Predictive Model for PICC Occlusion Risk for Patients in Intensive Care Units: A Retrospective Clinical Study.

Context: Peripherally inserted central catheters (PICCs) have a high incidence of catheter occlusion, but research exploring the risk factors for such an occlusion for patients in intensive care units (ICUs) is lacking.

Objective: The study intended to examine the impact of multiple risk factors on the occurrence of PICC catheter occlusion to find evidence that can help clinical medical staff identify patients at an early stage who are at high risk of a catheter occlusion.

Design: The research team performed a retrospective, observational clinical study.

Setting: The study took place at a tertiary general hospital, the Second Affiliated Hospital and Yuying Children's Hospital of Wenzhou Medical University in Wenzhou, China.

Participants: Participants were 300 patients with a PICC who received treatment in the hospital's adult ICU between January 2019 and April 2022.

Groups: According to the time of catheterization, the research team numbered the 1~300 participants and then selected one starting number to divided them into two groups according to the random number table. These two groups were: (1) a training group with 225 participants and (2) validation group with 75 participants.

Outcome Measures: The main outcome measure was the evaluation of the factors impacting patients who had had a PICC occlusion during catheter retention, including complete and incomplete occlusions, to build a risk prediction model of PICC occlusion. A secondary outcome measure was the occurrence of extubation of the PICC discharge of the ICU patient. The research team performed a univariate analysis of the training group's data and a multivariate logistic regression analysis of the risk factors. The team: (1) built a risk prediction model of PICC occlusion using the independent risk factors for catheter occlusion for PICC patients in an ICU and (2) used the Hosmer-Lemeshow goodness-of-fit test to test the prediction model. A two tailed using p>0.05 indicated that the model had a good fit. Then, the team applied the model to the validation group and evaluated the model's predictive ability using a receiver operating characteristic (ROC) curve. The team considered an area under the curve (AUC) >0.5 to have predictive value. The larger the area was, the better the predicted value was. The incidence of PICC occlusion in the training group was 18.22%, including 10 participants with complete occlusion and 31 with partial occlusion. The team used the SPSS 22.0 and R software for statistical analysis.

Results: The univariate analysis showed that 13 factors were associated with PICC occlusion, including: (1) an age ≥65 years (P < .001), a BMI of ≥24 kg/m2 (P < .001), (2) a BMI of ≥24kg/m2 (P = .002), (3) diabetes (P < .001), (4) stroke (P < .001), (5) hypertension (P < .001), (6) malignant tumors (P < .001), (7) a history of deep vein thrombosis (P < .001), (8) limb activity (P < .001), (10) flushing and sealing pipe frequency of Q8h (P = .035), (11) retention time (P < .001), (12) an increased platelet count (P = .036), (13) blood transfusions (P < .001), and (14) intravenous nutrition (P < .001). The independent risk factors for PICC occlusion included: (1) age ≥65 years-OR=1.224, P = .028; (2) BMI ≥24 kg/m2-OR=1.679, P = .004; (3) diabetes-OR=1.343, P = .017; (4) malignant tumors-OR=2.736, P < .001; (5) blood transfusions-OR=1.947, P < .001), and (6) intravenous nutrition-OR=2.021, P < .001. The frequency of flushing and sealing the pipe (Q8h)-OR=-2.145, P = .002-was a protective factor. In the training group, the area under the curve (AUC) for predicting a PICC occlusion was 0.917. The Hosmer-Lemeshow test of the prediction model showed that no significant differences existed in the test results within the model (χ2 = 5.830, P = .666), indicating that the model passed the internal validation. The ideal and calibration curves of the prediction model were highly coincident, and the model was well calibrated. The Hosmer-Lemeshow test of the validation group showed that no significant differences existed in the test results outside the model, suggesting that the model had high consistency.

Conclusions: Age ≥65 years, BMI ≥24 kg/m2, diabetes, malignant tumors, blood transfusions, and intravenous nutrition were independent risk factors for PICC occlusion, while the frequency of flushing and sealing pipe (Q8h) was a protective factor. This prediction model had an outstanding ability to discriminate in identifying patients with a high-risk of PICC occlusion in the ICU.