Natural history of gastroesophageal reflux: A prospective cohort study in a stratified, randomized population in Beijing

To explore the natural history of and risk factors for gastroesophageal reflux (GER) in a prospective cohort in Beijing, China.

(PPIs) for GERD 6 have all been increasing. A few longitudinal studies focusing directly on changes in the prevalence of GERD over time in the same populations 4,[7][8][9][10][11] showed that weekly GER symptoms disappeared in more than half of patients 6 months after treatment in Spain; 7 while symptoms remained stable at a lower prevalence of 2.5% to 2.7% during a 1-year follow-up of the general Chinese population. 8 The long-term prevalence of GER increased from 5.5% to 10.5% from 1994 to 1999 in Singapore, 9 and the prevalence of at least weekly GER was increased by 47.0% (from 11.6% to 17.1%) from 1995 to 2006 in a large, prospective, population-based cohort study in Norway. 4 However, neither of the two longitudinal studies in Sweden showed a significant increase in the prevalence of reflux symptoms during a 7-year and 10-year resurvey. 10,11 These variable results suggest that the natural history of GER differs among populations because of the differences induced by geographic variations in the prevalence of GER, 12 or various factors that affect the trends in the change in GER over time.
In one study on the natural history of GERD and functional abdominal disorders, 10 61% and 29% of patients with initial predominant reflux symptoms continued to experience reflux at 1 and 7 years, respectively. The causes of the persistence of or recovery from GERD were not directly addressed in previous studies. 3,10 Additionally, people with occasional reflux symptoms (ie, once monthly or even fewer episodes) may seldom seek medical care, and such individuals are usually ignored by physicians and investigators. 7,13 Thus, little is known about the natural history of occasional reflux in the general population.

| PARTICIPANTS AND METHODS
This prospective, population-based cohort study included participants selected by a stratified randomized method from a cluster sample in Beijing, China. These participants were initially surveyed in 1996 14,28 and subsequently surveyed again in 2008.

| Initial survey
In the initial study, we selected the Beijing area as a cluster sample, which was stratified into urban, suburban, and rural areas. We then used a random sampling method, gradually drawing out districts, towns, and communities. The community was used as the unit of this study and the sample size in each unit was proportional to the population in this area. In total, 2486 residents aged 18 to 70 years were surveyed among 2500 potential participants by well-trained physicians and medical students in face-to-face interviews, 14,28 and these participants comprised the initial sample population. The response rate of the initial survey was 99.4%.

| Current survey
The current study was performed from September to November, 2008. Well-trained investigators held face-to-face interviews with the participants, and each completed questionnaire was checked by the principal investigators on site. The respondents then signed the questionnaire, indicating their approval. Because of the municipal construction of Beijing, some of the participants in the urban study unit had moved out to decentralized areas after the first survey, and 198 (16.7%) were resurveyed by the senior investigators or the specialists via telephone. The study flowchart is shown in Figure 1. Each respondent received a gift of about $5 after the investigation in appreciation.
Each participant had a unique code in both the initial and current surveys. The patient's data of the current survey were entered into the database by two independent individuals and were checked for consistency. If the data were not consistent, the original questionnaire was re-reviewed and the data were corrected.

| Questionnaire
The questionnaire included items on the particpants' general information, GER, and other gastrointestinal symptoms during the past year, reflux-associated factors, dietary habits and lifestyle, medical history, and previous history of gastroscopy and barium examination during the past 3 years. The intensity of heartburn, acid reflux, and regurgitation of food was recorded as none, mild, moderate, or severe, and its frequency as monthly, weekly, or daily. 14 In light of the accuracy of self-reported symptoms, 29 the recall time was set as the past 3 months in the current survey.

| Risk factors
In total 23 possible risk factors were selected according to the results in the initial study and previous publications. [14][15][16][17][18][19][20][21][22][23] All parameters were carefully defined during the training period of the investigators (Table S1). The questionnaires also included questions about anxiety and depression, such as "In the last 3 months, have you felt tense or wound up?" and "In the last 3 months, have you felt downhearted and low?". If participants answered to these questions with "Often" (≥2 days a week) or "Most of the time", they were suspected to be anxious or depressed. 30 GER was defined as heartburn and/or acid reflux and/or regurgitation of food at least once a week, and monthly reflux was defined as at least one of the abovementioned symptoms occurring at least 1-3 days per month. 1,8 If the participants had either no episodes or episodes on <1 day per month, they were considered reflux-free. Univariate associations were identified by the χ 2 test, variables that were statistically significant in the χ 2 test were included in a multivariate logistic regression model to identify the risk factors associated with GER persistence, aggravation, and new onset. A P value of ≤ 0.05 was considered statistically significant. The calculation of the rate of new-onset GER (per 1000 person-years) was based on symptom-free participants (excluding those with monthly reflux) at the time of the initial survey; therefore, it was not considered to be the true incidence of GER. 4,8 3 | RESULTS

| General data
In total, 1316 (52.9%) participants out of those in the initial survey were followed up in the current survey. Among them, 123 had died and four had impaired abilities and were unable to complete the questionnaire. Therefore, 1189 (47.8%) questionnaires were valid for the final analysis ( Figure 1). The sex, age, percentage of those doing heavy labor, educational level distribution, and family economic status of these 1189 respondents were comparable with those of the 2486 participants in the initial survey except that there were more farmers and fewer jobless or retired participants in the initial survey (Table 1)

| Change in GER over time
Altogether 98 respondents in the initial survey had GER. Of these, 22.4% had had GER over the past 12 years (defined as persistent GER), 15.3% had monthly reflux, and 62.2% became reflux-free over time ( Table 2). The GER group was more likely to have persistent GER over time than the monthly reflux and the no-reflux groups based on the initial survey (22.4% vs 11.9%, P = 0.02; and 22.4% vs 7.5%, P < 0.001, respectively).
There were no significant differences in the initial symptom patterns (ie, heartburn, acid reflux, or regurgitation) between participants with persistent GER and those with either monthly reflux or no reflux in the current survey (Table 3). Participants who initially had a single reflux symptom were more likely to be reflux-free over time than those with two or three symptoms (74.1% vs 47.7%, P = 0.007). Considering any GER symptom with the most severe intensity and the highest frequency, we found that participants with initially mild to moderate daily symptoms were more likely to recover from GER (Table 3). Actually, participants with daily symptoms had a similar number and intensity of reflux symptoms as those with weekly GER.
In the initial survey, six among 1189 participants were identified as having GER only because of food regurgitation, which reduced the prevalence of GER from 8.2% to 7.7%; while in the current survey, nine participants were defined as having GER only because of food regurgitation, which reduced the prevalence of GER from 9.5% to 8.7%.
Neither postprandial nor nocturnal reflux affected the outcome of GER over time. GER with dysphagia, noncardiac chest pain, and the sensation of a foreign body in the throat initially did not affect the trend of GER over time. The disease course of GER in the initial survey did not affect the trend as well.

| Changing trend of monthly reflux over time
Among the 219 participants with monthly reflux in the initial survey, 11.9% developed aggravated GER over time ( Table 2). Participants who initially had monthly reflux were more likely to continue to have monthly reflux and less likely to become reflux-free than the no-reflux group (20.1% vs 8.9%, P < 0.001; and 68.0% vs 83.6%, P < 0.001, respectively). The trend of aggravation from monthly reflux to GER had no correlation with the initial symptom pattern, number, intensity, or episode time point of reflux (all P > 0.05, data not shown).

| Outcome of becoming reflux-free over time
In the initial survey, 872 participants reported that they had experienced no reflux symptoms during the previous year, 83.6% had no reflux over time (including no reflux and reflux on <1 day per month), 7.5% developed GER during the past 12-year period (defined as new-onset GER), and 8.9% developed monthly reflux (  ± 4.30 years, respectively. We found that groups with persistent and aggravated GER had more severe heartburn and acid reflux than the new-onset group in the current survey (all P < 0.05; Figure 2). There was no significant difference in the frequency of episodes among the three groups (Figure 3). There was no significant difference in dysphagia, noncardiac chest pain, and sensation of a foreign body in the throat among the three groups.

| DISCUSSION
The limited studies on the trends of GERD prevalence in the general population have produced variable results. 3,4,[7][8][9][10][11]31 This prospective, population-based cohort study showed that the prevalence of GER at least weekly and reflux symptoms at least monthly remained stable over the entire 12-year period. The initial GER symptom patterns of single, mild to moderate, and daily GER were associated with a greater chance of recovery from reflux over time. We found that more severe heartburn and acid reflux occurred in participants with persistent and aggravated GER than in those with new-onset GER in the current survey and the emotional depression was a risk factor for GER aggravation.
The long-term trends in the prevalence of GER in the general population, the fluctuation of symptoms, and recovery without treatment in a minority of patients with reflux esophagitis indicate that certain patient profiles affect the natural history of GERD. The crucial variable is the definition of GER (ie, number, severity, and frequency of symptoms evaluated in the studies), and these differences might affect the initial prevalence and consequently the trend and natural history of GER. 2,3,7,[9][10][11]32 Most previous population-based studies defined GER as heartburn and/or acid regurgitation, and we found that regurgitation of food is also common among Chinese patients in clinical practice and is a more troublesome symptom than heartburn and/or acid regurgitation for patients; therefore, we added this symptom along with heartburn and acid regurgitation as predominant symptoms by which to define GER in the initial survey. Among the 270 participants with GER in the initial survey of 2486 participants, the prevalence of regurgitation of food was 65.9%, and 47 (17.4%) participants had GER with weekly food regurgitation; however, it seemed that adding the regurgitation of food to the definition of GER did not significantly increase the prevalence of GER (which increased by 0.5% in the initial survey and 0.8% in the current survey, respectively). In this cohort study, the predominant reflux symptoms in the initial questionnaire were differentiated as mild, moderate, or severe in intensity and as monthly, weekly, or daily in frequency, thus ensuring a longitudinal comparative analysis of overall GER and individual reflux symptoms to assess the natural history of GER.
The sampling method, sample size, response rate, and data collec- In this study, 66.9% of participants remained in their initial refluxrelated status for 12 years, and one-third of them experienced changes in their symptoms with a lower onset rate and lower disappearance rate of GER. This kept GER prevalence stable at 8.2% to 9.5% in this cohort population, without significant differences among the different age groups. These data also reveal differences between Eastern and Western populations. 3,7-9 Some risk factors, such as frequently eating sweet food, a history of emotional stress and abuse, and anxiety increase the risk of persistent GER.
Occasional reflux is usually ignored. 13,35 We found that approximately 12% of participants with monthly reflux in the initial survey developed GER over time, and participants who performed heavy physical labor, were depressed, and had insomnia had a higher risk of reflux aggravation over time. Reflux-free individuals showed a lower trend (7.5%) of developing true GER over time. Higher education is likely to be a protective factor against GER onset over time. This finding is similar to that in the HUNT study. 27 Conversely, a 1-year longitudinal study from Hong Kong showed that a higher educational level was associated with GERD onset. 8 The difference in risk factors and the natural history of GER indicate that GERD is heterogeneous, but the effects of psychosocial factors cannot be ignored because they are associated with a high GER prevalence, persistence, aggravation, and new onset based on several previous studies and the present cohort data, especially for the depression, which is a risk factor for GER aggravation as found based on multiple logistic regression analyses. 8,14,16,24,26,36 Ruth et al noted that acid regurgitation at baseline was a predictive factor for the development of heartburn 10 years later (OR 10.0), 11 but the potential mechanism for the transition between these two symptoms remains unclear. A study of the general Chinese population showed that acid regurgitation in the initial survey was associated with the disappearance of GERD. 8 In the present study, we found that the symptom pattern did not affect the trend of GER or monthly reflux over time, while participants with GER persistence and aggravation had more severe heartburn and acid reflux than those with new-onset GER. Increased symptom severity was associated with a poor, disease-related quality of life. 7,37,38 These findings indicate that participants with multiple and severe symptoms require more comprehensive therapies and early lifestyle modifications to prevent their reflux from persisting and worsening.
There were some limitations in this study. In current study, we surveyed 198 participants (16.7%) via a telephone interview by our senior investigators or specialists, paying close attention in order to obtain accurate data. The response rate in this study was 47.8%, which was relatively higher, considering the initial sampling method and sample size compared with previous random sampling studies (initial samples of 699, 1290 and 337 participants, resurvey in 5, 7 and 11 years, with response rate of 34%, 79.2%, and 67%, respectively). 9-11 Different recall times reportedly might affect prevalence trends. 1,11 We set 3 months as the recall time in the current survey instead of 1 year in the initial survey, in light of the accuracy of selfreported symptoms. 29 There was a theoretical possibility of missing some respondents who had developed reflux symptoms that subsequently disappeared during those 9 months. In fact, the GER disappearance rate was low (64.6 per 1000 person-years) in this cohort population, and 13 participants developed new-onset GER with a GER duration of <1 year. We presumed that changes in the recall time did not alter the reliability of the study results. We did not collect data on anti-reflux therapy in the current survey. It remains unclear whether subjects with GER were prescribed medications or took ondemand therapy for their reflux symptoms 4,39,40 or medication for coexisting diseases, which might affect the natural history of GER in this general population, especially in participants with initial GER. In the initial study, <2% participants with and without GER were randomly selected for undergoing a gastroscopy examination, 14 while in the current survey we performed a routine gastroscopy only for those who had indications.
In conclusion, we prospectively resurveyed 1189 participants from a stratified randomized sample from the general population in the Beijing area and found that the prevalence of at least weekly GER and at least monthly reflux symptoms was stable over a 12-year period. The initial profiles of GER symptoms determined the natural history of the disease, and participants with monthly reflux were more likely than those with no reflux to develop GER over time. Comprehensive therapies and lifestyle modifications are recommended, and these measures should be taken as early as possible for subjects with reflux in the general population to ameliorate their GER and prevent reflux persistence and aggravation.