ORIGINAL ARTICLES An Examination of Cognitive Versus Behavioral Components of Recovery From Anorexia Nervosa Rachel Bachner-Melman, MA,*† Ada H. Zohar, PhD,‡ and Richard P. Ebstein, PhD*§ Abstract: Definitions of “full recovery” from anorexia nervosa (AN) vary, and rarely include the cognitive criteria of lack of body image distortion and fear of weight gain. We investigated the implications of including or excluding cognitive criteria of AN in the definition of “full recovery”. Current symptomatology and per- sonality characteristics associated with AN were assessed and com- pared in 42 behaviorally but not cognitively recovered women, 32 both behaviorally and cognitively recovered women, and 253 con- trols. On all measures included, the scores of the behaviorally recovered women were significantly more anorexic-like than those of the women recovered cognitively as well, who were indistin- guishable from controls. Criteria for recovery from AN need to be refined and standardized, and cognitive criteria incorporated, to characterize a minority who recover to the extent that their eating attitudes and personality profiles are indistinguishable from those of women with no history of an eating disorder. Key Words: Anorexia nervosa, recovery, personality. (J Nerv Ment Dis 2006;194: 697–703) A norexia nervosa (AN) is a serious disorder characterized by a refusal to maintain a normal weight, an intense fear of weight gain, a disturbed body image, and amenorrhea (APA, 1994). Whereas attitudes toward food and weight obviously improve with recovery from AN, recovered indi- viduals have generally been found to display residual behav- ioral and attitudinal disturbances characteristic of the disorder (Clinton and McKinlay, 1986; O’Dwyer et al., 1996; Windauer et al., 1993). Similarly, whereas research has shown that personality features associated with AN such as perfectionism, obsessiveness, harm avoidance, and low self- esteem tend to regress to the mean with symptomatic im- provement (Bloks et al., 2004; Bulik et al., 2000; Kennedy et al., 1990; Pollice et al., 1997; Ward et al., 1998), such characteristics have been found to persist to some extent after remission (Casper, 1990; O’Dwyer et al., 1996; Srinivasagam et al., 1995; Sullivan et al., 1998), which has implicated them in the etiology of the disorder. On the other hand, Ward et al. (1998) found recovered anorexics’ levels of harm avoidance to be in the normal range, Sutandar-Pinnock et al. (2003) found their level of perfectionism as measured by the Eating Disorders Inventory (though not by another scale) to be similar to that of controls, and Hulley and Hill (2001) found that six athletes treated in the past for an eating disorder scored similarly to controls on a battery of questionnaires measuring psychological health. One partial explanation for these inconsistencies is that the lack of consensus on criteria for recovery or outcome limits the comparability of findings. Some studies define recovery using the largely biological criteria of normal weight and regular menstruation (Pollice et al., 1997), others extend them to include the behavioral criteria of a lack of bingeing and purging symptoms (Bulik et al., 2000; Ro et al., 2004), and yet others add the absence of restrictive eating patterns (Brown et al., 2001, 2003; Srinivasagam et al., 1995). Outcome studies have traditionally relied on relatively general classification systems. The most common is a trichot- omy between good, fair, and poor outcome, as exemplified by the Morgan and Russell criteria (1975). In all these schemas, “recovery” or “good outcome” is a weak index of overall status, since patients with residual psychological, cognitive, and personality features of AN are included. More precise classification systems exist. Herzog et al. (1993) developed a Psychiatric Status Rating Scale (PSR) for AN based on DSM-IV criteria that implements a 6-point rating scale with 1 representing full recovery and 6 represent- ing active and severe AN. This scale, however, is not uni- formly implemented. Whereas some studies (Lowe et al., 2001) define good outcome as a PSR level of 1 (absence of all symptoms), others (Bloks et al., 2004; Herzog et al., 1993, 1999) include level 2 (presence of residual symptoms), which again includes those with lingering symptoms in the good outcome category. The most stringent definition of recovery, providing the greatest conceptual clarity, is the absence of all symptoms of AN, as used, for example, in long-term outcome studies by Lowe et al. (2001) and Strober et al. (1997). A lack of all *Department of Psychiatry, Hadassah University Medical Center, Ein Kerem, Jerusalem, Israel; †Psychiatry, Hadassah University Medical Center, Hebrew University of Jerusalem, Israel; ‡Behavioral Sciences, Ruppin Academic College, Israel; and §Research Laboratory, Sarah Herzog Memorial Hospital, Jerusalem, Israel. E-mail: msrbach@plutomail.huji. ac.il. This research was partially supported by the Israel Science Foundation founded by the Israel Academy of Sciences and Humanities and the Israel Association of University Women. Send reprint requests to Rachel Bachner-Melman, MA, Psychology Depart- ment, Hebrew University of Jerusalem, 12 HaGedud Haivri St., Jerusa- lem 92345, Israel. Copyright © 2006 by Lippincott Williams & Wilkins ISSN: 0022-3018/06/19409-0697 DOI: 10.1097/01.nmd.0000235795.51683.99 The Journal of Nervous and Mental Disease • Volume 194, Number 9, September 2006 697