Comparing diagnostic performance of Cantonese-Chinese version of Rome IV criteria and a short Reference Standard for functional dyspepsia in China

Before the translation process, official approval was sought from the Rome Foundation for the use and translation of R4DQ-FD. The official English version of the R4DQ-FD were translated to Cantonese by professional medical translators using the five-step forward–backward translation method [14] and according to the guidelines prescribed by the Rome Foundation [15]. The Cantonese R4DQ-FD were then pre-tested in ten FD patients before data collection. Details of the translation process are illustrated in Additional file 1: Appendix 1.

An organic upper GI disease referred to pathologies located in the upper GI tract, identifiable by conventional diagnostic procedures, and is likely to cause improvement or resolution of dyspeptic symptoms when it is improved or eliminated [1, 12, 24]. Examples include, but are not limited to, erosive or reflux oesophagitis, Barrett’s oesophagus, oesophageal carcinoma, gastric ulcer, and H. Pylori-associated gastritis. Patients with any organic upper GI diseases that were likely to explain their dyspeptic symptoms as indicated by the OGD reports were not classified as having Reference Standard or Rome IV-defined FD.

We approached a total of 749 subjects from the GI outpatient department of the Prince of Wales Hospital and the community, with 456 agreeing to participate. One-hundred-and-thirty-seven failed to fulfil the inclusion criteria due to missing valid medical reports (n = 97) and absence of dyspeptic symptoms (n = 40). Eventually, we received 220 completed questionnaires among the remaining 319 participants (Fig. 1). The response rate was 69.0%. No follow-up was needed for this diagnostic study.

Their mean age was 52.7 years (SD: 12.8), and 117 (80.4%) of them were female. The mean body mass index was 22.0 (SD: 3.8). Twelve (5.4%) and sixty (27.2%) patients were regular users of tobacco and alcohol, respectively. Although 156 (70.9%) of them reported having no other chronic conditions, 154 (70.0%) reported that their health status was poor or very poor. Regardless of the presence of organic upper GI diseases as indicated by medical records, 186 (84.5%) and 106 (48.2%) patients fulfilled the Reference Standard and the Rome IV criteria for FD, respectively. Details on the sociodemographic and health-related characteristics are reported in Table 3.

At least one organic upper GI disease was detected in twenty-six (14.0%) of the 186 patients who met the Reference Standard for FD, with erosive or reflux oesophagitis being the most common diagnosis (Table 4). Gastric erosion and gastric ulcer were also diagnosed in 2.1% of the sample. Likewise, erosive or reflux oesophagitis was the most prevalent organic upper GI finding in those who did not meet the Reference Standard. Barrett's oesophagus, gastric erosion, H. Pylori-associated gastritis, gastric ulcer, and unremoved polyps were found in 19.5% of the sample. There was a significant difference between patients who met the Reference Standard and those who did not in the prevalence of erosive or reflux oesophagitis (11.8% for patients meeting the criteria versus 32.4% for those not; p = 0.003), gastric ulcer (0.5% for patients meeting the criteria versus 8.8% for those not; p = 0.010), and unremoved polyps (0.0% for patients meeting the criteria versus 8.8% for those not; p = 0.003).

Of the 186 patients meeting the Reference Standard for FD, sixty-three (33.9%) met the criteria for PDS alone, fourteen (7.5%) met the criteria for EPS alone, and 109 (58.6%) met the criteria for both PDS and EPS (Table 5). Erosive or reflux oesophagitis was the dominant organic disease among the symptom subtypes. Significant difference was detected in the prevalence of erosive or reflux oesophagitis between the three groups of patients (9.5% for PDS patients versus 35.7% for EPS patients versus 8.3% for patients with both PDS and EPS; p = 0.018).

Among the seventy-nine patients with FD according to the Rome IV criteria, seventy-two met the Reference Standard for FD (Table 6), illustrating a sensitivity of 91.1% (95% CI 82.6%–96.4%) (Table 7). In the 141 patients who were not judged to have Rome IV-defined FD, fifty-three did not meet the Reference Standard, illustrating a specificity of 37.6% (95% CI 929.6%–46.1%). The positive LR of the Reference Standard for FD was 1.46 (95% CI 1.26–1.69), and the negative LR was 0.24 (95% CI 0.11–0.49). The AUC for the Reference Standard was 0.64 (95% CI 0.59–0.69).

The performance of the Reference Standard for PDS and EPS was also compared against corresponding criteria in the Rome IV criteria. Among the sixty-six patients judged to have PDS based on the Rome IV criteria, sixty-three met the Reference Standard for PDS (Table 6), illustrating a sensitivity of 95.5% (95% CI 87.3%–99.1%) (Table 7). There were 154 individuals without Rome IV-defined PDS, of whom sixty-six did not meet the Reference Standard, showing a specificity of 42.9% (95% CI 34.9%–51.1%). The positive LR for PDS was 1.67 (95% CI 1.44–1.93), and the negative LR was 0.11 (95% CI 0.04–0.33). The AUC for the Reference Standard was 0.69 (95% CI 0.62–0.76).

Among the forty-two patients with a diagnosis of EPS under the Rome IV criteria, thirty-three met the Reference Standard for EPS (Table 6), demonstrating a sensitivity of 78.6% (95% CI 63.2%–89.7%) (Table 7). For the 178 patients who did not have Rome IV-defined EPS, 106 of them did not meet the Reference Standard as well, illustrating a specificity of 59.6% (95% CI 52.0–66.8). The positive LR for the Reference Standard for EPS was 1.94 (95% CI 1.53–2.47), and the negative LR was 0.36 (95% CI 0.20–0.65). The AUC for the Reference Standard was 0.69 (95% CI 0.64–0.74).

Since no previous studies were designed to investigate the diagnostic performance of the Reference Standard relative to the Rome IV criteria, we cannot compare the findings in this diagnostic study to those in other publications. That said, the Reference Standard has been evaluated against the Rome II criteria and the Rome III criteria for FD [12]. When compared against the Reference Standard, the Rome II criteria had a sensitivity of 71.4% (95% CI 68.4%–74.2%), a specificity of 55.6% (95% CI 51.5%–59.7%), a positive LR of 1.61 (95% CI 1.45–1.78), and a negative LR of 0.51 (95% CI 0.45–0.58). The Rome III criteria had a sensitivity of 60.7% (95% CI 57.5%–63.9%), a specificity of 68.7% (95% CI 64.6%–72.6%), a positive LR of 1.94 (95% CI 1.69–2.22), and a negative LR of 0.57 (95% CI 0.52–0.63).

This cross-sectional study illustrated that the Reference Standard may be sufficient in ruling out patients without Rome IV-defined FD, PDS, or EPS, given relatively high sensitivity values ranging from 78.6 to 95.5%. Their negative LRs of around 0.20 implied that negative test results of the Reference Standard may moderately decrease patients’ post-test probabilities of having FD, PDS, or EPS diagnosis [26]. However, affected by the mediocre specificity values ranging from 37.6 to 59.6%, the Reference Standard may not be useful in ruling in patients with Rome IV-defined FD, PDS, or EPS. The positive LRs of less than 2.0 also revealed that positive test results of the Reference Standard may only slightly increase patients’ post-test probabilities of having an FD, PDS, or EPS diagnosis [26]. The AUC values of around 0.70 indicated that the Reference Standard had moderate accuracy in distinguishing between patients with and without Rome IV-defined FD, PDS, or EPS [27].

With its satisfactory performance in ruling out patients without FD, PDS, or EPS, the Reference Standard may reduce unnecessary initiation of FD treatments. Reduction of over-treatment may reduce treatment-associated adverse events. The use of prokinetics, the recommended first-line therapy for FD in the Asian guideline [8], is associated with adverse events of dystonia, parkinsonism-type movements, tardive dyskinesia, or even life-threatening arrhythmia [9, 28]. Proton pump inhibitors, the first-line therapy for FD in the North American guideline [9], may increase the risk of hip fracture, community-acquired pneumonia, and Clostridium difficile infection [29]. Also, second-line therapy, tricyclic antidepressants, may cause constipation, dry mouth, urinary retention, and somnolence [9, 30].

By ruling out FD effectively, it is expected that the financial burden on FD patients and healthcare systems would be relieved by minimising the chance of initiating unnecessary treatments. A study in 2013 estimated that each FD patient in the United States had to pay, on average, USD805 per year for regular consultations and treatment [31]. These calculations did not consider the indirect cost incurred by absence from work and loss of productivity. A retrospective study in Malaysia also showed that FD is associated with the highest healthcare burden compared to other functional GI disorders in secondary care [32].

In routine practice where consultation time is limited, the Reference Standard may be used as an initial screening tool for FD and FD symptom subtypes prior to confirmation by the Rome IV criteria. If service arrangement allows, the Rome IV criteria should be implemented to confirm the positive results made by the Reference Standard to avoid potential false-positive cases, given the mediocre specificity of the former instrument. Moreover, the Rome IV criteria may be adopted for confirming FD diagnosis among patients who present with persistent dyspepsia symptoms but indeed show negative test results in Reference Standard screenings, facilitating appropriate initiation of treatment. Besides the questionnaire-based criteria, physicians should consider additional patient information [5, 33], such as symptom duration and co-morbidity, when making diagnoses for dyspeptic patients, since patients who do not fully meet the Rome IV criteria may still be offered essential treatments for reducing symptoms and improving quality of life [5, 8]. Additional diagnostic workup may also supplement OGD and H. Pylori tests for differential diagnosis. For example, in areas with a high prevalence of hepatocellular carcinoma like Southern China, upper abdominal ultrasound may be valuable for differentiating epigastric pain caused by malignancy or FD [8].

Although the Rome diagnostic criteria were recognised in the Asian and North American FD guidelines for clinical research [8, 9], they were considered to have limited relevance to routine practice. To prepare for future updates, future research can investigate the feasibility of developing a concise edition of the Rome FD diagnostic criteria based on the Reference Standard, so as to facilitate FD diagnosis in outpatient clinics where consultation time is limited.

Furthermore, to introduce objectivity of FD diagnosis, the potential value of adding duodenal eosinophilia as a diagnostic marker should be investigated, since the phenomenon is closely associated with early satiety and PDS [34]. Gastric emptying is associated with the pathophysiological mechanism of FD [1, 28], so accelerated gastric emptying, delayed emptying, and fasting gastric volume may also be evaluated as potential motility markers for FD diagnosis [28, 35]. Alteration of the GI microbiota may also be chosen as another biomarker [28, 36], given its relationship with the occurrence of functional GI disorders.

Lastly, further diagnostic research may be conducted to investigate whether the Reference Standard is able to differentiate organic dyspepsia from FD. Evidence produced may contribute to reducing unnecessary tests and examinations in routine practice [13].

To the best of our knowledge, this study is the first to translate the Rome IV criteria for FD from English to Cantonese using the standardised forward–backward translation method. We held cognitive debriefing sessions with FD patients to test the clarity, adequacy of cultural adaptation, language usage, and acceptability of the draft translation. Each step in the translation process was monitored and approved by the official Rome Foundation. This study is also the first to compare the Reference Standard against the Rome criteria in terms of diagnostic performance and the first that is conducted in the China Greater Bay Area. Most importantly, this cross-sectional diagnostic study has low risk of bias and concerns over applicability in terms of the four domains in the QUADAS-2. These domains include: (i) patient selection; (ii) index test; (iii) reference standard; and (iv) flow and timing [17].

This study had certain limitations. First, given that only participants in Hong Kong were recruited for cognitive debriefing, local adaptations may be required before adopting the Cantonese R4DQ-FD in other Cantonese-speaking populations in China or overseas. Second, misclassifications of organic upper GI diseases might exist, since OGD and H. Pylori tests within five years were accepted for eligibility screening instead of referring the patient to concurrent diagnostic workup. Third, results from abdominal ultrasound were not included in the eligibility criteria, because the clinical value of the procedure in evaluating organic dyspepsia is limited [37]. Fourth, we used the prevalence of FD in Asia (30%) [38] for the sample size calculation because no information regarding the prevalence of FD in the China Greater Bay Area is available. This diagnostic study would have required a larger sample size if the prevalence of FD in the China Greater Bay Area was, in fact, lower than in Asia. Also, due to the lack of objective gold standard definitions of FD, the prevalence of FD used in this study may not reflect the true prevalence of FD in the continent. Fifth, the diagnostic performance indicators of a diagnostic test may vary between subgroups of participants with different demographical or clinical characteristics [39]. These characteristics may include but are not limited to age, gender, and symptoms’ severity and frequency [39]. However, due to the relatively small sample size, we did not conduct logistic regression analyses to explore the relationships between the diagnostic performance indicators of the Reference Standard and participant subgroups.