Introduction Most patients with upper gastrointestinal symptoms are now H pylori negative and no abnormalities are seen at endoscopy. Proton pump inhibitors (PPIs) are usually prescribed to these patients, though only a proportion benefit.

Aim To identify factors predicting response to PPI therapy in patients with upper GI symptoms and normal endoscopy.

Methods 105 patients with upper gastrointestinal symptoms of >3 months, normal endoscopy and HP-ve tests were prospectively recruited. Patients with hiatus hernia or taking NSAIDs were excluded. Character of symptoms and lifestyle factors were assessed by validated questionnaires and baseline severity over 7 days. The predominant symptom was retrosternal discomfort in 43, epigastric discomfort in 42, nausea/vomiting in 10, fullness/satiety in 1 and other symptoms in 6. Their median BMI was 26 (range 17.6–40.4) with 39% overweight and 20% obese. 24 h oesophageal pH monitoring and manometry of the lower oesophageal sphincter (LOS) by perfusion pull through technique (1 cm every minute) was performed. Patients were then randomised in double-blind fashion to 2 weeks daily treatment with 2/3 receiving lansoprazole 30 mg/day and 1/3 placebo. Symptoms were re-assessed throughout the second week of treatment, with the primary outcome being absolute change in symptom severity score (Glasgow Dyspepsia Score).

Results 102 individuals were treated (69 randomised to Lansoprazole, 33 to placebo). The mean pre-treatment symptom score in the active treatment group was 10.4 (SD 2.2) and in placebo 10.9 (SD2.7). The mean reduction in score following active treatment was 3.8 (SD 3.8) vs 0.6 (SD 2.1) following placebo (p<0.001). Within the PPI treated group, univariate analysis showed that BMI, LOS pressure, LOS abdominal length and per cent time oesophageal pH<4 were statistically significant predictors of response. Multivariate regression modelling revealed that in the best model only BMI (p=0.003) and LOS abdominal length (p=0.013) were independent predictors of response (R²=19.4%). Neither predominant symptom nor any other symptom characteristic predicted response.

Conclusion This study indicates the lack of value of detailed symptom assessment in predicting response to PPI therapy. The strong predictive value of BMI is likely to be due to its association with gastro-oesophageal reflux which will respond to PPI therapy and highlights the important role of BMI in reflux disease.