Current Medical Research & Opinion
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
0300-7995 doi:10.1185/03007995.2014.998814
Vol. 31, No. 3, 2015, 439–447
Article RT-0356.R1/998814 All rights reserved: reproduction in whole or part not permitted
Original article Risk of recurrent venous thromboembolism among deep vein thrombosis and pulmonary embolism patients treated with warfarin
Beth L. Nordstrom
Abstract
Evidera, Lexington, MA, USA
Michael A. Evans Geisinger Medical Center, Danville, PA, USA
Brian R. Murphy
Objective: Guidelines for warfarin treatment of venous thromboembolism (VTE) recommend targeting an international normalized ratio (INR) level of 2–3. This study examines the association between INR levels and VTE recurrence among warfarin-treated patients.
Evidera, Lexington, MA, USA
Edith A. Nutescu University of Illinois at Chicago, Chicago, IL, USA
Jeff R. Schein Brahim K. Bookhart Janssen Scientific Affairs LLC, Raritan, NJ, USA Address for correspondence: Beth L. Nordstrom PhD MPH, Evidera, 430 Bedford Street, Suite 300, Lexington, MA 02420, USA. Tel.: +1 781-960-0250; Fax: +1 781-761-0147; [email protected] Keywords: Deep vein thrombosis – International normalized ratio – Pulmonary embolism – Venous thromboembolism – Warfarin Accepted: 9 December 2014; published online: 24 December 2014 Citation: Curr Med Res Opin 2015; 31:439–47
Methods: A retrospective cohort study in the MedMining electronic health record database included adults treated with warfarin for VTE in 2004–2011. INR levels during warfarin use were categorized as below therapeutic range (52), in range (2–3), or above range (43), with time in each category estimated using the Rosendaal method. Recurrent VTE was noted from 30 days after the initial VTE to end of follow-up, which ranged up to 8 years. The incidence of recurrent VTE was calculated, and association with time-varying INR levels estimated using Cox models. Results: Of 1753 qualifying patients, 867 had deep vein thrombosis, and 886 had pulmonary embolism. Mean age was 58 years, and 50.7% were female. Across all follow-up time, VTE recurrences were observed in 134 (7.6%) patients, at a rate of 3.2 (95% confidence interval [CI]: 0.7–9.1) events per 100 person-years. The risk of VTE recurrence was greater during time spent with INR52 than with INR in the therapeutic range (hazard ratio [HR]: 3.37; 95% CI: 2.16–5.27). Low platelet counts also predicted greater risk of VTE recurrence (HR: 2.13; 95% CI: 1.24–3.67). Limitations: Exposure to warfarin and other anticoagulants was estimated based on prescription data and may be inaccurate. The study data include care within a single health system; thus, care received outside of the health system may be missing, and results may not be generalizable to the broader US population. Conclusions: Approximately 8% of patients experienced a recurrent VTE during follow-up. Subtherapeutic INR levels were associated with a more than three-fold increased risk of VTE recurrence.
Introduction Venous thromboembolism (VTE) affects an estimated 300,000–600,000 individuals in the United States annually1. VTE includes the two related conditions of deep vein thrombosis (DVT) and pulmonary embolism (PE). The wide list of risk factors for VTE include hereditary factors such as deficiencies of ! 2014 Informa UK Ltd www.cmrojournal.com
Risk of recurrent VTE Nordstrom et al.
439
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Current Medical Research & Opinion Volume 31, Number 3
March 2015
antithrombin as well as acquired risk factors such as cancer, surgery, obesity, heart failure, and ongoing hormonal treatment)2. For the treatment and prophylaxis of VTE, the American College of Chest Physicians (ACCP) recommends anticoagulation therapy consisting of a vitamin K antagonist in combination with unfractionated heparin (UH), low molecular weight heparin (LMWH), or fondaparinux, or more recently available combinations involving novel oral anticoagulants3. Warfarin, a common vitamin K antagonist, is prescribed for patients with VTE and is initially paired with heparin, LMWH, or fondaparinux until its full effect is established in the patient. The correct dose of warfarin for each patient is guided by frequent monitoring of the target international normalized ratio (INR). The recommended target INR range for VTE treatment is between 2.0 and 3.0. Because INR outside of this range can result in bleeding or a recurrent thrombotic event, it is imperative that patients’ INR be regularly monitored3–5. A patient is most at risk for a recurrent event during the first 6 months following an initial VTE. During those 6 months, the cumulative proportion of patients experiencing a recurrent event at 2 weeks is 2%, increasing to 6.4% at 3 months and reaching 8–10% at 6 months6,7. A systematic review estimated a rate of VTE recurrence of 3.3% per patient-year during the first 2 years after stopping anticoagulation therapy following a VTE, where the index VTE was provoked by transient risk factors8. VTE not provoked by known risk factors was more likely to be followed by a recurrence; the same review estimated a recurrence rate of 7.4% per patient-year in such cases8. The objective of this study was to evaluate the association between INR during warfarin therapy and the subsequent risk of recurrent VTE among patients treated with warfarin for an initial VTE event.
Patients and methods Data source and study design Data were obtained from the MedMining database of electronic health record (EHR) data for approximately 3.9 million patients from a large integrated health system in Pennsylvania. The health system includes an 870þ multispecialty physician group practice, three hospital campuses, 64 primary and specialty clinic sites, and a clinical reference laboratory. Available data included inpatient, outpatient, and prescription drug data, along with laboratory results and clinical characteristics, such as height, weight, and tobacco status. Data were extracted from 1 January 2004 through 31 March 2012 for all patients in the MedMining database 440
Risk of recurrent VTE Nordstrom et al.
with a diagnosis of VTE (International Classification of Diseases, 9th Revision, Clinical Modification [ICD9-CM] codes 452.xx, 453.xx, or 415.1x). Each patient was assigned an index date corresponding to the date of the first VTE event. Uncomplicated VTE is typically treated on an outpatient basis in this health system; for these outpatient events, the first date of outpatient diagnosis was designated as the index date. For inpatient events, to ensure that the date of the event could be ascertained, a primary or secondary diagnosis of VTE was required. To ensure that the date of the VTE event could be determined accurately, patients were required to have either a date of entry of the VTE diagnosis into the system or a record of the VTE as both an admission and a discharge diagnosis. For the latter subset, the date of the event was assumed to be the first day of the hospitalization. Eligible patients were required to have a first diagnosis of VTE between 31 January 2004 and 31 December 2011 (3 months prior to end of the study data), followed by warfarin initiation within the next 3 days, to be at least 18 years old on the index date, and to have at least two INR results available during the follow-up period. Patients were excluded from the analysis if their date of VTE event was unclear, if their earliest recorded VTE predated the study start date, if they received warfarin within 30 days prior to the index date, or if their first visit in the EHR system occurred less than 30 days prior to the index date.
Exposure to anticoagulation therapy Warfarin exposure was defined based on the start and stop dates noted in the EHR and was considered to be continuous as long as there was no more than a 30 day gap in exposure. Restarting warfarin after longer than a 30 day gap was considered a new episode of warfarin use. Longterm maintenance on warfarin therapy was not a requirement for the study cohort; patients were included who had as few as 2 days of warfarin treatment noted in the EHR during follow-up. Use of bridging therapies, including heparins, pentasaccharides, and direct thrombin inhibitors, on the warfarin start date was noted, including the type and duration of each therapy. For these drugs, any gap of 1 day or longer ended the episode of continuous use. Bridging therapy was classified with respect to whether it met the criteria for ACCP-recommended therapy overlap (i.e. therapy lasting for at least 5 days and until the INR is at least 2.0 and stable). For many patients, the optimal bridging therapy approach may differ from the standard ACCP recommendations, but the reasons for a reduction in bridging may not be identifiable through the database, and hence were not considered in this classification. www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 31, Number 3
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
INR measurement All available INR values were captured for each qualifying patient from the index date to end of follow-up, where the end of follow-up for each patient was defined as the date of last activity in the database or date of death for deceased patients. An INR measurement taken on the first day of warfarin exposure was treated as a baseline measurement rather than reflecting the INR during warfarin exposure. Each INR value was categorized as falling within the ACCP-recommended therapeutic range (2.0–3.0), below range (52.0), or above range (43.0).
Outcomes VTE diagnosis codes were identified from the index date to end of follow-up in the database for each patient. To allow for a complete assessment of recurrent events that may have happened either during the course of warfarin therapy or later, the outcome period included any available time after the end of warfarin treatment. The VTE outcomes were examined both combined and separately as DVT (ICD-9-CM codes 452.xx, 453.xx) and PE (ICD-9-CM code 415.1). ICD-9-CM codes indicating major bleed events, such as hemorrhage of various types (e.g. intracranial, gastrointestinal), menstrual disorders involving excessive bleeding, and blood in stool, were included. A complete list of the bleed outcome codes is available from the authors upon request. To be considered a recurrent VTE event rather than follow-up care for the index event, the VTE outcome was required to occur more than 30 days after the index date and to be associated with a visit reason indicating an initial anticoagulation visit. For patients with an index VTE that occurred during a hospitalization, outcome events were required to occur after discharge from the hospitalization for the initial VTE event to be included in the analysis, as it is not possible to determine whether an additional entry of a VTE diagnosis during a hospitalization represents a new event or the initial event. When a VTE recurrence outcome was found, the type (i.e. DVT or PE) and days from the index date to the date of the event were determined for the first VTE event that occurred during followup for each patient. The number of days to the first major bleed event was similarly calculated for those with one or more bleeds during follow-up.
Covariates Baseline characteristics examined included age on the index date, gender, body mass index (BMI) on the index date (taking the most recent value within 30 days before the index date if not available on that date), and smoking status on or last known prior to the index date. Comorbidities were noted from or within 30 days before ! 2014 Informa UK Ltd www.cmrojournal.com
March 2015
the index date; these included potential risk factors for VTE and bleeding. Blood pressure values, platelet counts, and creatinine clearance or estimated glomerular filtration rate were included.
Analysis Time to the first INR level of 2.0 or higher after the first documented VTE was estimated using Kaplan–Meier methods. The number of days spent in each INR category (52.0, 2.0–3.0, and 43.0) was estimated using the Rosendaal linear interpolation method9. For each patient, the percentage of time spent in each INR category out of the total period of exposure to warfarin was calculated to provide estimates of time in, below, and above the therapeutic range. The proportion of patients and incidence rates of recurrent VTE and of bleeding were calculated. A Cox proportional hazards model of time to first recurrent VTE event was constructed, with INR levels treated as a time-varying characteristic. The INR categories, bridging therapy exposure, and each of the baseline characteristics were entered into a Cox model, with the final covariates chosen using backward selection at p50.05. Hazard ratios (HRs) for each of the variables in the models with 95% confidence intervals (CIs) and p values were calculated.
Results A total of 1753 patients qualified for the study cohort (Table 1), 867 with an index DVT and 886 with an index PE. The median age of patients with an index DVT was 58 years, and with an index PE was 60 years. Approximately 51% of the patients were female. Among the patients with recorded BMI, most were overweight or obese. Overall, 37% of patients were noted as never smokers, and 18% as current smokers. As would be expected, DVT most often appeared as an outpatient diagnosis, whereas PE typically involved hospitalization for treatment. Cancer patients comprised 28% of the cohort, similarly distributed across DVT and PE; 17% had a history of respiratory failure, with a higher proportion among the patients with an index PE (19%) than DVT (15%). Hypertension, anemia, and ischemic heart disease appeared in 20% or more of the cohort. Platelet count results were unavailable for 31% of DVT patients but only 15% of patients with an index PE; 9% and 11%, respectively, had low platelet levels recorded. Blood pressure was high in 17% of DVT and 9% of PE patients. Estimates of creatinine clearance were unavailable for similar proportions of DVT and PE patients as were platelet counts, and showed at least a minor degree of renal impairment (i.e. creatinine clearance 590 mL/min/ Risk of recurrent VTE Nordstrom et al.
441
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 1. Baseline characteristics of study cohort. Type of index VTEa Patient characteristics
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Age, years 18–39 40–49 50–64 65–79 80þ Gender Male Female BMI, kg/m2 Below 18.5 (underweight) 18.5–24.9 (normal) 25.0–29.9 (overweight) 30.0 and above (obese) Unknown Smoking history Current smoker Former smoker Never smoked Unknown Treatment setting of index VTE Inpatient Outpatient Comorbidities Respiratory failure Malignancy Pregnancy within past year Varicose veins Obesityb Thrombophilia Laparoscopic cholecystectomy History of major bleeding Hypertensionb Ischemic heart disease Diabetes mellitus Anemia Liver disease Renal insufficiency Cerebrovascular disease Heart failure Platelets counts Low (5150 10/L) Normal (150–450 10/L) High (4150 10/L) Unknown Blood pressure Low (SBP 590 or DBP 560) Normal (90 SBP5140 and 60 DBP 90) High (SBP4140 or DBP490) Unknown Creatinine clearance, mL/min/1.73 m2 Normal (490) Mild renal impairment (60–90) Moderate renal impairment (30–59) Severe renal impairment (15–29) End-stage renal disease (515) Unknown
Overall n ¼ 1753
DVT n ¼ 867
PE n ¼ 886
277 (15.8) 280 (16.0) 509 (29.0) 489 (27.9) 198 (11.3)
140 (16.1) 147 (17.0) 257 (29.6) 220 (25.4) 103 (11.9)
137 (15.5) 133 (15.0) 252 (28.4) 269 (30.4) 95 (10.7)
865 (49.3) 888 (50.7)
426 (49.1) 441 (50.9)
439 (49.5) 447 (50.5)
28 (1.6) 255 (14.5) 460 (26.2) 634 (36.2) 376 (21.4)
17 (2.0) 122 (14.1) 234 (27.0) 306 (35.3) 188 (21.7)
11 (1.2) 133 (15.0) 226 (25.5) 328 (37.0) 188 (21.2)
315 (18.0) 524 (29.9) 648 (37.0) 266 (15.2)
163 (18.8) 225 (26.0) 339 (39.1) 140 (16.1)
152 (17.2) 299 (33.7) 309 (34.9) 126 (14.2)
915 (52.2) 838 (47.8)
243 (28.0) 624 (72.0)
672 (75.8) 214 (24.2)
302 (17.2) 492 (28.1) 20 (1.1) 73 (4.2) 283 (16.1) 22 (1.3) 0 (0.0) 325 (18.5) 788 (45.0) 351 (20.0) 335 (19.1) 414 (23.6) 169 (9.6) 202 (11.5) 230 (13.1) 165 (9.4)
129 (14.9) 242 (27.9) 8 (0.9) 50 (5.8) 143 (16.5) 10 (1.2) 0 (0.0) 166 (19.1) 395 (45.6) 156 (18.0) 166 (19.1) 211 (24.3) 96 (11.1) 123 (14.2) 119 (13.7) 81 (9.3)
173 (19.5) 250 (28.2) 12 (1.4) 23 (2.6) 140 (15.8) 12 (1.4) 0 (0.0) 159 (17.9) 393 (44.4) 195 (22.0) 169 (19.1) 203 (22.9) 73 (8.2) 79 (8.9) 111 (12.5) 84 (9.5)
177 (10.1) 1105 (63.0) 65 (3.7) 406 (23.2)
81 (9.3) 480 (55.4) 34 (3.9) 272 (31.4)
96 (10.8) 625 (70.5) 31 (3.5) 134 (15.1)
66 (3.8) 1381 (78.8) 228 (13.0) 78 (4.4)
41 (4.7) 637 (73.5) 151 (17.4) 38 (4.4)
25 (2.8) 744 (84.0) 77 (8.7) 40 (4.5)
612 (34.9) 478 (27.3) 198 (11.3) 33 (1.9) 18 (1.0) 414 (23.6)
277 (31.9) 197 (22.7) 95 (11.0) 19 (2.2) 11 (1.3) 268 (30.9)
335 (37.8) 281 (31.7) 103 (11.6) 14 (1.6) 7 (0.8) 146 (16.5)
BMI, body mass index; DBP, diastolic blood pressure; DVT, deep vein thrombosis; PE, pulmonary embolism; SBP, systolic blood pressure; VTE, venous thromboembolism. a Cell entries are n (%) unless otherwise specified. b Based on ICD-9 coding.
442
Risk of recurrent VTE Nordstrom et al.
www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 2. Anticoagulants and hospital length of stay following index VTE.
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Type of index VTEa
Other anticoagulants administered Unfractionated heparins Low molecular weight heparins Pentasaccharides (factor Xa inhibitors) Direct thrombin inhibitors Adequacy of bridging therapy Recommended bridging therapy: 5 days þ INR 2.0 Suboptimal bridging therapy, level 1: INR 2.0 but 55 days Suboptimal bridging therapy, level 2: 5 days but INR 52.0 Partial bridging therapy: 1–4 days þ INR 52.0 No bridging therapy Duration of bridging therapy, days Mean (SD) Median (range) Duration of warfarin use, days Mean (SD) Median (range) Number of days from index date to warfarin start Mean (SD) Median (range) Length of stay of index hospitalization, days Number of patients with an index hospitalization Mean (SD) Median (range)
Overall n ¼ 1753
DVT n ¼ 867
PE n ¼ 886
433 (24.7) 867 (49.5) 65 (3.7) 2 (0.1)
118 (13.6) 472 (54.4) 35 (4.0) 1 (0.1)
315 (35.6) 395 (44.6) 30 (3.4) 1 (0.1)
252 (14.4) 50 (2.9) 151 (8.6) 726 (41.4) 574 (32.7)
184 (21.2) 22 (2.5) 105 (12.1) 274 (31.6) 282 (32.5)
68 (7.7) 28 (3.2) 46 (5.2) 452 (51.0) 292 (33.0)
10.4 (62.2) 1 (0–1437)
16.4 (83.9) 2 (0–1437)
4.6 (26.6) 1 (0–546)
554.8 (628.1) 308 (2–2979)
550.6 (644.6) 286 (2–2928)
558.9 (612.0) 337 (2–2979)
0.4 (0.7) 0 (0–3)
0.3 (0.7) 0 (0–3)
0.5 (0.8) 0 (0–3)
915 (52.2) 7.2 (6.8) 5 (2–56)
243 (28.0) 8.9 (8.7) 6 (2–54)
672 (75.8) 6.6 (5.8) 5 (2–56)
DVT, deep vein thrombosis; INR, international normalized ratio; PE, pulmonary embolism; SD, standard deviation; VTE, venous thromboembolism. a Cell entries are n (%) unless otherwise specified.
1.73 m2) in slightly more than half of the patients with measurements recorded. More than 67% of the patients received some form of bridging therapy, most frequently a UH or LMWH (Table 2). A higher proportion of patients with an index DVT than PE received recommended bridging therapy, consisting of anticoagulation treatment that lasted for at least 5 days and until the INR reached the therapeutic range (21.2% and 7.7%, respectively). Mean duration of bridging therapy was 16.4 days for DVT patients and 4.6 days for PE patients. Warfarin exposure lasted for a median of 286 days for DVT and 337 days for PE, ranging from 2 days to more than 8 years in both groups. For patients with a hospitalization associated with the index VTE, the median length of stay was 6 days for DVT and 5 days for PE. Patients had a median of 12 INR values available during follow-up (Table 3). The mean percentage of time spent in the therapeutic range was 42.3% overall. The mean percentage of time below range was 45.9%, and above range was 11.9%. Using Kaplan–Meier methods that included patients who never achieved an in-range INR value, the median time to the first in-range level occurred at a median of 9 days after the index date. INR levels were, as expected, lower during the first 4 days of follow-up (with a mean of 13.4% of days in the therapeutic range) than on Day 5 and beyond. Starting on Day 5, the mean percentage ! 2014 Informa UK Ltd www.cmrojournal.com
of time in the 2–3 range was 43.2%, with 41.4% of time below range and 15.3% above range. Across the full cohort, 134 patients (7.6%) had records indicating a recurrent VTE event, for an incidence of 3.2 events per 100 person-years (Table 4). These outcome events were evenly split across patients with an index DVT and index PE. Most patients with a recurrence had the same event type for the recurrent event (i.e. DVT was typically followed by a new DVT, and PE by a new PE). A diagnosis of major bleeding was recorded in 411 (23.4%) patients, at a rate of 11.8 events per 100 person-years, with similar incidence between the DVT and PE patients. The highest proportion of VTE recurrences occurred during the first 3 months after the index date, although events continued to occur for years after the index event (Figure 1). The adjusted time-varying Cox model of recurrent VTE in the full cohort revealed that below-range INR levels were associated with an increased risk of recurrent VTE (HR: 3.37; 95% CI: 2.16–5.27; Table 5). The time when INR was not measured, indicating that it was unknown whether warfarin therapy continued, was predictive of a decreased risk of VTE (HR: 0.51; 95% CI: 0.28–0.92). A borderline association with increased risk of the outcome was found for patients who received suboptimal bridging therapy (HR: 1.79; 95% CI: 0.99–3.25). Low platelet Risk of recurrent VTE Nordstrom et al.
443
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 3. INR values during post-VTE warfarin exposure. Type of index VTEa
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Overall n ¼ 1753 Number of INR values during warfarin exposure Mean (SD) 19.9 (22.0) Median (range) 12 (2–167) Percentage of INR days in category: all follow-up timeb Time below range (52.0) Mean (SD) 45.9 (35.3) Median (range) 37 (0–100) Time in therapeutic range (2.0–3.0) Mean (SD) 42.3 (31.9) Median (range) 42 (0–100) Time above range (43.0) Mean (SD) 11.9 (19.8) Median (range) 3 (0–100) Percentage of INR days in category: Day 5 and later Time below range (52.0) Mean (SD) 41.4 (28.8) Median (range) 36 (0–100) Time in therapeutic range (2.0–3.0) Mean (SD) 43.2 (26.3) Median (range) 44 (0–100) Time above range (43.0) Mean (SD) 15.3 (18.5) Median (range) 11 (0–100) Ever have an INR in therapeutic range No 304 (17.3) Yes 1449 (82.7) Time to first in-range INR (days) Median (Kaplan–Meier) 9 Ever have an INR below therapeutic range 52.0 No 49 (2.8) Yes 1704 (97.2) Ever have an INR above therapeutic range 43.0 No 703 (40.1) Yes 1050 (59.9)
DVT n ¼ 867
PE n ¼ 886
18.4 (19.8) 12 (2–125)
21.4 (23.8) 14 (2–167)
45.0 (35.0) 37 (0–100)
46.7 (35.6) 38 (0–100)
42.3 (31.7) 41 (0–100)
42.2 (32.2) 42 (0–100)
12.7 (20.6) 3 (0–100)
11.1 (18.9) 2 (0–100)
41.3 (29.1) 35 (0–100)
41.6 (28.4) 37 (0–100)
42.9 (26.7) 44 (0–100)
43.6 (25.8) 44 (0–100)
15.8 (18.8) 11 (0–100)
14.8 (18.2) 11 (0–100)
136 (15.7) 731 (84.3)
168 (19.0) 718 (81.0)
9
9
27 (3.1) 840 (96.9)
22 (2.5) 864 (97.5)
327 (37.7) 540 (62.3)
376 (42.4) 510 (57.6)
DVT, deep vein thrombosis; INR, international normalized ratio; PE, pulmonary embolism; SD, standard deviation; VTE, venous thromboembolism. a Cell entries are n (%) unless otherwise specified. b Rosendaal method used to impute days in INR category.
counts also predicted recurrent VTE (HR: 2.13; 95% CI: 1.24–3.67). Cox models of recurrent DVT among the patients with an index DVT showed a similar pattern to the overall VTE results, with below-range INR (HR: 2.82; 95% CI: 1.42–5.63) predictive of an increased risk. Hypertension was associated with a decreased risk of recurrent DVT (HR: 0.36; 95% CI: 0.19–0.71). A PE outcome was observed in only 13 patients with an index DVT, and a DVT outcome in five patients with an index PE. Hence, Cox models of these outcomes are not presented. The Cox models of PE following an index PE again found a significant association between below-range INR levels and recurrent VTE (HR: 3.80; 95% CI: 1.97–7.34), although above-range levels also predicted an increased risk (HR: 2.77; 95% CI: 1.13–6.81). Also associated with an increased risk of recurrent PE were anemia (HR: 1.84; 444
Risk of recurrent VTE Nordstrom et al.
95% CI: 1.03–3.29) and low platelet counts (HR: 2.68; 95% CI: 1.32–5.44).
Discussion In this real-world cohort of patients treated with warfarin for a VTE, recurrent VTE events were observed in approximately 8% of patients during follow-up, which is in line with estimates from previous studies8. Consistent with previous work that has associated subtherapeutic INR levels with an increased risk of recurrent VTE, the present study shows that patients with subtherapeutic INR values across this cohort had more than a three-fold increase in VTE risk. Patients who lacked the ACCP-recommended bridging therapy also showed some indication of an increase in risk of recurrent VTE. www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 4. Incidence of study outcomes. Type of Index VTE Overall n ¼ 1753
DVT n ¼ 867
PE n ¼ 886
67 (3.8) 4264 1.6 0.1–6.5
54 (6.2) 2158 2.5 0.4–8.0
13 (1.5) 2106 0.6 0.0–4.9
67 (3.8) 4302 1.6 0.1–6.5
5 (0.6) 2265 0.2 0.0–4.1
62 (7) 2036 3 0.6–8.8
134 (7.6) 4163 3.2 0.7–9.1
59 (6.8) 2147 2.7 0.5–8.4
75 (8.5) 2016 3.7 1.0–9.8
411 (23.4) 3476 11.8 6.1–20.7
212 (24.5) 1806 11.7 6.0–20.6
199 (22.5) 1671 11.9 6.1–20.8
First VTE after index event DVT Number (%) of patients with event Person-years in denominator Incidence rate per 100 person-years 95% confidence interval PE Number (%) of patients with event Person-years in denominator Incidence rate per 100 person-years 95% confidence interval Any VTE Number (%) of patients with event Person-years in denominator Incidence rate per 100 person-years 95% confidence interval Bleeds Major bleeding Number (%) of patients with event Person-years in denominator Incidence rate per 100 person-years 95% confidence interval
DVT, deep vein thrombosis; PE, pulmonary embolism; VTE, venous thromboembolism.
35
% of Outcome Events
30 25 20 15 10 5
1+ 75
90 39 1– 45 0 45 1– 51 0 51 1– 57 0 57 1– 63 0 63 1– 69 0 69 1– 75 0
30
–3
33 1
70
–3
27 1
10
–2
–2
15 1
21 1
50 –1
91
–9
0
0
31
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Outcomes
Days to Event
Figure 1. Days from index venous thromboembolism to recurrent venous thromboembolism outcome.
We found a reduced risk of VTE among patients whose INR levels were not measured. Given that INR levels are typically tracked only through the duration of warfarin use, unmeasured levels should indicate the time after warfarin was discontinued, although some patients may have continued warfarin therapy outside of the health system reflected in this database. If warfarin treatment continued for the recommended duration, then unmeasured INR levels would indicate a time period when the patient is generally considered to be at lower risk of a recurrence, and hence no longer in need of anticoagulation therapy. ! 2014 Informa UK Ltd www.cmrojournal.com
While patients achieved a therapeutic INR at a median of 9 days after therapy initiation, the mean percentage of days in therapeutic range was below 50%, and considerable time was spent at INRs below the range. Supratherapeutic INR levels, which can place patients at risk of bleeding, were less common than subtherapeutic levels but were still observed during more than 10% of follow-up time. These results may be affected by the inclusion of shortterm warfarin users, who may have been either treated with a different anticoagulant within this health system or discharged to primary care outside of the health system, where warfarin therapy and INR monitoring could have continued but was not reflected in the study data. Warfarin is not the only anticoagulation therapy available for the treatment of VTE. The treatments examined in the present study as bridging therapies, which are generally used only during the start of warfarin treatment to boost anticoagulation until warfarin takes effect, were in some cases used for months or years. For such patients, the heparin or other non-warfarin anticoagulant may have been intended as their primary anticoagulation therapy, making INR levels less relevant, as INR is used to measure the anticoagulation effect of warfarin only. Bridging therapy was found to be below the level recommended by ACCP guidelines in many of these warfarintreated patients. Heightened bleeding risk may have been a concern in many of these patients, potentially preventing the use of recommended levels of anticoagulation therapy in an effort to avoid major bleeds. Patients whose index Risk of recurrent VTE Nordstrom et al.
445
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 5. Final multivariable time-varying proportional hazards model of recurrent VTE, full cohort. Full cohort, n ¼ 1746a
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Variableb INR level (time-varying) Below range: 52.0 In range: 2.0–3.0 Above range: 43.0 Not measured: unknown if warfarin continued Adequacy of bridging therapy Recommended bridging therapy: 5 days þ INR 42.0 Suboptimal bridging therapy Platelet counts Low Normal High Unknown
n (%)
HR (95% CI)
p Value
1693 (97.0)
50.0001
1025 (58.7) 1270 (72.7)
3.37 (2.16–5.27) Reference 1.31 (0.62–2.79) 0.51 (0.28–0.92)
1494 (85.6)
Reference 1.79 (0.99–3.25)
177 (10.1) 65 (3.7) 400 (22.9)
2.13 (1.24–3.67) Reference 0.20 (0.03–1.47) 1.06 (0.70–1.59)
0.48 0.026 0.056 0.0065 0.11 0.78
HR, hazard ratio; CI, confidence interval; INR, international normalized ratio; VTE, venous thromboembolism. a Patients with 52 INR levels available prior to the date of the outcome event are excluded from the analysis. b Covariates were chosen using backward selection at p50.05.
VTE was a DVT received more bridging therapy than those with a PE, which is an unexpected finding given the greater severity of PE relative to DVT. Unusual patterns of bridging therapy may relate to the very short-term use of warfarin treatment seen for some patients in the database, which could indicate either that a patient was not maintained on warfarin therapy and received a ‘bridging’ therapy for maintenance rather than bridging, or that a patient was discharged to primary care outside of the health system covered by this database, and hence follow-up data on post-discharge anticoagulation therapy were missing. In addition, the hospitalized DVT patients had a slightly longer length of stay than the hospitalized PE patients. In this health system, uncomplicated DVT is typically treated on an outpatient basis. The hospitalized DVT events may therefore have been even more severe than the typical PE, leading to the longer length of stay and potentially to the greater duration of bridging therapy. There are limitations inherent in the data used for this study. The use of EHR data allows for the investigation of treatment patterns and outcomes as they occur in realworld medical practice, but it has some unavoidable limitations. The data reflect practice patterns and outcomes in a single state and a relatively homogeneous healthcare system; results may not be generalizable to the broader US population. Medical care and drug prescriptions that occur outside of the EMR system are not recorded in the database, which could lead to underestimations of VTE and bleed events and of warfarin use. Detailed information on variable drug dosing and patient compliance is not available; this prevents a precise analysis of doses of warfarin and other anticoagulants. Some patients may have been prescribed warfarin or other medications but never filled the prescriptions or took the drugs. The timing of events that occur during a hospitalization with respect to 446
Risk of recurrent VTE Nordstrom et al.
INR levels is uncertain in some cases, as the dates of diagnoses that arise while in the hospital are not always tracked in the database. Some patients received only short-term warfarin therapy lasting a week or less and may have received long-term treatment with heparin or another non-warfarin anticoagulant. While inclusion of such patients allows for the examination of a real-world cohort of patients started on warfarin following a VTE, short-term warfarin users may have treatment patterns and outcomes that differ from those on longer-term warfarin therapy. Our study confirms the difficulty in maintaining a high quality of anticoagulation control in patients treated with warfarin for VTE and demonstrates an association between subtherapeutic INR levels and an increased risk of recurrent VTE. Novel treatment approaches that offer a more sustained anticoagulation level over time may allow benefits in terms of reducing the number of events and simplifying therapy.
Transparency Declaration of funding Janssen Scientific Affairs LLC, Raritan, NJ, USA provided the funding for this study. Declaration of financial/other relationships M.A.E. has disclosed that he has served on speakers bureaus for Janssen Scientific Affairs LLC and Roche Pharmaceuticals. E.A.N. has disclosed that she was a consultant to, received grant support from, and served on a speaker forum for Janssen Scientific Affairs LLC. J.R.S. and B.K.B. have disclosed that they are employees of Janssen Scientific Affairs LLC and shareholders of Johnson & Johnson. B.L.N. and B.R.M. have disclosed that they are employees of Evidera, a consultancy that received
www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 31, Number 3
funding from Janssen Scientific Affairs LLC for the study on which this manuscript is based. CMRO peer reviewers on this manuscript have no relevant financial or other relationships to disclose.
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Acknowledgments The authors acknowledge Chameleon Communications International for editorial review, with funding from Janssen Scientific Affairs LLC.
References 1. Beckman MG, Hooper WC, Critchley SE, Ortel TL. Venous thromboembolism: a public health concern. Am J Prev Med 2010;38:S495-501 2. Kahn SR, Lim W, Dunn AS, et al.; American College of Chest Physicians. Prevention of VTE in nonsurgical patients: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest 2012;141(2 Suppl):e195226S 3. Guyatt GH, Akl EA, Crowther M, et al.; American College of Chest Physicians Antithrombotic Therapy and Prevention of Thrombosis
! 2014 Informa UK Ltd www.cmrojournal.com
4.
5.
6.
7. 8.
9.
March 2015
Panel. Executive summary: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest 2012;141(2 Suppl):7-47S Ansell J, Hirsh J, Hylek E, et al.; American College of Chest Physicians. Pharmacology and management of the vitamin K antagonists: American College of Chest Physicians evidence-based clinical practice guidelines (8th edition). Chest 2008;133(6 Suppl):160-98S Kearon C, Kahn SR, Agnelli G, et al.; American College of Chest Physicians. Antithrombotic therapy for venous thromboembolic disease: American College of Chest Physicians evidence-based clinical practice guidelines (8th edition). Chest 2008;133(6 Suppl):454-545S Heit JA, Mohr DN, Silverstein MD, et al. Predictors of recurrence after deep vein thrombosis and pulmonary embolism: a population-based cohort study. Arch Intern Med 2000;160:761-8 Prandoni P, Lensing AW, Cogo A, et al. The long-term clinical course of acute deep venous thrombosis. Ann Intern Med 1996;125:1-7 Iorio A, Kearon C, Filippucci E, et al. Risk of recurrence after a first episode of symptomatic venous thromboembolism provoked by a transient risk factor: a systematic review. Arch Intern Med 2010;170:1710-16 Rosendaal FR, Cannegieter SC, van der Meer FJ, Brie¨t E. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost 1993;69:236-9
Risk of recurrent VTE Nordstrom et al.
447
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
0300-7995 doi:10.1185/03007995.2014.998814
Vol. 31, No. 3, 2015, 439–447
Article RT-0356.R1/998814 All rights reserved: reproduction in whole or part not permitted
Original article Risk of recurrent venous thromboembolism among deep vein thrombosis and pulmonary embolism patients treated with warfarin
Beth L. Nordstrom
Abstract
Evidera, Lexington, MA, USA
Michael A. Evans Geisinger Medical Center, Danville, PA, USA
Brian R. Murphy
Objective: Guidelines for warfarin treatment of venous thromboembolism (VTE) recommend targeting an international normalized ratio (INR) level of 2–3. This study examines the association between INR levels and VTE recurrence among warfarin-treated patients.
Evidera, Lexington, MA, USA
Edith A. Nutescu University of Illinois at Chicago, Chicago, IL, USA
Jeff R. Schein Brahim K. Bookhart Janssen Scientific Affairs LLC, Raritan, NJ, USA Address for correspondence: Beth L. Nordstrom PhD MPH, Evidera, 430 Bedford Street, Suite 300, Lexington, MA 02420, USA. Tel.: +1 781-960-0250; Fax: +1 781-761-0147; [email protected] Keywords: Deep vein thrombosis – International normalized ratio – Pulmonary embolism – Venous thromboembolism – Warfarin Accepted: 9 December 2014; published online: 24 December 2014 Citation: Curr Med Res Opin 2015; 31:439–47
Methods: A retrospective cohort study in the MedMining electronic health record database included adults treated with warfarin for VTE in 2004–2011. INR levels during warfarin use were categorized as below therapeutic range (52), in range (2–3), or above range (43), with time in each category estimated using the Rosendaal method. Recurrent VTE was noted from 30 days after the initial VTE to end of follow-up, which ranged up to 8 years. The incidence of recurrent VTE was calculated, and association with time-varying INR levels estimated using Cox models. Results: Of 1753 qualifying patients, 867 had deep vein thrombosis, and 886 had pulmonary embolism. Mean age was 58 years, and 50.7% were female. Across all follow-up time, VTE recurrences were observed in 134 (7.6%) patients, at a rate of 3.2 (95% confidence interval [CI]: 0.7–9.1) events per 100 person-years. The risk of VTE recurrence was greater during time spent with INR52 than with INR in the therapeutic range (hazard ratio [HR]: 3.37; 95% CI: 2.16–5.27). Low platelet counts also predicted greater risk of VTE recurrence (HR: 2.13; 95% CI: 1.24–3.67). Limitations: Exposure to warfarin and other anticoagulants was estimated based on prescription data and may be inaccurate. The study data include care within a single health system; thus, care received outside of the health system may be missing, and results may not be generalizable to the broader US population. Conclusions: Approximately 8% of patients experienced a recurrent VTE during follow-up. Subtherapeutic INR levels were associated with a more than three-fold increased risk of VTE recurrence.
Introduction Venous thromboembolism (VTE) affects an estimated 300,000–600,000 individuals in the United States annually1. VTE includes the two related conditions of deep vein thrombosis (DVT) and pulmonary embolism (PE). The wide list of risk factors for VTE include hereditary factors such as deficiencies of ! 2014 Informa UK Ltd www.cmrojournal.com
Risk of recurrent VTE Nordstrom et al.
439
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Current Medical Research & Opinion Volume 31, Number 3
March 2015
antithrombin as well as acquired risk factors such as cancer, surgery, obesity, heart failure, and ongoing hormonal treatment)2. For the treatment and prophylaxis of VTE, the American College of Chest Physicians (ACCP) recommends anticoagulation therapy consisting of a vitamin K antagonist in combination with unfractionated heparin (UH), low molecular weight heparin (LMWH), or fondaparinux, or more recently available combinations involving novel oral anticoagulants3. Warfarin, a common vitamin K antagonist, is prescribed for patients with VTE and is initially paired with heparin, LMWH, or fondaparinux until its full effect is established in the patient. The correct dose of warfarin for each patient is guided by frequent monitoring of the target international normalized ratio (INR). The recommended target INR range for VTE treatment is between 2.0 and 3.0. Because INR outside of this range can result in bleeding or a recurrent thrombotic event, it is imperative that patients’ INR be regularly monitored3–5. A patient is most at risk for a recurrent event during the first 6 months following an initial VTE. During those 6 months, the cumulative proportion of patients experiencing a recurrent event at 2 weeks is 2%, increasing to 6.4% at 3 months and reaching 8–10% at 6 months6,7. A systematic review estimated a rate of VTE recurrence of 3.3% per patient-year during the first 2 years after stopping anticoagulation therapy following a VTE, where the index VTE was provoked by transient risk factors8. VTE not provoked by known risk factors was more likely to be followed by a recurrence; the same review estimated a recurrence rate of 7.4% per patient-year in such cases8. The objective of this study was to evaluate the association between INR during warfarin therapy and the subsequent risk of recurrent VTE among patients treated with warfarin for an initial VTE event.
Patients and methods Data source and study design Data were obtained from the MedMining database of electronic health record (EHR) data for approximately 3.9 million patients from a large integrated health system in Pennsylvania. The health system includes an 870þ multispecialty physician group practice, three hospital campuses, 64 primary and specialty clinic sites, and a clinical reference laboratory. Available data included inpatient, outpatient, and prescription drug data, along with laboratory results and clinical characteristics, such as height, weight, and tobacco status. Data were extracted from 1 January 2004 through 31 March 2012 for all patients in the MedMining database 440
Risk of recurrent VTE Nordstrom et al.
with a diagnosis of VTE (International Classification of Diseases, 9th Revision, Clinical Modification [ICD9-CM] codes 452.xx, 453.xx, or 415.1x). Each patient was assigned an index date corresponding to the date of the first VTE event. Uncomplicated VTE is typically treated on an outpatient basis in this health system; for these outpatient events, the first date of outpatient diagnosis was designated as the index date. For inpatient events, to ensure that the date of the event could be ascertained, a primary or secondary diagnosis of VTE was required. To ensure that the date of the VTE event could be determined accurately, patients were required to have either a date of entry of the VTE diagnosis into the system or a record of the VTE as both an admission and a discharge diagnosis. For the latter subset, the date of the event was assumed to be the first day of the hospitalization. Eligible patients were required to have a first diagnosis of VTE between 31 January 2004 and 31 December 2011 (3 months prior to end of the study data), followed by warfarin initiation within the next 3 days, to be at least 18 years old on the index date, and to have at least two INR results available during the follow-up period. Patients were excluded from the analysis if their date of VTE event was unclear, if their earliest recorded VTE predated the study start date, if they received warfarin within 30 days prior to the index date, or if their first visit in the EHR system occurred less than 30 days prior to the index date.
Exposure to anticoagulation therapy Warfarin exposure was defined based on the start and stop dates noted in the EHR and was considered to be continuous as long as there was no more than a 30 day gap in exposure. Restarting warfarin after longer than a 30 day gap was considered a new episode of warfarin use. Longterm maintenance on warfarin therapy was not a requirement for the study cohort; patients were included who had as few as 2 days of warfarin treatment noted in the EHR during follow-up. Use of bridging therapies, including heparins, pentasaccharides, and direct thrombin inhibitors, on the warfarin start date was noted, including the type and duration of each therapy. For these drugs, any gap of 1 day or longer ended the episode of continuous use. Bridging therapy was classified with respect to whether it met the criteria for ACCP-recommended therapy overlap (i.e. therapy lasting for at least 5 days and until the INR is at least 2.0 and stable). For many patients, the optimal bridging therapy approach may differ from the standard ACCP recommendations, but the reasons for a reduction in bridging may not be identifiable through the database, and hence were not considered in this classification. www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 31, Number 3
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
INR measurement All available INR values were captured for each qualifying patient from the index date to end of follow-up, where the end of follow-up for each patient was defined as the date of last activity in the database or date of death for deceased patients. An INR measurement taken on the first day of warfarin exposure was treated as a baseline measurement rather than reflecting the INR during warfarin exposure. Each INR value was categorized as falling within the ACCP-recommended therapeutic range (2.0–3.0), below range (52.0), or above range (43.0).
Outcomes VTE diagnosis codes were identified from the index date to end of follow-up in the database for each patient. To allow for a complete assessment of recurrent events that may have happened either during the course of warfarin therapy or later, the outcome period included any available time after the end of warfarin treatment. The VTE outcomes were examined both combined and separately as DVT (ICD-9-CM codes 452.xx, 453.xx) and PE (ICD-9-CM code 415.1). ICD-9-CM codes indicating major bleed events, such as hemorrhage of various types (e.g. intracranial, gastrointestinal), menstrual disorders involving excessive bleeding, and blood in stool, were included. A complete list of the bleed outcome codes is available from the authors upon request. To be considered a recurrent VTE event rather than follow-up care for the index event, the VTE outcome was required to occur more than 30 days after the index date and to be associated with a visit reason indicating an initial anticoagulation visit. For patients with an index VTE that occurred during a hospitalization, outcome events were required to occur after discharge from the hospitalization for the initial VTE event to be included in the analysis, as it is not possible to determine whether an additional entry of a VTE diagnosis during a hospitalization represents a new event or the initial event. When a VTE recurrence outcome was found, the type (i.e. DVT or PE) and days from the index date to the date of the event were determined for the first VTE event that occurred during followup for each patient. The number of days to the first major bleed event was similarly calculated for those with one or more bleeds during follow-up.
Covariates Baseline characteristics examined included age on the index date, gender, body mass index (BMI) on the index date (taking the most recent value within 30 days before the index date if not available on that date), and smoking status on or last known prior to the index date. Comorbidities were noted from or within 30 days before ! 2014 Informa UK Ltd www.cmrojournal.com
March 2015
the index date; these included potential risk factors for VTE and bleeding. Blood pressure values, platelet counts, and creatinine clearance or estimated glomerular filtration rate were included.
Analysis Time to the first INR level of 2.0 or higher after the first documented VTE was estimated using Kaplan–Meier methods. The number of days spent in each INR category (52.0, 2.0–3.0, and 43.0) was estimated using the Rosendaal linear interpolation method9. For each patient, the percentage of time spent in each INR category out of the total period of exposure to warfarin was calculated to provide estimates of time in, below, and above the therapeutic range. The proportion of patients and incidence rates of recurrent VTE and of bleeding were calculated. A Cox proportional hazards model of time to first recurrent VTE event was constructed, with INR levels treated as a time-varying characteristic. The INR categories, bridging therapy exposure, and each of the baseline characteristics were entered into a Cox model, with the final covariates chosen using backward selection at p50.05. Hazard ratios (HRs) for each of the variables in the models with 95% confidence intervals (CIs) and p values were calculated.
Results A total of 1753 patients qualified for the study cohort (Table 1), 867 with an index DVT and 886 with an index PE. The median age of patients with an index DVT was 58 years, and with an index PE was 60 years. Approximately 51% of the patients were female. Among the patients with recorded BMI, most were overweight or obese. Overall, 37% of patients were noted as never smokers, and 18% as current smokers. As would be expected, DVT most often appeared as an outpatient diagnosis, whereas PE typically involved hospitalization for treatment. Cancer patients comprised 28% of the cohort, similarly distributed across DVT and PE; 17% had a history of respiratory failure, with a higher proportion among the patients with an index PE (19%) than DVT (15%). Hypertension, anemia, and ischemic heart disease appeared in 20% or more of the cohort. Platelet count results were unavailable for 31% of DVT patients but only 15% of patients with an index PE; 9% and 11%, respectively, had low platelet levels recorded. Blood pressure was high in 17% of DVT and 9% of PE patients. Estimates of creatinine clearance were unavailable for similar proportions of DVT and PE patients as were platelet counts, and showed at least a minor degree of renal impairment (i.e. creatinine clearance 590 mL/min/ Risk of recurrent VTE Nordstrom et al.
441
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 1. Baseline characteristics of study cohort. Type of index VTEa Patient characteristics
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Age, years 18–39 40–49 50–64 65–79 80þ Gender Male Female BMI, kg/m2 Below 18.5 (underweight) 18.5–24.9 (normal) 25.0–29.9 (overweight) 30.0 and above (obese) Unknown Smoking history Current smoker Former smoker Never smoked Unknown Treatment setting of index VTE Inpatient Outpatient Comorbidities Respiratory failure Malignancy Pregnancy within past year Varicose veins Obesityb Thrombophilia Laparoscopic cholecystectomy History of major bleeding Hypertensionb Ischemic heart disease Diabetes mellitus Anemia Liver disease Renal insufficiency Cerebrovascular disease Heart failure Platelets counts Low (5150 10/L) Normal (150–450 10/L) High (4150 10/L) Unknown Blood pressure Low (SBP 590 or DBP 560) Normal (90 SBP5140 and 60 DBP 90) High (SBP4140 or DBP490) Unknown Creatinine clearance, mL/min/1.73 m2 Normal (490) Mild renal impairment (60–90) Moderate renal impairment (30–59) Severe renal impairment (15–29) End-stage renal disease (515) Unknown
Overall n ¼ 1753
DVT n ¼ 867
PE n ¼ 886
277 (15.8) 280 (16.0) 509 (29.0) 489 (27.9) 198 (11.3)
140 (16.1) 147 (17.0) 257 (29.6) 220 (25.4) 103 (11.9)
137 (15.5) 133 (15.0) 252 (28.4) 269 (30.4) 95 (10.7)
865 (49.3) 888 (50.7)
426 (49.1) 441 (50.9)
439 (49.5) 447 (50.5)
28 (1.6) 255 (14.5) 460 (26.2) 634 (36.2) 376 (21.4)
17 (2.0) 122 (14.1) 234 (27.0) 306 (35.3) 188 (21.7)
11 (1.2) 133 (15.0) 226 (25.5) 328 (37.0) 188 (21.2)
315 (18.0) 524 (29.9) 648 (37.0) 266 (15.2)
163 (18.8) 225 (26.0) 339 (39.1) 140 (16.1)
152 (17.2) 299 (33.7) 309 (34.9) 126 (14.2)
915 (52.2) 838 (47.8)
243 (28.0) 624 (72.0)
672 (75.8) 214 (24.2)
302 (17.2) 492 (28.1) 20 (1.1) 73 (4.2) 283 (16.1) 22 (1.3) 0 (0.0) 325 (18.5) 788 (45.0) 351 (20.0) 335 (19.1) 414 (23.6) 169 (9.6) 202 (11.5) 230 (13.1) 165 (9.4)
129 (14.9) 242 (27.9) 8 (0.9) 50 (5.8) 143 (16.5) 10 (1.2) 0 (0.0) 166 (19.1) 395 (45.6) 156 (18.0) 166 (19.1) 211 (24.3) 96 (11.1) 123 (14.2) 119 (13.7) 81 (9.3)
173 (19.5) 250 (28.2) 12 (1.4) 23 (2.6) 140 (15.8) 12 (1.4) 0 (0.0) 159 (17.9) 393 (44.4) 195 (22.0) 169 (19.1) 203 (22.9) 73 (8.2) 79 (8.9) 111 (12.5) 84 (9.5)
177 (10.1) 1105 (63.0) 65 (3.7) 406 (23.2)
81 (9.3) 480 (55.4) 34 (3.9) 272 (31.4)
96 (10.8) 625 (70.5) 31 (3.5) 134 (15.1)
66 (3.8) 1381 (78.8) 228 (13.0) 78 (4.4)
41 (4.7) 637 (73.5) 151 (17.4) 38 (4.4)
25 (2.8) 744 (84.0) 77 (8.7) 40 (4.5)
612 (34.9) 478 (27.3) 198 (11.3) 33 (1.9) 18 (1.0) 414 (23.6)
277 (31.9) 197 (22.7) 95 (11.0) 19 (2.2) 11 (1.3) 268 (30.9)
335 (37.8) 281 (31.7) 103 (11.6) 14 (1.6) 7 (0.8) 146 (16.5)
BMI, body mass index; DBP, diastolic blood pressure; DVT, deep vein thrombosis; PE, pulmonary embolism; SBP, systolic blood pressure; VTE, venous thromboembolism. a Cell entries are n (%) unless otherwise specified. b Based on ICD-9 coding.
442
Risk of recurrent VTE Nordstrom et al.
www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 2. Anticoagulants and hospital length of stay following index VTE.
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Type of index VTEa
Other anticoagulants administered Unfractionated heparins Low molecular weight heparins Pentasaccharides (factor Xa inhibitors) Direct thrombin inhibitors Adequacy of bridging therapy Recommended bridging therapy: 5 days þ INR 2.0 Suboptimal bridging therapy, level 1: INR 2.0 but 55 days Suboptimal bridging therapy, level 2: 5 days but INR 52.0 Partial bridging therapy: 1–4 days þ INR 52.0 No bridging therapy Duration of bridging therapy, days Mean (SD) Median (range) Duration of warfarin use, days Mean (SD) Median (range) Number of days from index date to warfarin start Mean (SD) Median (range) Length of stay of index hospitalization, days Number of patients with an index hospitalization Mean (SD) Median (range)
Overall n ¼ 1753
DVT n ¼ 867
PE n ¼ 886
433 (24.7) 867 (49.5) 65 (3.7) 2 (0.1)
118 (13.6) 472 (54.4) 35 (4.0) 1 (0.1)
315 (35.6) 395 (44.6) 30 (3.4) 1 (0.1)
252 (14.4) 50 (2.9) 151 (8.6) 726 (41.4) 574 (32.7)
184 (21.2) 22 (2.5) 105 (12.1) 274 (31.6) 282 (32.5)
68 (7.7) 28 (3.2) 46 (5.2) 452 (51.0) 292 (33.0)
10.4 (62.2) 1 (0–1437)
16.4 (83.9) 2 (0–1437)
4.6 (26.6) 1 (0–546)
554.8 (628.1) 308 (2–2979)
550.6 (644.6) 286 (2–2928)
558.9 (612.0) 337 (2–2979)
0.4 (0.7) 0 (0–3)
0.3 (0.7) 0 (0–3)
0.5 (0.8) 0 (0–3)
915 (52.2) 7.2 (6.8) 5 (2–56)
243 (28.0) 8.9 (8.7) 6 (2–54)
672 (75.8) 6.6 (5.8) 5 (2–56)
DVT, deep vein thrombosis; INR, international normalized ratio; PE, pulmonary embolism; SD, standard deviation; VTE, venous thromboembolism. a Cell entries are n (%) unless otherwise specified.
1.73 m2) in slightly more than half of the patients with measurements recorded. More than 67% of the patients received some form of bridging therapy, most frequently a UH or LMWH (Table 2). A higher proportion of patients with an index DVT than PE received recommended bridging therapy, consisting of anticoagulation treatment that lasted for at least 5 days and until the INR reached the therapeutic range (21.2% and 7.7%, respectively). Mean duration of bridging therapy was 16.4 days for DVT patients and 4.6 days for PE patients. Warfarin exposure lasted for a median of 286 days for DVT and 337 days for PE, ranging from 2 days to more than 8 years in both groups. For patients with a hospitalization associated with the index VTE, the median length of stay was 6 days for DVT and 5 days for PE. Patients had a median of 12 INR values available during follow-up (Table 3). The mean percentage of time spent in the therapeutic range was 42.3% overall. The mean percentage of time below range was 45.9%, and above range was 11.9%. Using Kaplan–Meier methods that included patients who never achieved an in-range INR value, the median time to the first in-range level occurred at a median of 9 days after the index date. INR levels were, as expected, lower during the first 4 days of follow-up (with a mean of 13.4% of days in the therapeutic range) than on Day 5 and beyond. Starting on Day 5, the mean percentage ! 2014 Informa UK Ltd www.cmrojournal.com
of time in the 2–3 range was 43.2%, with 41.4% of time below range and 15.3% above range. Across the full cohort, 134 patients (7.6%) had records indicating a recurrent VTE event, for an incidence of 3.2 events per 100 person-years (Table 4). These outcome events were evenly split across patients with an index DVT and index PE. Most patients with a recurrence had the same event type for the recurrent event (i.e. DVT was typically followed by a new DVT, and PE by a new PE). A diagnosis of major bleeding was recorded in 411 (23.4%) patients, at a rate of 11.8 events per 100 person-years, with similar incidence between the DVT and PE patients. The highest proportion of VTE recurrences occurred during the first 3 months after the index date, although events continued to occur for years after the index event (Figure 1). The adjusted time-varying Cox model of recurrent VTE in the full cohort revealed that below-range INR levels were associated with an increased risk of recurrent VTE (HR: 3.37; 95% CI: 2.16–5.27; Table 5). The time when INR was not measured, indicating that it was unknown whether warfarin therapy continued, was predictive of a decreased risk of VTE (HR: 0.51; 95% CI: 0.28–0.92). A borderline association with increased risk of the outcome was found for patients who received suboptimal bridging therapy (HR: 1.79; 95% CI: 0.99–3.25). Low platelet Risk of recurrent VTE Nordstrom et al.
443
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 3. INR values during post-VTE warfarin exposure. Type of index VTEa
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Overall n ¼ 1753 Number of INR values during warfarin exposure Mean (SD) 19.9 (22.0) Median (range) 12 (2–167) Percentage of INR days in category: all follow-up timeb Time below range (52.0) Mean (SD) 45.9 (35.3) Median (range) 37 (0–100) Time in therapeutic range (2.0–3.0) Mean (SD) 42.3 (31.9) Median (range) 42 (0–100) Time above range (43.0) Mean (SD) 11.9 (19.8) Median (range) 3 (0–100) Percentage of INR days in category: Day 5 and later Time below range (52.0) Mean (SD) 41.4 (28.8) Median (range) 36 (0–100) Time in therapeutic range (2.0–3.0) Mean (SD) 43.2 (26.3) Median (range) 44 (0–100) Time above range (43.0) Mean (SD) 15.3 (18.5) Median (range) 11 (0–100) Ever have an INR in therapeutic range No 304 (17.3) Yes 1449 (82.7) Time to first in-range INR (days) Median (Kaplan–Meier) 9 Ever have an INR below therapeutic range 52.0 No 49 (2.8) Yes 1704 (97.2) Ever have an INR above therapeutic range 43.0 No 703 (40.1) Yes 1050 (59.9)
DVT n ¼ 867
PE n ¼ 886
18.4 (19.8) 12 (2–125)
21.4 (23.8) 14 (2–167)
45.0 (35.0) 37 (0–100)
46.7 (35.6) 38 (0–100)
42.3 (31.7) 41 (0–100)
42.2 (32.2) 42 (0–100)
12.7 (20.6) 3 (0–100)
11.1 (18.9) 2 (0–100)
41.3 (29.1) 35 (0–100)
41.6 (28.4) 37 (0–100)
42.9 (26.7) 44 (0–100)
43.6 (25.8) 44 (0–100)
15.8 (18.8) 11 (0–100)
14.8 (18.2) 11 (0–100)
136 (15.7) 731 (84.3)
168 (19.0) 718 (81.0)
9
9
27 (3.1) 840 (96.9)
22 (2.5) 864 (97.5)
327 (37.7) 540 (62.3)
376 (42.4) 510 (57.6)
DVT, deep vein thrombosis; INR, international normalized ratio; PE, pulmonary embolism; SD, standard deviation; VTE, venous thromboembolism. a Cell entries are n (%) unless otherwise specified. b Rosendaal method used to impute days in INR category.
counts also predicted recurrent VTE (HR: 2.13; 95% CI: 1.24–3.67). Cox models of recurrent DVT among the patients with an index DVT showed a similar pattern to the overall VTE results, with below-range INR (HR: 2.82; 95% CI: 1.42–5.63) predictive of an increased risk. Hypertension was associated with a decreased risk of recurrent DVT (HR: 0.36; 95% CI: 0.19–0.71). A PE outcome was observed in only 13 patients with an index DVT, and a DVT outcome in five patients with an index PE. Hence, Cox models of these outcomes are not presented. The Cox models of PE following an index PE again found a significant association between below-range INR levels and recurrent VTE (HR: 3.80; 95% CI: 1.97–7.34), although above-range levels also predicted an increased risk (HR: 2.77; 95% CI: 1.13–6.81). Also associated with an increased risk of recurrent PE were anemia (HR: 1.84; 444
Risk of recurrent VTE Nordstrom et al.
95% CI: 1.03–3.29) and low platelet counts (HR: 2.68; 95% CI: 1.32–5.44).
Discussion In this real-world cohort of patients treated with warfarin for a VTE, recurrent VTE events were observed in approximately 8% of patients during follow-up, which is in line with estimates from previous studies8. Consistent with previous work that has associated subtherapeutic INR levels with an increased risk of recurrent VTE, the present study shows that patients with subtherapeutic INR values across this cohort had more than a three-fold increase in VTE risk. Patients who lacked the ACCP-recommended bridging therapy also showed some indication of an increase in risk of recurrent VTE. www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 4. Incidence of study outcomes. Type of Index VTE Overall n ¼ 1753
DVT n ¼ 867
PE n ¼ 886
67 (3.8) 4264 1.6 0.1–6.5
54 (6.2) 2158 2.5 0.4–8.0
13 (1.5) 2106 0.6 0.0–4.9
67 (3.8) 4302 1.6 0.1–6.5
5 (0.6) 2265 0.2 0.0–4.1
62 (7) 2036 3 0.6–8.8
134 (7.6) 4163 3.2 0.7–9.1
59 (6.8) 2147 2.7 0.5–8.4
75 (8.5) 2016 3.7 1.0–9.8
411 (23.4) 3476 11.8 6.1–20.7
212 (24.5) 1806 11.7 6.0–20.6
199 (22.5) 1671 11.9 6.1–20.8
First VTE after index event DVT Number (%) of patients with event Person-years in denominator Incidence rate per 100 person-years 95% confidence interval PE Number (%) of patients with event Person-years in denominator Incidence rate per 100 person-years 95% confidence interval Any VTE Number (%) of patients with event Person-years in denominator Incidence rate per 100 person-years 95% confidence interval Bleeds Major bleeding Number (%) of patients with event Person-years in denominator Incidence rate per 100 person-years 95% confidence interval
DVT, deep vein thrombosis; PE, pulmonary embolism; VTE, venous thromboembolism.
35
% of Outcome Events
30 25 20 15 10 5
1+ 75
90 39 1– 45 0 45 1– 51 0 51 1– 57 0 57 1– 63 0 63 1– 69 0 69 1– 75 0
30
–3
33 1
70
–3
27 1
10
–2
–2
15 1
21 1
50 –1
91
–9
0
0
31
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Outcomes
Days to Event
Figure 1. Days from index venous thromboembolism to recurrent venous thromboembolism outcome.
We found a reduced risk of VTE among patients whose INR levels were not measured. Given that INR levels are typically tracked only through the duration of warfarin use, unmeasured levels should indicate the time after warfarin was discontinued, although some patients may have continued warfarin therapy outside of the health system reflected in this database. If warfarin treatment continued for the recommended duration, then unmeasured INR levels would indicate a time period when the patient is generally considered to be at lower risk of a recurrence, and hence no longer in need of anticoagulation therapy. ! 2014 Informa UK Ltd www.cmrojournal.com
While patients achieved a therapeutic INR at a median of 9 days after therapy initiation, the mean percentage of days in therapeutic range was below 50%, and considerable time was spent at INRs below the range. Supratherapeutic INR levels, which can place patients at risk of bleeding, were less common than subtherapeutic levels but were still observed during more than 10% of follow-up time. These results may be affected by the inclusion of shortterm warfarin users, who may have been either treated with a different anticoagulant within this health system or discharged to primary care outside of the health system, where warfarin therapy and INR monitoring could have continued but was not reflected in the study data. Warfarin is not the only anticoagulation therapy available for the treatment of VTE. The treatments examined in the present study as bridging therapies, which are generally used only during the start of warfarin treatment to boost anticoagulation until warfarin takes effect, were in some cases used for months or years. For such patients, the heparin or other non-warfarin anticoagulant may have been intended as their primary anticoagulation therapy, making INR levels less relevant, as INR is used to measure the anticoagulation effect of warfarin only. Bridging therapy was found to be below the level recommended by ACCP guidelines in many of these warfarintreated patients. Heightened bleeding risk may have been a concern in many of these patients, potentially preventing the use of recommended levels of anticoagulation therapy in an effort to avoid major bleeds. Patients whose index Risk of recurrent VTE Nordstrom et al.
445
Current Medical Research & Opinion Volume 31, Number 3
March 2015
Table 5. Final multivariable time-varying proportional hazards model of recurrent VTE, full cohort. Full cohort, n ¼ 1746a
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Variableb INR level (time-varying) Below range: 52.0 In range: 2.0–3.0 Above range: 43.0 Not measured: unknown if warfarin continued Adequacy of bridging therapy Recommended bridging therapy: 5 days þ INR 42.0 Suboptimal bridging therapy Platelet counts Low Normal High Unknown
n (%)
HR (95% CI)
p Value
1693 (97.0)
50.0001
1025 (58.7) 1270 (72.7)
3.37 (2.16–5.27) Reference 1.31 (0.62–2.79) 0.51 (0.28–0.92)
1494 (85.6)
Reference 1.79 (0.99–3.25)
177 (10.1) 65 (3.7) 400 (22.9)
2.13 (1.24–3.67) Reference 0.20 (0.03–1.47) 1.06 (0.70–1.59)
0.48 0.026 0.056 0.0065 0.11 0.78
HR, hazard ratio; CI, confidence interval; INR, international normalized ratio; VTE, venous thromboembolism. a Patients with 52 INR levels available prior to the date of the outcome event are excluded from the analysis. b Covariates were chosen using backward selection at p50.05.
VTE was a DVT received more bridging therapy than those with a PE, which is an unexpected finding given the greater severity of PE relative to DVT. Unusual patterns of bridging therapy may relate to the very short-term use of warfarin treatment seen for some patients in the database, which could indicate either that a patient was not maintained on warfarin therapy and received a ‘bridging’ therapy for maintenance rather than bridging, or that a patient was discharged to primary care outside of the health system covered by this database, and hence follow-up data on post-discharge anticoagulation therapy were missing. In addition, the hospitalized DVT patients had a slightly longer length of stay than the hospitalized PE patients. In this health system, uncomplicated DVT is typically treated on an outpatient basis. The hospitalized DVT events may therefore have been even more severe than the typical PE, leading to the longer length of stay and potentially to the greater duration of bridging therapy. There are limitations inherent in the data used for this study. The use of EHR data allows for the investigation of treatment patterns and outcomes as they occur in realworld medical practice, but it has some unavoidable limitations. The data reflect practice patterns and outcomes in a single state and a relatively homogeneous healthcare system; results may not be generalizable to the broader US population. Medical care and drug prescriptions that occur outside of the EMR system are not recorded in the database, which could lead to underestimations of VTE and bleed events and of warfarin use. Detailed information on variable drug dosing and patient compliance is not available; this prevents a precise analysis of doses of warfarin and other anticoagulants. Some patients may have been prescribed warfarin or other medications but never filled the prescriptions or took the drugs. The timing of events that occur during a hospitalization with respect to 446
Risk of recurrent VTE Nordstrom et al.
INR levels is uncertain in some cases, as the dates of diagnoses that arise while in the hospital are not always tracked in the database. Some patients received only short-term warfarin therapy lasting a week or less and may have received long-term treatment with heparin or another non-warfarin anticoagulant. While inclusion of such patients allows for the examination of a real-world cohort of patients started on warfarin following a VTE, short-term warfarin users may have treatment patterns and outcomes that differ from those on longer-term warfarin therapy. Our study confirms the difficulty in maintaining a high quality of anticoagulation control in patients treated with warfarin for VTE and demonstrates an association between subtherapeutic INR levels and an increased risk of recurrent VTE. Novel treatment approaches that offer a more sustained anticoagulation level over time may allow benefits in terms of reducing the number of events and simplifying therapy.
Transparency Declaration of funding Janssen Scientific Affairs LLC, Raritan, NJ, USA provided the funding for this study. Declaration of financial/other relationships M.A.E. has disclosed that he has served on speakers bureaus for Janssen Scientific Affairs LLC and Roche Pharmaceuticals. E.A.N. has disclosed that she was a consultant to, received grant support from, and served on a speaker forum for Janssen Scientific Affairs LLC. J.R.S. and B.K.B. have disclosed that they are employees of Janssen Scientific Affairs LLC and shareholders of Johnson & Johnson. B.L.N. and B.R.M. have disclosed that they are employees of Evidera, a consultancy that received
www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 31, Number 3
funding from Janssen Scientific Affairs LLC for the study on which this manuscript is based. CMRO peer reviewers on this manuscript have no relevant financial or other relationships to disclose.
Curr Med Res Opin Downloaded from informahealthcare.com by Biblioteka Uniwersytetu Warszawskiego on 03/23/15 For personal use only.
Acknowledgments The authors acknowledge Chameleon Communications International for editorial review, with funding from Janssen Scientific Affairs LLC.
References 1. Beckman MG, Hooper WC, Critchley SE, Ortel TL. Venous thromboembolism: a public health concern. Am J Prev Med 2010;38:S495-501 2. Kahn SR, Lim W, Dunn AS, et al.; American College of Chest Physicians. Prevention of VTE in nonsurgical patients: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest 2012;141(2 Suppl):e195226S 3. Guyatt GH, Akl EA, Crowther M, et al.; American College of Chest Physicians Antithrombotic Therapy and Prevention of Thrombosis
! 2014 Informa UK Ltd www.cmrojournal.com
4.
5.
6.
7. 8.
9.
March 2015
Panel. Executive summary: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest 2012;141(2 Suppl):7-47S Ansell J, Hirsh J, Hylek E, et al.; American College of Chest Physicians. Pharmacology and management of the vitamin K antagonists: American College of Chest Physicians evidence-based clinical practice guidelines (8th edition). Chest 2008;133(6 Suppl):160-98S Kearon C, Kahn SR, Agnelli G, et al.; American College of Chest Physicians. Antithrombotic therapy for venous thromboembolic disease: American College of Chest Physicians evidence-based clinical practice guidelines (8th edition). Chest 2008;133(6 Suppl):454-545S Heit JA, Mohr DN, Silverstein MD, et al. Predictors of recurrence after deep vein thrombosis and pulmonary embolism: a population-based cohort study. Arch Intern Med 2000;160:761-8 Prandoni P, Lensing AW, Cogo A, et al. The long-term clinical course of acute deep venous thrombosis. Ann Intern Med 1996;125:1-7 Iorio A, Kearon C, Filippucci E, et al. Risk of recurrence after a first episode of symptomatic venous thromboembolism provoked by a transient risk factor: a systematic review. Arch Intern Med 2010;170:1710-16 Rosendaal FR, Cannegieter SC, van der Meer FJ, Brie¨t E. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost 1993;69:236-9
Risk of recurrent VTE Nordstrom et al.
447
