The Alignment of Dietary Intake and Symptom-Reporting Capture Periods in Studies Assessing Associations between Food and Functional Gastrointestinal Disorder Symptoms: A Systematic Review

Food ingestion is heavily implicated in inducing symptoms of irritable bowel syndrome (IBS) and functional dyspepsia (FD), which affect over one-third of adults in developed countries. The primary aim of this paper was to assess the alignment of dietary assessment and symptom-reporting capture periods in diet-related studies on IBS or FD in adults. Secondary aims were to compare the degree of alignment, validity of symptom-reporting tools and reported significant associations between food ingestion and symptoms. A five-database systematic literature search resulted in 40 included studies, from which data were extracted and collated. The food/diet and symptom capture periods matched exactly in 60% (n = 24/40) of studies, overlapped in 30% (n = 12/40) of studies and were not aligned in 10% (n = 4/40) of studies. Only 30% (n = 12/40) of studies that reported a significant association between food and global gastrointestinal symptoms used a validated symptom-reporting tool. Of the thirty (75%) studies that reported at least one significant association between individual gastrointestinal symptoms and dietary intake, only four (13%) used a validated symptom tool. Guidelines to ensure that validated symptom-reporting tools are matched with fit-for-purpose dietary assessment methods are needed to minimise discrepancies in the alignment of food and symptom tools, in order to progress functional gastrointestinal disorder research.


Introduction
More than one-third of adults in developed countries have chronic unexplained gastrointestinal (GI) symptoms that are classified as a functional GI disorder (FGID) [1,2]. FGIDs are characterised by the absence of any structural or biochemical explanation for gastrointestinal symptoms [3,4]. Despite the The assessment methods most commonly used to prospectively examine dietary intake are 24 hour (h) recalls, weighed food records and estimated food records. Diet histories, food-frequency questionnaires (FFQ) and short assessment screeners are the main methods of examining intake retrospectively. Within each method, considerable variation exists depending on whether the validated tool has been modified for use in the population group, the frequency and duration of implementation, and the skills and experience of researchers [18]. The cost, time and burden to participants and researchers of the more specific, current or "real-time" methods can be prohibitive [18] and thus retrospective dietary assessment methods are often more feasible, but the capture period may be inappropriate for the study hypothesis and research question. For example, a food frequency questionnaire that assesses usual intake over the previous three months is not suitable to investigate associations between dietary intake and postprandial gastrointestinal symptoms immediately after eating.
Given these potential inadequacies, the aim of this systematic review was to investigate dietary assessment methodology in functional gastrointestinal research by critically assessing the alignment of the dietary assessment capture period and FGID symptom-reporting period in studies that examined relationships and reported on significant associations between dietary intake and FD or IBS symptoms.

Search Strategy
A systematic literature search was conducted in five databases (MEDLINE, Medline in process, EMBASE, PsycINFO and the Cochrane central register of controlled trials) for studies that examined associations between FGID symptoms and food or food components in adults (18 years and older). The search terms related to the following three domains (i) symptoms of functional dyspepsia OR irritable bowel syndrome OR presumed gastroduodenal symptoms (e.g., epigastric pain, early satiety, postprandial fullness, belching, nausea or vomiting, diarrhoea and constipation) and (ii) a dietary factor, nutrient, food or food component and (iii) a dietary assessment method. The search strategy is described in detail in Supplementary Table S1. The search was restricted to English language, human studies and timeframe limited from January 2000 to April 2019.

Primary and Secondary Outcomes
The primary outcome was the alignment of the symptom-reporting timeframe and the dietary intake capture period of the dietary assessment tool. The secondary outcome was the proportions of studies with full, partial or no overlap between dietary intake and symptom assessment that reported significant associations between food ingestion and FGID symptoms. The participants, interventions, comparisons, outcomes and study design (PICOS) used to systematically review relationship between food ingestion and symptom reporting in functional gastrointestinal disorders is shown in Supplementary Table S2.

Inclusion Criteria
Studies were included if (i) participants were adults (18 years or older) who were reported as being diagnosed with IBS or FD using Rome criteria (I or II or III or IV) and had a negative upper endoscopy or colonoscopy and (ii) the paper reported on at least one FGID (functional dyspepsia and irritable bowel syndrome) symptom and (iii) the paper reported on a dietary factor, either nutrient, food or food component and the method of assessing the dietary factor and (iv) the paper reported on the relationship between at least one dietary intake factor and at least one FGID symptom. The review considered experimental and epidemiological study designs including randomized controlled trials, non-randomized controlled trials, pre-post studies, prospective and retrospective cohort studies, case control studies and analytical cross-sectional studies.

Exclusion Criteria
Studies were excluded if participants were children or adolescents under 18 years old, if the ROME criteria was not reported as being used for diagnosis of the FGID, if they did not exclude participants who had a positive upper endoscopy for FD (to excluded peptic ulcer, oesophagitis, coeliac disease or cancer) or colonoscopy for IBS (to rule out the presence of inflammatory bowel disease or cancer) or did not have a separate study arm for participants with only FGID diagnosis. Studies were excluded if they reported on eating but did not report on a dietary factor and the relationship between the dietary factor/s and FGID symptoms. Study types that were excluded were letters to editor, conference abstracts, reviews, case reports, and research theses.

Data Extraction
Data extraction and reporting of results was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist [19] (Supplementary Table S3). All studies identified in the database searches were retrieved, consolidated, screened for duplication and then assessed for inclusion using Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia: available at www.covidence.org). Two reviewers independently assessed relevance for inclusion by screening titles and abstracts against inclusion criteria (GD and MP). Conflicts were resolved by a third reviewer and full texts of included abstracts were retrieved (KD). The full texts of retrieved studies were subsequently assessed by two independent reviewers (KD and GD) to determine whether they met inclusion criteria. Conflicts were resolved by a third reviewer (TB).
Data were extracted from included studies using a purpose-built excel spreadsheet tool developed by the authors, with the following variables extracted: study location (Country); study design; year and duration of study; participant characteristics (e.g., age, gender); dietary assessment method/s; dietary outcomes; food intake/consumption or administration method or variable; FGID diagnosis; FGID symptoms, FGID symptom-reporting tool or method. The tool was piloted on the first five studies, with minor amendments made before finalising extraction of data from all included studies.
The dietary instrument/s reported were extracted into the spreadsheet, along with the individual FGID symptoms reported (either by participants or as study outcome measures) and whether the instrument was reported as being validated (V) or modified from a validated instrument (m-V). The timeframe or sequence of symptom/s reporting and collection was also extracted, and whether the tool was a self-report tool or researcher-administered tool.
Dietary assessment methods were categorised using the features shown in Table 1. Additionally, studies that reported on FGID symptom responses to a dietary protocol were included if the dietary intake measures, the dietary regime and the associated symptom-reporting methods were described. If the dietary assessment method was reported as not validated, validated or modified from a validated, this information was extracted. A method was considered to be validated if the paper stated that a validated dietary assessment method had been use in it's original format with a comparable study population. If validation was not mentioned, this was recorded as not reported (NR). The timeframe/sequence of dietary data collection was also extracted.
The reported relationship between dietary intake and FGID symptoms was categorised by the effect type, based on the analysis and study type. Categories were: "group by time" effect for significant differences in symptom responses to food between the control and intervention group (or between intervention arms) over the intervention period; "group" effect for significant difference in symptom responses to food between groups cross-sectionally or for retrospectively collected data; "time" effect for a significant difference in a cohort over the intervention period; "not significant if the study reported no significant group or time effect"; "not measured"; and "not applicable".
Two researchers (GD and KD) extracted data independently and compared extracted data, with inconsistencies rechecked to achieve consensus or referred to a third reviewer for adjudication (TB).

Critical Appraisal
Study quality appraisal was conducted to ensure that findings from respective studies could be contextualized. Articles meeting inclusion criteria were assessed for study quality, using the Academy of Nutrition and Dietetics Quality Criteria Checklist for Primary Research standardized tool [20]. This 10-point checklist includes four relevance and six validity questions to assess population bias, study blinding, the description of the intervention and assessment tool, statistical methods, and study funding. For each criterion, outcomes are rated as being absent, present, unclear or not applicable. No studies were excluded based on quality ratings. Quality assessment was conducted independently by two reviewers (KD and TB), and consensus was achieved without a third reviewer.

Data Analysis
Descriptive statistics were used to analyse the study characteristics and results of included studies. For the study characteristics, mean and range calculations were applied to discrete data. Counts, sums and proportions were applied to summarise and analyse categorical data, and to generate frequency tables. Tables, graphs and numerical summary tools were used to characterise the data and to identify and display patterns when comparing dietary intake capture data and symptom-reporting data. The protocol for this review was registered with the International Prospective Register of Systematic Reviews (PROSPERO, University of York Centre for Reviews and Dissemination), with registration number CRD42018105776XXX. Table S1) identified 12,626 citations ( Figure 1). After removal of duplicates, 11,659 citations were screened on titles and abstracts. Following the screening on title and abstract, 140 articles remained for full text screening. Forty articles met the inclusion criteria and were included in this systematic review ( Figure 1) . Reasons for the exclusion of the other 100 articles are also shown in Figure 1. Table S1) identified 12,626 citations ( Figure 1). After removal of duplicates, 11,659 citations were screened on titles and abstracts. Following the screening on title and abstract, 140 articles remained for full text screening. Forty articles met the inclusion criteria and were included in this systematic review ( Figure 1) . Reasons for the exclusion of the other 100 articles are also shown in Figure 1. The results of this review have been divided into sections. Firstly, the study characteristics, dietary assessment methods, symptom-reporting data and study quality of the 40 included studies are described in Section 3.1, 3.2 and Table 2. Secondly, diagnostic criteria and symptom data are then outlined in Section 3.3, Table 3 and Figure 2. Thirdly, we examined the dietary assessment methods used in the included studies, and the results of this analysis are described in Section 3.4 and Figure  3; Figure 4. Finally, the alignment of the dietary assessment method (DAM) and symptom reporting was reviewed in Section 3.5 and Figure 5.

Study Characteristics
A total of 10,997 (mean 275 per study, range 20 to 4763) participants were involved in the studies, including control group participants. As shown in Table 2, eight included studies were conducted The results of this review have been divided into sections. Firstly, the study characteristics, dietary assessment methods, symptom-reporting data and study quality of the 40 included studies are described in Sections 3.1 and 3.2 and Table 2. Secondly, diagnostic criteria and symptom data are then outlined in Section 3.3, Table 3 and Figure 2. Thirdly, we examined the dietary assessment methods used in the included studies, and the results of this analysis are described in Section 3.4 and Figure 3; Figure 4. Finally, the alignment of the dietary assessment method (DAM) and symptom reporting was reviewed in Section 3.5 and Figure 5.       Options: "never", "1-3 times", "4-6 times", "7-9 times" or "> 10 times" Self-report Baseline (usual intake) Symptoms: Distressing postprandial fullness, early satiation and/or epigastric pain or epigastric burning Measures: Four-item rating scale (i.e., "never" or "rarely", "sometimes", "often", and "always")  Severity of symptoms (score 1-12) calculated as the product of severity (mild, moderate, severe (score 1-3)) and frequency (one day or less per week, two to three days per week, four to five days per week, more than five days per week (score 1-4)) Self-report
The most commonly used instruments for the assessment of global symptomology were the IBS-SSS (n = 10 studies), validated for a capture period of the previous 10 days; the Gastrointestinal Stool Form (BSF), which was used in three studies. Non-validated visual assessment scales were used in 11 studies, and other non-validated symptom rating scales were used in eight studies.
The original validated IBS-SSS was the sole symptom assessment method used in four studies [27,47,59,60] and was used in a modified format in one study [23]. The IBS-SSS was used in conjunction with visual analogue scales (VASs) for individual symptoms in another three studies [32,34,35] and with the BSF used in two studies [26,40].
The validated "adequate relief" tool was used as the sole assessment method in three studies [28,53,57], and in addition to the validated GSRS in one study [55]. The GSRS was used in conjunction with non-validated individual symptom assessment tools in two other studies [28,56]. The validated Birmingham IBS Symptom Score was the overall symptomology assessment method in two studies [41,44]. The validated patient assessment of the upper gastrointestinal symptom severity index (PAGI-SYM) was the symptom assessment tool in one study [37] and the modified Rome III criteria in one study [58]. A non-validated "overall symptoms" score was used in addition to VAS for individual symptoms in two studies [38,42]. Overall symptom rating scales that were not reported as being validated were used to assess symptoms in six studies [22,25,30,39,43,45].
Visual analogue scales (VASs) for individual symptoms were the single symptom assessment method used in five studies [31,36,[49][50][51]-none of which reported the VAS as being validated. VASs were used in addition to the validated BSF (n = 1 study) [48]. Other individual symptom rating scales that were not reported by authors as being validated were used as the single symptom assessment method in three studies [21,46,52], and four studies assessed symptom induction in relation to food consumption using a non-validated tool [24,29,33,54].
The reporting period was highly variable between studies: with daily reporting ranging from three days [45] and 42 days [31]; weekly reporting for between three [48] and eight weeks [53]; and monthly reporting for between one [56] and 16 months [40]. Overall, validated tools were used to measure global FGID symptomology, but the availability of validated tools for assessing individual symptoms was very limited.

Dietary Assessment Methods (DAMs)
Prospective dietary assessment methods included food diaries of between two and 42 days duration (n = 13 studies [22,24,26,28,29,31,35,38,41,43,45,46,55], and weighed food record (n = 2) over a period of seven days [50,51]. Retrospective dietary assessment methods included food frequency questionnaires (n = 12) [21,25,27,30,32,33,37,39,44,52,54,56,58]-of which, five reported some form of validation [25,30,32,44,56]. Dietary protocols or regimes were used in 12 studies [23,34,36,40,42,[47][48][49]53,57,59,60]. The dietary factors assessed using the respective dietary assessment methods are shown in Figure 3 and detailed in Table 2.  symptom-reporting data collection that was longer but overlapped with dietary intake data collection [28,45,55]. Of the studies in which dietary assessment and symptom reporting were aligned at an interval of weeks to months, the data collection capture period matched exactly in three studies [32,42,48] and two studies included a dietary intake data collection that was longer but overlapped with symptom reporting [35,49]. Capture periods were aligned in the nine studies that reported usual dietary intake and current symptoms [21,27,30,33,37,39,44,54,58]. The capture period for dietary assessment and symptom reporting was not aligned in four studies [22,26,52,56]. Twenty five out of the 33 studies (76%) that reported on dietary intake and global gastrointestinal symptoms reported at least one significant association. Of these, 12 (36%) used a validated symptom-reporting tool, 11 (33%) used non-validated tools and three (9%) were not capture period aligned (Table 2). Of the seven studies (24%) that reported non-significant results, six (18%) used a validated global symptom assessment tool and one (3%) used a non-validated tool. Thirty out of 31 (97%) of the studies that reported on associations between individual gastrointestinal symptoms and dietary intake reported at least one significant association. Of these, four (13%) used a validated symptom assessment tool, 23 (74%) used a non-validated symptom assessment tool, and three (10%) were not aligned. The remaining study did not use a validated tool and did not report a significant finding ( Figure 5). Overall, a small proportion of studies that reported significant associations between dietary factors and symptoms used a validated tool for assessing individual or global symptoms and showed alignment between the capture periods for dietary intake and symptom reporting.

Discussion
In comparing the alignment of dietary assessment and symptom-reporting capture periods, this systematic review is a novel addition to the FGID literature. Over 80% of included studies assessed dietary factors that would be expected to induce symptoms within a timeframe from minutes to hours after consumption, but the majority of these involved dietary assessment methods suited to a longer capture period. Additionally, validated tools were routinely used to measure global IBS and FD but not available for individual FGID symptom assessment. In combination, these results highlight the need for validated symptom-reporting tools that are matched with fit-for-purpose dietary assessment methods.
Although diet is now routinely implicated in symptom induction for both IBS and FD [11,12], only 40 studies were identified that had addressed associations between dietary intake assessment and FGID symptom reporting. Collation of the study characteristics showed a high representation from the United Kingdom, Scandinavia and Australasia, and a higher representation of IBS-focused studies (75%) rather than FD-specific studies (25%). Both features are consistent with broader FGID literature [61]. The high proportion of intervention studies (62.5%) compared to retrospective or cross-sectional study designs contributed to the positive study ratings for 55% of included studies.
The use of Rome criteria for the diagnosis of the FGID was an inclusion criterion, and all 40 studies employed the most current version at the time of study implementation. The recent release of the Rome IV criteria is the likely reason that no studies reported using these diagnostic criteria [62]. There was some overlap between symptoms that were reported on in FD-specific and IBS-specific studies. As expected, the symptoms reported only in IBS-specific studies were bowel related and the symptoms only reported in FD-specific studies were epigastric pain, postprandial fullness/distress
Rome criteria (versions I, II or III) were used in all studies to determine the patient's diagnosis, as indicated in the study inclusion criteria and displayed in Table 3. No studies used Rome IV (2016), the most recent update of the Rome criteria. All 40 studies involved self-reporting of symptoms by participants and three [36,54,57] also involved documentation by researchers of responses to dietary challenges. The most commonly reported (or collected) symptoms for IBS were abdominal pain (n = 22), bloating (n = 18) and gas or wind (n = 18). Symptoms reported or collected only in IBS studies were overall GI symptoms (n = 11) and bowel specific symptoms including urgency, constipation, diarrhoea and borborygmi (Figure 2). The most commonly reported or collected symptoms in FD-specific studies were postprandial fullness/distress (n = 8), nausea (n = 6) and bloating (n = 5). The symptoms that were reported only in FD studies were postprandial fullness/distress (n = 8), epigastric pain (n = 4) and hunger (n = 1) ( Table 2).
The most commonly used instruments for the assessment of global symptomology were the IBS-SSS (n = 10 studies), validated for a capture period of the previous 10 days; the Gastrointestinal Symptom Rating Scale (GSRS) (n = 3), validated for capture over the previous seven days; the Birmingham IBS Symptom Score (n = 2 studies), validated over a capture period of the previous four weeks; and the "adequate relief" question, validated for use over the previous week (n = 4). The only individual symptom assessment instrument reported as being validated was the validated Bristol Stool Form (BSF), which was used in three studies. Non-validated visual assessment scales were used in 11 studies, and other non-validated symptom rating scales were used in eight studies.
The original validated IBS-SSS was the sole symptom assessment method used in four studies [27,47,59,60] and was used in a modified format in one study [23]. The IBS-SSS was used in conjunction with visual analogue scales (VASs) for individual symptoms in another three studies [32,34,35] and with the BSF used in two studies [26,40].
The validated "adequate relief" tool was used as the sole assessment method in three studies [28,53,57], and in addition to the validated GSRS in one study [55]. The GSRS was used in conjunction with non-validated individual symptom assessment tools in two other studies [28,56]. The validated Birmingham IBS Symptom Score was the overall symptomology assessment method in two studies [41,44]. The validated patient assessment of the upper gastrointestinal symptom severity index (PAGI-SYM) was the symptom assessment tool in one study [37] and the modified Rome III criteria in one study [58]. A non-validated "overall symptoms" score was used in addition to VAS for individual symptoms in two studies [38,42]. Overall symptom rating scales that were not reported as being validated were used to assess symptoms in six studies [22,25,30,39,43,45].
Visual analogue scales (VASs) for individual symptoms were the single symptom assessment method used in five studies [31,36,[49][50][51]-none of which reported the VAS as being validated. VASs were used in addition to the validated BSF (n = 1 study) [48]. Other individual symptom rating scales that were not reported by authors as being validated were used as the single symptom assessment method in three studies [21,46,52], and four studies assessed symptom induction in relation to food consumption using a non-validated tool [24,29,33,54].
The reporting period was highly variable between studies: with daily reporting ranging from three days [45] and 42 days [31]; weekly reporting for between three [48] and eight weeks [53]; and monthly reporting for between one [56] and 16 months [40]. Overall, validated tools were used to measure global FGID symptomology, but the availability of validated tools for assessing individual symptoms was very limited.

The Alignment of Dietary Assessment and Symptom Reporting
The dietary assessment and symptom-reporting capture periods matched exactly in 15 out of 27 prospective studies ( Figure 4). Of the studies in which alignment matched at a daily level, 10 out of 16 matched exactly [24,29,36,38,43,46,50,51,57,59], three studies included a dietary intake data collection that was longer but inclusive of symptom reporting [23,31,53] and three studies involved symptom-reporting data collection that was longer but overlapped with dietary intake data collection [28,45,55]. Of the studies in which dietary assessment and symptom reporting were aligned at an interval of weeks to months, the data collection capture period matched exactly in three studies [32,42,48] and two studies included a dietary intake data collection that was longer but overlapped with symptom reporting [35,49]. Capture periods were aligned in the nine studies that reported usual dietary intake and current symptoms [21,27,30,33,37,39,44,54,58]. The capture period for dietary assessment and symptom reporting was not aligned in four studies [22,26,52,56].
Twenty five out of the 33 studies (76%) that reported on dietary intake and global gastrointestinal symptoms reported at least one significant association. Of these, 12 (36%) used a validated symptom-reporting tool, 11 (33%) used non-validated tools and three (9%) were not capture period aligned (Table 2). Of the seven studies (24%) that reported non-significant results, six (18%) used a validated global symptom assessment tool and one (3%) used a non-validated tool. Thirty out of 31 (97%) of the studies that reported on associations between individual gastrointestinal symptoms and dietary intake reported at least one significant association. Of these, four (13%) used a validated symptom assessment tool, 23 (74%) used a non-validated symptom assessment tool, and three (10%) were not aligned. The remaining study did not use a validated tool and did not report a significant finding ( Figure 5). Overall, a small proportion of studies that reported significant associations between dietary factors and symptoms used a validated tool for assessing individual or global symptoms and showed alignment between the capture periods for dietary intake and symptom reporting.

Discussion
In comparing the alignment of dietary assessment and symptom-reporting capture periods, this systematic review is a novel addition to the FGID literature. Over 80% of included studies assessed dietary factors that would be expected to induce symptoms within a timeframe from minutes to hours after consumption, but the majority of these involved dietary assessment methods suited to a longer capture period. Additionally, validated tools were routinely used to measure global IBS and FD but not available for individual FGID symptom assessment. In combination, these results highlight the need for validated symptom-reporting tools that are matched with fit-for-purpose dietary assessment methods.
Although diet is now routinely implicated in symptom induction for both IBS and FD [11,12], only 40 studies were identified that had addressed associations between dietary intake assessment and FGID symptom reporting. Collation of the study characteristics showed a high representation from the United Kingdom, Scandinavia and Australasia, and a higher representation of IBS-focused studies (75%) rather than FD-specific studies (25%). Both features are consistent with broader FGID literature [61]. The high proportion of intervention studies (62.5%) compared to retrospective or cross-sectional study designs contributed to the positive study ratings for 55% of included studies.
The use of Rome criteria for the diagnosis of the FGID was an inclusion criterion, and all 40 studies employed the most current version at the time of study implementation. The recent release of the Rome IV criteria is the likely reason that no studies reported using these diagnostic criteria [62]. There was some overlap between symptoms that were reported on in FD-specific and IBS-specific studies. As expected, the symptoms reported only in IBS-specific studies were bowel related and the symptoms only reported in FD-specific studies were epigastric pain, postprandial fullness/distress and hunger. The symptoms that were reported on in both FD-specific and IBS-specific studies were consistent with diagnostic criteria or the overlapping symptom and diagnostic profiles for these two FGIDs [62].
Validated instruments were used to assess global IBS and FD symptomology over capture periods ranging from one to four weeks in less than half of the included studies, and the only individual symptom-reporting tool reported as being validated was the Bristol Stool Form, used in 8% of studies. To date, individual symptom-rating scales are not validated in the same manner as the overall symptom-reporting instruments [20]. However, the high proportion of included studies (n = 19, 48%) that measured individual symptoms on a meal-by-meal or daily basis shows that such tools are more closely aligned to dietary intake capture periods for prospectively collected dietary intake data. This finding highlights the need for validated individual FGID symptom-reporting tools and for a closer consideration of symptom-reporting and dietary intake capture period alignment in future food-related FGID studies.
As expected, food diaries and weighed food records were more common in studies that investigated symptom induction or change in symptoms prospectively over time in relation to a dietary intervention. Food frequency questionnaires were generally used in studies that retrospectively investigated presence of symptoms over previous weeks or months. Food frequency questionnaires are considered suitable for assessing changes in dietary intake over time, and for ranking consumption between individuals, but lack the sensitivity to be used in studies focused on symptom induction [18]. The included dietary protocol studies that compared a specific regime to a usual care control group (or over time) were either conducted in highly controlled laboratory settings, and investigated a limited number of dietary factors, or were implemented in a free-living setting with regular dietetic support. The dietary assessment methods in these studies were less well described, with protocol adherence being the only dietary assessment measure in seven of the 12 studies, and four studies not reporting on protocol adherence in the findings.
The small proportion (17%) of studies that assessed habitual diet were suited to the use of the FFQ dietary assessment method. As food chemical intolerance symptom induction can take several weeks to become evident, FFQs may also be useful in studies investigating this dietary factor. The remaining studies assessed dietary factors that would be expected to induce symptoms within a much shorter timeframe (minutes to hours) after consumption. For example, FODMAPs [12] and fibre [11] induce symptoms from within hours up to several days and allergens have an immediate effect. However, the majority of these studies employed dietary assessment methods suited to a longer capture period, which brings the clinical significance of findings into question.
An important additional consideration in reviewing dietary assessment tools used in functional gastrointestinal disorder research is the limitations of the food composition database linked to the dietary assessment tool. A common limitation of these country-specific food composition databases is the absence or lack of specific information on fibre (e.g., resistant starches) and fermentable carbohydrates (e.g., oligosaccharides) [63]. Gaps in food composition data availability result in incomplete dietary intake information, which in turn affects the completeness and quality of dietary intake data used in FGID research. When considered in combination with the limited use of validated individual symptom-reporting tools, these findings highlight the need for improvements in the methodology of diet-focused functional gastrointestinal research.
Although the capture periods for food and symptom reporting matched or overlapped in 90% (n = 36) of studies, only 30% of studies that reported on global gastrointestinal symptoms and 13% that reported on individual gastrointestinal symptoms that identified an association with food ingestion used a validated symptom tool. Therefore, guidelines to ensure validated symptom-reporting tools are matched with fit-for-purpose dietary assessment methods are needed to minimise discrepancies in the alignment of food and symptom tools in order to progress functional gastrointestinal disorder research.
Recommendations that would be addressed include: (i) An explicit explanation of hypothesised relationship between food and FGID symptoms; (ii) A clear differentiation between assessment of global symptomology and presence or induction of individual symptom; (iii) The selection of symptom-reporting tools be based on whether the study is investigating symptom induction versus presence (or retrospective assessment) of symptoms; (iv) The use of symptom-reporting tools that have been validated for use over the timeframe they were implemented (wherever possible); (v) The selection of appropriate dietary assessment methodologies and the implementation of these methods by researchers with dietary assessment expertise; (vi) The explicit explanation of the rationale for selected dietary assessment methodology, including the hypothesised association with symptom induction and the food composition database (and included nutrients) used for nutrient analysis; (vii) For dietary protocol implementation studies, more detailed reporting of dietary intake assessment than reporting of adherence alone; (viii) Protocols for the assessment of symptom induction in free-living settings, including dietary assessment methods that have a low participant and research burden but high specificity and sensitivity; (ix) The reporting of both significant and non-significant findings to reduce reporting bias.

Conclusions
This review summarises the body of research from studies aimed at assessing food and symptom relationships in IBS or FD. The findings indicate that the dietary assessment tools and symptom-reporting instruments used in these studies are often mismatched. The recommendations produced from this review are aimed at ensuring validated symptom-reporting tools are matched with fit-for-purpose dietary assessment methods to minimise discrepancies in the alignment of food and symptom tools. The implementation of these recommendations in future research will improve the determination of relationships between food ingestion and the presence or induction of FGID symptoms.

Limitations
This study is the first to report on the alignment of symptom reporting and dietary intake assessment methods in FGID research, addressed through studies that focused on the two most common FGIDs, IBS and FD. The findings will facilitate a more structured integration of dietary assessment into FGID research. Limitations associated with the review process are related to an inadequate description of methods in included studies and the need to subsequently extract and categorise this incomplete data. Examples of such limitations include the inadequate description of dietary assessment methods in dietary protocol studies and the reporting of only significant results rather than both significant and non-significant results, particularly in intervention studies with a control group. Recommendations to improve or reduce the impact of these limitations have been outlined in this review. It is also acknowledged that the wide range and large number of dietary factors and symptoms reported within studies increased the likelihood of at least one significant association being identified. Another consideration in FGID research is that study participants are likely to have made dietary changes prior to study participation, which may affect the assessment of usual dietary intake. For example, the reported intake in an FFQ (over previous three months) may not be sensitive to foods that have been reduced or eliminated by those with FGIDs prior to that dietary assessment capture period. It is important to take these factors into consideration when assessing study findings, along with the reported misalignment of dietary assessment and symptom-reporting instruments.
Supplementary Materials: The following are available online at http://www.mdpi.com/2072-6643/11/11/2590/s1: Table S1: Systematic review protocol of dietary factors that influence functional dyspepsia, Table S2: Participants, interventions, comparisons, outcomes and study design (PICOS) used to systematically review relationship between food ingestion and symptom reporting in functional gastrointestinal disorders, Table S3: PRISMA checklist for systematic review of the alignment of dietary intake and symptom-reporting capture periods in studies assessing associations between food and functional gastrointestinal disorder symptoms.
Author Contributions: K.D., T.B. and S.K. initiated the project, all authors contributing to define the database search criteria. M.P., G.D. and K.D. performed the article search. All authors contributed to article screening, data analyses and the writing of the manuscript. All authors approved the final manuscript.
Funding: This research received no external funding.