Eating disorders (EDs) are complex mental illnesses characterised by markedly abnormal eating habits that affect physical health, which may impact on normal daily activities. The four main types of eating disorder are: anorexia nervosa (AN), bulimia nervosa (BN), binge-eating disorder (BED) and atypical or non-specific eating disorder (with the latter satisfying most ED criteria, but not all).
EDs may co-exist with other health problems but there is scarce literature on related physical health conditions that may develop later in life. An ED can involve excessive eating and/or purging through self-induced vomiting and/or laxative misuse, which may cause severe weight change and chemical imbalances within the body. Other complications may occur including osteoporosis and fracture from reduced bone mineral density; heart problems due to chemical imbalance and reduced muscle strength and reduced fatty deposits; gastrointestinal problems from complication of slowing of the digestive system. Previous studies have reported higher incidence of self-harm and substance misuse and elevated risk of premature death, especially by suicide.
Using a large electronic primary care patient health record database with linkage to hospital and death records we aim to find out what conditions an individual with an eating disorder is more likely to develop than a person without an eating disorder. We also aim to determine the risk of premature death by natural causes and by external causes. This will be the largest primary care cohort of individuals with an eating disorder to be conducted. One of its aims will be to find out what future physical health outcomes they develop. It will help in closer monitoring and work towards reducing these risks after an eating disorder has been diagnosed.
The aim of the proposed study is to examine whether eating disorders are associated with increased risk of premature death, osteoporosis, fracture and organ failure as well as co-occurring medical conditions and complications such as diabetes, gastro-intestinal problems, substance abuse and psychiatric comorbidities. Using a matched cohort study design, individuals diagnosed with an eating disorder (including anorexia nervosa, bulimia nervosa, binge eating disorder and atypical eating disorder) between 1st January 1998 and 30th November 2018 will be identified using Read, SNOMED and local EMIS codes. Individuals will be eligible for linkage to IMD, HES and ONS mortality records and will have been continuously registered with an up-to-standard general practice for at least one year prior to their diagnosis. Individuals with a diagnosed ED will be extracted and matched on age, gender and practice with up to 20 comparators without a recorded history of eating disorders on the matched individuals diagnosed index date.
Physical and psychiatric health before and after ED diagnosis will be identified, including diabetes mellitus (both type 1 and type 2), gastro-intestinal problems, depression, anxiety disorders, obsessive compulsive disorder (OCD), borderline personality disorder and drug/alcohol misuse. Physical outcomes including osteoporosis and bone fracture, and organ failure, including heart, renal and liver failure will also be identified. All event data will be extracted from contacts through primary and secondary care data sources for individuals with and without an ED. In addition, non-fatal self-harm episodes and alcohol/drug misuse will be identified. Incidence rates will be estimated for all-cause mortality and specific causes of death including natural, unnatural and suicide, non-fatal self-harm and physical health outcomes. Cox regression models, estimating adjusted hazard ratios for each of the fatal and non-fatal adverse outcomes, will indicate whether individuals with ED have elevated morbidity and mortality risk.
We will estimate incidence rates for all-cause mortality, specific causes of death including natural, unnatural and suicide, non-fatal self-harm, osteoporosis, bone fractures and organ failure (including heart, renal, liver) among patients with and without an eating disorder diagnosis. Cox regression models will be used to estimate adjusted hazard ratios for all-cause mortality, death from specific causes (natural, unnatural and suicide), non-fatal self-harm, osteoporosis, bone fractures and organ failure (heart, renal and liver).
Darren Ashcroft - Chief Investigator - University of Manchester
Cathy Morgan - Corresponding Applicant - University of Manchester
Carolyn Chew-Graham - Collaborator - Keele University
Matthew Carr - Collaborator - University of Manchester
Roger Webb - Collaborator - University of Manchester
HES Accident and Emergency;HES Admitted Patient Care;HES Outpatient;ONS Death Registration Data;Patient Level Index of Multiple Deprivation