ORIGINAL ARTICLE
Warfarin Treatment and Outcomes of Patients With Atrial Fibrillation in Rural and Urban Settings Meytal Avgil Tsadok, PhD;1 Cynthia A. Jackevicius, PharmD, MSc;2,3,4,5 Vidal Essebag, MD, PhD;6 Mark J. Eisenberg, MD, MPH;7 Elham Rahme, PhD;1 & Louise Pilote, MD, PhD1 1 Division of Clinical Epidemiology, McGill University Health Centre, Montreal, Quebec, Canada 2 Department of Pharmacy Practice and Administration, College of Pharmacy, Western University of Health Sciences, Pomona, California 3 Institute for Clinical Evaluative Sciences, Toronto, Canada 4 Department of Health Policy, Management and Evaluation, Faculty of Medicine, University of Toronto, Toronto, Canada 5 University Health Network, Toronto, Canada 6 Division of Cardiology, McGill University Health Centre, Montreal, Quebec, Canada 7 Division of Cardiology, Jewish General Hospital, McGill University Health Centre, Montreal, Quebec, Canada
Abstract Background: Warfarin is an effective agent in the prevention of stroke in Funding: Financial support for this study was provided by an operating grant from the CIHR (grant number MOP-84304). For further information, contact Louise Pilote, MD, MPH, PhD, McGill University Health Centre, 687 Pine Ave West, V Building, Montreal, Quebec, H3A 1A1 Canada; e-mail: [email protected]. doi: 10.1111/jrh.12110
patients with atrial fibrillation (AF). However, it requires close monitoring with regular visits to health care facilities. To date, it is unknown whether there is a difference in warfarin utilization and outcomes between urban and rural settings. Methods: We used administrative databases to compare warfarin utilization patterns and stroke and major bleeding outcomes in rural and urban settings in a population-based cohort study of patients ࣙ 65 years admitted to hospital with a diagnosis of AF in the province of Quebec, Canada, from 1999 to 2007. Patients’ postal codes were used to differentiate between rural and urban settings. Results: The cohort comprised 18,198 rural (21.8%) and 65,315 urban (78.2%) patients, with similar mean age of 79 years and a similar burden of comorbidities. Overall, there was marked underutilization of warfarin in both rural and urban settings. Warfarin-filled prescription rates were slightly higher in the rural setting (adjusted OR: 1.16, 95% CI: 1.12-1.20). In multivariable Cox regression analyses, the risk for stroke and major bleeding in rural settings was similar to that in urban settings (stroke: adjusted HR: 1.01, 95% CI: 0.95-1.09; major bleeding: adjusted HR: 1.03, 95% CI: 0.95-1.12). Conclusions: Patients in rural settings were slightly more likely to fill a prescription for warfarin, but they experienced similar stroke and major bleeding rates to their urban counterparts.
Key words atrial fibrillation, observational study, stroke prevention, utilization of health services, warfarin.
Atrial fibrillation (AF), the most common chronic cardiac arrhythmia, is associated with a 5-fold increase in stroke risk.1 Warfarin has historically been the gold standard therapy for most AF patients since warfarin was found to reduce the risk of stroke by two-thirds compared to no antithrombotic therapy.2 However, the narrow therapeutic index of warfarin and its association with an increased
310
risk of bleeding necessitate frequent monitoring and dose adjustments according to the international normalized ratio (INR), which may be inconvenient for patients and requires them to attend health care facilities on a regular basis. Patients in rural settings have been found to differ from patients in urban settings in cardiovascular risk
c 2015 National Rural Health Association The Journal of Rural Health 31 (2015) 310–315
Warfarin Treatment in AF—Regional Variance
Avgil Tsadok et al.
factor prevalence, outcomes, and mode of treatment.3-6 In Canada, approximately 20% of the population lives in rural areas. A pan-Canadian report showed that people living in rural and remote areas of Canada are at a disadvantage in health status, access to care, and highly specialized health professionals.7 Patients who live in rural settings often need to travel long distances to receive medical care or laboratory monitoring of INR, and therefore differences between warfarin prescribing and monitoring may exist among rural versus urban patients. Recently, new antithrombotic agents have been introduced. These agents do not require routine monitoring, so patients are no longer tied to clinical facilities for INR testing. These agents may be more convenient in rural settings where patients may be treated less often with warfarin due to their inability to present for INR monitoring. However, to date, it is unknown whether there is a difference in warfarin treatment and outcomes between urban and rural settings. In this study, we aimed to compare warfarin utilization patterns, stroke, and major bleeding events in rural and urban settings, using a population-based cohort study of patients discharged alive from hospitalization with the diagnosis of AF in Quebec, Canada.
Methods Study Design and Population We conducted a retrospective cohort study of patients with AF, using administrative data with linkages between prescription drug claims, physician claims, and hospital discharge databases. Cohort formation has been described in detail elsewhere.8,9 In brief, participants were Quebec residents 65 years and older, discharged alive from hospital with a primary diagnosis of AF or a major comorbid diagnosis (secondary diagnosis) of AF (International Classification of Diseases – 9th/10th (ICD-9/10) revision, code 427.3, 427.31, or 427.32 /I48), between January 1, 1999, and March 31, 2007. We linked the provincial hospital discharge database (Maintenance et Exploitation des ´ ´ pour l’Etude ` Hospitaliere ` – MedDonnees de la Clientele Echo) to the provincial physician and prescription claims ´ ´ database (la Regie de l’assurance maladie du Quebec – ´ RAMQ). The Quebec prescription claims database has been previously determined to be a reliable source of filled medication prescriptions.10 We used the Med-Echo database to obtain information on patients’ characteristics at cohort entry such as comorbidities and the length of hospital stay, and to calculate the CHADS2 score (congestive heart failure, hypertension, age ࣙ 75 years, diabetes mellitus, prior stroke, or transient ischemic attack [TIA]) and the HAS-BLED score
c 2015 National Rural Health Association The Journal of Rural Health 31 (2015) 310–315
(hypertension, abnormal renal function, abnormal liver function, history of stroke, history of bleeding, labile INR, age ࣙ 65 years, drug therapy [antiplatelet agents, nonsteroidal antiinflammatory drugs], and alcohol intake). Since our databases do not provide information on labile INR and alcohol intake, we calculated a modified HAS-BLED score, with a maximum score of 7 rather than 9.11,12 We used the RAMQ database to obtain information on medication prescriptions filled after discharge from the index hospital admission. Initial treatment was determined according to the first prescription dispensed within 7 days postdischarge. The time window for the first prescription was selected to capture most patients with a prescription at discharge while minimizing the potential for survival bias.13 Patients’ residential postal codes were used to differentiate between rural and urban settings. Postal codes were determined during the index admission, according to the RAMQ database. The Canadian postal code is a 6character alphanumeric string, with the first 3 digits of the postal code called “forward sortation areas” (FSA). The FSA digits specify if the FSA is urban or rural: zero indicates a rural region, where there are no letter carriers (ie, residents go to a post office or corner postal box to pick up their mail), while all other digits indicate urban areas.14 To determine the rural or urban settings, we used the FSA digit (0 or ࣙ 1). Previous studies concluded that postal code locations are a reasonably accurate proxy for address location in Canada, and they can be used to differentiate between rural and urban settings.15-18
Outcomes Our outcomes were first hospital admission or emergency room visit for stroke/TIA or a major bleeding event that required hospitalization. Stroke was defined as cerebral thrombosis, embolism or artery occlusion, including TIA and retinal infarct (ICD-9/10 codes available upon request). Bleeding events included all gastrointestinal hemorrhage, intracranial hemorrhage, intraocular hemorrhage, aortic aneurysm dissection/rupture, hematuria, hemoptysis, epistaxis, and unspecified hemorrhage (ICD9/10 codes available upon request). Follow-up started at the index discharge date and the criteria to end a patient’s follow-up were a first diagnosis of stroke/TIA, bleeding event, death, or end of study period, whichever came first.
Statistical Analysis Descriptive statistics were used to compare demographic characteristics, comorbidities, and pattern of medication
311
Warfarin Treatment in AF—Regional Variance
Avgil Tsadok et al.
Table 1 Baseline Demographic and Clinical Characteristics by Setting
Mean age at admission, years (SD) Age ࣙ 75 years (%) Male sex (%) AF as primary diagnosis (%)a : Primary treating physician during hospitalization (%) General Internist Cardiologist Median length of index hospitalization, days (IQR) Comorbidities, within 1 year prior to AF (%) Hypertension CAD Congestive heart failure Diabetes Hyperlipidemia Valvular disease AMI Chronic kidney disease Acute kidney disease Previous stroke History of bleeding event Liver diseases Mean CHADS2 score (±SD) Low risk (0), % Moderate risk (1), % High risk (ࣙ2), % Mean HAS-BLED score (±SD) Low risk (1), % Moderate risk (2), % High risk (ࣙ3), % Filled prescription for other medication (within 30 days postdischarge), (%): Rate control drugs Rhythm control drugs ASA (Acetylsalicylic acid (Aspirin)) Clopidogrel NSAIDs (nonsteroidal antiinflammatory drugs)
Urban (n = 65,315)
Rural (n = 18,198)
79.0 (1.9) 68.6 46.1 24.9
78.7 (1.9) 66.6 51.0 27.3
43.6 10.7 22.5 8 (4,15)
65.8 6.5 11.4 7 (4,14)
58.7 45.8 28.2 23.4 22.8 19.4 17.2 16.4 8.6 7.4 6.4 3.2 1.93 ± 1.11 8.7 27.6 63.7 2.30 ± 0.97 21.8 39.4 38.8
55.2 44.2 28.9 22.9 22.6 18.8 15.4 15.4 8.7 7.6 5.9 3.0 1.88 ± 1.12 9.6 29.0 61.3 2.25 ± 0.96 23.4 40.2 36.5
70.5 23.9 30.6 3.9 1.5
69.1 21.9 29.8 3.9 1.6
SD, standard deviation; AF, atrial fibrillation; IQR, interquartile range; CAD, coronary artery disease; AMI, acute myocardial infarction. a About 1% of the population had both primary and secondary diagnosis of AF; we considered this population as having primary diagnosis of AF.
usage between rural and urban settings. Continuous variables are presented as mean ± SD and were compared by Student’s t test. Dichotomous variables are presented as percentages and were compared by chi-square test. Incidence rates of stroke and bleeding events were calculated as the number of events per 100 person-years of followup. To identify independent determinants of warfarinfilled prescriptions, multivariable analysis with logistic regression was performed. To assess determinants of stroke and bleeding outcomes, we conducted intention to treat analyses using Cox proportional hazards models, including exposure to warfarin, defined as a warfarin
312
prescription filled within 7 days of discharge. We considered warfarin treatment as a time-fixed binary variable, assuming that patients remain on the same treatment throughout the follow-up period. This method is akin to intention to treat design in clinical trials.19 All models were adjusted for age, sex, length of index hospitalization, type of AF diagnosis (primary vs secondary), primary treating physician, comorbidities within 1 year prior to admission (hypertension, coronary artery disease, congestive heart failure, diabetes, hyperlipidemia, valvular disease, acute myocardial infarction, liver diseases, acute and chronic kidney diseases, previous stroke, and a history of bleeding event), and use
c 2015 National Rural Health Association The Journal of Rural Health 31 (2015) 310–315
Warfarin Treatment in AF—Regional Variance
Avgil Tsadok et al.
Results A total of 83,513 patients with AF were observed during the study period, with a mean follow-up of 2.8 ± 2.2 years (range 0-8 years). The population was considerably less rural than urban (21.8% vs 78.2%). Mean age was 79 years in both populations and male sex was more prevalent among the rural population (51.0% vs 46.1%). The distribution of AF diagnosis (primary vs secondary) was similar in both settings, with the majority of patients diagnosed with AF as a secondary diagnosis. In rural settings, most patients were primarily treated by a general physician at index hospitalization (65.8%), while only 11.4% were treated by a cardiologist and 6.5% by internist physicians (Table 1). Burden of comorbidities within 1 year prior to index admission was similar in both settings. CHADS2 score and HASBLED score distributions were comparable in rural and urban settings, with the majority of patients having a high-risk score for stroke (ࣙ2) in both populations (Table 1). Rates of warfarin prescriptions filled within 7 days postdischarge were slightly, but statistically significantly higher, among patients in rural compared with urban settings (Table 2). Patients who were primarily treated by cardiologists filled more prescriptions for warfarin compared to patients who were primarily treated by an internist or general physician, with no differences between settings. The mean initial warfarin prescription daily dose was slightly lower among rural patients (4.1 ± 2.7 vs 4.4 ± 3.3 mg, P < .001) (Table 2). In multivariable logistic regression analysis, rural patients had a slightly greater odds of filling warfarin prescriptions compared with urban patients (adjusted OR: 1.16, 95% CI: 1.12-1.20), independent of age, sex, comorbidities, and primary treating physician. Crude stroke and major bleeding incidences per 100person-years did not differ between settings (Figure 1). In multivariable Cox regression analysis adjusting for sex, age, type of treating physician, comorbidities, and warfarin and antiplatelet use, we did not find an association between rural setting and stroke events (adjusted
c 2015 National Rural Health Association The Journal of Rural Health 31 (2015) 310–315
Table 2 Warfarin Prescription Within 7 Days Postdischarge by Setting Urban
Rural
P Value
Patients with filled warfarin prescription (%) 47.9 51.6 Initial prescribed dose, mg (mean ± SD) 4.4 ± 3.3 4.1 ± 2.7 According to CHADS2 score (%) 0 44.0 46.6 1 45.2 49.8 ࣙ2 49.6 53.2 According to HAS-BLED score (%) 1 53.6 57.0 2 51.0 53.9 ࣙ3 41.5 45.5 According to primary treating physician: General Practitioner 50.7 53.8 Internist 44.5 51.1 Cardiologist 58.0 60.3
Warfarin Treatment and Outcomes of Patients With Atrial Fibrillation in Rural and Urban Settings Meytal Avgil Tsadok, PhD;1 Cynthia A. Jackevicius, PharmD, MSc;2,3,4,5 Vidal Essebag, MD, PhD;6 Mark J. Eisenberg, MD, MPH;7 Elham Rahme, PhD;1 & Louise Pilote, MD, PhD1 1 Division of Clinical Epidemiology, McGill University Health Centre, Montreal, Quebec, Canada 2 Department of Pharmacy Practice and Administration, College of Pharmacy, Western University of Health Sciences, Pomona, California 3 Institute for Clinical Evaluative Sciences, Toronto, Canada 4 Department of Health Policy, Management and Evaluation, Faculty of Medicine, University of Toronto, Toronto, Canada 5 University Health Network, Toronto, Canada 6 Division of Cardiology, McGill University Health Centre, Montreal, Quebec, Canada 7 Division of Cardiology, Jewish General Hospital, McGill University Health Centre, Montreal, Quebec, Canada
Abstract Background: Warfarin is an effective agent in the prevention of stroke in Funding: Financial support for this study was provided by an operating grant from the CIHR (grant number MOP-84304). For further information, contact Louise Pilote, MD, MPH, PhD, McGill University Health Centre, 687 Pine Ave West, V Building, Montreal, Quebec, H3A 1A1 Canada; e-mail: [email protected]. doi: 10.1111/jrh.12110
patients with atrial fibrillation (AF). However, it requires close monitoring with regular visits to health care facilities. To date, it is unknown whether there is a difference in warfarin utilization and outcomes between urban and rural settings. Methods: We used administrative databases to compare warfarin utilization patterns and stroke and major bleeding outcomes in rural and urban settings in a population-based cohort study of patients ࣙ 65 years admitted to hospital with a diagnosis of AF in the province of Quebec, Canada, from 1999 to 2007. Patients’ postal codes were used to differentiate between rural and urban settings. Results: The cohort comprised 18,198 rural (21.8%) and 65,315 urban (78.2%) patients, with similar mean age of 79 years and a similar burden of comorbidities. Overall, there was marked underutilization of warfarin in both rural and urban settings. Warfarin-filled prescription rates were slightly higher in the rural setting (adjusted OR: 1.16, 95% CI: 1.12-1.20). In multivariable Cox regression analyses, the risk for stroke and major bleeding in rural settings was similar to that in urban settings (stroke: adjusted HR: 1.01, 95% CI: 0.95-1.09; major bleeding: adjusted HR: 1.03, 95% CI: 0.95-1.12). Conclusions: Patients in rural settings were slightly more likely to fill a prescription for warfarin, but they experienced similar stroke and major bleeding rates to their urban counterparts.
Key words atrial fibrillation, observational study, stroke prevention, utilization of health services, warfarin.
Atrial fibrillation (AF), the most common chronic cardiac arrhythmia, is associated with a 5-fold increase in stroke risk.1 Warfarin has historically been the gold standard therapy for most AF patients since warfarin was found to reduce the risk of stroke by two-thirds compared to no antithrombotic therapy.2 However, the narrow therapeutic index of warfarin and its association with an increased
310
risk of bleeding necessitate frequent monitoring and dose adjustments according to the international normalized ratio (INR), which may be inconvenient for patients and requires them to attend health care facilities on a regular basis. Patients in rural settings have been found to differ from patients in urban settings in cardiovascular risk
c 2015 National Rural Health Association The Journal of Rural Health 31 (2015) 310–315
Warfarin Treatment in AF—Regional Variance
Avgil Tsadok et al.
factor prevalence, outcomes, and mode of treatment.3-6 In Canada, approximately 20% of the population lives in rural areas. A pan-Canadian report showed that people living in rural and remote areas of Canada are at a disadvantage in health status, access to care, and highly specialized health professionals.7 Patients who live in rural settings often need to travel long distances to receive medical care or laboratory monitoring of INR, and therefore differences between warfarin prescribing and monitoring may exist among rural versus urban patients. Recently, new antithrombotic agents have been introduced. These agents do not require routine monitoring, so patients are no longer tied to clinical facilities for INR testing. These agents may be more convenient in rural settings where patients may be treated less often with warfarin due to their inability to present for INR monitoring. However, to date, it is unknown whether there is a difference in warfarin treatment and outcomes between urban and rural settings. In this study, we aimed to compare warfarin utilization patterns, stroke, and major bleeding events in rural and urban settings, using a population-based cohort study of patients discharged alive from hospitalization with the diagnosis of AF in Quebec, Canada.
Methods Study Design and Population We conducted a retrospective cohort study of patients with AF, using administrative data with linkages between prescription drug claims, physician claims, and hospital discharge databases. Cohort formation has been described in detail elsewhere.8,9 In brief, participants were Quebec residents 65 years and older, discharged alive from hospital with a primary diagnosis of AF or a major comorbid diagnosis (secondary diagnosis) of AF (International Classification of Diseases – 9th/10th (ICD-9/10) revision, code 427.3, 427.31, or 427.32 /I48), between January 1, 1999, and March 31, 2007. We linked the provincial hospital discharge database (Maintenance et Exploitation des ´ ´ pour l’Etude ` Hospitaliere ` – MedDonnees de la Clientele Echo) to the provincial physician and prescription claims ´ ´ database (la Regie de l’assurance maladie du Quebec – ´ RAMQ). The Quebec prescription claims database has been previously determined to be a reliable source of filled medication prescriptions.10 We used the Med-Echo database to obtain information on patients’ characteristics at cohort entry such as comorbidities and the length of hospital stay, and to calculate the CHADS2 score (congestive heart failure, hypertension, age ࣙ 75 years, diabetes mellitus, prior stroke, or transient ischemic attack [TIA]) and the HAS-BLED score
c 2015 National Rural Health Association The Journal of Rural Health 31 (2015) 310–315
(hypertension, abnormal renal function, abnormal liver function, history of stroke, history of bleeding, labile INR, age ࣙ 65 years, drug therapy [antiplatelet agents, nonsteroidal antiinflammatory drugs], and alcohol intake). Since our databases do not provide information on labile INR and alcohol intake, we calculated a modified HAS-BLED score, with a maximum score of 7 rather than 9.11,12 We used the RAMQ database to obtain information on medication prescriptions filled after discharge from the index hospital admission. Initial treatment was determined according to the first prescription dispensed within 7 days postdischarge. The time window for the first prescription was selected to capture most patients with a prescription at discharge while minimizing the potential for survival bias.13 Patients’ residential postal codes were used to differentiate between rural and urban settings. Postal codes were determined during the index admission, according to the RAMQ database. The Canadian postal code is a 6character alphanumeric string, with the first 3 digits of the postal code called “forward sortation areas” (FSA). The FSA digits specify if the FSA is urban or rural: zero indicates a rural region, where there are no letter carriers (ie, residents go to a post office or corner postal box to pick up their mail), while all other digits indicate urban areas.14 To determine the rural or urban settings, we used the FSA digit (0 or ࣙ 1). Previous studies concluded that postal code locations are a reasonably accurate proxy for address location in Canada, and they can be used to differentiate between rural and urban settings.15-18
Outcomes Our outcomes were first hospital admission or emergency room visit for stroke/TIA or a major bleeding event that required hospitalization. Stroke was defined as cerebral thrombosis, embolism or artery occlusion, including TIA and retinal infarct (ICD-9/10 codes available upon request). Bleeding events included all gastrointestinal hemorrhage, intracranial hemorrhage, intraocular hemorrhage, aortic aneurysm dissection/rupture, hematuria, hemoptysis, epistaxis, and unspecified hemorrhage (ICD9/10 codes available upon request). Follow-up started at the index discharge date and the criteria to end a patient’s follow-up were a first diagnosis of stroke/TIA, bleeding event, death, or end of study period, whichever came first.
Statistical Analysis Descriptive statistics were used to compare demographic characteristics, comorbidities, and pattern of medication
311
Warfarin Treatment in AF—Regional Variance
Avgil Tsadok et al.
Table 1 Baseline Demographic and Clinical Characteristics by Setting
Mean age at admission, years (SD) Age ࣙ 75 years (%) Male sex (%) AF as primary diagnosis (%)a : Primary treating physician during hospitalization (%) General Internist Cardiologist Median length of index hospitalization, days (IQR) Comorbidities, within 1 year prior to AF (%) Hypertension CAD Congestive heart failure Diabetes Hyperlipidemia Valvular disease AMI Chronic kidney disease Acute kidney disease Previous stroke History of bleeding event Liver diseases Mean CHADS2 score (±SD) Low risk (0), % Moderate risk (1), % High risk (ࣙ2), % Mean HAS-BLED score (±SD) Low risk (1), % Moderate risk (2), % High risk (ࣙ3), % Filled prescription for other medication (within 30 days postdischarge), (%): Rate control drugs Rhythm control drugs ASA (Acetylsalicylic acid (Aspirin)) Clopidogrel NSAIDs (nonsteroidal antiinflammatory drugs)
Urban (n = 65,315)
Rural (n = 18,198)
79.0 (1.9) 68.6 46.1 24.9
78.7 (1.9) 66.6 51.0 27.3
43.6 10.7 22.5 8 (4,15)
65.8 6.5 11.4 7 (4,14)
58.7 45.8 28.2 23.4 22.8 19.4 17.2 16.4 8.6 7.4 6.4 3.2 1.93 ± 1.11 8.7 27.6 63.7 2.30 ± 0.97 21.8 39.4 38.8
55.2 44.2 28.9 22.9 22.6 18.8 15.4 15.4 8.7 7.6 5.9 3.0 1.88 ± 1.12 9.6 29.0 61.3 2.25 ± 0.96 23.4 40.2 36.5
70.5 23.9 30.6 3.9 1.5
69.1 21.9 29.8 3.9 1.6
SD, standard deviation; AF, atrial fibrillation; IQR, interquartile range; CAD, coronary artery disease; AMI, acute myocardial infarction. a About 1% of the population had both primary and secondary diagnosis of AF; we considered this population as having primary diagnosis of AF.
usage between rural and urban settings. Continuous variables are presented as mean ± SD and were compared by Student’s t test. Dichotomous variables are presented as percentages and were compared by chi-square test. Incidence rates of stroke and bleeding events were calculated as the number of events per 100 person-years of followup. To identify independent determinants of warfarinfilled prescriptions, multivariable analysis with logistic regression was performed. To assess determinants of stroke and bleeding outcomes, we conducted intention to treat analyses using Cox proportional hazards models, including exposure to warfarin, defined as a warfarin
312
prescription filled within 7 days of discharge. We considered warfarin treatment as a time-fixed binary variable, assuming that patients remain on the same treatment throughout the follow-up period. This method is akin to intention to treat design in clinical trials.19 All models were adjusted for age, sex, length of index hospitalization, type of AF diagnosis (primary vs secondary), primary treating physician, comorbidities within 1 year prior to admission (hypertension, coronary artery disease, congestive heart failure, diabetes, hyperlipidemia, valvular disease, acute myocardial infarction, liver diseases, acute and chronic kidney diseases, previous stroke, and a history of bleeding event), and use
c 2015 National Rural Health Association The Journal of Rural Health 31 (2015) 310–315
Warfarin Treatment in AF—Regional Variance
Avgil Tsadok et al.
Results A total of 83,513 patients with AF were observed during the study period, with a mean follow-up of 2.8 ± 2.2 years (range 0-8 years). The population was considerably less rural than urban (21.8% vs 78.2%). Mean age was 79 years in both populations and male sex was more prevalent among the rural population (51.0% vs 46.1%). The distribution of AF diagnosis (primary vs secondary) was similar in both settings, with the majority of patients diagnosed with AF as a secondary diagnosis. In rural settings, most patients were primarily treated by a general physician at index hospitalization (65.8%), while only 11.4% were treated by a cardiologist and 6.5% by internist physicians (Table 1). Burden of comorbidities within 1 year prior to index admission was similar in both settings. CHADS2 score and HASBLED score distributions were comparable in rural and urban settings, with the majority of patients having a high-risk score for stroke (ࣙ2) in both populations (Table 1). Rates of warfarin prescriptions filled within 7 days postdischarge were slightly, but statistically significantly higher, among patients in rural compared with urban settings (Table 2). Patients who were primarily treated by cardiologists filled more prescriptions for warfarin compared to patients who were primarily treated by an internist or general physician, with no differences between settings. The mean initial warfarin prescription daily dose was slightly lower among rural patients (4.1 ± 2.7 vs 4.4 ± 3.3 mg, P < .001) (Table 2). In multivariable logistic regression analysis, rural patients had a slightly greater odds of filling warfarin prescriptions compared with urban patients (adjusted OR: 1.16, 95% CI: 1.12-1.20), independent of age, sex, comorbidities, and primary treating physician. Crude stroke and major bleeding incidences per 100person-years did not differ between settings (Figure 1). In multivariable Cox regression analysis adjusting for sex, age, type of treating physician, comorbidities, and warfarin and antiplatelet use, we did not find an association between rural setting and stroke events (adjusted
c 2015 National Rural Health Association The Journal of Rural Health 31 (2015) 310–315
Table 2 Warfarin Prescription Within 7 Days Postdischarge by Setting Urban
Rural
P Value
Patients with filled warfarin prescription (%) 47.9 51.6 Initial prescribed dose, mg (mean ± SD) 4.4 ± 3.3 4.1 ± 2.7 According to CHADS2 score (%) 0 44.0 46.6 1 45.2 49.8 ࣙ2 49.6 53.2 According to HAS-BLED score (%) 1 53.6 57.0 2 51.0 53.9 ࣙ3 41.5 45.5 According to primary treating physician: General Practitioner 50.7 53.8 Internist 44.5 51.1 Cardiologist 58.0 60.3
