Prevalence of Gastroesophageal Reflux Disease and Proton Pump Inhibitor-Refractory Symptoms

In all, 1.3 million individuals were invited to complete the National GI Survey with the opportunity to participate in the study up until at least 70,000 surveys were completed. Ultimately, 124,674 (9.4%) individuals accessed the survey, of whom 71,812 (57.6%) completed the questionnaires and were included in the study. Table 1 lists the demographic information of the study cohort. Among 71,812 participants, we found that 32,878 (PW 44.1%) had ever experienced GERD symptoms in the past and 23,039 (PW 30.9%) reported being symptomatic in the past week. Table 2 summarizes the predictors of having had GERD symptoms in the past week. When compared to individuals aged 18–29 years, those 30–59 years of age had increased odds for reporting recent esophageal symptoms; no significant difference was seen for the ≥60 year old group. Females and those who identified as non-Hispanic white were more likely to have had recent GERD symptoms. Increasing levels of education as well as non-single marital status were associated with significantly higher odds for having experienced GERD symptoms in the past week. Individuals with income levels from $50,001-$100,000 were more likely to report recent GERD symptoms as compared to those with income levels ≤$50,000; conversely, those making ≥$200,001 were less likely to have such symptoms. Moreover, those with specific comorbidities, including IBS, Crohn’s disease, diabetes, endometriosis, gallstones, peptic ulcer disease, and thyroid disease were also more likely to experience GERD symptoms within the past 7 days. We found that 13,881 (PW 18.0%) out of 71,812 individuals met the modified Montreal definition for troublesome GERD symptoms (heartburn or regurgitation for ≥2 days in the past week): heartburn only, 6,751 (PW 47.5%); both heartburn and regurgitation, 5,426 (PW 39.8%); regurgitation only, 1,704 (PW 12.6%). Table 2 presents findings from the regression on having Montreal-defined GERD symptoms. Similar to the regression on GERD symptoms in the past 7 days, female gender, non-single marital status, and total household income level from $50,001-$100,000 were associated with increased odds for having Montreal-defined GERD symptoms. However, unlike the prior analysis, only those 30–49 years of age had increased odds for troublesome GERD symptoms vs. those who were 18–29 years old; no difference was seen for the 50–59 year old group. Moreover, we also found that participants who were ≥60 years of age had decreased odds for Montreal-defined GERD symptoms when compared to 18–29 year olds. As for race/ethnicity, non-Hispanic blacks and Asians remained at lower odds for having bothersome symptoms vs. non-Hispanic whites, while no differences were seen for the Latino and other racial/ethnic groups. Education level, while predictive of GERD symptoms in the past week, was largely not associated with Montreal-defined GERD. With respect to specific comorbidities, those that were predictive of GERD in the past week remained positively associated with Montreal-defined GERD symptoms, with cirrhosis being a new addition. We show findings from the regression on GERD PROMIS percentile scores among individuals who were symptomatic in the past week in Table 3 . Latinos and Asians had significantly higher GERD PROMIS scores vs. non-Hispanic whites. Those who were non-single and had IBS, celiac disease, cirrhosis, Crohn’s disease, diabetes, endometriosis, and thyroid disease also had more severe symptoms. Moreover, individuals who reported current PPI, histamine-2 receptor blocker, and antacid use also had worse symptoms. Conversely, increasing age, male gender, and higher education levels were associated with significantly lower GERD PROMIS scores. Among the 32,878 individuals who reported ever experiencing presumptive GERD, we had data on medication use from 29,274 respondents. Among the 29,274, we found that 9,234 individuals (PW 35.1%) were currently taking a medicine to manage their symptoms. Those on therapy reported taking the following: PPI, 4,935 (PW 55.2%); histamine-2 receptor blocker, 2,286 (PW 24.3%); antacids, 2,370 (PW 24.4%); other, 217 (PW 2.6%). Table 4 lists the frequency of use of each medicine class. Most of those on a PPI reported taking it daily (PW 68.1%) whereas more intermittent use was noted among those using histamine-2 receptor blockers and antacids. Among those taking a daily PPI (n=3,229), 1,858 (PW 54.1%) noted persistent, troublesome GERD symptoms (heartburn or regurgitation for ≥2 days in the past week). Symptomatic individuals reported the following symptoms even while on a PPI: both heartburn and regurgitation, 877 (PW 48.0%); heartburn only, 792 (PW 42.3%); regurgitation only, 189 (PW 9.7%). Table 5 presents predictors of PPI-refractory GERD symptoms. Individuals who were younger, female, Latino, divorced, separated, widowed, or married, and had IBS and Crohn’s disease had higher odds for remaining symptomatic while on a PPI. Conversely, those with GI cancer and ulcerative colitis were less likely to have PPI-refractory symptoms. No significant associations were largely seen between persistent symptoms and concomitant use of histamine-2 receptor blockers and antacids.

In October 2015 our group conducted the “National Gastrointestinal (GI) Survey,” a population-based audit of GI symptoms in over 71,000 community-dwelling Americans. 12 – 15 The survey was administered via MyGiHealth , a mobile app that utilizes AEGIS (Automated Evaluation of GI Symptoms), an automated algorithm that has previously been described in detail. 16 AEGIS asked users to “Select any symptom(s) you experienced in the past week” and “Please check any of these GI symptom(s) that you have EVER experienced.” Answer options included the following eight symptoms as well as a “none of these” option: heartburn, acid reflux, or gastroesophageal reflux; abdominal pain; bloating/gas; constipation; diarrhea; disrupted swallowing; fecal incontinence; nausea and vomiting. We chose these symptoms based on the National Institutes of Health (NIH) Patient-Reported Outcomes Measurement Information System (PROMIS) framework. 17 – 19 For each reported GI symptom, AEGIS guided respondents through corresponding GI PROMIS questionnaires to measure severity 20 ; the GERD PROMIS item bank and other questions related to “heartburn, acid reflux, or gastroesophageal reflux” are presented in the Supplementary File . Participants were also presented questions regarding demographics, socioeconomic status, and medical comorbidities. We aimed to recruit a representative sample of Americans for the National GI Survey by enacting quotas for age, sex, and region of country (Northeast, South, Midwest, and West). We partnered with Cint, a survey research firm that utilizes a reward system to incentivize respondents to participate in surveys. Potential respondents were sent an email through Cint research panels inviting them to complete an online survey. Along with the link to the survey, the email also included the following templated text, which was subject to editing from individual research panels: “Based on the information stored in your [research panel] profile, we believe we have a survey that you will qualify & earn from. The survey takes approximately 15 minutes and if you successfully complete it, your account will be credited with [incentive].” Cint’s reward symptom is based on the length of the interview and requires certain thresholds to be met before panelists can redeem rewards. This structure is meant to encourage long-term participation and discourage professional respondents who seek to take surveys only for financial gain. Participant recruitment for the National GI Survey occurred from October 14, 2015 to November 4, 2015. Survey initiations were distributed by Cint until we reached our sample size goal of approximately 70,000 respondents, allowing us to create a dataset with robust explanatory power and for examining the prevalence and predictors of both common as well as less common GI symptoms. Users who clicked the survey link in the invitation were brought to a home page asking them to complete a “GI Survey”; no specific mentions of GERD were made on the initial screen. All individuals ≥18 years of age were included in the study. Our primary outcome was prevalence of having had GERD symptoms (heartburn, acid reflux, or gastroesophageal reflux) in the past, categorized by having ever experienced GERD symptoms, GERD symptoms within the past 7 days, and troublesome GERD symptoms as determined by a modified Montreal definition (heartburn or regurgitation occurring ≥2 days in the last week). 1 To determine whether respondents met the Montreal definition, we leveraged 2 items from PROMIS ( Supplementary File ): heartburn: In the past 7 days, how often did you feel burning in the red area shown in the picture (behind the breastbone)?; and regurgitation: In the past 7 days, how often did you have regurgitation – that is food or liquid coming back into your throat or mouth without vomiting? Answer options for both questions included never, one day, 2–6 days, once a day, or more than once a day. Of note, while the Global Consensus Group states that in population-based studies troublesome GERD symptoms can be determined by the presence of mild symptoms on ≥2 days a week or moderate/severe symptoms occurring ≥1 day a week, our survey only assessed for the former, hence our use of a modified Montreal definition. As a secondary outcome, we assessed for GERD symptom severity as determined by PROMIS in those who reported GERD in the past week. 13 , 19 , 20 Another secondary outcome was prevalence of PPI-refractory GERD symptoms, defined as heartburn or regurgitation for ≥2 days in the past week among those currently taking a daily PPI. We also examined participants’ medication use, demographics, and past medical history. Respondents were asked which medicines they were currently taking for their GERD symptoms and frequency of use ( Supplementary File ): PPI (dexlansoprazole, esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole); histamine-2 receptor blocker (cimetidine, famotidine, ranitidine); antacids (e.g., Tums, Rolaids, Mylanta, Maalox); other medicine. Demographic information elicited via the survey included age, gender, race/ethnicity, education, marital status, employment status, and income level. They were also asked to identify comorbid conditions that had been “diagnosed by a doctor and can affect the gastrointestinal system,” including irritable bowel syndrome (IBS), chronic idiopathic constipation, cancer of the GI tract, celiac disease, cirrhosis, Crohn’s disease, ulcerative colitis, diabetes, endometriosis, gallstones, HIV/AIDS, pancreatitis, peptic ulcer disease, and thyroid disease. All statistical analyses were performed in Stata 13.1 (StataCorp LP, College Station, TX). We used data from the 2010 US Census (age, sex) and 2015 US Census Bureau’s American Community Survey (race/ethnicity) to create population weights and applied them to the sample data to produce population estimates 10 , 21 ; the actual weights used in the analyses were previously described elsewhere. 13 , 14 This was done to adjust for over- and under-sampling of subgroups in the National GI Survey, thereby decreasing bias due to nonresponse and underrepresented groups in the population. A two-tailed p-value of less than .05 was considered statistically significant. We performed population-weighted (PW) multivariable regression models to adjust for potentially confounding factors and to calculate adjusted p-values, odds ratios (OR), and 95% confidence intervals (CI). These regression models were performed on our primary and secondary outcomes, adjusted by relevant demographic, socioeconomic medication, and comorbidity variables described above. We used logistic and linear multivariable regression models for binary and continuous outcomes, respectively. This study was approved by the Cedars-Sinai Institutional Review Board (Pro54744).

In this population-based survey, we found that GERD symptoms are very common in the community. More than 2 of 5 Americans have experienced heartburn or regurgitation in the past while nearly 1 of 3 experienced these symptoms in the last week. Additionally, among those managing their symptoms with a daily PPI, we found that more than half still have persistent, troublesome GERD symptoms. Our prevalence of presumptive GERD is largely in line with prior estimates from other US cross-sectional and population-based studies. Namely, the prevalence of weekly GERD symptoms from past studies conducted in the 1990s and early 2000s ranged from 18% to 28% with a sample size-weighted mean of 20%, 7 while we found that 31% of respondents in our study reported GERD symptoms in the past week. Although it is difficult to make direct comparisons in the prevalence rates given the varying definitions of GERD and different populations, 7 our data suggests that the prevalence of GERD symptoms may be increasing. 22 This increasing burden is likely related in large part to the obesity epidemic. At the time of the National GI Survey in 2015, approximately 40% of Americans were obese (as compared to 30% in 1999), 23 , 24 and obesity has been shown to increase the odds of GERD up to 3-fold. 25 – 27 This is problematic as GERD leads to decrements in quality of life, mental health, and social function. 2 – 4 It is also associated with significant healthcare utilization, as GERD is the second leading physician diagnosis among the GI disorders with more than 5.5 million office and emergency room visits in 2014. 5 Aside from examining the prevalence of GERD symptoms in the past week, we also determined how many individuals in the community have such symptoms as determined by a modified Montreal definition 1 ; this allows for a more precise, criterion-definition of GERD. Here, we found that 18% of individuals reported either heartburn or regurgitation at least 2 days out of the week, which the Global Consensus Group considers troublesome. The prevalence of GERD symptoms using the modified Montreal definition is much higher in our study than a previous US population study, which found that 6% and 3% of respondents experienced heartburn and regurgitation, respectively, at least twice per week. 9 To our knowledge, the only other studies that employed the Montreal definition examined non-US populations, with prevalence rates of 3% and 16% in China 28 and Japan, 29 respectively. These rates are lower than that noted in our study, and the true difference is likely even more pronounced, as we used a modified Montreal definition that did not include moderate-to-severe symptoms of heartburn or regurgitation occurring ≥1 day a week; our study was only able to assess for those who had either symptom to any degree on ≥2 days a week. In our study, we found a number of predictors of GERD symptoms. For instance, men were less likely to have had both GERD symptoms in the last week and Montreal-defined GERD when compared to women. Prior data in the literature has been equivocal on this point, as some studies indicate a higher prevalence of GERD in men, others in women, and others demonstrate no difference at all. 30 – 32 Similarly, while some studies have demonstrated an association of GERD with increasing age, other studies have not. 30 In our present study, we found that increasing age is associated with increased odds for GERD symptoms up to a point, after which the risk decreases. Namely, when compared to 18–29 year olds, those 30–49 years of age have a higher prevalence of troublesome, Montreal-defined GERD symptoms, while those aged ≥60 years have a lower prevalence. As for race/ethnicity, there are also mixed findings. One study showed that blacks may experience more heartburn and Asians less heartburn as compared to whites, 33 while another study demonstrated no difference in GERD prevalence among blacks and whites. 8 Although our study confirmed that Asians have a lower prevalence of GERD symptoms, our finding that GERD is also less common among non-Hispanic blacks vs. non-Hispanic whites is counter to prior findings. 8 , 33 Further research examining the etiologies behind disparities in GERD symptoms with age, gender, and race/ethnicity are needed. Prior studies have demonstrated a higher prevalence of GERD among those with IBS 34 – 36 and diabetes, 37 – 39 which was confirmed with our study. We also found that those with IBS and diabetes have more severe GERD symptoms as measured by NIH PROMIS vs. those without the disorders. Moreover, we noted a higher prevalence of GERD symptoms among those with other comorbidities, most of which have not been previously reported or rigorously studied: Crohn’s disease, endometriosis, and thyroid disease. Individuals with these diagnoses also have higher GERD PROMIS scores. Aside from determining the prevalence and predictors of GERD symptoms, we also systematically assessed medication use. For those managing their condition with daily PPIs, we noted that 54% still have persistent GERD symptoms, which is comparable to previous observational estimates in primary care and community-based settings (45%, range 30% to 60%). 6 With respect to independent predictors of PPI-refractory GERD symptoms, we found that Latinos are 2.44-times more likely to have persistent symptoms while on PPIs when compared to non-Hispanic whites. The reason behind this finding is unclear but may be secondary to physiologic or even cultural etiologies. Women 6 and those with IBS 40 have been previously noted to be more likely to have PPI-refractory GERD, which our study confirmed. Associations have also been found between PPI-refractory symptoms and stress, 41 anxiety, somatization, and functional GI disorders. 42 , 43 The increased prevalence of functional and psychosomatic disorders among women may account in part for their increased odds of persistent GERD symptoms while on PPIs as seen in our study. 43 In regards to age, a Japanese study found that older age was more associated with PPI-resistance, 44 which is contrary to our findings. The higher prevalence of PPI-refractory GERD symptoms among younger individuals in our population may again be explained by the higher prevalence of somatization and functional disorders among this group, 43 where PPIs are unlikely to improve esophageal symptoms. While the degree to which our findings reflect inadequate acid suppression from PPIs vs. representing a surrogate for true underlying functional disease or non-acid reflux remains unclear, our results nonetheless demonstrate the need for further research and development of novel therapies for those with PPI-refractory GERD symptoms. Preliminary data has shown benefit with the potassium competitive acid blocker vonaprazan 45 , 46 as well as gastric-retained bile acid sequestrants. 47 , 48 This study has strengths and limitations. The National GI Survey is among the largest US population-based studies focused on GI symptoms. We have information on more than 71,000 participants, with more than 32,000 individuals who reported ever experiencing GERD symptoms and nearly 1,900 individuals who reported PPI-refractory GERD symptoms in the last week. Another strength is our use of a novel online digital health tool that employs validated NIH PROMIS item banks and GERD-specific questions to systematically gather comprehensive information from respondents. 16 , 19 , 49 This digital platform also allowed us to efficiently recruit a large, highly diverse, representative population in a very short period of time. However, the online collection of data can also be considered a limitation. As the data were based on individuals’ responses to an online survey, there are concerns in regard to generalizability, particularly among middle aged and elderly individuals with poor access to the Internet and/or limited computer skills. Our study may also have selected for older individuals who were more functional and independent. Of note, though, the Pew Research Institute reports that 82% and 63% of those 50–64 and ≥65 years old, respectively, used the Internet in 2015. 50 Nevertheless, our survey may have underestimated the prevalence of GERD and PPI-refractory GERD symptoms among older individuals. Conversely, our description of the study as a “GI Survey” to potential respondents may have led to an overestimation of GERD symptom prevalence, as those without GI issues may have opted to not complete the survey. We attempted to minimize participation bias by incentivizing users through Cint’s reward system to fully complete the survey. Our results are also largely consistent with prior population-based studies, supporting the validity of our findings. We also would not have expected this to impact our regression analyses, as it is unlikely that there are systematic differences between survey responders and non-responders with GERD symptoms. Additionally, we were not able to confirm pathologic reflux with esophageal pH monitoring, but instead relied on cardinal symptoms reported by respondents. Our survey methodology may have therefore led to misclassification of which respondents truly did or did not have objective GERD; however, any discrepancies would be consistent with misclassifications made in clinical practice where a diagnosis of GERD is typically based on patient-reported symptoms without confirmatory, objective pH or impedance testing. Along the same lines, there may have been misclassification as our GI symptom screener question solely defined GERD as “heartburn, acid reflux, or gastroesophageal reflux” and did not specifically mention regurgitation. While some respondents may have equated “gastroesophageal reflux” with regurgitation, we nonetheless may have underestimated the prevalence of GERD, particularly for those who only experience regurgitation symptoms. Another limitation is that we did not assess whether respondents were taking PPIs correctly (i.e., 30–60 minutes before a meal), whether use of the PPI was guided by a physician or was self-administered (i.e., over-the-counter), or if they engaged in lifestyle modifications; this may have led to an overestimation of PPI-refractory symptoms. Finally, our study did not collect data on some demographic and lifestyle factors that can impact GERD symptom prevalence, such as body mass index, waist circumference, and alcohol and tobacco use; this will be addressed in our forthcoming National GI Survey 2. We also did not examine the impact of GERD symptoms on quality of life or healthcare utilization, as the primary goal of the National GI Survey was to assess the prevalence and distribution of the 8 cardinal GI symptoms. Further research updating our understanding of the burden imposed by GERD is warranted. In conclusion, in this large population-based survey of community-dwelling Americans, we found that GERD symptoms are very common, with 2 of 5 having ever had such symptoms in the past and 1 of 3 experiencing symptoms in the past week. We also found an uneven distribution of GERD symptoms, as women, non-Hispanic whites, and those with comorbidities such as IBS, Crohn’s disease, diabetes, and endometriosis, among others, are more likely to be symptomatic. Additionally, we noted that more than half of those on daily PPI therapy continue to experience persistent heartburn and/or regurgitation symptoms. Due to the significant impact of GERD on quality of life and its considerable economic burden, further research is needed to further explore these associations as well as guide the development of novel therapies for those with PPI-refractory GERD symptoms.

Gastroesophageal reflux disease (GERD) involves classic symptoms of heartburn and/or regurgitation. 1 It is a highly prevalent disease with significant economic impact and reduction in patient health-related quality of life. 2 – 5 Although there are a number of available effective prescription and over-the-counter therapies, 45% of patients on a proton pump inhibitor (PPI) experience persistent GERD symptoms despite treatment. 6 Previous estimates of the prevalence of weekly GERD symptoms in the US range from 18% to 28%. 7 These estimates, however, are based largely on two populations: residents of Olmstead County, Minnesota, and employees of the Houston Veterans Affairs (VA) Medical Center. 7 Neither group is representative of the current US demographics as Olmstead County is 90% Caucasian 7 and the Houston VA employee population is 43% African American. 8 Another US population-based study of 21,128 adults found that 22% and 16% of Americans experienced heartburn and regurgitation within the past month, respectively. 9 Of note, while this study was conducted nationally, the cohort was 82% non-Hispanic white; data from the US Census Bureau’s American Community Survey in 2017 shows that 61% of the population is non-Hispanic white. 10 As GERD prevalence varies with race/ethnicity, these prior studies may provide inaccurate estimates of the current prevalence of GERD symptoms in the US. 11 Given the significant impact of heartburn and regurgitation on health-related quality of life and healthcare utilization along with the evolving demographics of the US, it is important to understand the current burden and distribution of GERD symptoms in the US population. Moreover, the high prevalence of persistent GERD symptoms despite PPI therapy (referred to as ‘PPI-refractory GERD symptoms’ in this paper) also highlights the need for a better understanding of the predictors of the disease and response to therapies as we aim to reduce its overall burden and maximize benefits from future adjunctive, novel therapies. Therefore, the aims of this study were to determine the prevalence and predictors of GERD and PPI-refractory GERD symptoms in a large, representative sample of community-dwelling Americans.