Abstract
Objective
We sought to determine whether racial and ethnic disparities existed in inferior vena cava (IVC) filter (IVCF) placement rates among Black and Latino patients for the treatment of acute proximal lower extremity (LE) deep vein thrombosis (DVT) in the United States from 2016 to 2019.
Methods
We performed a retrospective review of National Inpatient Sample data to identify adult patients with a primary discharge diagnosis of acute proximal LE DVT from January 2016 to December 2019, including self-reported patient race and ethnicity. IVCF placement rates were identified using International Classification of Diseases, 10th revision, codes. Weighted multivariable logistic regression was used to compare IVCF use by race and ethnicity. The regression model was adjusted for patient demographics (ie, sex, primary payer, quartile classification of household income), hospital information (ie, region, location, teaching status, bed size), weekend admission, and clinical characteristics (ie, modified Charlson comorbidity index, hypertension, atrial fibrillation, diabetes mellitus type 2, congestive heart failure, dyslipidemia, coronary artery disease, smoking, obesity, alcohol abuse, chronic kidney disease, pulmonary embolism, malignancy, contraindications to anticoagulation, including other major bleeding).
Results
Of 134,499 acute proximal LE DVT patients, 18,909 (14.1%) received an IVCF. Of the patients who received an IVCF, 12,733 were White (67.3%), 3563 were Black (18.8%), and 1679 were Latino (8.9%). IVCF placement decreased for all patient groups between 2016 and 2019. After adjusting for the U.S. population distribution, the IVCF placement rates were 11 to 12/100,000 persons for Black patients, 7 to 8/100,000 persons for White patients, and 4 to 5/100,000 persons for Latino patients. The difference in IVCF placement rates was statistically significant between patient groups (Black patients vs White patients, P < .05; Black patients vs Latino patients, P < .05; Latino patients vs White patients, P < .05).
Conclusions
This nationwide study showed that Black patients have higher IVCF placement rates compared with White and Latino patients. Given the known long-term complications and uncertain benefits of IVCFs, coupled with the 2010 U.S. Food and Drug Administration safety warning regarding adverse patient events for these devices, proactive measures should be taken to address this disparity among the Black patient population to promote health equity. Future work should assess whether clinician bias might be perpetuating this disparity.
Keywords: Deep vein thrombosis, Health equity, Inferior vena cava filter, Minority health, Venous thromboembolism
Article Highlights.
-
•
Type of Research: A retrospective nationwide study using National Inpatient Sample data
-
•
Key Findings: Of 134,499 patients with proximal deep vein thrombosis, 18,909 received inferior vena cava filters (IVCFs) between 2016 and 2019. After adjusting for the U.S. population distribution, the IVCF placement rates were 11 to 12/100,000 persons for Black patients, 7 to 8/100,000 persons for White patients, and 4 to 5/100,000 persons for Latino patients. The differences in placement rates were statistically significant between all patient groups (P < .05).
-
•
Take Home Message: This nationwide study showed that Black patients had higher IVCF placement rates compared with White and Latino patients. Given the known long-term complications and uncertain benefits of IVCFs, coupled with the 2010 U.S. Food and Drug Administration safety warning regarding adverse patient events for these devices, proactive measures should be taken to address this disparity among Black patients to promote health equity.
The efficacy of inferior vena cava filters (IVCFs) in decreasing venous thromboembolism-associated morbidity and mortality is unclear.1 However, the long-term complications associated with IVCFs are associated with significant morbidity and mortality.2 In the United States, overuse of IVCFs and low retrieval rates have forced clinicians to care for a patient population with complications such as IVC thrombosis and filter migration and embolization. Studies have estimated that IVCF placement rates in 2012 in the United States were 25 times that of an equivalent population in Europe (224,700 vs 9070).3 Additionally, other studies have reported low retrieval rates, ranging from 10% to 34%.3, 4, 5 From 1979 to 2005, studies showed increased IVCF placement rates in the United States, a trend that continued through 2010 until finally declining from 2011 to 2014.6, 7, 8 This has been attributed to a 2010 U.S. Food and Drug Administration (FDA) warning about IVCF complications.9
IVCF complications carry serious risks for patients. With the risk of complications ranging from device migration to potential thrombosis, the use of these devices should be reserved for patients with an absolute indication, such as active bleeding and a possible extremely high risk of pulmonary thromboembolism.9,10 Several studies have reported the long-term complications and uncertain benefits of IVCFs. The PREPIC (Prévention du Risque d’Embolie Pulmonaire par Interruption Cave) randomized study, which assessed the 8-year follow-up of patients with permanent IVCFs for the prevention of pulmonary embolism, found that these devices did reduce the risk of nonfatal pulmonary embolism.11 However, they also increased the risk of subsequent deep vein thrombosis (DVT). Thus, the investigators concluded that systemic use in the general population could not be recommended.11
Few studies have assessed the racial and ethnic disparities in IVCF placement for patients with acute proximal lower extremity (LE) DVT. Thus, given the numerous studies documenting adverse IVCF placement outcomes, coupled with the lack of reported literature examining IVCF rates in minority populations, we investigated whether racial and ethnic disparities existed for Black and Latino patients in the United States between 2016 and 2019.
Methods
Adult hospital discharges with a primary diagnosis of acute proximal LE DVT were identified from 2016 to 2019 in the National Inpatient Sample (NIS), using International Classification of Diseases, 10th revision, codes. The proximal LE DVT variable was generated by combining International Classification of Diseases, 10th revision, clinical modification codes for IVC DVT, iliac DVT, femoral DVT, popliteal DVT, and unspecified proximal DVT (Supplementary Methods and Supplementary Table I, online only). Proximal DVT is defined as thrombosis in the deep veins from the popliteal vein to the IVC. The NIS database is part of the Healthcare Cost and Utilization Project and the Agency for Healthcare Research and Quality. The database is derived from billing data submitted from hospitals to state-level organizations across the United States and is the largest publicly available all-payer inpatient care database in the United States, containing information on >7 million hospital stays each year.
All patients in our sample were admitted emergently. We excluded patients admitted electively or transferred from another acute care hospital to avoid duplication of entries. We also excluded patients with data missing for the variables of interest. Self-reported patient race and ethnicity was the primary exposure, and IVCF placement was the primary outcome for this study. However, for the purposes of our study, only Black, Latino, and White patient populations were included in the analysis because the number of Asian, Native American, and “other” identifying patient groups were few in comparison (<5% of total sample).
Weighted population estimates were calculated using the 2010 U.S. Census data.12 Crude rates, proportions, and percentages of IVCF placement in the racial/ethnic groups were calculated. Weighted multivariable logistic regression stratified by year was used to compare IVCF use stratified by race/ethnicity, with specific comparisons conducted to assess differences between Latino vs White patients and Black vs White patients. The model was adjusted for patient demographics (ie, sex, primary payer, quartile classification of household income), hospital information (ie, region, location, teaching status, bed size), weekend admission, and clinical characteristics (ie, modified Charlson comorbidity index [CCI], hypertension, atrial fibrillation, diabetes mellitus type 2, congestive heart failure, dyslipidemia, coronary artery disease, smoking, obesity, alcohol abuse, chronic kidney disease, pulmonary embolism, malignancy, contraindications to anticoagulation, including other major bleeding). Linear fit lines with trends were analyzed for temporal changes in the rates of IVCF placement for the different racial/ethnic groups. Prism, version 9.3.0 (GraphPad), was used to create graphic representations, including linear trends.
A hospital is defined as a teaching hospital if it has one or more Accreditation Council for Graduate Medical Education–approved residency programs, is a member of the Council of Teaching Hospitals, or has a ratio of full-time equivalent interns and residents to beds of ≥0.25. The teaching status can be nonteaching, teaching, or missing (teaching status considered missing if the data source that contributed discharge data to the NIS prohibited the release of hospital identifiers). Hospital location designations are defined by the Core Based Statistical Area (CBSA). Hospitals residing in counties with a CBSA type of metropolitan are defined as urban, and hospitals with a CBSA type of micropolitan or noncore are defined as rural. Urban teaching hospitals are those that meet both the previously stated teaching definition and the urban definition. No rural teaching hospital definition is available because rural teaching hospitals are rare. A hospital's bed size is defined using the region of the U.S. hospital (Northeast, Midwest, Southern, Western), the urban vs rural designation of the hospital, and teaching status. The NIS assesses the number of short-term acute care beds set up and staffed in a hospital, which ranges depending on these criteria.13
Ethics statement
The present study was deemed exempt from institutional review board assessment because it was conducted using the NIS, a publicly available database that contains only de-identified patient information.
Results
Of 164,450 patients with a primary diagnosis of acute proximal LE DVT from 2016 to 2019 in the United States, 134,499 were included in our final analysis. Of the 134,499 patients with proximal DVT, 18,909 (14.1%) received an IVCF. Of the 18,909 patients who received an IVCF, 12,733 were White (67.3%), 3563 were Black (18.8%), and 1679 were Latino (8.9%; Fig 1). The median patient age was 67 years, 49.4% were women, 58.1% used Medicare, 12.5% used Medicaid, and 72.4% had been treated at urban teaching hospitals (Table).
Fig 1.
Flow diagram for study selection. DVT, Deep vein thrombosis; DRG, diagnosis-related group; IVCF, inferior vena cava filter; LE, lower extremity.
Table.
Baseline characteristics of study population stratified by year (n = 134,499)
| Characteristic | 2016 (n = 32,142) | 2017 (n = 33,296) | 2018 (n = 34,250) | 2019 (n = 34,811) | Total (n = 134,499) |
|---|---|---|---|---|---|
| Age, years | 67 (54-78) | 67 (55-78) | 67 (55-78) | 67 (56-78) | NA |
| Sex | |||||
| Female | 15,880 (49.4) | 16,396 (49.2) | 17,000 (49.6) | 17,103 (49.1) | 66,379 (49.4) |
| Male | 16,262 (50.6) | 16,900 (50.8) | 17,250 (50.4) | 17,708 (50.9) | 68,120 (50.6) |
| Race and ethnicitya | |||||
| Native American or Alaskan Native | 122 (0.4) | 108 (0.3) | 106 (0.3) | 119 (0.3) | 455 (0.3) |
| Asian | 381 (1.2) | 467 (1.40) | 501 (1.5) | 491 (1.4) | 1840 (1.4) |
| Black | 6129 (19.1) | 6449 (19.4) | 6530 (19.1) | 6874 (19.7) | 25,982 (19.3) |
| Latino | 2536 (7.9) | 2648 (8.0) | 3044 (8.9) | 2866 (8.2) | 11,094 (8.2) |
| Other | 845 (2.6) | 963 (2.9) | 895 (2.6) | 889 (2.6) | 3592 (2.7) |
| White | 22,129 (68.8) | 22,661 (68.1) | 23,174 (67.8) | 23,572 (67.7) | 91,536 (68.1) |
| Hospital location | |||||
| Rural | 2007 (6.2) | 1984 (6.0) | 1843 (5.4) | 1990 (5.7) | 7824 (5.8) |
| Urban, nonteaching | 8547 (26.6) | 7632 (22.9) | 7041 (20.6) | 6028 (17.3) | 29,248 (21.8) |
| Urban, teaching | 21,588 (67.2) | 23,680 (71.1) | 25,366 (74.1) | 26,793 (77.0) | 97,427 (72.4) |
| Hospital bed size | |||||
| Small | 5543 (17.3) | 5915 (17.8) | 6577 (19.2) | 6745 (19.4) | 24,780 (18.4) |
| Medium | 9329 (29.0) | 9939 (29.9) | 10,178 (29.7) | 10,381 (29.8) | 39,827 (29.6) |
| Large | 17,270 (53.7) | 17,442 (52.4) | 17,495 (51.1) | 17,685 (50.8) | 69,892 (52.0) |
| Hospital region | |||||
| Northeast | 6892 (21.4) | 7068 (21.2) | 7018 (20.5) | 7068 (20.3) | 28,046 (20.9) |
| Midwest | 6987 (21.7) | 7163 (21.5) | 7589 (22.2 | 7633 (21.9) | 29,372 (21.8) |
| South | 12,402 (38.6) | 13,079 (39.3) | 13,427 (39.2) | 13,734 (39.5) | 56,642 (39.1) |
| West | 5861 (18.2) | 5986 (18.0) | 6216 (18.1) | 6376 (18.3) | 24,439 (18.2) |
| Primary payer | |||||
| Medicare | 18,465 (57.4) | 19,490 (58.5) | 20,052 (58.5) | 20,197 (58.0) | 78,204 (58.1) |
| Medicaid | 4010 (12.5) | 4103 (12.3) | 4302 (12.6) | 4424 (12.7) | 16,839 (12.5) |
| Private insurance | 7729 (24.0) | 7735 (23.2) | 7828 (22.9) | 8025 (23.1) | 31,317 (23.3) |
| Self-pay | 1041 (3.2) | 1104 (3.3) | 1175 (3.4) | 1266 (3.6) | 4586 (3.4) |
| No charge/other | 112 (0.3) | 98 (0.3) | 101 (0.3) | 89 (0.3) | 400 (0.3) |
| Missing | 785 (2.4) | 766 (2.3) | 792 (2.3) | 810 (2.3) | 3153 (2.3) |
| Incomeb | |||||
| Quartile 1 | 9577 (29.8) | 9621 (28.9) | 9761 (28.5) | 9977 (28.7) | 38,936 (28.9) |
| Quartile 2 | 7822 (24.3) | 8388 (25.2) | 8683 (25.4) | 8604 (24.7) | 33,497 (24.9) |
| Quartile 3 | 7725 (24.0) | 8094 (24.3) | 8368 (24.4) | 8685 (24.9) | 32,872 (24.4) |
| Quartile 4 | 7018 (21.8) | 7193 (21.6) | 7438 (21.7) | 7545 (21.7) | 29,194 (21.7) |
| All-patient DRG severity | |||||
| Minor | 2131 (6.6) | 1913 (5.7) | 1712 (5.0) | 2613 (7.5) | 8369 (6.2) |
| Moderate | 8132 (25.3) | 7915 (23.8) | 6472 (18.9) | 8009 (23.0) | 30,528 (22.7) |
| Major | 13,798 (42.9) | 14,334 (43.1) | 14,957 (43.7) | 14,400 (41.4) | 57,489 (42.7) |
| Extreme | 8079 (25.1) | 9132 (27.4) | 11,108 (32.4) | 9787 (28.1) | 38,106 (28.3) |
| CCI | |||||
| 0 | 7901 (24.6) | 7867 (23.6) | 7729 (22.6) | 7318 (21.0) | 30,815 (22.9) |
| 1 | 6173 (19.2) | 6146 (18.5) | 6384 (18.6) | 6373 (18.3) | 25,076 (18.6) |
| 2 | 5210 (16.2) | 5462 (16.4) | 5438 (15.9) | 5477 (15.7) | 21,587 (16.0) |
| 3 | 3640 (11.3) | 3590 (10.8) | 3642 (10.6) | 3970 (11.4) | 14,842 (11.0) |
| 4 | 2574 (8.0) | 2617 (7.9) | 2725 (8.0) | 2770 (8.0) | 10,686 (7.9) |
| ≥5 | 6644 (20.7) | 7614 (22.9) | 8332 (24.3) | 8903 (25.6) | 31,493 (23.4) |
CCI, Charlson comorbidity index; DRG, diagnosis-related group; NA, not applicable.
Data presented as median (range) or number (%); unweighted data are presented.
Race and ethnicity self-reported by each patient; however, the categories were preset by the data source.
According to median income stratified by zip code.
IVCF rates stratified by race and ethnicity
Overall, the IVCF placement rates decreased across all racial and ethnic groups from 2016 through 2019. In the White patient population, the IVCF placement rates were 15.8% in 2016 and had declined to 12.9% in 2019. Similarly, among the Black and Latino patient populations, the placement rates decreased from 14.9% and 16.6% in 2016 to 12.89% and 14.1% in 2019, respectively.
When combining all other patient populations into an “other” racial group, 5887 patients (4.4% of the total 134,499 patients) were included in this analysis. The IVCF placement rate in this other group was 15.9% (934 of 5887).
After adjusting for the overall national population distribution using the 2010 U.S. Census, the differences in the IVCF placement rates were statistically significant among the racial and ethnic groups. Black patients had an IVCF placement rate of 11 to 12/100,000 persons compared with an IVCF placement rate of 7 to 8/100,000 persons for White patients. Latino patients had the lowest IVCF placement rate at four to five per 100,000 persons during the same period (Fig 2). The 95% confidence intervals (CIs) for the individual annual rates during the study period are presented in Supplementary Table II (online only). A statistically significant difference in the IVCF placement rates per 100,000 persons was found between the racial and ethnic groups (Black patients vs White patients, P < .05; Black patients vs Latino patients, P < .05; Latino patients vs White patients, P < .05).
Fig 2.
Racial and ethnic disparities in inferior vena cava filter (IVCF) placement in patients with acute proximal lower extremity (LE) deep vein thrombosis (DVT) in the United States (2016-2019).
IVCF placement trends stratified by race and ethnicity for 2016 to 2019
The adjusted odds of changes in the trends of IVCF placement over time for patients with acute proximal LE DVT among racial and ethnic minoritized groups compared with White patients was 0.93 (95% CI, 0.87-1.01; P = .08) in 2016 and remained nonsignificant in 2019 at 0.99 (95% CI, 0.92-1.07; P = .83; Fig 3, A). However, the adjusted odds for changes in trends of IVCF placement over time for Black patients compared with White patients was significantly different between 2016 (odds ratio [OR], 0.88; 95% CI, 0.80-0.96; P = .004), although they became similar in 2019 (OR, 0.94; 95% CI, 0.86-1.03; P = .17; Fig 3, B). Finally, the adjusted odds for changes in trends of IVCF placement over time in Latino patients compared with White patients remained similar between 2016 (OR, 1.04; 95% CI, 0.91-1.19; P = .56) and 2019 (OR, 1.08; 95% CI, 0.95-1.23; P = .25; Fig 3, C).
Fig 3.
Change in trends of inferior vena cava filter (IVCF) placement in racial and ethnic groups compared with White patients from 2016 to 2019. A, All racial and ethnic minority groups vs White patients. B, Black vs White patients. C, Latino vs White patients. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs; error bars) of IVCF placement stratified by year, with 95% CIs (A,pink; B,orange; and C,green). Dashed horizontal line indicates reference standard (White patients).
When all racial/ethnic groups were compared to Black patients, the adjusted odds for changes in trends of IVCF placement over time was significantly different during 2016 (OR, 1.14; 95% CI, 1.05-1.25; P = .003) but became similar in 2019 (OR, 1.08; 95% CI, 0.99-1.17; P = .09; Fig 4, A). Similarly, the adjusted odds for changes in trends of IVCF placement over time in all racial/ethnic groups compared to Latino patients remained similar from 2016 (OR, 0.93; 95% CI, 0.82-1.06; P = .27) through 2019 (OR, 0.92; 95% CI, 0.81-1.04; P = .18; Fig 4, B).
Fig 4.
Change in trends of inferior vena cava filter (IVCF) placement in all racial and ethnic groups compared with Black and Latino patients from 2016 to 2019. A, All racial and ethnic minority groups vs Black patients (reference, dashed horizontal line; Black patients). B, All racial and ethnic minority groups vs Latino patients (dashed horizontal line is reference; Latino patients). Adjusted odds ratios (ORs) and 95% confidence intervals (CIs; error bars) of IVCF placement stratified by year, with 95% CIs in pink and blue.
Thrombolysis and thrombectomy
Between 2016 and 2019 in our study sample, a significant difference was found in the number of Black, Latino, and White patients who received thrombolysis in the setting of DVT (n = 1014, n = 587, and n = 4656, respectively; χ2 = 66.04; P = .0005). Similarly, we also found a significant difference when comparing Black and White patients alone (n = 1014 and n = 4656, respectively; χ2 = 61.76; P = .0005). However, when directly comparing Latino and White patients, the difference was no longer significant (n = 587 and n = 4656, respectively; χ2 = 0.85; P = .36).
Between 2016 and 2019 in our study sample, we also found a significant difference in the number of Black, Latino, and White patients who received thrombectomy in the setting of DVT (n = 558, n = 356, and n = 2851, respectively; χ2 = 69.98; P = .0005). Similarly, we also found a significant difference when comparing Black and White patients alone (n = 558 and n = 2851, respectively; χ2 = 67.18; P = .0005). However, when directly comparing Latino and White patients, the difference was no longer significant (n = 356 and n = 2851, respectively; χ2 = 0.29; P = .59).
Cancer patients
We investigated the annual rates of IVCF placement in Black, Latino, other, and White patients with DVT with and without cancer. We also considered the odds of increased IVCF placement in cancer patients from 2016 to 2019 in these same racial categories. Finally, we studied the odds of increased IVCF placement among cancer patients for the 4 years combined from 2016 to 2019 in these same racial groups. This separate multivariate subgroup analysis of racial disparities in IVCF placement in cancer patients only did not reveal significant racial disparities in the cancer subpopulation. However, a significant difference was found between the cancer and noncancer patients (Supplementary Table III (online only), Supplementary Table IV (online only), Supplementary Table V (online only), online only).
Discussion
This study showed that racial and ethnic disparities existed in the placement of IVCFs for patients with a principal diagnosis of acute proximal LE DVT in the United States between 2016 and 2019. Black patients were found to have the highest IVCF placement rate per 100,000 persons compared with both White and Latino patients. Although the reasons behind this racial and ethnic disparity needs further investigation, factors such as clinician bias related to concern for medication adherence could play a role. Given the 2010 U S. FDA safety warning regarding these devices, combined with previously reported literature recommending implantation rates similar to or lower than the rate observed in Europe (3 per 100,000 persons based on data from five large European countries), these higher placement rates in the Black community are especially concerning and suggest that IVCFs are still overused in the United States.1,9,10 Of equal concern is the disproportionate overuse of a procedure that has an unfavorable risk/benefit ratio for Black patients. In addition, the linear trends of IVCF placement did not show a statistically significant change among all patient groups during the study period, indicating the persistence of risk factors and practices leading to higher IVCF placement rates in the Black patient population.
Our findings of higher IVCF placements rates for Black patients is consistent with other previously reported literature on IVCF use in minoritized communities. One recent study found that prophylactic IVCF placement was threefold higher for Black patients than for White patients who were undergoing metabolic and bariatric surgery.14 Another recent study assessing IVCF placement among Medicare enrollees with acute venous thromboembolism demonstrated that Black patients experienced a longer wait, on average, for IVCF retrieval procedures than their White counterparts.15 Yet another study in 2022 assessing IVCF usage at a single tertiary care center during a 19-year period found a racial disparity in retrieval rates, with White patients having a 13.1% retrieval rate compared with 9.1% for non-White patients for IVCFs that were potentially retrievable.16 However, our finding that higher IVCF placements rates were observed for Black patients is counter to other procedure literature, in which Black patients have been reported to receive fewer surgical procedures with proven benefit such as early lung cancer surgery, heart valve replacement, coronary artery bypass grafting for symptomatic and severe coronary artery disease, and carotid endarterectomy.17 Overuse of a treatment with a safety warning of adverse patient events is mostly unprecedented. Our findings would align with the broader perspective of Black patients receiving lower quality health care and should call attention to quality of care equity indicators in hospitals.
This study also showed that Latino patients had the lowest IVCF placement rate per 100,000 persons compared with both Black and White patients in the United States between 2016 and 2019. This rate was similar to those observed in Europe.1 Although few data are available on IVCF placement rates in this community, potential contributing factors could be limited patient–provider communication and patient preference. Previously reported literature has shown that limited English proficiency and language discordance between patients and their physicians results in poor comprehension and poor interactive communication.18, 19, 20
Study limitations
Study limitations include the possibility of unmeasured confounders not assessed and adjusted for in our multivariable regression analysis, which might be responsible for the differences in outcomes seen between the racial and ethnic groups. Second, a lack of information on DVT characteristics, patient medications and compliance, eligibility for IVCF placement, and possible contraindications could play a role in the different rate of IVCF placement. Trends in IVCF placement for other racial and ethnic groups or subpopulations of existing groups, including Latino patients of different national heritages, were not investigated owing to the lack of categorization and/or small sample size. Additionally, the accuracy of our dataset relies on appropriate coding of diagnoses and procedures, which makes it subject to the limitations of administrative datasets. Because our data were derived from an entirely inpatient sample, IVCF placement trends could be different when considering patients undergoing the procedure in an outpatient setting, although these are rare. Next, no method was available to determine the percentage of Black, Latino, and White patients in the dataset when racial and ethnic data were not collected. The collection of race and ethnicity varies by both state and hospital regarding the completeness and categories collected. Finally, our analysis was limited to identification of hospital admissions with the presence of acute proximal DVT (through diagnosis codes) and placement of IVCF (through procedure codes). Because of the inherent limitations of the dataset, we were unable to accurately identify the primary cause of hospital admissions such as trauma or malignancy. To balance for the higher risk of DVT for patients with cancer and malignancy, these variables were included in the multivariate analysis. In addition, we performed a separate multivariate subgroup analysis to determine racial disparities in IVCF placement in cancer patients only, which did not show significant racial disparities in the cancer subpopulation.
Conclusions
This nationwide, retrospective study showed that Black patients had the highest IVCF placement rate per 100,000 persons compared with White and Latino patients in the United States from 2016 to 2019. Given the known long-term complications and uncertain benefits with IVCFs, coupled with the 2010 U S. FDA safety warning regarding adverse patient events for these devices, proactive measures are required to address this overuse among Black patients to promote health care equality. Furthermore, given the relatively high IVCF placement rates in the United States compared with those observed in Europe, our findings suggest that these implantable devices are still overused for all U.S. patients. Future work should assess whether clinical bias for adherence to other treatments could be perpetuating this disparity in minoritized communities.
Author contributions
Conception and design: JJ, MK, EC, IS, VL, RB
Analysis and interpretation: JJ, MK, BU, CR, RM, DS, EC, IS, VL, HZ, ER, EP, RB
Data collection: JJ, MK, RB
Writing the article: JJ, MK, BU, CR, RM, DS, HZ, ER, EP, RB
Critical revision of the article: JJ, MK, EC, IS, VL, HZ, ER, EP, RB
Final approval of the article: JJ, MK, BU, CR, RM, DS, EC, IS, VL, HZ, ER, EP, RB
Statistical analysis: MK, HZ
Obtained funding: Not applicable
Overall responsibility: RB
Disclosures
R.B. has received research support from the National Heart, Lung, and Blood Institute and has equity interest in Thrombolex Inc. J.J.J., M.U.K., B.A.U., C.M.R., R.M., D.S., E.C., I.S., V.L., H.Z., E.J.R., and E.J.P.-S. have no conflicts of interest.
Footnotes
Research support was provided by the National Institutes of Health Medical Research Scholars Program, a public–private partnership supported jointly by the National Institutes of Health and contributions to the Foundation for the National Institutes of Health from the American Association for Dental Research and the Colgate-Palmolive Company.
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest.
Appendix
Additional material for this article may be found online at www.jvsvenous.org
Supplementary Methods
Case selection and missing data variables
The total proximal deep vein thrombosis (DVT) variable was generated by combining the International Classification of Diseases, 10th revision, clinical modification codes for inferior vena cava (IVC) DVT, iliac DVT, femoral DVT, popliteal DVT, and unspecified proximal DVT (Supplementary Table I, online only). All patients without proximal DVT were excluded from the analysis to facilitate evaluation of a smaller dataset. Patients with missing values for the following variables of interest were excluded from the analysis: age, sex, race, quartile classification of household income, expected primary insurance payer, all patient-refined diagnosis-related groups, weekend admission date, hospital bed size, and location and teaching status of the hospital.
Primary outcome and Charlson comorbidity index
Our primary outcome (IVCF placement for DVT treatment) and comorbidities (ie, hypertension, atrial fibrillation, diabetes mellitus type 2, congestive heart failure, dyslipidemia, coronary artery disease, smoking, obesity, alcohol abuse, chronic kidney disease, pulmonary embolism, malignancy, and contraindications to anticoagulation, including other major bleeding) were accounted for in our analyses using the International Classification of Diseases, 10th revision, clinical modification, codes listed in Supplementary Table I (online only). The comorbidities were also characterized by calculating the Charlson comorbidity index (CCI). 11 The CCI was calculated for all patients with acute proximal DVT, grouping the values into 0, 1, 2, 3, 4, and ≥5.
Statistical analysis
To account for confounders of exposure–outcome associations, a multivariable regression model was adjusted for by the following variables: patient demographics (ie, sex, primary payer, quartile classification of household income), hospital information (ie, region, location, teaching status, bed size), weekend admission, and clinical characteristics (ie, modified CCI, hypertension, atrial fibrillation, diabetes mellitus type 2, congestive heart failure, dyslipidemia, coronary artery disease, smoking, obesity, alcohol abuse, chronic kidney disease, pulmonary embolism, malignancy, contraindications to anticoagulation, including other major bleeding). The variance inflation factor was used to assess multicollinearity. Because of the high variance inflation factor (cutoff <5) for age in the patient demographic, age was excluded from the multivariate regression analysis. The median age and weighted data according to the variables of interest were obtained on an annual basis. Prespecified adjusted OR estimates are presented for each year without adjustment for multiple comparisons. A test for linear trend of crude DVT use over time was performed with variance-weighted least squares regression, with year as a continuous variable. The linear trend for the adjusted odds of DVT over time was determined by fitting a logistic regression model with an interaction term for race by year (continuous variable), overall for all racial and ethnic minority groups and stratified by racial and ethnic subgroups.
Appendix (online only)
Supplementary Table I (online only).
Administrative codes used for construction of variables not provided by National Inpatient Sample (NIS)
| Diagnosis | ICD-10 codes |
|---|---|
| Inferior vena cava filter | 06H03DZ, 06V03DZ, 06V03ZZ |
| IVC DVTa | I82.220 |
| Iliac DVTa | I82.421, I82.422, I82.423, I82.429 |
| Femoral DVTa | I82.411, I82.412, I82.413, I82.419 |
| Popliteal DVTa | I82.431, I82.432, I82.433, I82.439 |
| Unspecified proximal DVTa | I82.4Y1, I82.4Y2, I82.4Y3, I82.4Y9 |
| Hypertension | I10, I11.0, I11.9, I12.0, I13.0, I13.11, I13.2, I15.0, I15.1, I15.2, I15.8-I16.1, I16.9, I67.4 |
| Atrial fibrillation | I48.91, I48.1, I48.2, I48.0 |
| Diabetes mellitus type 2 | E11 |
| Congestive heart failure | I50 |
| Dyslipidemia | E78.0, E78.1, E78.2, E78.3, E78.4, E78.5, E78.6, E78.70, E78.79, E78.9, E78.81, E78.89 |
| Coronary artery disease | I25.1, I25.82, I25.83, I25.84. I25.89, I25.9, I20.0, I20.1, I20.8, I23.7, I24.8, I24.9, I25.5, I25.6 |
| Smoking | Z72.0, F17.299, F17.210, F17.218, F17.219, T65.222 A |
| Obesity | E66 |
| Alcohol | F10.10 |
| Chronic kidney disease | N18, N18.1, N18.2, N18.3, N18.4, N18.5, N18.6, N18.9 |
| Pulmonary embolism | I26 |
| Any malignancy, including lymphoma and leukemia, except for malignant neoplasm of skin | C00.x-C26.x, C30.x-C34.x, C37.x-C41.x, C43.x, C45.x-C58.x, C60.x-C76.x, C81.x-C85.x, C88.x, C90.x-C97.x |
| Other major bleeding | E0789, E214, D500, D62, D698, D699, H31309, H31319, H0289, H05239, H47029, H4313, I8501 I8511, R58, K2211, K226, NAK250, K250, K5660, K252, K252, K5660, K254, K254, K5660, K256, K256, K5660, K260, K260, K5660, K262, K262, K5660, K264, K264, K266, K266, K5660, K270, K270, K272, K272, K274, K274, K276, K276, K280, K280, K282, K282, K284, K284, K286, K286, K2901, K2941, K2951, K2961, K2921, K2961, K2971, K2991, K2981, K5281, K661, K625, K5521, K920, K921, K922, N421, N920, O036, O031, O036, O046, O046, O046, O046, O046, O046, O046, O046, O046, O071, O081, O200, O200, O200, O208, O208, O208, O209, O209, O209, O4401, O4402, O4403, O4401, O4402, O4403, O4410, O4411, O4412, O4413, O4411, O4412, O4413, O458X9, O458X1, O458X2, O458X3, O4591, O4592, O4593, O458X1, O458X2, O458X3, O4591, O4592, O4593, O46009, O46019, O46029, O46099, O45001, O45002, O45003, O45011, O45012, O45013, O45021, O45022, O45023, O45091, O45092, O45093, O46001, O46002, O46003, O46011, O46012, O46013, O46021, O46022, O46023, O46091, O46092, O46093, O670, O45001, O45002, O45003, O45011, O45012, O45013, O45021, O45022, O45023, O45091, O45092, O45093, O46001, O46002, O46003, O46011, O46012, O46013, O46021, O46022, O46023, O46091, O46092, O46093, O468X9, O468X1, O468X2, O468X3, O678, O468X1, O468X2, O468X3, O4690, O4691, O4692, O4693, O679, O4691, O4692, O4693, O717, O720, O43211, O43212, O43213, O43221, O43222, O43223, O43231, O43232, O43233, O720, O720, O43211, O43212, O43213, O43221, O43222, O43223, O43231, O43232, O43233, O721, O722, O860, O902, O860, P544, R040, R041, 792XXA, D7801, D7802, D7821, D7822, E3601, E3602, G9731, G9732, G9751, G9752, H59111, H59112, H59113, H59119, H59121, H59122, H59123, H59129, H59311, H59312, H59313, H59319, H59321, H59322, H59323, H59329, H9521, H9522, H9541, H9542, I97410, I97411, I97418, I9742, I97610, I97611, I97618, I9762, J9561, J9562, J95830, J95831, K9161, K9162, K91840, K91841, L7601, L7602, L7621, L7622, M96810, M96811, M96830, M96831, N9961, N9962, N99820, N99821, D7801, D7802, D7821, D7822, E3601, E3602, G9731, G9732, G9751, G9752, H59111, H59112, H59113, H59119, H59121, H59122, H59123, H59129, H59311, H59312, H59313, H59319, H59321, H59322, H59323, H59329, H9521, H9522, H9541, H9542, I97410, I97411, I97418, I9742, I97610, I97611, I97618, I9762, J9561, J9562, J95830, J95831, K9161, K9162, K91840, K91841, L7601, L7602, L7621, L7622, M96810, M96811, M96830, M96831, N9961, N9962, N99820, N99821 |
DVT, Deep vein thrombosis; ICD-10, International Classification of Diseases, 10th revision.
Inferior vena cava DVT, iliac DVT, femoral DVT, popliteal DVT, and unspecified proximal DVT codes were used to create the total proximal DVT value.
Supplememtary Table II (online only).
Annual inferior vena cava (IVC) filter (IVCF) placement rate stratified by population and race/ethnicity
| Race/ethnicity | 2016 | 2017 | 2018 | 2019 | Total | 2010 U.S. census population | Rate/100,000 population (95% CI) |
|||
|---|---|---|---|---|---|---|---|---|---|---|
| 2016 | 2017 | 2018 | 2019 | |||||||
| Black | 4500 (900) | 4580 (916) | 4410 (882) | 4325 (865) | 17,815 (3563) | 38,929,319 | 11.56 (11.2-11.9) | 11.76 (11.4-12.1) | 11.33 (11.0-11.7) | 11.11 (10.8-11.4) |
| Latino | 2085 (417) | 1985 (397) | 2325 (465) | 2000 (400) | 8395 (1679) | 50,477,594 | 4.13 (4.0-4.3) | 3.93 (3.8-4.1) | 4.61 (4.4-4.8) | 3.96 (3.8-4.1) |
| White | 17,100 (3420) | 16,570 (3314) | 15,380 (3076) | 14,615 (2923) | 63,665 (12,733) | 223,553,265 | 7.65 (7.5-7.8) | 7.41 (7.3-7.5) | 6.88 (6.8-7.0) | 6.54 (6.4-6.6) |
CI, Confidence interval.
Data presented as weighted numbers, with unweighted numbers in parentheses.
Supplementary Table III (online only).
Annual rates of inferior vena cava (IVC) filter (IVCF) placement in patients with deep vein thrombosis (DVT) stratified by cancer
| IVCF |
No IVCF |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 2016 | 2017 | 2018 | 2019 | Total | 2016 | 2017 | 2018 | 2019 | Total | |
| White patients | ||||||||||
| No cancer | 2654 (14.8) | 2502 (13.7) | 2261 (12.2) | 2126 (11.4) | 9543 | 15,254 (85.2) | 15,734 (86.3) | 16,280 (78.8) | 16,594 (88.6) | 63,862 |
| Cancer | 766 (18.2) | 812 (18.4 | 815 (17.6 | 797 (16.4) | 3190 | 3455 (81.9) | 3613 (81.7) | 3818 (82.4) | 4055 (83.6) | 14,941 |
| Black patients | ||||||||||
| No cancer | 679 (13.4) | 681 (12.9 | 645 (12.2 | 640 (11.7) | 2645 | 4391 (86.6) | 4603 (87.1) | 4625 (87.8) | 4854 (88.4) | 18,473 |
| Cancer | 221 (20.9) | 235 (20.2 | 237 (18.8 | 225 (16.3) | 918 | 838 (79.1) | 930 (79.8) | 1023 (81.2) | 1155 (83.7) | 3946 |
| Latino patients | ||||||||||
| No cancer | 304 (15.2) | 302 (14.3 | 353 (14.8 | 280 (12.8) | 1239 | 1698 (84.8) | 1810 (85.7) | 2032 (85.2) | 1914 (87.2) | 7454 |
| Cancer | 113 (21.2) | 95 (17.7 | 112 (17) | 120 (17.9) | 440 | 421 (78.8) | 441 (82.3) | 547 (83) | 552 (82.1) | 1961 |
| Other | ||||||||||
| No cancer | 172 (16.8) | 162 (14.4 | 158 (14.7) | 140 (12.9) | 632 | 853 (83.2) | 966 (85.6) | 917 (85.3) | 942 (87.1) | 3678 |
| Cancer | 60 (18.6) | 79 (19.3) | 81 (18.8) | 82 (19.7) | 302 | 263 (81.4) | 331 (80.7) | 346 (81) | 335 (80.3) | 1275 |
Data presented as number (%).
Supplementary Table IV (online only).
Odds of increased inferior vena cava (IVC) filter (IVCF) placement in cancer patients stratified by race from 2016 to 2019
| 2016 |
2017 |
2018 |
2019 |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | P value | OR | 95% CI | P value | OR | 95% CI | P value | OR | 95% CI | P value | |
| All vs White patients | 1.09 | 0.94-1.27 | .24 | 1.04 | 0.90-1.2 | .59 | 0.97 | 0.84-1.12 | .65 | 1.03 | 0.9-1.17 | .7 |
| Black vs White patients | 1.1 | 0.92-1.33 | .29 | 1.07 | 0.89-1.28 | .47 | 0.98 | 0.82-1.18 | .84 | 0.93 | 0.78-1.1 | .38 |
| Latino vs Black patients | 1.22 | 0.95-1.56 | .11 | 0.98 | 0.75-1.28 | .88 | 0.9 | 0.72-1.13 | .38 | 1.17 | 0.93-1.47 | .18 |
| All vs Black patients | 0.92 | 0.77-1.11 | .38 | 0.94 | 0.78-1.12 | .47 | 1.01 | 0.84-1.2 | .95 | 1.11 | 0.94-1.31 | .22 |
| All vs Latino patients | 0.83 | 0.65-1.06 | .14 | 1.04 | 0.80-1.34 | .79 | 1.1 | 0.88-1.38 | .39 | 0.85 | 0.68- 1.07 | .16 |
CI, Confidence interval; OR, odds ratio.
Supplementary Table V (online only).
Odds of increased inferior vena cava (IVC) filter (IVCF) placement among cancer patients combined for 2016 to 2019
| 2016-2019 |
|||
|---|---|---|---|
| OR | 95% CI | P value | |
| All vs White patients | 1.04 | 0.96-1.11 | .33 |
| Black vs White patients | 1.03 | 0.94-1.12 | .57 |
| Latino vs Black patients | 1.06 | 0.94-1.2 | .3 |
| All vs Black patients | 0.98 | 0.90-1.08 | .72 |
| All vs Latino patients | 0.95 | 0.84-1.07 | .36 |
CI, Confidence interval; OR, odds ratio.
References
- 1.Wang S.L., Lloyd A.J. Clinical review: inferior vena cava filters in the age of patient-centered outcomes. Ann Med. 2013;45:474–481. doi: 10.3109/07853890.2013.832951. [DOI] [PubMed] [Google Scholar]
- 2.Alkhouli M., Morad M., Narins C.R., Raza F., Bashir R. Inferior Vena Cava Thrombosis. JACC Cardiovasc Interv. 2016;9:629–643. doi: 10.1016/j.jcin.2015.12.268. [DOI] [PubMed] [Google Scholar]
- 3.Alkhouli M., Bashir R. Inferior vena cava filters in the United States: Less is more. Int J Cardiol. 2014;177:742–743. doi: 10.1016/j.ijcard.2014.08.010. [DOI] [PubMed] [Google Scholar]
- 4.Angel L.F., Tapson V., Galgon R.E., Restrepo M.I., Kaufman J. Systematic review of the use of retrievable inferior vena cava filters. J Vasc Interv Radiol. 2011;22:1522–1530.e3. doi: 10.1016/j.jvir.2011.08.024. [DOI] [PubMed] [Google Scholar]
- 5.Dabbagh O., Nagam N., Chitima-Matsiga R., Bearelly S., Bearelly D. Retrievable inferior vena cava filters are not getting retrieved: where is the gap? Thromb Res. 2010;126:493–497. doi: 10.1016/j.thromres.2010.08.024. [DOI] [PubMed] [Google Scholar]
- 6.Stein P.D., Kayali F., Olson R.E. Twenty-one-year trends in the use of inferior vena cava filters. Arch Intern Med. 2004;164:1541–1545. doi: 10.1001/archinte.164.14.1541. [DOI] [PubMed] [Google Scholar]
- 7.Moore P.S., Andrews J.S., Craven T.E., Davis R.P., Corriere M.A., Godshall C.J., et al. Trends in vena caval interruption. J Vasc Surg. 2010;52:118–125. doi: 10.1016/j.jvs.2009.09.067. [DOI] [PubMed] [Google Scholar]
- 8.Saeed M.J., Turner T.E., Brown D.L. Trends in Inferior Vena Cava Filter Placement by Indication in the United States From 2005 to 2014. JAMA Intern Med. 2017;177:1861–1862. doi: 10.1001/jamainternmed.2017.5960. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Reddy S., Lakhter V., Zack C.J., Zhao H., Chatterjee S., Bashir R. Association Between Contemporary Trends in Inferior Vena Cava Filter Placement and the 2010 US Food and Drug Administration Advisory. JAMA Intern Med. 2017;177:1373. doi: 10.1001/jamainternmed.2017.2719. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.US Food and Drug Administration Inferior vena cava (IVC) filters: initial communication: risk of adverse events with long term use. https://wayback.archive-it.org/7993/20161022180008/http:/www.fda.gov/MedicalDevices/Safety/AlertsandNotices/ucm221676.htm Available at:
- 11.PREPIC Study Group Eight-year follow-up of patients with permanent vena cava filters in the prevention of pulmonary embolism: the PREPIC (Prevention du Risque d’Embolie Pulmonaire par Interruption Cave) randomized study. Circulation. 2005;112:416–422. doi: 10.1161/CIRCULATIONAHA.104.512834. [DOI] [PubMed] [Google Scholar]
- 12.U.S. Census Bureau Overview of Race and Hispanic Origin. 2010. https://www.census.gov/content/dam/Census/library/publications/2011/dec/c2010br-02.pdf Available at:
- 13.Agency for Healthcare Research and Quality - Healthcare Cost and Utilization Project. NIS Description of Elements. https://hcup-us.ahrq.gov/db/vars/hosp_bedsize/nisnote.jsp Available at:
- 14.Edwards M.A., Mazzei M., Zhao H., Reddy S., Bashir R. Racial disparities in inferior vena cava filter use in metabolic and bariatric surgery patients: Nationwide insights from the MBSAQIP database. Am J Surg. 2021;221:749–758. doi: 10.1016/j.amjsurg.2020.02.062. [DOI] [PubMed] [Google Scholar]
- 15.Williams A.O., Sridharan N., Rojanasarot S., Chaer R., Anderson N., Wifler W., et al. Population-Based Disparities in Inferior Vena Cava Filter Procedures Among Medicare Enrollees With Acute Venous Thromboembolism. J Am Coll Radiol. 2022;19:722–732. doi: 10.1016/j.jacr.2022.03.011. [DOI] [PubMed] [Google Scholar]
- 16.Axley J.C., May M.M., Novak Z., Aucoin V.J., Spangler E.L., McFarland G.E., et al. Clinical practice and volume trends of inferior vena cava filter usage at a single tertiary care center during a 19-year period. J Vasc Surg Venous Lymphat Disord. 2022;10:887–893. doi: 10.1016/j.jvsv.2022.01.005. [DOI] [PubMed] [Google Scholar]
- 17.Best M.J., McFarland E.G., Thakkar S.C., Srikumaran U. Racial disparities in the use of surgical procedures in the US. JAMA Surg. 2021;156:274–281. doi: 10.1001/jamasurg.2020.6257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Fernandez A., Schillinger D., Grumbach K., Rosenthal A., Stewart A.L., Wang F., et al. Physician language ability and cultural competence: an exploratory study of communication with Spanish-speaking patients. J Gen Intern Med. 2004;19:167–174. doi: 10.1111/j.1525-1497.2004.30266.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Pérez-Stable E.J., Nápoles-Springer A., Miramontes J.M. The effects of ethnicity and language on medical outcomes of patients with hypertension or diabetes. Med Care. 1997;35:1212–1219. doi: 10.1097/00005650-199712000-00005. [DOI] [PubMed] [Google Scholar]
- 20.Wilson E., Chen A.H., Grumbach K., Wang F., Fernandez A. Effects of limited English proficiency and physician language on health care comprehension. J Gen Intern Med. 2005;20:800–806. doi: 10.1111/j.1525-1497.2005.0174.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
Reference
- 1.Charlson M.E., Pompei P., Ales K.L., MacKenzie C.R. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40:373–383. doi: 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]