Association Between Use of Antithrombotic Medication and Hematuria-Related Complications

Research

JAMA | Original Investigation Association Between Use of Antithrombotic Medication and Hematuria-Related Complications

Christopher J. D. Wallis, MD, PhD; Tristan Juvet, MD; Yuna Lee, MD, MEd; Rano Matta, MD, MSc; Sender Herschorn, MD; Ronald Kodama, MD; Girish S. Kulkarni, MD, PhD; Raj Satkunasivam, MD, MPH; William Geerts, MD; Anne McLeod, MD; Steven A. Narod, MD; Robert K. Nam, MD, MSc

Related article page 1250

IMPORTANCE Antithrombotic medications are among the most commonly prescribed Supplemental content medications.

OBJECTIVE To characterize rates of hematuria-related complications among patients taking antithrombotic medications.

DESIGN, SETTING, AND PARTICIPANTS Population-based, retrospective cohort study including all citizens in Ontario, Canada, aged 66 years and older between 2002 and 2014. The final follow-up date was December 31, 2014.

EXPOSURES Receipt of an oral anticoagulant or antiplatelet medication.

MAIN OUTCOMES AND MEASURES Hematuria-related complications, defined as emergency department visit, hospitalization, or a urologic procedure to investigate or manage gross hematuria.

RESULTS Among 2 518 064 patients, 808 897 (mean [SD] age, 72.1 [6.8] years; 428 531 [53%] women) received at least 1 prescription for an antithrombotic agent over the study period. Over a median follow-up of 7.3 years, the rates of hematuria-related complications were 123.95 events per 1000 person-years among patients actively exposed to antithrombotic agents vs 80.17 events per 1000 person-years among patients not exposed to these drugs (difference, 43.8; 95% CI, 43.0-44.6; P < .001, and incidence rate ratio [IRR], 1.44; 95% CI, 1.42-1.46). The rates of complications among exposed vs unexposed patients (80.17 events/1000 person-years) were 105.78 for urologic procedures (difference, 33.5; 95% CI, 32.8-34.3; P < .001, and IRR, 1.37; 95% CI, 1.36-1.39), 11.12 for hospitalizations (difference, 5.7; 95% CI, 5.5-5.9; P < .001, and IRR, 2.03; 95% CI, 2.00-2.06), and 7.05 for emergency department visits (difference, 4.5; 95% CI, 4.3-4.7; P < .001, and IRR, 2.80; 95% CI, 2.74-2.86). Compared with patients who were unexposed to thrombotic agents, the rates of hematuria-related complications were 191.61 events per 1000 person-years (difference, 117.3; 95% CI, 112.8-121.8) for those exposed to both an anticoagulant and antiplatelet agent (IRR, 10.48; 95% CI, 8.16-13.45), 140.92 (difference, 57.7; 95% CI, 56.9-58.4) for those exposed to anticoagulants (IRR, 1.55; 95% CI, 1.52-1.59), and 110.72 (difference, 26.5; 95% CI, 25.9-27.0) for those exposed to antiplatelet agents (IRR, 1.31; 95% CI, 1.29-1.33). Patients exposed to antithrombotic agents, compared with patients not exposed to these drugs, were more likely to be diagnosed as having bladder cancer within 6 months (0.70% vs 0.38%; odds ratio, 1.85; 95% CI, 1.79-1.92).

CONCLUSIONS AND RELEVANCE Among older adults in Ontario, Canada, use of antithrombotic medications, compared with nonuse of these medications, was significantly associated with higher rates of hematuria-related complications (including emergency department visits, hospitalizations, and urologic procedures to manage gross hematuria). Author Affiliations: Author affiliations are listed at the end of this article. Corresponding Author: Robert K. Nam, MD, MSc, Division of Urology, Sunnybrook Health Sciences Centre, Room MG-406, 2075 Bayview Ave, Toronto, ON M4N 3M5, Canada JAMA. 2017;318(13):1260-1271. doi:10.1001/jama.2017.13890 ([email protected]).

1260 (Reprinted) jama.com

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Antithrombotic Medication Use and Hematuria-Related Complications Original Investigation Research

ntithrombotic agents are among the most commonly prescribed medications for older adults in Key Points North America.1 Oral anticoagulants are indicated for A Question Is there an association between the use of oral primary and secondary prevention of stroke and systemic em- antithrombotic agents and hematuria-related complications? bolism, as well as treatment of venous thromboembolism.2 Findings In this cohort study that included 2 518 064 older adults Antiplatelet agents are indicated for primary and secondary in Ontario, Canada, use of antithrombotic medications, compared prevention of cardiovascular disease.3 Despite proven ben- with nonuse of these medications, was significantly associated efits, antithrombotic agents are among the medications with hematuria-related complications (including emergency most commonly associated with adverse events4 and have department visits, hospitalizations, and urologic procedures). contributed to nearly half of all adverse drug events.5 Fur- Meaning Use of antithrombotic medications was associated with ther, the rates of these adverse events are increasing.4 To date, a significant increase in rates of hematuria-related complications. published randomized clinical trials and observational studies of antithrombotic agents have focused on intracranial hemorrhage, gastrointestinal bleeding, and all-cause bleed- excluded individuals who died and those who emigrated prior ing as adverse events.6,7 to the index date. We further excluded patients diagnosed as To our knowledge, a complication that has not been having a cancer (other than nonmelanomatous skin cancer) examined as the primary outcome in patients treated with prior to the index date and those with prior endoscopic uro- antithrombotic agents is hematuria. While hematuria repre- logic procedures as these are likely to significantly affect a pa- sents a less life-threatening adverse event than intracranial or tient’s likelihood of hematuria-related complications. We also gastrointestinal bleeding, it is common and involves diagnos- excluded patients older than the age of 105 years. tic evaluation including abdominal imaging and invasive From linked databases, we collected demographic testing8,9 and management. The prevalence, severity, and information including patient age at the time of each pre- risk factors for hematuria associated with the use of anti- scription, geographic location (local health integration thrombotic agents are largely unknown. To better character- networks12), sex, geographically derived socioeconomic sta- ize this association, this analysis examined rates of gross tus, rurality, and general comorbidity (Johns Hopkins aggre- hematuria-related complications including hospitalization, gate disease group13). The Johns Hopkins aggregate disease emergency department visits, and urologic interventions group has better discrimination than the Charlson score in over a 13-year period among patients who received antico- comorbidity assessment.14 agulant or antiplatelet therapy from a population-based cohort of adults aged 66 years or older in Ontario, Canada. Exposure The primary exposure was use of any oral antithrombotic agent, including anticoagulant and antiplatelet medications for which Methods the first prescription occurred during the study interval (eTable 1intheSupplement). We operationalized antithrombotic ex- We conducted a population-based, retrospective cohort study posure in an intermittent, time-varying fashion (examples of of patients aged 66 years or older in Ontario, Canada, between operationalization of this exposure are in the eFigure in the April 2002 and December 2014 using data from the Institute Supplement). Age 66 years was selected for cohort inclusion of Clinical Evaluative Sciences (ICES). In Ontario, medical care to allow for a 1-year look-back to ensure that patients were not is reimbursed by a single, government-operated health insur- exposed to antithrombotic agents prior to study entry. ance system (Ontario Health Insurance Plan). The cost of pre- On the date of filling their first prescription during the study scription medications is covered for all patients starting at age interval, patients were considered “exposed” and remained 65 years through the Ontario Drug Benefit. exposed until 14 days following the prescription end date This study was designed and conducted according to (washout period). Fifteen days following the prescription end the Strengthening the Reporting of Observational Studies date, patients were considered to be “unexposed.” When the in Epidemiology guidelines10 and Reporting of Studies Con- washout period coincided with prescription renewal, pa- ducted Using Observational Routinely-Collected Health Data tients had continuous, ongoing exposure. When patients dis- Statement.11 The Sunnybrook Health Sciences Centre Re- continued antithrombotic therapy and then restarted after dis- search Ethics Board approved this study. Individual informed continuation, a new exposure period commenced. Similarly, consent was waived owing to the anonymous, aggregated na- when patients switched medications, exposure and outcome ture of the data. time was allocated to each medication during the prescription period plus the 14-day washout period. Study Patients We identified all residents of Ontario born before 1936, who Outcomes would be aged 66 years or older during the study interval We measured counts of total hematuria-related complica- (2002-2014), based on date of birth using unique identifiers tions, which was the sum of the counts of 3 specific end points (ICES key number). The index date was defined as each per- including the occurrence of emergency department visits for son’s 66th birthday. To include only those patients actively re- gross hematuria, hospital admissions with a primary diagno- ceiving medical care in Ontario during the study interval, we sis of gross hematuria, or urologic procedures to manage gross

jama.com (Reprinted) JAMA October 3, 2017 Volume 318, Number 13 1261

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Research Original Investigation Antithrombotic Medication Use and Hematuria-Related Complications

hematuria. No other adverse events were included in this total. the multivariable models such that an effect estimate could be Specific diagnostic and procedural codes are provided in eTable derived for each medication. As concurrent antiplatelet and 2intheSupplement. anticoagulant exposure may have a synergistic effect, the association of combination therapy was also assessed. Data Sources We linked the following validated data sets using patients’ Subgroup Analyses unique ICES key numbers: the Ontario Drug Benefit data- We conducted several preplanned subgroup analyses to un- base, which provides information on all outpatient prescrip- derstand modifying risks, given the known association be- tion pharmaceuticals15; the Canadian Institute for Health tween hematuria and benign prostatic hypertrophy (BPH)22 and Information Discharge Abstract Database, which contains medical kidney disease.23 As a surrogate for BPH, we exam- records for all hospitalizations16; the Canadian Institute ined patients’ prescriptions for BPH-related medications. for Health Information National Ambulatory Care Reporting We assessed the association of receipt of BPH-related medi- System, which contains records for emergency department cations in the year prior to antithrombotic exposure on rates visits; the Ontario Health Insurance Plan database, which of hematuria-related complications among men (eTable 4 in tracks claims paid for physician billings, laboratories, and the Supplement). Then, among all patients, we assessed for the out-of-province providers (physicians, allied health, and association of medical kidney disease by including consulta- hospitals)17; the Ontario Cancer Registry, a population-based tion with a nephrologist. registry estimated to be 95% or more complete18; and the We identified patients diagnosed as having bladder can- Registered Persons database for demographic information19 cer within 6 months after an episode of hematuria using the (validation details in eTable 3 in the Supplement). Ontario Cancer Registry. We calculated the standardized incidence ratio (SIR) of bladder cancer among patients Statistical Analysis receiving antithrombotic agents by identifying the age- and We measured the total number of complications each indi- sex-stratified expected number of bladder cancer cases vidual experienced rather than the first presentation of a based on the Ontario population. We then calculated the SIR complication. Patients contributed data until the date of as the ratio of the observed number of bladder cancer cases death or until the last date of follow-up. Patients with miss- divided by the expected number of cases. Logistic regres- ing data were excluded from the multivariate analysis. We sion analysis was also performed to calculate the odds ratio calculated incidence density rates of each complication for being diagnosed as having cancer between the exposed using the total count of the complication as the numerator and unexposed group, as patients are only diagnosed as and number of person-years exposed as the denominator, having cancer once. stratified by exposure to antithrombotic agents. Multivari- Statistical significance was set at P < .05 based on a 2-tailed able negative binomial regression was used to study the comparison. All analyses were performed using SAS Enter- association between exposure to antithrombotic agents prise Guide 6.1 (SAS Institute Inc). (operationalized in the time-varying manner described above) and complications due to the skewed nature of 20,21 health services data. We expressed this as the incidence Results rate ratio (IRR), the ratio of incidence density rate during antithrombotic agent exposure to the rate during unex- We identified 4 184 141 residents of Ontario born before 1936, posed periods. Each rate ratio was adjusted for the associa- who would be aged 66 years or older during the study inter- tion of patient age, sex, comorbidity, income quintile, val (2002-2014). Following exclusion of individuals who region of residence, and rurality with tests for interaction. died (n = 915 892) and emigrated (n = 317 141) prior to the We tested for interaction between exposure variables using index date, those diagnosed as having a cancer (other than separate interaction terms in the models (ie, exposure nonmelanomatous skin cancer) prior to the index date A × exposure B). Where there was statistical significance, (n = 348 302), those with prior endoscopic urologic proce- we included the interaction term in the final model and dures (n = 84 674), and those older than the age of 105 years expressed the results using the interaction. This only (n = 68), the final cohort comprised 2 518 064 individuals occurred for the interaction between antithrombotic expo- aged 66 years and older. Of these, 607 323 died after the sure and age at prescription. As a result, we expressed the index date, and 208 159 were lost to follow-up: 111 845 had a IRR for antithrombotic exposure in a stratified fashion date of last contact of more than 2 years before the end of the according to the patient’s age at the time of prescription. study period, and 96 314 had loss of continuous health insur- We compared differences in the IRR of hematuria-related ance coverage. complications between medications using pairwise tests for Of these participants, 808 897 filled at least 1 prescrip- heterogeneity between aspirin and other antiplatelet agents tion for an antithrombotic agent, while 1 709 167 were not ex- and between each of the 4 anticoagulants. Many patients re- posed during the study interval. Patients who filled a prescrip- ceived more than 1 agent during the study interval, and some tion for an antithrombotic agent were older (median age, 70 received them concurrently. While assessing the association years vs 66 years; P < .001), more likely to be male (47.0% vs of each medication, concurrent exposures were handled using 41.9%; P < .001), had a lower level of income (20.8% vs 17.9%; each medication as an independent exposure variable within P < .001), and had higher levels of comorbidity (30.6% vs 18.4%;

1262 JAMA October 3, 2017 Volume 318, Number 13 (Reprinted) jama.com

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Antithrombotic Medication Use and Hematuria-Related Complications Original Investigation Research

Table 1. Demographics of Study Cohort, Stratified by Exposure to Antithrombotic Agents (Study Interval Between 2002-2014)

No. (%) Unexposed During Ever Exposed During Characteristic Total Study Intervala Study Intervala P Value Sample size 2 518 064 1 709 167 808 897 Age, y Mean (SD) 70.16 (6.41) 69.23 (6.00) 72.12 (6.79) <.001 Median (IQR) 66 (66-73) 66 (66-70) 70 (66-77) <.001 Age, categorical 66-69 y 1 642 981 (65.2) 1 254 546 (73.4) 388 435 (48.0) 70-74 y 339 136 (13.5) 180 692 (10.6) 158 444 (19.6) 75-79 y 257 889 (10.3) 127 262 (7.4) 130 623 (16.1) <.001 80-84 y 157 432 (6.3) 77 235 (4.5) 80 197 (9.9) ≥85 y 120 626 (4.8) 69 432 (4.1) 51 194 (6.3) Sex Female 1 421 828 (56.5) 993 297 (58.1) 428 531 (53.0) <.001 Male 1 096 236 (43.5) 715 870 (41.9) 380 366 (47.0) Income quintile 1: Lowest 474 591 (18.8) 306 713 (17.9) 167 878 (20.8) 2 516 527 (20.5) 342 915 (20.1) 173 612 (21.5) 3 497 811 (19.8) 336 594 (19.7) 161 217 (19.9) <.001 4 490 999 (19.5) 340 295 (19.9) 150 704 (18.6) 5: Highest 514 187 (20.4) 363 335 (21.3) 150 852 (18.6) Missing 23 949 (1.0) 19 315 (1.1) 4634 (0.6) Comorbidity (ADG category)b 0-2 483 388 (19.2) 386 160 (22.6) 97 228 (12.0) 3-5 764 884 (30.4) 549 985 (32.2) 214 899 (26.6) <.001 6-8 707 808 (28.1) 458 864 (26.8) 248 944 (30.8) ≥9 561 984 (22.3) 314 158 (18.4) 247 826 (30.6) Select relevant medical diagnoses in the 5 y prior to index date Atrial fibrillation 81 963 (3.3) 18 375 (1.1) 63 588 (7.9) <.001 Ischemic stroke 19 101 (0.8) 5536 (0.3) 13 565 (1.7) <.001 TIA 97 545 (3.9) 37 614 (2.2) 59 931 (7.4) <.001 DVT or PE 11 834 (0.5) 4191 (0.2) 7643 (0.9) <.001 Valvular heart disease 30 796 (1.2) 10 822 (0.6) 19 974 (2.5) <.001 Myocardial infarction 98 242 (3.9) 35 205 (2.1) 63 037 (7.8) <.001 Angina 384 786 (15.3) 164 120 (9.6) 220 666 (27.3) <.001 PAD 85 831 (3.4) 37 421 (2.2) 48 410 (6.0) <.001 Contact with urologist prior to age 66 y No 1 901 967 (75.5) 1 355 229 (79.3) 546 738 (67.6) <.001 Yes 616 097 (24.5) 353 938 (20.7) 262 159 (32.4)

(continued)

P < .001) (Table 1). Patients exposed to antithrombotic agents plications, urologic procedures were the most common were significantly more likely to have been diagnosed as hav- (105.78 events/1000 person-years; difference, 33.5; 95% ing a relevant medical condition in the 5 years preceding the CI, 32.8-34.3; P < .001), followed by hospitalizations index date than those who were not exposed (Table 1). (11.12 events/1000 person-years; difference, 5.7; 95% CI, 5.5- Over a median follow-up of 7.3 years, gross hematuria- 5.9; P < .001) and emergency department visits (7.05 related complication rates were higher during active ex- events/1000 person-years; difference, 4.5; 95% CI, 4.3-4.7; posure to any antithrombotic agent (123.95 events/1000 P < .001; Table 2). The crude rate ratio for the develop- person-years) than during unexposed periods (80.17 ment of hematuria-related complications was 1.44 (95% CI, events/1000 person-years; difference, 43.8; 95% CI, 1.42-1.46) for exposure to antithrombotic agents compared 43.0-44.6; P < .001; Table 2). Of the hematuria-related com- with unexposed periods. Despite the fact that urologic

jama.com (Reprinted) JAMA October 3, 2017 Volume 318, Number 13 1263

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Research Original Investigation Antithrombotic Medication Use and Hematuria-Related Complications

Table 1. Demographics of Study Cohort, Stratified by Exposure to Antithrombotic Agents (Study Interval Between 2002-2014) (continued)

No. (%) Unexposed During Ever Exposed During Characteristic Total Study Intervala Study Intervala P Value Geographic region (local health integration network) Erie St Clair 135 385 (5.4) 87 891 (5.1) 47 494 (5.9) South West 199 549 (7.9) 136 629 (8.0) 62 920 (7.8) Waterloo Wellington 122 346 (4.9) 85 880 (5.0) 36 466 (4.5) Hamilton Niagara 298 521 (11.9) 200 685 (11.7) 97 836 (12.1) Haldimand Brant Central West 120 468 (4.8) 83 785 (4.9) 36 683 (4.5) Mississauga Halton 178 018 (7.1) 123 852 (7.2) 54 166 (6.7) Toronto Central 217 274 (8.6) 147 169 (8.6) 70 105 (8.7) Abbreviations: ADG, aggregate Central 296 960 (11.8) 204 867 (12.0) 92 093 (11.4) <.001 disease groups; DVT, deep vein thrombosis; IQR, interquartile Central East 296 284 (11.8) 200 062 (11.7) 96 222 (11.9) range; PAD, peripheral arterial South East 120 322 (4.8) 81 968 (4.8) 38 354 (4.7) disease; PE, pulmonary embolism; Champlain 239 674 (9.5) 164 629 (9.6) 75 045 (9.3) TIA, transient ischemic attack. a Differences between participants North Simcoe Muskoka 92 450 (3.7) 61 119 (3.6) 31 331 (3.9) who were exposed and unexposed North East 134 038 (5.3) 83 881 (4.9) 50 157 (6.2) were compared using the χ2 test for North West 49 366 (2.0) 31 827 (1.9) 17 539 (2.2) categorical data and the Wilcoxon rank sum test for continuous data. Missing 17 409 (0.7) 14 923 (0.9) 2486 (0.3) b Aggregate disease groups are Rural area score-based comorbidity Yes 370 167 (14.7) 245 095 (14.3) 125 072 (15.5) assessments based on resource No 2 130 309 (84.6) 1 449 063 (84.8) 681 246 (84.2) <.001 use of health care (range, 0-34). Higher scores indicate higher levels Missing 17 588 (0.7) 15 009 (0.9) 2579 (0.3) of comorbidity.

procedures were the most common complication, the asso- Patients taking a combination of any antiplatelet agent and ciation between antithrombotic agent use and complica- any anticoagulant experienced significantly increased rates of tions, when compared with unexposed periods, was highest hematuria-related complications, particularly for hospitaliza- for emergency department visits (crude rate ratio, 2.80; tions (range of rate ratios, 2.68-4.16) and emergency depart- 95% CI, 2.74-2.86), followed by hospitalizations (crude rate ment visits (range of rate ratios, 6.03-10.48) (Table 5). ratio, 2.03; 95% CI, 2.00-2.06) and urologic procedures (crude rate ratio, 1.37; 95% CI, 1.36-1.39). These associations Sensitivity Analysis persisted on multivariable analysis with increasing age, While clot evacuation, control of bladder bleeding, and cys- male sex, and increasing comorbidity being significantly toscopy are unlikely to be indicated for patients without a associated with rates of hematuria-related complications history of bladder cancer in the absence of hematuria, ure- (Table 3). The highest rate was for emergency visits among thral catheterization may be indicated owing to urinary adults aged 85 years and older (adjusted rate ratio, 4.74; retention. Thus, we performed a sensitivity analysis exclud- 95% CI, 4.51-4.99; Table 3). ing catheterization from the definition of urologic proce- We then examined each antithrombotic medication indi- dures. While the overall event rate was lower, in regression vidually: 315 639 individuals filled prescriptions for aspirin models, the direction and magnitude of association were (dose ≥82 mg), 275 887 for other antiplatelet agents, 15 102 similar to the primary analysis, with antithrombotic exposure for apixaban, 43 451 for dabigatran, 87 912 for rivaroxaban, associated with a rate ratio of hematuria-related complica- and 320 347 for warfarin. Hematuria-related complications tions ranging from 1.39 to 2.36, depending on patient age were more common during exposure to anticoagulants than (all P < .001; eTable 5 in the Supplement). antiplatelet agents, and patients experienced the lowest rates of complications during exposure to older medications (aspi- Subgroup Analyses rin and warfarin; Table 2). Among anticoagulants, in multi- To examine the association of BPH and medical kidney dis- variable models, exposure to dabigatran (and not warfarin) ease on hematuria, we used surrogate measures including was associated with the lowest rate of complications, while prescriptions of BPH medications and consultations with rivaroxaban had the highest rate for each age group (Table 4). nephrologists. Prescription of a BPH-related medication in Other antiplatelet agents, including clopidogrel, prasugrel, the year prior to antithrombotic prescription was also associ- tricagrelor, ticlopidine, and dipyridamole, were associated ated with an increased rate of hematuria-related complica- with higher rates of hematuria-related complications than tions (range of adjusted rate ratios, 1.67-1.93; all P < .001). acetylsalicylic acid (≥82-mg dosage; Table 4). This association persisted across secondary outcomes and

1264 JAMA October 3, 2017 Volume 318, Number 13 (Reprinted) jama.com

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Antithrombotic Medication Use and Hematuria-Related Complications Original Investigation Research

Table 2. Incidence Density Rates of Hematuria-Related Complications for Each Antithrombotic Agent Exposure (Study Interval Between 2002-2014)

Any Hematuria-Related Complication Emergency Department Visits Hospitalizations Urologic Procedures Sample Exposure Time No. of Incidence No. of Incidence No. of Incidence No. of Incidence Variable Size (Person-Years) Events Density Ratea Events Density Ratea Events Density Ratea Events Density Ratea No antithrombotic 1 709 167 15 864 708 1 271 901 80.17 39 846 2.51 85 941 5.42 1 146 077 72.24 Any antithromboticb 808 897 2 600 520 322 322 123.95 18 322 7.05 28 905 11.12 275 076 105.78 Difference from 43.8 4.5 5.7 33.5 no antithrombotic (43.0-44.6) (4.3-4.7) (5.5-5.9) (32.8-34.3) (95% CI) Aspirin 315 639 795 686 75 033 94.3 3180 4.00 5353 6.73 66 498 83.57 Difference from 8.3 0.9 0.5 8.9 no aspirin (7.7-9.0) (0.7-1.0) (0.3-0.7) (8.3-9.5) (95% CI)c Other antiplatelet 275 887 706 988 91 927 130.03 4621 6.54 7823 11.07 79 477 112.42 Difference from 45.4 3.5 5.0 36.8 no other antiplatelet (44.6-4.2) (3.3-3.7) (4.8-5.3) (36.1-37.6) (95% CI)c Apixiban 15 102 6612 1085 164.09 76 11.50 132 19.96 877 132.64 Difference from 77.8 8.3 13.7 55.7 no apixiban (68.9-86.7) (5.8-10.9) (10.4-17.1) (47.5-63.9) (95% CI)c Dabigatran 43 451 57 675 8319 144.24 626 10.85 932 16.16 6760 117.21 Difference from 58.1 7.7 10.0 40.4 no dabigatran (55.2-61.0) (6.9-8.6) (8.9-11.0) (37.7-43.0) (95% CI)c Rivaroxaban 87 912 40 668 7672 188.65 766 18.84 922 22.67 5984 147.14 Difference from 102.5 15.7 16.5 70.3 no rivaroxaban (98.7-106.3) (14.4-17.0) (15.0-17.9) (66.9-73.8) (95% CI)c Warfarin 320 347 887 691 123 100 138.67 8804 9.92 12 468 14.05 101 816 114.70 Difference from 55.0 7.1 8.2 39.6 no warfarin (54.2-55.7) (6.9-7.3) (7.9-8.4) (39.0-40.3) (95% CI)c a Incidence density rates are expressed as the number of events per 1000 c Each drug has different nonexposure time; difference in incidence density rate person-years. for each drug is not based on “No antithrombotic” row. b Totals in the “any antithrombotic” category may not equal the total of the component exposures owing to concurrent exposure.

individual antithrombotic medication exposures. Similarly, CI, 1.62-1.73) for patients in the exposed group, compared nephrology consultation was significantly associated with an with the unexposed group, and for kidney cancer was 1.50 increased risk of hematuria-related complications (range of (95% CI, 1.4-1.6). Specific frequency of cancers and BPH out- adjusted rate ratios, 1.88-2.26; all P < .001). comes are detailed in eTable 6 in the Supplement and the dis- In addition, 12 108 of 425 350 individuals (2.85%; 95% CI, tribution of duration of time for each antithrombotic medica- 2.80%-2.90%) who presented with a hematuria-related com- tion are detailed in eTable 7 in the Supplement. plication were subsequently diagnosed as having bladder cancer within 6 months. A significantly higher proportion of patients exposed to antithrombotic agents (0.70%) were Discussion diagnosed as having bladder cancer than those who were unexposed to these agents (0.38%; unadjusted odds ratio, In this population-based cohort study among 2 518 064 older 1.85; 95% CI, 1.79-1.92). We calculated the standardized inci- adults in Ontario, Canada, treatment with antithrombotic dence ratio of bladder cancer among patients receiving anti- medications, compared with nonuse of these medications, thrombotic agents, compared with the general Ontario popu- was significantly associated with increased rates of lation, with age and sex adjustment. Those receiving hematuria-related complications (including emergency antithrombotic prescriptions had significantly more bladder department visits, hospitalizations, and urologic procedures cancer diagnoses than expected (SIR, 2.38; 95% CI, 2.32- to manage gross hematuria). While there was variation 2.44). Among women, the SIR was 2.17 (95% CI, 2.06-2.30) between medications, this association was present for all and among men it was 2.33 (95% CI, 2.26-2.40). Standardized medications examined. Readily identifiable factors, includ- incidence ratios of other urologist-managed cancers (prostate ing patient age, male sex, comorbidity, and preexistent uro- cancer: SIR, 0.75; 95% CI, 0.73-0.77, and kidney cancer: SIR, logic disease, were significantly associated with rates of gross 0.64; 95% CI, 0.59-0.68) were not elevated among patients hematuria. Patients taking antithrombotic agents were more prescribed antithrombotic agents. The unadjusted odds ratio likely to be diagnosed as having bladder cancer, both com- for being diagnosed as having prostate cancer was 1.65 (95% pared with unexposed individuals and with the general

jama.com (Reprinted) JAMA October 3, 2017 Volume 318, Number 13 1265

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Research Original Investigation Antithrombotic Medication Use and Hematuria-Related Complications

Table 3. Multivariable Negative Binomial Regression Models Assessing the Association Between Exposure to Antithrombotic Agents (Primary Exposure) and Hematuria-Related Complications

Association of Antithrombotic Medication Exposure, Stratified by Age at Prescription Patients 66-69 y Patients 70-74 y Patients 75-79 y Patients 80-84 y Patients ≥85 y Variable Unexposed Exposed Unexposed Exposed Unexposed Exposed Unexposed Exposed Unexposed Exposed Sample size 1 254 546 388 435 180 692 158 444 127 262 130 623 77 235 80 197 69 432 51 194 Exposure time 9 166 367 563 189 2 966 576 605 332 2 008 765 607 366 1 082 624 473 134 640 373 351 500 (person-years) Any Hematuria-Related Complication No. of 663 547 59 870 274 000 77 477 189 025 83 645 96 960 62 888 48 369 38 442 events Incidence 72.39 106.31 92.36 127.99 94.1 137.72 89.56 132.92 75.53 109.37 density ratea Adjusted 1.18 1.56 1.78 1.82 1.65 rate ratio (1.15-1.20) (1.53-1.60) (1.74-1.81) (1.77-1.86) (1.60-1.69) (95% CI)b P Value <.001 <.001 <.001 <.001 <.001 Emergency Department Visits No. of 18 098 3159 8409 3929 6521 4643 4087 3729 2731 2862 events Incidence 1.97 5.61 2.84 6.49 3.25 7.64 3.78 7.88 4.27 8.14 density ratea Adjusted 2.17 2.75 3.55 3.97 4.74 rate ratio (2.07-2.28) (2.63-2.88) (3.40-3.70) (3.79-4.15) (4.51-4.99) (95% CI)b P Value <.001 <.001 <.001 <.001 <.001 Hospitalizations No. of 45 542 5468 17 178 6636 12 179 7155 6937 5697 4105 3949 events Incidence 4.97 9.71 5.79 10.96 6.06 11.78 6.41 12.04 6.41 11.24 density ratea Adjusted 1.51 1.83 2.09 2.29 2.35 rate ratio (1.46-1.56) (1.78-1.89) (2.03-2.16) (2.22-2.37) (2.26-2.45) (95% CI)b P Value <.001 <.001 <.001 <.001 <.001 Urologic Procedures No. of 599 896 51 238 248 400 66 909 170 316 71 844 85 932 53 455 41 533 31 630 events Incidence 65.45 90.98 83.73 110.53 84.79 118.29 79.37 112.98 64.86 89.99 density ratea Adjusted 1.11 1.48 1.67 1.69 1.47 rate ratio (1.08-1.13) (1.45-1.52) (1.64-1.71) (1.64-1.73) (1.42-1.51) (95% CI)b P Value <.001 <.001 <.001 <.001 <.001 a Incidence density rates are expressed as the number of events per 1000 use (type 3 P < .001). Test for trend using Cochran-Armitage test for age person-years. group, P < .001. Rate ratios comparing antithrombotic exposed and b Models adjusted for association of participant sex, comorbidity, rurality, unexposed periods, with stratification by patient age owing to a significant income quintile, and geographic region of residence. Among all these interaction between these 2 variables. variables, a significant interaction was found between age and antithrombotic

population. As there is no putative mechanistic linkage, these cant burden of clinically significant, and potentially life- data suggest that use of antithrombotic agents was likely threatening, urologic disease.26,27 This analysis identified an unmasking otherwise clinically silent bladder cancers. association of asymptomatic bladder cancer among patients In 1994, a prospective study of 243 patients receiving oral undergoing antithrombotic therapy, with an SIR of 2.4, anticoagulation concluded that there was no association be- although the absolute rate of bladder cancer remains low. tween anticoagulant use and hematuria.8 These data are out- Compared with the control group, in this study, positive as- dated given the introduction of novel anticoagulant and anti- sociations for being diagnosed as having prostate and kidney platelet agents. More recently, observational postmarketing cancer were found between the exposed and unexposed surveillance reports have included other sources of “extracra- groups, but there was no increase in their SIRs. This is be- nial bleeding”24,25 but none have explicitly examined the as- cause the SIR was based on comparison with the general popu- sociation between hematuria and treatment with antithrom- lation, which is more heterogeneous than the control group botic agents. that was used in the analysis. Historically, hematuria, particularly among patients A previous population-based study in Ontario showed taking antithrombotic agents, was associated with a signifi- that randomized clinical trials significantly underestimate

1266 JAMA October 3, 2017 Volume 318, Number 13 (Reprinted) jama.com

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Antithrombotic Medication Use and Hematuria-Related Complications Original Investigation Research

Table 4. Multivariable Negative Binomial Regression Models Assessing the Association Between Exposure to Specific Antithrombotic Agents (Primary Exposure) and Any Hematuria-Related Complication

Association of Antithrombotic Medication Exposure, Stratified by Age at Prescription Patients 66-69 y Patients 70-74 y Patients 75-79 y Patients 80-84 y Patients ≥85 y Variable Unexposed Exposed Unexposed Exposed Unexposed Exposed Unexposed Exposed Unexposed Exposed Aspirin (n = 315 639) Exposure time 10 160 989 160 791 3 348 981 204 022 2 291 823 195 057 1 210 425 140 412 657 324 95 404 (person-years) No. of 801 233 12 811 334 912 20 065 227 890 20 062 108 228 13 842 46 927 8253 events Incidence 78.85 79.68 100.00 98.35 99.44 102.85 89.41 98.58 71.39 86.51 density rate Adjusted 0.90 1.16 1.28 1.26 1.18 rate ratio (0.87-0.94) (1.13-1.20) (1.23-1.32) (1.21-1.32) (1.12-1.25) (95% CI)a P value <.001 <.001 <.001 <.001 <.001 Other Antiplatelet Agents (n = 275 887) Exposure time 10 280 093 135 302 3 387 590 156 773 2 290 944 161 751 1 174 989 135 987 624 623 117 175 (person-years) No. of 799 873 14 457 332 886 20 759 221 924 23 889 103 146 19 327 44 467 13 495 events Incidence 77.81 106.85 98.27 132.41 96.87 147.69 87.79 142.12 71.19 115.17 density ratea Adjusted 1.15 1.53 1.75 1.79 1.59 rate ratio (1.11-1.19) (1.48-1.58) (1.69-1.81) (1.72-1.86) (1.52-1.67) (95% CI)a P value <.001 <.001 <.001 <.001 <.001 Apixiban (n = 15 102) Exposure time 10 779 951 922 3 521 671 1 420 2 368 676 1651 1 187 243 1334 601 075 1284 (person-years) No. of 868 318 112 354 288 172 230 650 319 100 404 255 39 478 227 events Incidence 80.55 121.50 100.60 121.09 97.38 193.20 84.57 191.09 65.68 176.76 density ratea Adjusted 1.15 1.33 1.89 2.05 2.13 rate ratio (0.82-1.62) (1.05-1.70) (1.51-2.36) (1.62-2.59) (1.69-2.67) (95% CI)b P value .42 .02 <.001 <.001 <.001 Dabigatran (n = 43 451) Exposure time 10 740 552 7699 3 509 541 11 044 2 363 691 13 956 1 188 749 13 461 605 020 11 514 (person-years) No. of 862 319 826 352 158 1474 230 149 2293 101 067 2109 40 211 1617 events Incidence 80.29 107.28 100.34 133.47 97.37 164.30 85.02 156.67 66.46 140.43 density ratea Adjusted 1.04 1.44 1.80 1.71 1.62 rate ratio (0.90-1.20) (1.28-1.62) (1.63-1.99) (1.52-1.91) (1.44-1.83) (95% CI)b P value .60 <.001 <.001 <.001 <.001 Rivaroxaban (n = 87 912) Exposure time 10 701 607 6212 3 529 453 8444 2 385 506 9472 1 200 592 8610 607 402 7929 (person-years) No. of 857 232 921 354 118 1629 232595 2008 102 285 1709 40 321 1405 events Incidence 80.10 148.25 100.33 192.92 97.50 211.99 85.20 198.49 66.38 177.19 density ratea Adjusted 1.46 2.18 2.39 2.35 2.34 rate ratio (1.28-1.70) (1.97-2.40) (2.17-2.65) (2.13-2.60) (2.09-2.63) (95% CI)b P value <.001 <.001 <.001 <.001 <.001

(continued)

jama.com (Reprinted) JAMA October 3, 2017 Volume 318, Number 13 1267

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Research Original Investigation Antithrombotic Medication Use and Hematuria-Related Complications

Table 4. Multivariable Negative Binomial Regression Models Assessing the Association Between Exposure to Specific Antithrombotic Agents (Primary Exposure) and Any Hematuria-Related Complication (continued)

rates of hemorrhage associated with warfarin therapy.28 Strengths Compared with warfarin, recent studies have shown lower This sample was larger than recent nationwide cohort stud- rates of intracranial bleeding for each of the direct oral anti- ies assessing bleeding events from antithrombotic treatment coagulants individually and, when combined, major bleeding in Denmark25 and France.37 In addition to the large sample size, is also reduced.29 In contrast, this analysis demonstrated an this study has significant strengths owing to its population- increased rate of hematuria-related complications associated based nature. First, this study was performed in Ontario, with rivaroxaban use and comparable rate between warfarin Canada, a jurisdiction in which all relevant medications and and other newer anticoagulants. A similar result has been health services are available free of cost to seniors and are sys- observed for gastrointestinal bleeding and abnormal uterine tematically tracked in administrative databases. Second, as all bleeding in menstruating women.30,31 There is no clear bio- patients older than the age of 66 years in the largest province logic explanation why there were observed differences in of Canada were identified, these results are generalizable. Un- hematuria complications between different antithrombotic like randomized clinical trials with strict inclusion and exclu- agents. Further study into the exact mechanisms of the sion criteria or institutional reports representing tertiary care causes of hematuria from the urinary tract based on their patterns, these results represent the population spectrum of mechanism of action will be necessary. clinical practice. Third, all hospitalizations and emergency de- Long-term antithrombotic decisions often require com- partment visits occurring anywhere in the province of Ontario plex risk-benefit considerations. Among patients with a were captured. This eliminates recall bias and minimizes se- clear indication for anticoagulation, these medications are lection bias. Thus, the outcome ascertainment is more robust associated with improved survival.2,3 Over time, significant than institutional studies. Fourth, as patients may stop anti- changes in the prescribing of antithrombotic medications thrombotic medications owing to adverse events, they may have occurred and overall use has significantly expanded.32 transition from exposed to unexposed states and back. The use Gross hematuria can be significantly distressing to of a time-varying exposure used herein allows for accurate at- patients33 and may contribute to subsequent poor adher- tribution of exposure for each patient at the time of each he- ence with antithrombotic therapy,34,35 which has been maturia-related event. For this reason, survival analysis could shown to increase the risk of stroke and death.35 Similarly, not be conducted with competing risk analysis with morbid- antithrombotic discontinuation due to bleeding contributes ity and mortality, which would cause the exposure to cease. to increased rates of thrombotic events.36 Thus, the persis- tent risk of thrombotic events and relative infrequency of Limitations hematuria requiring hospitalization suggest that ongoing This study had 5 limitations. First, owing to funding eli- anticoagulant use is often warranted. Research to produce gibility for prescription medications in Ontario, the cohort patient-specific decision aids based on cost-benefit or cost- was restricted to patients aged 66 years and older. Given effectiveness analyses incorporating reductions in throm- the interaction between age and the association of antithrombotic events, as well as hematuria and other adverse events, botic therapies with hematuria-related complications, would be valuable. these results are not directly applicable to younger patients.

1268 JAMA October 3, 2017 Volume 318, Number 13 (Reprinted) jama.com

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Antithrombotic Medication Use and Hematuria-Related Complications Original Investigation Research

Table 5. Multivariable Negative Binomial Regression Models Assessing the Association Between Exposure to Combined Antithrombotic Therapy and Hematuria-Related Complications

Association of Antithrombotic Medication Exposure, Stratified by Age at Prescription Patients 66-69 y Patients 70-74 y Patients 75-79 y Patients 80-84 y Patients ≥85 y Variable Unexposed Exposed Unexposed Exposed Unexposed Exposed Unexposed Exposed Unexposed Exposed Exposure time 7 688 136 4010 1 994 048 6225 1 263 218 7500 630 170 6364 354 710 4867 (person-years) Any Hematuria-Related Complication No. of 530 442 598 174 771 1182 108 154 1679 50 052 1211 22 855 880 events Incidence 69.00 149.14 87.65 189.89 85.62 223.86 79.43 190.30 64.43 180.79 density ratea Adjusted 1.39 1.96 2.36 2.06 2.03 rate ratio (1.18-1.62) (1.75-2.19) (2.15-2.60) (1.83-2.32) (1.76-2.34) (95% CI)b P value <.001 <.001 <.001 <.001 <.001 Emergency Department Visits No. of 14 749 83 5705 137 4060 176 2385 114 1455 112 events Incidence 1.92 20.70 2.86 22.01 3.21 23.47 3.79 17.91 4.10 23.01 density ratea Adjusted 6.03 7.35 8.08 6.94 10.48 rate ratio (4.52-8.05) (5.95-9.09) (6.72-9.72) (5.48-8.79) (8.16-13.45) (95% CI)b P value <.001 <.001 <.001 <.001 <.001 Hospitalizations No. of 37 556 88 11 368 180 7324 239 3794 168 2052 119 events Incidence 4.89 21.95 5.70 28.92 5.80 31.87 6.02 26.40 5.79 24.45 density ratea Adjusted 2.68 3.68 4.16 3.82 3.75 rate ratio (2.08-3.47) (3.04-4.45) (3.51-4.92) (3.17-4.61) (2.97-4.75) (95% CI)b P value <.001 <.001 <.001 <.001 <.001 Urologic Procedures No. of 478 127 427 157 694 863 96 767 1264 43 870 929 19 348 649 events Incidence 62.19 106.50 79.08 138.64 76.60 168.53 69.62 145.99 54.55 133.34 density ratea Adjusted 1.09 1.56 2.00 1.75 1.65 rate ratio (0.93-1.27) (1.39-1.76) (1.82-2.22) (1.55-1.99) (1.41-1.93) (95% CI)b P value .30 <.001 <.001 <.001 <.001 a Incidence density rates are expressed as the number of events per 1000 income quintile, and geographic region of residence. Rate ratios compare person-years. combined antithrombotic exposed and unexposed periods, with stratification b Models adjusted for association of participant sex, comorbidity, rurality, by patient age due to a significant interaction between these 2 variables.

Second, the databases preclude capture of the use of come definition was restricted to hospitalizations and emer- over-the-counter low-dose aspirin and nonsteroidal anti- gency department visits, excluding outpatient office physi- inflammatories. Low-dose aspirin has been associated with cian interactions. Hospitalizations, emergency department lower rates of major bleeding compared with higher-dose visits, and urologic interventions are likely to capture most therapy.38,39 Thus, as high-dose aspirin (≥82 mg) had the significant episodes of gross hematuria and the validity of lowest rate ratio for hematuria-related complications of these diagnoses has been well established in Ontario,16 all agents examined, low-dose aspirin is unlikely to be associ- while this is not true for diagnostic fields associated with out- ated with a clinically significant risk of hematuria. Low- patient consultations. However, the specific diagnoses and molecular-weight heparins were not included in the analysis procedures examined in this study have not been directly owing to differing indications and their typical short dura- validated. For urological procedures, these were used as sur- tions of use. Thus, these results should only be applied to rogates for actual complications without an ability to ascertain oral antithrombotic agents. Third, exposure ascertainment the indication for each intervention. While there are alterna- relied on data on prescriptions filled as a surrogate for medi- tive indications for urethral catheterization (namely, acute uri- cation use without verification of medication consumption. nary retention), few alternatives exist for cystoscopy, clot re- Nonadherence estimates have ranged from 25% to 55%40 and moval/irrigation, or control of bladder bleeding in the absence would bias these results toward the null. Fourth, the out- of a history of bladder cancer. A sensitivity analysis removing

jama.com (Reprinted) JAMA October 3, 2017 Volume 318, Number 13 1269

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Research Original Investigation Antithrombotic Medication Use and Hematuria-Related Complications

urethral catheterization from the outcome definition did not substantively change the findings. Fifth, the available data sets Conclusions lacked information on the doses of antithrombotic medications, cumulative dose exposure, patient weight or body mass Among older adults in Ontario, Canada, use of antithrom- index, international normalized ratio values for patients tak- botic medications, compared with nonuse of these medica- ing warfarin, and alcohol consumption. However, within the tions, was significantly associated with higher rates of hema- study, a proportion of patients served as their own controls for turia-related complications (including emergency department the time they were not taking an antithrombotic medication, visits, hospitalizations, and urologic procedures to manage which would minimize this bias. gross hematuria).

ARTICLE INFORMATION Role of the Funder/Sponsor: The funding studies Conducted using Observational Accepted for Publication: September 5, 2017. organizations had no role in the design and conduct Routinely-collected health Data (RECORD) of the study; collection, management, analysis, statement. PLoS Med. 2015;12(10):e1001885. Author Affiliations: Division of Urology, and interpretation of the data; preparation, review, Sunnybrook Health Sciences Centre, University of 12. Ontario's Local Health Integration Networks. or approval of the manuscript; and decision to http://www.lhins.on.ca/. Accessed March 17, 2017. Toronto, Toronto, Ontario, Canada (Wallis, Juvet, submit the manuscript for publication. Matta, Herschorn, Kodama, Satkunasivam, Nam); 13. Johns Hopkins Bloomberg School of Public Institute for Health Policy, Management & Disclaimer: The opinions, results, and conclusions Health. The Johns Hopkins ACG Case-Mix System Evaluation, University of Toronto, Toronto, Ontario, reported are those of the authors. No endorsement Reference Manual Version 7.0. Baltimore, MD: The Canada (Wallis, Kulkarni, Nam); Department of by the Institute of Clinical Evaluative Sciences or Johns Hopkins University Bloomberg School of Public Medicine, St Michael’s Hospital, University of any of its funders or partners is intended or should Health; 2005. be inferred. Toronto, Toronto, Ontario, Canada (Lee); Division of 14. Austin PC, van Walraven C, Wodchis WP, Urology, University Health Network, University of REFERENCES Newman A, Anderson GM. Using the Johns Hopkins Toronto, Toronto, Ontario, Canada (Kulkarni); Aggregated Diagnosis Groups (ADGs) to predict Department of Medicine, Sunnybrook Health 1. National Center for Health Statistics. Health, mortality in a general adult population cohort in Sciences Centre, University of Toronto, Toronto, United States, 2015. Hyattsville, MD: Centers for Ontario, Canada. Med Care. 2011;49(10):932-939. Ontario, Canada (Geerts, McLeod); Dalla Lana Disease Control and Prevention; 2016. 15. Levy AR, O’Brien BJ, Sellors C, Grootendorst P, School of Public Health, University of Toronto, 2. Hart RG, Pearce LA, Aguilar MI. Meta-analysis: Toronto, Ontario, Canada (Narod). Willison D. Coding accuracy of administrative drug antithrombotic therapy to prevent stroke in claims in the Ontario Drug Benefit database. Can J Author Contributions: Drs Wallis and Nam had full patients who have nonvalvular atrial fibrillation. Clin Pharmacol. 2003;10(2):67-71. access to all the data in the study and take Ann Intern Med. 2007;146(12):857-867. 16. Juurlink DN, Preyra C, Croxford R, et al. responsibility for the integrity of the data and the 3. Baigent C, Blackwell L, Collins R, et al; accuracy of the data analysis. Canadian Institute for Health Information Discharge Antithrombotic Trialists’ (ATT) Collaboration. Abstract Database: A Validation Study. Toronto, ON, Study concept and design: Wallis, Juvet, Narod, Nam. Aspirin in the primary and secondary prevention of Acquisition, analysis, or interpretation of data: All Canada: Institute for Clinical Evaluation Sciences; vascular disease: collaborative meta-analysis of 2006. authors. individual participant data from randomised trials. Drafting of the manuscript: Wallis, Herschorn, Lancet. 2009;373(9678):1849-1860. 17. Williams JI, Young W. A summary of studies on Narod, Nam. the quality of health care administrative databases Critical revision of the manuscript for important 4. Shehab N, Lovegrove MC, Geller AI, Rose KO, in Canada. In: Goel V, Williams J, Anderson G, et al, intellectual content: All authors. Weidle NJ, Budnitz DS. US emergency department eds. Patterns of Health Care in Ontario, Canada: The Statistical analysis: Wallis, Satkunasivam, Nam. visits for outpatient adverse drug events, ICES Practice Atlas. Ottawa, ON, Canada: Canadian Obtained funding: Nam. 2013-2014. JAMA. 2016;316(20):2115-2125. Medical Association; 1996:339-345. Administrative, technical, or material support: 5. Budnitz DS, Lovegrove MC, Shehab N, Richards 18. Robles SC, Marrett LD, Clarke EA, Risch HA. An Wallis, Nam. CL. Emergency hospitalizations for adverse drug application of capture-recapture methods to the Study supervision: Lee, Kulkarni, Geerts, Narod, events in older Americans. N Engl J Med. 2011;365 estimation of completeness of cancer registration. Nam. (21):2002-2012. J Clin Epidemiol. 1988;41(5):495-501. Conflict of Interest Disclosures: All authors have 6. Ruff CT, Giugliano RP, Braunwald E, et al. 19. Iron K, Zagorski BM, Sykora K, Manuel DG. completed and submitted the ICMJE Form for Comparison of the efficacy and safety of new oral Living and Dying in Ontario: An Opportunity for Disclosure of Potential Conflicts of Interest. anticoagulants with warfarin in patients with atrial Improved Health Information. Toronto, ON, Canada: Dr Herschorn reports grants and/or personal fees fibrillation: a meta-analysis of randomised trials. ICES Investigative Report; 2008. from Astellas, Ipsen, Pfizer, and Allergan. Dr Geerts Lancet. 2014;383(9921):955-962. 20. Diehr P, Yanez D, Ash A, Hornbrook M, Lin DY. reports personal fees and/or other funding from 7. Gaist D, García Rodríguez LA, Hellfritzsch M, Bayer, Pfizer, Leo Pharma, Bristol-Myers Squibb, Methods for analyzing health care utilization and et al. Association of antithrombotic drug use costs. Annu Rev Public Health. 1999;20:125-144. Sanofi, and Janssen. No other disclosures with subdural hematoma risk. JAMA. 2017;317(8): were reported. 836-846. 21. Moineddin R, Meaney C, Agha M, Zagorski B, Glazier RH. Modeling factors influencing the Funding/Support: Dr Wallis is supported by the 8. Culclasure TF, Bray VJ, Hasbargen JA. Canadian Institute of Health Research Banting demand for emergency department services in The significance of hematuria in the anticoagulated Ontario: a comparison of methods. BMC Emerg Med. and Best Doctoral Award. Dr Nam is supported patient. Arch Intern Med. 1994;154(6):649-652. by the Ajmera Family Chair in Urologic Oncology. 2011;11:13. This study was funded by the Ajmera Family Chair 9. Matulewicz RS, Meeks JJ. Blood in the urine 22. Vasdev N, Kumar A, Veeratterapillay R, in Urologic Oncology and Sunnybrook Foundation. (hematuria). JAMA. 2016;316(14):1508. Thorpe AC. Hematuria secondary to benign This study used deidentified data from the Institute 10. von Elm E, Altman DG, Egger M, Pocock SJ, prostatic hyperplasia: retrospective analysis of 166 of Clinical Evaluative Sciences Data Repository, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. men identified in a single one stop hematuria clinic. which is managed by the Institute for Clinical The Strengthening the Reporting of Observational Curr Urol. 2013;6(3):146-149. Evaluative Sciences with support from its funders Studies in Epidemiology (STROBE) statement: 23. Moreno JA, Martín-Cleary C, Gutiérrez E, et al. and partners: Canada’s Strategy for Patient- guidelines for reporting observational studies. Ann Haematuria: the forgotten CKD factor? Nephrol Dial Oriented Research (SPOR), the Ontario SPOR Intern Med. 2007;147(8):573-577. Transplant. 2012;27(1):28-34. Support Unit, the Canadian Institutes of Health 11. Benchimol EI, Smeeth L, Guttmann A, et al; Research, and the Government of Ontario. 24. Yao X, Abraham NS, Sangaralingham LR, et al. RECORD Working Committee. The REporting of Effectiveness and safety of dabigatran, rivaroxaban,

1270 JAMA October 3, 2017 Volume 318, Number 13 (Reprinted) jama.com

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Antithrombotic Medication Use and Hematuria-Related Complications Original Investigation Research

and apixaban versus warfarin in nonvalvular atrial gastrointestinal bleeding: a systematic review and systemic anticoagulation in hospitalized inpatients fibrillation. J Am Heart Assoc. 2016;5(6):e003725. meta-analysis. Gastroenterology. 2013;145(1): with gastrointestinal bleeding: a prospective study. 25. Larsen TB, Skjøth F, Nielsen PB, Kjældgaard JN, 105-112.e15. Am J Gastroenterol. 2015;110(2):328-335. Lip GY. Comparative effectiveness and safety of 31. Martinelli I, Lensing AW, Middeldorp S, et al. 37. Maura G, Blotière PO, Bouillon K, et al. non-vitamin K antagonist oral anticoagulants and Recurrent venous thromboembolism and abnormal Comparison of the short-term risk of bleeding and warfarin in patients with atrial fibrillation: uterine bleeding with anticoagulant and hormone arterial thromboembolic events in nonvalvular atrial propensity weighted nationwide cohort study. BMJ. therapy use. Blood. 2016;127(11):1417-1425. fibrillation patients newly treated with dabigatran 2016;353:i3189. 32. Adelborg K, Grove EL, Sundbøll J, Laursen M, or rivaroxaban versus vitamin K antagonists: 26. Mariani AJ, Mariani MC, Macchioni C, Stams Schmidt M. Sixteen-year nationwide trends in a French nationwide propensity-matched cohort UK, Hariharan A, Moriera A. The significance of antithrombotic drug use in Denmark and its study. Circulation. 2015;132(13):1252-1260. adult hematuria: 1,000 hematuria evaluations correlation with landmark studies. Heart. 2016;102 38. Yu J, Mehran R, Dangas GD, et al. Safety and including a risk-benefit and cost-effectiveness (23):1883-1889. efficacy of high- versus low-dose aspirin after analysis. J Urol. 1989;141(2):350-355. 33. Groninger H, Phillips JM. Gross hematuria: primary percutaneous coronary intervention in 27. Avidor Y, Nadu A, Matzkin H. Clinical assessment and management at the end of life. ST-segment elevation myocardial infarction: the significance of gross hematuria and its evaluation in J Hosp Palliat Nurs. 2012;14(3):184-188. HORIZONS-AMI (Harmonizing Outcomes With Revascularization and Stents in Acute Myocardial patients receiving anticoagulant and aspirin 34. Raparelli V, Proietti M, Cangemi R, Lip GY, Lane treatment. Urology. 2000;55(1):22-24. Infarction) trial. JACC Cardiovasc Interv. 2012;5(12): DA, Basili S. Adherence to oral anticoagulant 1231-1238. 28. Gomes T, Mamdani MM, Holbrook AM, therapy in patients with atrial fibrillation: focus on Paterson JM, Hellings C, Juurlink DN. Rates of non-vitamin K antagonist oral anticoagulants. 39. Huang ES, Strate LL, Ho WW, Lee SS, Chan AT. hemorrhage during warfarin therapy for atrial Thromb Haemost. 2017;117(2):209-218. Long-term use of aspirin and the risk of fibrillation. CMAJ. 2013;185(2):E121-E127. gastrointestinal bleeding. Am J Med. 2011;124(5): 35. Jackevicius CA, Tsadok MA, Essebag V, et al. 426-433. 29. Chai-Adisaksopha C, Crowther M, Isayama T, Early non-persistence with dabigatran and Lim W. The impact of bleeding complications in rivaroxaban in patients with atrial fibrillation. Heart. 40. Rodriguez RA, Carrier M, Wells PS. patients receiving target-specific oral 2017;103(17):1331-1338. Non-adherence to new oral anticoagulants: a reason for concern during long-term anticoagulants: a systematic review and 36. Sengupta N, Feuerstein JD, Patwardhan VR, meta-analysis. Blood. 2014;124(15):2450-2458. anticoagulation? J Thromb Haemost. 2013;11(2): et al. The risks of thromboembolism vs. recurrent 390-394. 30. Holster IL, Valkhoff VE, Kuipers EJ, Tjwa ET. gastrointestinal bleeding after interruption of New oral anticoagulants increase risk for

jama.com (Reprinted) JAMA October 3, 2017 Volume 318, Number 13 1271

Downloaded From: https://jamanetwork.com/ on 10/01/2021