Machine learning enhances prediction of illness course: a longitudinal study in eating disorders.

BACKGROUND: Psychiatric disorders, including eating disorders (EDs), have clinical outcomes that range widely in severity and chronicity. The ability to predict such outcomes is extremely limited. Machine-learning (ML) approaches that model complexity may optimize the prediction of multifaceted psychiatric behaviors. However, the investigations of many psychiatric concerns have not capitalized on ML to improve prognosis. This study conducted the first comparison of an ML approach (elastic net regularized logistic regression) to traditional regression to longitudinally predict ED outcomes. METHODS: Females with heterogeneous ED diagnoses completed demographic and psychiatric assessments at baseline (n = 415) and Year 1 (n = 320) and 2 (n = 277) follow-ups. Elastic net and traditional logistic regression models comprising the same baseline variables were compared in ability to longitudinally predict ED diagnosis, binge eating, compensatory behavior, and underweight BMI at Years 1 and 2. RESULTS: Elastic net models had higher accuracy for all outcomes at Years 1 and 2 [average Area Under the Receiving Operating Characteristics Curve (AUC) = 0.78] compared to logistic regression (average AUC = 0.67). Model performance did not deteriorate when the most important predictor was removed or an alternative ML algorithm (random forests) was applied. Baseline ED (e.g. diagnosis), psychiatric (e.g. hospitalization), and demographic (e.g. ethnicity) characteristics emerged as important predictors in exploratory predictor importance analyses. CONCLUSIONS: ML algorithms can enhance the prediction of ED symptoms for 2 years and may identify important risk markers. The superior accuracy of ML for predicting complex outcomes suggests that these approaches may ultimately aid in advancing precision medicine for serious psychiatric disorders.

Perfectionism, anorexia nervosa, and family treatment: How perfectionism changes throughout treatment and predicts outcomes.

OBJECTIVE: This study uses data from a multisite randomized clinical trial to study the role of perfectionism in family-based treatment (FBT) for adolescent anorexia nervosa (AN). The main aim is to examine the role of baseline perfectionism in treatment response. METHOD: Adolescents (N = 158; ages 12-18; 89.2% female) and their families were randomized to receive either FBT or systemic family treatment for AN. Eating disorder (ED) pathology, obsessive-compulsive symptoms, and perfectionism were assessed at baseline, end of treatment, and 6- and 12-month follow-up. Linear regression analyses were used to test whether perfectionism and obsessive-compulsive symptoms at baseline predict ED pathology at all timepoints. An independent samples t test was used to test whether there was a significant difference in the change in perfectionism in either treatment group. RESULTS: Baseline maladaptive perfectionism significantly predicted ED pathology but not ideal body weight at all timepoints. The model that included obsessive-compulsive symptoms also predicted ED pathology at all timepoints except 12-month follow-up. Perfectionism scores did not change during treatment regardless of treatment type. DISCUSSION: Baseline perfectionism predicted treatment response in this study. Interventions might target perfectionism to improve treatment response in AN.

Not all body image constructs are created equal: Predicting eating disorder outcomes from preoccupation, dissatisfaction, and overvaluation.

OBJECTIVE: Diverse terminology has been used to operationalize body image disturbance in eating disorders. However, the differential validity of these terms and their underlying constructs to predict outcomes among heterogeneous eating disorders is unknown. This study evaluated the validity of body image constructs to predict eating disorder and negative psychological symptoms concurrently and prospectively over 2 years in a transdiagnostic clinical sample. METHODS: Women with heterogeneous eating disorder diagnoses (n = 448) completed assessments at baseline, 12-month, and 24-month follow-up. Cross-sectional and cross-lagged generalized linear models examined effects of three body image constructs (i.e., weight and shape preoccupation, overvaluation, and dissatisfaction) on concurrent and subsequent outcomes (i.e., global eating disorder symptoms, binge eating, purging, fasting, self-esteem, and depression). RESULTS: In concurrent analyses, preoccupation was significantly associated with all outcomes (ps = .01 to <.001), overvaluation with all outcomes (ps = .01 to <.001) except binge eating (p = .06), and dissatisfaction with all outcomes (ps < .001) except purging (p = .38). In prospective analyses, preoccupation predicted Eating Disorder Examination global (p = .003) and fasting (p < .001), overvaluation predicted binge eating (p = .01), and body dissatisfaction did not predict any outcomes. DISCUSSION: Preoccupation, overvaluation, and dissatisfaction are differentially related to eating disorder and psychiatric outcomes, indicating that no one body image construct can capture clinical risk in eating disorders. Preoccupation was the most consistent concurrent and longitudinal predictor; this construct may warrant further attention in assessment and diagnosis. Further investigation of these constructs in diverse samples is encouraged.

Therapist adherence to family-based treatment for adolescents with anorexia nervosa: A multi-site exploratory study.

OBJECTIVE: This exploratory study is the first to examine family-based treatment (FBT) adherence and association to treatment outcome in the context of a large-scale, multi-centre study for the treatment of adolescents with anorexia nervosa. METHOD: One hundred and ninety recorded FBT sessions from 68 adolescents with anorexia nervosa and their families were recruited across multiple sites (N = 6). Each site provided 1-4 tapes per family over four treatment time points, and each was independently rated for therapist adherence. RESULTS: There were differences in adherence scores within and between sites. ANOVA produced a main effect for site, F(5, 46) = 8.6, p < .001, and phase, F(3, 42) = 12.7, p < .001, with adherence decreasing in later phases. Adherence was not associated to end of treatment percent ideal body weight after controlling for baseline percent ideal body weight (r = .088, p = .48). CONCLUSIONS: Results suggest that FBT can be delivered with adherence in phase one of treatment. Adherence was not associated with treatment outcome as determined using percent ideal body weight.

An investigation of indirect effects of personality features on anorexia nervosa severity through interoceptive dysfunction in individuals with lifetime anorexia nervosa diagnoses.

OBJECTIVE: This study examined a hypothesized pathway by which interoceptive dysfunction accounted for associations between personality features (harm avoidance, self-directedness, and perfectionism) and anorexia nervosa (AN) severity (indicated by drive for thinness, eating disorder-related preoccupations and rituals, and body mass index). METHOD: The study sample (n = 270, mean age = 28.47, 95.2% female, 98% White/Caucasian) consisted of probands and biological relatives who met DSM-IV criteria for lifetime diagnoses of AN (omitting criterion D, amenorrhea) drawn from the Price Foundation Anorexia Nervosa Affected Relative Pairs Study (AN-ARP). Participants completed measures assessing personality, interoceptive dysfunction, and eating pathology. RESULTS: Associations between personality features of low self-directedness and high perfectionism and indicators of AN severity (drive for thinness and eating disorder-related preoccupations and rituals) were significant, as were the hypothesized indirect pathways through interoceptive dysfunction. Neither harm avoidance nor body mass index was significantly related to other study variables, and the proposed indirect pathways involving these variables were not significant. DISCUSSION: Findings suggest that certain personality features may relate to AN severity, in part, through their associations with interoceptive dysfunction. Future research should examine prospective associations and the value of interventions targeting interoceptive dysfunction for interrupting the link between personality and AN severity.

Associations between dimensions of anorexia nervosa and obsessive-compulsive disorder: An examination of personality and psychological factors in patients with anorexia nervosa.

OBJECTIVE: Anorexia nervosa (AN) and obsessive-compulsive disorder (OCD) are highly comorbid. However, the factors that account for this comorbidity are poorly understood. We examined the core dimensions of AN and OCD and psychological and personality factors shared by both disorders. METHOD: In path analyses (N = 732 women with either current AN or recovered from AN), we examined which factors were uniquely and independently associated with the core dimensions of AN and OCD. We also examined recovery from AN as a moderator. RESULTS: When individuals with AN reported greater concern over mistakes, they endorsed more severity in both AN and OCD core dimensions. These unique associations existed above and beyond all other transdiagnostic personality and psychological factors and regardless of AN recovery status. CONCLUSIONS: Concern over mistakes partially accounts for severity in the core dimensions of both AN and OCD. Concern over mistakes may represent an important target in the aetiology of AN and OCD.

Are parental self-efficacy and family flexibility mediators of treatment for anorexia nervosa?

OBJECTIVE: Family-based treatment (FBT) for adolescent anorexia nervosa (AN) promotes faster weight restoration when compared to other treatments. However, the mechanisms through which this occurs are not clarified. This study explored the trajectories of parental self-efficacy and perceived family flexibility during FBT and systemic family therapy (SyFT). We also explored whether parental self-efficacy mediates the effects of treatment on weight gain early in treatment. METHOD: 158 adolescents (12-18 years old; 89% girls) and their parents were randomized to FBT or SyFT. Parental self-efficacy as well as adolescents' and parental perceptions of the family's flexibility were collected at baseline and at sessions 2, 4, 6, and 8. RESULTS: Over time, only parents in FBT reported significantly greater self-efficacy. The change in maternal self-efficacy over the first 8 weeks of treatment was a significant mediator of session 10 weight gain. There were no significant group differences in perceived flexibility by session 8. DISCUSSION: Both parents in FBT and mothers in SyFT understand early the need to change their family's rules and roles. However, the specific strategies of FBT appear to mediate early weight gain in AN.

Food portion size area mediates energy effects on expected anxiety in anorexia nervosa.

A study in which adolescent patients with anorexia nervosa (n = 24) rated their expected food-anxiety in response to images of portions of food (potatoes, rice pizza, and M&Ms) showed that lower energy-dense foods elicited higher expected anxiety per kilocalorie than higher energy-dense foods. However, the area of the portion sizes could be an unmeasured variable driving the anxiety response. To test the hypothesis that area mediates the effects of energy content on expected anxiety, the same images of portions were measured in area (cm2), and standardized values of expected anxiety were regressed from standardized values of energy and area of portions. With regression of expected anxiety from portion size in area, M&Ms, which had the highest energy density of the four foods, elicited the highest expected anxiety slope (ß = 1.75), which was significantly different from the expected anxiety slopes of the other three foods (ß range = 0.67 - 0.96). Area was confirmed as a mediator of energy effects from loss of significance of the slopes when area was added to the regression of expected anxiety from energy x food. When expected anxiety was regressed from food, area, energy and area by energy interaction, area accounted for 5.7 times more variance than energy, and ß for area (0.7) was significantly larger (by 0.52, SE = 0.15, t = 3.4, p = 0.0007) than ß for energy (0.19). Area could be a learned cue for the energy content of food portions, and thus, for weight gain potential, which triggers anxiety in patients with anorexia nervosa.

Does family-based treatment reduce the need for hospitalization in adolescent anorexia nervosa?

OBJECTIVE: We examined the timing and number of days of hospitalization during the course of treatment, hospitalization effects on outcome, and predictors and moderators of the use of hospitalization in adolescents with anorexia nervosa (AN). METHOD: Data used in this study were collected from 158 adolescents (ages 12 to 18 years of age) who met DSM-IVTR criteria for AN (exclusive of the amenorrhea criteria) randomized to receive either Family Based Treatment (FBT) or Systemic Family Therapy (SyFT) in a 7 site study. RESULTS: The trajectory of hospital day use is similar in the first 5 weeks irrespective of treatment allocation. However, days of hospitalization continued to increase throughout SyFT but leveled off in FBT after ∼5 weeks of treatment. Early hospitalization was a negative predictor for improvements in percent weight change for both treatment groups (t(1)=2.6, p = 0.011). Co-morbid psychopathology predicted early hospital use in both treatments. Higher levels of eating related obsessions and depression moderated hospitalization rates suggesting that FBT reduces early hospitalization rates compared to SyFT for these subgroups. DISCUSSION: These data support and extend findings from previous studies by identifying patterns of hospital use, and predictors and moderators of treatment effect for early hospitalization use in adolescent AN. © 2016 Wiley Periodicals, Inc.(Int J Eat Disord 2016; 49:891-894).

A national survey of eating disorder training.

OBJECTIVE: Eating disorders (EDs) result in the highest mortality rate of all psychiatric disorders, and in the United States, approximately one in twenty females suffers from an eating disorder. However, training provided within residency programs to address the needs of these patients is sparse. The objective of this study was to conduct a national survey that assesses the amount of EDs training for trainees across five ACGME accredited specialties: internal medicine, pediatrics, family medicine, psychiatry, and child and adolescent psychiatry. The results of the survey will be used to develop strategies to improve eating disorder education among residents. METHOD: Eight hundred eighty training coordinators were contacted using information available on the ACGME website and asked to complete the survey. RESULTS: Of the 637 responding programs, 514 did not offer any scheduled or elective rotations for EDs. Of the 123 programs offering rotations, only 42 offered a formal, scheduled rotation. Child and adolescent psychiatry offered the most clinical experiences, and pediatric programs offered the greatest number of didactic hours on EDs. DISCUSSION: Training in EDs is limited. Simulated patient encounters, massive open online courses, web-based curricula, dedicated rotations and clinical experiences, didactic curricula, and brief-training programs may help to improve eating disorder diagnostic and treatment skills among trainees.

Primary amenorrhea in anorexia nervosa: impact on characteristic masculine and feminine traits.

Animal studies indicate that gonadal hormones at puberty have an effect on the development of masculine and feminine traits. However, it is unknown whether similar processes occur in humans. We examined whether women with anorexia nervosa (AN), who often experience primary amenorrhea, exhibit attenuated feminization in their psychological characteristics in adulthood due to the decrease/absence of gonadal hormones at puberty. Women with AN were compared on a number of psychological characteristics using general linear models on the basis of the presence/absence of primary amenorrhea. Although women with primary amenorrhea exhibited lower anxiety scores than those without primary amenorrhea, in general, results did not provide evidence of attenuated feminization in women with AN with primary amenorrhea. Future research should utilize novel techniques and direct hormone measurement to explore the effects of pubertal gonadal hormones on masculine and feminine traits.

A cost effectiveness analysis of stepped care treatment for bulimia nervosa.

BACKGROUND: The cost effectiveness of various treatment strategies for bulimia nervosa (BN) is unknown. AIMS: To examine the cost effectiveness of stepped care treatment for BN. METHOD: Randomized trial conducted at four clinical centers with intensive measurement of direct medical costs and repeated measurement of subject quality of life and family/significant other time involvement. Two hundred ninety-three women who met DSM-IV criteria for BN received stepped care treatment or cognitive behavioral therapy. Cost effectiveness ratios were compared. RESULTS: The cost per abstinent subject was $12,146 for stepped care, and $20,317 for cognitive behavioral therapy. Quality of life ratings improved significantly with treatment, and family/significant other time burden diminished substantially. DISCUSSION: In this trial, stepped care for BN appeared cost effective in comparison to cognitive behavioral therapy. Treatment was associated with improved quality of life and diminished time costs of illness.

Perplexities of treatment resistance in eating disorders.

BACKGROUND: Treatment resistance is an omnipresent frustration in eating disorders. Attempts to identify the features of this resistance and subsequently develop novel treatments have had modest effects. This selective review examines treatment resistant features expressed in core eating disorder psychopathology, comorbidities and biological features. Novel treatments addressing resistance are discussed. DESCRIPTION: The core eating disorder psychopathology of anorexia nervosa becomes a coping mechanism likely via vulnerable neurobiological features and conditioned learning to deal with life events. Thus it is reinforcing and ego syntonic resulting in resistance to treatment. The severity of core features such as preoccupations with body image, weight, eating and exercising predicts greater resistance to treatment. Bulimia nervosa patients are less resistant to treatment with treatment failure related to greater body image concerns, impulsivity, depression, severe diet restriction and poor social adjustment. For those with binge eating disorder overweight in childhood and high emotional eating predicts treatment resistance. There is suggestive data that a diagnosis of an anxiety disorder and severe perfectionism may confer treatment resistance in anorexia nervosa and substance use disorders or personality disorders with impulse control problems may produce resistance to treatment in bulimia nervosa. Traits such as perfectionism, cognitive inflexibility and negative affect with likely genetic influences may also affect treatment resistance. Pharmacotherapy and novel therapies have been developed to address treatment resistance. Atypical antipsychotic drugs have shown some effect in treatment resistant anorexia nervosa and topiramate and high doses of SSRIs are helpful for treatment of resistant binge eating disorder patients. There are insufficient randomized controlled trials to evaluate the novel psychotherapies which are primarily based on the core psychopathological features of the eating disorders. CONCLUSION: Treatment resistance in eating disorders is usually predicted by the severity of the core eating disorder psychopathology which develops from an interaction between environmental risk factors with genetic traits and a vulnerable neurobiology. Future investigations of the biological features and neurocircuitry of the core eating disorders psychopathology and behaviors may provide information for more successful treatment interventions.

Longitudinal stability of binge-eating type in eating disorders.

OBJECTIVE: The purpose of this study was to examine the 2-year longitudinal stability of objective bulimic (binge eating) episodes (OBEs) and subjective bulimic (binge eating) episodes (SBEs) in a multisite eating disorders sample. METHOD: Participants included 288 females with eating disorder symptoms who were assessed every 6 months using the Eating Disorder Examination. RESULTS: Markov modeling revealed considerable longitudinal variability between types of binge eating over 6-month time intervals with relatively higher probability estimates for consistency between OBEs and SBEs than specific transitions between types for the overall sample as well as for eating disorder diagnostic groups. Transition patterns examining all five time points indicated notable variability in binge-eating patterns among participants. DISCUSSION: These findings suggest that although longitudinal patterns of binge types are variable among individuals with eating disorders, consistency in OBEs and SBEs was the most common pattern observed.

The Yale-Brown-Cornell eating disorders scale self-report questionnaire: a new, efficient tool for clinicians and researchers.

OBJECTIVE: The YBC-EDS is a semistructured interview assessing core preoccupations and rituals related to eating disorders. METHOD: We developed and conducted an examination of the reliability and validity of a self-report questionnaire (SRQ) version of the YBC-EDS. Convergent validity of YBC-EDS-SRQ with the YBC-EDS was examined for 112 eating disordered patients. RESULTS: All subscales and total scores were significantly intercorrelated. Thirty-one additional patients completed YBC-EDS-SRQ at admission and again 1 week later. All correlations revealed significant test-retest reliability. Discriminant validity of the SRQ was evaluated for a smaller subset of participants who completed the Beck Depression Inventory (BDI) and State Trait Anxiety Inventory (STAI). There were no significant correlations between various symptom dimensions of the YBC-EDS-SRQ and the BDI and STAI. DISCUSSION: Taken together, these findings indicate that the self-report form of the YBC-EDS is both valid and reliable. The SRQ can serve as a useful and efficient assessment of eating disorder patients for clinicians and researchers.

An examination of early childhood perfectionism across anorexia nervosa subtypes.

OBJECTIVE: To examine childhood perfectionism in anorexia nervosa (AN) restricting (RAN), purging (PAN), and binge eating with or without purging (BAN) subtypes. METHOD: The EATATE, a retrospective assessment of childhood perfectionism, and the eating disorder inventory (EDI-2) were administered to 728 AN participants. RESULTS: EATATE responses revealed general childhood perfectionism, 22.3% of 333 with RAN, 29.2% of 220 with PAN, and 24.8% of 116 with BAN; school work perfectionism, 31.2% with RAN, 30.4% with PAN, and 24.8% with BAN; childhood order and symmetry, 18.7% with RAN, 21.7% with PAN, and 17.8% with BAN; and global childhood rigidity, 42.6% with RAN, 48.3% with PAN and 48.1% with BAN. Perfectionism preceded the onset of AN in all subtypes. Significant associations between EDI-2 drive for thinness and body dissatisfaction were present with four EATATE subscales. DISCUSSION: Global childhood rigidity was the predominate feature that preceded all AN subtypes. This may be a risk factor for AN.

Retrospective maternal report of early eating behaviours in anorexia nervosa.

This exploratory study assessed whether maternal recall of childhood feeding and eating practices differed across anorexia nervosa (AN) subtypes. Participants were 325 women from the Genetics of Anorexia Nervosa study whose mothers completed a childhood feeding and eating questionnaire. Multinomial logistic regression analyses were used to predict AN subtype from measures related to childhood eating: (i) infant feeding (breastfed, feeding schedule, age of solid food introduction), (ii) childhood picky eating (picky eating before age 1 year and between ages one and five) and (iii) infant gastrointestinal problems (vomiting and colic). Results revealed no significant differences in retrospective maternal report of childhood feeding and eating practices among AN subtypes.

Understanding the association of impulsivity, obsessions, and compulsions with binge eating and purging behaviours in anorexia nervosa.

OBJECTIVE: To further refine our understanding of impulsivity, obsessions, and compulsions in anorexia nervosa (AN) by isolating which behaviours--binge eating, purging, or both--are associated with these features. METHODS: We conducted regression analyses with binge eating, purging, and the interaction of binge eating with purging as individual predictors of scores for impulsivity, obsessions, and compulsions in two samples of women with AN (n = 1373). RESULTS: Purging, but not binge eating, was associated with higher scores on impulsivity, obsessions, and compulsions. Purging was also associated with worst eating rituals and with worst eating preoccupations. CONCLUSION: Our results suggest that purging, compared with binge eating, may be a stronger correlate of impulsivity, obsessions, and compulsions in AN.

Affect Modulated Startle Response in Anorexia Nervosa, Restricting Type: Implications for Theory and Practice.

OBJECTIVE: Individuals with anorexia nervosa (AN), restricting type demonstrate unique emotional responses to hedonically positive stimuli beyond eating disorder (ED)-related stimuli. The goal of this study was to evaluate differences in responses to five types of emotionally positive stimuli among acutely ill anorexia nervosa (IAN), restricting type patients, weight-recovered anorexia patients (WRAN), and healthy controls (HCs) using affect modulated startle response (AMSR) as an objective measure. METHOD: A total of 28 participants were recruited (n=28). Fourteen participants were recruited as IAN using the Diagnostic and Statistical Manual of Mental Disorders-V (DSM-V) criteria, seven were WRAN, and seven were HC females. All participants were female and aged between 8 and 18 years. The participants viewed images depicting negative, neutral, standardized, and non-eating disorder (ED)-related positive stimuli. Additionally, four categories of ED-related stimuli (high-calorie food, body image, success, and parent-child relationships) were presented to all participants during a standard AMSR paradigm. RESULTS:  No significant between-group differences were found for any of the four ED stimulus categories; all groups showed an inhibited startle response to the four ED-related categories. In contrast, IAN and WRAN showed reduced hedonic responses to standardized positive stimuli relative to HC-replicating previous results. Reduced hedonic response to the standardized (non-ED) positive stimuli was highly correlated with self-reported social anxiety, low self-esteem, body dissatisfaction, asceticism, interpersonal problems, and ineffectiveness. CONCLUSION: AN patients had a reduced hedonic response to some non-ED-related positive stimuli, which correlated with several anxiety-related traits. In contrast, their early automatic responses to high-calorie food, normal weight models, images of success, and positive parent-child relationships did not differ from HC, suggesting these stimuli are either being evaluated as highly interesting or hedonically positive.

Dissecting the Shared Genetic Architecture of Suicide Attempt, Psychiatric Disorders, and Known Risk Factors.

BACKGROUND: Suicide is a leading cause of death worldwide, and nonfatal suicide attempts, which occur far more frequently, are a major source of disability and social and economic burden. Both have substantial genetic etiology, which is partially shared and partially distinct from that of related psychiatric disorders. METHODS: We conducted a genome-wide association study (GWAS) of 29,782 suicide attempt (SA) cases and 519,961 controls in the International Suicide Genetics Consortium (ISGC). The GWAS of SA was conditioned on psychiatric disorders using GWAS summary statistics via multitrait-based conditional and joint analysis, to remove genetic effects on SA mediated by psychiatric disorders. We investigated the shared and divergent genetic architectures of SA, psychiatric disorders, and other known risk factors. RESULTS: Two loci reached genome-wide significance for SA: the major histocompatibility complex and an intergenic locus on chromosome 7, the latter of which remained associated with SA after conditioning on psychiatric disorders and replicated in an independent cohort from the Million Veteran Program. This locus has been implicated in risk-taking behavior, smoking, and insomnia. SA showed strong genetic correlation with psychiatric disorders, particularly major depression, and also with smoking, pain, risk-taking behavior, sleep disturbances, lower educational attainment, reproductive traits, lower socioeconomic status, and poorer general health. After conditioning on psychiatric disorders, the genetic correlations between SA and psychiatric disorders decreased, whereas those with nonpsychiatric traits remained largely unchanged. CONCLUSIONS: Our results identify a risk locus that contributes more strongly to SA than other phenotypes and suggest a shared underlying biology between SA and known risk factors that is not mediated by psychiatric disorders.