A systematic review was conducted at each PICO. Dr. M.C. of the National Evidence-based Healthcare Collaborating Agency and a working group member of each subject selected appropriate search key words and conducted a literature search from July to August 2018 using the Ovid-MEDLINE, EMBASE, Cochrane Library, KoreaMed, and KMBASE databases; the key words were listed in the appendix. The common inclusion criteria for the studies were as follows: (1) adult subjects or patients as the study population; (2) written in English or Korean; (3) systematic reviews and meta-analyses, randomized or non-randomized trials, and observational studies; (4) published between 2008 and 2018; and (5) studies with proper results (e.g., symptom improvement, development of gastric cancer, eradication rates) reported. The common exclusion criteria were as follows: (1) studies on children or teenagers; (2) studies with no proper results reported; (3) duplicated publication; (4) impossible to obtain original text; and (5) expert opinion, case series or report, narrative review, or guidelines. Two independent members reviewed the literatures and selected the final studies according to the inclusion and exclusion criteria. The literature selection process was summarized in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) plot for each PICO. An example of the process of literature selection for key question 1 is shown in Fig. 1. The quality of finally selected studies was assessed using quality assessment tools according to the study design. Randomized controlled trials (RCTs) were evaluated using the Cochrane risk of bias (RoB) tool [16], while non-randomized clinical studies were assessed using the Risk of Bias Assessment Tool for Non-Randomized Study (RoBANS) [17]. If the assessments were not consistent as the two paired working members, the two members and the chairman coordinated a final evaluation.

After systematic literature review, the evidence table was organized and meta-analysis was conducted. Evidence profiles were created based on ‘Grading of Recommendations, Assessment, Development and Evaluation’ (GRADE) (Table 1) [18]. GRADEpro software was used to rank the quality of evidence according to four categories: high, moderate, low, and very low. The quality assessment of the evidence was then used to determine the strength of the supporting evidence that informs a recommendation [19]. The recommendations presented in this guideline are summarized in Table 2.

A literature search was conducted to utilize resources and economic evaluation. The cost of H. pylori eradication, which uses antibiotics and proton pump inhibitors (PPIs) for 7 to 14 days, was not significantly different between the treatment methods. The tailored therapy based on H. pylori susceptibility to antibiotics may be cost-effective in a high clarithromycin-resistant region compared to standard empirical triple therapy. There are reports that tailored treatment is superior in terms of cost-effectiveness if the eradication rate of H. pylori is lowered below 75.3%. Therefore, the tailored treatment based on H. pylori susceptibility to antibiotics was added to the recommendation [20,21].

The aim of expert consensus by modified Delphi methods was to determine the extent to which experts agreed about draft recommendation including evidence [22]. The first round of Delphi process was conducted with e-mail voting and 44 experts were invited, with 30 participating in the first agreement. The first round was investigated using the 9-likert scale’s self-reporting questionnaire asking the extent of agreement, along with evidence data from the Development Committee for each recommendation. In the response scale, one point was “completely disagreeable” and nine points were “very agreeable.” Consent was considered when the ratio of points from 7 to 9 (high agreement) was more than two-thirds. Eight recommendations were agreed upon for a total of 12 recommendations. Indication of H. pylori eradication therapy with IDA and atrophy/IM and eradication regimen of H. pylori with standard triple therapy and sequential therapy failed to reach agreement, so the Development Committee revised these 4 recommendations after the first voting and conducted the second round of voting in a face-to-face agreement (December 14, 2019). A second round of voting was conducted anonymously, and recommendations for atrophy gastritis/IM were rejected by 48% of 23 respondents, with the remaining recommendations passed and finally 11 recommendations were adopted.

In the second round of face-to-face voting for the expert consensus process, various drafts were adopted after the anonymous vote. After the process of expert agreement and external review, the draft was revised with their opinions.

This guideline was provided on the Korean College of Helicobacter and Upper Gastrointestinal Research (http://www.hpylori.or.kr) and the Korean Association of Internal Medicine (http://www.kaim.or.kr) websites. In addition, this guideline will be published in Korean and English as a paper and will be spread throughout the academic symposium.

Can H. pylori eradication increase the hemoglobin in patients with idiopathic iron deficiency anemia?

Anemia is a major health problem and mostly caused by iron deficiency [23,24]. The estimated prevalence of anemia is 24.8% (95% CI, 22.9% to 26.7%), affecting 1.62 billion people (95% CI, 1.50 to 1.74 billion) globally [23], and concentrated in preschool children and women. H. pylori infection causes diverse gastrointestinal diseases, including chronic gastritis, peptic ulcer disease and cancer. Furthermore, H. pylori chronic gastritis can induce decreasing gastric acid secretion and gastric ascorbic acid, which are essential for the absorption of dietary iron [25,26].

H. pylori has been associated with IDA. A recent meta-analysis revealed a significantly higher likelihood of IDA in subjects with H. pylori-infection (pooled odds ratio [OR], 1.72; 95% CI, 1.23 to 2.42) [27]. However, this association was strong in children, but subgroup analysis of adult population revealed weaker association with significant heterogeneity (pooled OR, 1.70; 95% CI, 1.01 to 2.85) [23].

The role of H. pylori infection in IDA was shown in studies with H. pylori eradication therapy combined with iron supplementation to treat the IDA. However, these studies were heterogeneous because of confounders including age (children or adolescent vs. adults), sex and different study setting in terms of different definition of IDA or outcomes (quantitative assessment of ferritin or hemoglobin or qualitative assessment, such as recovery from anemia). Meta-analysis including children, adolescent, or adults showed significant increase of ferritin after the eradication, not hemoglobin. Meta-analysis of s even RCTs showed increased ferritin, standardized mean difference (SMD) 0.53 (95% CI, 0.21 to 0.85), but not hemoglobin, SMD 0.36 (95% CI, −0.07 to 0.78) [27]. However, there were limited studies which showed the efficacy of H. pylori eradication in adult population. Non-randomized comparative study in adults with IDA showed the additional effect of H. pylori eradication on iron supplement in adult patients with IDA and H. pylori-positive chronic gastritis [28]. In a prospective observational study, H. pylori infection was correlated with low serum ferritin level and after eradication, serum ferritin increased; however, the sample size was too small [29]. In this study, serum ferritin in premenopausal women was significantly lower than that of post-menopausal women, but not different in men. In other observational studies with adult patients with IDA, IDA was resolved after 38.1% of eradication of H. pylori-eradicated patients (32/84). This was more frequent in men and postmenopausal women compared with premenopausal women (75% vs. 23%, p < 0.01) [30]. Despite the very low level of evidence, it was decided as a “weak recommendation” because short-term treatment of H. pylori infection has the potential for long-term benefits and low risk for serious harm.

Is H. pylori eradication helpful to prevent metachronous recurrence after endoscopic resection of gastric adenoma?

Many studies have reported that the incidence rate of metachronous cancer decreased with H. pylori eradication after ER of early gastric cancer (EGC) [31–33]. Thus, H. pylori should be eradicated to prevent metachronous recurrence after ER of EGC. However, there was no definite guideline about H. pylori eradication after ER of gastric adenoma. Until now, there were two RCTs about H. pylori eradication to prevent metachronous gastric cancer after ER of gastric tumors including EGC and adenoma (Supplementary Table 1) [31,33]. Three retrospective studies about H. pylori eradication after ER of gastric adenoma were reported [34–36]. All of them were conducted in Korea. According to studies, the incidence of metachronous recurrence was lower in H. pylori eradicated group than non-eradicated group (3.24% vs. 4.87% [33]; 7.69% vs. 14.29% [31]; 7.76% vs. 10.80% [34]; 8.20% vs. 19.44% [35]; 4.71% vs. 11.36% [36]). When meta-analysis included five studies, the effect of H. pylori on prevention of metachronous recurrence after ER of gastric adenoma was statistically significant (OR, 0.55; 95% CI, 0.34 to 0.92) (Fig. 2).

According to studies, H. pylori eradication is helpful to prevent metachronous recurrence after ER of gastric adenoma. Therefore, H. pylori eradication is indicated after ER for H. pylori-positive gastric adenoma. However, RCT focused on gastric adenoma is required to support this recommendation.

Is H. pylori eradication helpful in symptom relief in patients with functional dyspepsia?

In the meta-analysis of RCTs, when H. pylori was eradicated in dyspeptic patients, the symptom improvement was not significant in the short-term (3 months) follow-up, but symptoms were significantly improved in the long-term (6 to 12 months) follow-up [37,38]. Based on these results, the Maastricht V guidelines in Europe and the United States and Canadian guidelines strongly recommend the eradication of H. pylori as the first-line treatment for dyspepsia [11,39].

In the present guideline, 18 RCTs from January 1997 to December 2017 were selected and meta-analysis was conducted to evaluate the long-term effects of H. pylori eradication in patients with dyspepsia (Supplementary Table 2) [40–57]. In a meta-analysis of 4,672 patients from 18 RCTs, the risk ratio (RR) of persistence of dyspeptic symptoms in the control group was 1.18 (95% CI, 1.07 to 1.31) compared with the eradication group. Although statistically significant, the number of patients needed for treatment (number needed to treat [NNT]) was 14, and heterogeneity among studies was moderate (I² = 34%) (Supplementary Fig. 1) [58].

Because of the heterogeneity among the studies, subgroup analysis was performed by region to analyze five RCTs from the Asian regions and 13 RCTs from outside Asian regions. Meta-analysis from RCTs from outside Asia regions showed significant improvement of dyspeptic symptoms in eradication group (RR, 1.22; 95% CI, 1.08 to 1.38; I² = 33%). However, because of analysis of RCTs from Asia, the effect of eradication on improvement of dyspeptic symptoms was not significant (RR, 1.10; 95% CI, 0.92 to 1.31; I² = 32%).

In summary, eradication of H. pylori improved dyspeptic symptoms significantly, however, the clinical effect was not large due to the improvement of symptoms in 1 of 14 treated patients (NNT = 14) and the result of subgroup analysis of RCTs conducted in Asia was not statistically significant. The prevalence of H. pylori in Korea is estimated to be 54% (95% CI, 50.1% to 57.8%) according to a study that estimates the prevalence of H. pylori worldwide [59]. In areas with high prevalence of H. pylori, costs, adverse effects associated with eradication therapy, the risk of emergence of resistance strains, and re-infection are thought to be higher than those of low prevalence regions. Therefore, in the present guideline, it was decided to make weak recommendations despite the high level of evidence for H. pylori eradication in patients with functional dyspepsia. The RCTs, including cost-effectiveness analysis of eradication therapy in patients with functional dyspepsia in areas with high prevalence of H. pylori, including in Korea, are likely to be needed.

Is H. pylori eradication effective to prevent gastric cancer in the presence of chronic atrophic gastritis and intestinal metaplasia?

H. pylori eradication can reduce the risk of gastric cancer development. However, it is controversial whether the eradication can be beneficial in individuals with pre-neoplastic lesions including chronic atrophic gastritis (CAG) and IM.

Recently published two meta-analyses showed that individuals with non-atrophic or CAG benefited from eradication in reducing the risk of gastric cancer, whereas individuals with IM and/or dysplasia did not [60,61]. The effect of H. pylori eradication can be affected by the degree of mucosal atrophy. H. pylori eradication can be more beneficial to subjects with mild mucosal atrophy than those with extensive atrophic gastritis [62]. Maastricht V guideline recommended that the risk for gastric cancer can be reduced more effectively by eradicating H. pylori before atrophy and IM develop [11].

However, the two meta-analyses included the studies that evaluated the effect of H. pylori eradication on the occurrence of metachronous gastric cancers after ER of EGC, as well as the studies in the general population. The effect of H. pylori eradication may be different between in the general population and in the high-risk group. Moreover, a population-based cohort study in China, which was not included in the previous meta-analyses, showed that H. pylori eradication can benefit individuals with IM and/or dysplasia at baseline, suggesting H. pylori eradication can benefit an entire population regardless of the baseline gastric histopathology [63].

When the meta-analysis was performed using the RCTs in the general population only, H. pylori eradication significantly reduced the incidence of gastric cancer, as in previous studies (Fig. 3). In a subgroup analysis that included only subjects with CAG or IM, eradication had no effect on the prevention of gastric cancer, and two studies involving subjects without CAG and IM did not demonstrate significant gastric cancer prevention effect (Supplementary Fig. 2). However, in the latter case, the number of incidences of gastric cancer were small, and there were limitations in drawing an accurate conclusion. In expert consensus based on these analyses, only 48.0% agreed in the first e-mail questionnaire, and only 63.3% agreed in the second face-to-face meeting. In other words, there is no firm evidences or expert agreement to recommend eradication of H. pylori in subjects with CAG or IM, so re-discussion is needed after more researches have been accumulated. The indications for H. pylori eradication treatment presented above are summarized in Table 3.

In patients undergoing H. pylori eradication for the first time, one of the following four regimens can be used: (1) 14-day standard triple therapy; (2) non-bismuth quadruple therapy; (3) 7-day standard triple therapy after clarithromycin resistance test; and (4) bismuth quadruple therapy.

What is the recommended salvage regimen after failure of previous H. pylori eradication therapy?

In the last decade, the efficacy of PPI, clarithromycin, and amoxicillin triple therapy has decreased mainly due to clarithromycin resistance [145,146]. As a result, it has become a common situation in clinical practice to choose a salvage regimen after failure of one or more eradication attempts. Moreover, the selection of a rescue regimen may be more complicated with the emergence of alternative first-line treatments such as sequential or CT.

In the systematic literature review, there were 36 RCTs that compared different combinations of antibiotics, different durations of a regimen, or different PPIs as salvage therapy after one or more eradication failures between 2008 and 2017 (Supplementary Table 8) [147–182]: 24 RCTs evaluated second line regimens, five evaluated third line regimens, five compared different durations of a regimen, and two compared different doses or kinds of PPIs. Most studies for second-line regimens were conducted after failure of clarithromycin-based triple therapy; there was no RCT that evaluated second-line regimens after failure of non-bismuth or bismuth quadruple therapy, and there was no RCT that evaluated third-line regimen after failure of first-line clarithromycin-based triple therapy followed by second-line bismuth quadruple therapy. Meta-analyses were conducted and there were more than three RCTs. The recommendations on the salvage regimen were primarily based on those RCTs and their meta-analyses. However, when no suitable trial was found, the most relevant cohort studies, published systematic reviews of cohort studies, and RCTs published before 2008 were referenced for the recommendations. All eradication rates presented below are from ITT analyses. The evidences which were included in the meta-analysis of regimens for salvage therapy are summarized in the Supplementary Table 9.

In the 2013 revised Korean guidelines, bismuth quadruple therapy (PPI, bismuth, tetracycline, and metronidazole) for 7 to 14 days was recommended. The systematic review conducted for the present guidelines identified 15 RCTs which compared 31 treatment arms as 2nd line treatment after failure of 1st line PPI-clarithromycin-amoxicillin triple therapy [147,152,156,161–163,165–168,174–176,179,182]. Of those, nine studies adopted bismuth quadruple therapy [156,161,162,165,167,175, 176,179,182], and eight studies adopted levofloxacin triple therapy (PPI, amoxicillin, and levofloxacin) [147,152,165,166,168,175,176,182]; of those, four studies compared bismuth quadruple therapy and levofloxacin triple therapy directly [165,175,176,182].

Bismuth quadruple therapy showed pooled eradication rate of 75.5% (95% CI, 71.6% to 79.1%) in the meta-analysis of the nine studies (Fig. 8). Regarding the treatment duration, four studies treated for 7 days, two for 10 days, and three for 14 days. Because 10- and 14-day regimens showed similar efficacy, subgroup analysis was conducted to compare 7 day versus 10- to 14-day bismuth quadruple therapy. In result, 10- to 14-day therapy showed significantly higher eradication rates (pooled eradication rate, 81.6%; 95% CI, 76.9% to 85.6%; I² = 29.6%) than 7-day therapy (pooled eradication rate, 68.4%; 95% CI, 53.0% to 73.5%; I² = 73.8%) (p < 0.01). The systematic review also identified three RCTs comparing 14-day versus 7-day bismuth quadruple therapies. Meta-analysis of these RCTs also showed a significant eradication rate with 14-day regimen than 7-day regimen (risk difference [RD], 0.09; 95% CI, 0.02 to 0.15; p < 0.01).

Levofloxacin triple therapy showed pooled eradication rate of 73.1% (95% CI, 68.4% to 77.3%) in the meta-analysis of the eight studies (Fig. 9). In the subgroup analysis comparing treatment duration, the 10- to 14-day regimen in four studies showed a significantly higher eradication rate (78.5%; 95% CI, 71.9% to 84.0%) than 7-day regimen in four studies (69.1%; 95% CI, 61.6% to 74.9%; p = 0.04). One factorial RCT reported 10-day therapy showed significantly higher eradication rate with 7-day therapy (87.5% vs. 67.5%, p < 0.01) [177]. Meanwhile, there was no significant difference in the dose of levofloxacin between 500 mg daily and 1,000 mg daily in this study (p = 1.00).

There was no significant difference in the eradication rates between bismuth quadruple therapy and levofloxacin triple therapy in the meta-analysis of four RCTs. There were non-significant contradictory trends favoring levofloxacin triple therapy in ITT analysis (bismuth quadruple vs. levofloxacin triple: RD, −0.06; 95% CI, −0.14 to 0.02; p = 0.16) but favoring bismuth quadruple therapy in PP analysis (RD, 0.02; 95% CI, −0.05 to 0.10; p = 0.58) (Supplementary Fig. 10). These results may be because of low tolerability of bismuth quadruple therapy. In two systematic reviews published in 2006, 10-day levofloxacin triple therapy showed superior efficacy compared to 7-day bismuth quadruple therapy [183,184]. In our meta-analysis, two regimens were treated for the same duration (7 days vs. 7 days, 10 days vs. 10 days, or 14 days for 14 days). Thus, it can be suggested that bismuth quadruple therapy and levofloxacin triple therapy may have similar efficacy when the two treatments are administered for the same duration.

A major limitation of levofloxacin triple therapy is that efficacy of the regimen is substantially reduced in the presence of levofloxacin resistance [165]. In Korea, the resistance rate for levofloxacin in H. pylori strains has been increasing rapidly as high as 28.1% [14,185,186]. Very recently, the nationwide antibiotic resistance profile of H. pylori in Korean population was reported [2]. According to this report, resistance rate against levofloxacin was 37.0%. Thus, bismuth quadruple therapy would be favored over levofloxacin triple therapy in Korea. However, it should also be noted that resistance rate against metronidazole was also as high as 29.5% in the same study. Because resistance to metronidazole can be overcome with increased duration and dose, 14-day course would be preferred to 10- to 14-day course for bismuth quadruple therapy as a salvage regimen in Korea [187]. Therefore, bismuth quadruple therapy for 14 days is recommended as a second-line therapy after failure of standard triple therapy.

There was no RCT comparing salvage regimens after failure of first-line non-bismuth quadruple therapy. There was a meta-analysis of cohort studies in which most studies included were a levofloxacin triple regimen [188]. In this study, the pooled eradication rate of levofloxacin triple therapy from five studies including 86 patients was 81% (95% CI, 71% to 91%; I² = 28%). Another meta-analysis conducted in Maastricht V guidelines showed 81% (six studies; 95% CI, 73% to 90%; I² = 19%) after failure of sequential therapy and 78% (three studies; 95% CI, 58% to 97%; I² = 67%) after failure of CT [11]. However, these results may not be directly applicable to Korean population because of high levofloxacin resistance rates as previously discussed [2,14,185,186].

Bismuth quadruple therapy showed eradication rate of 84% (95% CI, 63% to 106%; I² = 56%) in the meta-analysis conducted in Maastricht V [11]. However, the analysis included only two cohort studies of similar sample sizes with each other. One of them was a study conducted in Korea where 14-day bismuth quadruple therapy achieved successful eradication in 10 of 14 patients.

Therefore, bismuth quadruple therapy is recommended as a second-line therapy after failure of first-line non-bismuth quadruple therapy based on currently available evidences. However, more data is required to support this recommendation.

The most common situation in which bismuth quadruple therapy fails in Korea is failures as a second-line regimen after failure of first-line standard triple therapy. This approach was recommended by previous Korean guidelines 2013 and Maastricht IV guidelines [8,189]. The scenario in which second-line bismuth quadruple therapy fails after failure of first-line non-bismuth quadruple therapy is also expected to be a more common situation. In these cases, it is not recommended to use clarithromycin again in the third-line regimen [187]. It would be also inappropriate to use clarithromycin after failure of first-line bismuth quadruple therapy because this regimen had been chosen as first-line when clarithromycin resistance was suspected [8]. Treatment regimen may be decided based on antibiotics susceptibility tests, either by culture, PCR, or sequencing analysis. Generally, it is recommended not to use clarithromycin, fluoroquinolone, and rifabutin again in the presence of resistance to respective drugs, while amoxicillin and metronidazole may be re-used [187]. However, it is noteworthy that benefits of susceptibility-guided therapy over empirical regimen was evident only in the first-line treatment, not in the second-line setting in a recent systematic review [139]. Neither was it in the third-line treatment in a recent RCT [190]. The first-line and salvage treatment regimens and algorithms for H. pylori treatment combined with the regimens are summarized in Table 4 and Fig. 10, respectively.