Citation
Variation in Quality of Care in Use of Vena Cava Filters for Hospitalized Venous Thromboembolism Patients

Material Information

Title:
Variation in Quality of Care in Use of Vena Cava Filters for Hospitalized Venous Thromboembolism Patients
Creator:
Chen, Ming
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Master's ( M.S.P.)
Degree Grantor:
University of Florida
Degree Disciplines:
Pharmaceutical Sciences
Pharmaceutical Outcomes and Policy
Committee Chair:
BROWN,JOSHUA D
Committee Co-Chair:
XIAO,HONG
Graduation Date:
12/17/2017

Subjects

Subjects / Keywords:
thromboembolism
venous

Notes

General Note:
Objective: (1) To examine the association between VCF use and patient and hospital characteristics, and explain the variation at the patient level; (2) To delineate the distribution of four quality of care measures by hospital quartiles of VCF rates, and explain the variation at the hospital level. Setting: Nationwide Readmission Database (2013-2014) Method: Non-elective index admission of hospitalized VTE patients aged >=18 were included. (1) A hierarchical logistic regression which accounted for the random effects of hospital-level factors was used to evaluate the association between VCF use and patient-level and hospital-level risk factors. (2) Hospitals were stratified by VCF rates into four quartiles, representing an ascending order of VCF rates from the lowest quartile to the highest. Results: (1) There were 212,395 VTE hospitalizations, with 12.18% (n=25,877) receiving VCF placement. Significant association from adjusted are: age over 80 (aOR: 2.53, 95% CI: 2.25-2.85), >=13 comorbid conditions (aOR: 3.85, 95% CI: 3.25- 4.27), and privately-owned hospitals (aOR: 1.21, 95% CI: 1.08-1.36). Optimal goodness-of-fit was achieved with a combination of random effects and patient-level fixed effects. (2) Overall, 12.29% of VTE patients had VCF placements at 1,614 hospitals. When stratified by quartiles, the proportion of VCF placements ranged from 0% to 7.28% in the lowest quartile of hospitals to 15.84% to 46.84% in the highest. Conclusion: (1) Patient-level variation in use of VCF for hospitalized VTE patients is associated with patient and hospital characteristics. (2) Hospital-level variation also existed. Several patient case-mix and hospital characteristics attributed to the variation.

Record Information

Source Institution:
UFRGP
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
12/31/2019

Downloads

This item has the following downloads:


Full Text

PAGE 1

VARIATION IN QUALITY OF CARE IN USE OF VENA CAVA FILTERS FOR HOSPITALIZED VENOUS THROMBOEMBOLISM P ATIENTS By MING CHEN A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE R EQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN PHARMACY UNIVERSITY OF FLORIDA 2017

PAGE 2

2017 Ming Chen

PAGE 3

To mo m and dad, with the unconditional support and love where ver I am To grandmother, who passed aw ay before this thesis decision; w ish you would have a life in heaven without sufferi ng from the pain of the disease To my boyfriend who loves me, and accompanies me to walk thr ough the ups and downs in life

PAGE 4

4 ACKNOWLEDGMENTS Two years ago, I decided to go to Houston in Texas two days before April 15 ( The 415 Resolution ) At that time, my boyfriend had withdrawn the most prestigious f ellowship at Florida and decided to go to Austin in Texas One week later, an unforgettable email from Dr. Segal on beha lf of the faculty at Florida made me and my boyfriend to re consider our future and I went to the University of Florida. During these two years study t he u nforgettable experiences at Florida always reminded me of this decision. I am thankful for the expe riences at Florida which help me understand the rules of education systems, importance of decisions, humanity, personality and approaches to dealing with things inside and outside the graduate study. All these would benefit my following life. For t his the sis it serves as a summary of my research interest through e ducation at Florida that trains me to work on large population based secondary database to investigate the real world healthcare issues through statistical methods First, I would like to thank m y primary advisor Dr Joshua Brown and thesis member Dr. Hong Xiao, drive me to keep learning and realize I have a long way to go. Without the correct research direction under his supervision I would get lost in the various disease areas and methodologies. Dr. Xiao guides me to start research with a prospective look at big pictures, which would foster my thinking for long term objective s Secondly, I thank Drs. Amie Goodin, Driss Raissi and Qio ng Han, who invested efforts in our collaborated works. I learned how to publish a paper and work with collaborators and got benefited from the communication with Dr. Goodin on manuscript revision s which would contribute to my following accomplishments I also appreciated the clinical insights from Drs. Raissi and Han.

PAGE 5

5 Additionally, thank you very much to Dr. Richard Segal without his thoughtful suggestions, considerate support and encouragement, I could not have gone so far. I am also grateful to other faculty in the department who enrich my knowledge in graduate school. Thank you to Mom and Dad, wherever I am, for your unconditional love and support both physically and mentally Thank you, grandmother, who suffered from disease you make me further unde rstand the meaning of real world study, and there are two areas complicated with venous thromboembolism that I would like to investigate in future. JING my boyfriend in Austin, Texas as a young computational chemist who love s and trust s me, and provides valuable thoughts in our life Without you, I could not have been to Austin ten times and like Austin even Texas so much. A lso thank you RXG8 R at Austin, in my hardest time we could talk on my future study You would never know how important this talk i s to give me hope and drag me out of the bottom of a big black hole. T hank you for my friends at Florida who accompanied me with the beautiful moments when we chatted had hotpots, made cookies, watched games for Gators and worked outs. Thank you for my most close friends in California and Netherlands, who made me understand there is always place in my mind reserved for you. I look forward to the day we meet each other again play volleyball and travel around Finally, I must thank myself, never give in a nd never give up. The dreams would never fade away.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF ABBREVIATIONS ................................ ................................ ............................. 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION TO VENA CAVA FILTER UTILIZATION IN VENOUS THROMBOEMBOLISM PATIENTS ................................ ................................ ........ 12 The E pidemiology of Venous Thromboembolism ................................ ................... 12 Treatment Management of Venous Thromboembolism and Vena Cava Filters ...... 1 2 Use of Vena Cava Filters in Venous Thromboembolism Patients ........................... 13 Variation in Use of Vena Cava Filters ................................ ................................ ..... 16 2 PATIENT LEVEL VARIATION IN USE OF VENA CA VA FILTERS FOR HOSPITALIZED VENOUS THROMBOEMBOLISM PATIENTS ASSOCIATED WITH PATIENT AND HOSPITAL CHARACTERISTICS ................................ ......... 18 Introduction ................................ ................................ ................................ ............. 18 Methods ................................ ................................ ................................ .................. 20 Data Source ................................ ................................ ................................ ..... 20 Cohort Selection ................................ ................................ ............................... 20 Study Variables ................................ ................................ ................................ 21 Statistical Analysis ................................ ................................ ............................ 22 Results ................................ ................................ ................................ .................... 23 Discussion ................................ ................................ ................................ .............. 25 Conclusion ................................ ................................ ................................ .............. 30 3 HOSPITAL LEVEL VARIATION IN USE OF VENA CAVA FILTERS FOR HOSPITALIZED VENOUS THROMBOEMBOLISM PATIENTS ATTRIBUTABLE TO PATIENT AND HO SPITAL CHARACTERISTICS ................................ ............. 42 Introduction ................................ ................................ ................................ ............. 42 Method ................................ ................................ ................................ .................... 43 Data ................................ ................................ ................................ .................. 43 Study Population ................................ ................................ .............................. 43 Stratification by Quartile ................................ ................................ ................... 44 Patient Case Mix and Hos pital Level Characteristics ................................ ....... 44 Main Outcome ................................ ................................ ................................ .. 45 Statistical Analysis ................................ ................................ ............................ 45

PAGE 7

7 Results ................................ ................................ ................................ .................... 46 Discussion ................................ ................................ ................................ .............. 48 Conclusion ................................ ................................ ................................ .............. 49 4 CONCLUSIONS AND FUTURE DIRECTIONS ................................ ...................... 57 LIST OF REFERENCES ................................ ................................ ............................... 59 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 64

PAGE 8

8 LIST OF TABLES Table page 2 1 Patient and Hospital Characteristics by Vena Cava Filter Use (VCF) in all Deep Vein Thrombosis and Pulmonary Embolism Patients, from the 2013 2014 Nationwide Readmissions Database ................................ ......................... 31 2 2 The Association between Vena Cava Filter Utilization and Patient and Hospital Characteristics in Patients with Deep Vein Thrombosis or Pulmonary Embolism, Hierarchical Logistic Regression (n=212, 395) ................................ .. 38 2 3 Comparison of Fit Statistics for Hierarchical Regression Modeling in the Association of Vena Cava Filter Utilization and Patient and Hospital Characteristics ................................ ................................ ................................ .... 41 3 1 Comparison of VCF utilization and overall cases by quartile, based on the hospital quartile of VCF utilization ................................ ................................ ...... 50 3 2 Comparison of VCF utilizatio n and quality of care measures by quartile, based on the hospital quartile of VCF utilization ................................ ................. 51 3 3 Comparison of VCF utilization and patient case mix characteristics by quartile, based on the hospital quartile of VCF utilization ................................ ... 53 3 4 Comparison of VCF utilization and hospital characteristics by quartile, based on the presence of VCF utilization over the VTE cases per hospital .................. 56

PAGE 9

9 LIST OF ABBREVIATIONS DVT HCUP IVC NRD PE USFDA VCF VTE Deep Vein Thrombosis Healthcare Cost and Utilization Project Inferior Vena Cava Nationwide Readmission Database Pulmonary Embolism U.S. Food and Drug Administratio n Vena Cava Filter Venous Thromboembolism

PAGE 10

10 Abstract of Presented to the Graduate School of t he Un iversity o f Flo rida in Partial Fulfillmen t o f the Requirements for the Degree o f Master of Science in P harmacy VARIATION IN QUALITY O F CARE IN USE OF VE NA CAVA FI LTERS FOR HOSPITALIZED VENOUS T HROMBOEMBOLISM PATIENTS By Ming Ch en December 2017 Chair: Joshua D. Brown Major: Pharmaceutical Sciences Concentration: Pharmaceutical Outcomes and Policy Objective: (1) To exami ne the association between VCF use and patient and hospital characteristics, and expl ain the variation at the patient level; (2) To delineate the distribution of four quality of care measures by hospital quartiles of VCF rates, and explain the variation at the hospital level. Setting: Nationwide Readmission Database (2013 2014) Method: Non There were 212,395 VTE hospitalizations, with 12.18% (n=25,877) receiving VCF placement. Significant association from adjusted are: age over 80 (aOR: 2.53, 95% CI: 2.25

PAGE 11

11 privately owned hospitals (aOR: 1.21, 95% CI: 1.08 1.36). Optimal goodness of fit was achieved w ith a combination of random effects and patient level fixed effects. (2) Overall, 12.29% of VTE patients had VCF placements at 1,614 hospitals. When stratified by quartiles, the proportion of VCF placements ranged from 0% to 7.28% in the lowest quartile o f hospitals to 15.84% to 46.84% in the highest. Conclusion: (1) Patient level variation in use of VCF for hospitalized VTE patients is associated with patient and hospital characteristics. (2) Hospital level variation also existed. Several patient case mix and hospital characteristics attributed to the variation.

PAGE 12

12 CHAPTER 1 INTRODUCTION TO VENA CAVA FILTER UTILIZATION IN VENOUS THROMBOEM BO LISM PATIENTS The Epidemiology o f Venous Thromboembolism Venous thromboembolism (VTE) includes both deep vein thrombo sis (DVT) and pulmonary embolism (PE). DVT mostly occurs in deep veins of legs, including c alf veins, femoral vein or popliteal veins, but can also develop in the veins of the arms and in the mesenteric and cerebral veins (Goldhaber & Bounameaux, 2012) When blood clots in veins travel to the lung, the lift threatening consequence of DVT is pulmonary embolism (Kesieme, Kesieme, Jebbin, Irekpita, & Dongo, 2011) The incidence of venous thromboembolism (VTE) is approximately 100 events per 10,000 person year, most of which are recurrent events and an estimate of two thirds are DVT cases (Hann & Streiff, 2005) The risk of VTE incr ease by age, and clinical f actors associated with an increased risk of VTE are presence of certain cancers by site (e.g., brain, lung, stomach, colon, pancreas and kidney) or metastatic cancers (Khorana et al., 2006 ; Khorana, Francis, Culakova, & Lyman, 2005; Levitan et al., 1999) certain types of surgery of which requires use of anesthesia at least 30 minutes, trauma and use of oral contraceptives and hormone replacement therapy (Anderson & Spencer, 2003) Ri sk of mortality is highest in patients with PE (Huerta, Johansson, Wallander, & Garcia Rodriguez, 2007) Treatment Management of Venous Thromboembolism a nd Vena Cava Filters According to American College of Chest Physician (ACCP) guideline in 2016 anticoagulants are recommend ed treatment for VTE patients as long term (first 3 months) therapy or extended therapy (after 3 months and no scheduled stop date),

PAGE 13

13 including the oral anticoagulants and low molecular weight heparin (Kearon et al., 2016) ACCP guideline s also recommend that the use of VCFs should be indicated for acute VTE patients who are contraindicated to anticoagulants because of the risk of bleeding. For VTE patients who have been treated with anticoagulants, the use of VCFs are not recommended (Kearon et al., 2016) A v ena cava filter is a cage like medical device that is percutaneously placed in the vena cava via femoral vein and jugular vein by interventional radiologist s and/or tr auma, vascular and/or cardiovascular surgeons to trap and /or prevent the blood clots in veins from travelling to the lung. U.S. Food and Drug Administration (USFDA) approved indication for use of vena cava filters for both prevention VTE and secondary VTE, including: Pulmonary thromboembolism when anticoagulants are contraindicated ; Failure of anticoagulant therapy in thromboembolic diseases ; Emergency treatment following massive pulmonary embolism where anticipated benefits of conventional therapy are red uced ; Chronic, recurrent pulmonary embolism where anticoagulant therapy has failed or is contraindicated U se of Vena Cava Filters i n Venous Thromboembolism Patients Previous studies showed concerns on the clinical outcomes and complications associated wit h VCFs A prospective study with 6 month follow up f or patients after VCF placement at a hospital setting in Japan reported the complications including filter thrombosis, IVC thrombosis, filter dislocation, equipment fracture, and catheter infection by dif ferent types of VCFs (Miyahara et al., 2006) The complication varied by brand and model of VCFs, with a range between 0% and 69% (Athanasoulis et al., 2000;

PAGE 14

14 Crowther, 2007; Van Ha et al., 2008; Yunus et al., 2008) In a randomized study (PREPIC) with eight year follow up for patients at the high risk of proximal DVT, the safety and efficacy of the preventive use of VCF for PE was examined compared to heparin therapy Even though no short term or long term mo rtality associated with VCFs were found, there was a decreased risk of occurrence of PE without major complications and a lagged risk of recurrent DVT However, this study was restricted to non retrievable VCFs (Dec ousus et al., 1998) With respect to the retrievable VCFs and the effect of time of retrievals on the rates of complications, i n a case report study by using Canada Registry on a single type of VCF (the Gunther Tulip Retrievable Filter, GTF) found if GT F s were removed within 25 days, there were no major complications (Millward et al., 2001) For the same t ype device GTF a case report on the prophylactic use of a retrievable VCF on a young aged trauma patient fai led retrievals occurred at 21 days and 25 days because of the risk of IVC laceration (Binkert, Bansal, & Gates, 2005) A second randomized study (PREPIC2) examined the safety and efficacy of retrievable VCFs, with 3.0% recurrent PE in the filter group and 1.5% recurrent PE in the control group at either 3 month or 6 month inter val for retrievals, respectively Other complications such as major bleeding, DVT, and death were not significantly different in intervals for VCF retrievals at these two time periods (Mismetti et al., 2015) Desp ite the lack of evidence, s tudies found there was an increased use of VCFs from 1979 to 2006 in the national hospital discharge survey setting, and a greater rate of VCF use was found in prevention VTE compared to secondary VTE (Stein, Matta, & Hull, 2011) A recent study using the nationwide inpatient sample data found that an

PAGE 15

15 increased use of VCFs from 2005 to 2010 (before 2010) however, there was a decreased use from 2010 to 2014 (after 2010) with an increased use of VCFs for secondary VTE compared with prevention VTE (V. Wadhwa et al., 2017) This is likely due to the warning of VCF use by USFDA i n 2010. USFDA warned the use of VCFs for a long time stay in patients, which would increase potential risk of adverse clinical outcomes, after receiving over 900 ad verse events caused by VCFs In 2010, USFDA released a recommendation on retrievable VCFs, which emphasized the need of retrievable VCFs by implanting physicians and clinicians for patients with VCF. Once the protection of PE is no longer needed, the VCFs should be retrieved (Morales et al., 2013) In 2014, USFDA u pdated recommendations with the time to retrievals after VCF placement; physicians should consider removing VCFs within 25 to 54 days (Ghatan & Ryu, 2016; Morales et al., 2013) Regarding the retrievable rates of VCF, a mean of 34% retrievable rate was reported by a systematic review with articles extracted from MEDLINE between 1966 to 2011 (Angel, Tapson, Galgon, Restrepo, & Kaufman, 2011) Taken the effect of the USFDA warning in 2010 on the rates of VCF use and retrievals into account a retrospective study at Truven Marketscan data setting found a significant increased r etrievable rate from 14.0% in 2010 to 38.2% in 2014 (P<0.001) however, the retrieval estimates were underestimated due to the claims data characteristics. After correcting the assumption of underestimation, the corrected estimates of retrievable rates app roximately ranged from 26% to 30%, which was consistent with previous studies (Angel et al., 2011; Brown, Raissi, Han, Adams, & Talbert, 2017; Desai, Naddaf, Pan, Hood, & Hodgson, 2016)

PAGE 16

16 Variation in Use o f Vena C ava Filters The variation in use of VCF for hospitalized VTE patients associated with patient characteristics and hospital level factors have been examined by using state level inpatient data in California and Kentucky, respectively. White s study found a range from 0% to 38.96% of VCF use between hospitals, and 18.49% residual variation was attributable to between hospital variation (White et al., 2013) A wide variation was also found using Kentucky inpatient data, from 0.4% to 15.2% between hospitals, and 12% residual variation in VCF use was accounted for b etween hospital variation. Several cancers by site was associated with an increased use of VCF, such as brain, cervical and stomach cancers. The variation of VCF use also depended on age, and older patients had a higher likelihood of VCF use (Brown & Talbert, 2017) The variation in use of VCF between hospitals could be accounted for case mix of patients and hospital characteristics (Brown & Talbert, 2016) With a restriction on cancer patients w ith VTE, the variation of VCF use was examined at the single state discharge lev el, which could be explained by between hospital variation and cancer sites. It was also likely due to variation among physician preference s within different hospitals (Ho, Brunson, White, & Wun, 2015) Another study found g eographic v ariation in use of VCF among 33 states in the United States us ing the State Inpatient Data. A 3 fold variation in VCF use was found at the state level after adjusting for the variation of VTE, including both PE and DVT. The eastern states were prone to overuse the VCFs, with increased rates of VCF use for prevention VTE (Meltzer et al., 2013) In summary, previous studies primarily focus ed on the variation of VCF use in hospitalized VTE patients or cancer patients with VTE at a single state or multiple state

PAGE 17

17 levels. Risk fact ors associated with the VCF use at individual patient level data were patient level demo socioeconomic characteristics comorbidities and hospital level factors; at the hospital level data the variation of VCF use could be explained by patient case mix and hospital level factors. However, t h e knowledge and evidence of the variation in use of VCF from a national perspective is absen t. Hence, this thesis sought to use a national administrative discharge database to identify the association between patient and hospital characteristics and the use of VCF, and explain both patient level and hosp ital level variation in VCF use. Also, quality of care measures were delineated at the hospital quartiles of VCF rates setting.

PAGE 18

18 CHAPTER 2 PATIENT LEVEL VARIATION IN USE OF V ENA CAVA FI L T ERS FOR HOSPITA LIZED VENOUS THROMBOEMBOLISM PATIENTS ASSOCIATED WITH PATIENT AND HOSPITAL CHARACTERISTICS Introduction Inferior Vena Cava Filters (VCFs) may be used as secondary prevention following venous thromboembolism (VTE), and are also used prophylactically for the prevention of pulmonary embolism (PE) in patients meeting certain high risk criteria such as failed or contraindicated anticoagulation (Kearon et al., 2016) Despite the increased risk for VTE recurrence following VCF placement (Group, 2005; Naddour et al., 2017) device related complications (Everhart, Vaccaro, Worley, Rogstad, & Seleznick, 2017) thrombosis (DVT) and PE patients (Wilbur & Shian, 2017) VCFs continue to be employed prophylactically and therapeutically in VTE patients. In 2005, the PREPIC study conducted a randomized controlled t rial of permanent VCF placements in DVT patients and followed up for eight years; it was reported that VCFs reduced PE risk but increased DVT, with overall no effect on survival rates (Group, 2005) A 2015 randomized controlled trial (the PREPIC2 study) of retrievable VCFs reached similar conclusions: where VTE patients were given either retrievabl e VCFs and anticoagulation or anticoagulation alone there were no discernible differences in recurrent PE after three months (Mismetti et al., 2015) Evidence on VCF effectiveness from studies in particular subgrou ps remains mixed. One observational study in 2012 reported that in hospital mortality rates were lower in stable patients that received VCFs as well as in unstable patients receiving thrombolytic therapy (Stein, Matta, Keyes, & Willyerd, 2012) A 2014 observational study also reported a decre ase

PAGE 19

19 in in hospital mortality in unstable adults with PE that received VCFs (Stein & Matta, 2014) Recent studies of VCF usage in cancer patients with VTE have demonstrated clinically poor outcomes. One 2017 popula tion based study of hospital discharge records found a higher 180 day risk of recurrent DVT in patients with cancer who received VCFs while these patients received no reduction of either 180 day PE risk or 30 day mortality (Brunson, Ho, White, & Wun, 2017) Another 2017 retr ospective cohort analysis of cancer and non cancer Canadian patients found that 21% of VCF placements occurred in patients with no anticoagulation contraindication and that a greater proportion of these patients (31%) had active cancer(s) and a high short term mortality rate, which suggests possible inappropriate VCF placement in end of life care settings in this population (Wassef, Lim, & Wu, 2017) Additionally, a 2017 meta analysis of VCF placements from randomized controlled trials in cancer and non cancer PE patients concluded that there is no evidence for routine VCF use in these populations as evidenced by a lack of reduction in both absolute and relative risk following VCF placement (Jerjes Sanchez et al., 2017) Emerging evidence in the literature suggests that VTE patients with either DVT or PE a re more likely to receive a VCF placement if they have certain types of cancers, if they have characteristics deemed high risk for bleeding from anticoagulation therapies, or if they are treated in hospitals that have certain characteristics (White et al., 2013) The purpose of this study is to expand on previ ous work to examine the association between patient and hospital characteristics and VCF utilization in venous thromboembolism patients using a large sample retrospective cohort design.

PAGE 20

20 Methods Data Source The Agency for Healthcare Research and Quality (A HRQ) 2013 and 2014 Nationwide Readmissions Database (NRD) was used in this retrospective cohort study. The NRD database includes all payer administrative hospital discharge claims from 21 states in 2013 and 22 states in 2014 with unique patient identifiers to facilitate follow up within each calendar year, which accounts for 85% of the discharges from all the State Inpatient Database. In addition to patient demographic information (e.g., age, gender, race, and zip code income), and diagnostic and procedural information from the NRD Core File, information on severity of illness and comorbidities were also identified by linkages via patient identification number in the NRD Severity File. Hospital characteristics (e.g., bed size, ownership, teaching status, rur al or urban) were included from the NRD Hospital File and linked via hospital identifiers to patient data within each calendar year. Cohort Selection The study cohort was defined as all VTE cases in patients aged 18 and older, based on International Class ification of Disease Codes version 9 (ICD 9), with primary diagnoses of either DVT (ICD 9 Codes 451.xx and 453.xx) or PE (ICD 9 Codes 415.1x). Index hospitalization for VTE patients were identified as the first hospitalization within each calendar year. El ective index hospitalizations were excluded. VCF placements were identified by ICD 9 procedural code 38.7 at any point in the 2013 2014 data following initial PE or DVT diagnosis. Hospitals that did not have the capacity to conduct VCF placement, defined a s performing zero VCF procedures within the study period, were excluded to avoid bias of hospital level characteristics. Hospitals were also

PAGE 21

21 required to have a minimum of 55 VTE hospitalizations during the study period to calculate reliable point estimates and 95% confidence intervals. Study Variables Patients with VTE were categorized as DVT or PE according to blood clot location. DVT patients were further classified by location of DVT to lower (ICD 9 Codes 451.0x, 451.1x, 451.2x, 451.81, 453.4, 453.5, 453.6), deep (ICD 9 Codes 451.1x cc451.81, 453.4, 453.5, 453.72, 453.82), proximal (ICD 9 Codes 451.81, 453.41, 453.51) and migrant DVT (ICD 9 Codes 453.1). Patient demographic, admission related, and socioeconomic variables were also included. Age was categorized by 10 year intervals, (e.g., 18 to 29, 30 to 39, 39 to 40, etc.). Admission day was categorized as on the weekend or on a weekday. Patient health insurance was classified as Medicare, Medicaid, private insurance, self pay, no charge, or other type of health insurance. Median ho usehold income was reported as a quartile classification of the large metropolitan areas, small metropolitan areas, and the micropolitan with other areas (rural). Count of chronic condit ions were aggregated into the following categories: Comorbidity measures included 28 Agency for Healthcare Research Quality (AHRQ) comorbidities and other comorbidities of interest. AHRQ comorbidities and were identified from the NRD severity data file, and other comorbidities of interest were selected based on previously published coding algorithms (Brown & Talbert, 2017) Comorbidities of interest included: hyperlipidemia, chronic obstructive pulmonary

PAGE 22

22 disorder (COPD), stroke, sepsis, infection, trauma, bleeding, thrombolysis, se. Hospital characteristics included hospital ownership (e.g., Government, non Federal; privately owned non profit; private, investment owned), bed size, teaching status, and urban United States) (HCUP NRD 2017) Urban and rural classification were based on population thresholds of residents in the hospital county, where categories were group (metropolitan area with <1 million residents), and micropolitan and other rural areas. Statistical Analysis Bivariate analysis via chi square analysis was performed to comp are patient and hospital characteristics between VCF users and non users following DVT or PE diagnosis at hospitalization. Bivariate testing for any patient characteristic or condition rarchical logistic regression model was constructed to analyze variance in the binary outcome VCF use, due to patient demographic and clinical characteristics and hospital characteristics being at hierarchical levels. Adjusted odds ratios and 95% confidenc e intervals were calculated after controlling for these factors, with a priori significance set at 0.05. Four hierarchical models with hospital as random effects and potential risk factors at the patient level and/or at the hospital level as fixed effects were specified to test assumptions of the original model and to explore goodness of fit across

PAGE 23

23 methodological alternatives. The intraclass correlation coefficient (ICC) was calculated to quantify the variance of VCF use between hospitals and within hospit als. To account for variation of VCF use by hospitals, a random effect only model was built as Model 1. Model 2 only included level 1 (patient level) fixed effects, that is, the patient characteristics potentially associated with VCF use as fixed effects. Model 3 was restricted to level 2 (hospital level) fixed effects. A full model including both level 1 and level 2 was subsequently performed as Model 4. C statistics were used as a measure of discrimination ability between VCF use and no VCF use. Akaike in formation criterion (AIC) and Bayesian information criterion (BIC) were calculated to assess the model fit across the four specified hierarchical models. All analysis was conducted using SAS version 9.4 (SAS Institute, Cary, NC). Results There were 212,39 5 VTE patients in the NRD database with a primary diagnosis of either DVT (n=89,482) or PE (n=122,913). Of these 12.18% (n=25,877) of these underwent VCF placement during the study period and n=186,518 did not receive a VCF. There were 1,524 hospitals, wit h a range from 0.60% of VTE patients receiving VCF to 47.10 in these hospitals. Table 2 1 provides a summary of patient and hospital characteristics, by patients who underwent the VCF procedure and patients who did not. In bivariate analysis, patients who received VCF were older (p<0.001), were admitted on a weekday more frequently than on the weekend (p=0.013), had Medicare more often than private insurance (63.82% VCF Medicare vs. 52.56% no VCF Medicare; p<0.001), and had a greater number of chronic condi tions (p<0.001). DVT and PE patients receiving VCF also were more likely to be treated in large metropolitan

PAGE 24

24 hospitals (p<0.001), non teaching hospitals (p<0.001), and private, investment owned hospitals (p<0.001), than patients who did not receive VCF. T able 2 2 details results from the hierarchical logistic regression results estimating the relationship between patient and hospital characteristics associated with VCF utilization. Following multivariate adjustment, several patient and hospital level chara cteristics that were associated with VCF placement remained statistically significant. DVT and PE Patients with private insurance were 1.07 times more likely to receive VCFs when compared with patients with Medicare (95% CI: 1.02 1.12), while patients with Medicaid insurance were 7% less likely to receive VCFs when compared with patients with Medicare (95% CI: 0.88 0.99). Older patients, specifically those over 80 years old, were 2.53 times more likely to receive VCFs than younger patients aged 18 to 29 yea rs old (95% CI: 2.25 2.85). Patients who had many chronic conditions were also more likely to receive VCFs; those with 13 or greater conditions were 3.85 times more likely to receive VCF (95% CI: 3.25 4.57) compared to patients without chronic conditions. A solid tumor without metastasis diagnosis resulted in 1.66 times greater likelihood of receiving VCF placement (95% CI: 1.57 1.75). At the hospital level, teaching status was not a significant predictor of whether or not a DVT or PE patient received VCF. Private, investment ownership of a hospital, however, was associated with a 1.21 times greater likelihood of VCF placement (95% CI: 1.08 1.36). Large metropolitan hospitals also placed VCFs 1.69 times more often than micropolitan/rural hospitals (95% CI: 1.44 1.99), as did small metropolitan hospitals (aOR: 1.48; 95% CI: 1.26 1.72), and hospitals with large bed sizes (aOR: 1.22; 95% CI: 1.10 1.35).

PAGE 25

25 Table 2 3 includes a comparison of the fit statistic results for four model fitting methodologies. Complete regression results for each model fitting are available from the authors. Taken in combination, AIC and BIC results (lower scores are better) also suggest that the full Model 4, which included fixed effects for hospital and patient characteristics and the random effects intercept for between hospital variation, was optimal. Model 4 also fit the data well with a c statistics of 0.80. In this final model, the ICC showed that 7.51% of the model variation was explained by the between hospital variation, i.e. th e random effects intercept. Discussion A previous study employing this design was limited to discharge data in a single state (Kentucky) (Brown & Talbert, 2017) Overall, the findings using this larger cohort and nationwide data suggest that DVT and PE patients who are treated at large, investment owned private hospitals may be more likely to receive a VCF placement than similar patients treated at smaller, nonprofit and/or Government hospitals. Patient characteristics such as having many comorbidities are also associated with whether or not they receive VCF placement and this supports findings from the Kentucky study. This study did not include as many cancer related comorb idities as in previous work, however, which were found to be significant patient level factors related to VCF utilization and so should be explored in future analyses using the larger, nationwide cohort. Additionally, it should be noted that the overall h ospital level factors, as gauged by hospital level random effects, were found to be a smaller contributor to variation in VCF utilization when compared with patient level factors in this study. These findings support evidence from the previous Kentucky stu dy that reached similar conclusions

PAGE 26

26 using a single state analysis of hospital variation in VCF utilization, where residual variation between hospital VCF utilization rate was reported to be low after controlling for patient case mix and hospital characteri stics (Brown & Talbert, 2016) Smaller residual variation in VCF use were also found in previous California study after controlling for patient characteristics, which reflected the small er between hospital variation in VCF use. Specifically, there were 18.49% residual variation in VCF use after controlling for patient clinical characteristics, and 18.47% residual variation after controlling both patient clinical and socioeconomic variable s, and 12.71% residual variation after controlling patient clinical, socioeconomic and hospital characteristics variables (White et al., 2013) C urrent study showed 7.51% between hospital variation and the previous Kentucky study showed 7.1% between hospital variation; both were with smaller variation than the 12.71% variation in California based study (White et al ., 2013) While the NRD data did not permit a state based analysis, comparing the results imply not only hospital level variation, but state based variation in VCF use as well. While anticoagulation is the mainstay of VTE treatment, VCF placement is indi cated for patients when anticoagulation is contraindicated, e.g., concurrent intracranial hemorrhage, massive gastrointestinal bleed or imminent planned surgery, or when patient has failed anticoagulation therapy, e.g., development of PE while on anticoagu lation therapy (Kearon et al., 2016) In other words, the decision for VCF placement is largely clinical, and the type of patient cohort at each institution or practice is likely to dictate the percentage of VTE pa tients receiving VCFs, leading to much variation in VCF usage across different institutions. Furthermore, current guideline encourages the usage of temporary instead of permanent VCF and call for retrieval as

PAGE 27

27 soon as clinically appropriate (Sutphin et al., 2015) E merging evidence suggests that the initiation of anticoagulation therapy is weakly associated with temporary VCF retrieval, which would indicate that there are other barriers to the timely retrieval of VCF s after placement (Brown et al., 2017) In the United States, there are two major guideline statements that address VCF implantation pra ctices in VTE patients: the American College of Chest Physicians (ACCP) guidelines on the management of VTE and the Society for Interventional Radiology (SIR) guidelines for the management of PE. ACCP guidelines do not call for prophylactic use and only re commend VCF placement in cases of contraindication for anticoagulation (Kearon et al., 2016) SIR guidelines, however, recommend prophylactic use in VTE patients at high risk for bleeding and/or are contraindicated for anticoagulation therapy (Caplin et al., 2011) The inconsistencies between these inter societal guidelines, along with loose adherence to these societal guidelines and the increase in availability of retrievable VCF, all have been blamed for significant overuse in IVC filters (Kaufman et a l., 2006) While the aforementioned factors may play a significant role in the variation of VCF filter placement across the United States, several other factors may have been largely overlooked (Knudson, 2013) VCF use varies from country to country, among the states within the United States and even at the county level despite adjusting for clinical and socioeconomic status (Alkhouli & Bashir, 2014; Bikdeli et al., 2016; White et al., 2013) Given that the efficacy of VCF filters and its (White, Zhou, Kim, & Romano, 2000) our study was designed to shed some light on potential underlying hospital and patient population related factors that may be contributing to whether

PAGE 28

28 patients receive or not receive VCFs as suggested by several papers (Brown & Talbert, 2017; Sarosiek, Crowther, & Sloan, 2013) In our study 12.18% of DVT and PE patien ts received a VCF placement during the study period, which is on the higher side of the VCF placement rates reported on the literature. This is a more frequent occurrence than the currently re usage of VCFs. Our results suggest that certain financial factors may have some level of influence on the rates of VCFs placed. VTE patients with private insurance are more likely to receive VCFs, and patients admitted to investment owned private hospitals are also more likely to have a VCF placed. The se findings may also explain the fact that Medicare (more generous reimbursement than Medicaid) patients also have a higher VCF placement rate than their Medicaid counterparts; however, this could be limited by selection bias given the fact that the Medica re population is generally older with overall higher comorbidities and increased rates of absolute or relative contraindications for anticoagulation therapy, leading to more VCF candidates in this population (Pickham et al., 2012) For a similar reason, such demographic factors, such as older age or Medicaid insurance, can also decrease the VCF retrieval rate (Smith, Shanks, Guy, Yang, & Dowell, 2015) Beyond f inanc ial factors, expertise availability may also have an impact on VCF placement rates as suggested by the higher placement rates in urban settings and during weekdays as compared with weekends. Several nonteaching hospitals have limited access to VCF imp lanting specialists, and the same is true for non urban hospitals. Many of these patients end up transferred to another facility for filter placement, or simply await regular working hours during the weekdays. These practice

PAGE 29

29 patterns could explain some of the dynamics of VCF filter placement across the country. Also, therapeutic VCF placement is not usually considered a weekend emergency and the implanting physician may defer placement to the first weekday in many clinical circumstances. As expected, the pr esence of more comorbidities did show a statistically significant correlation with increased VCF placement rate; a have overall clustering of underling VTE risk factors such as cancer and higher likelihood of having anticoagula tion contraindications. There are several limitations with the analysis and study design. First, follow up after the initial DVT or PE diagnosis was limited to the 2013 2014 study period, which means that VTE patients may have received a VCF placement at some point after the study period and these patients would not be labeled in our analysis as receiving VCF. Longitudinal tracking was not available for individual patients across calendar years due to a lack of patient identification linkage variable, so i ndividual patients can only be analyzed within a year. No medication use data was available; hence, we could not compare patients who are contraindicated to anticoagulation, or patients with anticoagulation failure. Additionally, patients who received a DV T or PE diagnosis at an NRD participating hospital may have received VCF at a hospital that does not report readmission statistics to NRD, which would render our estimates of VCF usage as conservative. There are also limitations with under reporting and mi sclassification inherent to the use of administrative data and diagnosis/procedure codes, so it is possible that VCF usage is conservatively recorded in this data. Also, information on race was not included in NRD so race cannot be examined using this data Last, the retrospective cohort study design allows for the construction of a highly powered

PAGE 30

30 statistical analysis with large sample sizes, but does present limitations to the generalizability of these results due to the possibility of selection bias. Con clusion This study was undertaken to assess whether a combination of patient and hospital level factors correlates with the rate of VCF placement in VTE patients. The findings in this study contribute to the growing body of evidence that VCFs may be inapp ropriately employed in certain populations and that hospitals should evaluate their PE patients. This study indicated the need for additional research to investigate whether the utilization of VCFs could be explained by hospital variation and patient characteristics at a national level.

PAGE 31

31 Table 2 1 Patient and Hospital Characteristics by Vena Cava Filter Use (VCF) in all Deep Vein Thrombosis and Pulmonary Embolism Pat ients, from the 2013 2014 Nationwide Readmissions Database n (% ) Characteristics VTE Patients Receiving VCFs (N=25,877) VTE Patients that Did Not Receive VCFs (N=186,518) P value % Receiving VCF Patient Characteristics Age 18~29 444(1.72) 7876(4.22) <.0001 5.34 30~39 848(3.28) 13200(7.08) 6.04 40~49 2061(7.96) 22784(12.22) 8.3 0 50~59 3770(14.57) 32800(17.59) 10.31 60~69 5432(20.99) 37907(20.32) 12.53 70~79 5832(22.54) 36582(19.61) 13.75 80+ 7490(28.94) 35369(18.96) 17.48 Gender Male 12282(47.46) 89095(47.77) 0.3581 12.12 Female 13595(52.54) 97423(52.23) 12.25 Admission day Workdays 20445(79.01) 145720(78.13) 0.0013 12.3 0 Weekends 5432(20.99) 40798(21.87) 11.75

PAGE 32

32 Table 2 1. Continued n (% ) Charact eristics VTE Patients Receiving VCFs (N=25,877) VTE Patients that Did Not Receive VCFs (N=186,518) P value % Receiving VCF Insurance Medicare 16515(63.82) 98027(52.56) <.0001 14.42 Medicaid 1982(7.66) 20119(10.79) 8.97 Private insurance 5788(22.3 7) 51888(27.82) 10.04 Self pay 729(2.82) 8614(4.62) 7.8 0 No charge 136(0.53) 1515(0.81) 8.24 Other 727(2.81) 6355(3.41) 10.27 Median household income 0 to 25th percentile 7022(27.14) 50616(27.14) 0.0688 12.18 26th to 50th percentile 6651(25.7 0) 48547(26.03) 12.05 51st to 75th percentile 6106(23.60) 44707(23.97) 12.02 76th to 100th percentile 6098(23.57) 42648(22.87) 12.51 Patient Location (Urban/rural status) Large metropolitan 15581(60.21) 109404(58.66) <.0001 12.47 Small metropol itan 7684(29.69) 59097(31.68) 11.51 Micropolitan/others 2612(10.09) 18017(9.66) 12.66

PAGE 33

33 Table 2 1. Continued n (% ) Characteristics VTE Patients Receiving VCFs (N=25,877) VTE Patients that Did Not Receive VCFs (N=186,518) P value % Receiving VCF Number of Chronic Conditions 0 353(1.36) 7049(3.78) <.0001 4.77 1~3 5282(20.41) 56207(30.13) 8.59 4~6 9888(38.21) 68439(36.69) 12.62 7~9 6956(26.88) 38188(20.47) 15.41 10~12 2672(10.33) 13161(7.06) 16.88 13+ 726(2.81) 3474(1.86) 17.29 DVT Type Lower DVT 21052(81.35) 97409(52.22) <.0001 17.77 Deep DVT 20850(80.57) 98241(52.67) <.0001 17.51 Proximal DVT 13286(51.34) 55161(29.57) <.0001 19.41 Migrans DVT 32(0.12) 90(0.05) <.0001 26.23 AHRQ Comorbidity Acquired immune deficie ncy syndrome 67(0.26) 546(0.29) 0.342 10.93 Alcohol abuse 963(3.72) 5899(3.16) <.0001 14.03 Deficiency anemias 7283(28.14) 38158(20.46) <.0001 16.03

PAGE 34

34 Table 2 1. Continued n (% ) Characteristics VTE Patients Receiving VCFs (N=25,877) VTE Pati ents that Did Not Receive VCFs (N=186,518) P value % Receiving VCF AHRQ Comorbidity (continued) Rheumatoid arthritis/collagen vascular diseases 971(3.75) 7013(3.76) 0.952 12.16 Chronic blood loss anemia 756(2.92) 1557(0.83) <.0001 32.68 Congestive heart failure 2466(9.53) 21307(11.42) <.0001 10.37 Chronic pulmonary disease 5600(21.64) 40308(21.61) 0.9123 12.2 0 Coagulopath 3134(12.11) 12226(6.55) <.0001 20.4 0 Depression 2848(11.01) 21515(11.54) 0.0123 11.69 Diabetes, uncomplicated 5501(21.26) 357 99(19.19) <.0001 13.32 Diabetes with chronic complications 922(3.56) 7078(3.79) 0.0665 11.53 Drug abuse 475(1.84) 6078(3.26) <.0001 7.25 Hypertension 16224(62.7) 109133(58.51) <.0001 12.94 Hypothyroidism 3514(13.58) 23114(12.39) <.0001 13.2 0 Liver dis ease 850(3.28) 4802(2.57) <.0001 15.04 Lymphoma 477(1.84) 2833(1.52) <.0001 14.41

PAGE 35

35 Table 2 1. Continued n (% ) Characteristics VTE Patients Receiving VCFs (N=25,877) VTE Patients that Did Not Receive VCFs (N=186,518) P value % Receiving VCF AHRQ Comorbidity (continued) Fluid and electrolyte disorders 7576(29.28) 37457(20.08) <.0001 16.82 Metastatic cancer 3006(11.62) 12482(6.69) <.0001 19.41 Other neurological disorders 3062(11.83) 15210(8.15) <.0001 16.76 Obesity 4591(17.74) 36827(19 .74) <.0001 11.08 Paralysis 1042(4.03) 3993(2.14) <.0001 20.7 0 Peripheral vascular disorders 2013(7.78) 7246(3.88) <.0001 21.74 Psychoses 1056(4.08) 8595(4.61) 0.0001 10.94 Pulmonary circulation disorders 2429(9.39) 25516(13.68) <.0001 8.69 Renal fail ure 3795(14.67) 24326(13.04) <.0001 13.5 0 Solid tumor without metastasis 2171(8.39) 9720(5.21) <.0001 18.26 Peptic ulcer disease excluding bleeding 15(0.06) 45(0.02) 0.0030* 25 .00 Valvular disease 1139(4.4) 9835(5.27) <.0001 10.38 Weight loss 2110(8.15 ) 8138(4.36) <.0001 20.59 Other comorbidities

PAGE 36

36 Table 2 1. Continued n (% ) Characteristics VTE Patients Receiving VCFs (N=25,877) VTE Patients that Did Not Receive VCFs (N=186,518) P value % Receiving VCF AHRQ Comorbidity (continued) Hyperlipidemia 9115(35.22) 63043(33.8) <.0001 12.63 COPD 7425(28.69) 48292(25.89) <.0001 13.33 Stroke 1485(5.74) 6189(3.32) <.0001 19.35 Sepsis 666(2.57) 1835(0.98) <.0001 26.63 Infection/Pneumonia 5201(20.1) 28650(15.36) <.0001 15.36 Trauma 852(3.2 9) 2924(1.57) <.0001 22.56 Bleeding 2020(7.81) 3432(1.84) <.0001 37.05 Thrombolysis 2456(9.49) 4793(2.57) <.0001 33.88 Embolectomy 332(1.28) 321(0.17) <.0001 50.84 Unstable 1054(4.07) 2388(1.28) <.0001 30.62 Hospital Characteristics Bed Size Small 1807(6.98) 15613(8.37) <.0001 10.37 Medium 6765(26.14) 50992(27.34) 11.71 Large 17305(66.87) 119913(64.29) 12.61

PAGE 37

37 Table 2 1. Continued n (% ) Characteristics VTE Patients Receiving VCFs (N=25,877) VTE Patients that Did Not Receive V CFs (N=186,518) P value % Receiving VCF Ownership Government, nonfederal 2851(11.02) 21291(11.41) <.0001 11.81 Private, not profit 18483(71.43) 135931(72.88) 11.97 Private, invest own 4543(17.56) 29296(15.71) 13.43 Urban rural designation L arge metropolitan areas 16277(62.9) 112746(60.45) <.0001 12.62 Small metropolitan areas 8835(34.14) 66493(35.65) 11.73 Non metropolitan areas 765(2.96) 7279(3.9) 9.51 Teaching Status Non teaching 11134(43.03) 78359(42.01) 0.0019 12.44 Teaching 1 4743(56.97) 108159(57.99) 12 .00

PAGE 38

38 Table 2 2 The Association between Vena Cava Filter Utilization and Patient and Hospital Characteristics in Patients with Deep Vein Thrombosis or Pulmonary Embolism, Hierarchical Logistic Regression (n=212,395) Charac teristics Adjusted Odds Ratio 95% CI Patient Characteristics Age 18~29 Ref. Ref. 30~39 1.06 0.93 1.20 40~49 1.3 1.16 1.46 50~59 1.49 1.34 1.67 60~69 1.75 1.56 1.95 70~79 1.91 1.70 2.15 80+ 2.53 2.25 2.85 Gender Male Ref. Ref. Female 1 0.9 7 1.03 Admission day Workdays Ref. Ref. Weekends 0.99 0.96 1.03 Insurance Medicare Ref. Ref. Medicaid 0.93 0.88 0.99 Private insurance 1.07 1.02 1.12 Self pay 0.89 0.81 0.98 No charge 0.9 0.74 1.10 Other 1.02 0.93 1.11 Median household incom e 0 25th percentile Ref. Ref. 26th to 50th percentile 1 0.96 1.05 51st to 75th percentile 1.06 1.02 1.11 76th to 100th percentile 1.06 1.00 1.11 Patient Location (Urban/rural status) Micropolitan/other rural Ref. Ref. Small metropolitan 0.81 0.7 6 0.87 Large metropolitan 0.76 0.70 0.82 Number of Chronic Conditions 0 Ref. Ref. 1~3 1.57 1.40 1.77 4~6 2.21 1.96 2.49 7~9 2.79 2.45 3.17 10~12 3.34 2.90 3.85 13+ 3.85 3.25 4.57 DVT Type Lower DVT 3.32 3.04 3.63 Deep DVT 1.16 1.06 1.26

PAGE 39

39 Table 2 2. Continued Characteristics Adjusted Odds Ratio 95% CI DVT Type (continued) Proximal DVT 1.22 1.18 1.26 Migrans DVT 1.63 1.02 2.59 AHRQ Comorbidity Acquired immune deficiency syndrome 0.87 0.66 1.14 Alcohol abuse 1.18 1.09 1.28 Def iciency anemias 1.15 1.11 1.19 Rheumatoid arthritis/collagen vascular diseases 0.89 0.82 0.96 Chronic blood loss anemia 2.59 2.33 2.88 Congestive heart failure 0.68 0.64 0.71 Chronic pulmonary disease 0.21 0.19 0.22 Coagulopath 1.33 1.26 1.39 D epression 0.87 0.83 0.91 Diabetes, uncomplicated 0.95 0.92 0.99 Diabetes with chronic complications 0.73 0.67 0.79 Drug abuse 0.72 0.65 0.80 Hypertension 0.88 0.85 0.91 Hypothyroidism 0.92 0.88 0.96 Liver disease 1.08 0.99 1.17 Lymphoma 1.03 0. 93 1.15 Fluid and electrolyte disorders 1.24 1.20 1.28 Metastatic cancer 1.63 1.55 1.72 Other neurological disorders 1.18 1.13 1.24 Obesity 1.01 0.97 1.06 Paralysis 1.35 1.24 1.48 Peripheral vascular disorders 1.91 1.80 2.04 Psychoses 0.9 0.84 0 .97 Pulmonary circulation disorders 0.28 0.26 0.30 Renal failure 0.76 0.73 0.80 Solid tumor without metastasis 1.66 1.57 1.75 Peptic ulcer disease excluding bleeding 0.61 0.32 1.16 Valvular disease 0.72 0.67 0.77 Weight loss 1.19 1.12 1.26 Other comorbidities Hyperlipidemia 0.87 0.84 0.90 COPD 5.1 4.69 5.55 Stroke 1.17 1.09 1.27 Sepsis 1.24 1.11 1.40 Infection/Pneumonia 1.19 1.14 1.24 Trauma 2.03 1.86 2.22 Bleeding 3.53 3.30 3.77 Thrombolysis 3.15 2.97 3.34

PAGE 40

40 Table 2 2. Continued C haracteristics Adjusted Odds Ratio 95% CI AHRQ Comorbidity (continued) Embolectomy 4.45 3.72 5.33 Unstable 1.96 1.78 2.15 Hospital Characteristics Bed Size Small Ref. Ref. Medium 1.1 0 0.99 1.22 Large 1.22 1.10 1.35 Hospital Ownership Gove rnment, nonfederal Ref. Ref. Private, not profit 0.96 0.87 1.06 Private, invest own 1.21 1.08 1.36 Urban rural designation Non metropolitan areas Ref. Ref. Small metropolitan areas 1.48 1.26 1.72 Large metropolitan areas 1.69 1.44 1.99 Teaching S tatus Non teaching Ref. Ref. Teaching 0.98 0.91 1.04

PAGE 41

41 Table 2 3 Comparison of Fit Statistics for Hierarchical Regression Modeling in the Association of Vena Cava Filter Utilization and Patient and Hospital Characteristics Model 1: Model 2: M odel 3: Model 4: Random Effects Only Model 1 + Level 1 a Fixed Effects Model 1 + Level 2 b Fixed Effects Model 1 + Level 1 and Level 2 Fixed Effects Intercept (SE) 2.10 (0.015) 5.47(0.480) 1.82(0.046) 5.72(0.481) Hospital Random Effects (SE) 0.28(0.0 14) 0.28(0.015) 0.26(0.013) 0.27(0.014) ICC (%) c 7.82 7.93 7.41 7.51 C statistic 0.66 0.80 0.65 0.80 AIC d 154,280.70 132,038.50 154,210.00 131,962.70 BIC d 154,291.50 132,406.80 154,258.80 132,369.00 a Level 1: patient level characteristics. b Level 2: hospital level characteristics. c The intraclass correlation coefficient (ICC) accounts for the variance of VCF use between hospitals and within hospitals. d Akaike information criterion (AIC) and Bayesian information criterion (BIC) assess the model f it by penalizing the addition of parameters (e.g., smaller values indicate better fit). The Association between Vena Cava Filter Utilization and Patient and

PAGE 42

4 2 CHAPTER 3 HOSPITAL LEVEL VARIATION IN USE OF VENA CAVA FI L T ERS FOR HOSPITALIZED VENOUS THROM BOEMBOLISM PATIENTS ATTRIBUTABLE TO PATIENT AND HOSPITAL CHARACTERISTICS Introduction The utilization of Inferior Vena Cava Filters (VCFs) has been decreasing overall in the United States since 2010, but the proportion of patients receiving VCFs who have venous thromboembolism (VTE) has been increasing since 2005 (V. Wadhwa et al., 2017) In an analysis of the Nationwide Inpatient Sample Database between 2000 and 2009, Kuy, et al. (2013) reported that 84.7% of VCF placements occurred in patients with VTE, with majority of these placements (94.6%) occurring in urban hospitals (Kuy et al., 2014) A 2013 st udy of VTE patients found substantial variation in VCF utilization among California hospitals, much of which was attributed to hospital characteristics such as having a smaller number of beds, rural location, private ownership, and several patient level ch aracteristics (such as acute bleeding at admission, and metastatic cancer) (White et al., 2013) Interestingly, an older study using the Nationwide Inpatient Sample Database between 2006 and 2008 found no statistically significant relationships between VCF placements in VTE patients and cancer diagnoses, nor w ith hospital characteristics such as teaching status or hospital size (Meltzer et al., 2013) A recent study of VTE patients by hospitals in Kentucky (2016) found that patient case mix was a strong predictor varia tion in VCF utilization (Brown & Talbert, 2016) A follow up study employing hierarchical logistic regression analysis to control for both patient and hospital level characteristics amon g the population of Kentucky VTE patients found that patients with higher risks (e.g., older patients, having multiple

PAGE 43

43 comorbidities, metastatic tumors) were more likely to receive VCFs (Brown & Talbert, 2017) After controlling for patient and hospital level characteristics, between hospital variation in VCF utilization was approximately 7%. Most previous studies on this subject have been restricted to populations within s ingle states, or employ older data. The objective of this study is to build on previous work by employing a database to build a VTE patient cohort from a larger, multi state population to examine between hospital variation in VCF utilization by quartiles b ased on the VCF rates across hospitals. Method Data This retrospective cohort study was conducted by using the Nationwide Readmission Database (2013 2014) from the Agency for Healthcare Research and Quality (AHRQ). The NRD is a n all payer administrative i npatient database in the United States, with inpatient records drawn from 21 State Inpatient Database (SID) in 2013 and 22 States in 2014. Unique patient linkage numbers were identified to enable the follow up within each calendar year. This study was waiv ed from formal review and informed consent by the institutional review board of University of Florida because NRD data was publicly available and de identified. Use of the NRD follows regulations within the data use agreement as defined by the HCUP and Age ncy of Healthcare Research and Quality. Study Population VTE patients were identified by International Classification of Diseases, Ninth Revision, Clinical Modification (ICD 9) diagnoses codes, as of a primary diagnosis in PE (ICD 9, 415.1x) or DVT (ICD 9 451.xx and 453.xx). Patients aged 18 or older at the

PAGE 44

44 time of index hospitalization were included. Elective hospitalizations were excluded. VCFs were identified as ICD 9 procedure code 38.7. Stratification by Quartile Hospitals were stratified by VCF ra te s into four quartiles. VCF rate was calculated as the proportion of VTE patients per eligible hospital that underwent VCFs within each calendar year. These proportions were stratified into quartiles representing hospital quartiles with ascending order of VCF rate s from the lowest quartile to the highest. The number and rate of VCF placements were defined at the hospital level. Hospital quartiles were identified at the cluster level of hospitals, with a range of VCF rates. The number of VCFs by quartile wa s calculated as the sum of cumulative VCFs placed in each hospital within the quartile. Patient Case Mix a nd Hospital Level Characteristics Patient characteristics included gender, age, race, location of DVT, insurance status, median income of zip code, u rban/rural location, length of stay, weekend admission, and number of chronic conditions. Case mix comorbidities included 28 AHRQ comorbidities measure, and hyperlipidemia, chronic obstructive pulmonary disease (COPD), sepsis, infection, trauma, bleeding, stroke, thrombolysis, embolectomy and unstable based on previous coding algorithms (Quan et al., 2005; White et al., 2000) Hospital characteristics included teaching status, urban/rural location, bed size and owne rship. Hospitals with no VCF placements were deemed as no availability for the procedure and excluded. Hospitals with at least 55 VTE cases were included to ensure the precision of estimation for 95% confidence intervals (White et al., 2013)

PAGE 45

45 Main O utcome Primary outcome measures were rates of in hospital mort ality, mean scores of severity of illness and risk of mortality, and rates of 30 day readmissions of all causes and primary cause as VTE over the corresponding index hospitalizations by quartile. Secondary outcome measures were in hospital cost and length of stay Rate of in hospital mortality was calculated as the proportions of patients who died in index hospitalization The number of index VTE hospitalizations with alive discharge s between January and November annually were used as denominator to calcul ate the rates of 30 day readmissions of all causes and VTE as the primary cause. We further subgrouped these index VTE hospitalization s into those with VCF s and calculated the relevant rates of 30 day readmission s over the index hospitalizations In hospi tal cost were calculated by multiplying the in hospital total charges with the cost to charge ratio. All costs were adjusted by the annual inflation rates from U.S. Consumer Price Index to year 2016. Leng th of stay was measured in NRD for the days of in ho spital stay between the index admission and discharge. Statistical Analysis Descriptive analyses were performed to delineate the distribution of patient case mix and hospital level factors by quartile. T he numbers of VTE hospitalization, VCF placement and hospitals by quartile were tabulated, as well as the proportions of the VCF operative rates. Proportion of PE patients over VTE patients was calculated to indicate the severity of patient case mix by quartile. Proportions of patient case mix variables and hospital level factors were also tabulated by quartile Cochran Armitage trend test s (P trend test) w ere used to analyze the linear trend s in the rates of VCF placements, in hospital mortality, and the rates of 30 day readmissions over the

PAGE 46

46 corresponding i ndex hospitalizations across the quartiles of hospital s ANOVA test was used for the mean difference of the continuous variables (e.g., risk of mortality s cores cost and length of stay ). All analyses were performed using SAS version 9.4 and JMP 13 Pro (SA S Institute, Cary, NC). All tests were two tailed. Statistical significance was assessed at P value < 0.05. Results O verall 207,246 VTE patients were identified, with 25,470 (12.29%) undergoing VCF placement 1,6 1 4 hospitals meeting eligibility criteria pl aced VCFs The frequency of VCF placements ranged from 0% to 7.28% (Median: 4.95%, IQR: 2.86%) in the lowest hospital quartile, 7.29% to 10.97% quartile (Median: 9.14%, IQR: 1.95%) in the second, 10.98% to 15.83% (Median: 13.04%, IQR: 2.15%) in the third q uartile and 15.84% to 46.84% (Median: 19.35%, IQR: 4.74%) in the highest quartile of hospital specific VCF rates quartile. Table 3 1 showed a greater proportion of VCF placements occurred in hospitals with a lower proportion of PE patients (P value < 0.00 01), with the highest quartile having 53.35% of PE cases relative to 61.41% PE cases in the lowest quartile. This suggests that hospitals treating more DVTs than PEs conduct a greater proportion of VCF procedures. Table 3 2 presented the quality of care me asures by hospital quartiles Among hospitalized VTE patients with 30 day readmission including all cause readmission and readmission of VTE as primary cause at 30 days a higher proportion were found in the higher hospital quartiles, with a significant l inear trend across the quartiles (Both P<0.05). Mean severity of illness score was relatively higher in the second and third quartiles. Significant mean differences in in hospital cost and length of stay were found

PAGE 47

47 in all VTE inpatients and those eligible for 30 day readmissions (All P value < 0.05). Among hospitalized VTE patients with VCF, significant difference also existed. However, n o significance was found in VTE inpatients with VCF that were allowed for 30 day readmissions (All P v alue > 0.05). Com pared with hospitalized VTE patients with 30 day readmissions, either the all cause readmission or the readmission of VTE as primary diagnosis the correspond ing hospitalization s with VCF that were allowed for 30 day readmission had higher mean costs and l ength of stay. Table 3 3 detailed patient case mix characteristics stratified by hospital quartiles. VTE patients who received care in large metropolitan areas were significantly more likely to receive VCF placements (P value < 0.0001), and the opposite s ituations significantly occurred in patients with access to care in small metropolitan areas, micropolitan and other areas, respectively (P value < 0.0001). VTE patients with either lower or deep or proximal DVT were significantly more likely to received VCF placements (P value < 0.0001). There were significant trends of increased VCF use for VTE patients who underwent thrombolysis and/or embolectomy (P value < 0.0001). In addition, a greater proportion of deficiency anemia comorbidities were found in the highest quartile of VCF placements, as were the presence of other complicating comorbidities such as hypertension, renal failure and cancer (All with P value < 0.0001). In the highest quartile of VCF placements, patients had more Medicare insurance, and a greater proportion of patients were aged 80 or older (P value < 0.0001). These significant factors represented progressive effects on the VCF placement over the hospital quartile.

PAGE 48

48 Table 3 4 included the stratification of VCF placements in VTE cases by hos pital characteristics. Private, non profit and private, investment owned hospitals performed the majority of VCF placements in the highest quartiles. Hospitals in urban areas including both large and small metropolitan areas performed more VCF placements i n the highest quartile compared to rural areas. Significant trends were found in hospital locations, that is, the higher quartile, the more hospitals in large metropolitan areas and less hospitals in small metropolitan areas and non metropolitan areas (P v alue <0.001). Discussion When analyzed by quartiles, only 0.00% to 7.28% of VTE patients received VCFs in the lowest quartile but 15.84% to 46.84% of VTE patients received VCFs in the highest quartile. This 15 .55 % in overall variation of VCF utilization m ay primarily be explained in differences between patient case mix characteristics compared with hospital level characteristics, as noticeable significances were observed in case mix comorbidities while only hospital ownership showed significant association with VCF use across quartiles of VCF placements. A recent study found that hospitals in the southern region of the United States place VCFs more often than hospitals in other regions (Vibhor Wadhwa et al., 2017) This suggested that an additional layer of hospital characteristics, namely location, should be explored in follow up studies. Additionally, another study found that larger hospital bed size was associated with more VCF placements in a California VTE pa tient population (White et al., 2013) however, this as sociation was not observed in this national scale study, which indicated that the variations of VCF use might be geographical and regional. This study also reported more VCF placements in private

PAGE 49

49 hospitals and urban hospitals in California, which aligned w ith our findings using a national data (White et al., 20 13) To our knowledge, this study is the first national study to delineate the patient case mix and hospital level characteristics attributable to the variation of VCF utilization by hospital qu artile among VTE patient s. However, several limitations exist ed. First, NRD data did not contain indicators for state or region of the hospitalization, so a regional analysis of hospital variation was not conducted. Second, the descriptive analysis design did not include multivariate adjustment, hence, follow up stu dies controlling for both patient and hospital level characteristics and accounting for the clustering of hospitals via multilevel analysis should be conducted. Additionally, there were limitations associated with the use of claims data, which typically in cluded risk of conservative classification of diagnoses due to reliance on billing codes. Conc lusion Wide variations in use of VCFs were found at the hospital level, addressing the issue on the low quality of care. Considering that current treatment guidel ines study suggest that additional examination of risk factors associated with hospital variation in use of VCF among VTE patients is warranted.

PAGE 50

50 Table 3 1. Comparison o f VCF utilization and overall cases by quartile, based on the hospital quartile of VCF utilization Characteristics Hospital Quartiles of VCF Rates Lowest Quartile Second Quartile Third Quartile Highest Quartile P Value a Hospitals (N) 402 405 40 4 403 VTE Hospitalization (N) 41218 54840 59268 51920 PE / VTE (%) 61.41 59.11 56.33 53.35 <0.0001 VCF Placements N 2,018 5,001 7,832 10,619 VCF / VTE (%) 4.90 9.12 13.21 20.45 <0.0001 VCF per hospital, mean 5 12 19 26 VCF rate (min max, %) 0.00 7.28 7.29 10.97 10.98 15.83 15.84 46.84 VCF rate (median, %) 4.95 9.14 13.04 19.35 VCF rate (IQR, %) 2.86 1.95 2.15 4.74 a Cochran Armitage trend test

PAGE 51

51 Table 3 2 Comparison of VCF utilization and quality of care measures by quartile, based on the hospital quartile of VCF utilization Hospital Quartiles of VCF Rates Characteristics Lowest Quartile Second Quartile Third Quartile Highest Quartile P value a Hospitalized VTE Patients N 30 day all cause readmission, n (%) 4864 (11.80) 6758 (12.32) 7753 (13.08) 6897 (13.28) <0.0001 3 0 day VTE readmission, n (%) 848 (2.06) 1086 (1.98) 1258 (2.12) 1144 (2.20) 0.0337 Hospitalized VTE Patients with VCFs, N 30 day all cause readmission, n (%) 332 (16.45) 839 (16.78) 1315 (16.79) 1739 (16.38) 0.6437 30 day VTE readmission, n (%) 52 (2.58) 120 (2.40) 167 (2.13) 230 (2.17) 0.1962 In hospital mortality (%) 2.15 2.29 2.28 2.23 0.5696 Severity of illness (SOI), mean Hospitalized VTE Patients 2.26 2.28 2.30 2.30 <0.0001 30 day all cause readm ission 2.48 2.49 2.51 2.49 0.0843 30 day VTE readmission 2.18 2.23 2.23 2.24 0.4035 Hospitalized VTE Patients with VCFs 2.69 2.68 2.68 2.61 <0.0001 30 day all cause readmission 2.70 2.77 2.79 2.73 0.0626 30 day VTE readmission 2.29 2.56 2.51 2.50 0.2586 Risk of mortality, mean Hospitalized VTE Patients 2.00 2.02 2.05 2.06 <0.0001 30 day all cause readmission 2.20 2.23 2.26 2.27 0.0003 30 day VTE readmission 1.82 1.88 1.89 1.95 0.0156 Hospitalized VTE Patients with VCF 2.49 2.52 2.51 2.44 <0.0001 30 day all cause readmission 2.49 2.60 2.63 2.56 0.0225 30 day VTE readmission 2.06 2.35 2.34 2.28 0.2594

PAGE 52

52 Table 3 2. Continued Hospital Quartiles of VCF Rates Characteristics Lowest Quartile Second Quartile Third Quartile Highest Quar tile P value a Length of Stay (LOS), d, mean SD Hospitalized VTE Patients 4.70 4.71 4.95 5.28 5.17 5.51 5.36 5.43 <0.0001 30 day all cause readmission 5.56 5.38 5.85 5.24 6.11 7.06 6.31 6.74 <0.0001 30 day VTE readmission 4.3 3.6 9 4.65 3.43 5.08 5.93 5.19 4.35 <0.0001 Hospitalized VTE Patients with VCF 7.96 7.67 7.88 10.16 7.84 7.71 7.35 7.71 <0.0001 30 day all cause readmission 8.01 6.62 8.30 7.42 8.47 9.17 8.33 9.67 0.8590 30 day VTE readmission 5.92 4.58 6.23 4.32 7.08 6.07 6.81 6.00 0.4414 Cost, $, mean SD Hospitalized VTE Patients 10,07312,922 10,07311,748 11,414 15,091 11,810 15,665 <0.0001 30 day all cause readmission 11764 13408 11694 12440 13012 16227 13713 21931 <0 .0001 30 day VTE readmission 9,780 9,707 9,687 9,020 11,150 12,093 11,471 11,368 <0.0001 Hospitalized VTE Patients with VCF 22,67523,870 20,33520,871 21,909 23,626 20,421 24,169 <0.0001 30 day all cause readmission 21 500 19 993 20 7 82 18 821 21 747 24 384 21 871 37 191 0.8396 30 day VTE readmission 16,17510,173 17,42410,540 19,572 15,208 19,493 16,475 0.2880

PAGE 53

53 Table 3 3 Comparison of VCF utilization and patient case mix characteristics by quartile, based on the h ospital quartile of VCF utilization Hospital Quartiles of VCF Rates Patient Case Mix Characteristics (%) Lowest Quartile Second Quartile Third Quartile Highest Quartile P value a Age 18~29 4.32 3.86 4.11 3.55 <0.0001 30~39 7.08 6.79 6. 93 5.91 <0.0001 40~49 12.15 12.07 11.87 10.91 <0.0001 50~59 17.56 17.67 17.6 0 16.12 <0.0001 60~69 20.25 20.49 20.38 20.24 0.7839 70~79 19.71 19.96 19.46 20.49 0.029 0 80+ 18.92 19.16 19.65 22.78 <0.0001 Gender Female 51.77 52.51 52.06 52.65 0.04 89 Admission day Weekends 21.94 21.81 21.50 21.70 0.2349 Insurance Medicare 52.29 53.48 53.05 56.56 <0.0001 Medicaid 12.21 10.45 10.76 9.30 <0.0001 Private insurance 26.46 27.36 27.38 25.96 0.045 Self pay 4.44 4.82 4.54 4.03 <0.0001 No ch arge 0.76 0.70 0.76 0.96 0.0001 Other 3.84 3.19 3.51 3.19 <0.0001 Median household income 0 25th percentile (lowest) 27.24 28.03 26.87 28.15 0.1389 26th to 50th percentile (median) 27.30 25.86 26.27 25.65 <0.0001 51st to 75th percentile 24.27 25. 14 24.43 21.25 <0.0001 76th to 100th percentile (highest) 21.18 20.98 22.42 24.95 <0.0001 Patient Urban/rural status Large metropolitan 49.55 55.49 60.61 64.94 <0.0001 Small metropolitan 35.63 34.70 30.50 27.67 <0.0001 Micropolitan/others 14.83 9 .81 8.89 7.39 <0.0001 Number of Chronic Conditions 0 3.54 3.65 3.54 3.36 0.0636 1~3 29.94 29.05 28.87 29.17 0.0149 4~6 36.94 37.26 37.09 37.04 0.9753 7~9 20.59 21.26 21.44 21.24 0.0178 10~12 7.11 7.14 7.24 7.42 0.0509 13+ 1.87 1.64 1.81 1.77 0 .7645

PAGE 54

54 Table 3 3. Continued Hospital Quartiles of VCF Rates Patient Case Mix Characteristics (%) Lowest Quartile Second Quartile Third Quartile Highest Quartile P value a DVT Type Lower DVT 51.52 54.58 56.85 60.60 <0.0001 Deep DVT 51. 71 54.81 57.19 60.96 <0.0001 Proximal DVT 28.69 30.77 33.97 35.13 <0.0001 Migrans DVT 0.08 0.05 0.04 0.07 0.413 0 AHRQ Comorbidity Acquired immune deficiency syndrome 0.25 0.29 0.29 0.32 0.0759 Alcohol abuse 3.48 3.24 3.27 2.93 <0.0001 Deficiency anemias 19.42 20.45 21.52 23.55 <0.0001 Rheumatoid arthritis/collagen vascular diseases 3.6 3.58 3.89 3.84 0.0062 Chronic blood loss anemia 0.95 1.06 1.11 1.20 0.0002 Congestive heart failure 11.4 11.12 11.44 10.75 0.0132 Chronic pulmonary disease 22.39 21.83 21.24 21.08 <0.0001 Coagulopathy 6.56 6.94 7.46 7.76 <0.0001 Depression 12.40 11.80 11.73 10.70 <0.0001 Diabetes (uncomplicated) 19.27 19.96 19.64 20.02 0.0318 Diabetes (complicated) 3.24 3.47 3.45 3.57 0.0131 Drug abuse 3.64 3.25 2.9 6 2.66 <0.0001 Hypertension 57.49 58.64 58.89 60.57 <0.0001 Hypothyroidism 12.6 12.53 12.65 12.37 0.3869 Liver disease 2.66 2.66 2.63 2.53 0.2031 Lymphoma 1.55 1.42 1.56 1.67 0.0228 Fluid and electrolyte disorders 20.83 21.5 21.3 21.59 0.0209 Me tastatic cancer 6.77 6.79 7.55 7.57 <0.0001 Other neurological disorders 8.4 8.37 8.73 8.84 0.0026 Obesity 19.41 19.88 19.17 18.35 <0.0001 Paralysis 2.05 2.30 2.38 2.65 <0.0001 Peripheral vascular disorders 3.49 3.49 3.77 4.02 <0.0001 Psychoses 4.5 1 4.30 4.37 4.23 0.0895 Pulmonary circulation disorders 13.6 13.44 13.37 12.42 <0.0001 Renal failure 12.8 12.73 13.17 13.38 0.001 0 Solid tumor without metastasis 5.30 5.59 5.71 5.70 0.0067 Peptic ulcer disease w/o bleeding 0.04 0.02 0.03 0.02 0.153 0 Valvular disease 4.92 5.35 5.00 5.23 0.3019 Weight loss 4.62 4.33 4.70 4.65 0.1973

PAGE 55

55 Table 3 3. Continued Hospital Quartiles of VCF Rates Patient Case Mix Characteristics (%) Lowest Quartile Second Quartile Third Quartile Highest Quartile P v alue a Other comorbidities Hyperlipidemia 32.82 33.74 33.77 33.60 0.0279 COPD 26.75 26.46 25.92 25.76 <0.0001 Sepsis 0.99 1.16 1.18 1.27 0.0002 Infection 15.78 15.91 16.03 15.97 0.3826 Trauma 1.84 1.78 1.73 1.77 0.3431 Bleeding 2.28 2.53 2.73 2.65 <0.0001 Stroke 3.33 3.45 3.61 3.87 <0.0001 Unstable 1.47 1.68 1.72 1.60 0.1267 Thrombolysis 2.54 2.78 3.90 4.34 <0.0001 Embolectomy 0.18 0.22 0.37 0.44 <0.0001 a Cochran Armitage trend test

PAGE 56

56 Table 3 4 Comparison of VCF utilization and hospit al characteristics by quartile, based on the presence of VCF utilization over the VTE cases per hospital Hospital Quartiles of VCF Rates Hospital Characteristics (%) Lowest Quartile Second Quartile Third Quartile Highest Quartile P value a Bed Size Small 11.84 7.60 8.39 4.90 <0.0001 Medium 34.22 27.13 21.53 27.60 <0.0001 Large 53.94 65.27 70.07 67.50 <0.0001 Ownership Government, nonfederal 12.52 12.40 10.52 11.34 <0.0001 Private, not profit 71.91 71.36 75.35 69.12 <0.0001 Private, inv est own 15.57 16.24 14.12 19.54 <0.0001 Urban rural designation Large metropolitan areas 50.26 56.59 63.74 67.22 <0.0001 Small metropolitan areas 41.75 39.63 33.57 30.70 <0.0001 Non metropolitan areas 7.99 3.78 2.69 2.08 <0.0001 Teaching Status Teaching 50.31 60.00 65.79 50.69 0.0001 a Cochran Armitage trend test

PAGE 57

57 CHAPTER 4 CONCLUSIONS AND FUTURE DIRECTIONS These two studies evaluated variation in use of VCFs among hospitalized VTE patients at a national level Risk factors associated with the VCF use were examined, and the quality of care measures such as 30 day readmission rate, in hospital mortality, in hospital cost and length of day were also described at the setting of hospital quartiles of VCF rates These studies concluded wide varia tion in use of VCF associated with both patient and hospital characteristics, which indicated the needs for future studies on variation of quality of care. The process of completing this thesis involves a deep exploration of published studies with respect to use of VCF, the retrievable rates of VCF, risk factors, cancer patients with VTE, variation study, and quality of care indicators. The future directions are driv en by these exploration, and listed as below: Firstly, studies to examine the risk facto rs of VCF use associated with 30 day and 90 day readmission are warranted, with subgroup analysis to compare the readmission assoc iated with u se of VCF for prevention VTE and secondary VTE. Secondly, most of the studies used hierarchical generalized linea r model to examine the risk factors of readmission. However, competing risk of death could bias the rates of readmission after the index hospitalization of patients who underwent VCFs. A future study would take the effect of competing risk into account and evaluate the risk of readmission then compare the findings with the results of the previous point. Thirdly, to evaluate the association between 30 day readmission and 1 year mortality, which would be estimated using Cox proportional hazards models with readmission as a time dependent covariate and by using landmark analysis. The main

PAGE 58

58 outcome measures were all cause 30 day readmission to any hospital following VCF and 1 year mortality. Fourthly, the previous three points do not have anticoagulation medica tion involved. There is a need for the comparison of use in VCF and anticoagulant s for VTE patients associated with clinical outcomes. Fifthly, as US FDA warned the over use of VCF in 2010 to propose the retrievals of VCF, it suggests future study to examin e the rates of retrievable VCFs at either state level or national level setting. There would also be variation in the quality of care in retrievable rates of VCF use.

PAGE 59

59 LIST OF REFERENCES Alkhouli, M., & Bashir, R. (2014). Inferior vena cava filters in the United States: less is more. Int J Cardiol, 177 (3), 742 743. doi:10.1016/j.ijcard.2014.08.010 Anderson, F. A., Jr., & Spencer, F. A. (2003). Risk factors for venous thromboembolism. Circulation, 107 (23 Suppl 1), I9 16. doi:10.1161/01.CI R.0000078469.07362.E6 Angel, L. F., Tapson, V., Galgon, R. E., Restrepo, M. I., & Kaufman, J. (2011). Systematic review of the use of retrievable inferior vena cava filters. J Vasc Interv Radiol, 22 (11), 1522 1530 e1523. doi:10.1016/j.jvir.2011.08.024 Atha nasoulis, C. A., Kaufman, J. A., Halpern, E. F., Waltman, A. C., Geller, S. C., & Fan, C. M. (2000). Inferior vena caval filters: review of a 26 year single center clinical experience. Radiology, 216 (1), 54 66. doi:10.1148/radiology.216.1.r00jl1254 Bikdeli B., Wang, Y., Minges, K. E., Desai, N. R., Kim, N., Desai, M. M., . Krumholz, H. M. (2016). Vena Caval Filter Utilization and Outcomes in Pulmonary Embolism: Medicare Hospitalizations From 1999 to 2010. J Am Coll Cardiol, 67 (9), 1027 1035. doi:10.101 6/j.jacc.2015.12.028 Binkert, C. A., Bansal, A., & Gates, J. D. (2005). Inferior vena cava filter removal after 317 day implantation. J Vasc Interv Radiol, 16 (3), 395 398. doi:10.1097/01.RVI.0000150029.86869.DE Brown, J. D., Raissi, D., Han, Q., Adams, V. R., & Talbert, J. C. (2017). Vena Cava Filter Retrieval Rates and Factors Associated With Retrieval in a Large US Cohort. J Am Heart Assoc, 6 (9). doi:10.1161/JAHA.117.006708 Brown, J. D., & Talbert, J. C. (2016). Variation in the Use of Vena Cava Filters f or Venous Thromboembolism in Hospitals in Kentucky. JAMA Surg, 151 (10), 984 986. doi:10.1001/jamasurg.2016.1004 Brown, J. D., & Talbert, J. C. (2017). Hospital Variation and Patient Characteristics Associated With Vena Cava Filter Utilization. Med Care, 55 (1), 31 36. doi:10.1097/MLR.0000000000000599 Brunson, A., Ho, G., White, R., & Wun, T. (2017). Inferior vena cava filters in patients with cancer and venous thromboembolism (VTE) does not improve clinical outcomes: A population based study. Thromb Res, 153 57 64. doi:10.1016/j.thromres.2017.03.012 Caplin, D. M., Nikolic, B., Kalva, S. P., Ganguli, S., Saad, W. E., Zuckerman, D. A., & Society of Interventional Radiology Standards of Practice, C. (2011). Quality improvement guidelines for the performance of inferior vena cava filter placement for the prevention of pulmonary embolism. J Vasc Interv Radiol, 22 (11), 1499 1506. doi:10.1016/j.jvir.2011.07.012

PAGE 60

60 Crowther, M. A. (2007). Inferior vena cava filters in the management of venous thromboembolism. Am J Med, 120 (10 Suppl 2), S13 17. doi:10.1016/j.amjmed.2007.07.015 Decousus, H., Leizorovicz, A., Parent, F., Page, Y., Tardy, B., Girard, P., . Simonneau, G. (1998). A clinical trial of vena caval filters in the prevention of pulmonary embolism in patients wit h proximal deep vein thrombosis. Prevention du Risque d'Embolie Pulmonaire par Interruption Cave Study Group. N Engl J Med, 338 (7), 409 415. doi:10.1056/NEJM199802123380701 Desai, S. S., Naddaf, A., Pan, J., Hood, D., & Hodgson, K. J. (2016). Impact of con sensus statements and reimbursement on vena cava filter utilization. J Vasc Surg, 64 (2), 425 429. doi:10.1016/j.jvs.2016.01.046 Everhart, D., Vaccaro, J., Worley, K., Rogstad, T. L., & Seleznick, M. (2017). Retrospective analysis of outcomes following infe rior vena cava (IVC) filter placement in a managed care population. J Thromb Thrombolysis, 44 (2), 179 189. doi:10.1007/s11239 017 1507 z Ghatan, C. E., & Ryu, R. K. (2016). Permanent versus Retrievable Inferior Vena Cava Filters: Rethinking the "One Filter for All" Approach to Mechanical Thromboembolic Prophylaxis. Semin Intervent Radiol, 33 (2), 75 78. doi:10.1055/s 0036 1582123 Goldhaber, S. Z., & Bounameaux, H. (2012). Pulmonary embolism and deep vein thrombosis. Lancet, 379 (9828), 1835 1846. doi:10.1016/ S0140 6736(11)61904 1 Group, P. S. (2005). 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 112 (3), 416 422. doi:10.1161/CIRCULATIONAHA.104.512834 Hann, C. L., & Streiff, M. B. (2005). The role of vena caval filters in the management of venous thromboembolism. Blood Rev, 19 (4), 179 202. doi:10.1016/j.blre.2004.08.002 Ho, G., Brunson, A., White, R., & Wun, T. (2015). Vena cava filter use in cancer patients with acute venous thromboembolism in California. Thromb Res, 135 (5), 809 815. doi:10.1016/j.thromres.2015.02.002 Huerta, C., Johansson, S., Wallander, M. A., & Garcia Rodriguez, L. A. (2007). R isk factors and short term mortality of venous thromboembolism diagnosed in the primary care setting in the United Kingdom. Arch Intern Med, 167 (9), 935 943. doi:10.1001/archinte.167.9.935 Jerjes Sanchez, C., Rodriguez, D., Navarrete, A., Parra Cantu, C., Joya Harrison, J., Vazquez, E., & Ramirez Rivera, A. (2017). Inferior vena cava filters in pulmonary embolism: A historic controversy. Arch Cardiol Mex, 87 (2), 155 166. doi:10.1016/j.acmx.2017.01.007

PAGE 61

61 Kaufman, J. A., Kinney, T. B., Streiff, M. B., Sing, R. F., Proctor, M. C., Becker, D., . Venbrux, A. C. (2006). Guidelines for the use of retrievable and convertible vena cava filters: report from the Society of Interventional Radiology multidisciplinary consensus conference. J Vasc Interv Radiol, 17 (3), 4 49 459. doi:10.1097/01.rvi.0000203418.39769.0d Kearon, C., Akl, E. A., Ornelas, J., Blaivas, A., Jimenez, D., Bounameaux, H., . Moores, L. (2016). Antithrombotic Therapy for VTE Disease: CHEST Guideline and Expert Panel Report. Chest, 149 (2), 315 352. doi:10.1016/j.chest.2015.11.026 Kesieme, E., Kesieme, C., Jebbin, N., Irekpita, E., & Dongo, A. (2011). Deep vein thrombosis: a clinical review. J Blood Med, 2 59 69. doi:10.2147/JBM.S19009 Khorana, A. A., Francis, C. W., Culakova, E., Fisher, R. I., Kude rer, N. M., & Lyman, G. H. (2006). Thromboembolism in hospitalized neutropenic cancer patients. J Clin Oncol, 24 (3), 484 490. doi:10.1200/JCO.2005.03.8877 Khorana, A. A., Francis, C. W., Culakova, E., & Lyman, G. H. (2005). Risk factors for chemotherapy as sociated venous thromboembolism in a prospective observational study. Cancer, 104 (12), 2822 2829. doi:10.1002/cncr.21496 Knudson, M. M. (2013). Hospital specific risk factors for filter fever. JAMA Surg, 148 (7), 687 688. doi:10.1001/jamasurg.2013.2286 Kuy, S., Dua, A., Lee, C. J., Patel, B., Desai, S. S., Dua, A., . Patel, P. J. (2014). National trends in utilization of inferior vena cava filters in the United States, 2000 2009. J Vasc Surg Venous Lymphat Disord, 2 (1), 15 20. doi:10.1016/j.jvsv.2013.08. 007 Levitan, N., Dowlati, A., Remick, S. C., Tahsildar, H. I., Sivinski, L. D., Beyth, R., & Rimm, A. A. (1999). Rates of initial and recurrent thromboembolic disease among patients with malignancy versus those without malignancy. Risk analysis using Medic are claims data. Medicine (Baltimore), 78 (5), 285 291. Meltzer, A. J., Graham, A., Kim, J. H., Connolly, P. H., Karwowski, J. K., Bush, H. L., . Schneider, D. B. (2013). Clinical, demographic, and medicolegal factors associated with geographic variati on in inferior vena cava filter utilization: an interstate analysis. Surgery, 153 (5), 683 688. doi:10.1016/j.surg.2012.11.005 Millward, S. F., Oliva, V. L., Bell, S. D., Valenti, D. A., Rasuli, P., Asch, M., . Kachura, J. R. (2001). Gunther Tulip Retri evable Vena Cava Filter: results from the Registry of the Canadian Interventional Radiology Association. J Vasc Interv Radiol, 12 (9), 1053 1058.

PAGE 62

62 Mismetti, P., Laporte, S., Pellerin, O., Ennezat, P. V., Couturaud, F., Elias, A., . Group, P. S. (2015) Effect of a retrievable inferior vena cava filter plus anticoagulation vs anticoagulation alone on risk of recurrent pulmonary embolism: a randomized clinical trial. JAMA, 313 (16), 1627 1635. doi:10.1001/jama.2015.3780 Miyahara, T., Miyata, T., Shigemats u, K., Deguchi, J., Kimura, H., Ishii, S., & Nagawa, H. (2006). Clinical outcome and complications of temporary inferior vena cava filter placement. J Vasc Surg, 44 (3), 620 624. doi:10.1016/j.jvs.2006.05.019 Morales, J. P., Li, X., Irony, T. Z., Ibrahim, N G., Moynahan, M., & Cavanaugh, K. J., Jr. (2013). Decision analysis of retrievable inferior vena cava filters in patients without pulmonary embolism. J Vasc Surg Venous Lymphat Disord, 1 (4), 376 384. doi:10.1016/j.jvsv.2013.04.005 Naddour, M., Kalani, M. Hattab, Y., Gandhi, V., Singh, A. C., & Bajwa, O. (2017). Prognosis and Monitoring of VTE. Crit Care Nurs Q, 40 (3), 288 300. doi:10.1097/CNQ.0000000000000167 Pickham, D. M., Callcut, R. A., Maggio, P. M., Mell, M. W., Spain, D. A., Bech, F., & Staudenmay er, K. (2012). Payer status is associated with the use of prophylactic inferior vena cava filter in high risk trauma patients. Surgery, 152 (2), 232 237. doi:10.1016/j.surg.2012.05.041 Quan, H., Sundararajan, V., Halfon, P., Fong, A., Burnand, B., Luthi, J. C., . Ghali, W. A. (2005). Coding algorithms for defining comorbidities in ICD 9 CM and ICD 10 administrative data. Med Care, 43 (11), 1130 1139. Sarosiek, S., Crowther, M., & Sloan, J. M. (2013). Indications, complications, and management of inferior vena cava filters: the experience in 952 patients at an academic hospital with a level I trauma center. JAMA Intern Med, 173 (7), 513 517. doi:10.1001/jamainternmed.2013.343 Smith, S. C., Shanks, C., Guy, G., Yang, X., & Dowell, J. D. (2015). Social and De mographic Factors Influencing Inferior Vena Cava Filter Retrieval at a Single Institution in the United States. Cardiovasc Intervent Radiol, 38 (5), 1186 1191. doi:10.1007/s00270 014 1046 3 Stein, P. D., & Matta, F. (2014). Vena cava filters in unstable eld erly patients with acute pulmonary embolism. Am J Med, 127 (3), 222 225. doi:10.1016/j.amjmed.2013.11.003 Stein, P. D., Matta, F., & Hull, R. D. (2011). Increasing use of vena cava filters for prevention of pulmonary embolism. Am J Med, 124 (7), 655 661. doi :10.1016/j.amjmed.2011.02.021

PAGE 63

63 Stein, P. D., Matta, F., Keyes, D. C., & Willyerd, G. L. (2012). Impact of vena cava filters on in hospital case fatality rate from pulmonary embolism. Am J Med, 125 (5), 478 484. doi:10.1016/j.amjmed.2011.05.025 Sutphin, P. D ., Reis, S. P., McKune, A., Ravanzo, M., Kalva, S. P., & Pillai, A. K. (2015). Improving inferior vena cava filter retrieval rates with the define, measure, analyze, improve, control methodology. J Vasc Interv Radiol, 26 (4), 491 498 e491. doi:10.1016/j.jvi r.2014.11.030 Van Ha, T. G., Chien, A. S., Funaki, B. S., Lorenz, J., Piano, G., Shen, M., & Leef, J. (2008). Use of retrievable compared to permanent inferior vena cava filters: a single institution experience. Cardiovasc Intervent Radiol, 31 (2), 308 315. doi:10.1007/s00270 007 9184 5 Wadhwa, V., Trivedi, P. S., Chatterjee, K., Tamrazi, A., Hong, K., Lessne, M. L., & Ryu, R. K. (2017). Decreasing Utilization of Inferior Vena Cava Filters in Post FDA Warning Era: Insights From 2005 to 2014 Nationwide Inpati ent Sample. J Am Coll Radiol, 14 (9), 1144 1150. doi:10.1016/j.jacr.2017.04.022 Wadhwa, V., Trivedi, P. S., Chatterjee, K., Tamrazi, A., Hong, K., Lessne, M. L., & Ryu, R. K. (2017). Decreasing Utilization of Inferior Vena Cava Filters in Post FDA Warn ing Era: Insights From 2005 to 2014 Nationwide Inpatient Sample. Journal of the American College of Radiology doi:10.1016/j.jacr.2017.04.022 Wassef, A., Lim, W., & Wu, C. (2017). Indications, complications and outcomes of inferior vena cava filters: A ret rospective study. Thromb Res, 153 123 128. doi:10.1016/j.thromres.2017.02.013 White, R. H., Geraghty, E. M., Brunson, A., Murin, S., Wun, T., Spencer, F., & Romano, P. S. (2013). High variation between hospitals in vena cava filter use for venous thromboe mbolism. JAMA Intern Med, 173 (7), 506 512. doi:10.1001/jamainternmed.2013.2352 White, R. H., Zhou, H., Kim, J., & Romano, P. S. (2000). A population based study of the effectiveness of inferior vena cava filter use among patients with venous thromboembolism. Arch Intern Med, 160 (13), 2033 2041. Wilbur, J., & Shian, B. (2017). Deep V enous Thrombosis and Pulmonary Embolism: Current Therapy. Am Fam Physician, 95 (5), 295 302. Yunus, T. E., Tariq, N., Callahan, R. E., Niemeyer, D. J., Brown, O. W., Zelenock, G. B., & Shanley, C. J. (2008). Changes in inferior vena cava filter placement o ver the past decade at a large community based academic health center. J Vasc Surg, 47 (1), 157 165. doi:10.1016/j.jvs.2007.08.057

PAGE 64

64 BIOGRAPHICAL SKETCH Ming Chen joined the graduate program in Pharmaceutical Outcomes and Policy at the University of Flori da in August 2015. She started to get exposed to observational study and touch ed upon the secondary administrative claims database. Prior to it, she received both her Bachelor in Pharmaceutical Business Management in May 2011 with thesis focus on drug pric ing regulation, and her Master of Science degree in Social and Administrative Pharmacy from China Pharmaceutical University, Nanjing, China, in May 2014, with specialization in pharmaceutical policy in biosimilars, and orphan drug approvals. She completed her capstone on biosimilars from a regulatory perspective at Amgen China while a graduate student at China Pharmaceutical University Aft worked as a research associate in the Research Center of National Drug Policy and Ecosystem, China Pharmaceut ical University, Nanjing, China, and decided to pursue a further study abroad with concentration in pharmaceutical outcomes and policy research.