Table 3 Characteristics of nine studies included in the review

Garber et al. [5]

12 to 24-year-olds with AN (n = 66) and AAN (n = 50)

Randomised controlled trial. AN and AAN groups were statistically compared using t test. Associations between weight history variables (total weight loss, rate of weight loss and duration of weight loss) and indicators of illness severity on admission were examined (HR and serum phosphorus)

Groups did not differ by weight history or admission HR. Independent of admission weight, lower HR (p = 0.01) was associated with faster loss; lower serum phosphorus was associated with a greater amount (p = 0.04) and longer duration (p = 0.001)

Greater rate, duration and amount of weight loss was associated with increased illness severity, regardless of current weight

Meirer et al. [2]

10 to 17-years-olds with AN. 2 cohorts (2004, 2014) 2 groups, premorbid overweight history (n = 34 and n = 16 in 2004) and premorbid normal weight history (n = 72 and n = 50 in 2014)

Single centre retrospective cohort study. Proportion of AN patients with a history of excess weight over a 10-year period examined. Premorbid normal and overweight history groups statistically compared in terms of medical markers at admission (HR, BP, and duration of amenorrhea) using t-tests

Excess premorbid weight was similar in both cohorts (32% in 2004 versus 29.5% in 2014). The historically overweight subgroup had a lower HR at intake (64.77 versus 69.75, p = 0.03). The total decrease in BMI greater in premorbid overweight group (7 BMI points versus 3.8, p = 0.0001)

Greater rate and amount of weight loss was associated with higher premorbid weight. Screening of AN should not be limited to the underweight and number of AAN cases increasing

Sawyer et al. [15]

13 to 17-year-olds with AAN (n = 42) and AN (n = 118)

Cohort study. Medical markers (HR, BP, temperature, and weight) and psychological markers of adolescents AAN and AN were compared using independent t-tests for continuous variables and χ2 for categorical variables

AAN lost more weight (17.6 kg vs 11.0 kg) over a longer period (13.3 vs 10.2 months) than AN. No significant difference in bradycardia rates (24% vs. 33%;) BP (43% vs. 38%). No difference found in frequency of psychiatric comorbidities (38% vs 45%)

Many adolescents with AAN present with medical instability, despite being healthy weight. Physical and psychological morbidity similar across AN and AAN, with similar proportions requiring hospitalisation and rates of AAN increasing

Whitelaw et al. [4]

12 to 19-year-olds with AN (n = 118) and AAN (n = 53)

Retrospective and prospective cohort studies comparing total and recent weight loss with admission weight as predictors of physical (hypophosphatemia, bradycardia, hypothermia, and hypotension) and psychological complications

Greater total weight loss (p = 0.002) and greater recent weight loss (p = 0.006), but not admission weight, were associated with a lower HR. Greater total weight loss (p = 0.003) and greater recent weight loss (p = 0.02) were associated with bradycardia

Number of AAN cases increasing over years. Weight loss was a stronger predictor than degree of underweight of physical complications leading to hospitalisation. Greater attention to weight loss as a measure of starvation is recommended

Whitelaw et al. [38]

12 to 19-year-olds with AN (n = 73) and EDNOS (n = 26)

6-year retrospective cohort study comparing HR, BP, and biochemistry of AN and EDNOS adolescents using linear regression for continuous variables, or nonparametric rank sum test for non-normal data

Proportion of EDNOS diagnosis increased from 8 to 47% over 6-year period. Hypophosphatemia developed in 41% of AN and in 39% of EDNOS patients. Lowest HR in AN was 45 bpm compared with 47 in EDNOS

Huge increase in the proportion of healthy weight adolescents hospitalised. EDNOS patients experienced a similar degree of medical instability as AN

Hudson et al. [1]

5 to 13-year-olds with AN (n = 76), BN (n = 3), EDNOS (n = 83), OSFED (n = 46)

Prospective surveillance study. T-tests and Mann–Whitney U tests were used to compare BMI, weight loss, HR, BP, and temperature between diagnoses

35% of cases had medical instability at presentation (60% bradycardia, 54% hypotension, 34% dehydration, 26% hypothermia). 52% of cases required admission at diagnosis (73% to a paediatric ward). 41% of cases had medical instability in the absence of underweight

Anthropological indices alone are poor markers for medical instability, clinical assessment is essential. Doctors providing care for children have central role in both the recognition and management of early onset EDs

Ornstein et al. [29]

8 to 22-year-olds with AN (n = 66)

Retrospective chart review. 3 groups (phosphorous < 2.5 mg/dl, < 3.0 mg/dl, and > 3.0 mg/dl) created. Age, weight, % IBW, and phosphorous were compared among groups. Pearson correlation coefficient was used to analyse the relationship between phosphorus and % IBW

Patients who developed moderate hypophosphatemia were significantly more malnourished than those who did not (p = 0.02). Phosphorus nadirs were directly proportional to % IBW (p = 0.01)

Those patients most severely malnourished are at the greatest risk for developing moderate hypophosphatemia

Peebles et al. [3]

8 to 19-year-olds with AN (n = 330), BN (n = 162), and EDNOS (n = 818)

Retrospective chart review. HR, BP, temperature and QTc compared between diagnoses using t-tests and × 2

EDNOS patients had the greatest and fastest weight lost (p = 0.001 and p = 0.005) and 28.7% compared with 38.5% AN patients had bradycardia. AN had highest rate of bradycardia (p = 0.01) and hypotension (p < 0.001) and lowest phosphorous levels (p < 0.005)

Medical complications can occur EDNOS and AN. Medical severity of patients with EDNOS intermediate to that of patients with AN

Whitelaw et al. [38]

12 to 18-year-olds with AN (n = 29)

Retrospective chart review study. Logistic regression used to identify associations between %IBW and phosphorous

%IBW at admission was significantly associated with the subsequent development of hypophosphatemia (p = 0.007)

Significant correlation between %IBW and incidence of hypophosphatemia