Patterns of food and acid reflux in patients with low-grade oesophagitis—the role of an anti-reflux agent


Washington Department of Surgery, E Floor, West Block, Queen’s Medical Centre, Nottingham NG7 2UH, UK.



Food and acid have been shown to be refluxed independently of each other in healthy volunteers, and anti-reflux agents decrease the reflux of both parameters. Until now this phenomenon had not been studied in patients with low-grade oesophagitis, who are the group most likely to use anti-reflux medication.


To assess patterns of gastro-oesophageal reflux of acid and food in 12 ambulant patients with endoscopically proven oesophagitis of between grades I and II, but who were otherwise healthy. Also to assess the effectiveness of a single dose of an alginate-containing anti-reflux agent in controlling food and acid reflux in this patient group.


Oesophageal pH monitoring and external ambulatory gamma detection were used to study food and acid reflux. A pH electrode was positioned 5 cm above the cardia and the gamma detector was positioned externally over the pH electrode. The patients then received a technetium-99m labelled meal designed to provoke reflux. Thirty minutes later the patients were given a 20 ml dose of alginate (Liquid Gaviscon), or 20 ml of tap water. Incidence of reflux was monitored for approximately 4 h from the end of the meal. Allocation to treatment group was randomized, with patients receiving the alternative treatment on the second study day after approximately a 7-day washout period.


The mean percentage time oesophageal pH remained below 4 was 16.3 min for the control group and 5.4 min for the treatment group (P=0.03). Food reflux was detected 23.7% of the time in the control group compared to 12% of the time in the treatment group (P=0.02). The anti-reflux agent was also successful in decreasing the number of events, but the duration of the reflux events was not significantly different.


Patients with grades I and II oesophagitis reflux food and acid independently, and are predominantly either food refluxers or acid refluxers, but not both. Liquid alginate decreases the number of both food and acid reflux events, but does not change their duration.


Although pH monitoring is generally regarded as the ‘gold standard’ in the diagnosis of gastro-oesophageal reflux, global results of 24-h monitoring of both oesophageal pH and pressure correlate poorly with symptoms. In recent years, several groups have compared the results obtained from 24-h pH monitoring with those from gamma scintigraphic detection of reflux. Interestingly, there is an extremely poor correlation between the reflux of food and acid detected by the two techniques, and significantly more reflux episodes are recorded by scintigraphy than the number of pH drops greater than one unit, even at pH levels higher than 4. The two techniques appear to explore different phases of reflux, with scintigraphy detecting the reflux of buffered gastric contents.1[2][3]–4 It has further been suggested that scintigraphy reflects histological oesophagitis rather than endoscopic oesophagitis.5

The phenomenon of independent food and acid reflux has also been detected in ambulant normal subjects using conventional pH monitoring combined with ambulatory detection of radiolabelled food.6 Simultaneous food and acid reflux occurred for less than 1% of the time, implying incomplete mixing of food and acid in the stomach. The aim of the present study was to explore the relationship between food and acid reflux in ambulant patients with endoscopically-diagnosed low-grade oesophagitis, and to examine the effectiveness of raft-forming anti-reflux agents, which provide a physical barrier to reflux, in preventing the reflux of food and acid into the oesophagus.


Patient selection

The study was performed in 12 patients suffering from gastro-oesophageal reflux disease (GERD), age range 19–73 years. Patients with grade I or II oesophagitis (Savary–Miller classification) as reported by routine endoscopic evaluation were recruited on to the trial. These patients then underwent a medical examination to evaluate their general fitness to participate in the trial.

Patients were randomly assigned by a blocked, balanced randomization list to treatment order following a Williams Square design. On the morning of the trial, all females were given a pregnancy test and were only accepted if the result was negative.

Ethical committee approval was obtained from the Nottingham Hospitals Ethical Committee and the study was carried out in accordance with the Declaration of Helsinki. Patients were given both written and verbal information concerning the nature of the study and they were required to provide informed written consent before entering into the trial. Permission to administer radioisotopes to human subjects was given by ARSAC.

Study procedure

The pH probe (Radiometer, Copenhagen) was calibrated as per the manufacturers instructions, and then marked 5 cm from the end using approximately 0.5 MBq Tc-99m dried on to a square of filter paper, and secured using waterproof tape. The probe was sterilized before use.

Subjects were fasted overnight and, upon arrival in the department, the exact location of the gastro-oesophageal sphincter was located using manometry. The pH probe was then located 5 cm above the cardia. The Tc-99m tag on the pH probe was used to locate the cadmium telluride detector over the pH probe. Both the pH probe and the gamma detector were connected to a two channel solid state recorder (Novo Memolog; Vertec Scientific Ltd, Reading), the timebase was set to 15 s.

The patients were given the radiolabelled meal,7, 8 which consisted of a Spanish omelette, fizzy lemonade followed by apple pie and ice cream and tea or coffee. Thirty minutes later they were given either 20 ml tap water or 20 ml Liquid Gaviscon (1000 mg sodium alginate, 534 mg sodium bicarbonate, 320 mg calcium carbonate per 20 ml dose; Reckitt and Colman Products, batch no. N13901). Each dose of Liquid Gaviscon was freshly dispensed prior to administration. The patients were then allowed to resume their normal activities.

Recording of data took place for approximately 4 h, and at the end of this period the pH probe and gamma detector were removed and the calibration of the pH probe was checked. Probes were sterilized in ASCP (Galen Ltd, Craigavan).

The two treatments were administered 1 week apart and allocation to treatment order was randomized.

Data assessment

At the end of the recording period, the data were transferred to an Apple Macintosh computer for analysis using the communication software package Versaterm (Abelbeck Software). To assess food and acid reflux, the data were transferred to a spreadsheet (Microsoft Excel). For the purpose of acid reflux detection, a pH threshold of less than four units was used, in line with clinical practice. The spreadsheet counted the total number of events per hour, the percentage of time that the pH was lower than 4 and the average duration of a pH event. The number of pH events longer than 1 min, and the percentage of events that were longer than 1 min, were also computed. Analysis of the food reflux data from the gamma probe was more complex because the detector measured a small background count from the radiolabelled food in the stomach; this signal decayed as the stomach emptied during the recording period. Consequently, a decaying exponential baseline was subtracted from the data to level the baseline, prior to applying a visually-determined threshold to select the food reflux events. In practise the food reflux events were easy to identify and the threshold was not ambiguous. Normal threshold values were approximately 100 c.p.s. in all subjects; it would be expected that this threshold value would differ among the subjects due to variations in subject anatomy and gamma probe positioning. The spreadsheet then reported the number of food reflux events per hour, the percentage of time that a food reflux signal was present, the average duration of a food reflux event, and the number and percentage of events that were longer than 1 min. Statistical differences in the means of the parameters were determined using a non-parametric Wilcoxon test, because the means were not normally distributed.


Fourteen subjects with a history of gastro-oesophageal reflux were recruited from the Queen’s Medical Centre reflux clinics. Two patients were unable to complete the study. Data from the 12 subjects who successfully completed the trial were included in the analysis. No adverse effects were reported by any of the patients.

The 12 subjects who completed the study were in the age range 19–73 years, mean age 49.3 years. The study population comprised six males and six females with eight subjects classified as having grade I oesophagitis and four subjects as having grade II oesophagitis on entry into the study. In eight subjects less than 50% of the circumference of the oesophagus was affected, for one subject more than 50% was affected, and for a further three subjects the whole circumference was affected. The length of oesophagus affected was measured and showed a range of 1–10 cm, a mean of 2.75 cm and a median of 2 cm for the 12 subjects.

Table 1 shows the distribution of the values obtained for food and acid reflux in these patients. Acid was present in the oesophagus for 16.4±7.6% of the time during the 4-h postprandial period. This was significantly reduced by Gaviscon to 5.4±3.8 min (P=0.03). The patients experienced 3.2±1.3 acid reflux events per hour. Gaviscon reduced the number of such pH events to 1.4±0.5 events per hour (P=0.09) and the percentage of time taken up by such events was significantly reduced (P=0.03).

Table 1.  . pH and food reflux parameters in patients with low-grade oesophagitis and the effect of an anti-reflux agent Thumbnail image of

Liquid Gaviscon also reduced the number of acid events longer than 1 min, from 1.6±0.6 events per hour to 0.5±0.2 events per hour. However, the mean duration of the pH events was not significantly reduced (2.0±0.8 in control group vs. 1.3±0.5 in the treated group; P=0.19), nor was the percentage of the pH events that were longer than 1 min (32.2±8.8 in the control group vs. 28.9±9.3 in the treated group; P=0.2).

Broadly similar trends were seen in the food reflux data between the control and treatment groups. The number of food reflux events per hour was reduced from 13.6±1.8 to 8.4±1.8 by Liquid Gaviscon (P=0.02), and the percentage of time that a food reflux signal was detected was also reduced, from 23.7±5.4 to 12.0±4.5 (P=0.02). The number of events longer than 1 min (per hour) was reduced from 2.8±0.7 to 1.2±0.5. The average duration of a food reflux event was not changed significantly, from 1.1±0.2 to 0.8±0.3 (P=0.26). It is interesting that the average duration of a food reflux event was very similar to that of an acid reflux event. Finally, the total food reflux (defined as the integrated area under the count–time curve) was significantly reduced from 138.8±68.3  kct/min to 28.3 ± 12.6 kct/min.


This study was the first to use the combined techniques of pH monitoring and gamma monitoring in patients with endoscopically proven grade I and II oesophagitis. As with previous studies carried out on volunteer subjects, food and acid reflux events were found to occur independently of each other.6, 8Figure 1 shows a scattergram of percentage time below pH 4 compared to percentage time of food reflux for each subject. The correlation coefficient for this plot is −0.15, indicating a very low degree of correlation; indeed, but for one subject, there appears to be a good inverse correlation. Patients displayed either predominantly food or acid reflux, but not a combination of the two. The patients refluxed acid for approximately 16% of the postprandial period, whereas in a previous study with an identical protocol, healthy volunteers were found to reflux acid for only 3% of the time.6 Food reflux occurred for 24% of the postprandial period in the patients compared to 18% in healthy volunteers. Interestingly, three patients with endoscopically proven oesophagitis did not reflux acid during the postprandial period and three others refluxed acid for less than 5% of the recording time.

Figure 1.

. Correlation between food and acid reflux in untreated subjects.

The results show that treatment with Liquid Gaviscon reduces a range of parameters for both acid and food reflux ( Table 1). The number of pH events was reduced although this did not reach statistical significance, however, the number of food reflux events and the percentage of time taken up by such events was significantly reduced. The average duration of reflux events was not significantly reduced and was 1–2 min in both cases, which was not unexpected because alginate anti-reflux agents act via a barrier action. If the barrier is breached, there is no reason to expect the refluxed material to be cleared more rapidly. Thus we would expect the number, but not the duration, of reflux events to be reduced, which is consistent with the present results.

Although the use of ambulatory gamma monitoring for the detection of radiolabelled food reflux in gastro-oesophageal reflux patients is still in its infancy, the more traditional technique of gamma scintigraphy was used as early as 1982 to study this disease.9 Scintigraphic studies have also shown a poor correlation between food and pH reflux, and so it was initially concluded that scintigraphy simply was not sensitive enough to detect the small amount of refluxed food.10 In 1984, Cargill in France found that scintigraphy could be used to evaluate the reflux of food in the development of pulmonary disorders.11 Later studies demonstrated that the use of acidified food increased the correlation between food and acid reflux, leading to the conclusion that gamma scintigraphy was detecting the reflux of neutral materials which could not be detected by pH monitoring.12, 13 By 1986, largely due to the work of Kaul and co-workers, scintigraphic reflux could be detected in 92% and 79% of the patients with and without histologic evidence of oesophagitis.14 By 1990, the lack of correlation between pH monitoring and reflux of radiolabelled gastric contents as measured by scintigraphy became accepted as accurate findings and the conclusion was drawn that extended pH monitoring and scintigraphy measured different pathophysiologic phenomena and they detected reflux under different conditions.3, 15 Tolia and co-workers15 also commented that: ‘The ability of these tests to detect reflux may be complementary and they may be of greatest value when used together to enhance the sensitivity and specificity of the diagnostic evaluation’. This was confirmed independently by Shay and co-workers2 and Vandenplas and co-workers.1 Shay’s study concluded that the two methods agreed in only 25% of total reflux events. Scintigraphy was superior for the detection of reflux of buffered gastric contents and the detection of additional reflux events during acid clearing intervals, whereas only the pH probe detected reflux events after gastric emptying. Vandenplas and co-workers reported that out of 123 reflux episodes recorded with both techniques, only six occurred simultaneously. Significantly more reflux episodes were recorded by scintigraphy, particularly during the first half-hour period (n=62), if compared with the number of pH drops greater than one unit, even at pH levels higher than 4 (n=41; P<0.05). It was again concluded that the two techniques explored the reflux phenomenon in different ways.

In some cases, scintigraphy has been shown to be superior to pH monitoring, for example in the detection of reflux of gastric contents resulting in aspiration pneumonia in tube-fed elderly patients16 and pulmonary aspiration of gastric contents in asthmatic adults.17 Recent evidence from Nottingham using magnetic resonance imaging has clearly demonstrated that food and liquid form two distinct layers in the fundus of the stomach, where the motility patterns which could mix the two phases are relatively weak.18 Formation of chyme, which is a homogenous mixture of food, liquid and gastric secretions, occurs in the antrum from where it is emptied into the duodenum. This provides further evidence for a mechanism by which food and acid could be refluxed independently.

Although oesophageal pH monitoring is often quoted as being the ‘gold standard’ method for the diagnosis of gastro-oesophageal reflux, the analysis of the pH data has not been standardized. Although most groups define reflux of acid to have occurred if the oesophageal pH drops below pH 4,19 other groups advocate that the duration of each reflux event is a better indicator of disease and damage.20 Branicki and co-workers advocated a fall in pH of at least two units from baseline.21 In addition to these problems of analysis, there is no precise definition as to the discrimination between ‘normal’ and ‘abnormal’ reflux. Large numbers of reflux events per hour indicate that the lower oesophageal sphincter is acting as a poor barrier to reflux, but prolonged episodes reflect poor clearance. So-called ‘normal’ ranges have been given by Cheadle and co-workers.22 Using their values, five of the patients in the current study can be classified as having ‘normal’ amounts of acid reflux, three as having ‘normal’ numbers of reflux events per hour, and in 10 patients the duration of each event was ‘normal’. In three patients all parameters studied were normal despite the fact that these patients had endoscopic evidence of reflux oesophagitis. There are several possible explanations for the low levels of acid reflux:

These patients may be mainly nocturnal acid refluxers.

The meal may not have contained the exact ingredients which stimulated reflux for each particular patient. It may be necessary in the future to use patient questionnaires to identify the foods which precipitate the onset of symptoms.

Reflux disease is cyclical and can have periods of relative quiescence. Generally, the period of symptoms is triggered by a particular event, such as eating a food which is a stimulant of reflux, or overeating or over indulgence in alcohol, etc.

In contrast, all the subjects showed some degree of food reflux. It would be expected that food reflux would be a better indicator because reflux is more likely to occur when the stomach is full. The main difficulty with this approach is that food reflux detection is still not sufficiently widely used for scoring and severity criteria to be established, which is necessary before it can form a clinical test. It should, however, be pointed out that the scoring criteria used for pH monitoring are largely arbitrary and have come about through general acceptance rather than establishment of cause and effect.


Patients with grades I and II oesophagitis reflux food and acid independently, and are predominantly either food refluxers or acid refluxers, but not both. In this regard they are similar to normal subjects. Anti-reflux agents reduce the number of both food and acid reflux events, but do not change their duration. It appears that it is necessary to study both food and acid reflux in order to successfully understand the mechanism of GERD successfully.


The authors gratefully acknowledge the financial support for the study from Reckitt and Colman Products.