Approaches to Diagnosing IBS
The search strategy identified 3361 citations, of which 33 studies appeared to be eligible and were retrieved for evaluation. Twenty-two of these met all eligibility criteria (Figure S1 http://onlinelibrary.wiley.com/doi/10.1111/apt.13283/suppinfo). Agreement between investigators when assessing eligibility was excellent (94% agreement, K = 0.86). The twenty-two included studies evaluated a total of 7106 patients. Eleven of the studies were conducted in Europe, six in North America, four in Asia and one in Australasia. Thirteen of the studies were of cross-sectional design, and nine were case–control. As case–control studies are not representative of the true IBS prevalence, these were excluded when calculating the pooled prevalence of IBS, which was 43% (95% CI 37–50%) in the remaining 13 studies. Twenty-one were conducted in secondary care, with one in both primary and secondary care. Individual study characteristics are summarised in Table S1 http://onlinelibrary.wiley.com/doi/10.1111/apt.13283/suppinfo. The diagnostic tests utilised in the eligible studies are shown in Table 2, along with the number of studies assessing the accuracy of each test, total number of patients included and the positive and negative LRs with 95% CIs (pooled where appropriate). Study bias and applicability outcomes assessed, according to the QUADAS-2 tool, are shown in Table S2 http://onlinelibrary.wiley.com/doi/10.1111/apt.13283/suppinfo. Fourteen of the 22 studies were judged as high risk in one or more of the four key domains.
Five studies evaluated ≥2 of the Manning criteria, six studies ≥3, and six studies ≥4. The Rome I criteria were evaluated in three studies, the Rome II criteria in two studies and the Rome III criteria in only one study. All studies were cross-sectional in design, collected symptom data using a questionnaire completed by the patient, and utilised a reference standard of a normal colonoscopy or barium enema to confirm the diagnosis of IBS.
The Manning criteria. Pooled positive and negative LRs when using ≥2 of the Manning criteria in a total of 2452 patients were 2.20 (95% CI: 1.54–3.14) and 0.18 (95% CI: 0.10–0.32) respectively. In studies assessing ≥3 of the Manning criteria, accuracy was best in the study conducted by Dogan et al., with a positive LR of 7.15 (95% CI: 4.93–10.57) and a negative LR of 0.11 (95% CI: 0.07–0.17). However, this was not replicated in the five other studies, including in the original validation study. The pooled positive and negative LRs, in a total of 2966 patients, were 2.85 (95% CI: 1.95–4.15) and 0.36 (95% CI: 0.24–0.53) respectively (Figure 1a,b). Finally, when data were pooled from the six studies assessing the accuracy of ≥4 of the Manning criteria, positive and negative LRs, in a total of 2986 patients, were 3.43 (95% CI: 2.49–4.71) and 0.66 (95% CI: 0.60–0.73) respectively (Figure 2a,b).
(Enlarge Image)
Figure 1.
(a) Pooled positive likelihood ratio of ≥3 Manning criteria. (b) pooled negative likelihood ratio of ≥3 Manning criteria.
(Enlarge Image)
Figure 2.
(a) Pooled positive likelihood ratio of ≥4 Manning criteria. (b) pooled negative likelihood ratio of ≥4 Manning criteria.
The Rome criteria. Pooled positive and negative LRs in the three studies, containing 3006 patients, that reported on the Rome I criteria were 3.20 (95% CI: 2.29–4.47) and 0.22 (95% CI: 0.10–0.49) respectively. Pooled positive and negative LRs of the two studies, containing 2402 patients, that evaluated the Rome II criteria were 2.56 (95% CI: 1.64–4.00) and 0.25 (95% CI: 0.08–0.85) respectively. The positive and negative LRs in the one study, containing 1848 patients, that reported on the Rome III criteria were 3.35 (95% CI: 2.97–3.79) and 0.39 (95% CI: 0.34–0.46).
Visceral Hypersensitivity and Pain Perception as a Biomarker. One case control study and one cross-sectional study evaluated the role of visceral hypersensitivity using rectal barostat testing and pain perception during colonoscopy, in differentiating IBS from healthy controls (HCs) and miscellaneous gastrointestinal and medical conditions, in a total of 328 patients. Pooled positive and negative LRs in the two studies were 3.71 (95% CI: 2.74–5.02) and 0.16 (95% CI: 0.10–0.24) respectively. One case–control study, containing 138 patients, reported on colonic air insufflation to reproduce typical abdominal pain experienced in IBS as a biomarker in differentiating the disorder from colonic structural disease. This test performed poorly, with positive and negative LRs of 0.98 (95% CI: 0.80–1.60) and 1.09 (95% CI: 0.40–3.98) respectively.
Serum-based Biomarkers. The diagnostic accuracy of a serum-based 10 biomarker panel, which were selected by examining differences in biomarker expression between IBS patients and HCs (among them interleukin-1ß, anti-tissue transglutaminase and anti-neutrophil cytoplasmic antibody), was reported in one cross-sectional study and one case–control study containing a total of 760 patients, with a pooled positive LR of 3.03 (95% CI: 1.49–6.17) and pooled negative LR of 0.52 (95% CI: 0.43–0.64). In the study conducted by Jones et al., an additional 24 serum biomarkers, selected through a combination of gene chip human array, gene array data analysis and real-time quantitative polymerase chain reaction, were added to the original 10 biomarker panel. Positive and negative LRs of the 34 biomarker panel were 2.28 (95% CI: 1.71–3.17) and 0.30 (95% CI: 0.21–0.42) respectively.
Faecal Biomarkers. Volatile Organic Metabolites (VOMs): The diagnostic accuracy of VOMs, chemicals released in faeces and that can undergo changes in the presence of organic disease or changes in microbiota, were assessed in one case–control study containing 30 IBS patients with diarrhoea, 62 patients with active Crohn's disease and 48 patients with active ulcerative colitis. Using a receiver operator characteristics curve (ROC) to determine optimum performance, the positive and negative LRs in differentiating IBS from active inflammatory bowel disease were 4.83 (95% CI: 3.36–7.14) and 0.04 (95% CI: 0.01–0.21) respectively.
Chromogranins and Secretogranins: Chromogranins (Cg) and secretogranins (Sg) are proteins found in the secretory cells of the endocrine, enteric and immune system. In one case–control study, CgB and SgII and SgIII levels were measured in faecal samples from 82 IBS patients and 29 HCs. SgII and SGIII levels were higher in the IBS patients, and CgB levels were lower. SgII performed the most accurately using a cut-off of >0.16 nmol/g, with a positive LR of 3.89 (95% CI: 2.07–8.23) and negative LR of 0.25 (95% CI: 0.15–0.39).
Mucosal Intestinal Endocrine Cells as a Biomarker. Quantification of CgA cells was performed on biopsy samples taken from the duodenum during gastroscopy in one case–control study. Using a cut-off of <200 cells/mm, the positive and negative LRs in differentiating 203 IBS patients from 86 HCs, were 18.5 (95% CI: 7.58–47.3) and 0.14 (95% CI: 0.10–0.20). In a similarly designed study, rectal biopsies were taken from 50 patients with IBS and 27 HCs. Endocrine cell content was quantified and three endocrine cells, (peptide YY, oxyntomodulin and somatostatin) were validated as diagnostic tests. Using optimum performance determined on an ROC curve, peptide YY performed the best, with a positive LR of 7.56 (95% CI: 2.96–21.9) and a negative LR of 0.18 (95% CI: 0.09–0.33) at a cut-off of <30 cells/mm.
Sigmoid Muscularis Propria Thickness as a Biomarker. In a cross-sectional study of 175 female patients who were undergoing trans-vaginal ultrasound for investigation of gynaecological symptoms, sigmoid muscularis propria thickness was measured. A diagnosis of IBS was made using a cut-off for abnormal muscularis propria thickness of ≥3 mm. A clinical diagnosis of IBS was confirmed with the primary physician and/or Gastroenterologist following ultrasound. Positive and negative LRs were 14.9 (95% CI: 7.07–31.5) and 0.31 (95% CI: 0.17–0.51) respectively.
Combinations of Biomarkers. Faecal Calprotectin and Intestinal Permeability Ratio as a Biomarker: In the previously cited study from Tibble et al., all patients provided a stool sample for measurement of faecal calprotectin (FC) levels, in addition to undergoing a lactulose/L-rhamnose small intestinal permeability test. Using an FC level of <10 mg/L and a permeability ratio of <0.05, this biomarker combination was able to identify IBS patients with a positive LR of 8.64 (95% CI: 5.76–13.1) and a negative LR of 0.34 (95% CI: 0.28–0.39).
Bile Acid Secretion and Colonic Transit as a Biomarker: One case–control study used a 2-item model consisting of total faecal bile acid excretion and colonic transit to differentiate between 64 IBS patients with diarrhoea, 30 IBS patients with constipation and 30 HCs. Using the optimum cut-off on an ROC curve, the 2-item model was able to differentiate IBS from HCs with a positive LR of 2.78 (95% CI: 1.55–5.58) and a negative LR of 0.46 (95% CI: 0.33–0.65).
The use of psychological markers in differentiating IBS from health was evaluated in two case–control studies containing 714 patients. In the study conducted by Spiller et al. 319 IBS patients and 151 HCs completed the patient health questionnaire 12 (PHQ-12).
The PHQ-12 differs from the patient health questionnaire-15 (PHQ-15) in that the three specific gastrointestinal-related questions are removed. Using a cut-off score of >6, the positive LR for the PHQ-12 in differentiating IBS from health was 12.5 (95% CI: 6.55–24.6), and the negative LR was 0.35 (95% CI: 0.30–0.41). Using a cut-off score of >7, the anxiety component of the hospital anxiety and depression scale (HADS) was reported to have positive and negative LRs of 2.88 (95% CI: 2.20–3.86) and 0.37 (95% CI: 0.30–0.45) respectively. The depression component of the HADS demonstrated positive and negative LRs of 5.44 (95% CI: 3.01–10.1) and 0.68 (95% CI: 0.62–0.75) respectively.
In the previously described study conducted by Jones et al., participants were asked to complete the HADS, PHQ-15 and the perceived stress scale. Positive and negative LRs of these measures of psychological well-being combined in differentiating between IBS and HCs were 2.95 (95% CI: 2.04–4.48) and 0.35 (95% CI: 0.26–0.46) respectively.
Kruis Statistical Model. The accuracy of the Kruis statistical model, a scoring system that incorporates the clinical history, physical examination and blood tests (erythrocyte sedimentation rate (ESR), leucocyte count and haemoglobin level), was assessed in four cross-sectional studies, including a total of 1171 patients. A score of ≥44 was used as the cut-off to diagnose IBS. The pooled positive LR of these studies as assessed in a previous meta-analysis, as there have been no studies published in the interim, was 8.63 (95% CI: 2.89–25.8) and the pooled negative LR was 0.26 (95% CI: 0.17–0.41).
Other Statistical Models. Frigerio et al., in a cross-sectional study, lowered the predetermined cut-off point of haemoglobin level in the Kruis statistical model from 14 g/100 mL to 13 g/100 mL in males and from 12 g/100 mL to 11 g/100 mL in females. Positive and negative LRs for this modified model, containing 253 patients, were 7.73 (95% CI: 4.83–12.4) and 0.34 (95% CI: 0.22–0.49) respectively.
Although differing from the Kruis model in the items included, the model validated in the cross-sectional study by Bellentani et al., also incorporated the clinical history, physical examination and an ESR and leucocyte count. Positive and negative LRs for this statistical model, containing 254 patients, were 4.29 (95% CI: 2.86–6.66) and 0.30 (95% CI: 0.22–0.39) respectively.
A Combination of FC, Intestinal Permeability Ratio and the Rome I Criteria. In the previously described study by Tibble et al., if a positive result for the Rome I criteria was incorporated with FC levels of <10 mg/L and permeability ratio of <0.05, the positive and negative LRs were 26.4 (95% CI: 11.4–61.9) and 0.51 (95% CI: 0.45–0.56) respectively.
A Combination of Serum-based Biomarkers and Psychological Markers. Finally, in the study conducted by Jones et al., the serum-based 34 biomarker panel and psychological measures were combined to ascertain if this improved accuracy in diagnosing IBS. Positive and negative LRs for this combined approach in differentiating IBS from health were 7.14 (95% CI: 4.01–13.3) and 0.18 (95% CI: 0.12–0.25) respectively.
When individual study results were combined to obtain pooled positive and negative LRs for each of the approaches to diagnose IBS, using all available studies for each of the approaches they assessed, there were significant differences in the pooled positive LR between studies using symptom-based criteria alone (positive LR 2.85; 95% CI: 2.53–3.20), and studies that used a combination of symptoms, biomarkers and psychological markers (positive LR = 8.48; 95% CI: 4.64–15.5), but not between any of the other methods (Figure 3a). Negative LRs were not significantly different for any of the four approaches (Figure 3b).
(Enlarge Image)
Figure 3.
(a) Pooled positive likelihood ratios for all approaches to the diagnosis of IBS. (b) Pooled negative likelihood ratios for all approaches to the diagnosis of IBS.
Results
The search strategy identified 3361 citations, of which 33 studies appeared to be eligible and were retrieved for evaluation. Twenty-two of these met all eligibility criteria (Figure S1 http://onlinelibrary.wiley.com/doi/10.1111/apt.13283/suppinfo). Agreement between investigators when assessing eligibility was excellent (94% agreement, K = 0.86). The twenty-two included studies evaluated a total of 7106 patients. Eleven of the studies were conducted in Europe, six in North America, four in Asia and one in Australasia. Thirteen of the studies were of cross-sectional design, and nine were case–control. As case–control studies are not representative of the true IBS prevalence, these were excluded when calculating the pooled prevalence of IBS, which was 43% (95% CI 37–50%) in the remaining 13 studies. Twenty-one were conducted in secondary care, with one in both primary and secondary care. Individual study characteristics are summarised in Table S1 http://onlinelibrary.wiley.com/doi/10.1111/apt.13283/suppinfo. The diagnostic tests utilised in the eligible studies are shown in Table 2, along with the number of studies assessing the accuracy of each test, total number of patients included and the positive and negative LRs with 95% CIs (pooled where appropriate). Study bias and applicability outcomes assessed, according to the QUADAS-2 tool, are shown in Table S2 http://onlinelibrary.wiley.com/doi/10.1111/apt.13283/suppinfo. Fourteen of the 22 studies were judged as high risk in one or more of the four key domains.
Symptom-based Diagnostic Criteria
Five studies evaluated ≥2 of the Manning criteria, six studies ≥3, and six studies ≥4. The Rome I criteria were evaluated in three studies, the Rome II criteria in two studies and the Rome III criteria in only one study. All studies were cross-sectional in design, collected symptom data using a questionnaire completed by the patient, and utilised a reference standard of a normal colonoscopy or barium enema to confirm the diagnosis of IBS.
The Manning criteria. Pooled positive and negative LRs when using ≥2 of the Manning criteria in a total of 2452 patients were 2.20 (95% CI: 1.54–3.14) and 0.18 (95% CI: 0.10–0.32) respectively. In studies assessing ≥3 of the Manning criteria, accuracy was best in the study conducted by Dogan et al., with a positive LR of 7.15 (95% CI: 4.93–10.57) and a negative LR of 0.11 (95% CI: 0.07–0.17). However, this was not replicated in the five other studies, including in the original validation study. The pooled positive and negative LRs, in a total of 2966 patients, were 2.85 (95% CI: 1.95–4.15) and 0.36 (95% CI: 0.24–0.53) respectively (Figure 1a,b). Finally, when data were pooled from the six studies assessing the accuracy of ≥4 of the Manning criteria, positive and negative LRs, in a total of 2986 patients, were 3.43 (95% CI: 2.49–4.71) and 0.66 (95% CI: 0.60–0.73) respectively (Figure 2a,b).
(Enlarge Image)
Figure 1.
(a) Pooled positive likelihood ratio of ≥3 Manning criteria. (b) pooled negative likelihood ratio of ≥3 Manning criteria.
(Enlarge Image)
Figure 2.
(a) Pooled positive likelihood ratio of ≥4 Manning criteria. (b) pooled negative likelihood ratio of ≥4 Manning criteria.
The Rome criteria. Pooled positive and negative LRs in the three studies, containing 3006 patients, that reported on the Rome I criteria were 3.20 (95% CI: 2.29–4.47) and 0.22 (95% CI: 0.10–0.49) respectively. Pooled positive and negative LRs of the two studies, containing 2402 patients, that evaluated the Rome II criteria were 2.56 (95% CI: 1.64–4.00) and 0.25 (95% CI: 0.08–0.85) respectively. The positive and negative LRs in the one study, containing 1848 patients, that reported on the Rome III criteria were 3.35 (95% CI: 2.97–3.79) and 0.39 (95% CI: 0.34–0.46).
Biomarkers
Visceral Hypersensitivity and Pain Perception as a Biomarker. One case control study and one cross-sectional study evaluated the role of visceral hypersensitivity using rectal barostat testing and pain perception during colonoscopy, in differentiating IBS from healthy controls (HCs) and miscellaneous gastrointestinal and medical conditions, in a total of 328 patients. Pooled positive and negative LRs in the two studies were 3.71 (95% CI: 2.74–5.02) and 0.16 (95% CI: 0.10–0.24) respectively. One case–control study, containing 138 patients, reported on colonic air insufflation to reproduce typical abdominal pain experienced in IBS as a biomarker in differentiating the disorder from colonic structural disease. This test performed poorly, with positive and negative LRs of 0.98 (95% CI: 0.80–1.60) and 1.09 (95% CI: 0.40–3.98) respectively.
Serum-based Biomarkers. The diagnostic accuracy of a serum-based 10 biomarker panel, which were selected by examining differences in biomarker expression between IBS patients and HCs (among them interleukin-1ß, anti-tissue transglutaminase and anti-neutrophil cytoplasmic antibody), was reported in one cross-sectional study and one case–control study containing a total of 760 patients, with a pooled positive LR of 3.03 (95% CI: 1.49–6.17) and pooled negative LR of 0.52 (95% CI: 0.43–0.64). In the study conducted by Jones et al., an additional 24 serum biomarkers, selected through a combination of gene chip human array, gene array data analysis and real-time quantitative polymerase chain reaction, were added to the original 10 biomarker panel. Positive and negative LRs of the 34 biomarker panel were 2.28 (95% CI: 1.71–3.17) and 0.30 (95% CI: 0.21–0.42) respectively.
Faecal Biomarkers. Volatile Organic Metabolites (VOMs): The diagnostic accuracy of VOMs, chemicals released in faeces and that can undergo changes in the presence of organic disease or changes in microbiota, were assessed in one case–control study containing 30 IBS patients with diarrhoea, 62 patients with active Crohn's disease and 48 patients with active ulcerative colitis. Using a receiver operator characteristics curve (ROC) to determine optimum performance, the positive and negative LRs in differentiating IBS from active inflammatory bowel disease were 4.83 (95% CI: 3.36–7.14) and 0.04 (95% CI: 0.01–0.21) respectively.
Chromogranins and Secretogranins: Chromogranins (Cg) and secretogranins (Sg) are proteins found in the secretory cells of the endocrine, enteric and immune system. In one case–control study, CgB and SgII and SgIII levels were measured in faecal samples from 82 IBS patients and 29 HCs. SgII and SGIII levels were higher in the IBS patients, and CgB levels were lower. SgII performed the most accurately using a cut-off of >0.16 nmol/g, with a positive LR of 3.89 (95% CI: 2.07–8.23) and negative LR of 0.25 (95% CI: 0.15–0.39).
Mucosal Intestinal Endocrine Cells as a Biomarker. Quantification of CgA cells was performed on biopsy samples taken from the duodenum during gastroscopy in one case–control study. Using a cut-off of <200 cells/mm, the positive and negative LRs in differentiating 203 IBS patients from 86 HCs, were 18.5 (95% CI: 7.58–47.3) and 0.14 (95% CI: 0.10–0.20). In a similarly designed study, rectal biopsies were taken from 50 patients with IBS and 27 HCs. Endocrine cell content was quantified and three endocrine cells, (peptide YY, oxyntomodulin and somatostatin) were validated as diagnostic tests. Using optimum performance determined on an ROC curve, peptide YY performed the best, with a positive LR of 7.56 (95% CI: 2.96–21.9) and a negative LR of 0.18 (95% CI: 0.09–0.33) at a cut-off of <30 cells/mm.
Sigmoid Muscularis Propria Thickness as a Biomarker. In a cross-sectional study of 175 female patients who were undergoing trans-vaginal ultrasound for investigation of gynaecological symptoms, sigmoid muscularis propria thickness was measured. A diagnosis of IBS was made using a cut-off for abnormal muscularis propria thickness of ≥3 mm. A clinical diagnosis of IBS was confirmed with the primary physician and/or Gastroenterologist following ultrasound. Positive and negative LRs were 14.9 (95% CI: 7.07–31.5) and 0.31 (95% CI: 0.17–0.51) respectively.
Combinations of Biomarkers. Faecal Calprotectin and Intestinal Permeability Ratio as a Biomarker: In the previously cited study from Tibble et al., all patients provided a stool sample for measurement of faecal calprotectin (FC) levels, in addition to undergoing a lactulose/L-rhamnose small intestinal permeability test. Using an FC level of <10 mg/L and a permeability ratio of <0.05, this biomarker combination was able to identify IBS patients with a positive LR of 8.64 (95% CI: 5.76–13.1) and a negative LR of 0.34 (95% CI: 0.28–0.39).
Bile Acid Secretion and Colonic Transit as a Biomarker: One case–control study used a 2-item model consisting of total faecal bile acid excretion and colonic transit to differentiate between 64 IBS patients with diarrhoea, 30 IBS patients with constipation and 30 HCs. Using the optimum cut-off on an ROC curve, the 2-item model was able to differentiate IBS from HCs with a positive LR of 2.78 (95% CI: 1.55–5.58) and a negative LR of 0.46 (95% CI: 0.33–0.65).
Psychological Markers
The use of psychological markers in differentiating IBS from health was evaluated in two case–control studies containing 714 patients. In the study conducted by Spiller et al. 319 IBS patients and 151 HCs completed the patient health questionnaire 12 (PHQ-12).
The PHQ-12 differs from the patient health questionnaire-15 (PHQ-15) in that the three specific gastrointestinal-related questions are removed. Using a cut-off score of >6, the positive LR for the PHQ-12 in differentiating IBS from health was 12.5 (95% CI: 6.55–24.6), and the negative LR was 0.35 (95% CI: 0.30–0.41). Using a cut-off score of >7, the anxiety component of the hospital anxiety and depression scale (HADS) was reported to have positive and negative LRs of 2.88 (95% CI: 2.20–3.86) and 0.37 (95% CI: 0.30–0.45) respectively. The depression component of the HADS demonstrated positive and negative LRs of 5.44 (95% CI: 3.01–10.1) and 0.68 (95% CI: 0.62–0.75) respectively.
In the previously described study conducted by Jones et al., participants were asked to complete the HADS, PHQ-15 and the perceived stress scale. Positive and negative LRs of these measures of psychological well-being combined in differentiating between IBS and HCs were 2.95 (95% CI: 2.04–4.48) and 0.35 (95% CI: 0.26–0.46) respectively.
Combinations of Symptoms, Biomarkers and Psychological Markers
Kruis Statistical Model. The accuracy of the Kruis statistical model, a scoring system that incorporates the clinical history, physical examination and blood tests (erythrocyte sedimentation rate (ESR), leucocyte count and haemoglobin level), was assessed in four cross-sectional studies, including a total of 1171 patients. A score of ≥44 was used as the cut-off to diagnose IBS. The pooled positive LR of these studies as assessed in a previous meta-analysis, as there have been no studies published in the interim, was 8.63 (95% CI: 2.89–25.8) and the pooled negative LR was 0.26 (95% CI: 0.17–0.41).
Other Statistical Models. Frigerio et al., in a cross-sectional study, lowered the predetermined cut-off point of haemoglobin level in the Kruis statistical model from 14 g/100 mL to 13 g/100 mL in males and from 12 g/100 mL to 11 g/100 mL in females. Positive and negative LRs for this modified model, containing 253 patients, were 7.73 (95% CI: 4.83–12.4) and 0.34 (95% CI: 0.22–0.49) respectively.
Although differing from the Kruis model in the items included, the model validated in the cross-sectional study by Bellentani et al., also incorporated the clinical history, physical examination and an ESR and leucocyte count. Positive and negative LRs for this statistical model, containing 254 patients, were 4.29 (95% CI: 2.86–6.66) and 0.30 (95% CI: 0.22–0.39) respectively.
A Combination of FC, Intestinal Permeability Ratio and the Rome I Criteria. In the previously described study by Tibble et al., if a positive result for the Rome I criteria was incorporated with FC levels of <10 mg/L and permeability ratio of <0.05, the positive and negative LRs were 26.4 (95% CI: 11.4–61.9) and 0.51 (95% CI: 0.45–0.56) respectively.
A Combination of Serum-based Biomarkers and Psychological Markers. Finally, in the study conducted by Jones et al., the serum-based 34 biomarker panel and psychological measures were combined to ascertain if this improved accuracy in diagnosing IBS. Positive and negative LRs for this combined approach in differentiating IBS from health were 7.14 (95% CI: 4.01–13.3) and 0.18 (95% CI: 0.12–0.25) respectively.
Pooled Positive and Negative LRs for Each Approach Used to Diagnose IBS
When individual study results were combined to obtain pooled positive and negative LRs for each of the approaches to diagnose IBS, using all available studies for each of the approaches they assessed, there were significant differences in the pooled positive LR between studies using symptom-based criteria alone (positive LR 2.85; 95% CI: 2.53–3.20), and studies that used a combination of symptoms, biomarkers and psychological markers (positive LR = 8.48; 95% CI: 4.64–15.5), but not between any of the other methods (Figure 3a). Negative LRs were not significantly different for any of the four approaches (Figure 3b).
(Enlarge Image)
Figure 3.
(a) Pooled positive likelihood ratios for all approaches to the diagnosis of IBS. (b) Pooled negative likelihood ratios for all approaches to the diagnosis of IBS.
Source...