ObjectiveThe StRONG trial demonstrated the safety and efficacy of higher calorie refeeding (HCR) in hospitalized adolescents and young adults with malnutrition secondary to restrictive eating disorders. Here we compare refeeding outcomes in patients with atypical anorexia nervosa (atypical AN) versus anorexia nervosa (AN) and examine the impact of caloric dose.MethodPatients were enrolled upon admission and randomized to meal‐based HCR, beginning 2000 kcal/day and advancing 200 kcal/day, or lower calorie refeeding (LCR), beginning 1400 kcal/day and advancing 200 kcal every other day. Atypical AN was defined as %median BMI (mBMI) > 85. Independent t‐tests compared groups; multivariable linear and logistic regressions examined caloric dose (kcal/kg body weight).ResultsAmong n = 111, mean ± SD age was 16.5 ± 2.5 yrs; 43% had atypical AN. Compared to AN, atypical AN had slower heart rate restoration (8.7 ± 4.0 days vs. 6.5 ± 3.9 days, p = .008, Cohen's d = −.56), less weight gain (3.1 ± 5.9%mBMI vs. 5.4 ± 2.9%mBMI, p < .001, Cohen's d = .51) and greater hypomagnesemia (29% vs. 11%, p = .03, OR = 3.29). These suboptimal outcomes were predicted by insufficient caloric dose (32.4 ± 6.9 kcal/kg in atypical AN vs. 43.4 ± 9.8 kcal/kg in AN, p < .001, Cohen's d = 1.27). For every 10 kcal/kg increase, heart rate was restored 1.7 days (1.0, 2.5) faster (p < .001), weight gain was 1.6%mBMI (.8, 2.4) greater (p < .001), and hypomagnesemia odds were 70% (12, 128) lower (p = .02).DiscussionAlthough HCR is more efficacious than LCR for refeeding in AN, it contributes to underfeeding in atypical AN by providing an insufficient caloric dose relative to the greater body weight in this diagnostic group.Public significanceThe StRONG trial previously demonstrated the efficacy and safety of higher calorie refeeding in patients with malnutrition due to restrictive eating disorders. Here we show that higher calorie refeeding contributes to underfeeding in patients with atypical anorexia nervosa, including poor weight gain and longer time to restore medical stability. These findings indicate these patients need more calories to support nutritional rehabilitation in hospital.