Risk of falls and bleeding in elderly patients with acute venous thromboembolism

Authors


Abstract

Objective

Whether or not a high risk of falls increases the risk of bleeding in patients receiving anticoagulants remains a matter of debate.

Methods

We conducted a prospective cohort study involving 991 patients ≥65 years of age who received anticoagulants for acute venous thromboembolism (VTE) at nine Swiss hospitals between September 2009 and September 2012. The study outcomes were as follows: the time to a first major episode of bleeding; and clinically relevant nonmajor bleeding. We determined the associations between the risk of falls and the time to a first episode of bleeding using competing risk regression, accounting for death as a competing event. We adjusted for known bleeding risk factors and anticoagulation as a time-varying covariate.

Results

Four hundred fifty-eight of 991 patients (46%) were at high risk of falls. The mean duration of follow-up was 16.7 months. Patients at high risk of falls had a higher incidence of major bleeding (9.6 vs. 6.6 events/100 patient-years; = 0.05) and a significantly higher incidence of clinically relevant nonmajor bleeding (16.7 vs. 8.3 events/100 patient-years; < 0.001) than patients at low risk of falls. After adjustment, a high risk of falls was associated with clinically relevant nonmajor bleeding [subhazard ratio (SHR) = 1.74, 95% confidence interval (CI) = 1.23–2.46], but not with major bleeding (SHR = 1.24, 95% CI = 0.83–1.86).

Conclusion

In elderly patients who receive anticoagulants because of VTE, a high risk of falls is significantly associated with clinically relevant nonmajor bleeding, but not with major bleeding. Whether or not a high risk of falls is a reason against providing anticoagulation beyond 3 months should be based on patient preferences and the risk of VTE recurrence.

Introduction

Falls occur in 30–60% of elderly patients each year, and older patients have a higher susceptibility for fall-related injuries than younger patients [1]. Fear of fall-related bleeding is the most commonly cited reason for not providing anticoagulation to such patients [2, 3]. Whether or not a high risk of falls increases the risk of bleeding in elderly patients receiving anticoagulants is still a matter of debate. The few studies that have specifically addressed whether or not a high risk of falls is associated with bleeding in elderly patients receiving anticoagulants have been limited by a retrospective design [4-6], data derived from a single centre [5-7], and a failure to include clinically relevant nonmajor bleeding as an outcome [4, 6-8]. Moreover, these studies have focused mainly [5, 7, 8] or exclusively [4, 6] on patients with atrial fibrillation.

The incidence of acute venous thromboembolism (VTE) varies with age, increasing to 4.1 events per 1000 patient-years in patients ≥65 years of age [9]. The standard treatment for acute VTE (anticoagulation for ≥3 months) carries a risk of major bleeding of 2–3% per year, with the rate of major bleeding increasing approximately twofold in patients ≥65 years of age [10]. Although elderly patients with VTE may have a 40% higher incidence of bleeding than patients with atrial fibrillation [8], whether or not the risk of falls is associated with bleeding has never been determined in such patients.

Our aim was to determine whether or not the risk of falls is associated with major and clinically relevant nonmajor bleeding in a prospective multicentre cohort of elderly patients receiving anticoagulants for VTE. If such an association exists, a higher risk of falls could be a reason against extending anticoagulation beyond a duration of 3 months in elderly patients with VTE.

Methods

Cohort sample

The study was conducted between September 2009 and September 2012 as part of the Swiss cohort of elderly patients with venous thromboembolism (SWITCO65+), a prospective multicentre cohort study that assessed long-term medical outcomes and quality of life in elderly patients with acute VTE from all five university and four high-volume nonuniversity hospitals in Switzerland. Consecutive patients ≥65 years of age with an acute, objectively confirmed VTE were identified in the inpatient and outpatient services of all participating study sites. A detailed description of the study methods was published elsewhere [11]. The study was approved by the institutional review board at each participating centre.

Baseline data collection

Trained study nurses prospectively collected baseline demographics (age and gender), co-morbid conditions (history of major bleeding, recent major surgery, cerebrovascular disease, cardiac disease, diabetes mellitus, hypertension, active cancer, and chronic liver and renal diseases), laboratory findings (haemoglobin concentration and platelet count), concomitant antiplatelet therapy and VTE-related treatment (low molecular weight heparin, unfractionated heparin, fondaparinux and vitamin K antagonists) from all enrolled patients using standardized data collection forms. We assessed the risk of falls using two validated screening questions [12] as follows: (i) Did you fall during the last year? and (ii) If not, did you notice any problems with gait, balance, or mobility? Patients who had fallen in the past year have a positive likelihood ratio of future falls ranging from 2.3 to 2.8, whereas patients who had not fallen during the past year, but had noticed gait, balance or mobility problems, have a positive likelihood ratio of future falls ranging from 1.7 to 2.4 [12]. Patients who answered yes to at least one screening question were considered to be at high risk of falls. All other patients were considered to be at low risk of falls.

Study outcomes

The primary outcome for the current study was the time to the first major bleeding episode, which was defined as fatal bleeding, symptomatic bleeding in a critical site or organ (intracranial, intraspinal, intraocular, retroperitoneal, intra-articular, pericardial or intramuscular with compartment syndrome), bleeding with a reduction in haemoglobin concentration of ≥20 g L−1 or bleeding leading to a transfusion of ≥2 units of packed red blood cells [13]. Fatal bleeding was defined as any death that followed an intracranial haemorrhage or a bleeding episode leading to hemodynamic deterioration [14]. The secondary outcome for the current study was the time to the first clinically relevant nonmajor bleeding episode, which was defined as bleeding not meeting the definition of major bleeding, but requiring physician consultation or evaluation in the emergency department. Whether or not bleeding was considered fall-related was based on patient self-report, supplemented by information from the medical chart, if available.

Patient follow-up included one telephone interview and two surveillance face-to-face evaluations during the first year of participation in the study, then semi-annual contact alternating between face-to-face evaluations and telephone calls, as well as periodic reviews of the patient's hospital chart. During each visit/contact, study nurses interviewed patients to obtain information about the date, type and circumstances (i.e. following a fall) of bleeding episodes and assessed whether or not the patient had died. If a clinical event had occurred, supplemental information was obtained by reviewing medical charts and interviewing the patient's primary care physician and/or family members. We also collected international normalized ratio (INR) values throughout the period of follow-up. A committee of three blinded clinical experts adjudicated all outcomes and classified the cause of all deaths as definitely due to major bleeding or due to another cause; final classifications were made on the basis of the full consensus of this committee.

Statistical analysis

We calculated the overall incidence rates of the first major bleeding episode, a clinically relevant nonmajor bleeding episode and a fall-related bleeding episode in patients at low- versus high risk of falls. We compared the cumulative incidence of the first major bleeding episode, clinically relevant nonmajor bleeding episode and fall-related bleeding episode using the Kaplan–Meier technique and the log-rank test. In patients who received vitamin K antagonists within 14 days from the time of diagnosis, we also compared the percentage of time spent within one of three specified international INR ranges (<2.0, 2.0–3.0, >3.0) [15], excluding the first seven treatment days.

We examined the associations between the risk of falls and the time to the first major and clinically relevant nonmajor bleeding episodes using competing risk regression models, accounting for death as a competing event [16]. The strength of the association between the risk of falls and bleeding was reflected by the subhazard ratio (SHR), which is the ratio of hazards associated with the cumulative incidence function in the presence and absence of high risk of falls. We accounted for nonhaemorrhagic death as a competing event when analysing major bleeding and fall-related bleeding and for overall death when analysing clinically relevant nonmajor bleeding. We adjusted for risk factors that have been previously shown to be associated with major bleeding, including age, female gender, overt pulmonary embolism, history of major bleeding, recent major surgery, cerebrovascular disease, cardiac disease, diabetes mellitus, hypertension, active cancer, chronic liver disease, chronic renal disease, anaemia, low platelet count, concomitant antiplatelet therapy and periods of anticoagulation as a time-dependent covariate [17-34]. Because the risk of bleeding is higher in the early phase of anticoagulation [31], we excluded bleeding events occurring during the first 90 days following a VTE in a sensitivity analysis. We assumed missing values to be normal in the primary analyses, but we also tested the impact on study results by assuming missing values to be abnormal. All analyses were performed using Stata 12 (Stata Corporation, College Station, TX, USA).

Results

Study sample

Of 1863 screened patients ≥65 years of age with VTE, we excluded 462 patients who had ≥1 exclusion criterion and 398 who did not consent to participate (Fig. 1). After the exclusion of 12 patients who withdrew informed consent or in whom information on the falls risk was not available, our final study sample was comprised of 991 patients. The excluded patients were significantly older (median age, 78 vs. 75 years; < 0.001) and more likely to be women (59% vs. 47%; < 0.001) than the analysed patients.

Figure 1.

Patient flow chart.

Overall, 458 patients (46%) were at high risk of falls. Patients at high risk of falls were older and more likely to be female and to have a history of major bleeding, cerebrovascular disease, diabetes mellitus, hypertension and anaemia and to have received concomitant antiplatelet therapy than patients at low risk of falls (Table 1).

Table 1. Patient baseline characteristics
CharacteristicsbAll (n = 991)High risk of falls (n = 458)Low risk of falls (n = 533)P-valuea
Number (%) or median (range)
  1. VKA, vitamin K antagonist.

  2. a

    A Fisher's exact test was used for categorical variables, and a Wilcoxon rank-sum test was used for continuous variables.

  3. b

    Data were missing for history of major bleeding (0.1%), haemoglobin concentration (6.4%) and platelet count (6.4%).

  4. c

    Acute heart failure during the last 3 months, a known history of systolic or diastolic heart failure, left or right heart failure, forward or backward heart failure, left ventricular ejection fraction <40% or a myocardial infarction with or without ST elevation during the last 3 months.

  5. d

    Solid or haematologic cancer requiring chemotherapy, radiotherapy, surgery and/or palliative care during the last 3 months.

  6. e

    Serum haemoglobin concentration <13 g dL−1 for males or <12 g dL−1 for females.

  7. f

    Platelet count <150 000 μL−1.

  8. g

    Use of aspirin, clopidogrel, prasugrel or aspirin/dipyridamole.

Age, years75 (65–97)77 (65–97)73 (65–92)<0.001
Female gender463 (46.7)243 (53.1)220 (41.3)<0.001
Overt pulmonary embolism687 (69.3)324 (70.7)363 (68.1)0.41
History of major bleeding102 (10.3)59 (12.9)43 (8.1)0.02
Recent major surgery149 (15.0)65 (14.2)84 (15.8)0.53
Cerebrovascular disease93 (9.4)53 (11.6)40 (7.5)0.03
Cardiac diseasec119 (12.0)52 (11.4)67 (12.6)0.62
Diabetes mellitus155 (15.6)93 (20.3)62 (11.6)<0.001
Hypertension639 (64.5)316 (69.0)323 (60.6)0.006
Active cancerd178 (18.0)86 (18.8)92 (17.3)0.56
Chronic liver disease15 (1.5)6 (1.3)9 (1.7)0.80
Chronic renal disease187 (18.9)92 (20.1)95 (17.8)0.37
Anaemiae387 (39.1)207 (45.2)180 (33.8)0.002
Low platelet countf141 (14.2)67 (14.6)74 (13.9)1.00
Treatment
Platelet inhibitorsg323 (32.6)172 (37.6)151 (28.3)0.002
Pre-existing VKA therapy59 (6.0)29 (6.3)30 (5.6)0.69
Initial parenteral anticoagulation
Low molecular weight heparin468 (47.2)222 (48.5)246 (46.2)0.87
Unfractionated heparin330 (33.3)150 (32.8)180 (33.8) 
Fondaparinux159 (16.0)72 (15.7)87 (16.3) 
None34 (3.4)14 (3.1)20 (3.8) 
Initial VKA therapy846 (85.4)386 (84.3)460 (86.3)0.37

Comparison of bleeding rates

After a mean (±standard deviation) follow-up period of 16.7 (±9.5) months, 159 of 991 patients (16.0%) died and 103 (10.4%) experienced a first major bleeding episode, resulting in an overall incidence of 7.9 major bleed episodes per 100 patient-years. The characteristics of major bleeding episodes are shown in Table 2. Overall, 150 patients (15%) had a first clinically relevant nonmajor bleeding episode during the follow-up period, with an overall incidence of 12.0 nonmajor bleed episodes per 100 patient-years. Fifty patients (5.0%) had a fall-related first major or clinically relevant nonmajor bleeding episode, with an overall incidence of 3.7 fall-related bleed episodes per 100 patient-years.

Table 2. Description of first major bleeding events (N = 103)
Bleeding characteristicsaNumber (%)
  1. a

    More than one characteristic may apply.

  2. b

    Defined as an INR > 3.0.

Fatal6 (5.8)
Intracranial11 (10.7)
Intraspinal1 (1.0)
Intraocular1 (1.0)
Pericardial1 (1.0)
Retroperitoneal4 (3.9)
Intra-articular4 (3.9)
Intramuscular with compartment syndrome2 (1.9)
Gastrointestinal39 (37.9)
Musculocutaneous31 (30.1)
Urogenital10 (9.7)
Epistaxis2 (1.9)
Unknown localization2 (1.9)
Fall in haemoglobin concentration ≥20 g L−166 (64.1)
Transfusion of ≥2 units of packed red blood cells54 (52.4)
Excessive anticoagulation at the time of bleedingb29 (28.2)
Following a fall15 (14.6)

Patients at high risk of falls had a higher incidence of major bleeding episodes than patients at low risk of falls (9.6 vs. 6.6 bleed episodes per 100 patient-years; exact = 0.05). Although patients at high risk of falls were not more likely to be over-anticoagulated (INR > 3.0) at the time of major bleeding (32.3% vs. 33.3%; = 0.93), they were significantly more likely to have a fall-related major bleeding episode (21.4% vs. 6.4%; = 0.03).

Patients at high risk of falls had a significantly higher incidence of clinically relevant nonmajor bleeding episodes than low-risk patients (16.7 vs. 8.3 bleed episodes per 100 patient-years; exact < 0.001). The proportion of patients with clinically relevant nonmajor bleeding episodes who were over-anticoagulated at the time of bleeding (38.6% vs. 26.1%; = 0.30) or whose bleeding episodes were fall-related (23.1% vs. 20.3%; = 0.69) were similar. Patients at high risk of falls had a higher incidence of fall-related major or clinically relevant nonmajor bleeding episodes (5.5 vs. 2.3 bleed episodes per 100 patient-years; exact = 0.003). The 2-year cumulative incidence of bleeding episodes by falls risk is shown in Fig. 2 (panels a–c).

Figure 2.

(a) Kaplan–Meier estimates of the first major bleeding episode by risk of falls. The 2-year cumulative incidence of the first major bleeding episode was 13.5% for patients with a high risk of falls versus 10.9% for patients with a low risk of falls (P = 0.06 by the log-rank test). (b) Kaplan–Meier estimates of the first clinically relevant nonmajor bleeding episode by risk of falls. The 2-year cumulative incidence of the first clinically relevant nonmajor bleeding episode was 23.2% for patients with a high risk of falls versus 13.6% for patients with a low risk of falls (P < 0.001 by the log-rank test). (c) Kaplan–Meier estimates of the first fall-related major or clinically relevant nonmajor bleeding episode by risk of falls. The 2-year cumulative incidence of the first fall-related bleeding episode was 8.8% for patients with a high risk of falls versus 3.9% for patients with a low risk of falls (P = 0.002 by the log-rank test).

Amongst the 783 patients who received vitamin K antagonists within the first 14 days after VTE and in whom the percentage of time within a specified INR range could be calculated, patients spent 26.0%, 57.4% and 16.6% of time in a subtherapeutic (<2.0), therapeutic (2.0–3.0) and supra-therapeutic (>3.0) INR range, respectively. Patients at high risk of falls spent less time in the therapeutic INR range (54.0% vs. 60.2%; < 0.001) and more time in the subtherapeutic INR range than patients at low risk of falls (28.4% vs. 24.0%; = 0.003). Overall, patients at high risk of falls spent a nonsignificantly higher percentage of time in a supra-therapeutic INR range than patients at low risk of falls (17.5% vs. 15.8%; P = 0.15).

Association between risk of falls and bleeding

After adjustment, patients at high risk of falls had a somewhat higher risk of major bleeding, but the risk failed to achieve statistical significance [subhazard ratio (SHR) = 1.24, 95% confidence interval (CI) = 0.83–1.86; Table 3]. Patients at high risk of falls were significantly more likely to experience a clinically relevant nonmajor bleeding episode than patients at low risk of falls (adjusted SHR = 1.74, 95% CI = 1.23–2.46).

Table 3. Association between risk of falls and bleeding
 Adjusted subhazard ratioa (95% confidence interval)P-value
  1. a

    Competing risk regression analysis was used, accounting for death as a competing risk. Adjustments were made for age, gender, overt pulmonary embolism, history of major bleeding, recent major surgery, cerebrovascular disease, cardiac disease, diabetes mellitus, hypertension, active cancer, chronic liver disease, chronic renal disease, anaemia, low platelet count, concomitant antiplatelet therapy and anticoagulation as a time-dependent covariate.

Major bleeding
High risk of falls1.24 (0.83–1.86)0.30
Low risk of fallsReference 
Clinically relevant nonmajor bleeding
High risk of falls1.74 (1.23–2.46)0.002
Low risk of fallsReference 

Inclusion of the 59 major and 109 clinically relevant nonmajor bleeding episodes occurring after the first 90 days of anticoagulation in our analysis showed that a high risk of falls was not associated with major bleeding (adjusted SHR = 0.91, 95% CI = 0.55–1.53), but remained significantly associated with clinically relevant nonmajor bleeding (adjusted SHR = 1.80, 95% CI = 1.21–2.68). The results remained nearly the same when missing values were assumed to be abnormal.

Discussion

In this prospective multicentre cohort of elderly patients receiving anticoagulants for VTE, a high risk of falls was significantly associated with clinically relevant nonmajor bleeding, but not major bleeding. As expected, patients at high risk of falls had a higher incidence of fall-related bleeding than patients at low risk of falls. Given that patients at high risk of falls spent more time in a subtherapeutic INR range and were not more likely to be over-anticoagulated at the time of bleeding, our findings do not appear to be confounded by differences in the quality of anticoagulation. Restricting the analysis to bleeding events occurring after the first 90 days of anticoagulation did not change the results.

Our findings are difficult to compare with the results of prior studies examining the association between the risk of falls and bleeding because of differences in study populations, definitions of risk of falls and major bleeding, and duration of follow-up [4-8]. Two studies [4, 8] demonstrated a significant association between an increased risk of falls and major bleeding, but three studies did not [5-7]. In a retrospective study, Medicare beneficiaries with atrial fibrillation who were documented in the medical records to be at high risk of falls had a twofold increased risk of intracranial haemorrhage (mostly trauma-related) compared with control patients without a documented risk of falls [4]. Given that only 33.5% of patients at high risk of falls were receiving warfarin in that study, and the prescription of warfarin did not affect the risk of bleeding, the relationship between falls and anticoagulation-induced intracranial haemorrhage remains unclear. A prospective study enrolling patients ≥80 years of age, of whom 74% were anticoagulated for atrial fibrillation and 26% were anticoagulated for VTE, demonstrated that a history of ≥2 falls during the last year had a threefold increase in major bleeding episodes [8]; however, the study did not examine whether or not major bleeding occurred following a fall or there were differences in the quality of anticoagulation between patients with and without a history of falls.

According to current guidelines [35], patients >65 years of age with unprovoked VTE who fall frequently are considered to be at high risk of bleeding and should not receive anticoagulation beyond 3 months. Given the increased risk of clinically relevant nonmajor bleeding, the presence of an increased risk of falls could be a reason against extending anticoagulation beyond the duration of 3 months in elderly patients with acute VTE, especially if other bleeding risk factors are present. Because nonmajor bleeding is likely to be less relevant than a recurrent VTE event in most cases, the decision not to extend anticoagulation beyond 3 months should be made on an individual basis and should also include patient preferences and the risk of VTE recurrence. In a decision analysis of patients with unprovoked pulmonary embolism, long-term warfarin therapy was most beneficial in 65-year-old patients without a history of falls and was least beneficial in 80-year-old patients with a history of falls [36]. Elderly patients at increased risk of falls who receive anticoagulants for VTE should be instructed to take the following precautions to limit the risk of falling: wear stable shoes; exercise regularly; vitamin D supplementation; use walking aids; and discontinue unnecessary medications [4].

Although patients at high risk of falls had a significantly higher rate of fall-related bleeding episodes, <25% of bleeding events in high-risk patients were fall-related. Thus, in patient populations with a relative low baseline risk of bleeding (e.g. elderly patients without major bleeding risk factors), the absolute risk attributable to fall-related bleeding may be relatively low, even in patients at high risk of falls.

Our study had several strengths. First, in contrast to prior studies that exclusively or predominantly enrolled patients with atrial fibrillation, the current study is the first study to specifically determine the association between the risk of falls and bleeding in patients with VTE. This is important because the risk of bleeding may be higher in patients with VTE than patients with atrial fibrillation [8]. Secondly, given that virtually all patients with acute VTE receive anticoagulation for at least 3 months regardless of the bleeding risk, the potential for indication bias was low in the current study. Indication bias is a threat to the validity of studies of anticoagulation-related bleeding because physicians may choose not to prescribe anticoagulant treatment to patients who they perceive to be at high risk of bleeding or falls [7]. Thirdly, because our sample included elderly patients with multiple morbidities who were at risk of dying from nonhaemorrhagic causes before experiencing a bleeding event, we used competing risk regression to account for the competing risk of death [16]. Finally, in contrast to prior studies that evaluated the risk of falls based on chart review criteria [4-6], we prospectively identified patients at high risk of falls by two standardized, validated questions.

The current study had several potential limitations. First, the sample may not reflect the full prognostic spectrum of patients with VTE because the analysed patients were younger and more likely to be males than the excluded patients. Thus, we cannot exclude the possibility that the association between the risk of falls and bleeding would have been different in more severely ill patients. Secondly, the risk of falls was only assessed at the beginning of the study and may have changed over time. Thirdly, we did not assess the actual number of falls that occurred during follow-up and therefore could not validate the predictive performance of the two screening questions used in our study. Fourthly, the fall-related bleed episodes were mainly based on patient self-report, which may underestimate the real incidence. Finally, our study excluded patients with severe cognitive impairment who may be at the highest risk of falls [37]; thus, our results may not be applicable to such patients.

In conclusion, in elderly patients who receive anticoagulants because of VTE, a high risk of falls is significantly associated with clinically relevant nonmajor bleeding, but not with major bleeding. Such patients should be instructed to take precautions to limit the risk of falling. The decision whether or not a high risk of falls is a reason not to provide anticoagulation beyond 3 months in elderly patients with VTE should be based on patient preferences and the risk of VTE recurrence.

Conflict of interest statement

The authors have no conflict of interests or financial interest related to this work.

Acknowledgements

This study was supported by the Swiss National Science Foundation (grant 33CSCO-122659/139470). We thank all persons who made the study possible.

Ancillary