A Double-blind, Placebo Controlled Trial of Risperidone for the Treatment of Anorexia Nervosa

Eating Disorder Inventory -2 - Subscale : Drive for Thinness Subscale (DT). Lower scores are better on this scale and indicate less cognitive focus on drive for thinness.

The EDI 2 is a 91 item scale with 8 subscales - (Drive for thinness, Bulimia, body dissatisfaction, ineffectiveness, perfection, interpersonal distrust, interoceptive awareness and maturity fears.). The DT subscale was used for this outcome. Respondents rate each item as "usually , often, sometimes, rarely or never". Subscale scores are computed by summing all item scores for each subscale. There are 7 items in the DT subscale (questions 1,7,11,16,25,32 and 49). the subscale score range is 0-21. The EDI-2 was completed by subjects at baseline and then monthly during study participation (range 0 -18 weeks). Change in the DT subscale score was calculated using an estimate of change in score between week 0 and week 7 derived from the mixed effect model across all time points.

change in Eating Disorder Inventory (EDI) 2-score for Body Dissatisfaction (BD).

Lower scores are better on this scale. Higher scores indicate the subject has greater body dissatisfaction. BD is one of the 8 subscales of the EDI-2. 9 of the 91 questions in the EDI-2 scale constitute this subscale. The score range is 0-27. Subjects completed the EDI-2 at baseline and monthly during study participation (range 0 to 18 weeks). Change in the BD subscale score during the study was calculated using an estimate of change in score between week 0 and week 7 derived from the mixed effect model across all time points.

The Ease of Eating Scale (EOES) is a 14 item scale which measures Food avoidance behaviors (FABs). The scale is rated by staff observing a subject eating a meal or snack. 0 = normal eating behavior, maximum score 28.

Higher scores indicate more food avoidance behaviors, such as taking small bites, taking > 30 seconds between bites (slow eating), etc.

EOE was completed for each meal a subject ate in the program and scores were averaged for each week in the study and entered in the data base.

Change in EOES score was calculated by evaluating change over time. This measure was only used in Phase 1 of the study, for days the subjects were in the treatment program.

Color A Person Test (CAPT) - Subjects color an outlined image of a body to indicate body dissatisfaction (red (5)= very dissatisfied, Yellow, dissatisfied, black, neutral, green satisfied, blue very satisfied (1). The outline is divided into16 sections for scoring. The CAPT was completed at baseline and monthly during study participation.

Total CAPT scores were calculated by adding the total score and dividing by 16. Score range is 1-5. Lower scores indicate less body dissatisfaction.

Change in the CAPT score during the study was calculated using an estimate of change in score between week 0 and week 7 derived from the mixed effect model across all time points.

Body Image Software (BIS) - the subject adjusts a digital image of themselves on the computer using the direction to "adjust their image to how they see themselves right now", this determines their perception of their current image. Accuracy is measured by a smaller score between desired image and actual image.

Change in the BIS Average Distortion score during the study was calculated using an estimate of change in score between week 0 and week 7 derived from the mixed effect model across all time points.

There are no identifiable minimum/maximum values as there would be in a questionnaire scale. There are no subscales. The BIS program calculates the difference between their actual image and the size of the image they have adjusted the digital image to based on their perception of "how they see themselves right now"

Body Image Software (BIS) - the subject adjusts a digital image of themselves on the computer to "their desired image". The BIS program calculates the difference between their actual image, and how much they have adjusted the image to represent their "desired image". Accuracy is measured by a smaller score between desired image and actual image.

Change in BIS - Average Desired Thinness score was calculated using an estimate of change in score between week 0 and week 7 derived from the mixed effect model across all time points.

There are no identifiable minimum/maximum values as there would be in a questionnaire scale. . There are no subscales.

Body Image Software (BIS) - the subject adjusts a digital image of themselves on the computer to their desired image, and also completes a task that determines their perception of their current image. Accuracy is measured by a smaller score between desired image and actual image.

Change in BIS -PSE was calculated using an estimate of change in score between week 0 and week 7 derived from the mixed effect model across all time points.

There are no identifiable minimum/maximum values as there would be in a questionnaire scale. Interpreting the PSE is how it compares to a PSE = 0, which is no distortion in body size.

Body Image Software (BIS) - the subject adjusts a digital image of themselves on the computer to their desired image, and also completes a task that determines their perception of their current image. Accuracy is measured by a smaller score between desired image and actual image.

Change in BIS-DL was calculated using an estimate of change in score between week 0 and week 7 derived from the mixed effect model across all time points.

There are no identifiable minimum/maximum values as there would be in a questionnaire scale. There are no subscales. Interpreting the DL occurs by referencing it to DL= 0, which would reflect a total inability to detect size differences, which has never occurred in studies using the BIS program.

The mean survival time and its standard error were underestimated because the largest observation was censored and the estimation was restricted to the largest event time. These estimates were produced using Kaplan-Meier probabilities. This was measured weekly from 0-18 weeks.

The Multidimensional Anxiety Scale for Children (MASC) is a self report measure completed by the subject that measures anxiety symptoms.

Higher scores indicate greater anxiety. A score of over 50 is significant for anxiety

Change in MASC scores was calculated using an estimate of change in score between week 0 and week 7 derived from the mixed effect model across all time points.

Leptin levels were measured by serum blood draws, results reports in nanograms / ml (ng/ml).

Prolactin serum blood levels, measured in nanograms / ml

The mean survival time and its standard error were underestimated because the largest observation was censored and the estimation was restricted to the largest event time. These estimates were produced using Kaplan-Meier probabilities.