[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 54.147.241.171. Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Download PDF
Figure 1.
Event-free rates (Kaplan-Meier curves) of total bleeding complications (top) and thrombotic (bottom) events during outpatient anticoagulant treatment in elderly patients and in younger controls.

Event-free rates (Kaplan-Meier curves) of total bleeding complications (top) and thrombotic (bottom) events during outpatient anticoagulant treatment in elderly patients and in younger controls.

Figure 2.
Rates (per 100 patient-years) of total bleeding complications (gray bars), venous thromboembolic (solid bars), or arterial thrombotic events (open bars), according to the time spent in the different international normalized ratio (INR) categories, recorded during follow-up in elderly patients (top) and in younger controls (bottom).

Rates (per 100 patient-years) of total bleeding complications (gray bars), venous thromboembolic (solid bars), or arterial thrombotic events (open bars), according to the time spent in the different international normalized ratio (INR) categories, recorded during follow-up in elderly patients (top) and in younger controls (bottom).

Table 1. 
Information About the Oral Anticoagulant Treatment (OAT) Course of Elderly Patients and Controls Included in the Analysis
Information About the Oral Anticoagulant Treatment (OAT) Course of Elderly Patients and Controls Included in the Analysis
Table 2. 
Reasons for OAT* Withdrawal or Follow-up Interruption During the Observation Period
Reasons for OAT* Withdrawal or Follow-up Interruption During the Observation Period
Table 3. 
Deaths From All Causes Recorded During Follow-up
Deaths From All Causes Recorded During Follow-up
Table 4. 
Number of Elderly Patients and Controls With Presence of Comorbid Conditions, General Risk Factors, and Comedication
Number of Elderly Patients and Controls With Presence of Comorbid Conditions, General Risk Factors, and Comedication
Table 5. 
Number and Type of Bleeding and Thrombotic Events Recorded During Follow-up in Elderly Patients and Controls*
Number and Type of Bleeding and Thrombotic Events Recorded During Follow-up in Elderly Patients and Controls*
Table 6. 
Number of Bleeding and Thrombotic Events in Elderly Patients and Controls According to Sex, Indication for OAT, Anticoagulant Drug Used, Intended Anticoagulation Intensity, and Temporally Related INR Values*
Number of Bleeding and Thrombotic Events in Elderly Patients and Controls According to Sex, Indication for OAT, Anticoagulant Drug Used, Intended Anticoagulation Intensity, and Temporally Related INR Values*
Table 7. 
Risk Ratios (95% Confidence Intervals) Calculated by Multivariate (Poisson Regression) Analysis Taking Into Account Bleeding (Fatal, Major, and Minor) and Thrombotic Events in Elderly Patients and Controls
Risk Ratios (95% Confidence Intervals) Calculated by Multivariate (Poisson Regression) Analysis Taking Into Account Bleeding (Fatal, Major, and Minor) and Thrombotic Events in Elderly Patients and Controls
1.
Palareti  GLeali  NCoccheri  S  et al.  Bleeding complications of oral anticoagulant treatment: an inception-cohort, prospective collaborative study (ISCOAT). Lancet. 1996;348423- 428Article
2.
Beyth  RJLandefeld  CS Anticoagulants in older patients: a safety perspective. Drugs Aging. 1995;645- 54Article
3.
Levine  MNRaskob  GLandefeld  SHirsh  J Hemorrhagic complications of anticoagulant treatment. Chest. 1995;108276S- 290SArticle
4.
Scott  PJW Anticoagulant drugs in the elderly: the risks usually outweigh the benefits. BMJ. 1988;21261- 1263Article
5.
Kutner  MNixon  GSilverstone  F Physicians' attitudes toward oral anticoagulants and antiplatelet agents for stroke prevention in elderly patients with atrial fibrillation. Arch Intern Med. 1991;1511950- 1953Article
6.
Russmann  SGohlke-Barwolf  CJahnchen  ETrenk  DRoskmann  H Age-dependent differences in the anticoagulant effect of phenprocoumon in patients after heart valve surgery. Eur J Clin Pharmacol. 1997;5231- 35Article
7.
Shorr  RIRay  WADaugherty  JRGriffin  MR Concurrent use of nonsteroidal anti-inflammatory drugs and oral anticoagulants places elderly persons at high risk for hemorrhagic peptic ulcer disease. Arch Intern Med. 1993;1531665- 1670Article
8.
Hylek  EMSinger  DE Risk factors far intracranial hemorrhage in outpatients taking warfarin. Ann Intern Med. 1994;120897- 902Article
9.
Arnsten  JHGelfand  JMSinger  DE Determinants of compliance with anticoagulation: a case-control study. Am J Med. 1997;10311- 17Article
10.
Lau  HSBeuning  KSPostmalim  EKleinbeernink  LDeboer  APorsius  AJ Non-compliance in elderly people: evaluation of risk factors by longitudinal data analysis. Pharm World Sci. 1996;1863- 68Article
11.
Palareti  GPoggi  MGuazzaloca  GSavino  ACoccheri  S Assessment of mental ability in elderly anticoagulated patients: its reduction is associated with a less satisfactory quality of treatment. Blood Coagul Fibrinolysis. 1997;8411- 417Article
12.
Palareti  GManotti  CD'Angelo  A  et al.  Thrombotic events during oral anticoagulant treatment: results of the inception-cohort, prospective, collaborative ISCOAT study. Thromb Haemost. 1997;781438- 1443
13.
Tripodi  AChantarangkul  VAkkawat  BClerici  MMannucci  PM A partial factor V deficiency in anticoagulated lyophilized plasmas has been identified as a cause of the international normalized ratio discrepancy in the external quality assessment scheme. Thromb Res. 1995;78283- 292Article
14.
Rosendaal  FRCannegieter  SCVandermeer  FJMBriet  E A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost. 1993;69236- 237
15.
Kuller  LH Sudden death: definition and epidemiologic considerations. Prog Cardiovasc Dis. 1980;231- 12Article
16.
Fihn  SDCallahan  CMMartin  DCMcdonell  MBHenikoff  JGWhite  RH The risk for and severity of bleeding complications in elderly patients treated with warfarin. Ann Intern Med. 1996;124970Article
17.
Landefeld  CSGoldman  L Major bleeding in outpatients treated with warfarin: incidence and prediction by factors known at the start of outpatient therapy. Am J Med. 1989;87144- 152Article
18.
Stroke Prevention in Atrial Fibrillation Investigators, Bleeding during antithrombotic therapy in patients with atrial fibrillation. Arch Intern Med. 1996;156409- 416Article
19.
Hart  RGBoop  BSAnderson  DC Oral anticoagulants and intracranial hemorrhage: facts and hypotheses. Stroke. 1995;261471- 1477Article
20.
Landefeld  CSRosenblatt  MWGoldman  L Bleeding in outpatients treated with warfarin: relation to the prothrombin time and important remediable lesions. Am J Med. 1989;87153- 159Article
21.
Levine  MHirsh  J Hemorrhagic complications of long-term anticoagulant therapy for ischemic cerebral vascular disease. Stroke. 1986;17111- 1116Article
22.
Dahl  TAbildgaard  USandset  PM Long-term anticoagulant therapy in cerebrovascular disease: does bleeding outweigh the benefit? J Intern Med. 1995;237323- 329Article
Original Investigation
February 28, 2000

Oral Anticoagulation Treatment in the ElderlyA Nested, Prospective, Case-Control Study

Author Affiliations

From the Departments of Angiology and Blood Coagulation, University Hospital S Orsola-Malpighi, Bologna, Italy (Coordinating Center) (Drs Palareti, Guazzaloca, Legnani, and Coccheri); Research Centre, Hamilton Civic Hospitals, McMaster University, Hamilton, Ontario (Dr Hirsh); Centro Emostasi, Ospedale Regionale, Parma (Dr Manotti), Ambulatorio Emostasi Trombosi, IRCCS Ospedale S Raffaele, Milano (Dr D'Angelo), Servizio Prevenzione Trombosi, Cattedra di Cardiologia, Università di Padova, Padova (Dr Pengo), Centro Emofilia e Trombosi A. Bianchi Bonomi, IRCCS Ospedale Maggiore, Università di Milano, Milano (Dr Moia); and Istituto di Statistica, Università di Bologna, Bologna (Dr Musolesi), Italy.

Arch Intern Med. 2000;160(4):470-478. doi:10.1001/archinte.160.4.470
Abstract

Background  Whether elderly patients are at increased risk of complications during oral anticoagulant treatment (OAT) is still a matter of debate.

Method  Bleeding and thrombotic events occurring during OAT in 461 patients, aged 75 years or older when they started OAT, and in 461 patients younger than 70 years, matched for sex, OAT indication, and treating center, were examined in a prospective, multicenter, inception-cohort study.

Results  Bleeding rate was 9.9% and 6.6% patient-years in elderly and young patients, respectively (P = .07), and 2.1% and 1.1% for major bleeding (P = .19); 6 and 1 events, respectively, were fatal (all intracranial, relative risk, 6.4; P = .05). In the elderly, bleeding rate was lower (4.5%) for international normalized ratios (INRs) between 2.0 and 2.9; it was higher during the first 90 treatment days (P = .05) and when arterial vascular disease was the indication for OAT (P = .03). Thrombosis rate was 4.2% and 2.5% patient-years in elderly and young patients, respectively (P = .10); however, 13 and 5 events were fatal (relative risk, 2.8; P = .04). Thrombosis rate was lower (1.5%) for INRs between 2.0 and 2.9; it was higher during the first 90 treatment days (P<.001) and 6 of 7 venous events occurred at lower than 2.0 INRs.

Conclusions  A nonsignificant trend was noted toward a higher rate of both bleeding and thrombotic complications in elderly vs matched younger patients. Intracranial bleeding and fatal thrombotic events were significantly more frequent in the elderly. Our results also indicate that lower than 2.0 INRs do not preclude bleeding in the elderly nor offer adequate protection from thrombotic events. Moderate anticoagulation (2.0-3.0 INRs) in elderly patients seems the safest and most effective.

ORAL anticoagulant treatment (OAT) is effective in the prevention and treatment of thromboembolic complications in patients with cardiovascular disease. Some of the indications for OAT (eg, venous thromboembolism and atrial fibrillation) are particularly frequent in elderly people, the fastest growing population of our society. More than one third of all patients included in a recent collaborative prospective Italian study were older than 70 years when they started OAT, and 8% were older than 80.1 It is uncertain whether the risk of bleeding during OAT is higher in older patients,2,3 although many physicians are persuaded that the risk of OAT is higher in the elderly.4,5 Elderly subjects have the potential to be at higher risk for bleeding complications during OAT for the following reasons: they require lower anticoagulant doses than younger subjects, mainly because of reduced metabolic clearance6; they are more likely to be taking interacting drugs; they have a higher prevalence of comorbid conditions7; and they have been reported to have increased vascular fragility, a factor that may increase the risk of intracranial bleeding.8 Noncompliance with OAT has been reported to be similar in elderly and younger patients.9 Nevertheless, noncompliance could also contribute due to the complexity of the drug regimen, or a lack of a clear understanding of the purpose of the treatment by the elderly10 who are prone to mental impairment.11 The physician is therefore faced with a dilemma. The elderly are considered to be at higher risk of bleeding during OAT, but an increasing number of elderly patients are candidates for, and could benefit from, anticoagulants; this trend is expected to increase in the future. Therefore, there is a need for reliable information on the complication rates of OAT in the elderly and the modifiable factors that are responsible for these complications.

Accordingly, the aims of the present study are to: (1) analyze the rates of bleeding complications and of thrombotic events recorded during treatment in elderly patients (≥75 years at initiation of OAT) and (2) compare these outcomes in the elderly with an equal number of younger controls (aged <70 years at the initiation of OAT), matched for sex, main indication for OAT, and enrolling center. All patients and controls were part of the Italian Study on Complications of Oral Anticoagulant Therapy (ISCOAT), a collaborative, prospective, multicenter study, although the observation period in this article has been extended beyond the follow-up period of the 2 previous ISCOAT publications.1,12

PATIENTS, MATERIALS, AND METHODS
DESIGN AND PATIENTS

The ISCOAT has been described in detail elsewhere.1 A total of 34 anticoagulation clinics affiliated with the Italian Federation of Anticoagulation Clinics took part in ISCOAT. The Italian Federation of Anticoagulation Clinics are required to perform the following services: give extensive instructions to all new patients; periodically perform a prothrombin time test (results expressed as international normalized ratio [INR]); fix the date for the next visit and prescribe daily anticoagulant doses for the upcoming period; monitor changes in patient habits, diet, comedication, intercurrent illnesses, bleeding complications, and scheduled surgical or invasive procedures. These services are performed by conducting an interview at each visit, or calling the patient's home or the patient's family physician if the patient misses the date of the visit by more than 20 days. All the Italian Federation of Anticoagulation Clinics take part in the specifically designed external quality control program, which runs 3 times yearly and uses lyophilized plasma samples obtained from anticoagulated patients.13

From May 1, 1993, to October 31, 1994, consecutive patients receiving either warfarin sodium or acenocoumarol therapy for the first time and for no more than 30 days, were recruited in each center, independently of age, indication for anticoagulation, intended therapeutic range, or expected duration of treatment. The only exclusion criteria were pregnancy or the inability to obtain appropriate follow-up. The observation period, in this study, ended on August 31, 1996; it was earlier if treatment was discontinued, the patient stopped attending the center, or was censored after a major bleeding or thrombotic event.

ELDERLY PATIENTS AND SELECTION OF THE CONTROLS

From the total number of 2745 patients enrolled in the ISCOAT, we evaluated all patients who were aged 75 years or older when included in the ISCOAT and an equal number of patients younger than 70 years as controls. The controls were matched for sex, OAT indication, and, as far as possible, for the enrolling center. The elderly patients were identified and the controls selected in the coordinating center as follows: from the general ISCOAT database, the fields relevant for the selection were copied onto an electronic page and then printed; the fields considered were enrolling center, progressive individual number of the patients recruited in each center (sorted by ascending date of inclusion), sex, age at the time of inclusion, and the main indication for OAT (selected from a standard reference list used by all the participating centers). A member of the laboratory staff in the coordinating center, unaware of all the remaining information contained in the main database, was responsible for identifying the elderly patients and for selecting the controls. The printout list of all the ISCOAT patients was checked and sorted by ascending order of enrolling center and the date of inclusion; the list contained only the above-mentioned fields. The first patient aged 75 years or older was identified and then a control, matched for sex and main indication to OAT but younger than 70 years, was selected. This procedure was then repeated until all elderly patients and matched controls were identified for each center. Those elderly patients for whom suitable controls could not be found in the same center ("Results" section) were matched with controls identified among the patients from the next center, by following the progressive code order of the participating centers.

DATA COLLECTION, STUDY MONITORING, AND LABORATORY CONTROL OF ANTICOAGULATION

Data collection, study monitoring, and laboratory control of anticoagulation methods are described in detail elsewhere.1 The quality of anticoagulation monitoring during follow-up was investigated by calculating the observed percentage of time spent at different INR levels using a software program.14 This program allows the number of days (and patient-years) accumulated at different INR levels (in 0.5-INR intervals) to be calculated for all patients, or according to specific stratifications. The incidence of events at different achieved intensities of anticoagulation (different INR categories) can also be calculated by dividing the number of events occurring in patients with temporally related INR values in each category by the total number of patient-years accumulated in that range. An INR value was defined as "temporally related" to an outcome event when it was obtained at the time of the bleeding or thrombotic event or during the preceding 8 or 14 days for bleeding or thrombotic events, respectively.

OUTCOMES

Both bleeding and thrombotic complications that occurred during follow-up are considered in this article. Events were classified according to the diagnosis reported in hospital or emergency department records, death certificates, or autopsy results (when available); cases of superficial thrombophlebitis were diagnosed by the physician of the respective center.

Deaths for all causes were recorded and coded as follows: (1) due to an underlying or other disease; (2) sudden death, which was not considered a thrombotic event unless otherwise confirmed by autopsy15; (3) death caused by bleeding; or (4) death due to a thrombotic event, when caused by pulmonary embolism, acute myocardial infarction, or stroke; and when death was generically attributed to cardiovascular causes.

The use of a standardized classification of either bleeding or thrombotic events is promoted among all Italian Federation of Anticoagulation Clinics and was adopted by those participating in ISCOAT. Bleeds were classified as major and minor, using the following criteria. For major bleeding: fatal (death due to hemorrhage); intracranial (documented by computed axial tomographic scan and/or nuclear magnetic resonance imaging); ocular (with blindness); articular; retroperitoneal; if surgery or angiographic intervention was necessary to stop bleeding; if bleeding led to a reduction in the hemogloblin level of 2 g/dL or more and/or a need for a blood transfusion of 2 U or more. For minor bleeding: all cases of bleeding not considered as major. For minor bruising or ecchymoses, self-limiting epistaxis (not requiring emergency visit or tamponade), occasional hemorrhoidal bleeding, and microscopic hematuria were not considered as bleeding events.

Deep vein thrombosis (in whatever site, first event or recurrence), pulmonary embolism, stroke, acute myocardial infarction, peripheral or visceral arterial thromboembolism, thrombosis of a prosthetic cardiac valve, clinical worsening of peripheral obliterative arterial disease leading to amputation or reconstructive surgery were considered major thrombotic events. Transient cerebral ischemic attacks and superficial thrombophlebitis were all considered minor thrombotic events.

In the few cases in which the attribution of the events was unclear, the coordinating center requested a more detailed description of the event and adjudicated the event. If classification of one event was still uncertain, it was coded as major (this happened on a few occasions for hemorrhages).

STATISTICAL ANALYSIS

Differences between groups were assessed by the χ2 test. Cumulative incidence of hemorrhagic and thrombotic events was calculated separately using the Kaplan-Meier method; data were censored after the first complication, cessation of OAT, or discontinuation of monitoring by the center. The effect of various risk factors (sex, intended anticoagulation intensity, type of coumarin drug, timing of the event from the beginning of treatment, and temporally related anticoagulation intensity) was assessed by univariate analysis while the possible interaction between some factors was investigated by performing a Poisson regression analysis. Where indicated 95% confidence intervals (CIs) were calculated and a P≤.05 (2 tailed) was considered as statistically significant. The SOLO statistical software package (version 4.0; BMDP Statistical Software Inc, Los Angeles, Calif) was used for data processing.

RESULTS
ELDERLY PATIENTS AND CONTROLS

Among the 2745 patients included in the ISCOAT, 461 (200 men) were aged 75 years or older at inclusion (median age, 79 years; age range, 75-93 years). Following the procedure described in the "Patients, Materials, and Methods" section, 461 patients aged 70 years or younger were selected as controls (median age, 61 years; age range, 15-69 years); they were matched for sex, main indication for OAT, and enrolling center. Only 43 controls (9.3%) could not be matched for enrolling center.

Table 1 summarizes information about OAT of the cohort of 461 elderly patients and 461 matched controls. Thirty-eight elderly patients (8.0%) were aged older than 80 years when they started OAT. The most common indications for OAT were venous thromboembolism (n = 175 [38%]) and nonischemic heart disease (mainly atrial fibrillation) (n = 139 patients [30.1%]). Multiple indications for treatment were reported in 177 elderly patients (38.4%) and 129 controls (28.0%). About two thirds of the patients or controls were treated with warfarin, the remainder with acenocoumarol. The intended anticoagulation level was of moderate intensity (INR target values ≤2.8) for most patients (79.8%) and controls (78.1%).

As already discussed elsewhere,1,12 anticoagulation control was evaluated separately in patients and controls whose targeted INR was in the moderate (INR = 2.0-3.5) and high (INR = 2.5-4.5) intensity ranges. The total follow-up treatment period was 566 and 609 patient-years for the elderly and control patients, respectively. During the observation period (Table 2) 270 elderly patients (58.6%) and 248 controls (53.8%) withdrew from OAT. The distribution of the individual length of follow-up in the patients from both groups and the mean interval between visits are given in Table 1. The reasons for withdrawal were similar in both groups, except that there were more deaths from all causes (Table 3) in the elderly patients than in the controls (11.3% and 4.1%, respectively, relative risk [RR] 3.0; 95% CI, 1.9-4.0; P<.001).

One or more comorbid conditions or other risk factors for bleeding were reported in a substantial number of elderly patients and controls (38.6% and 39.5%, respectively) (Table 4). In addition more than 50% of patients in both groups were also taking 1 or more other drugs (Table 4).

BLEEDING EVENTS

The bleeding and thrombotic events that occurred in elderly patients and in controls during follow-up are listed in Table 5, and the event-free rates (Kaplan-Meier curves) are shown in Figure 1. Although the rate of all bleeding events was higher in the elderly, this trend was not statistically significant (RR, 1.44; 95% CI, 0.96-2.14; P = .07). However, the incidence of fatal episodes from bleeding—all intracranial—was significantly higher in the elderly (RR, 6.4; 95% CI, 1.02-40.6; P = .047). The highest rate of bleeding (Table 6) was recorded in patients whose main indication for OAT was the presence of cerebral vascular disease or peripheral arteriopathy (17.1%); RR vs all the other indications was 1.97; 95% CI, 1.08-3.56; P = .03. No differences were detected between the 2 groups for other variables, including sex, type of coumarin drug used, and intended anticoagulation intensity. In the elderly, the risk of hemorrhagic events during the course of therapy was higher during the first 90 days of treatment, decreasing considerably thereafter (15.4% vs 8.7%; RR, 1.78; 95% CI, 1.00-3.15; P = .049).

The incidence of bleeding events at different achieved intensities of OAT was investigated by dividing the number of events occurring in patients with temporally related INR values in 4 increasing INR categories (<2.0, 2.0-2.9, 3.0-4.4, and ≥4.5) by the total number of patient-years accumulated in these categories. As reported in Table 6 and shown in Figure 2, 10 (21.7%) of 46 bleeding events with available temporally related INR values occurred at low levels of anticoagulation intensity (10.0% patient-years in the <2.0 INR category). The rate of bleeding was lowest (P<.001) in the 2.0 to 2.9 INR category (4.5%). There was an increase in the bleeding incidence with increasing INR values, which became exponential for INR values greater than 4.5. A similar trend was also recorded in controls (Table 6 and Figure 2). Multivariate analysis confirmed that both in elderly patients and in controls, the risk of bleeding was markedly higher during the first 90 days of treatment, and when arterial vascular disease was the indication for OAT (Table 7). The risk of bleeding was significantly higher in elderly patients who received multiple-drug therapy. The quality of anticoagulation control did not differ between elderly patients and controls and was similar in patients who either did or did not experience bleeding events. In particular, elderly patients who bled had INR value3 of 4.0 for 4.9% of the time, compared with the 4.5% for those who did not bleed.

As regards the 6 elderly patients who had fatal intracranial hemorrhage, it should be noted that (1) only 1 had a time-related INR value below 3.0 (INR = 2.8), while the INR values in the remaining cases were 3.3, 3.9, 4.0, 4.3 and 5.9, respectively; (2) 4 had cerebral or peripheral arteriosclerosis as either a main or secondary indication for OAT, 2 of them with associated arterial hypertension (treated and apparently controlled); and (3) 1 patient had associated type 2 diabetes mellitus. The 1 control patient who died of intracranial hemorrhage had an associated condition of hypertensive blood pressure. In none of these 7 cases was the occurrence of a trauma before the bleeding event reported.

THROMBOTIC EVENTS

Though the rate of all thrombotic events was no different in elderly patients than in controls (P = .10), the elderly had a significantly higher number of fatal thrombotic complications (RR, 2.8; 95% CI, 1.04-7.50; P = .04). As listed in Table 6, the rates of the events in both groups were not different according to the following variables: sex, type of coumarin drug, main indications for OAT, and intended therapeutic ranges. In the elderly patients, 13 of 22 events with available time-related INR values occurred at an INR value less than 2.0 (RR, <2.0 INRs vs ≥2.0 INRs = 6.56; 95% CI, 3.13-13.70; P<.001); furthermore, all venous events, except one, occurred at INR values less than 2.0 (Figure 2). Conversely, in young controls only 2 of 14 events with available time-related INR values occurred at INR values less than 2.0: moreover, all of the 5 venous events (3, deep vein thrombosis recurrences; and 2, superficial phlebitis) had a time-related INR value higher than 2.0, although 2 of these events occurred in patients with thrombophilic alterations (lupus anticoagulant, protein S deficiency).

In the elderly patients, the rate of thrombosis was markedly higher during the first 90 days of treatment (RR vs >90 days = 3.72; 95% CI, 1.76-7.86; P<.001). These differences were not detected in the control group, in which a significantly higher rate of events was recorded in patients whose main indication for OAT was arterial vascular disease (RR vs all the other indications = 3.28; 95% CI, 1.24-8.70; P = .02). The multivariate analysis confirmed that the risk of thrombosis was markedly higher during the first 90 days of treatment both in elderly patients and controls (though the univariate analysis did not show this difference) as well as when arterial disease was the indication for OAT in controls (Table 7). Analysis of the laboratory control quality of treatment showed that elderly patients who had experienced thrombotic events had spent significantly more time at inadequately low anticoagulation intensity levels (<2.0 INRs) than those free from thrombotic complications (24.8% and 17.7%, respectively; RR, 1.40; 95% CI, 1.34-1.46; P<.001).

COMMENT

This large, collaborative, prospective, multicenter study showed a nonsignificant trend for overall rates of bleeding during OAT to be higher in patients aged 75 years or older than in younger controls who were aged younger than 70 years. A nonsignificant trend for higher rates of thrombotic events during OAT was also recorded in older patients. The elderly patients and controls were matched for sex, main indication for therapy, and treating center. A higher risk of fatal complications occurred in elderly patients (6 vs only 1) than in controls. These results agree with the findings of recent studies in which the risk of life-threatening or fatal bleeding was significantly higher in older than younger patients receiving OAT.6,16 In this study, all fatal bleeding events recorded in both groups were intracranial hemorrhages. Others have also reported that the risk of intracranial bleeding during OAT is higher in older persons.8,17,18 In their review, Hart et al19 conclude that predictors of intracerebral hematoma during OAT are advanced age, prior ischemic stroke, hypertension, and intensity of OAT. In our previous article about bleeding complications in the total ISCOAT population, no relationship was detected between intracranial bleeding and the INR.1 In contrast, in the present study, which focused on elderly patients, we found that only 1 of 6 patients who died of intracranial bleeding had an event-related INR value within the moderate-intensity therapeutic range (INR = 2.8); in 4 cases INR values ranged between 3.3 and 4.3, and in the final case the INR value was 5.9. Thus, in the elderly patients, a moderately high INR (3.0-4.5) is a risk factor for intracranial hemorrhage. The presence of cerebral or peripheral arteriosclerosis in 4 of these patients (2 of them with associated hypertensive blood pressures) may well be considered another risk factor for intracranial bleeding.

The lowest rate of bleeding occurred at INR values of 2.0 to 2.9. However, a substantial number of events (10.0%) occurred in association with very low INR values (<2.0), confirming previous reports1,20 that bleeding during OAT is not always related to the intensity of the OAT but that OAT can unmask a local bleeding source. The risk of bleeding markedly increased with INR values of 3.0 to 4.4, and became disproportionally high for INR values above 4.5.

In keeping with the results of the original ISCOAT,1 in this article we found that the rate of bleeding in elderly patients was higher during the first 90 days of each OAT course, almost double that recorded thereafter. This early effect resulted from the unmasking of cryptic lesions by the OAT as well as from poor anticoagulant control during the initiation of treatment. It also occurred in the controls, but was much less marked.

Elderly patients (but not younger controls) undergoing OAT for arterial vascular disease had a significantly higher incidence of bleeding than the others. This result agrees with previous findings1,17 and indicates that the risk of bleeding during OAT in these patients may outweigh the benefit.21,22 The multivariate analysis showed that comedication was significantly associated with a higher bleeding risk in elderly patients, confirming observations made by Beyth and Landefeld.2 Bleeding from peptic ulcer disease has been reported to be increased about 13-fold in elderly persons receiving OAT when comedicated with nonsteroidal anti-inflammatory drugs.7

Though the difference was not statistically significant, more thrombotic episodes occurred in elderly patients than in controls (24 vs 15, respectively); the number of fatal thrombotic episodes was, however, significantly higher in the elderly patients (13 vs 5, P = .04). As discussed elsewhere,12 the outcomes considered in this article are heterogeneous, including venous thromboembolism, arterial embolic or nonembolic events, and vascular death. Oral anticoagulants might be expected to be more effective in the treatment of venous thromboembolism than in arterial thrombosis. In this study, 7 (5 fatal) of the events occurring in elderly patients were of venous thromboembolism, the remaining were arterial. In the elderly group (1) more than half of all the events for which a time-related INR value was available occurred when the INR values were less than 2.0; (2) only 1 venous thromboembolic event occurred with an INR value higher than 2.0; (3) those patients who had thrombotic events spent significantly more time at very low anticoagulation levels (<2.0 INRs) than those free from thrombotic complications (24.8% and 17.7%, respectively); and (4) almost half of the events occurred during the first 90 days of anticoagulation treatment. The results were different in controls, in which most events occurred in conjunction with therapeutic INR values and the event rates were not different in the various INR categories. Possible reasons for this observation are (1) the prevalence of arterial events (10 of 15 cases), for which a moderate-intensity anticoagulation therapeutic range (INR value, 2.0-3.0) may be less protective, and (2) the presence of a thrombophilic condition (lupus anticoagulant and protein S deficiency) in 2 of 5 controls who experienced venous events.

CONCLUSIONS

This article, in line with the findings of Fihn et al,16 showed that elderly subjects undergoing OAT have a higher risk of more serious complications—both fatal thrombotic episodes of bleeding and thrombosis. As has already been pointed out,8,19 intracranial hemorrhage is the most feared and serious bleeding complication in elderly patients. Hylek and Singer8 contend that anticoagulation intensity is the dominant risk factor for this complication and recommend avoiding INR values above 4. Our results indicate that in elderly patients, intracranial hemorrhage occurs with lower INR values, suggesting that, from the perspective of bleeding complications, an INR range between 2.0 and 3.0 represents the safest intensity of OAT. The high rate of bleeding in elderly patients with an INR below this range is difficult to explain. It could result from the presence of underlying lesions that are more frequent in older than younger subjects and that become unmasked early during the first stage of anticoagulant treatment (the highest rate was recorded during the first 90 days of treatment). Conversely, the bleeding rate increased sharply, as expected, above 3.0 INRs in proportion to the degree of anticoagulation. Our results also suggest that an INR range of 2.0 to 3.0 in elderly patients is also effective. Thus, thrombosis was more frequent at INR less than 2.0 (especially recurrence of venous thromboembolism). However, arterial venous events seemed to be less dependent on the degree of the intensity level of OAT, occurring even at higher INR values.

An increasing number of elderly subjects are candidates for OAT. However, because many physicians who prescribe coumarin therapy believe the risk of OAT is higher in elderly patients, they use a low targeted therapeutic range, allowing the INR to fall below an INR of 2.0. Our findings suggest a low targeted INR range (<2.0) should be avoided. The findings of this study indicate that the therapeutic range in elderly patients should be an INR of 2.0 to 3.0, since this range is associated with the lowest bleeding or thrombotic events.

Accepted for publication April 22, 1999.

The software program used to analyze the international normalized ratio of the anticoagulant was provided by F. R. Rosendaal, MD, University Hospital, Leiden, the Netherlands.

The study examined patients enrolled in the ISCOAT (Italian Study on Complications of Oral Anticoagulant Therapy); the participants in the ISCOAT Study Group are listed below.

Clinical Centers and Principal Investigators

The centers and investigators are listed in decreasing order of number of patients enrolled.

Chairman: G. Palareti, MD (Bologna, current president of the Italian Federation of the Anticoagulation Clinics)

Cattedra e Divisione di Angiologia e Malattie della Coagulazione, Policlinico S Orsola-Malpighi, Bologna; Coordinating Centre: G. Palareti, MD; G. Guazzaloca, MD; M. Poggi, PhD; S. Coccheri, MD.

Centro Emostasi, Ospedale Regionale, Parma: C. Manotti, MD; and R. Quintavalla, MD.

Ambulatorio Emostasi Trombosi, IRCCS Ospedale S Raffaele, Milano: A. D'Angelo, MD; L. Crippa, MD.

Servizio Prevenzione Trombosi, Cattedra di Cardiologia, Università di Padova, Padova: V. Pengo, MD.

Ambulatorio Emostasi Sezione Trasfusionale, Ospedale di Merate, Merate: N. Erba, MD; D. Restifo, MD.

Centro Emofilia e Trombosi A.Bianchi Bonomi, IRCCS Ospedale Maggiore, Università di Milano, Milano: M. Moia, MD; P. Bucciarelli, MD.

Servizio di Coagulazione, Policlinico, Bari: N. Ciavarella, MD; C. Ettorre, MD.

Laboratorio Analisi, Ospedale di Lavagna, Lavagna: G. Devoto, MD.

Centro Emostasi e Trombosi, Istituto Medicina Interna e Vascolare, Università di Perugia, Perugia: M. Berrettini, MD; F. Poeta, MD.

Servizio di Fisiopatologia della Coagulazione, Arcispedale S Anna, Ferrara: G. Ballerini, MD.

Divisione di Ematologia, Ospedale Niguarda, Milano: F. Baudo, MD.

Laboratorio Analisi Ospedale Pediatrico Apuano, Massa: F. Veschi, PhD.

Centro Emostasi e Trombosi, Servizio Trasfusionale, Sassari: G. Piseddu, MD.

Laboratorio Analisi Istituto di Patologia Clinica, Ospedale di Cremona, Cermona: S. Testa, MD.

Servizio di Immunoematologia, Ospedale Maria Vittoria, Torino: M. Molinatti, MD.

Laboratorio Analisi e Centro Trombosi, Ospedale Valduce, Como: L. Frigerio, MD.

Servizio di Immunoematologia, Ospedale di Ivrea: M. Pagliarino, MD.

Centro Emostasi e Angiologia Medica, Medicina Interna, Ospedale di Gallipoli, Gallipoli: L. Ria, MD.

Centro Trasfusionale, Istituti Clinici Milano, Milano: L. Gatti, MD.

Servizio Trasfusionale, Ospedale di Molfetta, Molfetta: G. Malcangi, MD.

Centro Emostasi e Trombosi, Ospedale di Cosenza, Cosenza: E. Rossi, MD.

Laboratorio Analisi, Ospedale di Seriate, Seriate: C. Agazzi, MD.

Laboratorio Emocoagulazione, Ospedale di S Vito al Tagliamento, S Vito al Tagliamento: F. Fusco, MD.

Laboratorio Analisi, Ospedale di San Secondo Parmense, San Secondo Parmense: G. Bazzicalupo, MD.

Servizio di Cardiologia, Ospedale di Este, Este: F. Corbara, MD.

Clinica Medica Generale e Cardiologia, Università di Firenze, Firenze: D. Prisco, MD.

Emostasi e Trombosi, Divisione de Ematologia, Ospediale di Latina, Latina: S. Guarino, MD.

Centro Anticoagulati, Ospedale di Rimini, Rimini: E. Tiraferri, MD.

Divisione Ematologia, Ospedale Cardarelli di Napoli, Napoli: V. Brancaccio, MD.

Laboratorio Patologia Clinica, Ospedale di Benevento, Benevento: V. Rocco, MD.

Laboratorio Analisi, Ospedale di Castelnuovo Monti, Castelnuovo Monti: G. Labò, MD.

Istituto Medicina Interna, Università di Cagliari, Cagliari: F. Marongiu, MD.

Clinica Medica, Università di Pavia, Ospedale Multizonale di Varese, Varese: L. Steidl, MD.

II Laboratorio Analisi, Ospedale di Brescia, Brescia: R. Del Bono, MD.

Reprints: Gualtiero Palareti, MD, Department of Angiology and Blood Coagulation, University Hospital S Orsola-Malpighi, Via Massarenti 9, 40138 Bologna, Italy (e-mail: palareti@tin.it).

References
1.
Palareti  GLeali  NCoccheri  S  et al.  Bleeding complications of oral anticoagulant treatment: an inception-cohort, prospective collaborative study (ISCOAT). Lancet. 1996;348423- 428Article
2.
Beyth  RJLandefeld  CS Anticoagulants in older patients: a safety perspective. Drugs Aging. 1995;645- 54Article
3.
Levine  MNRaskob  GLandefeld  SHirsh  J Hemorrhagic complications of anticoagulant treatment. Chest. 1995;108276S- 290SArticle
4.
Scott  PJW Anticoagulant drugs in the elderly: the risks usually outweigh the benefits. BMJ. 1988;21261- 1263Article
5.
Kutner  MNixon  GSilverstone  F Physicians' attitudes toward oral anticoagulants and antiplatelet agents for stroke prevention in elderly patients with atrial fibrillation. Arch Intern Med. 1991;1511950- 1953Article
6.
Russmann  SGohlke-Barwolf  CJahnchen  ETrenk  DRoskmann  H Age-dependent differences in the anticoagulant effect of phenprocoumon in patients after heart valve surgery. Eur J Clin Pharmacol. 1997;5231- 35Article
7.
Shorr  RIRay  WADaugherty  JRGriffin  MR Concurrent use of nonsteroidal anti-inflammatory drugs and oral anticoagulants places elderly persons at high risk for hemorrhagic peptic ulcer disease. Arch Intern Med. 1993;1531665- 1670Article
8.
Hylek  EMSinger  DE Risk factors far intracranial hemorrhage in outpatients taking warfarin. Ann Intern Med. 1994;120897- 902Article
9.
Arnsten  JHGelfand  JMSinger  DE Determinants of compliance with anticoagulation: a case-control study. Am J Med. 1997;10311- 17Article
10.
Lau  HSBeuning  KSPostmalim  EKleinbeernink  LDeboer  APorsius  AJ Non-compliance in elderly people: evaluation of risk factors by longitudinal data analysis. Pharm World Sci. 1996;1863- 68Article
11.
Palareti  GPoggi  MGuazzaloca  GSavino  ACoccheri  S Assessment of mental ability in elderly anticoagulated patients: its reduction is associated with a less satisfactory quality of treatment. Blood Coagul Fibrinolysis. 1997;8411- 417Article
12.
Palareti  GManotti  CD'Angelo  A  et al.  Thrombotic events during oral anticoagulant treatment: results of the inception-cohort, prospective, collaborative ISCOAT study. Thromb Haemost. 1997;781438- 1443
13.
Tripodi  AChantarangkul  VAkkawat  BClerici  MMannucci  PM A partial factor V deficiency in anticoagulated lyophilized plasmas has been identified as a cause of the international normalized ratio discrepancy in the external quality assessment scheme. Thromb Res. 1995;78283- 292Article
14.
Rosendaal  FRCannegieter  SCVandermeer  FJMBriet  E A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost. 1993;69236- 237
15.
Kuller  LH Sudden death: definition and epidemiologic considerations. Prog Cardiovasc Dis. 1980;231- 12Article
16.
Fihn  SDCallahan  CMMartin  DCMcdonell  MBHenikoff  JGWhite  RH The risk for and severity of bleeding complications in elderly patients treated with warfarin. Ann Intern Med. 1996;124970Article
17.
Landefeld  CSGoldman  L Major bleeding in outpatients treated with warfarin: incidence and prediction by factors known at the start of outpatient therapy. Am J Med. 1989;87144- 152Article
18.
Stroke Prevention in Atrial Fibrillation Investigators, Bleeding during antithrombotic therapy in patients with atrial fibrillation. Arch Intern Med. 1996;156409- 416Article
19.
Hart  RGBoop  BSAnderson  DC Oral anticoagulants and intracranial hemorrhage: facts and hypotheses. Stroke. 1995;261471- 1477Article
20.
Landefeld  CSRosenblatt  MWGoldman  L Bleeding in outpatients treated with warfarin: relation to the prothrombin time and important remediable lesions. Am J Med. 1989;87153- 159Article
21.
Levine  MHirsh  J Hemorrhagic complications of long-term anticoagulant therapy for ischemic cerebral vascular disease. Stroke. 1986;17111- 1116Article
22.
Dahl  TAbildgaard  USandset  PM Long-term anticoagulant therapy in cerebrovascular disease: does bleeding outweigh the benefit? J Intern Med. 1995;237323- 329Article
×