Gastro-oesophageal reflux symptoms, oesophagitis and Barrett’s oesophagus in the general population: the Loiano–Monghidoro study
- R M Zagari1,
- L Fuccio1,
- M-A Wallander2,
- S Johansson3,
- R Fiocca4,
- S Casanova5,
- B Y Farahmand6,
- C C Winchester7,
- E Roda1,
- F Bazzoli1
- 1Department of Internal Medicine and Gastroenterology, University of Bologna, Italy
- 2Department of Public Health and Caring Science, Uppsala University, Sweden
- 3Section of Preventive Cardiology, Gothenburg University, Sweden
- 4Division of Anatomic Pathology, University of Genoa, Italy
- 5Department of Oncology and Haematology, Sant’Orsola Hospital, Bologna, Italy
- 6Institute of Environmental Medicine, Karolinska Institute, Stockholm, Sweden
- 7Department of General Practice and Primary Care, University of Aberdeen, UK
- Professor F Bazzoli, Department of Internal Medicine and Gastroenterology, Bologna University, Policlinico Sant’Orsola, Via Massarenti n. 9, 40138 Bologna, Italy;
- Revised 4 April 2008
- Accepted 8 April 2008
- Published Online First 18 April 2008
Objective: Existing endoscopy-based data on gastro-oesophageal reflux disease (GORD) in the general population are scarce. This study aimed to evaluate typical symptoms and complications of GORD, and their associated risk factors, in a representative sample of the Italian population.
Methods: 1533 adults from two Italian villages were approached to undergo symptom assessment using a validated questionnaire and upper gastrointestinal endoscopy. Data were obtained from 1033 individuals (67.4% response rate).
Results: The prevalence of reflux symptoms was 44.3%; 23.7% of the population experienced such symptoms on at least 2 days per week (frequent symptoms). The prevalence rates of oesophagitis and Barrett’s oesophagus in the population were 11.8% and 1.3%, respectively. Both frequent (relative risk (RR) 2.6; 95% confidence interval (CI) 1.7 to 3.9) and infrequent (RR 1.9; 95% CI 1.2 to 3.0) reflux symptoms were associated with the presence of oesophagitis. No reflux symptoms were reported by 32.8% of individuals with oesophagitis and 46.2% of those with Barrett’s oesophagus. Hiatus hernia was associated with frequent reflux symptoms and oesophagitis, and was present in 76.9% of those with Barrett’s oesophagus. We found no association between body mass index and reflux symptoms or oesophagitis.
Conclusions: GORD is common in Italy, but the prevalence of Barrett’s oesophagus in the community is lower than has been reported in selected populations. Both frequent and infrequent reflux symptoms are associated with an increased risk of oesophagitis. Individuals with oesophagitis and Barrett’s oesophagus often have no reflux symptoms.
Gastro-oesophageal reflux disease (GORD) is characterised by symptoms of heartburn and regurgitation, and by oesophageal complications such as oesophagitis, Barrett’s oesophagus and oesophageal adenocarcinoma.1 The prevalence of GORD appears to be increasing2 and several recent epidemiological studies based on the evaluation of reflux symptoms have examined potential risk factors that might explain this trend, focusing particularly on lifestyle factors such as diet and obesity.3 4 However, a difficulty facing epidemiological studies of GORD is that the presence of oesophagitis and Barrett’s oesophagus can be assessed only by endoscopy. Both oesophagitis and Barrett’s oesophagus are important because they are thought to indicate an increased risk of oesophageal adenocarcinoma, but, because of the difficulties associated with performing endoscopy in population-based studies, only one study has reported their population prevalence.5 This study in northern Sweden showed that 15.5% of the general population had oesophagitis and 1.6% had Barrett’s oesophagus.5 It was notable that 36.8% of individuals with oesophagitis and 43.7% of those with Barrett’s oesophagus reported no reflux symptoms during the past 3 months. Further studies are needed to improve current understanding of the prevalence of oesophagitis and Barrett’s oesophagus in the general population, particularly in relation to typical reflux symptoms. Such investigations may also clarify the role of putative risk factors for symptoms and complications of GORD, and provide useful information for the screening, prevention and treatment of GORD and its complications. The aim of the present study was to assess the prevalence of, and risk factors for, typical reflux symptoms, oesophagitis and Barrett’s oesophagus in a representative sample of the population of two villages in northern Italy.
The study population was recruited from Loiano and Monghidoro, two villages near Bologna in northern Italy. This population was originally sampled as part of the Multicentre Italian Study on Cholelithiasis (MICOL),6 7 a population-based epidemiological study that included two cross-sectional surveys estimating the prevalence and incidence of gallstone disease. In the first survey, carried out between 1985 and 1987, all 3547 inhabitants of Loiano and Monghidoro aged 18–69 years (total population = 5198) were identified from the electoral roll and invited to participate; of these, 2357 inhabitants took part (a response rate of 66.5%).8 Ten years later, in 1996, the second MICOL survey was carried out. Of the 2357 subjects who participated in the first MICOL survey, 129 had died and 72 had moved away; the remaining 2156 subjects were invited to participate in the second survey. Of these, 1533 (71.1%) individuals aged 28–80 years gave informed consent to undergo 13C-urea breath testing as part of an additional study of Helicobacter pylori infection.9 Between 2000 and 2004, the individuals who participated in the H pylori study were invited to take part in the present study. They were sent an explanatory letter and contacted by telephone to arrange an appointment. If they were unable to attend the first appointment, a second letter was sent and followed up with a second telephone call. This procedure was repeated until an appointment was made or participation was declined. All participants were asked to withdraw from treatment with proton pump inhibitors, H2-receptor antagonists and antibiotics at least 4 weeks before assessment.
During a single visit, participants underwent symptom evaluation, 13C-urea breath testing and upper gastrointestinal endoscopy, with gastric biopsy to detect H pylori infection. Data were also collected on age, sex, weight, height, smoking, consumption of alcohol and coffee, and use of aspirin, non-steroidal anti-inflammatory drugs (NSAIDs) and corticosteroids.
Trained physician interviewers completed a detailed symptom assessment for each patient using a validated gastrointestinal symptom questionnaire, that has previously been shown to be reproducible.10 This questionnaire recorded the frequency and severity of clinically relevant gastrointestinal symptoms during the previous 12 months, using the Rome II cut-off which defines clinically relevant symptoms as those occurring for “at least 12 weeks, not necessarily consecutive”.11 For the purposes of this study, the typical reflux symptoms of heartburn (described as “a burning pain or discomfort behind the breastbone in the chest”) and acid regurgitation (“a bitter or sour fluid coming into the throat or mouth”) were recorded as frequent (symptoms on at least 2 days per week) or infrequent (symptoms on fewer than 2 days per week).
Upper gastrointestinal endoscopy was conducted by three experienced endoscopists from the Gastroenterology and Endoscopy Unit of Sant’Orsola Hospital, who were blinded to each patient’s symptoms and medical history. Before starting the study, the endoscopists agreed on the standardised definition and terminology that would be used for upper gastrointestinal lesions; video clips were also used for this purpose.
Oesophagitis was graded using a modified Savary–Miller classification (grade I, single or multiple non-confluent erosions; grade II, confluent non-circumferential multiple erosions; grade III, circumferential erosions; grade IV, ulcer and/or stricture).12 Mucosal erythema of the oesophagus was excluded from this definition. Barrett’s oesophagus (grade V in some versions of the Savary–Miller classification) was recorded separately. Pairs of biopsy specimens were taken at 2 cm intervals from any areas of suspected Barrett’s oesophagus (endoscopically recognisable salmon-coloured oesophageal mucosa). Barrett’s oesophagus was confirmed when, on staining with haematoxylin/eosin and alcian blue, biopsy specimens were seen to contain specialised intestinalised epithelium with goblet cells.13 Biopsy specimens were independently examined by two experienced pathologists (RF and SC). Short-segment Barrett’s oesophagus was defined as less than 2 cm long, while long-segment Barrett’s oesophagus was defined as at least 2 cm in length.14
Hiatus hernia was recorded if the gastro-oesophageal junction was seen to extend at least 2 cm above the diaphragmatic hiatal impression during quiet respiration. Measurement of hiatus hernia was performed at the end of the endoscopic examination after deflation of the stomach.
Multiple gastric biopsies were taken to detect H pylori: one antral biopsy was taken for the rapid urease test (CP Test; Yamanouchi Pharma SPA, Carugate, Milan, Italy) and a further six biopsies were taken for histology (two from the antrum, one from the angulus, two from the corpus and one from the fundus). For histology, all biopsies were embedded in paraffin and sectioned, and sections were stained with haematoxylin/eosin and modified Giemsa stain. The 13C-urea breath test was performed as described previously.9 If two of the three assessments (the rapid urease test, histology and the 13C-urea breath test) were positive, H pylori infection was considered to be present.
We used relative risks estimated by prevalence ratios to measure the associations between a range of potential risk factors and reflux symptoms or oesophagitis. These were estimated using a modification of Cox’s proportional hazards model with equal exposure time using the PHREG procedure in SAS (Statistical Analysis System version 9.1). The multivariate analyses presented here were adjusted for age (32–44, 45–54, 55–64 and ⩾65 years), sex, body mass index (BMI) (underweight (<20 kg/m2), normal (20–24.9 kg/m2), overweight (25–29.9 kg/m2) and obese (⩾30 kg/m2)), smoking (never, former or current), alcohol (none/occasional, weekly or daily) and coffee consumption (none vs any), use of aspirin and NSAIDs, patient-reported history of peptic ulcer disease, H pylori infection and the presence of hiatus hernia.
In total, 1533 people were invited to enter this study and 1033 were enrolled, a participation rate of 67.4%. The study population was representative of the adult population of the two villages as registered on 31 December 2001 (total adult population, 6332), except that individuals under the age of 35 years were under-represented (2.5% of the study population vs 24.4% of the population of the two villages). Table 1 shows the characteristics of the study population. The mean age was 59.7 (standard deviation (SD) 13.3) years for men and 57.7 (SD 12.8) years for women. H pylori infection was detected in 57.7% (596 out of 1033) of the population, and 7.6% (79 out of 1033) of the population was using NSAIDs or aspirin. The study sample was similar to the general Italian population in terms of the levels of obesity, alcohol use and smoking.15
Typical reflux symptoms during the previous year were reported by 458 (44.3%) of the 1033 participants and 245 (23.7%) had such symptoms on two or more days per week (frequent symptoms). Table 2 shows the association between reflux symptoms and a range of demographic and other factors. A patient-reported history of peptic ulcer was significantly associated with both frequent and infrequent reflux symptoms, while hiatus hernia was significantly associated only with frequent symptoms. An association was observed between female sex and infrequent reflux symptoms. Factors that did not increase the risk of frequent or infrequent reflux symptoms included H pylori infection, BMI, current or former smoking, alcohol or coffee consumption, and use of NSAIDs, aspirin or corticosteroids. No association was found between BMI and reflux symptoms even after stratifying subjects by sex.
Oesophagitis was detected by endoscopy in 11.8% (122 out of 1033) of the study population. Of the 122 subjects with oesophagitis, 82.8% (n = 101) had grade I oesophagitis, 13.1% (n = 16) had grade II, 3.3% (n = 4) had grade III and 0.8% (n = 1) had grade IV oesophagitis (an oesophageal ulcer). Both frequent (RR 2.6; 95% CI 1.7 to 3.9) and infrequent (RR 1.9; 95% CI 1.2 to 3.0) reflux symptoms were associated with an increased risk of oesophagitis (table 3). As shown in fig 1, there was a greater prevalence and severity of oesophagitis among those with more frequent reflux symptoms. The prevalence of oesophagitis in subjects without reflux symptoms was 7.0% (40 out of 575).
Risk factors for oesophagitis were male sex and hiatus hernia (table 3). However, a sex-specific analysis showed that, in men, oesophagitis was significantly associated with frequent reflux symptoms (RR 2.7; 95% CI 1.6 to 4.5) but not infrequent reflux symptoms (RR 1.7; 95% CI 0.9 to 3.1). In women, oesophagitis was associated with both frequent (RR 2.2; 95% CI 1.0 to 4.6) and infrequent (RR 2.2; 95% CI 1.0 to 4.7) reflux symptoms. Hiatus hernia was a risk factor for oesophagitis in both men (RR 3.3; 95% CI 1.9 to 5.5) and women (RR 3.5; 95% CI 1.7 to 7.1). There was no significant association between an increased risk of oesophagitis and H pylori infection, lifestyle factors or BMI (table 3). These analyses were conducted with individuals with Barrett’s oesophagus included in the comparison group; the observed associations were unaffected when they were excluded.
Barrett’s oesophagus was suspected in 37 subjects (3.6% of the population; mean age 60 years; 59.5% men) on the basis of endoscopic appearance (salmon-coloured oesophageal mucosa). Of the 37 subjects, six (16.2%) had suspected long-segment Barrett’s oesophagus, 19 (51.4%) had reflux symptoms and 25 (67.6%) were H pylori positive. Barrett’s oesophagus was confirmed histologically in only 13 subjects (1.3% of the study population). Long-segment Barrett’s oesophagus was confirmed in two subjects (0.2% of the study population), and short-segment Barrett’s oesophagus was confirmed in 11 subjects (1.1% of the study population). Of the 13 subjects with Barrett’s oesophagus, six (46.2%) were men, and all were at least 50 years of age; seven (53.8%) had reflux symptoms, 10 (76.9%) were H pylori positive and 10 (76.9%) had hiatus hernia. The prevalence of Barrett’s oesophagus was 1.5% (7 out of 458) in subjects with reflux symptoms and 1.0% (6 out of 575) in those with no reflux symptoms.
Endoscopy-negative frequent reflux symptoms
Of the 245 subjects with frequent reflux symptoms, 186 (75.9%) had negative endoscopic findings (ie, no oesophagitis or Barrett’s oesophagus). When individuals with endoscopy-negative frequent reflux symptoms were compared with individuals with no reflux symptoms, only a patient-reported history of peptic ulcer (46 out 186, 24.7%) was a risk factor for endoscopy-negative reflux symptoms (RR 1.8; 95% CI 1.3 to 2.6). The presence of hiatus hernia (84 out of 186, 45.2%) was not associated with endoscopy-negative reflux symptoms (RR 1.2; 95% CI 0.9 to 1.6). The prevalence of obesity was higher in subjects with endoscopy-negative reflux symptoms (23.1%, 43 out of 186) than in those without reflux symptoms (16.2%, 93 out of 575), but this association was not significant in the study population as a whole (RR 1.3; 95% CI 0.8 to 1.9) or when analysed in men and women separately (data not shown). The prevalence of H pylori infection in individuals with endoscopy-negative reflux symptoms was 55.4% (103 out of 186), which is similar to that found in individuals with no reflux symptoms (57.2%, 329 out of 575). We found no association between endoscopy-negative reflux symptoms and the consumption of aspirin or NSAIDs (data not shown).
We found a high prevalence of frequent reflux symptoms (23.7%) and of oesophagitis (11.8%) in a sample of the general population in Italy. Furthermore, 1.3% of the population was found to have Barrett’s oesophagus. If we assume that all people with reflux symptoms on at least 2 days per week (and any other individuals with oesophagitis or Barrett’s oesophagus) have GORD,1 then the prevalence of GORD was 31.1% (321 out of 1033). Our results are similar to those reported in the Swedish population5 and, therefore, do not support the concept of a north–south gradient in the prevalence of GORD in Europe.3 Previous studies reported a lower prevalence of GORD in Italy16 17 than in northern European countries, but these data may have been biased by a low response rate,16 or an unrepresentative population.17
Although the incidence of Barrett’s oesophagus is increasing,18 we found a lower prevalence of Barrett’s oesophagus than has been reported in selected populations.19 20 This is in accordance with the results of a previous population based study.21 Also in accordance with this study,21 we found that about one-third of subjects with oesophagitis and half of those with Barrett’s oesophagus did not have typical reflux symptoms. The existence of silent GORD has important implications for the screening and prevention of oesophageal adenocarcinoma. As has been previously reported, patients with oesophageal cancer often have no long-standing history of typical reflux symptoms.22 For example, Sanfey and colleagues23 found that 41% of patients with adenocarcinoma of the oesophagus arising from Barrett’s oesophagus had never experienced reflux symptoms. Asymptomatic Barrett’s oesophagus, which might be a result of asymptomatic oesophagitis, may explain this finding. Therefore, the most important clinical consequence of silent Barrett’s oesophagus is the failure to recognise individuals who are at high risk of oesophageal cancer and who may benefit from endoscopic follow-up. No analysis of risk factors for Barrett’s oesophagus was possible in our study, because of the small proportion of people with this complication (n = 13). However, an association with age has been shown previously20 and all those with Barrett’s oesophagus in our study were at least 50 years of age.
We found no association between BMI and reflux symptoms or oesophagitis, either in the population as a whole or after stratifying by sex. Although our data are in accordance with another population based study,24 we have to acknowledge that increasing BMI is a well recognised risk factor for GORD.5 25 26 However, some clarification of our results may be gained from a recent meta-analysis that, evaluating US and European data separately,27 showed that the association between elevated BMI and the presence of GORD reported within the European studies was less clear than that reported in the US studies. A potential explanation may be that the relationship between BMI and percentage body fat differs between different populations and different ethnic groups. It has been reported, for example, that the percentage body fat of individuals with the same BMI differs between Hispanic and European American women.28
Other lifestyle factors, such as smoking, alcohol and coffee consumption, were not significantly associated with reflux symptoms or oesophagitis in our study.3 4 However, a recent population-based twin study has reported an association between frequent smoking and an increased risk of symptoms of GORD.29 The association between lifestyle factors and GORD is still unclear and our findings would support the suggestion that this association is unlikely to be strong.3 4 However, it is sensible to provide advice about lifestyle factors to patients with GORD, although the benefits remain unproven.
We found hiatus hernia to be significantly associated with more frequent reflux symptoms and with oesophagitis; hiatus hernia was also present in two-thirds of patients with Barrett’s oesophagus. This is in accordance with previous studies30 and supports the suggestion that hiatus hernia is likely to predispose individuals towards the development of more severe forms of GORD.
This study is only the second to apply endoscopy to a representative sample of the general population; in this case, the sample consisted of almost 20% of the total adult population of two villages in northern Italy. A further strength is the evaluation of subjects with reflux symptoms occurring at least 2 days per week, which is the recommended frequency threshold for a symptomatic GORD diagnosis,1 separately from those with less frequent symptoms.
The main limitation of this study is that individuals had to be willing to undergo endoscopy to be enrolled in the study. We cannot exclude the possibility that this would create selection bias, as individuals with upper gastrointestinal symptoms will perhaps be more willing to undergo endoscopy than individuals with no symptoms. This possible selection bias could account for the lack of association with BMI and other lifestyle factors discussed earlier. However, in a similar population-based study in Sweden, Aro and colleagues31 showed that endoscopy was not a major, clinically relevant, source of selection bias. They reported that, in their population-based sample of 1000 subjects, the difference in prevalence of reflux symptoms between subjects who agreed to undergo endoscopy and non-responders was relatively small (5–7%) and of minor clinical importance. A second limitation could have been the use of the Savary–Miller classification for the evaluation of oesophagitis, rather than the Los Angeles classification. However, as patients with only mucosal erythema were excluded from the diagnosis of oesophagitis, this should not have resulted in any misdiagnosis of oesophagitis.
In conclusion, typical reflux symptoms and oesophagitis are as frequent in southern as in northern Europe. The prevalence of Barrett’s oesophagus in the general population is lower than reported in selected populations. Both frequent and infrequent reflux symptoms are associated with an increased risk of oesophagitis. Approximately one-third of patients with oesophagitis and half of those with Barrett’s oesophagus report no reflux symptoms.
The authors would like to acknowledge the following collaborators: Drs D Beretti, M L Bianchi, L Ceroni, L De Luca, S Fossi, S Maltoni, C Martuzzi, G Nicolini, P Pozzato and L Ricciardiello. We are also grateful to Dr D Panuccio, Head of the Unit of Internal Medicine of the Hospital of Loiano, and to the primary care physicians of Loiano and Monghidoro – Drs De Stefano, L Gaggioli, C Gaggioli and C L Navarra – for their help in coordinating the study.
Funding: This study was supported by a grant from the Fondazione del Monte di Bologna e Ravenna.
Competing interests: None.
Ethics approval: The study was approved by the ethical committee of Policlinico Sant’Orsola, Bologna, Italy, on 19 September 2000.