The chief purposes of this study were to: (1) estimate the prevalences of self-reported symptoms consistent with GERD and dyspepsia among adult inhabitants of São Paulo; and (2) investigate potential patient characteristics associated with these symptoms.

The study protocol was reviewed by an institutional Ethics Committee of the Faculty of Public Health of the University of São Paulo (FPH/USP). For telephone interviews, interviewers explained that participation was voluntary and that there would be neither any penalty for refusing to participate (or withdrawing consent) nor compensation for participating. Telephone respondents provided informed oral consent, and these were recorded. All data were held confidential, and interviewees were advised that they could communicate with the principal investigator (M.R.D.O.L.) or the FPH/USP Ethics Committee.

The study targeted adult inhabitants of São Paulo who owned landline telephones. A probabilistic sample was conducted using a computer-assisted telephone interviewing survey system. Potential respondents were categorized as to gender and age groups. Data were stratified post-hoc into three homogeneous geographical regions of São Paulo according to the Social Exclusion Indices of the districts and postal codes.

Sample size: The following algebraic expressions were used to determine the desired sample size:

n = n₀•deff

n₀ = (1 - P)/[P•cv² (p)]

Where P denotes the prevalence of symptoms consistent with GERD or dyspepsia in the study population (estimated a priori at 20%); cv (p) signifies the maximum tolerated coefficient of variation when estimating the ratio in the sample (fixed at 17%); and deff is the design effect.

The deff value was fixed at 2. When contacting adults by landline, we used a raffle to determine which member of the household would be interviewed. The use of weights to compensate different probabilities of selecting each house member introduced a deff of 1.2. Further weights were introduced after stratification to minimize no-replies and noncoverage, leading to a deff of 1.6. Given a total deff of 2, we determined that 280 interviews were needed in each study domain. The smallest domain comprised men aged ≥ 50 years (10.8% of the adult population of São Paulo according to 2000 census data): 280/0.108 = 2592 (desired sample size rounds to 2600). Assuming a response rate of 85%, we needed to dial at least 3000 telephone numbers.

Further assuming that 30% of telephones would not be eligible, the total number of telephone numbers was conservatively estimated at 5000. To achieve the desired sample size, we used a systematic raffle of 10000 telephone listings in São Paulo, organized by area code and hence implicitly stratified by geographical region. Given the likelihood of nonresponses and ineligible telephones, we expected that 5000 telephone numbers would be reached for interviews after receiving a series of 25 first raffled replies. After discarding telephone numbers that did not correspond to homes or were inactive, we expected that the product of the proportions of telephones answered and of dwellers agreeing to participate in the survey would be 80%.

The questionnaire included four parts: (1) sociodemographic characteristics (e.g., weight, height, smoking) and comorbidities; (2) dietary habits; (3) presence of symptoms consistent with GERD or dyspepsia in the prior 3 mo, according to the Brazilian Consensus of GERD and Rome III definitions of each condition; and (4) use of medications and other therapies to manage symptoms. The survey required approximately 20 min for each interviewee to complete.

Interviewers were trained by a São Paulo field supervisor (M.R.D.O.L.) in November 2010. M.R.D.O.L. also served as a central authority to answer calls from interviewers in the field.

The interviewing software and survey instrument were developed and validated using two subsamples (50 individuals in each) of respondents. In a pilot study, the first 50 interviews were conducted initially by telephone, followed by face-to-face interviews in the household 2 to 7 d later. For the next 50 subjects, the sequence was reversed. The pilot study was conducted from November 23, 2010, through March 1, 2011. Validation of our survey data was based on the pilot-study findings of: (1) high agreement between interview modalities with regard to comorbidities (κ≥ 0.61; P < 0.001) and (2) no significant differences between mean age, body mass index (BMI), and the grade of evaluation of dietary quality (each P > 0.070) as reported by respondents across the two interview modalities. These items also showed high intraclass correlation coefficients (r_icc > 0.78; P < 0.001).

A two-step data-weighting procedure was performed. The first step adjusted for different probabilities of selection among respondents, and the second (post-stratification) weighting step adjusted for imbalances caused by potential nonresponse and noncoverage bias between the study sample and population of São Paulo.

In the first step, data from each respondent were weighted by the numbers of adults and of landline telephones in each household. The second step further accounted for biases due to nonresponse and incomplete landline coverage by comparing distributions of the study sample population to distributions for the city of São Paulo according to age, gender, and educational level (10 population strata; Table 1). Population characteristics for São Paulo were obtained from the National Survey on Households (Pesquisa Nacional por Amostra de Domicílios [PNAD] População). This survey was conducted in 2008 by the Instituto Brasileiro de Geografia e Estatística (IBGE; http://www.ibge.gov.br).

Proportions of females and better-educated subjects (i.e., those with ≥ 9 or ≥ 12 school years) were higher in the EpiGastro population. In order to correct for these sample imbalances, post-stratification weights were adjusted. To calculate population weights, composite distributions per gender, age, and educational attainment (number of school years completed) were obtained for the São Paulo and EpiGastro populations across 18 categories for each population (9 categories for each gender within each population; Table 2). By computing ratios between values in the two populations across the 18 categories of gender, age, and educational level, post-stratification weights were generated (Table 3).

Final weights were computed by multiplying the designed weight by the post-stratification weight. Notably, there was only slight variation between the final weights. Coefficients of variation for the final weights were < 1%, suggesting that introduction of the population weights did not affect the overall accuracy of the results.

Descriptive statistics were generated for the sample characteristics. Correlations between symptoms consistent with GERD or dyspepsia by respondent gender or age were assessed using χ² tests with a two-tailed α = 0.05 (a priori significance level of P < 0.05). Electronic data entry was supported by Delphi language (dbase). Statistical analyses were conducted using the Statistical Package for the Social Sciences Version 14.0 for Windows (SPSS 14.0).

Among 4570 households contacted, one adult from 3050 (66.7%) agreed to participate in the survey. A total of 1344 (29.4%) of all individuals screened refused to participate before receiving any explanation of the survey, and a further 176 (3.9%) did so after receiving an explanation, for a nonresponse rate of 33.3%.

Overall characteristics: The mean (SE) age of the entire study population was 42.6 (16.0) years, and the median was 41.0. Most subjects were women (53.1%), aged 18 through 49 years (66.7%), and had never smoked (60.5%; Table 4). Majorities of respondents were married or cohabitating (52.5%). A total of 26.5% of women were perimenopausal. Slightly more than one-half (57.1%) of the study sample had some high-school or college education, and nearly one-half were employed (Table 4).

Nutritional status and dietary habits: Nearly half of the study population reported that they had normal nutritional status (45.8%), while slightly more than half were either overweight (35.3%) or obese (16.3%).

Daily consumption of fruit (40.2%), green vegetables (43.3%), or legumes (35.2%) was each reported by more than one-third of respondents. One-third or more consumed fruit (34.4%), green vegetables (33.7%), or legumes (41.3%) 2-4 d each week. On the other hand, 5%-7% of respondents reported that they never consumed the above food groups. Most respondents reported that they drank coffee every day (56.2%), whereas 29.2% never consumed coffee.

GI conditions and comorbidities: The most frequently reported GI condition (by ≥ 2% of respondents) was gastritis (19.0%), followed by GERD (6.1%), hemorrhoids (4.8%), irritable-bowel syndrome (3.1%), and ulcer (2.0%). Other GI comorbidities included esophagitis (1.9%), colitis (0.6%), diverticulitis (0.6%), ulcerative colitis (0.4%), anorexia/bulimia (0.1%), and Crohn’s disease (0.1%).

The most frequent non-GI complaint was headache (19.3%). Signs and symptoms of the following conditions were also recorded: hypertension (19.1%), depression (12.9%), diabetes (6.7%), thyroid disorder (5.2%), heart disease (4.9%), and asthma (3.5%).

GERD: In all, 27.3% of respondents experienced a sensation of stomach burning: 20.8% at least once per month and 6.5% less frequently (Figure 1). Corresponding data for heartburn (or chest burning sensation), included 22.7% (total): 18.2% at least once monthly and 4.5% less frequently. A total of 17.3% of respondents reported regurgitation: 12.9% at least once monthly and 4.4% less frequently. Approximately 10% of subjects reported excessive salivation (i.e., sialorrrhea). Fewer than 5% of respondents reported difficulty swallowing (dysphagia; 4.2%) or painful swallowing (odynophagia; 2.4%).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Frequencies of symptoms consistent with gastroesophageal reflux disease. x/mo: Times per month.

Dyspepsia: Abdominal swelling or distension was reported by 26.8% of survey participants: 20.9% at least once, and 5.9% less than once, monthly (Figure 2). Corresponding data for bloating/sensation of fullness were 21.9%, 17.6%, and 4.3%, respectively. More than 20% of respondents reported satiety after consuming a small amount of food: 15.9% at least once monthly and 4.2% less than once monthly. Fewer than 20% of respondents reported abdominal pain or discomfort (13.0% at least once monthly and 4.4% less than once monthly) or nausea (11.5% and 4.1%, respectively). Fewer than 10% of respondents (9.4%) reported vomiting, including 6.0% at least once monthly.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Frequencies of symptoms consistent with dyspepsia. x/mo: Times per month.

GERD by gender: Compared to their male counterparts, significantly higher (> 1.5-fold) frequencies of women reported most symptoms consistent with GERD occurring at least twice per week, including heartburn (P = 0.047), a sensation of stomach burning (P = 0.012), regurgitation (P < 0.001), difficulty swallowing (P = 0.012), and/or painful swallowing (P = 0.009 vs men by χ² tests; Table 5).

GERD by age: No symptom of GERD correlated significantly with age groups (Table 6).

Most respondents (86.4%) reported that they used no medications that reduce GERD symptoms (Table 7). Among the minority of subjects who took medications, most used prescription medications (62.6% of those with “yes” responses about taking medications) or self-medicated (34.2%). The findings in Table 7 indicate that only about 14% of all respondents with available data used medications to manage GERD symptoms.

The vast majority of respondents (96.7%) took no other measures, including dietary modification or consumption of herbal teas, to manage GERD symptoms.

Dyspepsia by gender: Significantly higher (> 2-fold) frequencies of women (vs men) reported that symptoms compatible with dyspepsia occurred at least twice weekly (Table 8). These included impaired digestion (P < 0.001 vs men), a sensation of postprandial gastric fullness after consuming a small amount of food (P < 0.001), abdominal swelling or distension (P < 0.001), and abdominal pain (P < 0.001). Nausea and vomiting occurring at least twice weekly were also significantly more frequent among women: 9.0% compared to 2.7% of men (P < 0.001) for nausea and 3.0% compared to 1.4%, respectively, for vomiting (P = 0.040 each by χ² test).

Dyspepsia by age: No symptom consistent with dyspepsia correlated significantly with age groups (Table 9).

Most respondents used no medications that reduce dyspeptic symptoms (88.9%; Table 10). Of the minority (11.1%) who did, most reported that they received prescriptions from their physicians (57.4%) or self-medicated (40.8%). Most survey participants also reported that they did not use other, nonpharmacologic interventions (96.0%). Among the minority of respondents who did, 65.5% used herbal teas and 25.5% modified their diets.

Our a posteriori weighting and mixed-mode survey design may have helped to limit noncoverage and other forms of bias. By telephone company estimates performed for our survey, more than three-quarters of our target population owned landlines. A previous Brazilian study determined that a minimum landline coverage of 70% (in Southern and Central-West metropolitan areas) was necessary to avoid coverage bias[47]. One other potential bias is that most of our survey participants were women, who are more likely (vs men) to respond to health surveys in general, report physical symptoms, seek medical attention for constipation, and use laxatives to manage symptoms[48-50]. Women also seek medical consultation for dyspepsia more frequently than men in some societies[18,51,52].

The participation rate (67%) was acceptable. In one of the largest epidemiologic studies of upper GI disorders performed by telephone, the complete-survey nonresponse rate exceeded our rate of 33%[7]. Telephone surveys typically have higher nonresponse rates than household surveys[53]. Caller identification mechanisms tend to screen out younger and/or unmarried adults, members of ethno-racial minorities, and homes with young children[54]. Given its cross-sectional nature, our study also could not provide context as to the natural history of symptoms consistent with dyspepsia or GERD, or accurately account for waxing and waning symptoms over time. Development of intercurrent illnesses can influence patients’ GI symptom recall[5]. However, the survey reference interval of 3 mo used in our study was shorter than recall intervals employed in a number of other telephone surveys[12], potentially limiting recall bias.

On the other hand, it is somewhat concerning that only about 6% of our respondents reported physician diagnoses of GERD, yet more frequently reported symptoms compatible with such diagnoses. This disparity may reflect limited access to care (among members of lower-socioeconomic strata) and the fact that symptoms of dyspepsia and GERD often go unreported to, and/or uninvestigated by, physicians. An estimated 48% of cases of dyspepsia are uninvestigated in Brazil[55]. In other (US) populations, 75% of patients with FGIDs never consult with physicians[56].

In conclusion, approximately 21% of respondents from São Paulo reported highly frequent symptoms consistent with GERD (e.g., gastric burning sensation = 20.8%) or dyspepsia (e.g., abdominal swelling/distension = 20.9%). However, most respondents did not report that they had received diagnoses of these conditions, or medications to manage them, from their physicians. Women were significantly (approximately 1.5- to 2-fold) more likely to report symptoms of GERD and dyspepsia than men, but there was no significant association between advancing age and self-reported symptoms of these conditions. Further research is needed to assess the potential effects of these disorders on daily activities, diet, sleep, and worker productivity.

Symptoms of gastroesophageal reflux disease (GERD) and dyspepsia are distressing to many patients and often erode their quality of life. Ratios of prevalence to incidence of GERD are high and hence consistent with this condition’s chronic nature and potentially suboptimal management.

Limited data are available concerning the prevalences of self-reported symptoms consistent with GERD and dyspepsia in Brazil, particularly within the city of São Paulo.

Some telephone surveys have revealed that women more frequently experience symptoms of dyspepsia, whereas men more frequently experience symptoms of GERD. Overall, however, studies differ concerning associations between gender, age (and other sociodemographic characteristics) and prevalences of self-reported GERD and dyspepsia. Data are limited regarding self-reported symptoms of these conditions, and measures taken to manage them, among persons in South America.

By conducting telephone and face-to-face surveys of individuals in São Paulo concerning symptoms of GERD and dyspepsia, as well as measures taken to manage them, this study may help to further delineate the public-health dimensions and clinical challenges presented by these chronic conditions. Observations of substantial frequencies of self-reported symptoms of GERD and dyspepsia, coupled with low overall frequencies of medication use and/or lifestyle modification to manage these conditions, might suggest an unmet need in this urban Brazilian population.

GERD is an organic disorder characterized by symptoms of gastric-acid reflux, whereas dyspepsia is a “functional,” upper gastrointestinal condition characterized by abdominal distension and other forms of discomfort.

This manuscript “Epidemiology of upper gastrointestinal symptoms in Brazil (EpiGastro): A population-based study according to sex and age group” is a very interesting article.