Assessment of Risk for Recurrent Diverticulitis

Recurrence of acute diverticulitis is common, and—especially complicated recurrence—causes significant morbidity. To prevent recurrence, selected patients have been offered prophylactic sigmoid resection. However, as there is no tool to predict whose diverticulitis will recur and, in particular, who will have complicated recurrence, the indications for sigmoid resections have been variable. The objective of this study was to identify risk factors predicting recurrence of acute diverticulitis. This is a retrospective cohort study of patients presenting with computed tomography–confirmed acute diverticulitis and treated nonresectionally during 2006 to 2010. Risk factors for recurrence were identified using uniand multivariate Cox regression. A total of 512 patients were included. History of diverticulitis was an independent risk factor predicting uncomplicated recurrence of diverticulitis (1–2 earlier diverticulitis HR 1.6, 3 or more—HR 3.2). History of diverticulitis (HR 3.3), abscess (HR 6.2), and corticosteroid medication (HR 16.1) were independent risk factors for complicated recurrence. Based on regression coefficients, risk scoring was created: 1 point for history of diverticulitis, 2 points for abscess, and 3 points for corticosteroid medication. The risk score was unable to predict uncomplicated recurrence (AUC 0.48), but was able to predict complicated recurrence (AUC 0.80). Patients were further divided into low-risk (0–2 points) and high-risk (>2 points) groups. Low-risk and high-risk groups had 3% and 43% 5-year complicated recurrence rates, respectively. Risk for complicated recurrence of acute diverticulitis can be assessed using risk scoring. The risk for uncomplicated recurrence increases along with increasing number of previous diverticulitis. (Medicine 94(8):e557) Abbreviations: AUC = area under curve, CI = confidence interval, CT = computed tomography, HR = hazard ratio, SE = standard error. ppäniemi, MD, PhD, and Panu Mentula, MD, PhD morbidity and mortality involved in acute diverticulitis with lower stages having a rather benign course and higher stages having up to 32% mortality. Patients have been offered prophylactic sigmoid resection in order to prevent recurrence of acute diverticulitis and the concomitant morbidity associated with the higher stages. However, the indications for elective sigmoid resection have been highly variable. In history, two acute diverticulitis—or even one on a young patient—were considered an indication for a sigmoid resection. Recent evidence has shown, however, that the first diverticulitis is usually the most severe one, and the recurrent attacks are usually uncomplicated. This has been thought to be due to increased scar tissue around the diverticula, which would prevent or limit the perforation. Yet, small portion of patients will develop complicated recurrence, and prediction and prevention of that event are of paramount importance. Earlier studies have provided risk factors for recurrence, but clinical decision making has been difficult nevertheless. Current guidelines are unable to make strong recommendations for patient selection for prophylactic sigmoid resection. Aims of this study was to (1) identify independent risk factors predicting complicated recurrence of acute diverticulitis, and (2) form a simple score to be used in bed-side clinical setting to determine which patients are at high risk for complicated recurrence. METHODS This was a retrospective cohort study conducted at an academic teaching hospital that functions as both a secondary referral center and tertiary referral center serving a population of approximately 1.5 million. Data were gathered from a database established earlier, which comprise of all patient admissions due to acute colonic diverticulitis from 2006 to 2010. A total of 968 patient admissions were identified from the database. Patients without computed tomography (CT) verification were also excluded (n1⁄4 335) because clinical diagnostic accuracy is poor. Patients with obstruction (n1⁄4 17) or colon cancer mimicking diverticulitis (n1⁄4 17) were excluded. Further, patients who underwent emergency colonic resection were excluded (n1⁄4 76) because these patients are not at risk for recurrence. Readmissions within 30 days (n1⁄4 11) were considered an activation of the primary acute diverticulitis, and these were incorporated into index admissions. Data regarding patient demographics, comorbidities, medications, prior abdominal operations, history of acute diverticulitis, CT-scan findings at index admission, laboratory parameters at index admission, treatment at index admission, classification of acute diverticulitis at index admission, elective sigmoid resection during follow-up, as well as incidence and ing follow-up were extracted. verticulitis was defined as radiological ulitis (thickened bowel wall segment, www.md-journal.com | 1 TABLE 1. Basic Demographics of Patients Included in the Study Total n1⁄4 512 n (%) Sex (male/female) 191 / 321 (37%/63%) Age, years, median (range) 56 (22–95) Charlson comorbidity index, mean (range) 0.5 (0–7) Previous abdominal operations Appendectomy 113 (22%) Cholecystectomy 52 (10%) Hysterectomy 69 (14%) Colorectal resection 13 (3%) Medications Corticosteroid 18 (4%) Immunosuppressant 12 (2%) Chemotherapy 6 (1%) Anticoagulation 36 (7%) No of previous diverticulitis 0 384 (75%) 1 66 (13%) 2 25 (5%) 3 or more 37 (7%) Hinchey class (Wasvary et al) Hinchey 1a 323 (63%) Hinchey 1b 71 (14%) Hinchey 2 19 (4%) Hinchey 3 2 (0.4%) Hinchey 4 0 Hinchey unknown 97 (19%) Stage (Sallinen et al) Stage 1 408 (80%) Stage 2 66 (13%) Stage 3 35 (7%) Stage 4 2 (0.4%) Stage 5 1 (0.2%) Treatment of acute diverticulitis Medical, nonoperative 482 (94%) Medical and drainage of an abscess 22 (4%) Operative, but nonresectional 8 (2%) Follow-up, mo, median (range) 47 (21–84) Elective sigmoid resection in the 120 (23%) Medicine Volume 94, Number 8, February 2015 inflammation in pericolonic fat, and/or pericolic air bubbles) without distant intraor retroperitoneal air, abscess, fistula, obstruction, or peritonitis (Stage 1 diverticulitis). Complicated diverticulitis is defined as radiological signs of acute diverticulitis with an abscess, fistula, obstruction, peritonitis, large amount of fluid in the abdominal cavity, and/or distant intraor retroperitoneal air (Stage 2 or higher). Distribution of extraluminal air has been defined earlier, and showed to have an effect on treatment and morbidity. Recurrence of diverticulitis was defined as a CT-verified diverticulitis or clinical diverticulitis (fever, elevated C-reactive protein levels, left lower abdominal pain) with treatment (per oral or intravenous antibiotics). Recurrences were diagnosed and treated both as inand outpatient and at the hospital and general practitioners visits. To gather follow-up data, electronical patient records were analyzed from our hospital, as well as from the family practitioners/health care centers at the Helsinki area. Follow-up was determined as the time from discharge to the examination of the patient records. Statistical analyses were made using SPSS Statistics v. 21 (IBM, Armonk, NY). Cox univariate regression was used for univariate analyses. Continuous variables that had P< 0.20 were further modified to obtain categorical variables, which could be inserted into multivariate analyses. Variables that had P< 0.20 were considered for multivariate analysis. If variables were multicollinear, one that was most easily accessible in terms of clinical use and/or showed highest hazard ratio was selected. Multivariate analyses were performed using Cox regression analysis with forward likelihood ratio method. Kaplan–Meier with log-rank test was used to analyze recurrence-free time, with sensoring for death, loss of follow-up, or elective sigmoid resection. Appropriate permissions to conduct the study were obtained from the hospital and Helsinki health care centers’ institutional review boards.