Patient selection bias and inappropriate stent selection leading to false conclusions in the comparison of metal and plastic stents

In a recent article, Chung et al. compared the effectiveness of self-expandable metallic stents (SEMSs) with plastic stents in a cohort of patients with hepatocellular carcinoma [1]. They claimed that patient survival was better with plastic stents and that stent patency duration was similar between plastic stents and SEMSs.

There are two gray areas in this statement: the first relates to the selected SEMS types; the second, owing to the retrospective study design, relates to the nonrandomized composition of the patient groups, which leads to a patient selection bias in the SEMS group. 

The SEMS types used in the study were uncovered Bonastents or uncovered Niti-S biliary stents. The diameters of the SEMSs were 10 mm and the stent length was 5 – 10 cm. The plastic stents were Percuflex Amsterdam or Cotton–Leung (Amsterdam) biliary stents of 5 – 15 cm in length. It is not mentioned why the uncovered SEMSs were chosen, but uncovered biliary stents are prone to late occlusion due to tumor ingrowth or overgrowth at the ends [2] [3]. Therefore, the selection of fully covered SEMSs would have been more appropriate.

Second, as listed in table 1 of the article [1], the SEMS group consisted of patients with higher TNM classifications (stage IV, 50 % vs. 33 %, respectively) and more portal vein invasion (58.3 % vs. 38.3 %, respectively). Additionally, plastic stent-treated patients had more anticancer treatment than the SEMS-treated patients (51.7 % vs. 30.6 %, respectively). Although these comparisons were not significantly different, there is a selection bias in the SEMS patient group. Even though it is stated that there was no statistical significance, the P value of 0.057 for the portal vein invasion comparison is actually close to being significant. This might become a significant result if the patient number in the SEMS group (and therefore the statistical power of the study) were increased.

In the end, the authors note the limitations of the study: the small number of patients and nonrandomized stent selection. Because there was no randomization in stent selection in this retrospective study, we cannot rely solely on the statistical results, which could lead us to false answers.

• References

• 1 Chung KH, Lee SH, Park JM et al. Self-expandable metallic stents vs. plastic stents for endoscopic biliary drainage in hepatocellular carcinoma. Endoscopy 2015; 47: 508-516
• 2 Isayama H, Komatsu Y, Tsujino T et al. A prospective randomised study of ‘‘covered’’ versus ‘‘uncovered’’ diamond stents for the management of distal malignant biliary obstruction. Gut 2004; 53: 729-734
• 3 Petersen BT, Kahaleh M, Kozarek RA et al. A multicenter, prospective study of a new fully covered expandable metal biliary stent for the palliative treatment of malignant bile duct obstruction. Gastroenterol Res Pract 2013; 2013: 642428