disorders.

Dr. Smolak is the Cummings Professor of Psychology at Kenyon College, in Gambier, Ohio, and a researcher advisor for the National Eating Disorders Association (NEDA).

"The model for addiction [in the Goeree paper] appears rather simplistic. It seems that depression and indeed most forms of psychopathology would meet this definition, in that having the symptoms at an initial time point is a strong predictor of having them later on. I see how the authors' analysis makes a case for persistence of bulimic behaviors, but not for addiction. For psychopathology, demonstration of persistence may be necessary but is not sufficient for addiction," Dr. Hudson told Medscape Psychiatry.

Dr. Smolak said economists often use Nobel Prize winner Gary Becker's definition of addiction, which focuses on the stability of behavior. "Stability is certainly part of the medical and psychological definition, but the core component in medicine and psychology is physiological withdrawal.

"The Goeree study does not, of course, establish such withdrawal. So it continues to be a bit misleading to suggest that BN is analogous to addictions such as heroin use in the medical sense," she said in an interview.

Authors Respond

However, the study authors believe focusing on withdrawal to define addiction is too narrow and restrictive. "We would argue that simply focusing on 'withdrawal' [to define addiction] seems too limited, since it focuses on the short-run physical aspects of addiction," study investigator John C. Ham, PhD, told Medscape Psychiatry.

"For example," he added, "this would seem to imply that detox units would solve problems of addiction for drugs and alcohol, since individuals released from these units would no longer be suffering from withdrawal and thus, by this definition, there would be no need for rehab facilities to address mental aspects of addiction," he added.

The Goeree study, which has occasioned considerable discussion in patient support groups and Internet chat rooms, was published December 29, 2008 as a working paper of USC's Institute of Economic Policy Research. A USC spokesperson said that the study is under review for publication in a peer-reviewed journal. The study was brought to media attention by a March 18 press release from the USC press office.

Fair Comparison?

Goeree et al pointed out that economists had not previously used the NHLBI Growth and Health Study data but that psychologists have extensively analyzed them.

For example, Dr. Smolak highlighted work published in 2003 by Striegel-Moore and colleagues in the American Journal of Psychiatry using this database, which showed that prevalence rates of subjects meeting the criteria for BN were higher in white than black women (Striegel-Moore RH et al. Am J Psychiatry. 2003;160:1326-1331).

"These rates were measured at ages 19 through 24, and BN was assessed using a full clinical interview. This methodology has 2 advantages over that of Goeree et al. First, the peak onset for BN is around age 19. Second, clinical interviews are more reliable than self-report surveys," Dr. Smolak said.

However, Dr. Ham countered that it is possible that BN rates change at different ages, but that this does not necessarily imply that estimates for 1 age group are better than another.

While Dr. Ham said he does not dispute the fact that clinical interviews are superior to self-reported surveys, he added that the study conducted by Striegel-Moore et al had a small sample size and did not follow subjects longitudinally.

"We do not want to diminish the important work by Stiegel-Moore et al and would simply point out that we are looking at different questions. They were looking at bulimia over a lifetime, whereas we looked for current evidence and were interested in persistence over time," he said.

Dr. Hudson added that the NCS-R data also do not support the view that BN is more common among African American women.

According to Dr. Ham, the only published study he has seen by Dr. Hudson examining the question of race appeared in Biological Psychiatry in 2007.

Findings Misleading?

From an economist's perspective, he said, this study has "serious statistical problems" due to selection bias. "It includes only individuals with eating disorders, and it is well-known that one does not obtain unbiased estimates in such studies."

The BN measure used by Goeree et al (the Eating Disorder Inventory–BN scale) is widely used in research, but Dr. Smolak commented that psychologists "virtually never" use it to diagnose BN.

"Of course psychologists don't use it [the Eating Disorder Inventory-BN scale], but until funding is available to perform these detailed interviews on thousands of girls and reinterview them every 2 years, researchers will not use the same measure as psychologists. The scale was devised by doctors and age-adjusted by doctors for this sample," said Dr. Ham.

Dr. Smolak said the study's conclusion that black girls are more likely to suffer BN than whites is "misleading."

"It would be more appropriate to say black girls report more symptoms than white girls," Dr. Smolak said. However, she added, she found the Goeree study helpful in dispelling the notion that African American girls are immune to eating disorders.

But Dr. Ham argued that it is difficult in a large-scale statistical analysis to focus on anything but symptoms.

The authors report no relevant conflicts of interest. The study was supported by the National Science Foundation, the Claremont McKenna Lowe Institute for Political Economy, the USC College of Letters, Arts, and Sciences, and the Ministerio de Eucacion y Ciencia and Ministerio de Ciencia y Tecnologia of Spain.

Institute of Economic Policy Research Working Paper 08.13. Published December 29, 2008. Abstract