Drain Use is Associated with Increased Odds of Blood Transfusion in Total Shoulder Arthroplasty: A Population-Based Study

Abstract

Background: In the absence of evidence supporting its benefit, the American Academy of Orthopaedic Surgeons (AAOS) strongly recommends against closed wound drainage in TKA; however, drain usage remains common in other joints, including the shoulder. While an extensive body of research exists for drain use in lower extremity joint arthroplasty, large-scale data on drain use and its association with benefits and complications in shoulder arthroplasties is lacking. Such data may be particularly valuable given the rapidly increasing demand for shoulder arthroplasties.

Question/purpose: Using national claims data, we (1) evaluated the trends in frequency of drain usage in shoulder arthroplasty procedures over time, as well as the association between drain usage and (2) blood transfusion usage, (3) length of stay (LOS), and (4) readmission or early infection within 30 days.

Methods: This retrospective study used data from the nationwide Premier Healthcare claims database (2006-2016; n = 105,116, including total, reverse, and partial shoulder arthroplasties, in which drains were used in 20% [20,886] and no drain was used in 80% [84,230]). Included hospitals were mainly concentrated in the South (approximately 40%) with equal distributions among the Northeast, West, and Midwest (approximately 20% each). The Premier database contains detailed inpatient billing data on approximately 20% to 25% of US hospital discharges, which allows the creation of a variable indicating drain use by evaluating inpatient billing for drains. Baseline demographics differed minimally between patients receiving a drain compared with those who did not, with a median age of 70 years in both groups. The potential for selection bias was addressed by adjusting for measured confounders in mixed-effects models that estimated associations between drain use and blood transfusion usage, LOS, and readmission or (early) infection within 30 days. In addition, alternative statistical approaches were applied to address confounding, including propensity score analysis and instrumental variable analysis where a so-called "instrumental variable" is applied that mimics the treatment assignment process similar to a randomized study. We report odds ratios (OR; or % change for continuous variables) and 95% confidence intervals (CIs).

Results: The usage of drains decreased over time, from 1106 of 4503 (25%) in 2006 to 2278 of 14,501 (16%) in 2016. After adjusting for relevant covariates, drain use was associated with an increased usage of blood transfusions (OR, 1.49; 95% CI, 1.35-1.65; p < 0.001) while only associated with a small increase in LOS (+6%, 95% CI, +4% to +7%; p < 0.001). Drain use was not associated with increased odds for early postoperative infection or 30-day readmission. Propensity score analysis and instrumental variable analysis corroborated our main results.

Conclusions: Use of drains in patients undergoing shoulder arthroplasty is associated with an almost 50% increased odds for blood transfusions. Given that our findings parallel close to what is known in patients undergoing lower extremity joint arthroplasty, we believe that our results from a large national database are sufficient to discourage the routine use of drains in patients undergoing shoulder arthroplasty.

Level of evidence: Level III, therapeutic study.

Fig. 1

This figure shows trends in drain use by hospital and procedure type.