A range of responses was generated for each item on the Eating in the Absence of Hunger Questionnaire in Children and Adolescents (EAH-C). Responses ranged from 1 to 5 and most medians were 2 or 3. A principal components factor analysis of the EAH-C, including all 14 items, generated three factors that collectively accounted for 65.3% of the variance. The first factor included 6 items, accounting for 28.4% of the variance (Eigenvalue=6.3), and represented EAH in response to feeling sad or depressed, angry or frustrated, and anxious or nervous (EAH-C Negative Affect Eating). Four items comprised the second factor, which was related to EAH when the food looks, tastes or smells so good and when others are eating (EAH-C External Eating). This factor accounted for 19.8% of the variance (Eigenvalue=1.6). The third factor also consisted of four items (17.1% of the variance; Eigenvalue=1.3) and represented EAH when feeling tired and bored (EAH-C Fatigue/Boredom Eating); . Identical subscales were generated by a common factor analysis: Negative Affect scale: Eigenvalue=6.2, 41.9% of the variance; External Eating scale: Eigenvalue=1.6, 8.3% variance; and Fatigue/Boredom scale Eigenvalue=1.3, 6.3% variance. All three subscales demonstrated good internal consistency; Cronbach's alphas for the EAH-C Negative Affect, EAH-C External Eating, and EAH-C Fatigue/Boredom subscales were 0.88, 0.80, and 0.83, respectively.
Factor Loadings for the EAH-C.
Convergent and discriminant validity
After accounting for the contribution of sex, age and SES, obese children had significantly higher EAH-C Negative Affect subscales scores compared to non-obese children [F(1, 216) = 3.82, p ≤ 0.05]. A similar, but non-significant pattern was found between obese and non-obese children for the Fatigue/Boredom scale, after controlling for race [F(1, 217) = 6.60, p < 0.06]. Controlling for age and SES, no differences were found on the External Eating subscale based upon body weight [F(1, 217) = 0.02, p < 0.1.0; ].
Figure 1 Obese participants reported significantly higher EAH-C Negative Affect (p ≤ 0.05), but not External Eating (p > 0.1), or Fatigue/Boredom (p < 0.06) subscales compared to non-obese children. *p ≤ 0.05. Bar key: Filled = (more ...)
Participants endorsing loss of control (n=44) during the EDE interview had significantly higher scores on the EAH-C Negative Affect [F(1, 150) = 7.09, p < 0.01], External Eating [F(1, 150) = 4.07, p < 0.05] , and Fatigue/Boredom [F(1, 150) = 8.45, p < 0.01] scales compared to those without loss of control (n=107), suggesting that the subscales demonstrated good convergent validity with loss of control eating (). Moreover, all three EAH-C subscales were positively correlated with all EES-C subscales (p's < 0.01; ), indicative of good construct validity. All correlations remained positively significant for boys and girls separately, and for the non-obese children. For the obese children, the correlation between the EAH-C Boredom/Fatigue scale and the EES-C Anger, Anxiety and Frustration scale became a trend (r = 0.23, p = 0.06); all other relationships remained significantly related for the obese children. All positive relationships also remained significant when analyzing Caucasian and African American children separately, except for the EAH-C Boredom/Fatigue subscale which was unrelated to the EES-C Anger, Anxiety and Frustration scale (r = 0.21, p = 0.10) for African American participants only.
Figure 2 Participants endorsing loss of control reported significantly higher EAH-C Negative Affect (p < 0.01), External Eating (p < 0.05), and Fatigue/Boredom (p < 0.01) subscales compared to those without loss of control. *p<0.05; (more ...)
Table 3 Product-moment correlation coefficients between the EAH-C Negative Affect, External and Fatigue/Boredom subscales and the EES-C Depression, Anger, Anxiety and Frustration, and Unsettled subscales, the Children's Depression Inventory, and the State-Trait (more ...)
All three EAH-C subscales were also positively related with the Children's Depression Inventory total score as well as state and trait anxiety (p's < 0.01; ). Higher EAH-C scores were associated with more symptoms of depression and anxiety. When analyzing the children separately based upon sex, all relationships remained positively significant for the girls. However, for the boys, the EAH-C Boredom/Fatigue scale discriminated from depressive symptoms (r = 0.13, p = 0.20), and the EAH-C External scale was unassociated with state anxiety (r = 0.05, p = 0.67) and depressive symptoms (r = 0.12, p = 0.25). Further, while all relationships remained significant and positive for the non-obese children, for the obese participants, correlations for the EAH-C External (r = 0.23, p = 0.06) and Boredom/Fatigue (r = 0.22, p = 0.07) scales with depressive symptoms became trends. The EAH-C External scale discriminated from trait anxiety for the obese children only (r = 0.08, p = 0.49). We also examined Caucasian and African American children separately. Findings remained the same for Caucasian children as compared to the entire sample. However, for African American participants, the EAH-C Negative Affect (r = 0.11, p = 0.38) and External (r = 0.16, p = 0.20) scales discriminated from state anxiety.
Given the wide age range of our sample, we conducted a second set of analyses examining the 37 younger (≤12y) and 189 older (>12y) children separately. Examining the younger group, the 30 non-overweight participants did not differ from the seven overweight children with regard to their EAH-C scores (p > 0.1 in all cases). After controlling for sex, younger children with loss of control (9.5 ± 3.6) had significantly higher EAH-C Fatigue/Boredom scores compared to those without loss of control (5.8 ± 2.1); [F (1, 19) = 9.85, p < 0.01]. No differences were found based upon loss of control status for the other EAH-C subscales (p > 0.4 in all cases). By contrast, all three EAH-C subscales positively correlated with the three EES-C scales (r's: 0.40-0.67, p < 0.05 in all cases). The EAH-C Negative Affect scale was positively related to trait anxiety (r = 0.34, p ≤ 0.05) and depressive symptoms (r = 0.40, p < 0.05); the EAH-C Fatigue/Boredom subscale positively correlated with depressive symptoms (r = 0.55, p < 0.01); and the External Eating subscale was positively related to trait anxiety (r = 0.48, p<0.01) and depressive symptoms (0.38, p < 0.05). Results from the older participants did not differ considerably from those of the entire sample.
To assess the stability of the EAH-C, a convenience sample of 115 children completed the measure a second time during their next scheduled visit to the NIH. The average interval period between the two administrations was varied: 150 (SD: 130) days ranging between 5 and 565 days. Intra-class correlations demonstrated good agreement between the first and second administrations of the questionnaire; scores for the EAH-C Negative Affect (0.65), EAH-C Fatigue/Boredom subscale (0.70), and the EAH-C External Eating subscale (0.69) were all significantly correlated (p < 0.01 in all cases). After accounting for length of interval between administrations, correlations for all three scales remained significant at p < 0.01.