Trends in inferior vena cava filter placement and retrieval at a tertiary care institution

Objective: The aim of this study was to examine practice patterns of inferior vena cava (IVC) filter insertion and retrieval at a tertiary care institution. Methods: A retrospective review of all IVC filter procedures performed at the University of Pennsylvania and entered into the Penn cohort of the Vascular Quality Initiative registry between January 2013 and September 2017 was performed. Data collected included demographics, venous thromboembolism risk factors, indications for filter placement, and presence and timing of retrieval. Trend analysis and multivariable logistic regression were performed to evaluate factors associated with failure to retrieve the filter. Results: During the study period, 627 IVC filters were inserted. The mean age was 52.8 +/- 16.9 years, and 49.3% were male; 39.2% were placed for a major indication, whereas 58.1% were placed for prophylaxis. There was a significant decline in overall frequency of filter placement during the period observed, with a 33% decrease from 2015 to 2016 and a 26% decrease from 2016 to 2017 (P < .001), with an overall retrieval rate of 44.9%. In contrast, there was a corresponding increase in filter retrieval, with a 20% increase in 2015 and a 68% increase in 2016 (P= .02). In evaluating trends separated by indication, there was a significant decline in prophylactic filter placement (P < .001) and a trend toward an increase in retrieval of prophylactic filters (P= .09). Whereas there was not a significant change in number of filter insertions for major indication (P= .06), filter retrievals for major indication filters increased (P= .01). Multivariable regression analysis revealed that longer time to follow-up (odds ratio [OR], 1.08; P < .001) and discharge to rehabilitation facility (OR, 6.14; P < .001) were predictive of failure to retrieve the filter. In contrast, filter placement at a later date within our study period (OR, 0.90; P < .001) and prophylactic indication for filter placement (OR, 0.36; P < .001) were protective from filter nonretrieval. Conclusions: These results show both a decline in overall IVC filter placement and an increase in overall IVC filter retrieval at our institution. These trends are predominantly due to a decrease in prophylactic filter placement as well as an overall increase in filter retrieval. Further study should be dedicated to increasing the retrieval rate in this population of patients.