The GOOD trial was a 14-day multicentre randomised double-blind double-dummy non-inferiority trial comparing efficacy between Gaviscon® and omeprazole 20 mg. It recruited 90 general practitioners (GPs) so as to obtain 75 active investigators. Patients recorded symptoms 4 times a day for 2 weeks and the time of taking each treatment (morning, midday, evening and bedtime); they also recorded any onset of heartburn, and if so, at what time of day (morning, midday, evening, bedtime) and any experience of relief, and if so, at what interval after first taking the treatment. The GPs performed three mandatory assessments: D0 (inclusion visit), D7, and D14.
The trial ran from August 27 to November 29, 2010, and respected the ethical principles of the Seoul revision (2008) of the Helsinki Declaration and Good Clinical Practice. The study protocol received approval by the Comité de Protection des Personnes d'Île-de-France VIII ethics committee on May 3, 2010, and was registered (N° A 100 546-10) by the Agence Française de Sécurité Sanitaire des Produits de Santé (French health products approval authority). All patients were duly informed of the trial objectives and signed an informed consent form.
Included patients were male or female, aged between 18 and 60 years, with 2 to 6 days of GERD episodes per week, with heartburn, with or without regurgitation, not taking alginate/antacid or PPI treatment for at least the preceding 2 months, and able to understand the study and to complete the self-administered questionnaires. Women of child-bearing age had to have effective birth control. Exclusion criteria were: atypical digestive or extradigestive symptoms without heartburn; gastric or duodenal ulcer; history of upper digestive tract surgery or of upper digestive tract or otorhinolaryngologic neoplasm; known hypersensitivity to at least one component of Gaviscon® or of omeprazole; known hypersensitivity to benzimidazoles; and treatment with clopidogrel, atazanavir combined with ritonavir, ketoconazole, or itraconazole. Breastfeeding women and women who knew that they were pregnant as well as patients who had participated in a therapeutic trial within the month preceding inclusion in this trial were also excluded.
Included patients were randomly allocated to one of two groups: Gaviscon® (4 × 10 mL/day), or omeprazole 20 mg/day. Randomisation by blocks of 3 (2 + 1) was double-blind. Successive blocks were balanced by 2 s.
Gaviscon® suspension in a 150-mL bottle (Reckitt Benckiser Healthcare France) was administered orally at a daily dose of 10 mL (2 teaspoonfuls), 4 times a day (after the three main meals and at bedtime). Omeprazole (omeprazole MYLAN® 20 mg, Mylan, France) in enteric coated capsule form was administered at a daily dose of 20 mg in the morning. Maximum treatment duration was 14 days.
The active substances of the Gaviscon® oral suspension were sodium alginate and sodium bicarbonate. The placebo was composed of hydrogenated glucose syrup, xanthane gum, methyl parahydroxybenzoate (E218), propyl parahydroxybenzoate (E216), erythrosine (E217), fennel flavour, titanium oxide, and purified water. The placebo was developed so as to have the same aspect, colour, odour and flavour as the aniseed Gaviscon® suspension.
All Gaviscon® group patients also received a capsule of omeprazole-placebo every morning for 14 days, and all omeprazole 20 mg group patients also received 10 mL of Gaviscon®-placebo 4 times a day (after the three main meals and at bedtime) for 14 days.
Study drugs were packaged per patient and per site according to the randomisation list.
The primary outcome was the mean time to onset of the first 24-h heartburn-free period after initial dosing. This outcome was assessed by the GP, based on the self-administered questionnaire filled in 4 times a day by the patient. Mean time to onset was calculated as the difference between 2 time-points: the time of taking the treatment for the first time and the date and time (morning, midday, evening, before bedtime) at which a 24-h heartburn-free period had been achieved.
Secondary outcomes were: (a) the mean number of days without heartburn by D7 as assessed from the patient's self-administered questionnaire; (b) patient's overall qualitative self-assessment of pain relief on D7 on a 5-point Likert scale; and (c) pain intensity on D7 and D14, assessed by the patient on a 100-mm visual analog scale (VAS).
Adverse events (AEs) were collected at the two study visits (D7 and D14). An AE was defined as an untoward medical event that occurred during the study period, whether or not related to the study procedure or study products.
Severe AE (SAE) was defined as an untoward medical event that resulted in death, was life-threatening, required inpatient admission or prolongation of hospitalization, or resulted in severe or persistent disability or incapacity.
Descriptive statistical analyses were performed on the data collected on D0, D7 and D14, and those of the self-administered questionnaire (D0 to D7, and D7 to D14), for the intention-to-treat (ITT) and per protocol (PP) populations. The PP population included all patients from the ITT population who attended at least one of the study visits, except those with major protocol deviations liable to interfere with the primary outcome result. As the objective of the GOOD trial was to determine whether Gaviscon® was non-inferior to omeprazole 20 mg, the PP population was the reference for efficacy analysis, and the efficacy results presented here are those for the PP population.
Inter-group statistical comparison used appropriate one-tailed tests: Chi2 test for qualitative variable (or Fisher's exact test in case of sample size < 5), Student t-test for Gaussian quantitative variables, and non-parametric Wilcoxon test for semi-quantitative or non-Gaussian quantitative variables.
Group comparison used analysis of variance (ANOVA) when variable distribution was normal. Inter-group comparability was checked at inclusion for heartburn frequency and severity, GERD duration, age, regurgitation and alcohol consumption. In case of non-comparability, analysis of covariance (ANCOVA) was performed, introducing into the model the variable or variables that were non-comparable at baseline.
If the distribution of time to a 24-h heartburn-free period was non-normal, a non-parametric Wilcoxon test was used, with the Hodges-Lehmann median estimated with its 95% confidence interval (CI).
Statistical analysis was carried out on SAS version 8.2 (SAS Institute, North Carolina, USA). The significance threshold was set at 5%.
Choice of the lower limit
In non-inferiority studies, the lower limit is classically set at 50% of the reference substance effect in comparison to placebo [12
]. Previous studies showed that the time to a 24-h heartburn-free period was 19-21 days with placebo [13
], 4-5 days with rabeprazole 20 mg [13
], and 2 days with pantoprazole 20 mg or omeprazole 20 mg [14
] (at least 16 days' difference between placebo and PPI). Halving this difference gives a non-inferiority limit of 8 days, which was neither ethically nor clinically acceptable.
The most recent study [14
] reported a value of 1.8 ± 0.8 day for omeprazole 20 mg. The GOOD trial hypothesized that omeprazole 20 mg provides a time to onset of the first 24-h heartburn-free period of 2 ± 1 days. As patients recorded symptoms 4 times a day, the non-inferiority of Gaviscon®
would be demonstrated by a clinically relevant value of 0.5 days less than for omeprazole 20 mg.
For the non-inferiority test, the mean time difference in onset of the first 24-h heartburn-free period between treatment groups was used, with its 95%CI. If the lower limit (0.5 days) was within this confidence interval, the non-inferiority hypothesis was taken to be confirmed.
Sample size calculation
For an α-risk of < 5% and power of 95%, the requisite sample size was 88 assessable patient data sets per group. To allow for incomplete recording of symptoms by patients and loss to follow-up (< 10%), 30% extra patients were to be recruited: i.e., 120 patients per group. In all, 240 patients were required in order to meet the primary endpoint. 90 investigation centres were set up, recruiting three patients each with a 3.5 month deadline.