Drug therapies for reducing gastric acidity in people with cystic fibrosis

Sze May Ng; Helen S Moore

doi:10.1002/14651858.CD003424.pub4

. 2016 Aug 22;2016(8):CD003424. doi: 10.1002/14651858.CD003424.pub4

Drug therapies for reducing gastric acidity in people with cystic fibrosis

Sze May Ng ^1,^✉, Helen S Moore ¹

Editor: Cochrane Cystic Fibrosis and Genetic Disorders Group

PMCID: PMC7170416 PMID: 27546383

Abstract

Background

Malabsorption of fat and protein contributes to poor nutritional status in people with cystic fibrosis. Impaired pancreatic function may also result in increased gastric acidity, leading in turn to heartburn, peptic ulcers and the impairment of oral pancreatic enzyme replacement therapy. The administration of gastric acid‐reducing agents has been used as an adjunct to pancreatic enzyme therapy to improve absorption of fat and gastro‐intestinal symptoms in people with cystic fibrosis. It is important to establish the evidence regarding potential benefits of drugs that reduce gastric acidity in people with cystic fibrosis. This is an update of a previously published review.

Objectives

To assess the effect of drug therapies for reducing gastric acidity for: nutritional status; symptoms associated with increased gastric acidity; fat absorption; lung function; quality of life and survival; and to determine if any adverse effects are associated with their use.

Search methods

We searched the Cochrane Cystic Fibrosis and Genetic Disorders Group Trials Register which comprises references identified from comprehensive electronic database searches, handsearches of relevant journals, abstract books and conference proceedings.

Most recent search of the Group's Trials Register: 12 May 2016.

Selection criteria

All randomised and quasi‐randomised trials involving agents that reduce gastric acidity compared to placebo or a comparator treatment.

Data collection and analysis

Both authors independently selected trials, assessed trial quality and extracted data.

Main results

The searches identified 39 trials; 17 of these, with 273 participants, were suitable for inclusion, but the number of trials assessing each of the different agents was small. Seven trials were limited to children and four trials enrolled only adults. Meta‐analysis was not performed, 14 trials were of a cross‐over design and we did not have the appropriate information to conduct comprehensive meta‐analyses. All the trials were run in single centres and duration ranged from five days to six months. The included trials were generally not reported adequately enough to allow judgements on risk of bias.

However, one trial found that drug therapies that reduce gastric acidity improved gastro‐intestinal symptoms such as abdominal pain; seven trials reported significant improvement in measures of fat malabsorption; and two trials reported no significant improvement in nutritional status. Only one trial reported measures of respiratory function and one trial reported an adverse effect with prostaglandin E2 analogue misoprostol. No trials have been identified assessing the effectiveness of these agents in improving quality of life, the complications of increased gastric acidity, or survival.

Authors' conclusions

Trials have shown limited evidence that agents that reduce gastric acidity are associated with improvement in gastro‐intestinal symptoms and fat absorption. Currently, there is insufficient evidence to indicate whether there is an improvement in nutritional status, lung function, quality of life, or survival. Furthermore, due to the unclear risks of bias in the included trials, we are unable to make firm conclusions based on the evidence reported therein. We therefore recommend that large, multicentre, randomised controlled clinical trials are undertaken to evaluate these interventions.

Plain language summary

Drugs for reducing stomach acid for people with cystic fibrosis

Review question

We reviewed the evidence for using drugs to reduce stomach acid in people with cystic fibrosis.

Background

Cystic fibrosis causes damage to the lungs and the pancreas. The pancreas produces enzymes which are needed for the body to digest and absorb food. If the pancreas is damaged this can affect how people can absorb food. It can also increase acidity in the stomach, which may lead to heartburn and peptic ulcers. There are drugs that can reduce the amount of acid in the stomach. Trials of these drugs have shown that they can improve problems in the stomach and digestive system and in the absorption of fat. This is an updated version of the review.

Search date

The evidence is current to: 12 May 2016.

Study characteristics

The review included at 17 trials with a total of 273 children and adults. Seven of the trials were limited to children and four trials enrolled only adults, while the remainder enrolled people of any age. All the trials were run in single centres and lasted from five days to six months.

Most trials compared an intervention to placebo (a dummy treatment with no active medication). Six trials compared proton pump inhibitors (drugs which reduce the amount of acid made in the stomach, e.g. omeprazole and esomeprazole) to placebo and seven trials compared a H2 receptor antagonist (a second group of medicines that reduce the amount of acid made in the stomach, e.g. ranitidine and cimetidine) to placebo. One of the trials had three arms and compared a proton pump inhibitor to both a H2 receptor agonist and a placebo. In the remaining five trials, one compared pancrelipase (a combination of three enzymes (lipase, protease, and amylase) normally produced by the pancreas) to a combination of pancrelipase and misoprostol (a drug which protects the lining of the gut from stomach acid); one compared misoprostol to placebo and one trial compared enprostil (similar drug to misoprostol) to ranitidine. Two trials used sodium bicarbonate ‐ one compared to placebo and the second compared to calcium carbonate.

Key results

We were not able to combine the results from these trials due to their design. Individual trials reported some improvements in abdominal pain and fat absorption. However, the trials did not report improvements for other outcomes such as lung function, quality of life, or survival. The different drugs studied caused some adverse events; mainly diarrhoea (two people withdrew from one trial because of this) and bloating due to wind. As we could not combine the results from these trials, we were not able to reach firm conclusions about whether people with cystic fibrosis would benefit from taking these drugs. New long‐term trials are needed to examine the benefits and possible adverse effects for people with cystic fibrosis taking drugs to reduce stomach acid.

Quality of the evidence

There were 14 trials which had a cross‐over design (where people taking part are given first one treatment and then another treatment) and we did not have the appropriate information to analyse the results properly. Few of the included trials reported clearly on aspects of trial quality or gave enough information to allow us to judge whether any factors might cause a potential risk of bias to the results.

Background

Description of the condition

Cystic fibrosis (CF) is an autosomal recessive inherited disease that is most common in Caucasian populations having an estimated incidence of 1 in 2500 live births (CF Foundation 1994). It affects the epithelial chloride channels. Defects cause reduced epithelial salt and water secretion, leading to viscous mucus in various organs such as the lungs and pancreas. As a result, exocrine pancreatic insufficiency occurs in the majority of people with CF causing maldigestion of fat and protein. Malabsorption of fat and protein lead to considerable energy losses in the stools and can contribute to poor nutritional status in people with CF (Murphy 1991). Malnutrition affects growth, increases the severity of pulmonary disease, and ultimately plays an important role in the shortening of lifespan in people with CF (Kraemer 1978). Furthermore, nutritional status is also related to lung function and exercise performance in people with CF (Neijens 1985).

Absorption in the small intestine may also be affected by reduced bicarbonate secretion in the small bowel and pancreas, contributing to a lower duodenal pH (Weber 1984). This may impair and irreversibly inactivate pancreatic enzymes, which in turn may lead to increased gastric acidity resulting in heartburn, epigastric pain and gastric or duodenal ulcers.

Description of the intervention

Oral pancreatic enzyme therapy to improve the nutritional status and decrease maldigestion due to pancreatic insufficiency in people with CF has been available for several decades. Despite adequate pancreatic replacement therapy, many people with CF continue to excrete large amounts of fat and protein in their stools, which may contribute to malnutrition and weight loss.

How the intervention might work

Orally administered pancreatic enzymes may be inactivated by gastric acid in people with CF with pancreatic insufficiency leading to fat and protein malabsorption (Zentler‐Munro 1985). The decrease in pancreatic bicarbonate secretion associated with CF impairs the release of enteric‐coated pancreatic enzymes and reduces their effectiveness (Regan 1979). Studies have suggested that drug therapy which reduces gastric acid may improve the effectiveness of pancreatic replacement therapy (DiMagno 2001). Agents to reduce gastric acidity should provide a duodenal environment more conducive to efficient enzyme function. In adults with pancreatic dysfunction secondary to chronic pancreatitis, H2 receptor antagonists or proton pumps that reduce gastric acidity improved fat malabsorption (Bruno 1994). Proton pump inhibitors and H2 receptor antagonists do this directly by inhibiting acid production from parietal cells. Synthetic prostaglandins can also be used to reduce gastric acid secretion and stimulate bicarbonate secretion from the small bowel.

Why it is important to do this review

Improvements in fat absorption and overall nutrition are important in the treatment of people with CF. The major goals in the treatment of people with CF are to improve the quality of life, achieve normal growth, improve respiratory status and increase life expectancy. It is therefore important to establish whether adjunct drug therapy to reduce gastric acidity in people with CF who are being treated with pancreatic enzymes is beneficial.

This version of the review is an update of previous versions (Ng 2003; Ng 2012; Ng 2014).

Objectives

To test the hypotheses that in people with CF, agents that reduce gastric acidity:

improve nutritional status, as assessed by weight, height and other indices of growth;
improve symptoms associated with increased gastric acidity such as heartburn and epigastric pain;
improve lung function, quality of life and survival;
do not have unacceptable adverse effects.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs) or quasi‐randomised controlled trials (published and unpublished).

Types of participants

Children and adults with defined CF, diagnosed clinically and by sweat or gene testing including all ages and all degrees of severity.

Types of interventions

Agents that reduce gastric acidity compared to placebo or a comparator treatment. The major drug groups are:

proton pump inhibitors;
H2 receptor antagonists.

Other drug therapies such as prostaglandin E2 analogues and sodium bicarbonate which reduce gastric acidity will also be considered. All doses and routes of administration will be considered.

Types of outcome measures

Primary outcomes

Measures of nutritional status as assessed by weight, height and other indices of growth
Symptoms related to increased gastric acidity such as epigastric pain, heartburn
Complications of increased gastric acidity such as gastric or duodenal ulcers

Secondary outcomes

Faecal fat, faecal nitrogen excretion and other measures of fat malabsorption
Measures of lung function
Measures of quality of life
Mortality
Any adverse effects reported

Search methods for identification of studies

Relevant trials were identified from the Group's Cystic Fibrosis Trials Register using the term: antacids.

The Cystic Fibrosis Trials Register is compiled from electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL) (updated each new issue of The Cochrane Library), weekly searches of MEDLINE, a search of Embase to 1995 and the prospective handsearching of two journals ‐ Pediatric Pulmonology and theJournal of Cystic Fibrosis. Unpublished work is identified by searching the abstract books of three major cystic fibrosis conferences: the International Cystic Fibrosis Conference; the European Cystic Fibrosis Conference and the North American Cystic Fibrosis Conference. For full details of all searching activities for the register, please see the relevant sections of the Cystic Fibrosis and Genetic Disorders Group Module.

Date of the most recent search of the Group's Cystic Fibrosis Trials Register: 12 May 2016.

Data collection and analysis

Selection of studies

Two authors independently selected the trials to be included in the review. Disagreement did not arise on the suitability of any trials for inclusion in the review or their quality. However, if this should occur during future updates of this review, we will reach consensus by discussion.

Data extraction and management

The authors independently extracted data using standard data acquisition forms. Again, disagreement did not arise between the authors. We will resolve any future disagreements on any extracted data by discussion.

All drugs which reduce gastric acidity were regarded as the intervention to be compared to the placebo and not split into individual drug comparisons.

We planned to group outcome data into those measured at 1, 3, 6 and 12 months and annually thereafter. However, we were unable to group data into the time points we had originally planned and so report results narratively from the end of each trial. The duration of each trial is detailed in the table 'Characteristics of included studies'. For future updates of this review, if outcome data are recorded at other time periods, then consideration will be given to examining these as well.

Assessment of risk of bias in included studies

In order to establish a risk of bias for each included study, each author assessed the methodological quality of each trial using criteria suggested by Jüni (Jüni 2001). In particular, authors examined details of the generation of allocation sequence, the concealment of treatment allocation schedule, whether the trial was blinded, whether intention‐to‐treat analyses were possible from available data and if the number lost to follow up or subsequently excluded from the trial was recorded. In future updates of the review, we will resolve any disagreements that occur by discussion.

Measures of treatment effect

For future updates of the review, if appropriate data are available, the authors will use the following methods to analyse binary and continuous outcome measures. For binary outcome measures, the authors plan to calculate a pooled estimate of the treatment effect for each outcome across the studies, (the odds of an outcome among treatment allocated participants to corresponding odds among controls). For continuous outcomes, the authors will record either a mean change from baseline for each group or mean post‐treatment or intervention values and standard deviation (SD) for each group. They will calculate a pooled estimate of treatment effect by calculating the mean difference where appropriate.

Unit of analysis issues

Ideally, when conducting a meta‐analysis combining results from cross‐over trials, the authors would have liked to use the methods that are recommended by Elbourne and Curtin (Curtin 2002; Elbourne 2002). However, due to restrictions on the data that were available from the papers the only method that the authors were able to use was to treat the cross‐over trials as if they were parallel trials. When trials did not report data in such a way that a correct analysis could be performed, the results were described narratively within the text rather than combine the results using inappropriate techniques. The primary authors of the cross‐over trials have been approached for further individual patient data and order of treatment, but no data have yet been made available.

Dealing with missing data

In order to allow an intention‐to‐treat analysis, the authors will seek data on the number of participants with each outcome event, by allocated treated group, irrespective of compliance and whether or not the participant was later thought to be ineligible or otherwise excluded from treatment or follow up.

Assessment of heterogeneity

Due to the lack of trials it was not possible to investigate heterogeneity between trial results using the standard Chi² test and I² statistic (Higgins 2003). For future updates of the review we will test for heterogeneity using the Chi² test and I² statistic. We will consider the ranges of I² to relate to the degree of heterogeneity as follows:

0% to 40%: might not be important;
30% to 60%: may represent moderate heterogeneity;
50% to 90%: may represent substantial heterogeneity;
75% to 100%: considerable heterogeneity.

Assessment of reporting biases

We examined the publications for potential reporting biases but were not able to compare the original trial protocols with the final publications. We were therefore unable to identify any potential reporting biases.

Data synthesis

Where between trial variability is not statistically significant we will carry out a fixed effect analysis and if the between trial heterogeneity is statistically significant we will perform a random effects analysis.

Subgroup analysis and investigation of heterogeneity

If there is significant heterogeneity between the trials, the authors plan to perform a subgroup analysis investigating different classes of drugs.

Sensitivity analysis

The authors also plan to perform a sensitivity analysis based on methodological quality of the trials including and excluding quasi‐randomised trials, when it is appropriate to do so.

Results

Description of studies

Results of the search

A total of 39 trials were identified from the searches; 17 trials were included and 22 trials were excluded.

Included studies

There was a large variation between the trials identified in terms of design, intervention, duration of treatment and outcome measures. SeeCharacteristics of included studies table for full details.

We included 17 trials in this review. All of the included trials were published, but four of them were published only as abstracts in conference proceedings (Bowler 1993; Chung 2000; Lubin 1979; Weber 1981).

Trial design

Fourteen trials were cross‐over in design (Bowler 1993; Boyle 1980; Carroccio 1992; Chalmers 1985; Chung 2000; Durie 1980; Francisco 2002; Heijerman 1990; Heijerman 1991; Heijerman 1993; Proesmans 2003; Robinson 1988; Robinson 1990; Weber 1981). Three trials were of a parallel design (Lubin 1979; Schoni 1984; DiMango 2014).

Trial duration ranged from 12 days (three days in each of the four treatment arms in this cross‐over trial) (Francisco 2002) to one year (again in cross‐over trials with treatment periods of six months in each arm) (Carroccio 1992; Chalmers 1985); duration was not stated in one trial (Lubin 1979).

Participants

The included trials reported results from a total of 273 participants. Sample size varied from five (Chung 2000) to 38 participants (Schoni 1984), with participants' ages ranging from six months (Robinson 1990) to 42 years (Heijerman 1993). Seven trials were limited to children (Bowler 1993; Chung 2000; Lubin 1979; Robinson 1988; Robinson 1990; Schoni 1984; Weber 1981); and three trials enrolled only adults (Heijerman 1990; Heijerman 1991; Heijerman 1993).

In one trial participants were grouped according to acetylation phenotype (slow or fast) and matched according to their age before the trial (Schoni 1984).

Interventions

Most trials compared an intervention to placebo.

Six trials compared proton pump inhibitors with placebo (Chung 2000; DiMango 2014; Francisco 2002; Heijerman 1991; Heijerman 1993; Proesmans 2003). Participants in two trials were randomised to receive either omeprazole or placebo; but in the Francisco trial this was only the case for the adult participants (Chung 2000; Francisco 2002). In a third trial, participants received either esomeprazole 40 mg or placebo twice daily (DiMango 2014). In the later Heijerman trial, participants were given either pancrealipase plus omeprazole or pancrealipase and placebo (Heijerman 1993). In a further trial, one group of participants received omeprazole and the other treatment group received no omeprazole (Proesmans 2003). In a cross‐over trial consisting of four comparative treatment periods participants received: firstly a high dose of pancreatic enzyme, pancrelipase (Pancrease®) (four capsules, three times daily) in conjunction with a proton pump inhibitor, omeprazole; secondly, high‐dose pancrelipase in conjunction with placebo; thirdly, low‐dose pancrelipase (two capsules, three times daily) in conjunction with omeprazole; and fourthly, low‐dose pancrelipase in conjunction with placebo (Heijerman 1991).

Seven trials compared a H2 receptor antagonist with placebo (Bowler 1993; Boyle 1980; Carroccio 1992; Chalmers 1985; Durie 1980; Francisco 2002; Schoni 1984). In one trial, participants received one of three treatments in a randomly‐assigned sequence: high‐dose pancreatic enzyme (Creon®) with the H2 receptor antagonist, ranitidine; high‐dose pancreatic enzyme with placebo; or lower‐dose pancreatic enzyme with placebo (Bowler 1993). In another trial, participants received pancreatic enzyme alone (Viokase®) or pancreatic enzyme (Viokase®) and the H2 receptor antagonist, cimetidine (Boyle 1980). Carroccio investigated famotidine or placebo given in addition to normal enzyme therapy (Carroccio 1992). In the Chalmers trial, participants received either cimetidine or placebo in addition to normal enzyme therapy (Chalmers 1985). The trial by Durie contained four treatment groups, who received each treatment in a random order; the four interventions that were used were: pancrelipase; sodium bicarbonate; cimetidine; and a combination of sodium bicarbonate and cimetidine (Durie 1980). As part of a trial, already mentioned in the previous paragraph, participants who were pancreatic insufficient were randomised to receive high‐dose or low‐dose ranitidine compared with placebo in a cross‐over trial (Francisco 2002). In the trial by Schoni, the effectiveness of cimetidine compared to placebo was evaluated (Schoni 1984).

One trial compared the normal dose pancrelipase and prostaglandin E2 analogue misoprostol (100 μg every six hours) with the normal dose pancrease alone (Robinson 1988), a further trial compared the prostaglandin E2 analogue misoprostol with placebo (Robinson 1990). One trial compared sodium bicarbonate with placebo (Weber 1981).

A further trial compared two active treatment arms, the H2 receptor antagonist (ranitidine) with prostaglandin E2 analogue (enprostil) (Heijerman 1990). One trial compared sodium bicarbonate with calcium carbonate (Lubin 1979).

Outcomes

Data for measuring fat malabsorption were presented in different ways such as percentage fat malabsorption (three‐day fat excretion/three times mean daily fat intake), faecal fat excretion (percentage daily fat intake), percentage fat absorption, faecal weight (g/24 h), and faecal fat (g/24 h). One trial used a gastro‐intestinal symptom score which incorporated a subjective measure for the frequency and severity of symptoms (Heijerman 1990). Gastro‐intestinal complaints were each scored: none = 0; little = 1; moderate = 2; severe = 3; and very severe = 4. These scores were then combined to give an overall complaints score which was analysed as a continuous variable. Another trial relied on participants reporting relief from chronic abdominal pain (Robinson 1988).

Excluded studies

In total, 22 trials were excluded for a variety of reasons. Six trials were not RCTs (Cameron 1982; Cox 1981; Dudley 1981; Lloyd‐Still 1992; Miller 1985; Miller 1986); seven trials examined agents which are not used to reduce gastric acidity (Ansaldi‐Balocco 1988; Geus 1999; Hoffman 1987; Kerr 1982; Mitchell 1981; Stead 1988; Tsang 1994); five trials compared cisapride (a pro‐kinetic agent) with a drug for reducing gastric acidity (Cucchiara 1996; Koletzo 1989; Prinsen 1985; Santamaria 1989; Smith 1988); one trial compared the accuracy of fat balance trials at home versus at a clinic (Francisco 1996); one trial contained four treatment periods, three of which were not relevant to the review, so no comparison could be made (Gow 1981); one trial compared the pharmacokinetics of oral versus intravenous famotidine (Maish 1998); and one trial assessed lipase, bile acid and trypsin concentration (Zentler‐Munro 1985).

Risk of bias in included studies

In order to assess the risk of bias in the included trials, the authors assessed the methodological quality of the included trials using criteria suggested by Jüni for the following dimensions: concealment of allocation; generation of the randomisation sequence; intention‐to‐treat; and level of blinding reported (Jüni 2001). The dimensions, concealment of allocation, generation of the randomisation sequence and intention‐to‐treat were categorised as adequate, unclear or inadequate which related to a low, unclear or high risk of bias respectively; RCTs were categorised according to whether double blinding had been reported or not. There is a decreasing risk of bias to the results when more people are blinded to an intervention.

Allocation

All trials stated that allocation was randomised, but no trials described the method of randomisation used. We therefore judged the risk of bias due to the generation of the randomisation sequence as unclear in all trials.

Concealment of allocation was adequate in only one trial in which the randomisation procedure was carried out at the pharmacy in the hospital (Robinson 1990). The trial organisers were unaware of this sequence until the completion of the trial. We judged this trial to have a low risk of bias. However, concealment of allocation, and also the risk of bias, was unclear in the remaining 16 trials (Bowler 1993; Boyle 1980; Carroccio 1992; Chalmers 1985; Chung 2000; DiMango 2014; Durie 1980; Francisco 2002; Heijerman 1990; Heijerman 1991; Heijerman 1993; Lubin 1979; Proesmans 2003; Robinson 1988; Schoni 1984; Weber 1981).

Blinding

Double‐blinding was reported in 13 of the 17 included trials; however, two trials did not report blinding of investigators (Boyle 1980; Lubin 1979). In one trial, Robinson reported that parents of 5 out of 15 children in the treatment group were able to correctly identify the period in which misoprostol was administered by improvements in symptoms such as abdominal pain and offensive bulky bowel motions (Robinson 1990). In two trials the level of blinding was not discussed (DiMango 2014; Durie 1980) and a further trial was not blinded (Proesmans 2003).

We considered the risk of bias from blinding to decrease if more people were blinded to the intervention and the risk of bias to be unclear if this aspect of trial quality was not discussed.

Incomplete outcome data

No trials explicitly stated that an intention‐to‐treat analysis was performed. Neither did any trial state that any participants deviated from the randomised group to which they were assigned. In six trials there were no participant withdrawals reported, but it was not explicitly stated that no participants withdrew (DiMango 2014; Heijerman 1991; Heijerman 1993; Lubin 1979; Schoni 1984; Weber 1981). Therefore we judged the risk of bias to be unclear in these four trials. The remaining 11 included trials experienced participant withdrawals (Bowler 1993; Boyle 1980; Carroccio 1992; Chalmers 1985; Chung 2000; Durie 1980; Francisco 2002; Heijerman 1990; Proesmans 2003; Robinson 1988; Robinson 1990). In 10 trials there were a low number of participant withdrawals and these were described, but in the remaining trial there was a discrepancy between the number of participants entering and completing the trial which was not addressed (Bowler 1993). We judged the risk of bias to be low in the 10 trials which accounted for participant withdrawals, but high in the trial which did not give any reason for the missing data from one participant.

Selective reporting

We were unable to identify any selective reporting in the included trials, but did not have any access to the original trial protocols to definitely confirm this; we therefore conclude that there is an unclear risk of bias due to selective reporting.

Other potential sources of bias

Durie states that the participants were selected from a restricted geographical area in proximity to the hospital on the basis of an assessment by clinical personnel of their ability to perform a clinical study accurately at home (Durie 1980). This may present a potential risk of bias due to pre‐selection of participants by clinic personnel.

Effects of interventions

We were unable to group data into the time points we had originally planned and so report results narratively from the end of each trial. The duration of each trial is detailed in the table Characteristics of included studies.

Drugs for reducing gastric acidity versus placebo

Primary outcomes

1. Measures of nutritional status

Data on growth indices and nutritional status were available from two trials comparing an H2 receptor antagonist with a placebo (Chalmers 1985; Schoni 1984). We were unable to pool data from the trial by Chalmers because there were insufficient data to calculate the SD (Chalmers 1985). Both trials reported that there were no significant improvements in height, weight and skinfold thickness between the treatment and the control group (Chalmers 1985; Schoni 1984). Body mass index and z scores for weight and height were not reported in either trial.

2. Measures of gastro‐intestinal symptoms

These measures were recorded in nine trials (Bowler 1993; Boyle 1980; Carroccio 1992; Chalmers 1985; Heijerman 1991; Heijerman 1993; Robinson 1988; Robinson 1990; Weber 1981), but not reported in eight of the trials.

In the 1988 trial, Robinson looked at relief from chronic abdominal pain in the treatment group compared with the control group (Robinson 1988). There were six participants who reported relief in the treatment group and none in the control group. The symptoms returned in five of the six participants after misoprostol was stopped at the end of the trial period.

3. Complications of increased gastric acidity such as gastric or duodenal ulcers

These outcomes were either not reported or the data were not available in any of the trial reports.

Secondary outcomes

1. Measures of fat malabsorption

In the 1993 trial, Heijerman reported the results for faecal fat excretion and found that there was no statistically significant difference between the treatment and control groups (Heijerman 1993). However, previously in his 1991 trial, Heijerman found significantly lower faecal fat excretion (percentage of daily fat intake) (P < 0.01) when omeprazole was combined with higher dose pancrelipase enzymes (four capsules, three times daily) (Heijerman 1991). Improvement in faecal fat excretion when omeprazole was added to lower dose pancrelipase (two capsules, three times daily) occurred in seven of the nine adults studied, but was not significant for the whole group. Proesmans reported that the effect of omeprazole on daily faecal fat loss (g/day) was statistically significant (P < 0.01) (Proesmans 2003).

The 1988 Robinson trial found a significant reduction of fat malabsorption in the treatment group (P < 0.02), only the P value was available from this trial (Robinson 1988). The later Robinson trial reported that there was a significant reduction in fat malabsorption while taking misoprostol (P < 0.01) with no change in daily fat intake (Robinson 1990).

Boyle showed that cimetidine significantly reduced the mean stool weight (P < 0.005) (Boyle 1980). Carroccio also showed that famotidine significantly reduced faecal weight (P < 0.0001) (Carroccio 1992). Weber showed no significant difference between sodium bicarbonate and placebo (Weber 1981).

No significant improvement was found for faecal fat (g/day) between treatment and placebo groups in the Weber trial (Weber 1981). Chalmers reported a significant reduction in faecal fat in the treatment group (P < 0.05), but no statistical difference was found for mean faecal wet weight (g/day) (Chalmers 1985). A trial by Bowler showed no significant difference in faecal fat or faecal weight when comparing both high lipase enzyme and with the addition of ranitidine to control (Bowler 1993). A trial by Boyle comparing cimetidine with placebo reported a significant difference in faecal fat (g/day) (P < 0.05) (Boyle 1980). A trial by Francisco reported no significant difference in fat absorption between high‐dose or low‐dose ranitidine compared with placebo groups in children or between omeprazole and placebo groups in the adult participants (Francisco 2002). Chung also reported no significant difference in malabsorption between high‐dose omeprazole or placebo groups (Chung 2000).

Boyle reported a significant improvement in fat absorption in the treatment group compared with the placebo group (P < 0.05) (Boyle 1980).

Durie calculated faecal fat and nitrogen as g/24 h and percentage of intake (Durie 1980). Cimetidine and sodium bicarbonate, both individually and in combination with each other, significantly improved fat and nitrogen excretion. However, the results from the combination of cimetidine and sodium bicarbonate were no better than those from cimetidine and sodium bicarbonate individually. Durie suggested that this meant that response to the single drug regimens were maximal (Durie 1980).

2. Measures of lung function

Data for forced expiratory volume in one second (FEV₁), forced vital capacity (FVC), forced expiratory flow rate (FEF_25‐75%) were not available in any of the trials.

Both the Chalmers and the DiMango trials reported objective measures of lung function narratively (Chalmers 1985; DiMango 2014). Chalmers found no significant difference in mean change in peak flow or Crispin‐Norman score after six months of treatment (Chalmers 1985) and the DiMango abstract reported that neither the esomeprazole group or the placebo group demonstrated a significant change in FEV₁ after 24 weeks (DiMango 2014). One trial measured lung function (Schoni 1984) and another trial measured Crispin‐Norman scores (Carroccio 1992), but no data were reported.

3. Measures of quality of life

Only the Di Mango abstract, mentioned quality of life and reported that neither the esomeprazole group or the placebo group demonstrated a significant change in CFQ‐R (DiMango 2014).

4. Mortality

These outcomes were either not reported or the data were not available in any of the trial reports.

5. Adverse effects

The trial by Robinson reported diarrhoea as an adverse effect experienced by participants on the prostaglandin E2 analogue, misoprostol (Robinson 1990). Two participants had to be withdrawn from the trial because of diarrhoea that did not subside after five days.

Durie reported that two participants complained of gaseous abdominal distension while taking sodium bicarbonate and one participant was forced to withdraw from the trial because of possible neurological complications due to cimetidine (Durie 1980).

DiMango reported that participants receiving esomeprazole experienced a shorter time to first exacerbation compared with placebo, but the difference was not statistically significant. Further, five of eight participants receiving esomeprazole compared with two of six participants receiving placebo developed a pulmonary exacerbation during the six‐month follow up (P = 0.11) (DiMango 2014).