Dietary guideline adherence for gastroesophageal reflux disease

We performed an analysis of a case–control study of Barrett’s esophagus; the study recruited members of the Kaiser Permanente, Northern California (KPNC) population, an integrated health services delivery organization, between 2002 and 2005. The KPNC membership contains approximately 3.3 million persons whose demographics closely approximate the underlying census population of Northern California [17, 18]. The study included Barrett’s esophagus patients as cases and two sex and age-matched control groups: GERD patients and population controls. The details of this original study were described previously in detail [19]. The present analysis contrasts the GERD and population control groups. The study and analyses were approved by the KPNC institutional review board and were performed in accordance with the Declaration of Helsinki. Informed consent was obtained.

GERD patients consisted of 317 eligible KPNC members who were selected from among all individuals with a physician-assigned GERD diagnosis (e.g. ICD-9 codes 530.11 reflux esophagitis and 530.81 gastroesophageal reflux) who also had a prescription for at least 90 days supply of anti-secretory medications and no diagnosis of Barrett’s esophagus (any visible columnar metaplasia) from an esophagogastroduodenoscopy close to the index date. Endoscopy and pathology reports were manually reviewed. The validity of identifying GERD patients using encounter and pharmacy databases was previously evaluated in a 100,000 member managed care organization [20]. All patients in the current study also ultimately received questionnaires regarding GERD symptoms and actual medication use.

Exposure to anti-secretory medications utilized a pharmacy database with information on prescription type, frequency of dispensing, and number of pills dispensed. We defined anti-secretory medications (e.g., histamine-2 receptor antagonists (H2 blocker) or proton pump inhibitors (PPI)) using data from the year prior to control selection (the reference dates).

Population controls were randomly selected from among members of the entire KPNC membership at the time the Barrett’s esophagus cases were identified. Individuals with Barrett's esophagus were excluded. For the current study, we also excluded population controls who, upon interview, reported heartburn or acid regurgitation once a month or more.

All study subjects completed: an in-person interview (most commonly at the subject’s home) of GERD symptoms, medication use, medical history, diet, tobacco use and alcohol use; phlebotomy; and anthropometric measurements.

We assessed dietary intake using a validated 110-item food frequency questionnaire (FFQ) [21‐24]. This questionnaire estimates average daily nutrient intake based on questions about frequency and portion size of a given food and usual eating habits for the year before the index date [25]. Dietary data were translated to daily intakes of specific nutrients using NutritionQuest (Berkeley, CA), and the raw data were used to estimate weekly consumption of particular food items. We excluded subjects with over 20 missing food items or with extremely high (>6000 kcal/day) or low (<400 kcal/day) total daily energy intakes.

We evaluated the associations between consumption of commonly discussed “reflux-triggering” dietary factors based on recommendations from the American College of Gastroenterology [17], National Institute of Health [18], and food items that have been studied in previous clinical research of GERD [26, 27]. Among those food items, data were available in our questionnaire for: coffee, tea, soft drinks, alcohol (total, wine, beer, and liquor), citrus, tomato products, total fat, fried foods, and large portion meals. “Soft drink” consumption was used as the proxy for “carbonated drinks” in the data; the questionnaire did not include diet sodas. Coffee and tea consumption did not distinguish between caffeinated or decaffeinated types, but excluded herbal tea. The food items and nutrients included total fat; fried foods (e.g., French fries, fried chicken, fish, and doughnuts); tomato products (e.g., fresh tomatoes, canned tomatoes, and tomato juice); and citrus (fruits and juice). The measurement units were mostly “servings per week”, except for fat (grams/day), coffee and tea (cups/day) and alcohol (drinks/day). The term ‘drink’ for alcohol was defined as: wine (7-8 oz), beer (12 ounces), liquor (one shot). Portion size reporting was facilitated by pictures depicting small (1/4 cup), medium (1/2 cup), large (1 cup), and extra-large (2 cups) servings. In this study, each person was scored on the proportion of foods that were reported as large or extra large. For instance, if a person indicated that half of what he/she consumed was in large or extra-large portions (1–2 cups), the person was considered to consume 50% of the meal in large portion. On the other hand, if a person always consumed small or medium portion meals (1/4-1/2cup), the person was categorized to consume 0% of the meal in large portion.

GERD symptoms were defined as either heartburn (a burning pain or discomfort behind the breastbone) or acid regurgitation (a bitter or sour-tasting fluid coming up into the throat or mouth) using a validated questionnaire [28]. Symptom severity was recorded as either mild (could be ignored), moderate (could not be ignored, but didn’t affect lifestyle), severe (could not be ignored and did affect lifestyle), or very severe (markedly affected lifestyle). Symptom frequency was coded as: less than once a month, less than once a week (but more than once a month), once or more a week, or daily. For the analysis of severity, we compared individuals who had moderate to severe symptoms (with at least monthly frequency) to asymptomatic population controls, and excluded individuals with mild severity (N = 75). For frequency analysis, we compared two categories (GERD symptoms greater than several times a week; more than once a month to once a week) to asymptomatic population controls.

We evaluated whether GERD was associated with the intake of certain beverages, nutrients, or food items using unconditional logistic regression to calculate odds ratios (ORs) as an estimate of the relative risk with GERD status as an independent variable and dietary factors as the dependent variable [29]. We evaluated the following additional variables as potential confounders: race/ethnicity (classified as white vs. non-white due to small sample sizes in the race/ethnic subgroups), smoking (ever vs. never and current vs. never), body mass index (BMI, kg/m²), medical center, recent alcohol use (number of drinks per week), aspirin or nonsteroidal anti-inflammatory drug (NSAID) use, a comorbidity index (the DxCg score) [30, 31], education, income, serum Helicobacter Pylori (H. Pylori) antibody status, and energy consumption (kcal/day). Confounders were included in the final model if their inclusion altered the odds ratio by >10% [29]. We evaluated for interaction by gender, race, long-term (2 + years) vitamin supplement use, NSAIDs use, H. pylori, or regular use (at least once a week) of GERD medication (antacid, H2 blocker, or PPI) using cross product terms in the logistic regression model and stratified analysis. All analyses were performed using STATA statistical software (College Station, TX).

Of the 316 GERD patients and 317 population controls from the original study, dietary data were available for 622 subjects (98% of interviewed subjects). Exclusion of six subjects with extreme values of caloric intake or ≥6 missing values on the food questionnaire and 126 population controls who reported acid regurgitation or heartburn more than once per month provided 308 GERD patients and 182 GERD-free population controls (Tables 1 and 2). The proportions of smokers and individuals who consumed large or extra-large portion sizes were similar between the groups.

Patients with moderate to severe GERD symptoms were over twice as likely as GERD-free controls to consume soft drinks, adjusting for smoking and education [OR = 2.11, 95% CI:1.16-3.84] (Table 3); there was no association with GERD frequency (Table 4).

Patients with moderate to severe GERD symptoms were also more likely to consume one or more cups of regular tea a day compared to asymptomatic controls [OR = 1.86; 95% CI:1.16-2.97 for severe GERD]. There was no association between GERD symptom severity or frequency and coffee consumption.

GERD patients with severe or frequent symptoms were less likely to consume beer, and there was also an inverse association between GERD symptom severity and general alcohol drinking, especially light drinking. For instance, patients with moderate to severe symptoms were almost half as likely to drink beer as those without GERD [OR = 0.54 95% CI 0.31-0.96]. Similarly, patients with frequent GERD symptoms (>1/week) were half as likely to drink beer [OR = 0.52 95% CI: 0.28-0.96]. There were non-significant, though positive association between GERD symptom severity and frequency and liquor drinking: individuals with severe GERD symptoms were 50% more likely to drink liquor compared to those without GERD [OR = 1.50 95% CI: 0.85-2.66].

Patients with severe or frequent GERD symptoms were less likely to consume citrus fruits compared to asymptomatic controls [OR = 0.62; 95% CI: 0.41-0.94; OR = 0.62; 95% CI: 0.40-0.98, respectively]. On the other hand, there was no association between the severity or frequency of GERD symptoms and consumption of tomato products, another acidic food which GERD patients are recommended to avoid (Tables 3 and 4).

Individuals with severe symptoms had higher total fat consumptions than those without GERD [OR = 1.77 95% CI: 1.07-2.93 for highest tertile; OR = 1.67 95% CI: 1.03-2.71 for second tertile]. Similar associations were observed for frequency of GERD symptoms and total fat. In addition, individuals with frequent symptoms were twice as likely to consume fried foods [OR = 2.10; 95% CI: 1.19-3.70 for most frequent; OR = 1.87 95% CI: 1.00-3.52 for monthly-weekly frequency]. There was no association between the severity or frequency of GERD symptoms and large portion sizes.