Expanding the food database with values of FODMAP content
The process to expand the food database with values of FODMAPs is illustrated in Fig. 1. First the five groups of FODMAPs; fructose, fructan, lactose, GOS and polyols, were added to the list of nutrients within the software program Dietist XP (version 3:2, Kostdata.se). Dietist XP is a nutrition calculation software program widely used in Sweden and is based on the official food database from the Swedish National Food Agency (2012-03-19). Second, all the existing 2039 food item or dishes in Dietist XP was manually checked against reports containing FODMAP-data from Swedish National Food Agency published in year 2007 to 2014 and published peer-reviewed papers. In addition, food items missing from the database but reported in the food diaries (see below) were added (Table 1). Non-carbohydrate foods such as meat, fish and egg were not included in the database as they consequently contain no FODMAPs. Baby food was not included as the database is intended for assessment of FODMAP intake in adults. Swedish sources were used when possible.
When a relevant FODMAP value was available it was assigned to the food item or dish in the database in Dietist XP. White bread is given as an example here: values per 100 g white bread: 0.27 g fructose , 1.02 g fructan  and 0.20 g GOS  were added to the database. All values assigned to the database are in fresh weight. A few values in dry weight have been converted to fresh weight using the data on water content from the food composition tables from Swedish National Food Agency (Swedish food composition tables, SFCT).
Single ingredient food items
The majority of the food items in the database are single ingredient food such as fruits, fruit-juices, vegetables, root-vegetables, milk-products as well as couscous, rice, oatmeal, pasta, potato- chips, rice cakes (plain or flavored), and legumes. When FODMAP-content data was missing the food item was assigned with the same value as a similar food item in Dietist XP, such as fresh broccoli and frozen broccoli. Similarly, brown beans and white beans originate from same plant family as kidney beans and were assigned the values for kidney beans, and bulgur was assigned with the same value as couscous.
Composite food items
The FODMAP values for composite foods, consisting of several FODMAP containing ingredients, were obtained through a recipe calculation procedure . The ingredients list from the data sources were matched with the most similar food item in the SFCT on which the database is built. As neither the SFCT nor the articles presenting data on FODMAP content present the exact amount of ingredients the food items were matched based on quantity of content. This was true for bread in particular, see section “Fructan” below.
Fructose is jointly absorbed with glucose using GLUT-5 and GLUT-2  transporters and fructose in excess of glucose is more likely to be malabsorbed . Fructose will be fermented whenever not properly absorbed. When calculating the total FODMAP value, fructose in excess of glucose is approximated by taking the monosaccharide content from the food records, in most foods free fructose is accompanied by glucose in various proportions . If the glucose content was higher than the fructose content the excess fructose value is set to 0 . Swedish fructose and glucose values from an appendix from the Swedish National Food Agency , analyzed using GC, were used when available. When data on a food item was missing, recent data on free fructose analyzed with different HPLC techniques and fresh weight were used [8,9,10, 16]. A few values were taken from the Finnish food composition database Fineli .
Swedish values for food items in dry weight were chosen for whole grain rye, soft and- crispbreads as well as values for wheat flour and other milling fractions of wheat [13, 14, 26, 27]. Analytical values of fructan content of Swedish sources were transformed from dry weight to fresh weigh using the water content in food items from SFCT . Values from Australia [8,9,10] and U.K were also used .
Total fructan content in a food item is measured using an enzymatic/spectrophotometric method . In a few cases (< 15) when data on total fructan content in a food item from enzymatic analyses was missing, values from separate HPLC analyses of the main fructo-oligosaccharides e.g. nystose (GF2), kestose (GF3) and 1F-β- fructofuranosylnystose (GF4) were summed up to give an approximation of the total fructan value . When there were discrepancies between papers regarding presence of fructan in a food item, such as bananas, the paper detecting fructan was chosen as the source of the assigned value to ensure that all fructan content in food was captured.
Breads from SFCT were grouped depending on their main type of flours e.g. white bread, granary bread, whole grain/meal wheat bread, sifted rye breads containing both rye and wheat flour, whole meal/grain rye breads and yeast fermented crisp bread with sourdough or non-leavened, non-yeast crisp bread. Content of yeast or sourdough was one of the main factors to take into account in the matching as fermentation is known to influence the fructan value . A white bread in Sweden generally contains wheat flour, water, wheat gluten, yeast, with or without sourdough, and differs slightly from the Australian bread which often contains added soy flour . The ingredients were matched to the description of bread type in published data and/or the company’s website to see the type of ingredients and type of flours in the product and the mean value was assigned, such as the values for gluten-free breads from Whelan  and Biesiekierski .
The majority of the analyzed Australian cereals and muesli came from large multinational companies that are sold internationally, e.g. Kellogg’s cornflakes, Weetabix, All-Bran and Rice Crispies. Other cereals and muesli was matched based on the description of the product , especially noting if the muesli contains dried fruit which is the main source of fructose in the food item. The same procedure was used for biscuits.
The values for composite foods containing wheat flour were calculated with 1.3 g of fructan/100 g wheat flour, the mean value from three wheat cultivars analyzed in Sweden .
Most values for lactose were derived from Swedish  and Finnish  dairy companies who have comprehensive data of lactose content in their products although a few Swedish values came from the Swedish National Food Agency.
GOS and polyols
Australian data regarding GOS and polyols in fruits, vegetables and chewing gum [8, 10, 21] were used as no data from Europe was available. Carbohydrates separately analyzed with HPLC e.g., raffinose and stachyose were summed up to give a value of GOS. Sorbitol and mannitol were summed up and called polyols. Values for composite meals containing GOS were calculated by using information about components from food business sites. Values for cough-drops were found on business websites .
Calculation of FODMAP intake from food diaries
Collection of food records
The present study includes 117 estimated four-day food records randomly selected from the Swedish nationwide diet survey Riksmaten  performed from May 2010 until July 2011. Sample size was determined based on constraining the uncertainty around the estimated mean of fructan, as fructan intake is a key parameter for FODMAP diet treatment in clinical practice, to within 10% of the expected mean value. This ensures sufficiently precise estimates of average intake. Assuming an SD of 2.5 and a 95% confidence interval, approximately 110 individuals were required. The achieved sample size slightly exceeds this requirement at 117.
Eligible participants of Riksmaten, 5000 randomly selected Swedish adults aged 19–80 years, were sent a letter with written information about the study and were phoned a few days later by a trained interviewer from Statistics Sweden and asked if they wanted to participate in the Riksmaten study. A total of 1797 persons chose to participate. Participants in the present study were selected to serve as matched healthy controls for 117 patients in another study not yet published and individuals with diabetes, kidney-disease, lactose intolerance or celiac disease had been excluded . The mean age was 39 years (range 18 to 70 years), 89 were women (76%). Body Mass Index (BMI) was on average 24.8. The study Riksmaten was approved by the Regional Ethical Review board at Uppsala University (registration number: 2010/060). All participants gave oral consent.
All participants received written and oral information about keeping a food diary. Food intake was estimated using a portion size guide to facilitate the recording. Participants were instructed to record their food intake immediately after every meal and encouraged to eat as they normally do. Food intake was entered on a website belonging to the Swedish National Food Agency. If a participant was unable to use the computerized web log, a trained person at Statistics Sweden phoned the participant and recorded the food intake during the call. The food records covered all weekdays and seasons on a group level to cover weekday versus weekend and seasonal variations in food intake.
Calculation of FODMAP content in food records
The 117 four-day food records were entered into Dietist XP, version 3:2, with the expanded database, by a registered dietitian. Composite dishes were entered when available in the database and otherwise broken down into their ingredients. Each food record was summarized in Dietist XP and the average intake in grams per day and 95% confidence interval per FODMAP was calculated. Each FODMAP was tabulated by magnitude of most common eaten food item, see supplementary files.
Patient and public involvement
Patients or representatives from the public were not involved in the planning or execution of this research project.