Risk of Rebleeding, Vascular Events and Death After Gastrointestinal Bleeding in Anticoagulant and/or Antiplatelet Users

Carlos Sostres; Beatriz Marcén; Viviana Laredo; Enrique Alfaro; Lara Ruiz; Patricia Camo; Patricia Carrera-Lasfuentes; Ángel Lanas

Disclosures

Aliment Pharmacol Ther. 2019;50(8):919-929. 

In This Article

Results

We identified 871 patients using AP and/or AC agents who were admitted for upper or lower GI bleeding. The median time of follow-up was 24.9 months (IQR: 7.0–38.0); 53.8% were men, the mean age was 78.9 ± 8.6 years and 91.7% of patients had a Charlson comorbidity index of 3 points or higher. The distribution of bleeding was 38.7% upper GI, 46.7% lower GI and 14.6% obscure bleeding; More than a third (38.9%; 339/871) of patients only used an AC and 52.5% (457/871) only used an AP; whereas 8.6% (75/871) used both AP and AC agents. Demographic data according to the type of drug consumed showed a difference in terms of gender, with a significantly higher proportion of men in the AP group (P < .001), but not age (P = .797) or the Charlson Index (P = .350) (Table 1).

At the time of admission, 93.1% of patients interrupted treatment and 80.5% of these patients restarted therapy within a mean of 7.63 ± 6.41 days (median of 6 days; IQR: 4.0–9.0). Most patients (98.5%) resumed therapy within the first 30 days. The mean time to restart APs (n = 327) was 7.30 ± 5.88 days (median of 5 days; range 1–35) compared with 7.30 ± 7.44 days (median of 7 days; range 1–64) for ACs (n = 264) and 6.26 ± 3.41 days (median of 6 days; range 1–15) for AP + AC. No differences were found when these variables were analysed according to location (upper GI bleeding [n = 244]: mean = 7.84 ± 7.31 days, median of 6, range 1–64; lower GI bleeding [n = 332]: mean 7.26 ± 5.54 days, median of 6, range 1–38; P = .867). The distribution of patients according to therapeutic management strategies and AP/AC use after the index bleeding is summarised in Figure 1. The clinical and demographic baseline characteristics of patients according to the adopted therapeutic strategy after drug interruption are summarised in Table 1. The causes of the initial bleeding events are described in Table 2.

Figure 1.

Flowchart for AP/AC therapy management after the index bleeding

More than half the patients (n = 546; 62.7%) developed a new significant event during the follow-up period; 17.8% (155/871) experienced a thromboembolic event (84.2 per 1000 patient-years [ptyears], 95% CI: 71.5–98.5); 24.9% (217/871) experienced a recurrent GI bleeding event (117.9 per 1000 pt-years, 95% CI: 102.7–134.6) and 28.0% (244/871) died (132.5 per 1000 pt-years, 95% CI: 116.4–150.3). The incidence rates of outcome events during follow-up were lower for the AP group than for the AC group, rebleeding rates were 99.0 (95% CI: 80.8–120.0) and 138.0 (95% CI: 111.1–169.4) events per 1000 pt-years for AP and AC users respectively. The corresponding event rates were 76.9 (95% CI: 61.0–95.7) and 94.0 (95% CI: 72.1–120.5) per 1000 pt-years for vascular events, and 115.3 (95% CI: 95.6–137.9) and 159.2 (95% CI: 130.2–192.7) per 1000 pt-years for deaths. The frequencies of the type of event are summarised in Table 3. Only 15.2% of patients had at least one event during the first 90 days after the index event. We did not observe significant differences according to the main outcomes of interest (GI rebleeding, death or ischaemic events) between patients who did (n = 811) or did not (n = 60) interrupt the AP and/or AC therapy after the index bleeding, since a minority did not interrupt.

In 61.8% (134/217) of patients who developed a rebleeding event, the location was the same as that identified in the index bleeding. The distribution of causes and locations, compared with the initial index bleeding episode, is summarised in Table 4. The PPI therapy did not affect the distribution of rebleeding event types, although there was clearly a reduction in the proportion of rebleeding events due to peptic ulcers.

Resumption vs no Resumption of Therapy

Of the 811 patients who interrupted therapy at the time of the index GI bleeding event, the Kaplan-Meier survival analysis over the total follow-up period showed significant differences between those patients who resumed vs those who did not resume therapy for the main outcomes of interest (GI rebleeding, ischaemic events or death) (Figure 2D-F). If we consider only the first 90 days after the event, those differences remained statistically significant for ischaemic events and death outcomes (Figure 2A-C). Overall, resumption of therapy was associated with a higher risk of rebleeding (HR 2.184; 95% CI: 1.357–3.515) but a lower risk of an ischaemic event (HR 0.626; 95% CI: 0.432–0.906) or death (HR 0.606; 0.453–0.804) in a multivariable analysis that controlled for gender, age, Charlson index and therapy use (Table 5). When patients were divided according to the type of drug used, patients taking ACs who restarted therapy had a significantly higher risk of rebleeding and lower risk of ischaemic events and death; however, AP users had a significantly lower risk of death but no differences in ischaemic and rebleeding events (Table 5). In Table S1, we report the distribution of the 155 CV events based on the primary indication and type of drug therapy. In Table S2, we report all causes of death in the 244 patients who died during follow-up. When patients were divided according to upper or lower GI bleeding, therapy results were generally similar, but those who had an initial lower GI bleeding event had more favourable outcomes than those with upper GI bleeding. (Table 5). In the specific group of patients with small bowel + obscure bleeding, the risk of the different outcomes was consistent with those observed for the overall lower GI bleeding. (Table 5). Patients on dual AP therapy had a lower risk of death compared with those who were not on this therapy (15.9% vs 29.3%; P = .009).

Figure 2.

Kaplan-Meier survival analysis according to resuming therapy status over 811 patients who interrupted therapy at the time of the index GI bleeding event. A, Kaplan-Meier survival analysis showing 90-day recurrent gastrointestinal bleeding events in patients who resumed vs those who did not resumed therapy (P = .171, log-rank test). B, Kaplan-Meier survival analysis showing 90-day thrombotic events in patients who resumed vs those who did not resumed therapy (P = .001, log-rank test). C, Kaplan-Meier survival analysis showing 90-day death rates in patients who resumed vs those who did not resumed therapy (P < .001, log-rank test). D, Kaplan-Meier survival analysis showing overall recurrent gastrointestinal bleeding events in patients who resumed vs those who did not resumed therapy (P = .001, log-rank test). E, Kaplan-Meier overall survival analysis showing thrombotic events in patients who resumed vs those who did not resumed therapy (P = .007, log-rank test). F, Kaplan-Meier overall survival analysis showing death rates in patients who resumed vs those who did not resumed therapy (P < .001, log-rank test)

Early vs Late Resumption of Therapy

Of the 653 patients who resumed either AP or AC treatment, early resumption of therapy (≤7 days) was significantly associated with a lower rate of ischaemic events (13.6% vs 20.4%; P = .025), but a higher rate of GI rebleeding (30.6% vs 23.1%; P = .044) during the follow-up period, with no statistical differences between groups in terms of death (Table 6). However, time-to-event analysis did not show significant differences between the two groups in ischaemic events (Table 6). The stratified analysis by type of drug therapy or type of initial bleeding event did not show significant differences (data not shown). A sensitivity analysis performed with different time points for early vs late resumption of therapy did not show differences and the most discriminant cut-off point was 7 days (Table S3).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....