Healthcare-related (nosocomial) gastro-intestinal (GI) bleeding is rare outside the intensive-care setting: consequently, outside these areas the benefit of routine acid-suppressive therapy is small, according to a cohort study from a large US academic hospital centre.

Acid-suppressive drugs are widely prescribed to hospital in-patients, despite the facts that most patients do not have indications for them and there is increasing evidence that use is linked with increased rates of nosocomial infections. The authors of this study suggest that the widespread prescribing is by extrapolation from evidence-based use in critically-ill patients to reduce stress-related GI bleeding. They note that there are few published data on the incidence of nosocomial GI bleeding in the non-ICU population and whether these patients would benefit from acid-suppressive medication. This study was carried out to clarify the issue, and aimed to determine the incidence of nosocomial GI bleeding and effects of acid-suppressive drugs in hospital patients who were not critically ill.

They used records of admissions to a large US academic medical centre between January 2004 and December 2007, and extracted data on patients aged over 18 admitted for more than 3 days and without a primary diagnosis of GI bleeding. Exposure to acid-suppressive drugs (any proton-pump inhibitor, PPI, or histamine-2 receptor blocker, H2RA) before a bleeding event was determined and censored at onset of GI bleeding. Primary outcome was nosocomial GI bleeding occurring outside an ICU setting; secondary outcome was clinically-significant nosocomial GI bleeding. Overall risk of nosocomial GI bleeding was calculated, plus the odds ratio (OR) for those receiving acid-suppressive therapy and the associated number needed to treat (NNT). Propensity score matching was used to identify exposed and non-exposed patients with a similar baseline risk of GI bleeding, to reduce the risk of confounding by indication.

There were 136,529 adult admissions during the study period; of these, after exclusions (stayed <3 days, n = 56,430; diagnosis GI bleeding, n = 812), there were 79,287 eligible for the analytic cohort. The median age of the cohort was 56 years (range, 18-107 years), and 31,798 (41%) were men. Exposure to acid-suppressive therapy was high, with prescriptions for a majority (59%) of patients (81% PPI, 39% H2RA – some received both). Within the cohort, 18,983 exposed patients were successfully matched with 18,983 non-exposed, and baseline characteristics of the two matched groups were similar.

There were 1,776 potential cases of nosocomial GI bleeding, and after review the primary outcome of nosocomial GI bleeding was identified in 224 admissions (0.29%); the secondary outcome of clinically significant GI bleeding occurred in 176 admissions (0.22%). The incidence of the primary outcome was similar in the propensity-matched and the overall cohorts, and after adjustment for residual imbalance, receipt of acid-suppressive medication was associated with a reduction in the risk of GI bleeding with an adjusted odds ratio of 0.63 (95% CI, 0.42-0.93).

Using the calculated values for incidence and effect, the estimated NNT was 770 to prevent one episode of nosocomial bleeding and 834 to prevent one episode of clinically relevant bleeding.

Based on their analysis, the authors conclude that in this study cohort nosocomial GI bleeding was rare outside the intensive care setting. Acid suppressive therapy reduced the relative risk by about 37%, but because the event was so rare the number of patients needed to treat to prevent one episode was high – 770 for all bleeds and 834 for clinically significant bleeds. They note the limitations of their observational study, and suggest further research to confirm and extend the results. Meanwhile, they suggest that their study supports recommendations to avoid routine use of acid-suppressive drugs in hospitalised patients who are not critically ill.