This observational study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the Cambridge Central Research Ethics Committee (11/EE/0312). Written informed consent was obtained from all subjects.
Fifty ulcerative colitis (UC) patients were recruited from gastrointestinal outpatient clinics at Lister Hospital (Stevenage), Addenbrooke’s Hospital (Cambridge) and Queen Elizabeth II (QE2) Hospital (Welwyn Garden City), all in the United Kingdom. Fifty healthy control subjects were recruited from the general population in the Cambridgeshire area. Controls were matched with cases for gender and age (within 5 years).
All participants completed a 7-day, estimated food diary and a EuroQol EQ-5D-5 L health-related quality of life questionnaire. In addition, patients with ulcerative colitis completed a disease specific quality of life questionnaire, namely the Inflammatory Bowel Disease Questionnaire (IBDQ), as well as a disease activity form, i.e. the Simple Clinical Colitis Activity Index (SCCAI).
Eligible case patients included those aged 18 – 80 with a confirmed diagnosis of ulcerative colitis by histological and/or radiological techniques and with disease in remission at the time of enrolment in the trial (minimum of 3 months). Clinical remission was defined as SCCAI < 5 . Exclusion criteria were active UC (i.e. SCCAI ≥ 5), anaemia of any kind (haemoglobin < 12 g/dL), other chronic disease, any form of iron therapy, proton-pump inhibitor therapy, eating disorders, pregnancy or breast-feeding. Patients who had received over-the-counter iron supplementation, erythropoiesis stimulating agents (EPO) or blood transfusions in the previous 28 days were also excluded.
The control group was recruited from the general population and consisted of healthy subjects aged 18–80 years old with no history of chronic disease, gastrointestinal disease or anaemia. Further exclusion criteria were as above.
Quality of life questionnaires
The EuroQol EQ-5D-5 L is self-administered and is comprised of five dimensions of health: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression [15, 16]. Each dimension has five levels: (1) no problems, (2) slight problems, (3) moderate problems, (4) severe problems, (5) inability/extreme problems. An ‘index value’ from 0–1 is generated using the Crosswalk Index Value Calculator [17, 18]. In addition, the questionnaire captures a self-rating of health status by a visual analogue scale (VAS) which is limited at 100 (best imaginable health) and 0 (worst imaginable health).
The McMaster IBDQ is specific to inflammatory bowel disease and has been extensively validated [19–21]. This questionnaire is more sensitive to changes in quality of life for patients with inflammatory bowel disease compared to general health questionnaires, and is able to display changes in disease activity for patients with ulcerative colitis or Crohn’s disease . The McMaster IBDQ consists of 32 questions designed to assess four different domains: emotional, bowel, social and systemic. Each question is scored on a seven point Likert scale (1, worst function; 7, best function) and the final score ranges from 32 to 224. Higher scores indicate a better quality of life. The questionnaire was self-administered.
The Simple Clinical Colitis Activity Index (SCCAI) was used to determine whether patients remained in disease remission following referral by their gastrointestinal consultant and to provide a formal IBD disease activity score upon enrolment in the study . The SCCAI consists of 4 questions regarding bowel movements, one question regarding general wellbeing and an area to report any extra-colonic features (one point per manifestation). A score of 5 or more indicates disease relapse .
All subjects completed a 7-day estimated diet diary which has been validated with data from the European Prospective Investigation into Cancer (EPIC) . Subjects were asked to record everything they had to eat or drink and to use household measurements to describe portion size. Subjects also collected all food packaging to increase the accuracy of the dietary data entry. Data from the diet diaries were recorded using in-house programme DINO (Diet In Nutrients Out) which is an all-in-one dietary recording and analysis system written in Microsoft Access . DINO was developed at MRC Human Nutrition Research and incorporates the Department of Health’s Nutrient Databank  and McCance and Widdowson’s Composition of Foods series . Quality was assured by a random 10% check of all entered diaries. After data entry, each participant’s mean energy and nutrient intake over the seven diary days were calculated.
Additionally, as before  fortificant iron intakes were calculated separately as detailed below and comprised foods that are fortified with iron on a voluntary basis by manufacturers (e.g. breakfast cereals), as well as foods containing white flour (e.g. bread), which is restored with iron by law. Fortificant iron was calculated as follows: (i) for food products which are fortified with iron to varying levels at the discretion of the manufacturer, e.g. breakfast cereals, the level of ‘natural’ (i.e. not added) non-haem iron was estimated based on the ingredient information using the McCance and Widdowson food composition tables , and this value was then subtracted from the total iron content declared by the manufacturer to estimate fortificant iron content; and (ii) for other foods containing white wheat flour, which is fortified with iron as a requirement under the UK Bread and Flour Regulation 1995, fortificant iron was estimated by considering the content of white flour per food portion and using the standard ‘restoration’ formula of 1.65 mg Fe/100 g white flour to derive the content of fortificant iron per food portion. All values for the different categories of dietary iron were calculated as total daily intakes for each subject.
A sample size of 50 subjects per group was calculated to give 88% power to detect a 2 point decrease in the disease specific quality of life score for each 1 mg increase in iron intake. This sample calculation was based on our previous data  demonstrating a similar effect size in a linear regression analysis with residual standard error of 26 for the quality of life score and standard deviation for iron intake of 5.25.
Unless otherwise stated, all statistical analysis was performed using GraphPad Prism 6 for Windows (GraphPad Software, San Diego, California, USA). Normality of the data was tested with the D’Agostino and Pearson omnibus normality test. Comparisons between case patients and controls in terms of diet and quality of life were made using the Mann–Whitney nonparametric test. Results are presented as means with standard deviations (SD). Significance of associations between the intakes of the different fractions of dietary iron and quality of life were assessed using simple linear regression analysis (Gaussian distribution of X-data not assumed) and multiple linear regression (with SPSS Statistics 21). Comparison of the slopes for the respective associations of quality of life and dietary iron intakes, between UC patients with disease in remission and those with disease in relapse, were performed by analysis of covariance (ANCOVA). Statistical significance was assumed at p < 0.05.