Short‐term treatment with proton pump inhibitors, H2‐receptor antagonists and prokinetics for gastro‐oesophageal reflux disease‐like symptoms and endoscopy negative reflux disease

Kirsten E Sigterman; Bart van Pinxteren; Peter A Bonis; Joseph Lau; Mattijs E Numans

doi:10.1002/14651858.CD002095.pub5

. 2013 May 31;2013(5):CD002095. doi: 10.1002/14651858.CD002095.pub5

Short‐term treatment with proton pump inhibitors, H2‐receptor antagonists and prokinetics for gastro‐oesophageal reflux disease‐like symptoms and endoscopy negative reflux disease

Kirsten E Sigterman ¹, Bart van Pinxteren ², Peter A Bonis ³, Joseph Lau ⁴, Mattijs E Numans ^5,^✉

Editor: Cochrane Upper GI and Pancreatic Diseases Group

PMCID: PMC7066537 PMID: 23728637

Abstract

Background

Approximately 25% of adults regularly experience heartburn, a symptom of gastro‐oesophageal reflux disease (GORD). Most patients are treated empirically (without specific diagnostic evaluation e.g. endoscopy. Among patients who have an upper endoscopy, findings range from a normal appearance, mild erythema to severe oesophagitis with stricture formation. Patients without visible damage to the oesophagus have endoscopy negative reflux disease (ENRD). The pathogenesis of ENRD, and its response to treatment may differ from GORD with oesophagitis.

Objectives

Summarise, quantify and compare the efficacy of short‐term use of proton pump inhibitors (PPI), H2‐receptor antagonists (H2RA) and prokinetics in adults with GORD, treated empirically and in those with endoscopy negative reflux disease (ENRD).

Search methods

We searched MEDLINE (January 1966 to November 2011), EMBASE (January 1988 to November 2011), and EBMR in November 2011.

Selection criteria

Randomised controlled trials reporting symptomatic outcome after short‐term treatment for GORD using proton pump inhibitors, H2‐receptor antagonists or prokinetic agents. Participants had to be either from an empirical treatment group (no endoscopy used in treatment allocation) or from an endoscopy negative reflux disease group (no signs of erosive oesophagitis).

Data collection and analysis

Two authors independently assessed trial quality and extracted data.

Main results

Thirty‐four trials (1314 participants) were included: fifteen in the empirical treatment group, fifteen in the ENRD group and four in both. In empirical treatment of GORD the risk ratio (RR) for heartburn remission (the primary efficacy variable) in placebo‐controlled trials for PPI was 0.37 (two trials, 95% confidence interval (CI) 0.32 to 0.44), for H2RAs 0.77 (two trials, 95% CI 0.60 to 0.99) and for prokinetics 0.86 (one trial, 95% CI 0.73 to 1.01). In a direct comparison PPIs were more effective than H2RAs (seven trials, RR 0.66, 95% CI 0.60 to 0.73) and prokinetics (two trials, RR 0.53, 95% CI 0.32 to 0.87).

In treatment of ENRD, the RR for heartburn remission for PPI versus placebo was 0.71 (ten trials, 95% CI 0.65 to 0.78) and for H2RA versus placebo was 0.84 (two trials, 95% CI 0.74 to 0.95). The RR for PPI versus H2RA was 0.78 (three trials, 95% CI 0.62 to 0.97) and for PPI versus prokinetic 0.72 (one trial, 95% CI 0.56 to 0.92).

Authors' conclusions

PPIs are more effective than H2RAs in relieving heartburn in patients with GORD who are treated empirically and in those with ENRD, although the magnitude of benefit is greater for those treated empirically.

Plain language summary

Short‐term treatment with medications for heartburn symptoms

Patients with only mild or intermittent heartburn may have adequate relief with lifestyle modifications and with antacids, although other options are available. The two most commonly used drugs for treatment of heartburn are H2‐receptor antagonists (H2RAs) and proton pump inhibitors (PPIs). These drugs act by suppressing the release of acid from the stomach. This review found that in the short term PPIs relieve heartburn better than H2RAs in patients who are treated without specific diagnostic testing. Although the difference is smaller, this is also true for patients with gastro‐oesophageal reflux disease (GORD), who have a normal upper endoscopy . In summary, proton pump inhibitor drugs appear to be more effective than H2‐receptor antagonists for relieving heartburn.

Summary of findings

Background

Description of the condition

Approximately one in four adults in Western society experience heartburn at least monthly, while 5% suffer from it daily (Corder 1996; Isolauri 1995; Locke 1997; Nebel 1976; Thompson 1982). Heartburn is associated with reduced quality of life (Dimenas 1996), with billions of dollars spent annually on healthcare costs associated with its evaluation and treatment. Heartburn, especially if accompanied by acid regurgitation, is the typical clinical manifestation of gastro‐oesophageal reflux disease (GORD), a term referring to both symptoms and oesophageal mucosal damage resulting from reflux of gastric acid into the oesophagus. A consensus of experts defined GORD as a condition that develops when the reflux of stomach contents causes troublesome symptoms and/or complications. Novel aspects of the new definition include a patient‐centred approach that is independent of endoscopic findings (Vakil 2006). The disease was sub‐classified into oesophageal and extra‐oesophageal syndromes. The accuracy of symptoms in diagnosis of GORD depends on the nature and severity of symptoms and the reference standard used. One study (Klauser 1990) found that clearly dominant heartburn had a positive predictive value of 81% in diagnosing GORD (when defined by oesophageal pH (acidity) monitoring). Accuracy was reduced when the presence of heartburn alone was considered. Monitoring of oesophageal pH has been proposed as a reference standard for GORD, but is inconvenient and the result is negative in more than one third of patients with chronic heartburn. Furthermore, many patients with a normal pH do respond to antacids and have symptoms reproducible by acid infusion, while some even have oesophageal mucosa damage (Rodriguez 1999; Shi 1995). The main importance of the use of endoscopy lies in diagnosing oesophagitis with possible complications such as bleeding, stricture formation, Barrett's metaplasia and adeno‐carcinoma. However, between half and two‐thirds of patients presenting with typical GORD symptoms have no endoscopic abnormalities (Joelsson 1989; Johansson 1986b; Johnsson 1987; Robinson 1998; Tefera 1997). Their condition is referred to as endoscopy negative reflux disease (ENRD). The severity and chronicity of symptoms in patients with ENRD is similar to that of patients with oesophagitis (Dent 1998; Johansson 1986b).

Description of the intervention

Several drugs are available for treatment of GORD. The most commonly used are H2‐receptor antagonists (H2RAs) and proton pump inhibitors (PPIs). Prokinetic agents are used much less commonly.

How the intervention might work

H2‐receptor antagonists and proton pump inhibitors improve symptoms by reducing gastric acid secretion and hence oesophageal acid exposure. By contrast, prokinetic agents work principally by increasing lower oesophageal sphincter tone, thereby reducing reflux.

Why it is important to do this review

There is considerable variability in the choice of initial therapy and the use of endoscopy across healthcare settings. The variability is in part related to an incomplete understanding of response to treatment in clinically relevant subgroups of patients. This concern is particularly relevant for patients with ENRD, since most studies of GORD therapy have focused on patients with oesophagitis (Chiba 1997). Similarly, the degree to which one option or another is better in patients who do not undergo investigation (i.e. are treated empirically) is incompletely understood.

Objectives

Summarise, quantify and compare the efficacy of the short‐term use of proton pump inhibitors, H2‐receptor antagonists and prokinetics in adults with suspected gastro‐oesophageal reflux disease who are treated empirically and in those with endoscopy negative reflux disease.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials with a single‐ or double‐blinded design, in which one of the intervention types was contrasted with placebo or another intervention type.

Types of participants

Adults
Either gender
Predominant heartburn (a retrosternal burning sensation), diagnosed as GORD or reflux‐like dyspepsia
Classifiable in one of the following two groups:
- empirical treatment group: no endoscopy performed or endoscopy results not used in allocating treatment;
- endoscopy negative reflux disease group: on endoscopy either a normal oesophageal mucosa or diffuse erythema (i.e. no erosive oesophagitis).

Types of interventions

Short‐term treatment (one to twelve weeks) with proton pump inhibitors (esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole and dexlansoprazole), H2‐receptor antagonists (cimetidine, famotidine, nizatidine and ranitidine) or prokinetics (cisapride, domperidone and metoclopramide).

Types of outcome measures

We studied the following comparisons for both the empirical treatment group and the ENRD group:

PPI versus placebo;
H2RA versus placebo;
prokinetic versus placebo;
PPI versus H2RA;
PPI versus prokinetic;
H2RA versus prokinetic.

Primary outcomes

Heartburn remission (defined as no more than one day per week with mild heartburn).

Secondary outcomes

(Partial) symptom relief; quality of life.

Studies using other types of symptomatic outcome measures were not included in our formal analysis.

Search methods for identification of studies

Electronic searches

We constructed the original search strategy in Appendix 1 by using a combination of MeSH subject headings and text words relating to the symptoms of gastro oesophageal reflux disease (GORD) and the associated pharmacological interventions. We applied the standard Cochrane search strategy filter for identifying randomised controlled trials to this search strategy.

For the updated review, the Cochrane Highly Sensitive Search Strategy for identifying randomised trials in MEDLINE, sensitivity maximising version; Ovid format (Cochrane Handbook), was combined with the search terms in the Appendices to identify randomised controlled trials in MEDLINE. The MEDLINE search strategy was adapted for use in the other databases searched.

We identified new reports of trials for the updated review by searching MEDLINE January 1966 to November 2011 (Appendix 2), EMBASE January 1988 to November 2011 (Appendix 3), and evidence‐based medicine reviews (including Cochrane DSR, ACP Journal Club, DARE, CCTR, CMR, HTA, and NHSEED to November 2011; Appendix 4). We did not confine our search to English language publications. Searches in all databases were first conducted in December 2005, updated in November 2008 and in November 2011.

Searching other resources

We handsearched reference lists from trials selected by electronic searching to identify further relevant trials.

Data collection and analysis

Selection of studies

We screened the titles and abstracts of trials identified using the search strategy first. Two review authors independently assessed the full articles of selected trials to confirm eligibility, assess quality and extract data using a data extraction form.

Data extraction and management

We recorded the following features:

Setting;
Country of origin;
Method of randomisation;
Adequacy of allocation concealment;
Details of blinding of participants and outcome assessors;
Inclusion and exclusion criteria used;
Baseline comparability between treatment groups;
Treatments compared and number of participants in each arm;
Outcome data in two‐by‐two tables or change in group means and standard deviations, when appropriate;
Drop‐outs reported and their reasons.

Assessment of risk of bias in included studies

Two review authors independently assessed the full articles of selected trials to assess the risk of bias using the method described in the Cochrane Handbook.

Measures of treatment effect

We expressed the impact of interventions as risk ratios together with 95% confidence intervals. We attempted meta‐analysis only if there were sufficient trials of similar comparisons reporting the same outcomes. risk ratio were combined for binary outcomes.

Unit of analysis issues

We did not encounter any unit of analysis issues.

Dealing with missing data

So far we made no attempt to retrieve missing data; where applicable it is mentioned in the relevant sections.

Assessment of heterogeneity

We originally evaluated statistical heterogeneity between studies by using the Chi² test comparing numbers of participants symptom‐free and considering significant for P values less than 0.10. For this update, we have classified studies secondarily with the I² statistic. The I² statistic describes the percentage of total variation across studies that is due to heterogeneity rather than chance (Higgins 2003). This method does not inherently depend on the number of studies in the meta‐analysis. A value of 0% indicates no observed heterogeneity, and larger values show increasing heterogeneity.

Assessment of reporting biases

If a study protocol described an unpublished or unavailable trial, we assessed whether the prespecified primary and secondary outcomes of interest to this review were predicted in the results in the outcomes section. We did not search for unpublished studies. In one of our comparisons (PPI versus placebo in ENRD) this 2013 update counted 10 studies for the first time. Following Cochrane guidelines this would implicate testing for “small study effects” However, this is a comparison tested to potentially host only substantial heterogeneity until now. All studies are still comparable in size. We considered testing on "small study effects" not adding substantial information.

Data synthesis

We compared the efficacy of PPI, H2RA and prokinetics with placebo and with each other. For each study, we calculated risk ratios and their 95% confidence intervals from the extracted data, and considered the finding statistically significant when the confidence interval did not include one. We calculated the number needed to treat for an additional beneficial outcome by taking the inverse of the absolute risk difference. We conducted all analyses on an intention‐to‐treat (ITT) basis, i.e. including all participants randomised. For the purposes of our formal meta‐analysis, we calculated a pooled estimate of the risk ratios for heartburn relief if appropriate, using a random‐effects model, which provides a more conservative estimate of the overall treatment response by incorporating between‐study heterogeneity.

Subgroup analysis and investigation of heterogeneity

Where we detected significant heterogeneity, we investigated possible explanations and summarised the data using a random‐effects model. We stratified data on an ITT basis, and where possible we performed subgroup analysis of double‐ versus single‐blind conditions, dose, drug class and duration of therapy. We compared change in quality of life in individual studies using group means.

Sensitivity analysis

We performed additional pooling using a fixed‐effect model to test whether point estimates were similar. If we found no difference, we have reported only the random‐effects values, as these are more conservative.

Results

Description of studies

Results of the search

The search strategy defined above generated over 3000 references (Figure 1). After screening titles, abstracts and, if necessary, the full paper, we included 34 trials in our analysis. From 19 trials we extracted data on outcome of empirical treatment for gastro‐oesophageal reflux disease (GORD), and from 19 trials data on outcome of treatment for endoscopy negative reflux disease (ENRD); four trials presented data on both groups (Armstrong 2001; Bate 1997; Galmiche 1997; Venables 1997). Thirty‐one other studies did not present dichotomous outcome measures, or did not match our inclusion criteria, and were subsequently excluded. They are listed in the table Characteristics of excluded studies.

Included studies

Design

All included studies used a double‐blinded and parallel group design. In three of them (Behar 1978; Bright‐Asare 1980; McCallum 1977) the term 'randomisation' was not stated explicitly, although some form of allocation concealment was suggested. The other studies were all described as randomised. In three studies a cross‐over design was used. From two of them (Johansson 1986b; McCallum 1977) we extracted only data from the first treatment period, while the third study (Watson 1997) did not provide these data, so we used data from the full cross‐over study (a carry‐over effect was not anticipated).

Setting

The 34 trials were conducted in North America, Europe, Australia, South Africa, China and Japan. All but five (Bright‐Asare 1980; Johansson 1986b; McCallum 1977; Tan 2011; Watson 1997) were multicentre trials. In six (Armstrong 2005; Carlsson 1998; Hatlebakk 1999; Rush 1995; Talley 2002; Venables 1997), participants were exclusively recruited by primary care physicians. Recruitment in another (Bardhan 1999) was from both primary and secondary care centres. One trial studied patients referred to a regional ambulatory pH monitoring service (Watson 1997), and one included only patients referred for possible anti‐reflux surgery (Johansson 1986b). The other studies provided no details on participant recruitment.

Participants

Empirical Treatment Group

We extracted data from a total of 6734 participants in nineteen trials. The mean number randomised per trial in this group was 354 (range 34 to 994). The mean age of all participants was 51 years (range 18 to 87), with 54% male. One trial included participants with symptoms of heartburn and regurgitation (McCallum 1977), one with two of the following: heartburn, epigastric pain and regurgitation (Hallerback 1998), one with long‐standing symptoms of GORD (Johansson 1986b) and another with symptoms of heartburn, acid eructation or pain on swallowing/dysphagia (Van Zyl 2004).

In all other trials, the primary inclusion criterion was heartburn meeting various criteria concerning severity, frequency and duration. Positive Bernstein testing was additionally required for all participants in two studies (Behar 1978; Bright‐Asare 1980), and only for participants with no signs of oesophagitis in two other studies (Sabesin 1991; Sontag 1987). In one trial reflux had to be demonstrable on x‐ray or oesophagoscopy (McCallum 1977). In seven trials endoscopy was either not performed (Armstrong 2005; Castell 1998; Rush 1995; Talley 2002; Van Zyl 2004) or its findings were not described (Bright‐Asare 1980; McCallum 1977). Participants with circumferential oesophagitis or oesophageal ulcer were excluded from seven trials (Armstrong 2001; Bardhan 1999; Bate 1997; Galmiche 1997; Hallerback 1998; Johnsson 2003; Venables 1997). One trial excluded participants with continuous (but non‐circumferential) mucosal breaks (Hatlebakk 1999). Other common reasons for exclusion were: Barrett's oesophagus, oesophageal stricture, peptic ulcer disease and the recent use of antisecretory drugs.

Endoscopy Negative Group

We extracted data from 6406 participants in nineteen trials. The mean number of participants randomised per trial was 337 (range 19 to 947). The mean age of the participants was 48 years (range 18 to 80), with 41% male. One trial included participants with both heartburn and regurgitation (Riemann 1991), three with either heartburn or regurgitation (Fujiwara 2005; Tan 2011; Watson 1997) and one with heartburn, regurgitation or dysphagia (Schenk 1997). In all other trials, heartburn was the primary inclusion criterion. Additional positive Bernstein testing was required in one study (Robinson 1991), a normal 24‐hour pH study in another (Watson 1997). Participants with any degree of erosive oesophagitis were excluded from all studies in this group. Other common exclusion criteria were: Barrett's oesophagus, oesophageal stricture, peptic ulcer disease and the recent use of antisecretory drugs.

Intervention

Empirical Treatment Group

Ten trials studied a proton pump inhibitor (PPI). This included fourteen treatment arms studying esomeprazole (20 mg twice and 40 mg once daily), omeprazole (10, 20 and 40 mg once daily) and pantoprazole (20 and 40 mg once daily) versus placebo (two studies; two and eight weeks), versus H2‐receptor antagonists (H2RAs) (seven studies; two and four weeks), or versus prokinetics (one study; four weeks).

Fourteen trials studied an H2RA, including fifteen treatment arms: cimetidine (300 and 400 mg four times daily), famotidine (20 mg twice and 40 mg once daily), nizatidine (150 mg twice daily) and ranitidine (150 mg twice and 300 mg once daily) versus placebo (six studies; two, six, eight and twelve weeks), versus PPIs (seven studies; two and four weeks), or versus prokinetics (one study; eight weeks).     Five trials studied a prokinetic agent (five treatment arms): metoclopramide (10 mg four times daily) and cisapride (10 mg four times and 20 mg twice daily) versus placebo (four studies; four and eight weeks), versus PPI (two studies; four and eight weeks), or versus H2RA (one study; eight weeks).

Endoscopy Negative Group

Seventeen trials studied PPIs (28 treatment arms): esomeprazole (20 and 40 mg once daily), omeprazole (10, 20 and 40 mg once daily), lansoprazole (15 and 30 mg once daily), pantoprazole (40 mg once daily), rabeprazole (20 mg once daily) and dexlansoprazole (30 and 60mg once daily) versus placebo (ten trials; two, four and eight weeks), versus H2RAs (five studies; four and eight weeks), or versus prokinetics (one study; four weeks).

Seven trials studied H2RAs (eight treatment arms): cimetidine (200 or 400 mg four times daily), famotidine (20 mg twice or 40 mg once daily), nizatidine (150 mg twice daily) and ranitidine (150 mg twice daily) versus placebo (two studies; two and six weeks), or versus PPIs (five studies; four and eight weeks).

Prokinetics were studied in one trial (one treatment arm), where cisapride (10 mg four times daily) was compared with a PPI (four weeks).

Outcome

Symptomatic outcome measures were used in all trials, since this was one of the inclusion criteria for this review. Data on heartburn outcome were provided in most detail, often expressed in terms of severity and frequency, using measures such as visual analogue scales (VAS), four‐grade Likert or own‐symptom scores. In some studies a distinction was made between outcome for daytime and night time heartburn. Many studies provided limited data on regurgitation, dysphagia and other symptoms. The primary efficacy variable of this review was remission of heartburn, defined as no more than one day with mild heartburn per week. Eleven trials used a quality of life instrument in assessing therapeutic response. The Gastro‐intestinal Symptom Rating Scale (GSRS) was used in seven (Armstrong 2001; Armstrong 2005; Carlsson 1998; Fujiwara 2005; Galmiche 1997; Lind 1997; Talley 2002), the Psychological General Well‐Being index (PGWB index) in two (Galmiche 1997; Lind 1997), the Short‐Form Health Survey (SF‐36) in four (Armstrong 2001; Fujiwara 2005; Rush 1995; Tan 2011) and the heartburn‐specific questionnaire in one (Rush 1995).

Excluded studies

We excluded 31 potentially eligible trials. See the Characteristics of excluded studies table for details.

Risk of bias in included studies

The risk of bias assessments are shown in Figure 2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

Six of the more recent trials (Armstrong 2005; Fass 2009; Katz 2003; Lind 1997; Talley 2002; Tan 2011) were classified as having a low risk of selection bias, indicating adequate allocation concealment. In all the other trials, little or no information was provided on allocation concealment. In those cases we classified the risk of bias as unclear.

Blinding

Performance bias and detection bias were evaluated for the 2013 update. In 17 studies (Armstrong 2001; Armstrong 2005; Bardhan 1999; Bate 1996; Castell 1998; Fass 2009; Galmiche 1997; Hatlebakk 1999; Johnsson 2003: Kahrilas 2005; Katz 2003; Richter 2000b; Rush 1995; Talley 2002; Tan 2011; Van Zyl 2004; Venables 1997) blinding was assessed as being adequate. In one trial (Fujiwara 2005) no blinding was described, so we classified it as having a high risk of bias. None of the other trials provided enough information to classify the risk of bias.

Incomplete outcome data

The quality of data reporting by most trials was poor; ITT analysis was reported in only a few studies, although in most cases data could be re‐analysed on an ITT basis from the data presented. Data could be analysed only per protocol from four studies (Bardhan 1999; McCallum 1977; Riemann 1991; Schenk 1997).

Selective reporting

There was insufficient information to permit judgement of low or high risk of reporting bias in any trials. All are assessed as being at unclear risk. Fass 2009 measured quality of life as a secondary outcome, but did not report the results.

Other potential sources of bias

Galmiche 1997, testing PPI versus prokinetics, used an inadequate dose of omeprazole (10mg) in one treatment arm, which could have decreased the relative efficacy of the PPI. Castell 1998 selected participants by means of a placebo run‐in period, which could have increased the relative efficacy of the cisapride. Johansson 1986a used a cross‐over design.

We found no obvious potential source of bias in the remaining trials, apart from the fact that the use of antacids as rescue medication was allowed in most of the included trials. Since a higher use of rescue medication can be expected, and was observed, in the study group randomised to receive the less effective drug or placebo, the clinical outcome in this group may improve, thereby decreasing the relative efficacy of the more effective drug. This effect may also account for the high healing rate observed in placebo groups from oesophagitis trials (Chiba 1997).

Effects of interventions

See: Table 1; Table 2

Summary of findings for the main comparison. PPIs, H2RAs or prokinetics for heartburn remission in gastro‐oesophageal reflux disease‐like symptoms.

PPIs, H2RAs or prokinetics for heartburn remission in gastro‐oesophageal reflux disease‐like symptoms
Patient or population: patients with heartburn remission in gastro‐oesophageal reflux disease‐like symptoms   Settings:   Intervention: PPIs, H2RAs or prokinetics
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect   (95% CI)	No of Participants   (studies)	Quality of the evidence   (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	PPIs, H2RAs or prokinetics
PPI versus placebo   Symptomatic outcome measures	75 per 100	28 per 100   (24 to 33)	RR 0.37   (0.32 to 0.44)	760   (2 studies)	⊕⊕⊕⊕   high¹
H2RA versus placebo	59 per 100	46 per 100   (36 to 59)	RR 0.77   (0.6 to 0.99)	1013   (2 studies)	⊕⊕⊕⊝   moderate²
Prokinetic versus placebo	See comment	See comment	Not estimable	322   (1 study)	⊕⊕⊕⊕   high
PPI versus H2RA	68 per 100	45 per 100   (41 to 49)	RR 0.66   (0.60 to 0.73)	3147   (7 studies)	⊕⊕⊕⊝   moderate³
PPI versus prokinetic	59 per 100	32 per 100   (19 to 52)	RR 0.53   (0.32 to 0.87)	747   (2 studies)	⊕⊕⊝⊝   low^4,5
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).   CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence   High quality: Further research is very unlikely to change our confidence in the estimate of effect.   Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.   Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.   Very low quality: We are very uncertain about the estimate.

PPIs, H2RAs or prokinetics for for heartburn remission in endoscopy negative reflux disease
Patient or population: patients with heartburn remission in endoscopy negative reflux disease   Settings:   Intervention: PPIs, H2RAs or prokinetics for
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect   (95% CI)	No of Participants   (studies)	Quality of the evidence   (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	PPIs, H2RAs or prokinetics for
PPI versus placebo	87 per 100	62 per 100   (57 to 68)	RR 0.71   (0.65 to 0.78)	3710   (10 studies)	⊕⊕⊕⊝   moderate¹
H2RA versus placebo	78 per 100	66 per 100   (58 to 74)	RR 0.84   (0.74 to 0.95)	514   (2 studies)	⊕⊕⊕⊝   moderate²
PPI versus H2RA	57 per 100	45 per 100   (36 to 56)	RR 0.78   (0.62 to 0.97)	960   (4 studies)	⊕⊕⊝⊝   low^3,4
PPI versus prokinetic	54 per 100	39 per 100   (30 to 50)	RR 0.72   (0.56 to 0.92)	302   (1 study)	⊕⊕⊝⊝   low⁵
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).   CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence   High quality: Further research is very unlikely to change our confidence in the estimate of effect.   Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.   Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.   Very low quality: We are very uncertain about the estimate.

Date	Event	Description
1 June 2012	New search has been performed	Searches rerun and two new studies identified and included.
1 June 2012	New citation required but conclusions have not changed	Two new studies identified and included. Conclusions not changed.

Date	Event	Description
17 September 2010	New citation required but conclusions have not changed	Review is being republished to reflect change in authorship with 2009 update.
5 October 2009	New search has been performed	Updated, 1 study added (Uemura 2008)
30 October 2008	Amended	Converted to new review format.
18 May 2006	New search has been performed	Minor update
12 March 2006	New citation required and conclusions have changed	Conclusions changed
11 March 2006	New citation required and conclusions have changed	Substantive amendment
1 January 2006	New search has been performed	New studies found and included or excluded

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Heartburn remission	12		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 Empirical treatment	2	760	Risk Ratio (M‐H, Random, 95% CI)	0.37 [0.32, 0.44]
1.2 Endoscopy negative reflux disease	10	3710	Risk Ratio (M‐H, Random, 95% CI)	0.71 [0.65, 0.78]
2 Overall improvement	5		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Endoscopy negative reflux disease	5	1231	Risk Ratio (M‐H, Random, 95% CI)	0.62 [0.55, 0.69]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Heartburn remission	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
1.1 Empirical treatment	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
2 Painfree at day	2		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Empirical therapy	2	428	Risk Ratio (M‐H, Random, 95% CI)	0.63 [0.51, 0.77]
3 Painfree at night	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
3.1 Empirical treatment	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
4 Overall improvement	2		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
4.1 Empirical treatment	2	429	Risk Ratio (M‐H, Random, 95% CI)	0.71 [0.56, 0.91]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Painfree at day	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
1.1 Empirical treatment	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]

Methods	Randomised   Double‐blinded   ITT
Participants	n = 220   Heartburn   No grade 4 oesophagitis
Interventions	Four weeks   PAN 40 mg   NIZ 150 mg bid
Outcomes	Complete heartburn relief   Adequate heartburn control   Quality of life
Notes	GORD/ENRD
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Patients were allocated to one of two treatment groups on the basis of a block design randomization list."
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement.
Blinding (performance bias and detection bias)   All outcomes	Low risk	Stated: double‐blinded. Patients received blinded plastic bottles with placebo and identical study medication. Blinding of outcome assessment ensured, and unlikely that the blinding could have been broken.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	"Data analysis for the primary efficacy endpoint was conducted according to the intent‐to‐treat (ITT) principle." Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups: "Seven patients randomized to nizatidine and five randomized to pantoprazole did not have symptom relief data at 28 days."
Selective reporting (reporting bias)	Unclear risk	Unclear
Other bias	Unclear risk	Unclear

Methods	Randomised   Double‐blinded   Per Protocol
Participants	n = 677   Heartburn   No circumferential oesophagitis
Interventions	2 weeks   OME 20 mg   OME 10 mg   RAN 300 mg
Outcomes	Asymptomatic
Notes	GORD
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"The patients were allocated to treatment according to a computer generated randomisation list. At each centre patients were allocated to the next available treatment number."
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding (performance bias and detection bias)   All outcomes	Low risk	Double‐blind, double‐dummy treatment.
Incomplete outcome data (attrition bias)   All outcomes	High risk	No ITT data extractable. Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups: 161 (24%) continued in the study on maintenance treatment, and 197 (29%) discontinued the study at some stage mainly because of unwillingness to continue (21), adverse events (51), and loss to follow‐up (58). There were no differences with respect to these outcomes between the three initially randomised groups.
Selective reporting (reporting bias)	Unclear risk	Unclear
Other bias	Unclear risk	Unclear

Methods	Randomised Double‐blind ITT unclear
Participants	n = 947 Heartburn for at least four of seven days, for at least six months Endoscopy: Normal mucosa, no oesophagitis
Interventions	Four weeks Dexlansoprazole MR 30 mg Dexlansoprazole MR 60 mg Placebo
Outcomes	24‐hour heartburn‐free days Mean severity of heartburn Sustained resolution Days without rescue medication
Notes	ENRD
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Patients were randomised using Interactive Voice Responce System in a 1:1:1 ratio.
Allocation concealment (selection bias)	Low risk	Patients were randomised using Interactive Voice Responce System in a 1:1:1 ratio.
Blinding (performance bias and detection bias)   All outcomes	Low risk	Adequate blinding described
Incomplete outcome data (attrition bias)   All outcomes	High risk	ITT analysis described, but unclear whether it was performed. Results are described in percentages.
Selective reporting (reporting bias)	Unclear risk	Unclear. Quality of life was measured, but not reported
Other bias	Unclear risk	Unclear

Methods	Randomised   ITT
Participants	n = 106
Interventions	Four weeks   FAM 20 mg bid   OME 20 mg
Outcomes	Complete relief   Partial relief   GSRS   SF‐36
Notes	ENRD
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'Randomized' is stated, but Insufficient information about the sequence generation process to permit judgement.
Allocation concealment (selection bias)	Unclear risk	Method of concealment is not described.
Blinding (performance bias and detection bias)   All outcomes	High risk	No blinding described.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups.
Selective reporting (reporting bias)	Unclear risk	Unclear
Other bias	Unclear risk	Unclear

Methods	Randomised   Double‐blinded   ITT(m)
Participants	n = 423   Two of the following: heartburn, epigastric pain, regurgitation   No oesophageal ulcer
Interventions	Two weeks   RAN 150 mg bid   Placebo
Outcomes	Complete symptom relief   Improvement
Notes	GORD
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Sequence generation process not described.
Allocation concealment (selection bias)	Unclear risk	Method of concealment not described.
Blinding (performance bias and detection bias)   All outcomes	Unclear risk	Double‐blinded was stated, but method was not described.
Incomplete outcome data (attrition bias)   All outcomes	High risk	After randomisation 4 of 427 withdrew and were not analysed or accounted for
Selective reporting (reporting bias)	Unclear risk	Unclear
Other bias	Unclear risk	Unclear

Methods	Randomised   Double‐blinded   Cross‐over   ITT
Participants	n = 42   GORD symptoms
Interventions	Eight weeks   RAN 150 mg bid   Placebo
Outcomes	Improvement
Notes	GORD
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Sequence generation process not described.
Allocation concealment (selection bias)	Unclear risk	Method of concealment allocation not described.
Blinding (performance bias and detection bias)   All outcomes	Unclear risk	Double‐blinded was stated, but the method was not described.
Incomplete outcome data (attrition bias)   All outcomes	High risk	No drop‐outs and no missing data reported.
Selective reporting (reporting bias)	Unclear risk	Unclear
Other bias	High risk	Cross‐over study

Methods	Randomised   Double‐blinded   APT
Participants	n = 509   Heartburn   No oesophagitis
Interventions	Four weeks   OME 20 mg   OME 10 mg   Placebo
Outcomes	Heartburn‐free   Max one day per week mild heartburn   Heartburn control   Quality of life
Notes	ENRD
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"The randomization was performed by computer."
Allocation concealment (selection bias)	Low risk	"The randomization was performed by computer and was carried out separately for each centre and within blocks of five consecutive patient numbers. The block size was not disclosed to the investigators until the study was completed."
Blinding (performance bias and detection bias)   All outcomes	Unclear risk	"Double‐blind treatment" stated, unclear how it was performed, insufficient information to permit judgement.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups. "The effect of treatment on heartburn after 4 weeks was analysed with the All Patients Treated (APT) method, in which all patients randomized to treatment were included."
Selective reporting (reporting bias)	Unclear risk	Unclear
Other bias	Unclear risk	Unclear

Methods	Randomised Double‐blinded Per protocol
Participants	n= 175 Heartburn or regurgitation Endoscopy: no erosive oesophagitis
Interventions	Eight weeks ESO 20 mg Placebo
Outcomes	Complete resolution of symptoms Mild symptoms (acceptable) GerdQ‐Score SF‐36
Notes	ENRD
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Adequate randomisation described
Allocation concealment (selection bias)	Low risk	"Randomization was performed by drawing a sealed envelope that contained a pre‐assigned randomized code generated by computer on entry to the study. Both the investigators and patients were blinded to the assigned treatment throughout the study."
Blinding (performance bias and detection bias)   All outcomes	Low risk	"Both the investigators and patients were blinded to the assigned treatment throughout the study.The esomeprazole and placebo tablets were identical in appearance to maintain blinding, and only the study investigators had access to the code key, stored on a word file in a locked computer, which decoded the patient’s treatment. The patient’s randomized treatment was not recorded in the patient’s file."
Incomplete outcome data (attrition bias)   All outcomes	High risk	Per Protocol analysis, drop‐out > 10%.
Selective reporting (reporting bias)	Unclear risk	Unclear
Other bias	Unclear risk	Unclear

Methods	Randomised Double‐blinded ITT
Participants	n = 284 Heartburn Grade M or N oesophagus according to Hoshihara's modified version of the LA classification (= no erosive changes)
Interventions	Four weeks OME 10 mg OME 20 mg Placebo
Outcomes	Complete resolution of heartburn
Notes	ENRD
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information about the sequence generation process to permit judgement.
Allocation concealment (selection bias)	Unclear risk	Method of concealment not described.
Blinding (performance bias and detection bias)   All outcomes	Unclear risk	Insufficient information to permit judgement.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups. The proportion of missing outcomes compared with observed event risk not enough to have a clinically relevant impact on the intervention effect estimate.
Selective reporting (reporting bias)	Unclear risk	Unclear
Other bias	Unclear risk	Unclear

Study	Reason for exclusion
Bennett 1983	No dichotomous outcome variable extractable.
Croxtall 2010	Design not RCT, one RCT written about is included.
Fass 2010	Patients with sleep disturbances.
Fiasse 1980	No dichotomous outcome variable extractable.
Greaney 1981	No dichotomous outcome variable extractable.
Hansen 2006	Open design and only long‐term outcome.
Haruma 2009	Patients with dyspepsia.
Herwaarden 1999	No dichotomous outcome variable extractable.
Hongo 2008a	Only selected group of patients with non‐erosive reflux disease grade M included.
Hongo 2008b	Only selected group of patients with non‐erosive reflux disease grade M included.
Jebbink 1993	No dichotomous outcome variable extractable.
Johnson 2010	Only selected group of patients (with a history of GORD‐related sleep disturbances).
Johnsson 1993	No dichotomous outcome variable extractable.
Johnsson 1998	No dichotomous outcome variable extractable.
Juul‐Hansen 2008	Patients were excluded if nonresponsive on lansoprazole.
Kinoshita 2010	Only selected group of ENRD patients (only grade M) with minimal changes and only patients not responding to antacids during the observation period were included.
Kushner 2009	Only selected group of patients: Only included if responsive to antacids, H2RAs or PPI, excluded if a history of GORD confirmed by endoscopy or if a history of pathologic pH monitoring.
Maddern 1986	No dichotomous outcome variable extractable.
Marrero 1994	No dichotomous outcome variable extractable.
Masharova 2011	No control group treated with placebo or H2RAs. In treatment group PPI dose was increased if needed.
Meineche 2004	Outcome not stratified, no outcome for patients with heartburn available.
Nakamura 2010	Design not blinded.
Powell‐Jackson 1978	No dichotomous outcome variable extractable.
Richter 1996	Only patients refractory to ranitidine treatment included.
Richter 1999	No dichotomous outcome variable extractable.
Robinson 1987	No dichotomous outcome variable extractable.
Scheiman 2009	Lack of information: only abstract available.
Schindlbeck 1995	No dichotomous outcome variable extractable.
Talley 2007	Patients with heartburn were excluded.
Van Zanten 2005	Patients with heartburn as exclusive symptom were excluded.
Van Zanten 2007	Patients with predominant symptoms of heartburn were excluded.

PERMALINK

Short‐term treatment with proton pump inhibitors, H2‐receptor antagonists and prokinetics for gastro‐oesophageal reflux disease‐like symptoms and endoscopy negative reflux disease

Kirsten E Sigterman

Bart van Pinxteren

Peter A Bonis

Joseph Lau

Mattijs E Numans

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Results

Description of studies

Results of the search

1.

Included studies

Design

Setting

Participants

Empirical Treatment Group

Endoscopy Negative Group

Intervention

Empirical Treatment Group

Endoscopy Negative Group

Outcome

Excluded studies

Risk of bias in included studies

2.

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Summary of findings for the main comparison. PPIs, H2RAs or prokinetics for heartburn remission in gastro‐oesophageal reflux disease‐like symptoms.

Summary of findings 2. PPIs, H2RAs or prokinetics for heartburn remission in endoscopy negative reflux disease.

Empirical treatment for gastro‐oesophageal reflux disease (GORD)

Heartburn remission

1.1. Analysis.

3.

2.1. Analysis.

4.

3.1. Analysis.

4.1. Analysis.

5.