Rebleeding and survival after acute lower gastrointestinal bleeding

Transcription

1 The American Journal of Surgery 188 (2004) Scientific paper Rebleeding and survival after acute lower gastrointestinal bleeding Thomas Anthony, M.D.*, Pradeep Penta, B.S., Robert D. Todd, B.S., George A. Sarosi, M.D., Fiemu Nwariaku, M.D., Robert V. Rege, M.D. Department of Surgery, University of Texas Southwestern Medical Center and the VA North Texas Health Care System, 4500 South Lancaster Rd., Dallas, TX 75216, USA Manuscript received June 10, 2004; revised manuscript July 7, 2004 Presented at the 28th Annual Symposium of the Association of VA Surgeons, Richmond, Virginia, April 25 27, 2004 Abstract Background: Previous studies of acute lower gastrointestinal bleeding (LGIB) have focused on evaluation and therapy. Measurement of long-term outcome has been rare. The purpose of this study was to document rebleeding and survival rates in patients with acute LGIB. Methods: A retrospective review of all patients undergoing technetium-labeled red blood cell scans for LGIB from January of 1997 to December of 2002 was performed. Rebleeding was defined as identification of enteric bleeding requiring a transfusion 2 or more weeks after the initial bleeding episode. Results: A total of 119 patients met inclusion criteria. Rebleeding was documented in 14 of 102 patients surviving for more than 2 weeks. The actuarial rebleeding rate was 15% at 2 years. No factors were identified that portended a higher likelihood of rebleeding. The 30-day mortality was 18% and the median survival was 60 months for the entire cohort. Of the 36 patients in whom cause of death was documented, 4 died of surgical complications and a single patient died as a direct result of hemorrhage. Conclusions: Rebleeding after an initial episode of LGIB occurs in a small percentage of individuals. Although survival is poor for patients with LGIB, few patients die as a direct consequence of hemorrhage Excerpta Medica Inc. All rights reserved. Keywords: Lower gastrointestinal bleeding; Rebleeding; Survival Despite the frequency with which patients are admitted, evaluated, and treated for lower gastrointestinal bleeding (LGIB), there are few reports of long-term outcomes for these individuals. Most prior reports have stressed the evaluation process, including tests available for localization and their characteristics (sensitivity, specificity, and so forth). Longitudinal studies of outcomes are rare for patients who have experienced acute LGIB. Knowledge of long-term survival and rebleeding rates would be beneficial in defining the efficacy of therapy. The actual rate of rebleeding after untreated LGIB attributed to diverticular disease may be as high as 25% at 4 years [1]. Several factors may contribute to the likelihood of a patient rebleeding after an initial episode * Corresponding author. Tel.: ; fax: address: Thomas.Anthony@med.va.gov of acute LGIB, including underlying comorbid diseases known to increase the chance of bleeding in general (end-stage renal disease and cirrhosis), medication use (nonsteroidal anti-inflammatory agents, antiplatelet agents, or anticoagulants), and therapy used to treat the initial episode of bleeding [2 6]. However, the relative contributions of each of these factors in determining the ultimate rate of rebleeding is not clear. Mortality associated with LGIB also is characterized poorly. Reported rates have varied from 1% to 10% (1,7 9). Because patients undergoing evaluation for acute LGIB typically have a number of comorbid conditions that may contribute to mortality rates, it is extremely important that process-specific mortality be considered. Prior studies that have considered process-specific survival have suggested that few patients actually died as a direct result of LGIB [1]. The purpose of the present study was to document the rebleeding rate after an initial episode of bleeding, identify /04/$ see front matter 2004 Excerpta Medica Inc. All rights reserved. doi: /j.amjsurg

2 486 T. Anthony et al. / The American Journal of Surgery 188 (2004) Table 1 Cohort description Mean age (years) 70 (45 88) African American race 38/119 (32%) Hematochezia 93/105 (89%) NSAID use 63/119 (53%) Anticoagulant use 22/119 (18%) Renal insufficency/failure 31/119 (26%) Cirrhosis 8/119 (7%) NSAID, anticoagulant, renal insuffiency, 89/119 (75%) or cirrhosis Mean pulse (beats/min) 83 (47 125) Orthostasis 69/103 (58%) Hypotension 30/111 (25%) Mean HCT 31 (17% to 47%) Mean platelet count 247,000 (30,000 to 719,000) Mean PT 15.6 (11 32) Mean PTT (seconds) 31 (19 118) Median number of units transfused (U) 3 (0 25) All factors recorded at the time of initial LGIB episode. NSAID Nonsteroidal anti-inflammatory drug; HCT hematocrit; PT prothrombin time; PTT partial thromboplastin time. factors associated with rebleeding, and to define overall and disease-specific survival rates in patients with acute LGIB. Patients and Methods An institutional review board approved, retrospective cohort study of LGIB patients was performed at a single VA tertiary-care institution. The population for this study was drawn from all patients who underwent technetium-99 labeled red blood cell (TLRBC) scans for presumed lower GI related bleeding between January 1, 1997, and December 31, This method of surveying the population of patients was chosen because most patients with acute significant LGIB at this institution are evaluated with a TLRBC scan. Patients were identified accordingly via the use of the Nuclear Medicine Department caseload summaries for this period. After confirmation of the performance of TLRBC scanning, patients were cross-referenced with a computerized medical record system to obtain additonal information. Patients in whom TLRBC was performed for upper GI bleeding or in whom a source was identified proximal to the ligament of Treitz were excluded. Clinical and laboratory parameters recorded from patient records examined in this study were documented during the emergency room visit or during the initial evaluation of the patient. The variables assessed included the following: patient age; sex; history of hematochezia or bright red blood per rectum; evidence of volume depletion (low resting blood pressure, orthostatic hypotension, and high pulse rate); the presence of bright red blood per rectum physical examination; prescan prothrombin and partial thromboplastin time; initial hematocrit and platelet count; indication for TLRBC scan; use and number of transfusions before the scan; use of angiography, colonoscopy, or other diagnostic modalities; treatment by surgery (including type of resection); angiographic embolization; potential bleeding diatheses (cirrhosis, chronic renal failure); and use of nonsteroidal anti-inflammatory medications, other antiplatelet agents, or anticoagulants at the time of evaluation. The follow-up time was calculated as the time elapsed between first presentation with LGIB and the recorded date of last contact documented in the medical record. Rebleeding was defined as the receipt of 1 or more red blood cell transfusions 2 or more weeks after the evaluation for LGIB and evidence of a GI source. Statistics The univariate association between clinical, laboratory, and treatment variables and rebleeding was examined using Pearson chi-squared analysis for categoric variables and Student t test for continuous variables. Actuarial survival and rebleeding curves and time to rebleeding were calculated by the Kaplan-Meier method. Statistical separation of survival curves was assessed by the log-rank method. All statistics were performed using SPSS 11.0 (Chicago, IL). Results A total of 119 patients underwent TLRBC scans over the period of time evaluated in this study. The vast majority of patients were men (117 patients; 97%), with an average age of 70 years (range years). The most common indication for performing a TLRBC scan was bright red blood per rectum, which was present in 102 (86%) patients. Ninety-nine patients (83%) had undergone transfusion before TLRBC scanning. The median number of units transfused was 3 (range 0 25). Additional pretreatment information is recorded in Table 1. Fig. 1 documents the disposition and diagnostic tests used in the 119 patients who underwent TLRBC for acute LGIB. Of the 119 TLRBC scans, 51 (43%) were determined to be positive. Additional tests In addition to TLRBC scans, the majority of the patients in this cohort (100 patients, 84%), received additional diagnostic tests. There was no difference in the frequency of utilization of additional tests between patients with positive and negative TLRBC scans (82% versus 85%; P 0.67). Of the 51 positive scans, therapy was undertaken in 25 (50%). From the 68 negative studies, despite the large number of additional tests used, only 1 patient underwent further therapy (1.4%, P compared with positive TLRBC scans). Of the entire cohort, 93 (78%) patients underwent endoscopy, most commonly colonoscopy (46 patients) or colonoscopy combined with esophagogastroduodenoscopy (29 patients). There were 18 angiograms performed; embolization was undertaken during 2 of these procedures. Overall, in only 19 (16%) cases did patients undergo TLRBC

3 T. Anthony et al. / The American Journal of Surgery 188 (2004) Fig. 1. Diagnosis and treatment of 119 patients with acute LGIB. scan without additional endoscopy or angiography; 7 patients went directly to surgery after a positive localized TLRBC scan. In 10 cases in which TLRBC scans were negative, patients were referred for interval endoscopy; in 2 cases in which TLRBC scans were positive, no therapy and no further testing was undertaken. One of these patients suffered a massive aspiration of oral bowel preparation and died, the second patient refused further diagnostic evaluation or therapeutic intervention and died as a result of exsanguination. Initial treatment A total of 26 (22%) patients underwent surgical or angiographic therapy directed at the site of their LGIB. Seven surgeries were based solely on findings from TLRBC scans. The additional 19 procedures were directed by additional tests. The 30-day mortality for surgically or interventionally treated patients was 4 of 26 (15%). The sites of localization Table 2 Surgical procedures performed for treatment of acute LGIB Procedure Hemorroidectomy 1 Right hemicolectomy 10 Left hemicolectomy 4 Sigmoid colectomy 5 Total abdominal colectomy 4 Small bowel resection 1 n and the therapies used are listed in Table 2. The pathology responsible for bleeding is summarized in Table 3. Rebleeding Of the patients who survived their initial episode of LGIB by at least 2 weeks (102 patients), with a median follow-up period of 16 months, 14 patients (14%) experienced gastrointestinal rebleeding. Of these 14 patients, 6 had initial TLRBC scans that were positive; 1 patient had undergone angiographic embolization and surgery that attempted to address his initial episode of LGIB and 5 patients had failed to localize despite a positive TLRBC and additional tests. The remaining 8 patients who experienced rebleeding had negative TLRBC scans at initial evaluation. Thirteen patients who rebled had undergone at least 1 test in addition to their initial TLRBC scan (most commonly colonoscopy). The remaining patient had refused further diagnostic testing. Table 3 Cause of LGIB in 26 patients requiring intervention Pathology Diverticular disease 14 Angiodysplasia 3 Ischemic colitis 3 Bleeding after polypectomy 2 Cytomegalovirus colitis 1 Colonic varicies 1 Internal hemorroids 1 n

4 488 T. Anthony et al. / The American Journal of Surgery 188 (2004) Fig. 3. Overall survival: 119 patients with acute lower gastrointestinal bleeding. Fig. 2. Actuarial rebleeding rate. Over 50% of patients who experienced rebleeding from a GI source did so within 1 month (median time to rebleeding 3 wk), and over 80% within 12 months of the initial TLRBC scan. The actuarial rebleeding rate was 9% at 1 month after the initial bleeding episode, 13% at 1 year, and 15% at 2 years. Actuarial time to rebleeding is shown in Fig. 2. At the time of their second bleeding episodes, endoscopy was used most frequently as the initial diagnostic test (5 esophagogastroduodenoscopy and 5 colonoscopic evaluations). Two patients underwent angiography, and a second TLRBC scan was performed for 2 individuals. The source of the GI bleeding was identified in all but 3 cases. Diverticular disease (4 patients), esophageal varices (1 patient), esophagitis (1 patient), gastric varices (1 patient), gastrointestinal stromal tumor of the ileum (1 patient), colonic ischemia (1 patient), rectal ulcer after hemorrhoidectomy (1 patient), and colonic angiodysplasia (1 patient) were defined by either endoscopic visualization or surgical pathology. Thus, 8 of the 11 cases of rebleeding in which pathology could be defined had a source distal to the ligament of Treitz. Patients with rebleeding were treated by surgical intervention in 3 cases, embolization in 1 case, direct coagulation during colonoscopy in 1 case, and by esophageal banding in 1 case. All other patients [8] were managed expectantly. Six patients who rebled have died. In only 1 case did this occur within a month of the second episode of bleeding. This patient died of multisystem organ failure after a subtotal colectomy. No additional deaths could be attributed to hemorrhage. No patient has rebled a third time (median follow-up after rebleeding 8 mo). In an attempt to identify variables that were significantly associated with rebleeding, an analysis was performed considering potential sources that would result in an increased likelihood of a second episode of bleeding. Specific comorbid conditions (end-stage renal disease and cirrhosis), medication use (nonsteroidal anti-inflammatory drugs, antiplatelet agents, or anticoagulants), and localization of therapy for the initial episode of LGIB were considered in this analysis. Although there was a trend toward decreased rebleeding if the site of bleeding was localized during the initial episode and a specific intervention was used, this factor did not reach statistical significance. No other variable was associated with an increased rate of rebleeding. These results are summarized in Table 4. Survival A total of 22 patients survived fewer than 30 days (30-day mortality rate 18%); 4 of these patients underwent surgery or surgery and embolization. Overall, the 5-year actuarial survival was 49% for the entire group of patients with acute LGIB (median survival 60 mo). Fig. 3 shows the actuarial overall Table 4 Factors associated with rebleeding in 102 patients surviving by two weeks or more after initial episode of LGIB Factor as recorded at initial presentation Rebleeding from gastrointestinal source if factor present Rebleeding from gastrointestinal source if factor absent P value NSAID use 7/56 (13%) 7/46 (15%) 0.69 Anticoagulant use 1/20 (5%) 13/82 (16%) 0.21 NSAID, anticoagulant, renal insufficency, or cirrhosis 11/80 (14%) 3/22 (14%) 0.99 Initial TLRBC scan positive 6/46 (14%) 8/56 (13%) 0.85 Additional testing after initial TLRBC scan 13/88 (15%) 1/14 (7%) 0.34 Surgery or angiographic embolization 1/25 (4%) 13/77 (17%) 0.10 NSAID nonsteroidal anti-inflammatory drug.

5 T. Anthony et al. / The American Journal of Surgery 188 (2004) Fig. 4. (A) Overall survival: initial technetium-labeled red blood cell scan (positive vs. negative). (B) Overall survival: invasive therapy (surgery or angiography) for initial bleeding episode versus none. survival curves for the entire cohort. In all, 40 deaths have occurred within the follow-up period of this study; the cause of death was available in 36 cases (90%). The most common reasons for death were sepsis (8 patients), myocardial dysfunction (8 patients), and metastatic cancer (6 patients). Apart from the 4 cases in which patients died after surgery or surgery and embolization, only 1 additional patient died as a direct result of bleeding (this patient refused surgery). Fig. 4A shows the relationship between an initial TLRBC scan status and survival. Fig. 4B shows the relationship between the use of invasive therapy versus expectant management and survival. There was no significant difference in survival if the initial TLRBC scan was positive versus if the initial TLRBC scan was negative (P 0.88), or in patients receiving surgical or angiographic treatment versus those not receiving invasive treatment (P 0.67). Similarly, 30-day survival rates were not different between undergoing interventions versus those treated expectantly (85% versus 81%, P 0.65). A separate analysis censuring patients surviving fewer than 30 days did not identify a survival advantage for patients treated surgically/interventionally compared with patients treated noninvasively (P 0.67). Comments GI bleeding distal to the ligament of Treitz is a commonly encountered problem. The estimated incidence of the need for hospitalization owing to LGIB is approximately 20 per 100,000 patients, but increases dramatically as the population considered ages. For example, the incident rate increases to 205 per 100,000 for patients 80 years of age or older [1]. Although much literature has focused on the optimal strategy for evaluation and treatment of initial episodes of LGIB, much less is known concerning the longterm outcomes for these patients. In particular, little information exists concerning the fate of patients in whom the initial bleeding episode is nonlocalized and therefore untreated. How frequently these patients rebleed and how often mortality can be traced directly to rebleeding rarely has been addressed. This study sought to answer these questions and additionally to address the issue of whether specific treatment of the initial episode of LGIB resulted in a decreased rate of rebleeding and secondarily if there were other specific factors that could be identified that led to an increased likelihood of rebleeding. By using objective criteria to define rebleeding (transfusion of 1 U of blood 2 or more weeks after initial evaluation for LGIB and evidence of a GI source), the overall rebleeding rate in this cohort was 14%. Actuarial rebleeding rates were 9% at 1 month and 13% at 1 year. The rate of rebleeding appears to plateau after 2 years at approximately 15%. These findings are consistent with the findings of Longstreth [1], who documented a 9% rate of rebleeding at 1 year and a 10% rate of rebleeding at 2 years in patients with diverticulosis who were not treated after an initial episode of LGIB. In the present study, over half of the patients who experienced rebleeding did so within the first month, and by 1 year over 80% had done so. Because of the low rate of rebleeding for the entire cohort it was not possible to identify an approach to the primary bleeding episode that statistically reduced the likelihood of rebleeding, however, only 1 patient (4%) who had undergone surgical or interventional therapy subsequently had an additional episode of GI bleeding. It is also interesting to note that the majority of patients with multiple bleeding episodes eventually are localized and given a diagnosis (78%), and that in approximately one fourth of these cases the bleeding source proves to be proximal to the ligament of Treitz. However, as was the case with primary acute LGIB, diverticulosis appears to be the most common cause of secondary episodes of acute LGIB. Taken together, these results suggest that the incidence of rebleeding is relatively low and treatment success or failure (in terms of prevention of rebleeding) can be judged quickly. Occurrences of rebleeding can be localized accurately at a rate at least equivalent to primary episodes of LGIB bleeding. Although pathology distal to the ligament of Treitz is the most likely source of the rebleeding, the upper GI system should be considered if bleeding remains nonlocalized. Consistent with previous studies, although mortality after an episode of LGIB is substantial, rarely is death caused by uncontrolled hemorrhage [1]. Examining causes of death within 30 days of the initial acute LGIB, the mortality from hemorrhage was found to be exceedingly low. In our series

6 490 T. Anthony et al. / The American Journal of Surgery 188 (2004) a single patient (who refused intervention) died of exsanguination. Because acute LGIB rarely leads directly to mortality, therapy directed at the source of the bleeding should be carefully and thoughtfully undertaken. In patients whose hemodynamic stability can be maintained, therapy for nonlocalized disease (such as subtotal colectomy) probably should be avoided. Continued ongoing bleeding (requiring 4 U) that is well localized should be dealt with expeditiously by directed intervention [9,10]. Such directed therapy, although it does not impact survival, seems to result in the lowest rate of rebleeding [10]. Despite the relatively large size of the cohort included in this study, there were several limitations. This series of patients was enrolled from a single institution and reflects a number of factors associated with this institution. The most obvious example of this was a predominance of elderly, male patients, many of whom had numerous and serious medical comorbidities. These factors no doubt are reflected in the mortality and rebleeding rates reported in this study. Second, this study was limited in its ability to identify factors that led to an increased likelihood of rebleeding. The small number of patients who experienced second episodes of rebleeding did not allow for in-depth analysis of risk factors that might have helped identify a subset of the population at greater risk for bleeding. Although these limitations are significant, this study remains important in its definition of rebleeding rates and survival after an initial episode of LGIB in the context of our population of patients and our diagnostic and treatment algorithm. As technology to localize acute LGIB improves and additional diagnostic and therapeutic algorithms are developed, it will be imperative to prospectively document their effect on rebleeding and survival. References [1] Longstreth GF. Epidemiology and outcome of patients hospitalized with acute lower gastrointestinal hemorrhage: a population-based study [see comment]. Am J Gastroenterol 1997;92: [2] Wilcox CM, Clark WS. Association of nonsteroidal antiinflammatory drugs with outcome in upper and lower gastrointestinal bleeding. Dig Dis Sci 1997;42: [3] Holt S, Rigoglioso V, Sidhu M, et al. Nonsteroidal antiinflammatory drugs and lower gastrointestinal bleeding. Dig Dis Sci 1993;38: [4] Foutch PG. Diverticular bleeding: are nonsteroidal anti-inflammatory drugs risk factors for hemorrhage and can colonoscopy predict outcome for patients? Am J Gastroenterol 1995;90: [5] Farner R, Lichliter W, Kuhn J, Fisher T. Total colectomy versus limited colonic resection for acute lower gastrointestinal bleeding. Am J Surg 1999;178: [6] Cryer B, Kimmey MB. Gastrointestinal side effects of nonsteroidal anti-inflammatory drugs. Am J Med 1998;105:20S 30S. [7] Bramley PN, Masson JW, McKnight G, Herd K, Fraser A, Park K, et al. The role of an open-access bleeding unit in the management of colonic haemorrhage. A 2-year prospective study. Scand J Gastroenterol 1996;31: [8] Richter JM, Christensen MR, Kaplan LM, Nishioka NS. Effectiveness of current technology in the diagnosis and management of lower gastrointestinal hemorrhage [see comment]. Gastrointest Endosc 1995;41:93 8. [9] Vernava AM III, Moore BA, Longo WE, Johnson FE. Lower gastrointestinal bleeding. Dis Colon Rectum 1997;40: [10] McGuire HH Jr. Bleeding colonic diverticula. A reappraisal of natural history and management. Ann Surg 1994;220:653 6.