BackgroundA 30-40 cm Roux loop is widely accepted in Kasai portoenterostomy in infants with Ohi Type III biliary atresia (BA), regardless of patient's size, to avoid reflux cholangitis. However, the redundant length of the jejunal limb may lead to complications; hence, we compared the outcome of a shorter Roux loop with the standard loop length in Kasai portoenterostomy in infants with Ohi Type III BA. MethodsThere were 166 patients with Ohi Type III BA who underwent the Kasai procedure and were prospectively randomized into two groups: (1) the long Roux loop group (LRLG) contained 91 patients, where a standard 30-40 cm Roux loop was used regardless of the child's size and (2) the short Roux loop group (SRLG) included 75 patients in which the Roux loop length was based on the distance between hepatic hilum and umbilicus. The follow-up periods were 3 to 50 mo. Contrast-enhanced ultrasonography of the upper gastrointestinal tract was studied, and laboratory tests were included during the follow-up period. ResultThe mean Roux loop length of SRLG was 17.42 ± 3.91 cm (13-20 cm); it was significantly shorter than that of the LRLG (32.87 ± 3.91 cm [30-40 cm], P = 0.00). There were no significant differences between the LRLG and SRLG in terms of age, body weight at time of the operation, length of operation, blood loss during the operation, and length of postoperative hospital stay (64.50 ± 20.41 d versus 68.34 ± 17.59 d, 4.3 ± 0.8 kg versus 4.1 ± 1.2 kg, 187.6 ± 35.7 min versus 172.3 ± 20.4 min, 21.71 ± 9.41 mL versus 25.32 ± 10.82 mL, 9.01 ± 3.97 d versus 8.16 ± 2.40 d; P = 0.63, 0.47, 0.09, 0.89, and 0.61, respectively). Six months after the operation, the clearance of jaundice rate in the LRLG was 45.1% (41/91), while the rate was 50.7% (38/75) in the SRLG; there was no statistically significant difference (P = 0.47). The incidence of cholangitis in the LRLG was 42.9% (39/91), which was similar to that of the SRLG (46.7%, 35/75) (P = 0.62). The incidence of overall complications (such as intestinal obstruction, gastrointestinal bleeding, bile leak, and anastomotic stenosis) in the SRLG was lower than that of the LRLG, but the difference did not reach statistical significance (4/91, 4.4% versus 5/75, 6.7%; P = 0.76). According to the Kaplan–Meier survival analysis and log rank test, the native liver survival rate in the LRLG and the SRLG also showed no significant difference (P = 0.964, log rank test). Grade I reflux was detected in three patients in the LRLG and two patients in the SRLG (P = 0.54). However, the reflux subsided 6 months later in all five patients. ConclusionsA shorter Roux-loop length in Kasai portoenterostomy reconstruction had a similar efficacy as the conventional Roux-loop length.