Abstract
Background
Patients with atrial fibrillation (AF) are at risk for both thromboembolic and bleeding complications. While the risk for thromboembolism is higher among women with AF than men, the sex‐related differences in post‐discharge outcomes after hospitalization is not clearly understood.
Hypothesis
Compared to men, women hospitalized for AF are at a higher risk of both thromboembolic and bleeding complications.
Methods
We conducted a retrospective cohort study using data from the 2013 to 2014 Nationwide Readmission Database (NRD), to compare outcomes among men and women, ≥50 years of age after hospitalization for AF. The primary patient outcome was all‐cause rehospitalization at 90‐days after initial hospitalization. Survey‐weighted Cox proportional hazard regression models were used to estimate the hazard ratios (HR) and their 95% confidence intervals (CI) for bleeding events at 30, 60, 90, and 270 days after hospitalization.
Results
From the 28 million patients in the NRD, we identified 522 521 individuals with an index hospitalization for AF. Compared to men, women hospitalized for AF accounted for 53.3% of the cohort and had higher rates of thrombotic (1.7%, 1.4%) and bleeding complications (1.4%, 1.1%). After adjustment, the 90‐day risk among women vs men was significantly greater; all‐cause rehospitalization (24.2%, 17.0%; HR = 1.07, 95% CI = 1.05‐1.09), rehospitalization related to ischemic stroke (0.6%, 0.3%; HR 1.31, 95% CI = 1.14‐1.51), pulmonary embolism (0.4%, 0.2%; HR 1.21, 95% CI = 1.01‐1.45), and any thrombotic event (1.3%, 0.7%; HR 1.20, 95% CI = 1.09‐1.32).
Conclusions
Hospitalization for AF is common and frequently associated with both in‐hospital complications and readmission, which were more commonly observed among women with AF. Further research into epidemiological factors and treatment differences between men and women with AF is warranted.
Keywords: atrial fibrillation, outcomes, readmission
1. INTRODUCTION
Atrial fibrillation (AF) affects 5.2 million Americans, with 12 million projected to be affected by 2050, commensurate with the “gray tsunami” of an aging US population.1, 2 AF carries a 5‐fold higher risk of stroke,3, 4, 5 2‐fold risk for heart failure,6 and 50% higher odds of dying.7, 8, 9, 10 AF is a frequent cause of rehospitalizations11 and the excess annual national cost from treating AF patients exceeds $26 billion.12
The epidemiology and natural history of AF differ between men and women.8, 13, 14, 15 Although AF incidence rates are lower among women as compared to men, once AF develops, it confers a higher risk for thromboembolic complications and death among women.13, 15, 16 Unfortunately, compared to men, older women are also at higher risk for bleeding complications from use of oral anticoagulants.17 There is emerging literature highlighting gender‐based differences regarding vulnerability to thromboembolism and bleeding in the context of AF.
Even though AF frequently leads to hospital admission, few large contemporary studies have examined gender‐specific differences in the prognosis of patients hospitalized with AF, particularly with respect to rates and causes of rehospitalization. We sought to fill these knowledge gaps by performing a retrospective analysis of nationally representative hospitalization data from the 2013 and 2014 Nationwide Readmission Database (NRD).18, 19
2. METHODS
We conducted a claims‐based, retrospective, cohort study comparing index hospitalizations of AF among women and men aged 50 or older. We used the NRD from years 2013 and 2014 that contains nationally representative information on hospital admissions with patient linkage numbers to track readmissions within a state. In the 2013 NRD, there are approximately 14 million discharges from 2006 hospitals from 21 state inpatient databases; representing 49.3% of the US population and 49.1% of US hospitalizations.18 In the 2014 NRD, there are approximately 14 million discharges from 2048 hospitals from 22 state inpatient databases; representing 51.2% of the US population and 49.3% of US hospitalizations.19 The NRD includes all discharges as well as patients who have died in the hospital. Diagnoses and procedures during each hospitalization are categorized using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) codes.
2.1. Study sample
We identified all individuals in the NRD who were hospitalized with a principal diagnosis of AF using standard ICD‐9‐CM codes given in Supporting Information Table S1. First, we identified all the initial visits (referred to as index event of AF) for a primary diagnosis of AF. From 70 886 775 weighted hospitalizations, we identified 690 433 potential index visits of AF among those 50 years and older. Second, we excluded hospitalizations during the last 3 months of 2013 and in 2014 to allow a minimum of 3 months follow‐up duration in the cohort (n = 528 088). Since our primary analyses focused on post‐discharge prognosis, we excluded those AF hospitalizations that died at the index hospitalization (n = 5567 died during their index hospitalization), leaving 522 521 discharged alive and eligible. These exclusions were performed to minimize selection bias. The flow chart for patient selection is outlined in Figure S1.
2.2. Outcomes
The primary clinical outcome was rehospitalization related to thrombotic events, (assessed using standard ICD‐9‐CM codes) and the primary patient outcome was all‐cause rehospitalization. We defined thrombotic events as a composite measure of ischemic stroke (IS), transient ischemic event (TIA), deep vein thrombosis (DVT) or pulmonary embolism (PE). The secondary outcomes were all‐cause rehospitalization, bleeding, and thrombotic events at 30, 60, and 270‐days after hospitalization, along with individual components of the composite outcomes at 90‐days. We defined bleeding related rehospitalization as a composite measure of anemia, intracranial hemorrhage, gastrointestinal bleed, and bleeding requiring transfusion. Definitions for etiology of rehospitalization were determined using the same ICD‐9‐CM codes. However, to isolate admissions related to clinically significant anemia events, we considered only those rehospitalizations with a bleeding‐related ICD‐9‐CM diagnosis codes in the first or second position. The ICD‐9 diagnostic codes used are presented in Table S1.
2.3. Covariates
The main exposure group was sex (women as compared with men). The other patient‐level baseline covariates used were age (categories of 50‐64, 65‐74, and 75‐90), location (central metro with >1 million population, fringe metro with >1 million population, metro with population 250 000 to <1 million and micropolitan areas with population < 250 000), insurance provider (private/medicare and medicaid/self‐pay/no charge/other forms), median household national income quartiles ($1‐$37 999, $38 000‐$47 999, $48 000‐$63 999, and > =$64 000) and whether the patient was electively admitted. Hospital‐level covariates were bed size of the hospital (small, medium, and large), teaching status of the hospital (metro non‐teaching, metro teaching, and non‐metro) and whether the hospital was an urban hospital.
We also used 29 clinical comorbidities at index hospitalization already derived in the dataset from ICD‐9 CM diagnosis codes present at the index hospitalization to calculate a cumulative comorbidity using the Elixhauser comorbidity score.20 Admission characteristics during the index hospitalization were also measured, such as number of procedures, number of diagnoses, and severity of illness and risk of mortality based on validated Agency for Health Care Research and Quality (AHRQ) methods.21 Coincident diagnoses related to bleeding and thromboembolic outcomes were also described.
2.4. Statistical analysis
All comparisons were performed between women and men hospitalized for a primary diagnosis of AF. First, we compared baseline patient and hospital characteristics; categorical variables were compared using χ 2 tests and continuous variables were compared using Student's t test. Second, procedural characteristics, comorbidities and co‐diagnosis of bleeding components at baseline were compared using χ 2 tests. All counts and percentages were survey‐weighted. Third, we used survey‐weighted Cox proportional hazard regression models to derive age‐adjusted hazard ratios (HR) and 95% confidence intervals (95% CI) with corresponding P‐values to assess the difference in risk of outcomes by sex at 30, 60, 90, and 270 days. Prior to this, we tested proportionality assumption using Schoenfeld residuals. For survival analysis, we created dates of events using given months and a random date during that month and used the duration variable from the dataset to construct the event date. For those who did not have a rehospitalization, we assumed that they are alive until December 31, 2013, and therefore, were censored at December 31, 2013 and for 2014, at December 31, 2014. Fourth, we obtained a multivariable Cox proportional hazard model additionally adjusting for location, insurance, median household income national quartile, hospital size, hospital teaching status, and Elixhauser comorbidity score. Fifth, we created Kaplan‐Meier curves after truncating the data at 270‐days to obtain maximum follow‐up along with the HR and 95% CI, which were presented as graphs. We then tested the effect modification by age groups and magnitude of comorbidity at the time of index event in the risk of rehospitalization and all bleeding at 90‐days associated with sex along with P for interaction by adding an interactive term between sex and the interaction terms of age group and Elixhauser score categories. Analysis was performed using STATA 14.1 (StataCorp LP, College Station, Texas; 2009). Any frequency or count <10 were not presented as per the restrictions and the data user agreements with the AHRQ. Statistical significance for primary outcomes at 90‐days was P < 0.025.
3. RESULTS
3.1. Baseline characteristics
The characteristics of the 522 521 patients hospitalized for AF in the NRD and included in our study sample are shown in Table 1. There were 278 410 women (53.3%) and 244 111 (46.7%) men included in our study. The mean age of individuals in our sample was 72.8 years, and 76.6% had a moderate or higher AHRQ severity of illness. Of those with an index admission for AF, 13% were admitted electively and almost 3 out of 5 patients with AF were hospitalized at a large hospital center. Women admitted for AF were significantly older (75.6 vs 69.6, P < 0.0001) and carried a greater comorbid disease burden (3.31 vs 3.00, P < 0.0001) than did men. Women also had higher risk for mortality (1.94 vs 1.88, P < 0.0001) and severity of illness (2.10 vs 2.05, P < 0.0001), compared to men.
Table 1.
Baseline patient and procedural characteristics
| Total cohort | Female | Male | P | |
|---|---|---|---|---|
| N (%) | 522 521 (100) | 278 410 (53.3) | 244 111 (46.7) | — |
| Age, mean (SE) | 72.8 (0.1) | 75.6 (0.1) | 69.6 (0.1) | <0.0001 |
| Age, n (%) | — | — | — | <0.0001 |
| 50‐64 | 131 799 (25.2) | 46 640 (16.8) | 85 159 (34.9) | — |
| 65‐74 | 147 890 (28.3) | 72 406 (26.0) | 75 494 (30.9) | — |
| 75‐90 | 242 822 (46.5) | 159 364 (57.2) | 83 458 (34.2) | — |
| Location, n (%) | — | — | — | 0.004 |
| Central metro (>1 m) | 109 769 (21.0) | 59 455 (21.4) | 50 314 (20.7) | — |
| Fringe metro (>1 m) | 140 435 (26.9) | 74 079 (26.6) | 66 356 (27.3) | — |
| Metro (250 k‐1 m) | 108 588 (20.8) | 57 469 (20.7) | 51 120 (21.0) | — |
| Micropolitan | 87 084 (31.3) | 87 084 (31.3) | 75 699 (31.1) | — |
| Insurance, n (%) | — | — | — | <0.0001 |
| Private/Medicare | 480 521 (92.0) | 262 421 (94.3) | 218 099 (89.4) | — |
| Medicaid/self pay/no charge/other | 41 625 (8.0) | 15 843 (5.7) | 25 782 (10.6) | — |
| Median household Income national quartile, n (%) |
— | — | — | <0.0001 |
| $1–$37 999 | 131 958 (25.6) | 71 875 (26.2) | 60 083 (25.0) | — |
| $38 000–$47 999 | 143 510 (27.9) | 77 564 (28.2) | 65 946 (27.5) | — |
| $48 000–$63 999 | 126 027 (24.5) | 66 586 (24.2) | 59 441 (24.8) | — |
| > = $64 000 | 113 244 (22.0) | 58 703 (21.4) | 54 541 (22.7) | — |
| Elective admission, n (%) | 67 540 (12.9) | 29 624 (10.7) | 37 916 (15.6) | <0.0001 |
| AHRQ risk of mortality, mean (SE) | 1.91 (<0.01) | 1.94 (<0.01) | 1.88 (<0.01) | <0.0001 |
| AHRQ severity of illness, mean(SE) | 2.08 (<0.01) | 2.10 (<0.01) | 2.05 (<0.01) | <0.0001 |
| AHRQ severity of illness, n (%) | — | — | — | <0.0001 |
| Minor | 122 426 (23.4) | 59 922 (21.5) | 62 504 (25.6) | — |
| Moderate | 249 497 (47.7) | 74 962 (49.0) | 113 058 (46.3) | — |
| Major | 136 737 (26.2) | 74 962 (26.9) | 61 775 (25.3) | — |
| Extreme | 13 854 (2.7) | 7080 (2.5) | 6774 (2.8) | — |
| AHRQ comorbidities, mean (SE) | 3.17 (0.01) | 3.31 (0.01) | 3.00 (0.01) | <0.0001 |
| Elixhauser score, n (%) | — | — | — | <0.0001 |
| 0/1 | 101 439 (19.4) | 46 230 (16.6) | 55 209 (22.6) | — |
| 2 | 110 481 (21.1) | 56 903 (20.4) | 53 578 (21.9) | — |
| 3 | 108 248 (20.7) | 59 451 (21.4) | 48 797 (20.0) | — |
| 4+ | 202 353 (38.7) | 115 826 (41.6) | 86 527 (35.4) | — |
| Hospital characteristics | ||||
| Bed size, n (%) | — | — | — | <0.0001 |
| Small | 78 259 (15.0) | 43 621 (15.7) | 34 638 (14.2) | — |
| Medium | 137 389 (26.3) | 74 016 (26.6) | 63 373 (26.0) | — |
| Large | 306 873 (58.7) | 160 773 (57.7) | 146 100 (59.8) | — |
| Teaching status, n (%) | — | — | — | <0.0001 |
| Metro, non‐teaching | 186 839 (35.8) | 101 932 (36.6) | 84 907 (34.8) | — |
| Metro, teaching | 266 029 (50.9) | 136 933 (49.2) | 129 096 (52.9) | — |
| Non‐metro | 69 653 (13.3) | 39 545 (14.2) | 30 108 (12.3) | — |
| Urban hospital, n (%) | 264 302 (50.6) | 139 702 (50.2) | 124 600 (51.0) | 0.011 |
All values are weighted frequencies and percentages except first line of age, which is weighted mean and SE. P‐value is derived from χ 2 test for all comparisons except comparison of mean and SE of age, which is tested using survey linear regression.
3.2. Index hospitalization characteristics
Patient clinical and procedural characteristics during their index hospitalization are presented in Table 2. Very few patients hospitalized for AF suffered thromboembolic complications (1.5%) or bleeding (1.3%) during their index AF admission. Women had relatively higher rates of bleeding (1.4% vs 1.1%, P < 0.0001) and thrombotic events (1.7% vs 1.4%, P < 0.0001) during index AF hospitalizations compared to men.
Table 2.
Patient, clinical, and procedural characteristics at index hospitalization
| Event | Total cohort | Female | Male | P |
|---|---|---|---|---|
| N (%) | 522 521 (100) | 278 410 (53.3) | 244 111 (46.7) | — |
| Any bleeding, n (%) | 6597 (1.3) | 3958 (1.4) | 2639 (1.1) | <0.0001 |
| Anemia, n (%) | 4413 (0.8) | 2680 (1.0) | 1733 (0.7) | <0.0001 |
| Intracranial hemorrhage, n (%) | 168 (<0.1) | 109 (<0.1) | 59 (<0.1) | 0.056 |
| GI bleed, n (%) | 2050 (0.4) | 1186 (0.4) | 864 (0.4) | 0.013 |
| Any thrombotic event, n (%) | 8011 (1.5) | 4599 (1.7) | 3413 (1.4) | <0.0001 |
| Ischemic stroke, n (%) | 1779 (0.3) | 1025 (0.4) | 754 (0.3) | 0.030 |
| Transient ischemic attack (TIA), n (%) | 2963 (0.6) | 1777 (0.6) | 1186 (0.5) | <0.0001 |
| Deep vein thrombosis (DVT), n (%) | 792 (0.2) | 419 (0.2) | 374 (0.2) | 0.87 |
| Pulmonary embolism (PE), n (%) | 2738 (0.5) | 1496 (0.5) | 1242 (0.5) | 0.39 |
All values are weighted frequencies and percentages. P‐value is derived from χ 2 test.
Any bleeding = anemia, intracranial hemorrhage, GI bleed or transfusion. There are <10 events for transfusion and therefore not reported. Any thrombotic event = ischemic stroke, TIA, DVT, and PE.
3.3. Risk of rehospitalization
A total of 108 952 (20.9%) patients were readmitted within 90‐days after an index AF hospitalization and 8105 patients (1.6%) died during the rehospitalization within 90‐days (Table 3 ). At 90‐days, the risk of thrombotic events was 20% greater among women than men (HR = 1.20, 95% CI = 1.09‐1.32), while the risk of all‐cause rehospitalization was 7% greater (HR = 1.07, 95% CI = 1.04‐1.09). Of the bleeding outcomes examined, GI bleed (0.6%) and any—bleeding (1.1%), as well as was the most common reason for rehospitalization at 90 days, but not different among men and women. There was increased risk of IS (HR = 1.31, 95% CI = 1.14‐1.51) and PE (HR = 1.21, 95% CI = 1.01‐1.45) among women than men. However, women had significantly lower rates of fatal rehospitalization after adjusting for age (HR = 0.91, 95% CI = 0.84‐0.98); however, without adjusting for age was HR = 1.34, 95% CI = 1.25‐1.44) (not shown in tables).
Table 3.
Clinical outcomes
| Type of rehospitalization | Total, n (%) | Female, n (%) | Male, n (%) | Age‐adjusted,HR (95%CI) | P | MV, HR (95% CI) | P |
|---|---|---|---|---|---|---|---|
| At 30 days | |||||||
| All cause fatal | 3983 (0.8) | 2365 (0.8) | 1618 (0.7) | 0.91 (0.81‐1.01) | 0.084 | 0.88 (0.78‐0.98) | 0.022 |
| Ischemic stroke | 1121 (0.2) | 770 (0.3) | 351 (0.1) | 1.35 (1.10‐1.65) | 0.004 | 1.34 (1.09‐1.64) | 0.005 |
| TIA | 474 (0.1) | 281 (0.1) | 192 (0.1) | 0.92 (0.69‐1.24) | 0.60 | 0.91 (0.68‐1.23) | 0.55 |
| Deep vein thrombosis | 316 (0.1) | 193 (0.1) | 123 (0.1) | 1.01 (0.70‐1.46) | 0.96 | 1.00 (0.69‐1.47) | 0.99 |
| Pulmonary embolism | 902 (0.2) | 587 (0.2) | 315 (0.1) | 1.24 (0.97‐1.57) | 0.080 | 1.19 (0.94‐1.52) | 0.15 |
| Anemia | 926 (0.2) | 589 (0.2) | 336 (0.1) | 1.15 (0.91‐1.46) | 0.25 | 1.11 (0.87‐1.41) | 0.39 |
| Intracranial hemorrhage | 191 (<0.1) | 106 (<0.1) | 86 (<0.1) | 0.81 (0.51‐1.27) | 0.36 | 0.78 (0.49‐1.25) | 0.31 |
| GI bleed | 1617 (0.3) | 929 (0.3) | 688 (0.3) | 0.87 (0.74‐1.02) | 0.096 | 0.83 (0.71‐0.97) | 0.023 |
| Any bleed | 2701 (1.0) | 3277 (1.2) | 1996 (0.8) | 1.06 (0.96‐1.16) | 0.27 | 1.02 (0.93‐1.12) | 0.62 |
| Thrombotic or fatal | 6325 (1.2) | 3895 (1.4) | 2430 (1.0) | 1.01 (0.92‐1.10) | 0.88 | 0.98 (0.90‐1.07) | 0.68 |
| Thrombotic event | 2704 (0.5) | 1760 (0.6) | 944 (0.4) | 1.19 (1.03‐1.36) | 0.016 | 1.16 (1.01‐1.34) | 0.032 |
| All‐cause | 61 713 (11.8) | 38 127 (13.7) | 23 585 (9.7) | 1.08 (1.05‐1.11) | <0.0001 | 1.05 (1.02‐1.08) | 0.002 |
| At 60 days | |||||||
| All cause fatal | 6360 (1.2) | 3802 (1.4) | 2559 (1.0) | 0.92 (0.85‐1.01) | 0.066 | 0.88 (0.81‐0.96) | 0.004 |
| Ischemic stroke | 1856 (0.4) | 1245 (0.4) | 612 (0.3) | 1.25 (1.08‐1.47) | 0.004 | 1.23 (1.05‐1.44) | 0.010 |
| TIA | 748 (0.1) | 457 (0.2) | 291 (0.1) | 1.01 (0.79‐1.29) | 0.94 | 1.01 (0.78‐1.30) | 0.94 |
| Deep vein thrombosis | 467 (0.1) | 294 (0.1) | 173 (0.1) | 1.10 (0.80‐1.53) | 0.55 | 1.10 (0.79‐1.52) | 0.58 |
| Pulmonary embolism | 1312 (0.3) | 845 (0.3) | 467 (0.2) | 1.21 (0.99‐1.47) | 0.062 | 1.16 (0.95‐1.41) | 0.15 |
| Anemia | 1573 (0.3) | 967 (0.3) | 606 (0.2) | 1.05 (0.88‐1.24) | 0.61 | 1.00 (0.84‐1.19) | 0.98 |
| Intracranial hemorrhage | 345 (0.1) | 201 (0.1) | 144 (0.1) | 0.90 (0.62‐1.29) | 0.58 | 0.90 (0.62‐1.31) | 0.60 |
| GI bleed | 2607 (0.5) | 1590 (0.6) | 1017 (0.4) | 1.01 (0.89‐1.16) | 0.83 | 0.96 (0.85‐1.10) | 0.60 |
| Any bleed | 4460 (0.9) | 2715 (1.0) | 1745 (0.7) | 1.01 (0.91‐1.12) | 0.81 | 0.97 (0.88‐1.07) | 0.56 |
| Thrombotic or fatal | 10 024 (1.9) | 6187 (2.2) | 3837 (1.6) | 1.02 (0.95‐1.09) | 0.64 | 0.98 (0.91‐1.05) | 0.52 |
| Thrombotic event | 4207 (0.8) | 2731 (1.0) | 1477 (0.6) | 1.18 (1.06‐1.32) | 0.002 | 1.16 (1.04‐1.29) | 0.008 |
| All‐cause | 90 132 (17.2) | 55 864 (20.1) | 34 269 (14.0) | 1.10 (1.07‐1.12) | <0.0001 | 1.06 (1.04–1.09) | <0.0001 |
| At 90 days | |||||||
| All cause fatal | 8105 (1.6) | 4899 (1.8) | 3207 (1.3) | 0.95 (0.89‐1.03) | 0.22 | 0.91 (0.84‐0.98) | 0.015 |
| Ischemic stroke | 2383 (0.5) | 1626 (0.6) | 756 (0.3) | 1.35 (1.17‐1.55) | <0.0001 | 1.31 (1.14‐1.51) | <0.0001 |
| TIA | 956 (0.2) | 582 (0.2) | 374 (0.2) | 0.99 (0.80‐1.24) | 0.97 | 0.99 (0.79‐1.24) | 0.95 |
| Deep vein thrombosis | 613 (0.1) | 387 (0.1) | 226 (0.1) | 1.12 (0.85‐1.46) | 0.43 | 1.10 (0.83‐1.45) | 0.51 |
| Pulmonary embolism | 1593 (0.3) | 1041 (0.4) | 551 (0.2) | 1.26 (1.05‐1.51) | 0.012 | 1.21 (1.01‐1.45) | 0.037 |
| Anemia | 2110 (0.4) | 1296 (0.5) | 813 (0.3) | 1.04 (0.90‐1.21) | 0.56 | 1.00 (0.86‐1.16) | 0.99 |
| Intracranial hemorrhage | 457 (0.1) | 267 (0.1) | 191 (0.1) | 0.91 (0.67‐1.25) | 0.57 | 0.91 (0.66‐1.26) | 0.58 |
| GI bleed | 3330 (0.6) | 2017 (0.7) | 1313 (0.5) | 1.00 (0.89‐1.13) | 0.96 | 0.96 (0.85‐1.08) | 0.46 |
| Any bleed | 5768 (1.1) | 3499 (1.3) | 2269 (0.9) | 1.01 (0.92–1.10) | 0.86 | 0.97 (0.88‐1.06) | 0.47 |
| Thrombotic or fatal | 12 665 (2.4) | 7882 (2.8) | 4784 (2.0) | 1.04 (0.98‐1.11) | 0.15 | 1.01 (0.95‐1.07) | 0.85 |
| Thrombotic event | 5304 (1.0) | 3483 (1.3) | 1821 (0.7) | 1.23 (1.12‐1.35) | <0.0001 | 1.20 (1.09–1.32) | <0.0001 |
| All‐cause | 108 952 (20.9) | 67 442 (24.2) | 41 510 (17.0) | 1.10 (1.08‐1.13) | <0.0001 | 1.07 (1.04–1.09) | <0.0001 |
| At 270 days | |||||||
| All cause fatal | 12 920 (2.5) | 7839 (2.8) | 5081 (2.1) | 0.97 (0.92‐1.03) | 0.32 | 0.93 (0.87‐0.98) | 0.013 |
| Ischemic stroke | 4107 (0.8) | 2860 (1.0) | 1248 (0.5) | 1.45 (1.30‐1.61) | <0.0001 | 1.41 (1.27‐1.57) | <0.0001 |
| TIA | 1881 (0.4) | 1193 (0.4) | 688 (0.3) | 1.12 (0.96‐1.31) | 0.16 | 1.11 (0.94‐1.30) | 0.21 |
| Deep vein thrombosis | 918 (0.2) | 615 (0.2) | 303 (0.1) | 1.34 (1.07‐1.69) | 0.011 | 1.31 (1.04‐1.65) | 0.021 |
| Pulmonary embolism | 2278 (0.4) | 1489 (0.5) | 789 (0.3) | 1.26 (1.09‐1.46) | 0.002 | 1.23 (1.06–1.43) | 0.007 |
| Anemia | 3613 (0.7) | 2221 (0.8) | 1392 (0.6) | 1.05 (0.94‐1.17) | 0.42 | 1.00 (0.90‐1.11) | 0.99 |
| Intracranial hemorrhage | 978 (0.2) | 590 (0.2) | 388 (0.2) | 1.01 (0.82‐1.25) | 0.91 | 0.97 (0.78‐1.20) | 0.79 |
| GI bleed | 5351 (1.0) | 3266 (1.2) | 2085 (0.9) | 1.02 (0.93‐1.12) | 0.67 | 0.98 (0.89‐1.07) | 0.61 |
| Any bleed | 9617 (1.8) | 5874 (2.1) | 3743 (1.5) | 1.03 (0.96‐1.10) | 0.47 | 0.98 (0.91‐1.05) | 0.58 |
| Thrombotic or fatal | 20 380 (3.9) | 12 797 (4.6) | 7583 (3.1) | 1.08 (1.03‐1.13) | 0.002 | 1.04 (0.99‐1.09) | 0.11 |
| Thrombotic event | 8732 (1.7) | 5843 (2.1) | 2889 (1.2) | 1.31 (1.22‐1.40) | <0.0001 | 1.28 (1.19‐1.37) | <0.0001 |
| All‐cause | 152 262 (29.1) | 93 482 (33.7) | 58 421 (23.9) | 1.11 (1.09‐1.13) | <0.0001 | 1.07 (1.05–1.09) | <0.0001 |
Abbreviations: CI, confidence intervals; GI, gastrointestinal bleeding; HR, hazard ratios; TIA, transient ischemic event.
Survey‐weighted cox proportional hazard regression model was used to estimate hazard ratios (HR) and 95% confidence intervals (95% CI). MV denotes multivariable model was additionally adjusted for location, household income, and elixhauser comorbidity score. All HRs are reported with males as the reference value. All‐cause rehospitalization include both fatal and non‐fatal rehospitalizations. Both arterial thrombosis and bleeding requiring transfusion have events <10 and are not presented here. For composite events, such as any bleed and any bleed or fatal are computed using the first of the events. A thrombotic event is a composite of ischemic stroke, TIA, DVT or PE. Any Bleeding event is a composite of anemia, intracranial hemorrhage, GI bleed, or transfusion.
The risk of thrombotic events and IS at 30, 60, and 270‐days were similar to that at 90‐days. The increased risk of PE among women was not evident during 30 or 60‐days of follow‐up, while the increased risk extended to 270‐days (HR = 1.23, 95% CI = 1.06‐1.43). The increased risk of all‐cause rehospitalization among women than men persisted throughout the follow‐up until 270‐days. Kaplan‐Meier curves examining rehospitalization events up to 270 days are presented in Figure 1, Figures S2‐S11.
Figure 1.
A, Kaplan‐Meier curves for thrombotic events (ischemic stroke, transient ischemic attack, deep vein thrombosis or pulmonary embolism). B, Kaplan‐Meier curves for any rehospitalization (fatal or non‐fatal)
3.4. Stratified analysis
We observed significant effect modification by age group (P‐interaction <0.0001) and by magnitude of comorbidity (P‐interaction = 0.0002) for the risk of all‐cause rehospitalizations at 90‐days for women as compared to men (Figure 2 ). Rehospitalizations for any bleeding event also showed significant effect modification by age group (P = 0.002). There was no significant interaction by age (P‐interaction = 0.82) and Elixhauser comorbidity (P‐interaction = 0.13) for the increased risk of thrombotic events among women as compared to men.
Figure 2.
Stratified analysis at 90 days
4. DISCUSSION
Our analysis sought to assess the risk of continued real‐world morbidity in terms of rehospitalization among survivors of an AF‐related hospitalization and investigate sex‐differences in all‐cause rehospitalization, thrombotic rehospitalization, and bleeding related rehospitalization using a nationally represented claims‐based inpatient readmission database comprised of more than 70 million patients. We report two primary findings. First, we observed that rehospitalizations after discharge from index AF hospitalization were common, with more than 20% of AF survivors requiring readmission by 90‐days post‐discharge. Second, we observed that women were at significantly higher risk for post‐discharge thrombotic complications than men that resulted in rehospitalization.
AF is a common cause of hospital readmission,11 with prior large‐scale cohort studies suggesting that up to 15% of patients hospitalized for AF are readmitted within 30 days of hospital discharge.22 In our study, we observed that rehospitalizations were common among hospital survivors of AF, with fatal or non‐fatal bleeding‐related hospitalizations occurring more commonly than fatal or non‐fatal stroke or thrombotic event related hospitalizations at 90‐ and 270‐days after discharge. Moreover, hospital readmissions for AF are expensive and cause significant utilization of hospital resources.23 In a nationwide study of all Medicare beneficiaries between 1999 and 2013, it was found that rates of hospitalization for AF increased by approximately 1% per year between 1999 and 2013, and although geographic variation was noted, this trend was consistent nationwide. Median hospital length of stay remained unchanged, but median Medicare inpatient expenditure per beneficiary increased by 60%, to $ 4719 per stay.23 This emphasizes that AF poses a significant financial burden to institutions and hospital readmission rates are often used as a metric for hospital quality and can be associated with financial penalties.24 Our findings also suggest that timely follow‐up in the outpatient setting may be warranted for AF patients discharged from the hospital, as these patients may benefit from close surveillance and targeted interventions, such as initiation or changes to dosing of oral anticoagulants, to reduce thromboembolic and bleeding‐related rehospitalizations.11
As has been observed in community‐based cohorts,25, 26, 27 we found that women were at a higher risk for ischemic stroke than men. However, the total number of ischemic stroke events that occurred among women during the 90 days after hospital discharge was relatively small (n = 1626 [0.6%]). Women were also at a higher risk than men for thrombotic events that included ischemic stroke, TIA, DVT, or PE. Notably, the number of thrombotic hospitalizations in women was almost twice the number of thromboembolic events, with nearly 2.1% of women experiencing such a rehospitalization (n = 5843).
Although treatment data, including information on the relative proportion of men and women treated with anticoagulants, was not available in the NRD, bleeding complications likely reflect complications from AF treatment with antiplatelet agents and/or oral anticoagulants.17 Although women were at higher risk of stroke than were men, women were at significantly greater near‐term risk for bleeding than of stroke, as has been noted in prior community‐based cohort studies and clinical trials.28, 29 This finding may be explained in part by the fact that women with AF tended to be older and frailer (based on their comorbidity scores) than men with AF in the NRD, putting them at increased risk for bleeding from oral anticoagulants.
Recent European Society of Cardiology (ESC) guidelines recommend that providers use the CHA2DS2‐VASc risk score to identify men and women with AF for whom oral anticoagulation is appropriate, considering the patients' risk of stroke.27, 30, 31, 32 Although this risk calculator incorporates sex as a risk factor, current guidelines do not recommend assigning equal weight to sex as a contributor to thromboembolic risk. Nor do they comment on the appropriate timing of anticoagulation initiation or timing of follow‐up for hospitalized patients with AF. ESC guidelines also recommend assessment of bleeding risk in AF using the HAS‐BLED bleeding risk score as a simple, easy calculation, whereby a score of ≥3 indicates “high risk”, indicating some caution and review of the patient is needed.30, 31 However, the HAS‐BLED score does not account for sex‐related difference in bleeding risk, and unlike the CHA2DS2‐VASc score, does not have gender as an independent risk factor.30
Our findings add a further dimension to the existing guidelines by suggesting that initiation of anticoagulation at discharge may cause greater harm than good, particularly for women with AF. Since acute cardiovascular hospitalizations are frequently complicated by end‐organ dysfunction that can affect drug metabolism, that is, alterations in hepatic or renal clearance, hospitalized patients with AF may be particularly susceptible to anticoagulation‐related side effects.33 Our findings support the need for enhanced efforts by clinicians and hospital systems to address “hand‐offs” among patients discharged after an AF‐related hospitalization, particularly for older, female patients with AF. In particular, our results suggest that conventional stroke and bleeding risk assessment should be combined with a careful and individualized assessment of factors that might impact transitional care, that is, whether or not the patient is scheduled for an international normalized ratio (INR) test and evaluation of hepatic and renal clearance within 2 weeks of discharge. Further research with gender‐specific result reporting is needed to identify the patient, physician, and system‐level factors associated with successful transition from hospital to home among men and women with AF discharged on oral anticoagulation.
There are several notable strengths to this study. First, it includes demographic and administrative hospital data from across the nation and multiple hospital sites. Second, our sample includes a large racially and geographically diverse sample from across the United States. Third, we used a rigorous statistical protocol adjusting for multiple confounding factors. As we analyzed a fairly large sample size, our study was sufficiently powered to detect differences in our primary and secondary outcomes.
Our study findings should be interpreted in the context of some limitations. First, we used claims data to define participant characteristics and outcomes. Although our data included all rehospitalizations related to thromboembolic and bleeding events, we could not ascertain for these events outside of the context of readmission. Although ICD‐9 related codes are validated and events were adjudicated and coded by clinicians, their use introduces the possibility of misclassification bias. Second, although the patient‐level data is weighted to allow for national estimates, the sample size to allow intent‐level analysis remains small. Third, the lack of hospital identifiers precluded clustering by institution and may have introduced inter‐hospital variability. Fourth, separate outcomes data for women and men stratified by CHA2DS2‐VASc scores could not be analyzed due to lack of data. Last, we were unable to assess information, such as medications, laboratory values, and changes in medications due to intermediate events that may have occurred and affected the following events.
5. CONCLUSION
Patients admitted with a diagnosis of AF frequently experience thromboembolic complications after hospital discharge. Our study found that women with AF are older and carry a heavier burden of co‐morbidities, than men hospitalized with AF. Even after adjusting for comorbid diseases, we observed higher rates of thrombosis‐related complications after discharge among women admitted with AF compared to men. Further longitudinal studies are warranted to determine the optimal timing and processes of care associated with initiating oral anticoagulation among hospitalized patients with AF, particularly among older women.
CONFLICTS OF INTEREST
The authors declare no potential conflict of interests.
Supporting information
Table S1 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) codes used
Figure S1: Flow chart of inclusion and exclusion criteria
Figure S2: Kaplan‐Meier curves for fatal rehospitalization
Figure S3: Kaplan‐Meier curves for ischemic stroke
Figure S4: Kaplan‐Meier curves for TIA
Figure S5: Kaplan‐Meier curves for deep vein thrombosis
Figure S6: Kaplan‐Meier curves for pulmonary embolism
Figure S7: Kaplan Meier curves anemia
Figure S8: Kaplan‐Meier curves intracranial hemorrhage
Figure S9: Kaplan‐Meier curves gastrointestinal bleeding
Figure S10: Kaplan‐Meier curves for any thrombotic+fatal rehospitalization
Figure S11: Kaplan‐Meier curves for bleeding events (anemia, intracranial hemorrhage, GI bleed, or transfusion)
ACKNOWLEDGMENTS
This study was supported by the following grants: 1R01HL126911‐01A1 (DDM), KL2RR031981 (DDM). B. K. and E. P. are supported by Evans Research Foundation.
Kalesan B, Kundu A, Vaze A, et al. Sex‐differences in post‐discharge outcomes among patients hospitalized for atrial fibrillation. Clin Cardiol. 2019;42:84–92. 10.1002/clc.23111
Funding information Edward P. Evans Foundation; NIH Blueprint for Neuroscience Research, Grant/Award Number: 1R01HL126911‐01A1 KL2RR031981
REFERENCES
- 1. Miyasaka Y, Barnes ME, Gersh BJ, et al. Secular trends in incidence of atrial fibrillation in Olmsted County, Minnesota, 1980 to 2000, and implications on the projections for future prevalence. Circulation. 2006;114(2):119‐125. [DOI] [PubMed] [Google Scholar]
- 2. Poli D, Antonucci E. Epidemiology, diagnosis, and management of atrial fibrillation in women. Int J Women's Health. 2015;7:605‐614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Cairns JA, Connolly SJ. Nonrheumatic atrial fibrillation. Risk of stroke and role of antithrombotic therapy. Circulation. 1991;84(2):469‐481. [DOI] [PubMed] [Google Scholar]
- 4. Simons LA, McCallum J, Friedlander Y, Simons J. Risk factors for ischemic stroke. Stroke. 1998;29(7):1341‐1346. [DOI] [PubMed] [Google Scholar]
- 5. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham study. Stroke. 1991;22(8):983‐988. [DOI] [PubMed] [Google Scholar]
- 6. Wang TJ, Larson MG, Levy D, et al. Temporal relations of atrial fibrillation and congestive heart failure and their joint influence on mortality. Circulation. 2003;107(23):2920‐2925. [DOI] [PubMed] [Google Scholar]
- 7. Krahn AD, Manfreda J, Tate RB, Mathewson FAL, Cuddy TE. The natural history of atrial fibrillation: incidence, risk factors, and prognosis in the Manitoba follow‐up study. Am J Med. 1995;98(5):476‐484. [DOI] [PubMed] [Google Scholar]
- 8. Benjamin EJ, Wolf PA, D'Agostino RB, Silbershatz H, Kannel WB, Levy D. Impact of atrial fibrillation on the risk of death. Circulation. 1998;98(10):946‐952. [DOI] [PubMed] [Google Scholar]
- 9. Stewart S, Hart CL, Hole DJ, McMurray JJV. A population‐based study of the long‐term risks associated with atrial fibrillation: 20‐year follow‐up of the Renfrew/Paisley study. Am J Med. 2002;113(5):359‐364. [DOI] [PubMed] [Google Scholar]
- 10. Vidaillet H, Granada JF, Po‐H C, et al. A population‐based study of mortality among patients with atrial fibrillation or flutter. Am J Med. 2002;113(5):365‐370. [DOI] [PubMed] [Google Scholar]
- 11. Kundu A, O'Day K, Shaikh AY, et al. Relation of atrial fibrillation in acute myocardial infarction to in‐hospital complications and early hospital readmission. Am J Cardiol. 2016;117(8):1213‐1218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Kim MH, Johnston SS, Chu BC, Dalal MR, Schulman KL. Estimation of total incremental health care costs in patients with atrial fibrillation in the United States. Circ Cardiovasc Qual Outcomes. 2011;4(3):313‐320. [DOI] [PubMed] [Google Scholar]
- 13. Chugh SS, Havmoeller R, Narayanan K, et al. Worldwide epidemiology of atrial fibrillation: a global burden of disease 2010 study. Circulation. 2014;129(8):837‐847. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Heeringa J, van der Kuip DAM, Hofman A, et al. Prevalence, incidence and lifetime risk of atrial fibrillation: the Rotterdam study. Eur Heart J. 2006;27(8):949‐953. [DOI] [PubMed] [Google Scholar]
- 15. Cheng EY, Kong MH. Gender differences of thromboembolic events in atrial fibrillation. Am J Cardiol. 2016;117(6):1021‐1027. [DOI] [PubMed] [Google Scholar]
- 16. Wagstaff AJ, Overvad TF, Lip GYH, Lane DA. Is female sex a risk factor for stroke and thromboembolism in patients with atrial fibrillation? A systematic review and meta‐analysis. QJM. 2014;107(12):955‐967. [DOI] [PubMed] [Google Scholar]
- 17. Volgman AS, Manankil MF, Mookherjee D, Trohman RG. Women with atrial fibrillation: greater risk, less attention. Gend Med. 2009;6(3):419‐432. [DOI] [PubMed] [Google Scholar]
- 18. https://www.hcup-us.ahrq.gov/db/nation/nrd/NRD_Introduction_2013.jsp
- 19. https://www.hcup‐us.ahrq.gov/db/nation/nrd/NRD_Introduction_2014.jsp.
- 20. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8‐27. [DOI] [PubMed] [Google Scholar]
- 21. https://www.hcup-us.ahrq.gov/db/nation/nis/OverviewofSeveritySystems.pdf.
- 22. Munir MB, Sharbaugh MS, Ahmad S, et al. Causes and predictors of 30‐day readmissions in atrial fibrillation (from the Nationwide readmissions database). Am J Cardiol. 2017;120(3):399‐403. [DOI] [PubMed] [Google Scholar]
- 23. Freeman JV, Wang Y, Akar J, et al. National Trends in atrial fibrillation hospitalization, readmission, and mortality for Medicare beneficiaries, 1999‐2013. Circulation. 2017;35(13):1227‐1239. [DOI] [PubMed] [Google Scholar]
- 24. Sardar P, Kundu A, Nairooz R, Chatterjee S, Ledley GS, Aronow WS. Health resource variability in the achievement of optimal performance and clinical outcome in ischemic heart disease. Curr Cardiol Rep. 2015;17(2):1‐7. [DOI] [PubMed] [Google Scholar]
- 25. Gibson CL. Cerebral ischemic stroke: is gender important? J Cereb Blood Flow Metab. 2013;33(9):1355‐1361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Seshadri S, Beiser A, Kelly‐Hayes M, et al. The lifetime risk of stroke: estimates From the Framingham Study. Stroke. 2006;37(2):345‐350. [DOI] [PubMed] [Google Scholar]
- 27. O'Brien EC, Kim S, Hess PL, et al. Effect of the 2014 atrial fibrillation guideline revisions on the proportion of patients recommended for oral anticoagulation. JAMA Intern Med. 2015;175(5):848‐850. [DOI] [PubMed] [Google Scholar]
- 28. Pancholy SB, Sharma PS, Pancholy DS, Patel TM, Callans DJ, Marchlinski FE. Meta‐analysis of gender differences in residual stroke risk and major bleeding in patients with Nonvalvular atrial fibrillation treated with Oral anticoagulants. Am J Cardiol. 2014;113(3):485‐490. [DOI] [PubMed] [Google Scholar]
- 29. Schneeweiss S, Gagne JJ, Patrick AR, Choudhry NK, Avorn J. Comparative efficacy and safety of new oral anticoagulants in patients with atrial fibrillation. Circ Cardiovasc Qual Outcomes. 2012;5(4):480‐486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Lip GYH. Implications of the CHA(2)DS(2)‐VASc and HAS‐BLED scores for thromboprophylaxis in atrial fibrillation. Am J Med. 2011;124(2):111‐114. [DOI] [PubMed] [Google Scholar]
- 31. Camm AJ, Lip GY, De Caterina R, et al. 2012 focused update of the ESC guidelines for the management of atrial fibrillation: an update of the 2010 ESC guidelines for the management of atrial fibrillation developed with the special contribution of the European heart rhythm association. Eur Heart J. 2012;33(21):2719‐2747. [DOI] [PubMed] [Google Scholar]
- 32. Kundu A, Day KO, Lessard DM, et al. Recent trends in Oral anticoagulant use and post‐discharge complications among atrial fibrillation patients with acute myocardial infarction. Journal of Atrial Fibrillation. 2018;10(5):1749. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Kundu A, Sardar P, Chatterjee S, Aronow WS, Owan T, Ryan JJ. Minimizing the risk of bleeding with NOACs in the elderly. Drugs Aging. 2016;33(7):491‐500. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) codes used
Figure S1: Flow chart of inclusion and exclusion criteria
Figure S2: Kaplan‐Meier curves for fatal rehospitalization
Figure S3: Kaplan‐Meier curves for ischemic stroke
Figure S4: Kaplan‐Meier curves for TIA
Figure S5: Kaplan‐Meier curves for deep vein thrombosis
Figure S6: Kaplan‐Meier curves for pulmonary embolism
Figure S7: Kaplan Meier curves anemia
Figure S8: Kaplan‐Meier curves intracranial hemorrhage
Figure S9: Kaplan‐Meier curves gastrointestinal bleeding
Figure S10: Kaplan‐Meier curves for any thrombotic+fatal rehospitalization
Figure S11: Kaplan‐Meier curves for bleeding events (anemia, intracranial hemorrhage, GI bleed, or transfusion)