The study outline has been published earlier[11]. In brief, during the period July 2008 and July 2011, patients who met the Rome II criteria for diagnosis were recruited. The IBS patients were registered according to their subtypes: constipation or diarrhea. An individual diagnostic workup was performed including, but not mandatory, blood tests, stool samples, breath tests, endoscopy and histological examination, and X-ray or ultrasound investigations to ensure the exclusion of organic disease or other malabsorption diseases such as lactose intolerance or celiac disease. Exclusion criteria were patients with severe chronic disease, severe chronic constipation (defined as laxative users), patients taking antibiotics or nonsteroidal anti-inflammatory drugs (NSAIDs), and patients whom had previously had performed an FBT or used an FRD.

As previously described[11], the study was designed with a pre-registration period of 2 wk in which the patients followed their individual habitual IBS diet (HID). The patients were then randomized without stratification to continue HID with or without additional FRD (< 2 g fructose per meal) for 12 wk.

The randomization was assisted by The Scientific Department, University Hospital of North Norway, Tromsø.

Individual instructions for the FRD were given both verbally and through written information that included a table in Norwegian showing the fructose content in 91 common food ingredients (a similar table can be found at web site given in reference 15). For a short version of the table of instructions see Table 1.

In addition to daily VAS registrations of abdominal pain/discomfort, bloating, stool frequency and consistency for 4 wk, an SGA registration was completed once weekly for 12 wk. Early dropouts (defined as patients who registered for < 3 wk of the main 12-wk period) were replaced but late dropouts were not replaced. Data from patients that registered for > 3 of the 4 wk were included in the total registration. The main reason for choosing a 4-wk VAS registration was concern about compliance because the subject would have to perform daily registrations throughout the study. After the main registration period, the patients delivered their diaries, underwent FBT, and were instructed in the fructose-rich provocation test for a maximum of 7 d, or for a shorter time if the test provoked IBS symptoms. For the provocation test, patients were told to choose sucrose-rich food and to include ≥ 200 mL of fruit juice with only small amounts of sorbitol in each daily meal[15,16] (e.g., 200 mL orange juice, 8-9 g fructose with no sorbitol content/200 mL apple juice, 15 g fructose and 1 g sorbitol). Study patients were instructed to use the same information table as a guide for both reducing fructose load and ensuring a sufficient intake of fructose (30 g) during the provocation test[15]. The VAS and SGA scores (as compared to the last week of main registration)[11] were logged in a separate provocation diary.

The subjects filled in a symptom registration diary. Each day they marked on a VAS form (0-100 mm) the degree of pain and bloating experienced (0 mm for no symptoms, and 100 mm for maximal symptom score). In addition, they counted the number of stools and gave a description of the stool quality on a scale of 1-7 (Bristol scale)[13].

Based on the experiences from our first study[11], a diagnostic test based on a self-reported (subjective) intolerance to fructose in IBS was constructed. We defined fructose-related food intolerance as a combination of symptom relief associated with dietary fructose restriction and symptom exacerbation following a fructose provocation test. In our previous study[11] the Bland-Altman analysis showed that the technical detection limits (corresponding to 1.96 SD of mean bias) were 18 mm (18% on VAS scale of 100 mm) for pain/discomfort and 17 mm for bloating. Based on these boundaries a response to FRD was defined as > 25 mm relief, whereas > 25 mm worsening of the VAS score during provocation was considered a positive test[11].

Patients determined the SGA of abdominal relief once during every weekend of the study period by entering their assessment in their personal diary. The assessment was completed by answering the following question: Please consider how you have felt the past week with regards to your IBS, in particular your overall wellbeing, symptoms of abdominal discomfort, pain and altered bowel habit compared to how you felt before entering the study). How do you rate your relief (or worsening) of symptoms during the past week? The scale contained five possible answers: (1) completely relieved; (2) considerably relieved; (3) somewhat relieved; (4) unchanged; or (5) worse[17]. Using the SGA score, patients who were somewhat relieved in week 3 and 4, or completely/considerably relieved in at least 1 wk were considered to have responded to the FRD.

Hydrogen (H₂) and methane (CH₄) were measured by a Microlyzer (Quintron Instrument Co. Inc., Milwaukee, WI, United States) in end-expiratory breath samples. After an overnight fast, H₂ and CH₄ levels were measured before drinking 15 mL solution corresponding to 50 g fructose). Measurements were performed every 30 min until a gas peak was reached, or up to 4 h. A high load of fructose was used to minimize false-negative results as indicated by Choi et al[12]. Incomplete absorption was defined as an increase of H₂ > 20 ppm or CH₄ > 12 ppm, or a sum of combined peak increase > 15 ppm. Symptoms during and after the test were recorded.

The statistical analysis included all randomized patients (intention to treat). Patients where split into the two predefined groups according to the study protocol; either a normal IBS diet alone or combined with FRD. A test-retest analysis of SGA was performed by comparing scores at pre-registration with those at 1, 4 and 12 wk in the control group; ∆ values were run using a Wilcoxon signed rank test vs 0. Internal consistency was explored by analysis of variance (ANOVA) and Spearman’s correlation on ∆VAS (week 0 vs week 4) vs SGA score at 4 wk in all included patients. The former analysis also yielded information regarding scale linearity and precision of the SGA measure. Finally, the face validity denoting whether the questions made sense was performed in all of the patients and 10 healthy volunteers.

Patient inclusion in this multicenter study is described in detail in a previous publication[11]. In brief, 310 patients admitted to hospital with IBS symptoms were screened, and 108 did not meet the inclusion criteria. The 202 patients included were randomized, and 182 completed the main registration period of 12 wk. All early dropouts were replaced. All patients reported a combination of constipation and diarrhea. A total of 88 patients were randomized to FRD. Among these, we experienced missing data from 11 patients; nine due to a missing provocation diary and two that missed markings for SGA change in week 4 of the main diary. The remaining 77 patients reported complete VAS and SGA data both during the pre- and main registration periods, as well as a complete registration during the provocation test. We found no significant differences in age, sex ratio, abdominal pain/discomfort, bloating, stool frequency or Bristol scale stool consistency between the FRD + HID and HID groups (Table 2).

Internal consistency was tested by calculating the VAS change for each of the 182 patients by comparing status at 4 wk with pre-registration. These ∆ values were compared to SGA scores at week 4 using the Spearman Rank correlation test. The analysis yielded ρ values of 0.59 (SGA vs pain/discomfort, P < 0.0005); 0.58 (SGA vs bloating, P < 0.0005); and 0.84 (bloating vs pain/discomfort, P < 0.0005). The graph for the control group illustrated in Figure 1 shows that SGA is a stable measure throughout the 3-mo study period. A test-retest analysis was performed by analyzing the control group SGA values in pairs. For each record, the differences between pre-registration week 0 and weeks 1, 4 and 12 were calculated. These delta values were analyzed with a Wilcoxon signed rank test using zero as the median for the null-hypothesis. The three ∆ values were not significantly different from zero (P = 0.41, 0.13, and 0.42 for pre-registration vs week 1, 4, and 12, respectively). Figure 1 shows the raw data distribution of VAS change registrations in the five SGA categories at 4 wk for all 182 study participants. The SGA scale is not linear; it discriminates best between the span of somewhat relieved and towards completely relieved. Two-way ANOVA of this dataset (VAS by SGA × Diet group) was performed and pairwise comparison is presented in Table 3. The traces for the control group in Figures 2 and 3 show that the VAS measure and the SGA rating were stable over time when used in an IBS setting.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Scale and precision of the subjective global assessment measure. At 4 wk, the change in VAS registration (compared to pre-registration values) was calculated. Box and whiskers plot of VAS change in the different subcategories of SGA at 4 wk. It is noted that the scale is not entirely linear, with best discrimination in the left part of the plot, while the right part shows smaller VAS differences between groups. SGA: Subjective global assessment; VAS: Visual analog scale.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Subjective global assessment of irritable bowel syndrome-related symptoms during the whole study. Mean registration (95%CI) for study groups. The control group showed stable mean value during the 2 + 12 wk registration. The mean effect of FRD was marked, showing stable improvement of symptom rating during the whole study. The SGA ratings were: 1: Completely relieved; 2: Considerably relived; 3: Somewhat relieved; 4: Unchanged; 5: Worse. FRD: Fructose-reduced diet; SGA: Subjective global assessment; IBS: Irritable bowel syndrome.

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Visual analog scale registrations of irritable bowel syndrome-related symptoms during the first 2 + 4 wk. Mean registration (95%CI) for the study groups. FRD: Fructose-reduced diet; VAS: Visual analog scale; IBS: Irritable bowel syndrome.

As shown in Table 4, for SGA, there was good sensitivity and specificity of 0.84 and 0.74, respectively, for identifying self-reported DFI. The inclusion of a provocation test in the diagnostic criteria improved the quality of the test criteria; especially the negative predictive value (Table 4). The sensitivity and specificity parameters for the FBT were low (Table 4).

Using our new criteria for the diagnosis of self-reported DFI, we established a diagnostic tool for fructose intolerance based on the results from the agreement testing (frequency analysis) (Table 4). As described in our earlier report[11], a discrepancy was found between the self-reported fructose intolerance and FBT. This was confirmed in the frequency analysis that gave a κ value of -0.13. There was a good agreement between the diagnosis of self-reported DFI and the SGA responses to FRD according to the criteria used (see methods) with a κ value of 0.61 (Table 4). When results from the provocation test were excluded from the diagnostic criteria, the κ value was less precise (κ = 0.53).

The prevalence of self-reported fructose intolerance, defined as a combination of response to FRD and a positive provocation test, was 56% (43 of 77 patients).

A substantial increase in the prevalence of irritable bowel syndrome (IBS) has been observed in the past 20 years. The main symptoms of IBS include abdominal pain, bloating, diarrhea or constipation. During the same period, an increase in the consumption of fructose, as well as processed food and additives has been seen in the general population. Fructose intolerance is regarded as a subgroup of IBS, and it has been proposed that the increase in IBS actually represents fructose intolerance as a result of the increased intake of this sugar. In the recently published FODMAP diet studies, which consist of a diet reduced in fructose and other carbohydrate types, about 50% of the IBS patients are reported to improve their symptoms and visual analog scale (VAS) score.

The self-reported intolerance to fructose intake has been described as fructose malabsorption (FM) due to small intestinal dysfunction. This was first reported in four patients with chronic diarrhea and colic in 1978. Despite later extensive studies the mechanisms behind this disease or medical condition are still unknown. One of the main unresolved problems is whether fructose intolerance is due to an overload of fructose intake and/or a defect in fructose absorption from the intestinal lumen.

Last year the authors published a report from the study Fructose malabsorption in Northern Norway (FINN study) in which the fructose breath test - the only objective test of FM - correlated poorly with self-reported fructose intolerance among IBS patients. In this second report from the FINN study, Subjective Global assessment of abdominal relief (subjective global assessment) was shown to be a valid end-point measure. This report also shows that a high dietary fructose load is the main explanation for this disease.

Validated end-point measures are necessary tools for future studies of self-reported fructose intolerance.

Fructose intolerance is defined as a self-reported intolerance to a normal load of fructose intake. FM is a similar concept, but mainly based on the abnormal intestinal absorption of fructose measured by FBT. The two concepts are used interchangeably in the literature and most likely describe the same medical condition.

Although the findings of this study replicate commonsense practice, this is a useful addition to literature in these days of evidence-based medicine. The sequence the authors followed is historically what happened with the lactose-intolerance studies, where the focus shifted from mucosal lactase measurements to lactose tolerance curves to symptom analysis