Skip navigation

Upper gastrointestinal disease


What is the prevalence of upper gastrointestinal symptoms?
Results
Comment
How effective is endoscopy in dyspepsia management?
Results
What happens to people with reflux oesophagitis on placebo?
Results
Comment
What are the consequences of long standing reflux disease?
Study
Results
Comment

If evidence-based medicine in the form of systematic reviews does anything, it makes you think. Reading any review is likely to inform, and not just in areas of ignorance: they inform also where we think we know something. Systematic reviews are also being done to answer questions that we think difficult, and using new methods to glean knowledge where previously there was only question or knowledge of the hand-me-down variety.

Bandolier has read with interest several different systematic reviews about upper gastrointestinal disease. Each examines a different aspect, and each raises perhaps as many questions as it answers.

What is the prevalence of upper gastrointestinal symptoms?


A systematic review searched MEDLINE and previous reviews on the subject [1]. Studies were included if conducted in the general population examining the prevalence of dyspeptic symptoms and reporting the period studied, sample size, response rate and definitions of questions.

Results


Ten studies were identified: six from Scandinavia, two from the USA and two from the UK. Samples were above 1000 individuals, all examined adults, and the response rate was generally above 70%. All the studies used a questionnaire, and the period over which symptoms were examined was from three months to lifetime, but most commonly one year.

Age and period made no obvious difference to the results. There were wide variations. Upper abdominal pain or discomfort affected between 8% and 54% in different studies. Heartburn and/or regurgitation affected 10% to 48% for heartburn, 9% to 45% for regurgitation, and 21% to 59% for either. Prevalence rates were lowest when the questions related only to epigastric pain (8-21%), were higher when questions included upper abdominal pain or discomfort (11-26%) and were highest when upper abdominal symptoms were included (29-54%).

Comment

Prevalence obviously depended on the questions asked, and how those questions were understood. The bottom line is that there is a lot of it about.

How effective is endoscopy in dyspepsia management?


A systematic review [2] set out to assess whether the literature could answer a number of questions relating to the effectiveness of endoscopy in improved patient outcomes or cost-effectiveness in the management of dyspepsia. Searching used three computerised databases for clinical studies (any study design) of patients with dyspepsia with information of the effects of endoscopy on:
  1. Patient outcomes (symptoms, quality of life, anxiety, satisfaction)
  2. Resource utilisation
  3. Clinical decision making
  4. Cost effectiveness

Results


For initial endoscopy in the management of dyspepsia, the weight of evidence only supports its use in clinical decision making. For the other three questions, the evidence does not support initial endoscopy.

There are important caveats, though. First, restriction of endoscopy to patients who had positive blood tests for Helicobacter pylori, who were older than 45 years, or who were taking NSAIDs would reduce the number of endoscopies by a large amount, a consistent finding in five non-randomised studies. Second, many studies had designs that were suboptimal, limiting our ability to generalise.

This review helps anyone trying to devise or review care pathways for patients with dyspepsia in primary care. The available evidence is presented, in some detail, and different weight may be placed on different aspects of the evidence depending upon local circumstances.

What happens to people with reflux oesophagitis on placebo?


What this paper sought was all trials of reflux oesophagitis with placebo controls between 1976 and 1990. It does not tell us whether or not the trials were randomised, but does tell us that only English language papers were used. To an extent, therefore, the searching and inclusion was less than satisfactory. Many more trials will have been published since 1990.

What the authors did was to extract the healing rates with placebo, both those symptom free and those with no worse than grade 1 oesophagitis. Grade 0 is given to normal oesophagus with no macroscopic damage. Grade 1 describes an oesophagus with a few areas of erythema, mucosal friability and contact bleeding. These are minor changes regarded as being present in normal oesophagus by some gastroenterologists.

Results


They found 22 studies. The healing rate for no worse than grade 1 oesophagitis varied widely between studies (Figure 1). Individual trials had healing rates between 0% and 63% at eight weeks, though only one of these trials had more than 20 patients given placebo. Overall 116/464 patients (25%) were healed at 4-6 weeks, 22/98 (22%) at 8 weeks, and 104/340 (31%) at 12 weeks or longer.

Figure 1: Healing rates with placebo in trials of reflux oesophagitis



Comment


This paper may not be the best ever, but it certainly makes one think. It is not uncommon for people to make a great play of different placebo event rates between individual trials to play up or play down a particular result from a single trial. We need to be cautious about the single trial reflex ( Bandolier 27 ), not just about extrapolating a result from one small trial, but on dismissing a result because we thing a placebo response is out of line.

It also makes it possible to consider using a generalised placebo healing rate together with healing rates from systematic reviews of PPIs and histamine antagonists [4] to calculate NNTs compared with placebo. Figure 2 shows the average eight-week healing rates. Not everyone would agree with this, but given there is clinical homogeneity between the trials, the implication is that the NNT for eight week healing of erosive reflux oesophagitis with a PPI is 1.8, while that for the same outcome with an H2A is 4.4. More formal work might underpin this conclusion.

Figure 2: Healing rates with H2As and PPIs in trials of reflux oesophagitis



Then there's a philosophical point. We talk glibly about placebo response rates, as if the placebo has caused the response. In this case we have no idea what would have happened without placebo. Is it credible that some psychological healing effect is at play here, or is this just the natural history of the disease?

What are the consequences of long standing reflux disease?


A study of the whole population of Sweden tells us that frequent and severe symptoms of reflux over a long period are associated with very much higher risks of oesophageal cancer [5].

Study


The study found every case of cancer of the oesophagus or gastric cardia newly diagnosed between late 1994 and 1997. The Swedish system and special organisation allowed the cases to be identified rapidly and to be paired with matched controls chosen at random from the Swedish population. Patients and controls were seen by trained interviewers blinded to the intention of the study, and they were asked a number of questions about lifetime experience of heartburn and regurgitation. Symptoms occurring within five years of diagnosis were disregarded. Diagnosis of cancer was by pre-set rules, and almost all cases were reviewed by a single pathologist.

Results


There were 618 patients with cancer and 820 controls. After adjusting for a mass of different things, the results showed that oesophageal adenocarcinoma, but not adenoma of the gastric cardia or squamous cell carcinoma of the oesophagus, was highly related to symptoms of reflux. The Table shows the odds ratios for the frequency, duration and severity of symptoms.

Table: Oesophageal adenocarcinoma - association with symptoms and their severity and duration
Symptom of reflux Comparison Odds ratio (95%CI)
Heartburn, regurgitation or both at least once a week Less than once a week 7.7 (5.3 to 11)
Heartburn, regurgitation or both at night at least once a week Less than once a week 11 (7.0 to 17)
Symptom of reflux Comparison Odds ratio (95%CI)
Reflux symptoms more than three times a week No symptoms 17 (8.7 to 28)
High reflux symptom score No symptoms 20 (12 to 35)
Duration of symptoms more than 20 years No symptoms 16 (8.3 to 28)

For frequency, symptom severity and for duration of symptoms there was a dose-response relationship - shown for frequency in Figure 3.

Figure 3: Odds ratios for frequency of reflux symptoms and development of oesophageal adenocarcinoma, compared with controls



Comment


This study was very good. It shows a clear association between reflux symptoms and oesophageal cancer, and probably establishes causality. What it does not show, and is careful to express, is that treating the symptoms will prevent the cancers developing. They point out that oesophageal cancer is becoming more common, while the prevalence of symptoms is more or less unchanged and while effective treatments have been introduced. It also points out that endoscopic surveillance would swamp the system. It poses many questions, but we have to wait a little longer for definitive answers.

References:

  1. RC Heading. Prevalence of upper gastrointestinal symptoms in the general population: a systematic review. Scandinavian Journal of Gastroenterology 1999 34 Suppl 231: 3-8.
  2. JJ Ofman, L Rabeneck. The effectiveness of endoscopy in the management of dyspepsia: a qualitative systematic review. American Journal of Medicine 1999 106: 335-46.
  3. F Pace et al. Meta-analysis of the effect of placebo on the outcome of medically treated reflux esophagitis. Scandinavian Journal of Gastroenterology 1995 30: 101-5.
  4. www.ebandolier.com/bandopubs/gordf/gord.html
  5. J Lagergren et al. Symptomatic gastroesophageal reflux as a risk factor for esophageal adenocarcinoma. New England Journal of Medicine 1999 340: 825-31.
previous or next story in this issue