Reversible prolonged skin bleeding time in acute gastrointestinal bleeding presumed due to NSAIDs

The purpose of this research was to look for a possible mechanism whereby NSAIDs, and particularly ASA, might cause gastrointestinal bleeding. A total of 34 hospitalized GI bleeders and 29 age- and sex-matched controls were studied. Skin bleeding time (SET) was measured within 6 h of coming to hospital and before any blood products were given. All patients and controls were questioned regarding current NSAID use. This history was supplemented by estimation of serum salicylate and of platelet cyclooxygenase activity to detect unreported current aspirin (ASA) use. Various aspects of platelet function were also tested by lumiaggregation in 28 controls and, after recovery, in 27 of the bleeders. Of 34 bleeders, 26 bled from the upper GI tract, (13 from peptic ulcer) and eight from the lower GI tract, 30 (88%) had a current intake of NSAIDs and of these 22 (73%) used ASA, some in combination with other NSAIDs, whereas 12 of 29 controls were using NSAID's, 11 of which were ASA. SET in the bleeders was 9.0 +/- 1.02 min versus 4.8 +/- 0.42 min in the controls (p < 0.001). SET measured 6.6 days later in 28 bleeders was 4.7 +/- 0.22 min (p < 0.0006), and of those tested after recovery all but one had fallen to 6.5 min or less. None had any residual constitutional platelet abnormalities as tested by lumiaggregation. By logistic regression, NSAID intake was strongly associated with prolonged SET to > 6 min (odds ratio [OR], 16.7; p < 0.0002), whereas NSAID intake (OR 14.6; p < 0.0003) and SET > 6 min (OR 1.8; p < 0.005) contributed to a bleeding outcome. Almost 90% of GI bleeders had recently consumed NSAIDs, mostly ASA, on an average 15 h before onset of bleeding. Although most of the nonbleeders who had used NSAIDs did not have a prolonged SET, most of the bleeders who used NSAIDs had an abnormal elevation of SET, suggesting a possible mechanism for GI bleeding. Retesting similar to 7 days after recovery from bleeding showed normalization of the SET, indicating that the defect was transient and spontaneously reversible.