MEDIA STATEMENT
5th September 2005

Lanas A et al. A nationwide study of mortality associated with hospital admission due to severe gastrointestinal events and those associated with nonsteroidal anti-inflammatory drug use. Am J Gastroenterol 2005;100:1685-93

The Aspirin Foundation welcomes new scientific studies which contribute to the evidence of the safety of aspirin but the study by Lanas et al. has shortcomings that cast doubt on the validity of its findings. In particular, the basis on which the authors attribute gastrointestinal (GI) events to drug treatment is open to criticism.

This observational study was a retrospective analysis of the incidence of hospital admissions and deaths in Spain due to serious upper and lower GI events. Its findings are consistent with what we already know about the frequency of serious GI events leading to hospital admission. The authors state (column 2, page 1690):

The overall rate of deaths due to GI complications and the rate of deaths associated with NSAID/aspirin use reported here are lower than some frequently quoted estimates from previous studies, despite the fact that our figures include both upper and lower GI complications and also refer to low dose aspirin. Our data, however, agree with some more conservative estimates from the United Kingdom and more recent data regarding recurrent rates of GI complications reported in the United States.

These estimates of GI events appear to be accurate but the authors do not provide adequate data on drug use and their estimates of risk are therefore unreliable. They claim that 36.3 percent of events were attributable to use of NSAIDs or aspirin. This is not a reliable estimate because:

• They estimated the frequency of GI events(a) and the level of NSAID and aspirin consumption(b) in different studies and their methodology is not completely reported in this paper.
• They do not appear to have checked whether the patients who experienced GI events were actually taking an NSAID or aspirin. Instead, they have inferred, from the findings of a previous study, that a proportion of GI events must be due to drug use.
• They do not report separate data for aspirin use (at any dose) and NSAID use, nor do they distinguish between the consumption of prescribed treatment and over-the-counter medicines.

In summary, this study confirms what we already know about the rates of GI events leading to hospital admission. However, it should be interpreted cautiously because:

• It was an observational study and should not be interpreted as drawing a causal inference between GI events and drug use.
• The method by which drug use was estimated, and how this related to GI events, is not adequately explained(c).

ENDS
Notes
a. The study drew on two separate and different sources of data about GI events.

The sources of data on deaths and admissions were:
• The minimum basic data set of computerised data from 26 general hospitals for 2001, selecting patients with a primary diagnosis of an upper or lower GI event which, in the clinical judgement of the admitting doctor, was the direct cause of the admission. New events (other than death due to the primary event) occurring in hospital were not recorded and cases of GI obstruction were excluded.
• Data from the IASIST database, drawn from 197 of the 269 hospitals in Spain. These data represent 77 percent of all admissions in Spain in 2001. Selection criteria were the same as above but also included severity of the event, length of stay and cost.

These datasets gave broadly similar results and therefore suggest some of the estimates are accurate, though data from the 26 hospitals suggested a 17 percent lower frequency of GI events overall, 20 percent fewer upper GI events and 16 percent fewer deaths.

b. Drug use was not recorded in the same dataset from which GI events were identified and there was no analysis of how well the two sets correlated. Data on drug use were obtained in two ways:

• For the patients identified in the 26 hospitals, charts were manually reviewed to identify drugs used in the month preceding death. However, these data do not appear to have been used in the analysis.
• A prospective study carried out at the same time and in the same 26 hospitals. This has been reported elsewhere in abstract form.

c. The prevalence of NSAID/aspirin use was estimated at 19 percent; the estimated rate of GI complications in patients who didn't use NSAID/aspirin was >120/100,000 pts/yr; that for previous users was >480/100,000 pts/yr. These figures do not mention deaths. The authors then state, without further explanation: 'Therefore, the proportion of complications and deaths attributed to NSAID/aspirin use was 36.3%'.