Abstract No. 528 Modeling socioeconomic and demographic variables affecting inferior vena cava filter placement and retrieval: a single-institution retrospective review at a safety net hospital from 2012 to 2019

Abstract

This study was conducted to investigate the factors affecting IVC filter retrieval rates at a county hospital with the intent to supplement guidelines for IVC filter placement. We conducted a retrospective review of all patients who underwent IVC filter placement at our institution between January 2012 and December 2019. During the study period 539 patients had IVC filters placed; of those, 405 (148 women, 257 men, median age 51 years), had sufficient data and were over the age of 18. The following variables were collected: age, gender, BMI, ZIP code, marital status, insurance type, primary language spoken, length of hospital stay, indication type (classic, expanded, or prophylactic)1, presence of trauma, discharge destination (home, long-term nursing, or other), date of IVC filter retrieval (if applicable), and the presence of risk factors (smoking history, alcohol use, hypertension, diabetes, venous thromboembolism). The patients’ ZIP codes were used in conjunction with US census data on median household income by ZIP code to estimate their socioeconomic status (SES). SES was log transformed to correct for non-normality. Variables that showed a significant difference between those who had their IVC filters removed and those who did not were tested for inclusion in a multivariate logistic regression model. Inclusion criteria was based on statistical significance at α ≤ .05 for each independent variable. All insurance data were included as dummy variables in a set with “uninsured” as the reference category. Inclusion of these covariates in this model predicts 31.5% of the variation in IVC filter retrieval rate (Nagelkerke R-squared = .315). The Table displays the patient characteristics that statistically significantly contribute (α ≤ .05) to predicting IVC filter retrieval rates. Various patient characteristics including demographics, comorbidities, insurance coverage, and clinical indications are predictive of IVC filter retrieval rates. These must be considered when developing protocol on IVC filter placement indications at institutions with certain patient populations, such as a county hospital.Tabled 1Odds Ratio95% Confidence IntervalP Value Age0.980.97–0.99.011 Ln SES2.21.15–4.18.017 Trauma3.11.74–5.4.000 Diabetes0.310.15–0.63.001 VTE0.530.31–0.9.018Insurance Commercial or self-pay1.970.94–4.12.072 Medicaid2.11.0–4.3.049 Medicare1.10.43–2.6.921Model AUC0.80.76–0.85Nagelkerke R-squared0.315AUC = area under the receiver operating characteristic curve; SES = estimated median household income. *** Open table in a new tab