Should We Use the Model for End-Stage Liver Disease (MELD) to Predict Mortality After Colorectal Surgery?

Abstract

We sought to determine the accuracy of the Model for End-Stage Liver Disease and the Mayo Clinic Postoperative Mortality Risk in Patients with Cirrhosis Calculator in patients with ascites who underwent colorectal surgery. The National Surgical Quality Improvement Program database was queried for patients with ascites who underwent a major colorectal operation. Predicted 90-day mortality rate based on the Model for End-Stage Liver Disease and 30-day mortality based on the Mayo Clinic Postoperative Mortality Risk in Patients with Cirrhosis Calculator were compared with observed 30-day mortality. The cohort contained 3137 patients with ascites who underwent a colorectal operation. The Model for End-Stage Liver Disease predicted that 252 (8 %) of patients with ascites undergoing colorectal operations would die within 90 days postoperatively, yet we observed 821 deaths (26 % mortality) within 30 days after surgery (p < 0.001). The Mayo Clinic Postoperative Mortality Risk in Patients with Cirrhosis Calculator predicted that 491 (16.6 % mortality) of patients with ascites undergoing colorectal operations would die within 30 days postoperatively, yet we observed 707 (23.9 % mortality) at 30 days (p < 0.01). We concluded that the current risk prediction models significantly under predict mortality in patients with ascites who underwent colorectal surgery.

Model for End‐Stage Liver Disease Underestimates Morbidity and Mortality in Patients with Ascites Undergoing Colectomy

“…While cirrhosis is not a direct variable in NSQIP, the variable of ascites is specific to liver disease, and specifically excludes malignant ascites. The presence of ascites as a marker for cirrhosis is previously reported [36, 38], and within NSQIP has generally been accepted as a sound substitute for cirrhosis, helping to identify patients with chronic liver disease for the study of surgical outcomes [9, 19].…”

“…Unlike previous studies, we investigated the effect of ascites on risk prediction for patients with chronic liver disease and compared patients of similar MELD stratum differentiated only by the presence of ascites. Other studies have also shown associations of increased postoperative risk with the presence of ascites [9, 24, 36] and that MELD underestimated risk for postoperative mortality for significant portion of patients (~ 25%) undergoing liver transplantation [23]. While it is not entirely clear the etiology of increased complications and death in patients with ascites, some theories exist.…”

Colorectal Surgery in Patients with Liver Cirrhosis: A Systematic Review

“…There was no uniform method for confirming diagnosis of liver cirrhosis. Five studies were based on coding of nationwide databases [12,13,[16][17][18]. The coding criteria differed between them.…”

“…There was heterogenous reporting of mortality rates which was available in only ten studies [5,10,11,13,16,[18][19][20][21][22] (Table 4). Rates ranged from 0 to 37%.…”

“…A total of 16 studies were analysed (with 8648 patients included) (Table 1) [4,5,[10][11][12][13][14][15][16][17][18][19][20][21][22][23] after excluding four studies due to mixture of cirrhosis and non-cirrhosis or chronic liver disease [24][25][26][27]. Of the 8648 patients, there were 4620 males and 4026 females.…”