Midline catheters are associated with lower catheter-related bloodstream infection but higher premature removal compared with PICCs in ICU patients following central venous catheter removal: a retrospective cohort study

Midline catheter (MCs) and peripherally inserted central catheters (PICCs) are alternatives following central venous catheter (CVC) removal in intensive care unit (ICU) patients. Limited data exist regarding their comparative safety in critically ill patients from resource-limited settings. To compare safety and efficacy of MCs versus PICCs in ICU patients following CVC removal at a Phu Tho province Hospital. We conducted a retrospective cohort study of adult ICU patients receiving MCs or PICCs within 6 h of CVC removal (01/08/2022- 31/01/2025). Primary outcome was catheter-related bloodstream infection (CRBSI). Secondary outcomes included thromboembolism, mechanical complications, and premature removal. Propensity score matching (1:1 nearest-neighbor, 0.2 SD caliper) on baseline illness severity (SOFA, APACHE II) and comorbidities was performed to address confounding by indication. After propensity score matching, multivariable regression analyses were performed using Firth penalized logistic regression for rare outcomes, conventional logistic regression for any complication, and Cox proportional hazards regression for premature catheter removal. Sensitivity analyses accounted for the matched-pair structure (exact McNemar test, conditional logistic regression, clustered Cox regression), re-estimated the propensity score including catheter-placement indication, and examined premature removal by reason. Among 494 patients (247 per group; 7,647 total catheter-days), MCs demonstrated significantly lower rates of CRBSI compared to PICCs (1.2% vs. 4.9%; p = 0.032) and DVT (0.8% vs. 4.0%; p = 0.034). After propensity score matching (200 pairs), CRBSI remained lower with MCs (1.0% vs. 5.5%; OR 5.76 PICC vs. MC, 95% CI 1.26–26.34; p = 0.011), while the DVT association was attenuated and no longer statistically significant (1.0% vs. 4.0%; p = 0.055). Other complications were more frequent with midline catheters, including SVT (3% vs. 1.5%), phlebitis and infiltration (6% vs. 3%), mechanical complications (6.5% vs. 4.5%) all p > 0.05. Premature catheter removal occurred more frequently with MCs in catheter-day analyses (12.15 vs. 6.89 per 1,000 catheter-days; IRR 1.76, 95% CI 1.09–2.85, p = 0.019. After matching, MCs use was associated with a lower risk of CRBSI after adjustment for measured confounders (adjusted OR 0.20 MC vs. PICC, 95% CI 0.04–0.71; p = 0.011), whereas no significant association was observed for DVT or overall catheter-related complications. MCs were associated with a higher hazard of premature catheter removal (adjusted HR 1.94, 95% CI 1.12–3.37; p = 0.018). In matched-pair-aware analyses, the CRBSI association was not robust: the exact McNemar matched OR was 0.18 (95% CI 0.02–0.83; p = 0.022), but the adjusted conditional OR was 0.107 (95% CI 0.009–1.206; p = 0.071), based on 13 informative pairs; DVT was not significant ( p = 0.109 and 0.152). The expanded propensity score model (128 pairs) did not show significant differences in CRBSI ( p = 0.313) or DVT ( p = 0.090). Premature removal by reason did not differ between groups (15.5% vs. 11.0%; McNemar p = 0.243). In ICU patients following CVC removal, MC use was associated with a numerically lower risk of CRBSI compared to PICCs, with this association persisting after propensity score matching and multivariable adjustment for measured baseline confounders. This association was consistent in direction but not statistically robust: it lost significance in the covariate-adjusted matched-pair-aware and expanded propensity score analyses and may be affected by residual confounding by indication; a similar direction of association was observed for DVT, without statistical significance after matching. Higher premature removal with MCs was seen in catheter-day analyses but not when compared by reason in matched pairs, so it is sensitive to the analytic approach. Both catheter types showed substantially higher complication rates during short-term use (< 16 days); this pattern, and the lower rates observed beyond 16 days, should be interpreted cautiously given the potential for survivorship bias rather than a true protective effect of prolonged catheter dwell time. These findings are hypothesis-generating; catheter selection should be individualized, and larger prospective studies are needed. Not applicable.