Table 2 Characteristics, outcome measures and results of included studies

Anderluh et al. (2009)[35] (Slovenia/UK)

Cross-sectional study using clinical sample and retrospective reports

N = 97 female patients. ANR = 35, ANBP = 32, BN = 30. Mean current BMI for AN across subtypes = 16.7. Mean age of AN patients = 27.75 years.

Aimed to define an eating disorders (ED) phenotype by retrospectively assessing lifetime ED symptoms to examine a lifetime pattern of illness.

Initial screening diagnosis by experienced clinician in inpatient service. Participants interviewed by trained researcher. Demographic information collected, weight and height measured.

Bewell-Weiss & Carter (2010) (Canada)[36]

Cross-sectional design using clinical sample

N = 153 (148 female) first-admission inpatients with AN. Mean age = 26.0 years. Mean BMI = 15.0 kg/m².

Aimed to amalgamate findings into a comprehensive regression model of predictors of excessive exercise in patients with AN.

All patients diagnosed by experienced clinician, with EDE. Exercise behaviour classified as excessive (34% of total N)-endorsed obligatory exercise, at least one hour daily, six days per week, for >1 month.

Davis & Claridge (1998) (Canada/UK)[37]

Cross-sectional design using clinical sample

83 female patients, AN = 34, BN = 49, Mean age = 28.1 years. Only patients without a history of another ED classification were included in the study.

Aimed to determine whether patients with eating disorders display addictive and OC personality characteristics relevant to weight preoccupation and excessive exercising.

Participants completed questionnaires and interview was conducted at time of admission to program.

Davis & Kaptein (2006) (Canada)[38]

Prospective design using clinical sample

50 inpatients-ANR. Mean age = 25.4 years. Mean BMI at admission = 14.05. BMI at discharge =19.6.

Aimed to determine whether AN patients represented a phenotype linked with OCD.

Completed questionnaires within first week of admission. Exercise interview conducted shortly after. Follow-up questionnaires were completed as soon as they attained target weight, determined by clinical team.

Davis et al. (1998) (Canada)[2]

Quasi-experimental design using clinical sample

Clinical sample (inpatient and outpatients) classified by DSM-III-R (1987) criteria. AN-Restrictor N =26; AN with bulimia = N = 16. EDNOS with low body weight N = 11. Mean AN BMI = 16.5. Classified as high level exercisers (N = 22, mean age = 27.1) or moderate/non exercisers (N = 31, mean age = 28.8).

Explored relationships between exercise levels, Obsessive Compulsive symptomatology, and restricted eating in AN. Discussed in relation to models of biological mechanisms in AN.

Questionnaires completed, physical activity interview for exercise classification. Height and weight measured after interview.

Davis et al. (1995) (Canada)[39]

Cross-sectional study using clinical and non-clinical sample

Clinical sample; Inpatients with AN, or had met criteria within past year for AN (N = 46, Mean age = 24.2, SD = 4.7). Non-clinical samples: 2 samples of women (n = 88, Mean age = 23.3, SD = 3.8 and n = 40, Mean age =24.7, SD = 3.2)

Aimed to investigate relationship between obsessive-compulsiveness and psychological and behavioural aspects of exercise in women with AN

Clinical sample tested within first 5 or 6 weeks of hospital admission, through questionnaires and interview.

Holtkamp et al. (2004) (Germany)[40]

Cross-sectional study using clinical sample

30 female adolescent inpatients with AN. Mean age = 14.6 years. Mean BMI = 14.4 kg/m².

Examined relationships between restricted diet, increased physical activity and psychopathology in acute stage of AN.

All patients diagnosed by experienced clinician, AN subtype diagnosed by trained interviewer blind to study hypothesis within three days of admission.

Naylor et al. (2011) (UK)[41]

Cross-sectional study comparing clinical and non-clinical samples

Clinical sample recruited from 4 eating disorder services: AN =30; BN = 24; EDNOS = 10. Mean BMI = 19.23, Mean age =29.98. Non-clinical university student sample: mean BMI = 20.86, Mean age = 20.32.

Aimed to explore exercise beliefs, obsessive beliefs and obsessive compulsive behaviours to understand the role of excessive exercise in eating disorders.

Clinical participants assessed and diagnosed by experienced clinicians using semi-structured interview, permission granted to gather diagnoses, BMI from medical file. Nonclinical sample reported height and weight.

Penas-Lledo et al. (2002) (Spain)[31]

Cross-sectional design using retrospective case notes from inpatient service

ANR patients = 35; ANBP patients = 28. BN patients = 61. Mean age = 20.25 years. Mean AN BMI = 16.9

Examined whether physical exercise is related to different aspects of psychopathology and if this association is different between diagnoses.

Data collected routinely by clinicians blind to the hypotheses. Participants coded as excessive exercisers if exercised at least 5× week (>1 hr per session), with aim to burn up calories.

Shroff et al. (2006)[42] USA

Cross-sectional design using clinical sample, and relatives with history of eating disorder.

AN probands and biological relatives who met lifetime diagnosis of AN, N = 431. BN probands and biological affected relatives, N = 750. AN Trios study- probands and parents, N = 749. Resulting sample size = 1857.

Explored features associated with excessive exercise across subtypes of eating disorders.

Participants from multi-site international Price Foundation Genetic study. Probands and relatives assessed for psychological and personality features that may underlie vulnerability to eating disorders.