Abdominal obesity increases the risk of reflux esophagitis: a systematic review and meta-analysis

Abstract

Background and objectives

The association between abdominal obesity and reflux esophagitis (RE) has been extensively evaluated, but the current findings are mixed and more convincing epidemiological evidence urgently needs to be established. To thoroughly explore this relationship, we summarized the latest studies, performed an updated meta-analysis, and examined the dose–response relationship.

Methods

We performed a systematic search of PubMed, Web of Science, and Embase up to 28 March 2021, using prespecified terms to identify studies investigating the association between abdominal obesity and RE. Odds ratios (ORs) with 95% confidence intervals (CIs), mean differences (MDs) or standardized mean differences (SMDs) with 95% CIs were taken as effect-size estimates.

Results

Forty-two observational studies, including 11 cohort studies, were meta-analyzed. Overall, a statistically significant association was observed between abdominal obesity and RE, by both the pooled OR (adjusted OR = 1.51, 95% CI: 1.37–1.66, p < .001) and the pooled SMD (SMD = 0.36, 95% CI: 0.30–0.42, p < .001). Moreover, this significant relationship persisted with subgroup stratification. In subgroup analyses, we found that study design, abdominal obesity measurement, adjustment for covariates and sex were possible sources of between-study heterogeneity. For the dose–response analyses, the risk of RE increased with the degree of abdominal obesity, and the increasing trend accelerated when waist circumference (WC) reached 87.0 cm.

Conclusion

This meta-analysis indicated a significant association between abdominal obesity and RE, and the risk of RE increased with abdominal obesity especially when the WC was over 87.0 cm.

Keywords:

• Abdominal obesity
• waist circumference
• visceral fat
• reflux esophagitis
• meta-analysis
• dose–response analyses

Background

Gastroesophageal reflux disease (GERD) is a highly prevalent gastrointestinal disease affecting 8–33% of the global population depending on the geographical region [1,2]. It causes economic losses of up to billions of dollars worldwide each year and consumes many medical resources [3]. Reflux esophagitis (RE) is an essential branch of GERD. Recent studies suggest that changes in the trans-diaphragmatic pressure gradient, motility disturbance of the esophagus, and the existence of esophageal hiatal hernia may destroy the balance of protective mechanisms and aggressive chemical substances, further resulting in endoscopic esophageal mucosal erosion and ulceration [4–6]. RE symptoms are associated with reduced health-related quality of life [1]. More importantly, chronic RE may be closely related to the occurrence and development of Barrett's esophagus and even esophageal adenocarcinoma [7,8].

Previous studies have demonstrated that obesity is a high-risk lifestyle factor associated with RE [9], and recent research indicates that the regional distribution of adipose tissue, especially abdominal obesity, is considered a more important pathogenic factor than total body obesity [10]. Accumulated abdominal fat increases the occurrence of reflux by increasing intra-abdominal pressure [11]. Over the past two decades, research has gradually intensified, although the results have been mixed [12–16]. Interestingly, in 2013, Siddharth and colleagues pooled the results of 17 independent studies that did not restrict the analysis based on study design and observed a significant association between abdominal obesity and RE (OR: 1.87, 95% CI: 1.51–2.31) [12]. However, the number of included studies was insufficient. In recent years, literature exploring the relationship between abdominal obesity and RE has flourished, including several high-quality cohort studies, which can further expand the sample size. Additionally, the dose–response analyses in the original research need to be improved, as they cannot adequately reflect the association between abdominal obesity and RE.

To provide more convincing epidemiological evidence for further studies, the present meta-analysis expanded the scope of the included studies to comprehensively assess the relationship between abdominal obesity and RE and further conducted a dose–response analysis between them. Meanwhile, we aimed to eliminate the influence of covariates on pooled ORs and explored possible sources of heterogeneity.

Method

Search strategy

The PubMed, Web of Science, and Embase databases were systematically searched for articles up to 28 March 2021. The combination of medical subject terms used included: (adiposity* OR abdominal obesity OR abdominal fat OR anthropomet* OR BMI OR body composition OR body fat distribution OR body fat patterning OR body mass index OR body size OR central obesity OR girth circumference OR intra-abdominal fat OR retroperitoneal fat OR visceral fat OR visceral obesity OR waist circumference OR waist–hip ratio OR visceral adiposity OR visceral adipose tissue) [All Fields] AND (Erosive esophagiti OR Esophag*, Reflux OR Esophag*, Peptic OR Gastroesophageal Reflux OR gastrooesophageal reflux OR Gastric Acid Reflux OR GERD OR Peptic Esophag*) [Title/Abstract]. Additionally, a manual search of the reference lists of major articles was performed to avoid potential omissions.

The search process was performed independently by two investigators (J. Z. and M. Y.), and the articles were not restricted by language or publication. If necessary, we contacted authors to try to obtain more relevant information. Reporting of the study conforms to the PRISMA statement [17], which is listed in Supplementary Table 1. This meta-analysis was based on published studies, and no prior ethics committee approval or informed consent was required.

Literature search

The present meta-analysis included studies that met criteria [1,2], and [3] with either [4] or [5] of the following criteria: (1) observational studies, including cross-sectional, case–control, and cohort studies, assessing of the association between abdominal obesity and RE; (2) abdominal obesity defined using visceral fat measured by abdominal computed tomography (CT), the waist–hip ratio (WHR), or waist circumference (WC); (3) RE diagnosed using an upper gastrointestinal endoscope; (4) the use of odds ratio (OR) or hazard ratio (HR) with a 95% confidence interval (95% CI) to measure the magnitude of the association between abdominal obesity and RE; and (5) continuous data sufficient to calculate the MD or SMD with the 95% CI.

Studies were excluded if they were published in the form of case reports or case series, editorials, and narrative reviews.

Data abstraction and quality assessment

Data extraction was performed using predefined forms, and the details were as follows: (1) characteristic information: first author, country, baseline age, gender, study publication time, study design, population type, and sample size. (2) Exposure assessment: measurement of abdominal obesity (CT, WHR, or WC) and the reference standard for abdominal obesity. (3) Information for calculation: adjusted OR or HR with the 95% CI or the sample size, mean, and standard deviation of continuous data. In addition, we recorded adjustments for confounding factors in each study in detail, such as body mass index (BMI), age, sex, smoking and drinking, and the presence of an esophageal hiatus hernia, for the next subgroup analysis.

Data were independently abstracted from each qualified article by two investigators (J. Z. and M. Y.). Disagreements in data abstraction were resolved by consensus and, if necessary, by a third author (Y. Z.).

The quality of cohort studies and case–control studies was assessed using the Newcastle–Ottawa Scale (NOS) tool [18]. The NOS mainly evaluates a study based on the following aspects: the selection of the study groups, the comparability of the study groups, and the assessment of exposure and outcome. The quality of cross-sectional studies was assessed using the criteria recommended by the Agency for Healthcare Research and Quality (AHRQ) [19]. The criteria consisted of 11 items, and the answers to each item were ‘yes’, ‘no’, or ‘unknown’.

Exposure and outcome assessment

When multiple indicators for abdominal obesity measurement were reported in the same study, we prioritized CT-measured visceral fat (visceral fat area (VFA cm²), visceral adipose tissue (VAT cm³), or visceral fat area to subcutaneous fat area ratio (V/S)), followed by WHR and lastly WC. Since the international community does not have a unified standard for the definition of abdominal obesity, the definition provided by the study or the lowest reference category of categorical variables was regarded as the benchmark when extracting data. We included more than one estimate from the studies (e.g., if a study reported an OR for persons with a WC 90–99.9 cm and an OR for persons with a WC ≥ 100 cm, both ORs were included in the summary estimate as WC ≥ 90 cm). The outcome measurement was defined as RE diagnosed according to endoscopy.

Statistical analysis

Data management and analysis were conducted using STATA 14.1 (StataCorp, College Station, TX). Effect-size estimates were expressed as either the OR or HR (for classified variables) or the MD or SMD (for continuous variables), with 95% CI for the RE risk. We converted HRs into RRs [20,21] to provide approximate OR values, which were combined with the reported ORs. Summary OR estimates were calculated based on the assumption of random effects; even in the absence of significant heterogeneity, all differences between studies were considered.

The generalized least squares regression proposed by Greenland and Longnecker was used to examine the dose–response relationship between abdominal obesity and RE [22]. The restricted cubic splines of exposure distributions with three knots (25th, 50th, and 75th percentiles) were used to test the nonlinear relationship.

The inconsistency index (I²) was used to quantify the heterogeneity between various studies. AnI² > 50% indicates significant heterogeneity, and the degree of heterogeneity increases with an increasing percentage. To explore potential sources of heterogeneity among the included studies, a considerable number of prespecified subgroup analyses were conducted based on population characteristics, research characteristics and whether covariates were adjusted. Sensitivity analysis was used to evaluate the impact of any single article on the stability of the overall results. In addition, publication bias was evaluated by Begg's funnel plots and Egger's regression asymmetry test, while the number of theoretically missing studies was estimated by the trim-and-fill method.

Result

Study selection

Among the initially identified 3622 unique publications searched by prespecified search strategies, 42 relevant studies (37 independent populations with more than 400,000 participants) met the inclusion criteria of the present meta-analysis. For studies with overlapping participants, different analysis methods were used to combine effect sizes to reduce the impact of autocorrelation. The detailed search strategy and inclusion and exclusion process are shown in Figure 1.

Figure 1. Flow chart of study selection in this meta-analysis.

Study characteristics and quality assessment

Among the 42 eligible studies, 22 were combined to calculate OR as effect-size estimates [14,16,23–42], and 20 were combined to calculate SMD [15,43–61]. The baseline characteristics of the eligible studies included in each outcome are recorded in Tables 1 and 2. Most of the included studies were carried out in Asia (11 in China [26,28,30,31,35,38,51,54,55,57,60], 6 in Japan [16,33,34,43,53,58], and 21 in Korea [14,24,25,27,28,32,36,37,39–42,44–46,48–50,52,59,61]), 2 studies were performed in Europe [15,56], 1 study was conducted in South America [23], and 1 study was carried out in Africa [47]. Twelve articles reported results upon stratification by sex [14,16,24,29,33,35,36,44,49,51,53,58], and 6 articles provided results for patients [16,23,27,33,43,60].

Table 1(A). The baseline characteristics of studies presenting adjusted effect size estimates with 95% CI.

First author	Year	Design	Exposure assessment	Reference	Basic diseases	Adjusted variables
Kang, M. S.	2007	C-S	WC	WC: 90 (for men), 80 (for women)	No	Age, sex, BMI, smoking, lifestyle choices, hiatal hernia, diabetes, use of aspirin or NSAIDs, use of hypertensive drugs
Lee, H. L.	2007	C-C	WHR	WHR: 0.8	No	Sex, BMI, percentage of body fat, obesity degree, smoking, hiatal hernia
Park, J. H.	2008	C-C	WC	WC: 90 (for men), 80 (for women)	No	Sex, smoking, drinking, reflux symptoms, hiatal hernia, fatty liver, homeostasis model
Chung, S. J.	2008	C-C	WC, CT	WC: 90 (for men), 80 (for women)	No	BMI, smoking, drinking, VAT, SAT
				VFA: 104.6 (for men), 52.1 (for women)
Koo, J. S.	2009	Cohort	WC	WC: 80	No	Sex, BMI, increase of BMI, smoking, drinking, triglyceride
Nam, S. Y.	2009	Cohort	WC	WC: 80	No	Age, smoking, drinking, gastrointestinal symptoms, diet, hiatal hernia, duodenal ulcer
Chua, C. S.	2009	C-C	WC	WC: 90 (for men), 80 (for women)	No	BMI
Nam, S. Y.	2010	C-S	WC, WHR, CT	WC: 80	No	Age, sex, BMI, VAT, WC, education, smoking, drinking, hiatal hernia, atrophic gastritis, Helicobacter pylori, diabetes, medication for hypertension
				WHR: 0.8
				VAT: 500
Wu, P.	2011	C-C	WC, WHR	WC: 90 (for men), 80 (for women)	No	Age, sex
				WHR: 0.9 (for men), 0.8 (for women)
Chiba, H.	2012	C-S	WC	WC: 85	No	Age, sex, BMI, WC, smoking, drinking, hiatal hernia, blood pressure, Helicobacter pylori, blood glucose, triglyceride, HDL-C, LDL-C
Sogabe, M.	2012	Cohort	WC	WC: 90	MS	Age, BMI, WC, hiatal hernia, MS type, fatty liver
Jung, J. G.	2013	C-S	WC	WC: 90 (for men), 80 (for women)	No	Sex, WC, smoking, drinking, diet, Helicobacter pylori IgG antibody
Loke, S. S.	2013	C-C	WC	WC: 90 (for men), 80 (for women)	No	NA
Hung, W. C.	2014	C-S	WC	WC: 90 (for men), 80 (for women)	No	Age, sex, BMI, smoking, drinking, tea drinking, exercise, hiatal hernia, diabetes, hypertension, hypertriglyceridemia, HDL-C
Sogabe, M.	2014	C-S	WC	WC: 80	MS	Age, BMI, WC, drinking, hiatal hernia, Helicobacter pylori, HDL, LDL, glycated hemoglobin
Ze, E. Y.	2017	C-S	CT	V/S: 1.165	No	BMI
Lee, S. W.	2017	C-C	WC (self-reported)	WC: 90 (for men), 80 (for women)	No	NA
Baeg, M. K.	2018	C-S	WC	WC: 90 (for men), 80 (for women)	MS	Age, sex, smoking, drinking, coffee, Helicobacter pylori, white blood cell count
Mazzoleni, F.	2019	Cohort	WC	WC: 94 (for men), 80 (for women)	Dyspepsia	Age, sex
Kim, T. J.	2019	Cohort	WC	WC: 90 (for men), 85 (for women)	No	Age, sex, smoking, drinking, exercise, acid inhibitor use, blood glucose, blood pressure, triglyceride, HDL-C, homeostasis model, C-reactive protein
Yang, H. J.	2019	Cohort	WC	WC: 56.5–80 (for men), 74.1–79.5 (for women)	No	Age, BMI, education, smoking, drinking, exercise, year of screening exam, hypertension, diabetes, medication for hypertension, medication for diabetes, medication for dyslipidemia
Hsieh, Y. H.	2019	C-S	WC	WC: 90 (for men), 80 (for women)	No	Age, sex, smoking, drinking, hiatal hernia, blood pressure, hyperglycemia

RE: reflux esophagitis; C-C: case–control study; C-S: cross-sectional study; CT: computed tomography; VFA: visceral fat area (cm²); VAT (cm3): visceral adipose tissue; SAT: superficial adipose tissue; V/S: VAT to SAT ratio; WHR: waist-to-hip ratio; WC: waist circumference (cm); BMI: body mass index; MS: metabolic syndrome; NA: not applicable; HDL-C: high density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol.

Table 1(B). The baseline characteristics of studies presenting adjusted effect size estimates with 95% CI.

First author	Year	Country	Baseline age	Gender	Effect estimate	Sample size	Cases	Effect size	Low limit	High limit
Kang, M. S.	2007	Korea	43.9	All	OR	2442	161	2.36	1.44	3.86
Lee, H. L.	2007	Korea	45.4	All	OR	3188	292	2.316	1.282	4.186
Lee, H. L.	2007	Korea	45.4	All	OR	3188	292	4.055	1.622	10.14
Park, J. H.	2008	Korea	45.0	All	OR	5037	1679	1.33	1.15	1.54
Chung, S. J.	2008	Korea	48.3	All	OR	1500	750	1.6	1.03	2.48
Chung, S. J.	2008	Korea	48.3	All	OR	1500	750	1.47	1.06	2.06
Chung, S. J.	2008	Korea	48.3	All	OR	1500	750	1.52	1.04	2.22
Koo, J. S.	2009	Korea	43.7	All	OR	1029	42	7.21	1.78	29.19
Koo, J. S.	2009	Korea	43.7	All	OR	1029	42	3.52	1.23	10.10
Nam, S. Y.	2009	Korea	48.8	M	OR	8571	552	1.15	0.59	2.23
Nam, S. Y.	2009	Korea	48.8	M	OR	8571	552	1.24	0.76	2
Chua, C. S.	2009	China	48.3	All	OR	854	427	1.41	1.05	1.89
Nam, S. Y.	2010	Korea	50.2	All	OR	4274	495	2.09	1.28	3.58
Nam, S. Y.	2010	Korea	50.2	All	OR	4274	495	2.10	1.31	3.39
Nam, S. Y.	2010	Korea	50.2	All	OR	4274	495	2.35	1.27	4.32
Wu, P.	2011	China	46.7	M	OR	372	182	3.41	1.52	7.62
Wu, P.	2011	China	46.7	W	OR	372	182	3.37	1.61	7.06
Chiba, H.	2012	Japan	34.0	All	OR	4990	728	1.276	1	1.628
Sogabe, M.	2012	Japan	56.9	M	OR	265	55	1.17	0.468	2.924
Jung, J. G.	2013	Korea	52.9	All	OR	296	50	1.1	0.466	2.603
Jung, J. G.	2013	Korea	52.9	All	OR	296	50	1.23	0.524	2.867
Loke, S. S.	2013	China	51.2	All	OR	1014	507	1.38	1.00	1.86
Hung, W. C.	2014	China	48.9	All	OR	12090	1922	1.14	1	1.31
Sogabe, M.	2014	Japan	58.3	W	OR	454	51	0.687	0.324	1.455
Ze, E. Y.	2017	Korea	47.2	All	OR	728	65	2.04	1.18	3.51
Lee, S. W.	2017	China	54.3	All	OR	200	100	6.22	3.34	11.57
Baeg, M. K.	2018	Korea	51.6	All	OR	10338	970	1.807	1.496	2.182
Baeg, M. K.	2018	Korea	51.6	All	OR	10338	970	1.585	1.308	1.92
Mazzoleni, F.	2019	Brazil	46.7	All	OR	351	39	1.861	0.687	5.042
Kim, T. J.	2019	Korea	47.3	All	HR	14725	1167	1.31	1.11	1.55
Yang, H. J.	2019	Korea	39.0	M	HR	142679	29509	1.13	1.06	1.21
Yang, H. J.	2019	Korea	39.0	W	HR	142679	29509	1.10	1.03	1.17
Hsieh, Y. H.	2019	China	50.1	All	OR	4895	2916	1.173	1.02	1.349

RE: reflux esophagitis; M: men; W: women; All: all genders; OR: odds ratio; HR: hazard ratio.

Table 2. The baseline characteristics of studies presenting continuous data.

First author	Year	Design	Exposure assessment	Basic diseases	Country	Baseline age	Gender	Sample size	Reflux esophagitis			Control
									Cases	Mean	SD	Cases	Mean	SD
Lee, H. L.	2009	C-C	CT: VFA	No	Korea	51.7	All	200	100	104.68	39.47	100	75.9	49.1
Lee, H. L.	2009	C-C	WHR	No	Korea	51.7	All	200	100	0.92	0.04	100	0.89	0.05
Mulholland, H. G.	2009	C-C	WHR	No	Britain	64.1	All	338	219	0.94	0.08	119	0.96	0.07
Tai, C. M.	2010	C-S	WHR	Obese	China	31.4	All	260	84	0.95	0.08	176	0.92	0.08
Tai, C. M.	2010	C-S	WC	Obese	China	31.4	All	260	84	123.7	16.8	176	117.3	15.1
Gunji, T.	2011	C-S	CT: VFA	No	Japan	51.7	M	9840	1831	104.7	52.8	8009	92.6	49.2
Gunji, T.	2011	C-S	WC	No	Japan	51.7	M	9840	1831	87.7	8.5	8009	85.3	8.1
Hsu, C. S.	2011	C-S	WC	No	China	51.7	All	741	131	83.8	10	610	80.4	9.4
Chung, S. J.	2011	C-C	WC	No	Korea	48.7	All	5616	2808	88.0	8.2	2808	85.9	7.8
Mokrowiecka, A.	2012	C-C	WHR	No	Poland	54.5	All	60	30	0.71	0.11	30	0.6	0.11
Wu, P.	2013	C-C	WHR	No	China	49.7	All	537	268	0.87	0	269	0.86	0
Wu, P.	2013	C-C	WC	No	China	49.7	All	537	268	82.8	0.6	269	81	0.6
Wu, Y. W.	2014	C-S	CT: VAT	No	China	55.0	All	458	178	72.8	27.6	280	61.7	29
Wu, Y. W.	2014	C-S	WC	No	China	55.0	All	458	178	90.7	8.4	280	87.1	9.6
Matsuzaki, J.	2015	C-C	CT: VFA	No	Japan	59.8	W	2608	34	105	46	949	64	38
Matsuzaki, J.	2015	C-C	WC	No	Japan	59.8	W	2608	34	85.9	11.1	949	78.3	10
Matsuzaki, J.	2015	C-C	CT: VFA	No	Japan	59.8	M	2608	182	117	47	1443	103	47
Matsuzaki, J.	2015	C-C	WC	No	Japan	59.8	M	2608	182	85.7	7.7	1443	83.3	8.4
Chung, H.	2016	C-S	WC	No	Korea	53.5	All	705	66	90.5	10	639	82.1	8.5
Hung, W. C.	2016	C-S	WC	No	China	49.4	M	11,952	1438	89.5	8.7	5739	87.6	8.7
Hung, W. C.	2016	C-S	WC	No	China	49.4	W	11,952	484	78.3	9.7	4429	77.0	9.3
Nam, S. Y.	2017	Cohort	CT: VAT	No	Korea	50.7	All	1503	83	1170	430	1420	915	458
Nam, S. Y.	2017	Cohort	WC	No	Korea	50.7	All	1503	83	90.9	7.3	1420	85.2	8
Park, J. K.	2017	C-S	WC	No	Korea	52.0	W	182,407	2858	81.38	8.4	76,377	78.66	7.72
Park, J. K.	2017	C-S	WC	No	Korea	52.0	M	182,407	15,965	89.35	7.6	87,207	87.3	7.44
Rafat, M. N.	2018	C-C	WHR	Patient	Egypt	48.7	All	130	40	0.91	0.06	40	0.83	0.05
Rafat, M. N.	2018	C-C	WC	Patient	Egypt	48.7	All	130	40	99	8	40	81	8
Rafat, M. N.	2018	C-C	WHR	Patient	Egypt	48.7	All	130	50	0.83	0.06	40	0.83	0.05
Rafat, M. N.	2018	C-C	WC	Patient	Egypt	48.7	All	130	50	86	9	40	81	8
Bang, K. B.	2018	Cohort	WC	No	Korea	41.0	All	1126	481	88	8	645	86	9.3
Nam, S. Y.	2019	Cohort	CT: VAT	No	Korea	51.4	All	163	83	1424	503	80	1118	440
Nam, S. Y.	2019	Cohort	WC	No	Korea	51.4	All	163	83	90.9	7.3	90	88.3	7.7
Takahashi, K.	2020	Cohort	CT: VFA	Obese	Japan	41.2	All	674	163	195.07	74.35	511	187.59	157.01
Chung, T. H.	2020	Cohort	WC	No	Korea	50.1	All	7123	276	87.1	6.9	6847	84.4	7.0
Tae, C. H.	2020	C-S	WC	No	Korea	45.2	W	1902	40	75.5	7.7	740	75.9	33.8
Tae, C. H.	2020	C-S	WC	No	Korea	45.2	M	1902	178	85.7	6.6	944	83.6	7.5

RE: reflux esophagitis; C-C: case–control study; C-S: Cross-sectional study; CT: computed tomography; VFA: visceral fat area (cm²); VAT (cm3): visceral adipose tissue; WHR: waist-to-hip ratio; WC: waist circumference (cm); M: men; W: women; All: all genders.

In terms of research characteristics, 11 articles adopted cohort studies as the experimental design [14,23–25,33,37,43,45,46,48,50], 14 adopted case–control studies [15,29,31,35,38–41,47,53,55,56,59,61], and the remaining 17 adopted cross-sectional studies [16,26–28,30,32,34,36,42,44,49,51,52,54,57,58,60]. Ten studies used CT-measured visceral fat to assess abdominal obesity [28,36,41,43,46,50,53,54,58,61], 9 studies used WHR [15,35,36,40,47,55,56,60,61], and 36 studies used WC [14,16,23–27,29–39,41,42,44–55,57–60].

Supplementary Table 2 shows the quality assessment of the cohort studies using the NOS tool. The average total score was 7.36 (range: 6–8), with a standard deviation of 1.25. Supplementary Table 3 shows the assessment of case-control studies, which had an average score of 6.86 (range: 5–9), with a standard deviation of 1.36. In addition, the quality assessment of the cross-sectional studies using AHRQ criteria is shown in Supplementary Table 4.

Meta-analyses

According to the pooled OR (adjusted OR= 1.51, 95% CI: 1.37–1.66, p < .001), abdominal obesity is significantly associated with an increased risk of RE. Meanwhile, in the continuous data, the pooled SMD also revealed a similar conclusion (SMD= 0.36, 95% CI: 0.30–0.42, p < .001).

To explore possible sources of potential heterogeneity in the overall results and to test the robustness of the results in this meta‐analysis, some prespecified subgroup analyses were conducted. Subgroup analyses of the two conclusions were discussed separately and presented only if more than one study was included.

Table 3 shows the overall and subgroup analyses of studies with adjusted ORs. Stratified by study design, significant associations between abdominal obesity and RE were noted across the three schemes, especially in case–control studies (adjusted OR = 1.92, 95% CI: 1.52–2.43, p < .001) and cross-sectional studies (adjusted OR =1.52, 95% CI: 1.29–1.80, p < .001). In addition, the association was still significant in cohort studies (adjusted OR = 1.15, 95% CI: 1.05–1.25, p = .002) (Figure 2). Stratified by abdominal obesity measurement, statistically significant associations between abdominal obesity and RE were still identified when abdominal obesity was defined by WC (adjusted OR = 1.40, 95% CI: 1.28–1.54, p < .001), WHR (adjusted OR = 2.47, 95% CI: 1.85–3.30, p < .001), and CT-measured visceral fat (adjusted OR= 1.74 95% CI: 1.47–2.06, p < .001). Moreover, the results were stable across geographic region and sex subgroups. Notably, the existence of historical diseases, such as MS (adjusted OR = 1.23, 95% CI: 0.65–2.32, p = .530), may have an impact on this correlation. Interestingly, the effect of abdominal obesity on increased risk of RE persisted regardless of whether the covariates were adjusted. Sensitivity analysis was conducted by omitting each article sequentially, which revealed that no single study had a significant impact on the pooled OR.

Figure 2. Association of abdominal obesity with reflux esophagitis in cohort studies.

Table 3. Overall and subgroup analyses of studies presenting adjusted effect size estimates with 95% CI.

Groups	Number of qualified studies	OR	95% CI	p	I^2
Overall analysis
Risk of RE	22	1.51	1.37–1.66	<.001	78.2%
Subgroup analysis
Characteristics of population
Geographic region
Asia	21	1.50	1.36–1.66	<.001	80.50%
South America	1	1.86	0.69–5.04	–	–
Gender
Man	6	2.04	1.34–3.10	<.001	86.50%
Woman	5	1.72	1.07–2.76	.025	71.90%
History of diseases
Health	19	1.49	1.35–1.65	<.001	79.50%
MS	3	1.23	0.65–2.32	.530	69.80%
Dyspepsia	1	1.86	0.69–5.04	–	–
Characteristics of research
Study design
Cohort	6	1.15	1.05–1.25	.002	49.40%
C-C	7	1.92	1.52–2.43	<.001	73.90%
C-S	9	1.52	1.29–1.80	<.001	73.40%
Measurement of central obesity
CT-measured visceral fat	3	1.74	1.47–2.05	<.001	0
WHR	3	2.47	1.85–3.30	<.001	0
WC	20	1.40	1.28–1.54	<.001	78.50%
Exposure assessment
By researchers	21	1.45	1.32–1.58	<.001	76.40%
Self-reported	1	6.22	3.34–11.58	–	–
Adjusted for covariates
Age	13	1.37	1.24–1.52	<.001	79.30%
Sex	13	1.64	1.43–1.89	<.001	73.70%
BMI	13	1.42	1.27–1.58	<.001	72.20%
Smoking status	14	1.41	1.28–1.55	<.001	78.60%
Drinking status	14	1.34	1.22–1.46	<.001	76.20%
Hiatal hernia	10	1.45	1.25–1.69	<.001	62.50%
Helicobacter pylori infection	5	1.60	1.34–1.92	<.001	50.60%

RE: reflux esophagitis; C-C: case–control study; C-S: cross-sectional study; CT: computed tomography; WHR: waist-to-hip ratio; WC: waist circumference (cm); OR: odds ratio; BMI: body mass index; MS: metabolic syndrome.

Table 4 shows overall and subgroup analyses of studies providing continuous data. When the comparison was stratified by abdominal obesity measurement, the MD was calculated as the index of effect scale. The measurements of WC (MD = 2.87, 95% CI: 2.46–3.28, p < .001), VFA (MD = 0.41, 95% CI: 0.20–0.62, p < .001), VAT (MD = 0.50, 95% CI: 0.35–0.64, p < .001) all suggested higher risks of RE in people with abdominal obesity, while WHR (MD = 0.03, 95% CI: 0.00–0.06, p < .028) failed to show a significant association. The results were stable with stratification by geographic region, sex and study design, which is consistent with the results in Table 3. Sensitivity analysis indicated that pooled SMD (SMD = 1.04, 95% CI: 0.81–1.28, p < .001) was affected when an independent article was included [62]. The possible reasons are inaccurate measurement of WC and participant selection bias. Accordingly, we omitted the article in the present meta-analyses.

Table 4. Overall and subgroup analyses of studies presenting continuous data.

Groups	Number of qualified studies	SMD	95% CI	p	I^2
Overall analysis
Risk of RE	20	0.36	0.30–0.42	<.001	89.3%
Subgroup analysis
Characteristics of population
Geographic region
Asia	17	0.36	0.30–0.42	<.001	88.40%
Europe	2	0.34	−0.89 to 1.58	.584	94.40%
Africa	1	0.72	−0.70 to 2.14	–	–
Gender
Man	5	0.26	0.24–0.28	<.001	12.20%
Woman	4	0.35	0.12–0.57	.003	92.30%
History of diseases
Health	18	0.37	0.31–0.44	<.001	89.90%
Obesity	2	0.20	−0.12 to 0.51	.215	75.00%
Characteristics of research
Study design
Cohort	5	0.35	0.18–0.52	<.001	80.70%
C-C	7	0.57	0.29–0.85	<.001	94.20%
C-S	8	0.29	0.24–0.35	<.001	84.00%
Exposure assessment
By researchers	18	0.38	0.28–0.47	<.001	89.50%
Self-reported	2	0.36	0.26–0.45	<.001	90.90%
Subgroup analysis	Number of qualified studies	MD	95% CI	p	I^2
Characteristics of research
Measurement of central obesity
VAT	3	0.50	0.35–0.64	<.001	16.10%
VFA	4	0.41	0.20–0.62	<.001	88.30%
WHR	6	0.03	0.00–0.06	.028	92.00%
WC	16	2.87	2.46–3.28	<.001	90.10%

RE: reflux esophagitis; C-C: case–control study; C-S: cross-sectional study; VFA: visceral fat area (cm²); VAT (cm³): visceral adipose tissue; WHR: waist-to-hip ratio; WC: waist circumference (cm); SMD: standardized mean difference; MD: mean deviation.

Strong evidence of between-study heterogeneity was revealed by the overall analyses. Through subgroup analyses, including analyses stratified by a cohort study design, the use of visceral fat measured by CT to define abdominal obesity, and adjustments for covariates, the heterogeneity of both the pooled OR and the pooled SMD were improved in some, but not all.

Dose–response analyses

In dose–response analyses, the association was consistently significant with different methods of measurement. The risk of RE was positively related to abdominal obesity (Table 5). The dose–response relationship for the association of WC with RE is illustrated in Figure 3, which shows a J-distribution. When the waist circumference reached 87 cm, the increasing trend of the RE risk was accelerated.

Figure 3. The dose–response plot for the association of abdominal obesity with reflux esophagitis.

Table 5. Dose–response relationship of abdominal obesity with risk of RE.

Measurement of abdominal obesity	Number of qualified studies	OR	95% CI	p	I^2
CT-measured visceral fat
Highest quantile (versus lowest quantile)	4	2.28	1.23–4.24	.009	70.2%
Mid quantile (versus lowest quantile)	4	1.65	1.37–2.00	<.001	1.6%
WHR
Highest quantile (versus lowest quantile)	2	2.79	1.57–4.99	.001	4.1%
Mid quantile (versus lowest quantile)	2	1.60	1.09–2.35	.016	19.3%
WC
Highest quantile (versus lowest quantile)	6	1.17	1.06–1.30	.003	68.1%
Mid quantile (versus lowest quantile)	5	1.12	1.03–1.21	.007	67.8%

RE: reflux esophagitis; CT: computed tomography; WHR: waist-to-hip ratio; WC: waist circumference (cm); OR: odds ratio.

Publication bias

Begg's and filled funnel plots of the association between abdominal obesity and RE were asymmetric (Figure 4), indicating that positive results may have been prioritized. Egger's tests further confirmed the publication bias (p < .01), which suggested that the meta-analysis may have been affected.

Figure 4. (A) The Begg’s funnel plots for studies presenting adjusted effect size estimates (OR or HR) with 95% CI. (B) The filled funnel plots for studies presenting adjusted effect size estimates (OR or HR) with 95% CI. (C) The filled funnel plots for studies presenting continuous data.

Discussion

To the best of our knowledge, this meta-analysis is the most comprehensive assessment exploring the relationship between abdominal obesity and RE. A total of 42 eligible studies with more than 400,000 participants were included in the present systematic review and meta-analysis. Our key finding suggests a 1.51-fold increased risk of RE among participants with abdominal obesity compared with individuals without abdominal obesity. Moreover, our subgroup analyses revealed that study design, abdominal obesity measurement, adjustment for covariates and sex were possible sources of between-study heterogeneity.

In previous literature, several prospective cohort studies have concluded that abdominal obesity was a significant risk factor for RE [37,43,45,50]. However, certain observational studies have yielded controversial findings. A cohort study including 8571 subjects in South Korea found that abdominal obesity was not independently associated with the risk of RE when adjusted for BMI (adjusted OR = 1.15, 95% CI: 0.59–2.23). However, after adjusting for WC, RE was the main consequence of increased BMI (adjusted OR = 4.15, 95% CI: 1.66–10.33) [14]. Similarly, a cross-sectional study conducted in Japan showed no significant correlation between abdominal obesity and RE (adjusted OR = 0.687, 95% CI: 0.324–1.455) [16]. After pooling the results of 17 independent studies, a meta-analysis by Siddharth and colleagues provided novel high-level evidence supporting abdominal obesity as a significant risk factor for the development of RE [12]. Nevertheless, the number of included studies was insufficient, and the dose–response analyses also required improvement.

To fill the evidence-gap, several improvements were applied. First, we added to the number of studies to include 42 studies with 11 cohort studies to lower selective biases. Second, ORs were all adjusted by multiple factors, which may have better reduced the influence of covariates on the results. Third, subgroup analyses were designed to be more detailed. Furthermore, dose–response analyses were conducted according to WC, WHR and CT-measured visceral fat separately.

With extensive research on body fat distribution, abdominal obesity is considered to be as important as general obesity in the pathogenesis of GERD [10]. Increased intra-abdominal pressure compromises the trans-diaphragm pressure balance, which is identified as the primary mechanism by which abdominal obesity increases the RE risk [4,63,64]. First, increased intra-abdominal pressure may exceed the retention pressure of the esophagogastric barrier, disrupt the valvular mechanism and lead to reflux [4,5]. Then, whenever the lower esophageal sphincter (LES) is opened, this pressure increases the speed of gastric contents returning to the esophagus [65,66]. An increased frequency of transient LES relaxation and delayed recovery of the LES also contribute to more acid exposure [4,67]. Accordingly, esophageal clearance was delayed, and esophageal injury occurred. Additionally, disorders of gastric emptying and gastrointestinal motility driven by excessive leptin levels [68], esophageal motility defects induced by fat accumulation at the gastroesophageal junction, and a high incidence of esophageal hiatal hernia are possible explanations for the higher prevalence of RE in central obesity [8,67].

In dose–response analyses, we observed increased abdominal obesity followed by a higher RE risk, which is consistent with the intra-abdominal pressure mechanism. The J-shaped curve shows the increasing trend of the RE risk. When gender and ethnic differences were considered, the significant peak in the trend close to 87 cm was consistent with the definition of abdominal obesity. However, the specific risk index needs to be verified by more literature in the future.

A healthy lifestyle is recommended as an alternative therapeutic method throughout the disease course of RE. Additionally, body fat distribution management is important [69–71]. The present meta-analysis summarized the results of previous studies and highlighted the importance of improving abdominal obesity. Considering clinical practice, this article may provide more credible epidemiological evidence for lifestyle intervention in the treatment of RE.

Several limitations in the current context warrant further discussion. First, the subgroup analysis showed that the included studies were mostly concentrated in Asian populations, which may be related to the higher incidence of abdominal obesity in Asian populations [72,73], indicating that the universality and difference of this correlation cannot be reasonably explained in multiple ethnic backgrounds, and more studies in different regions are needed to confirm our conclusions. Second, although the adjusted effect size estimates were incorporated in our study, some important confounding factors were not considered, such as diet, physical activity, and proton pump inhibitor usage. Third, although a considerable number of subgroup analyses were used to explore the source of heterogeneity, significance persisted in several subgroups, which limited the interpretation of pooled effect-size estimates. Fourth, obvious publication bias existed in the included studies, but since the relevant biological mechanisms were clearly explained and a positive dose–response relationship was provided in the present meta-analysis, the clinical significance of such publication bias may be low.

Conclusions

Our findings suggest that abdominal obesity and an unhealthy body fat distribution may independently increase the risk of RE. In addition, the degree of risk increases with the severity of abdominal obesity. However, to establish the causal relationship, large-scale and well-designed cohort studies are needed for future research.

Abbreviations
AHRQ	=	Agency for Healthcare Research and Quality
BMI	=	Body Mass Index
CI	=	confidence interval
CT	=	computed tomography
GERD	=	gastroesophageal reflux disease
HR	=	hazard ratio
LES	=	lower esophageal sphincter
MD	=	mean difference
NOS	=	Newcastle–Ottawa Scale
OR	=	odds ratio
RE	=	reflux esophagitis
SMD	=	standardized mean difference
VFA	=	visceral fat area
VAT	=	visceral adipose tissue
V/S	=	visceral fat area to subcutaneous fat area ratio
WC	=	waist circumference
WHR	=	waist–hip ratio

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All the data generated or analyzed in this study are included in this article and its supplementary information files.

Additional information

Funding

The project name is "Taishan scholar position construction fund", and grant number is "2018-35”.