Cohort study on the quality of oral anticoagulation therapy in chronic haemodialysis patients treated with phenprocoumon

BACKGROUND: Few studies have been published on the control of oral anticoagulation treatment in end stage renal disease (ESRD). METHODS: To analyse the quality of oral anticoagulation treatment control in ESRD patients treated with phenprocoumon we conducted a cohort study including all patients on chronic haemodialysis at a reference date. Data were collected retrospectively for 12 months and prospectively for 12 months preceding following the reference date. Endpoint was the percentage of INR in target range. RESULTS: 30 (27%) of 111 patients received oral anticoagulation treatment. The median frequency of INR measurements was every 6.5 days (range 1–16). In median 54% (range 17–74%) and 49% (range 21–65%) of INR measurements were within, 17% (range 0–45%) and 19% (range 4–56%) were above and 27% (range 8–83%) and 33% (range 9–57%) were below the target range in the retrospective and prospective dataset, respectively. The percentage of INR measurements within target range was significantly higher in patients with a target range width of 1.0 than in patients with a target range width of 0.5 (p = 0.04). There was no difference in the number of bleedings or thromboembolic events in patients with and without oral anticoagulation treatment. CONCLUSION: In our ESRD cohort, the percentage of INR in target range in patients treated with phenprocoumon seems comparable with published data on warfarin and


Background
The standard intervention for therapy and prevention of thromboembolic events is oral anticoagulation therapy (OAT) with vitamin K antagonists.OAT has a narrow therapeutic range with risk of bleeding in over-anticoagulation and risk of thromboembolism in under-anticoagulation [1].It is monitored by measurement of prothrombin time, expressed as international normalised ratio (INR) [2,3].An INR below 1.3 indicates a normal prothrombin time, while values above this reference indicate a prolonged prothrombin time and an increased risk of bleeding [4].To minimise complications, OAT dosage is continuously adapted to keep the INR within a defined target range.American College of Chest Physicians (ACCP) guidelines currently recommend a moderate intensity OAT (target INR 2.0-3.0),but state that a single therapeutic target range may not be optimal for all indications [2].Depending on the stability of INR results, INR should be monitored between daily at the initiation of OAT and at least once every 12 weeks when a stable dose of OAT has been established [5].INR control correlates negatively with the risk of adverse events [6,7].However, good INR control is difficult to achieve.In a meta-analysis of 47 studies in outpatients with atrial fibrillation (AF) mostly treated with warfarin, median percentage of INR in target range was 53% (range 34-68%) for retrospective studies, whereby 26% (10-51%) and 17% (14-29%) of all measurements were below and above target range, respectively [6].In recently published randomized controlled trials comparing safety and efficacy of factor Xa and thrombin inhibitors to warfarin in patients with AF or acute venous thromboembolism, 50 to 67% of INRs in the warfarin group were in target range [8][9][10][11][12][13][14][15].
Since patients on chronic haemodialysis suffer from a variety of co-morbidities associated with thromboembolic complications, the need for antithrombotic prophylaxis and therapy is high in this group of patients.The prevalence of AF in patients on dialysis varies between 7.7% and 27% [16][17][18][19][20][21][22], while in the general population its overall prevalence is 0.95%, ranging from 0.1% in patients younger than 55 years to 9% in patients older than 80 years [23].Additionally, vascular accesses such as arterio-venous graft and permanent tunnelled central venous catheter (PC) present an increased risk of local or systemic thromboembolism [24][25][26][27].However, only few studies have been published on the control of OAT in end stage renal disease (ESRD) and dialysis [21,28,29].There is to date no study evaluating phenprocoumon in haemodialysis patients.The aim of this study was to analyse the quality of OAT control in a chronic haemodialysis population treated with phenprocoumon and to document potential effects on thromboembolic and bleeding events.

Methods and study population
We conducted a cohort study including all patients who were on chronic haemodialysis at the University Hospital of Basel on 3 June 2008 (reference date).Patients quitting the program before or entering it after this date were not included.Data about oral anticoagulation control were collected retrospectively for 12 months preceding the reference date and prospectively for 12 months following the reference date (fig.1).Initiated primarily as a quality control survey, the study was performed without approval by the institutional review board and informed consent was not obtained from the study participants.Prior to data analysis this mistake was disclosed and reported to the Ethical Committee of both Cantons of Basel.The Ethical Committee has examined the issue and reviewed the study protocol on 2-23-2011.No ethical concerns have been raised apart from the failure to submit timely.

Data collection
In the dialysis programme at the University Hospital Basel all clinical data are prospectively and continuously collected in standardised flow sheets and in medical records.Baseline characteristics, indication for OAT, INR target range and INR measurements in the retrospective dataset were abstracted from the medical records by a single trained researcher (RG).The same researcher prospectively collected INR measurement and outcome data of the secondary endpoints for the prospective dataset.Preceding the statistical analysis, all data points were re-evaluated by an independent physician (CP).Both researchers had no influence on treatment decisions.The study was unblinded and the treating medical care team informed about the study.Co-morbidities were reported and defined as follows: diagnosis of diabetes mellitus (DM) with prescription of either oral antidiabetics or insulin, or an entry in the medical record of DM according to current diagnostic criteria [30].Diagnosis of coronary artery disease (CAD) was based either on a positive stress test, a positive coronary angiography, an aorto-coronary bypass or an entry in the medical record of an acute coronary event.Peripheral arterial disease (PAD) was defined by duplex ultrasound, angiography, a history of percutaneous angioplasty or bypass surgery, or an appropriate clinical event in the medical history.Diagnosis of an ischaemic cerebrovascular accident (CVA) was based on a medical record of a clinical event.Vascular disease was used to summarise any history of CAD, PAD and CVA.Active malignancies were based on a histological diagnosis.The diagnosis of autoimmune disease was based on the decision of the "Interdisciplinary Vasculitis Board" of the University Hospital Basel.Underlying renal pathologies were summarised in nine categories: Vascular nephropathy, diabetic nephropathy, glomerulonephritis, focal segmental glomerulosclerosis, analgesic nephropathy, polycystic kidney disease, interstitial nephropathy, other and unknown.Medication with platelet aggregation inhibitors (PAI) (acetylsalicylic acid (ASA) or clopidogrel) was recorded.

Measurements and definitions of OAT and INR
Primary endpoint was the percentage of INR values within the target range during OAT maintenance.The number of INR measurements within, above and below target range was recorded.If not defined otherwise for specific clinical reasons, INR target ranges were: for prosthetic mitral valve 2.5 to 3.5, for AF, prosthetic aortal valve, pulmonary embolism or deep venous thrombosis 2.0-3.0 and for PC and arterio-venous graft 2.0-2.5 [31][32][33].Secondary endpoints were patient outcome including the frequency of bleeding (cerebral, gastrointestinal, soft tissue) and thromboembolic events (venous thrombosis, pulmonary embolism, CVA, thrombosis of the arterio-venous fistula, catheter dysfunction).All INR measurements were performed in the central laboratory of the University Hospital Basel.Prothrombin time was measured in citrate plasma using a standardised reagent (Dade ® Innovin, Siemens) with an ISI value of 0.9, and a fully automated clotting detector (STA-R ® Evolution Coagulation Analyzer, Diagnostica Stago Inc.).The test is insensitive to therapeutic levels of Heparin (Heparin concentration <1U per ml citrate plasma).In most patients INR control was part of a weekly performed complete blood count and blood chemistry monitoring.Blood samples for INR measurements were taken after puncture of the vascular access.Subsequently, all of our patients, regardless of the OAT status, received intravenous low molecular weight heparin (enoxaparin (Clexane ® ), 60 U per kg bodyweight) to avoid clotting of the dialysis filter as reported previously [34].In view of the fact that a low intake of vitamin K may cause unstable OAT, all of our patients on haemodialysis were regularly seen by dieticians [35].The importance of regular dietary intake of vitamin K and the pitfalls of vitamin supplements that may contain vitamin K were discussed.Further, patients were instructed to keep vitamin K intake adequate and consistent avoiding excess of vitamin K rich food as well as diets low in vitamin K.The following definitions were used (fig. 2

Relation of OAT control and INR target range
In the retrospective and prospective dataset, the percentage of INR measurements within target range was significantly higher in patients with a target range width of 1.0 (59% and 50%, respectively) than in patients with a target range width of 0.5 (45% and 39%, respectively) (p = 0.04 and p = 0.02, respectively) (table 4).The percentage of INR measurements above target range was lower in patients with an upper limit of 3.0 (17% and 15%, respectively) than in patients with an upper limit of 2.5 (25% and 27%, respectively).However, the difference was not statistically significant (p = 0.5 and p = 0.1, respectively) (table 4).

Patient outcome
Two female patients (one with OAT) were lost to follow-up due to change to another dialysis centre.Of the remaining 109 patients in the prospective dataset, a total of 24 (22%) patients discontinued dialysis.12 (11%) patients received a kidney transplant; none of them had OAT.Twelve (11%) patients died during the observational period.The number of deaths was significantly higher in the OAT group (32% versus 5%, p = 0.001) (table 5).Overall, 10 patients (9%) had 11 thromboembolic events.The percentage of patients with at least one thromboembolic event was 4% in patients with and 11% in patients without OAT.In total, 7 patients (6%) had 9 bleeds.The percentage of patients with at least one bleed was 4% in patients with and 7% in patients without OAT.There was no statistically significant difference in number of bleeding or thromboembolic events between patients with and without OAT (table 5).Five of the nine bleeds occurred in patients who had a PAI therapy (all ASA).None of the patients with bleeds had a dual therapy (ASA + clopidogrel or PAI + OAT) (table 6).

Discussion
This is the first study analysing the quality of OAT control in patients treated with phenprocoumon in a chronic haemodialysis population.[29].Both individual and population based percentage of measurements in target range can be biased by more frequent measurements in patients with difficult adjustment of OAT, resulting in an underestimation of OAT control [37].In our study, we did not find a correlation between individual frequency of measurements and percent of INR in target range.To determine whether a high frequency of measurements per se leads to a better OAT control, a randomized study with predefined frequencies of measurement would be needed.
In our study, in spite of a higher frequency of INR measurements compared to other studies (measurements every 21-31 days [28,29]) the percentage of INR in target range did not exceed 54%.The indication for OAT may influence the percentage of INR measurements in target range both as an independent risk factor and via the height and width of target range [7,38,39].Obviously, as directly shown in our study, patients with a wider target range are more likely to have a high-

Original article
Swiss Med Wkly.2013;143:w13730 Swiss Medical Weekly • PDF of the online version • www.smw.ch2-3) [21,28,29].The absolute height of target range also can influence the percentage of INR measurements in target range.In our study the percentage of INR measurements above target range was lower in patients with a higher compared to patients with a lower upper limit of target range.Physicians are probably more carefully avoiding an INR above a predefined high upper limit of target range [38].Pharmacological differences between oral anticoagulants are another potential influencing factor that needs to be considered.While warfarin, the most investigated oral anticoagulant agent, has a half life of 20-60 hours, the half-life of phenprocoumon is 72-96 hours.The effect of this difference in pharmacokinetics has been discussed controversially [39].The longer half life of phenprocoumon could lead to a more stable blood level and INR in OAT maintenance and contrary to the need of more time to readjust INRs outside the range.Two recent cohort studies that directly compared phenprocoumon and warfarin in non-ESRD patients found that phenprocoumon allowed a better INR con-trol than warfarin [39,40].While Leiria et al. described the percentage of measurements in target range of 60.7% with phenprocoumon and 45.6% with warfarin (p = 0.001) [40], the difference in time in therapeutic range described by Jensen et al. was minimal (74% versus 70.2%, p = 0.008) [39].Thus, while there is a lack of consistent studies directly comparing the two agents, these data suggest that with respect to OAT control phenprocoumon seems to be at least not inferior to warfarin.Since OAT withdrawals and adjustment phases can lead to a substantial underestimation of OAT control, we studied OAT control for both overall exposure and OAT maintenance.In our study, the percentage of INR in target range in OAT maintenance was 5-10% higher than in overall OAT exposure.
Previously published studies on OAT in the general population have shown that the incidence of thromboembolic events correlates with the percentage of INR below target range, and the incidence of bleeding events increases with poor INR control, specifically when INR is very high [6,   Data are displayed as counts of events and percentage of exposed patients.# Fisher's exact Test.Due to the fact that repeated bleeds or thromboembolic events were possible in the same patient, numbers do add up to more than 100% (one patient with a second soft tissue bleed, one patient with gastrointestinal bleeding and posttraumatic intracerebral bleeding (medication with acetylsalicylic acid), one patient with a second ischaemic CVA; all in the No OAT group);.a one patient died from cardiac arrest, two patients died from acute infection with underlying severe peripheral arterial disease and previous limb amputation.For 5 patients who died out of hospital the definitive cause of death was indeterminate, and unfortunately no autopsy was performed; b two patients died after withdrawal of dialysis, one patient died from a metastatic small cell lung carcinoma, one patient died after repeated cerebrovascular insults; c INR at event: 1.5; d three strokes (two in the same patient) and one transient ischaemic attack; e one major bleeding event, INR at event: 7.7, no platelet aggregation inhibitor involved; f all major bleeding events.Definition of CVA, minor and major bleeding events as described in the RE-LY study protocol [52]., 41, 42].In our study, the number of patient years and thromboembolic or bleeding events observed was not sufficient to allow a correlation with INR control.However, we did not find evidence for excessive bleeding in patients with OAT as described by some authors for ESRD patients [16,[43][44][45].The higher number of bleedings in the patients without OAT in our study may be partly explained by the high prevalence of PAI in this group and the close INR monitoring in the OAT group.Recent studies indicate that while the efficacy of OAT for stroke prevention in patients with AF is superior to PAI, the bleeding risks of moderate dose OAT (target INR 2-3) and of PAI treatment might be similar [46,47].
The number of deaths was significantly higher in the OAT group, but only one of these deaths was directly associated with OAT.Although the patient characteristics did not significantly differ, many patients on OAT are part of a high risk population due to the medical condition requiring OAT, e.g.AF and mechanical heart valve replacement.Consistently, no patient in the OAT group was scheduled for a kidney transplant.Therefore, the higher number of deaths might not be attributed to the OAT but rather to the higher mortality risk of the exposed patient group.This study has several limitations: The number of patients under OAT was relatively small, limiting the power of the investigation.Additionally, we were not able to report on influencing factors such as dietary vitamin K intake and drug-drug interactions or on genetic factors influencing response to OAT [48][49][50][51].

Conclusions
While for the general population the new generation of oral anticoagulants might simplify OAT in the near future, its application in ESRD patients will be limited due to the lack of studies and/or to altered pharmacokinetics.Therefore, OAT with vitamin K antagonists will remain a basic strategy for therapy and prevention of thromboembolic events in ESRD.To allow informed treatment decisions, our results have to be confirmed in larger studies powered for analysing the factors influencing INR control and the impact of INR control on bleeding and thromboembolic events in ESRD patients treated with phenprocoumon.

Figures (large format)
Study design.OAT: Oral anticoagulation therapy

Figure 1
Figure 1 Study design.OAT: Oral anticoagulation therapy

Figure 2
Figure 2 Definition of oral anticoagulation therapy (OAT) exposure, adjustment phases and OAT maintenance.OAT: oral anticoagulation therapy.OAT exposure: time from the first dose of OAT to the end of OAT (temporary or definitive).Adjustment phase: time from first INR at start/restart of OAT to the 2nd INR value within target range.Intended withdrawal: time of temporary suspension of OAT (e.g., for scheduled surgery).OAT maintenance: time from 2nd INR value in target range after start/ restart to end of OAT (temporary or definitive).

Width of INR target range 0. 5 a
Width of INR target range 1.0 p-value°% in target range in retrospective dataset (n = 13 / n = 12) c 45 (17-71) 59 (33-74) 0.04 % in target range in prospective dataset (n = 10 / n = 14) c 39 (21-56) 50 as median and (range).°p-value of difference between groups in Mann-Whitney-U-Test; *OAT maintenance: Time from 2nd INR value in target range after start/restart to end of OAT (temporary or definitive) with normalisation of INR; a includes all patients with a target range of 2.0-2.5, 2.5-3.0 or 3.0-3.5;b includes all patients with a target range of 2.0-3.0 or 2.5-3.0;c number of patients in left/right column.

Figure 2
Figure 2 Definition of oral anticoagulation therapy (OAT) exposure, adjustment phases and OAT maintenance.OAT: oral anticoagulation therapy.OAT exposure: time from the first dose of OAT to the end of OAT (temporary or definitive).Adjustment phase: time from first INR at start/restart of OAT to the 2nd INR value within target range.Intended withdrawal: time of temporary suspension of OAT (e.g., for scheduled surgery).OAT maintenance: time from 2nd INR value in target range after start/restart to end of OAT (temporary or definitive).
[28]28,29] percentage of INR measurements in the target range was around 50% and comparable with previously published data on OAT control with warfarin and acenocoumarol in ESRD (37-50% in tar-get range)[21,28,29].In a secondary analysis by Limdi et al. of a prospective cohort study on OAT with warfarin in 53 ESRD patients, 40% of INR measurements were within the target range of 2.0-3.0[28].In a retrospective cohort study byTo etal.on 155 patients on haemodialysis, 11 patients receiving warfarin had INR values in the therapeutic range 50% of the time [21].In a retrospective study by Gompou et al. on INR deviations in 11 haemodialysis pa-

Table 1 :
Indication for oral anticoagulation therapy (OAT) and INR target range overall and by gender. of INRs in target range than patients with a narrower target range.Similar results could be seen by indirect comparison of studies with different target ranges.In the study by Gompou et al. the percentage of INR meas-urements was 37%, with an INR target range width of 0.5 (INR 2.0-2.5),while in the studies by Limdi et al. and To et al. the percentages of INR measurements in target range were 40% and 50% with a target range width of 1.0 (INR Included were all patients who were exposed to OAT in the retrospective and/or the prospective dataset.Data are displayed as counts and percentage (%) of exposed patients.°p-value of difference between male and female patients, Fisher's exact Test; a one patient with target INR 2.5-3.0;bonepatient with target INR 3.0-3.5;conepatient with myocardial dyskinesia with target INR 2.0-3.0,onepatient with vasculitis and repeated arterial occlusions with target INR 2.0-2.5, one patient with metastatic cancer with target INR 2.0-3.0.Swiss Medical Weekly • PDF of the online version • www.smw.cherpercentageIncluded were all patients who were exposed to OAT in the retrospective and/or the prospective dataset.Data are displayed as median [range] or counts and percentage (%).*Mann-Whitney-U-Test; # Fisher's exact Test; °p-values: comparing OAT and no-OAT; a vascular disease: patient suffered at least from one of CAD, PAD, CVA; b includes vesicoureteral reflux (n = 5), nephrectomy due to renal cell carcinoma (n = 3), multiple myeloma (n = 2), cystinosis (n = 2); PAI = platelet aggregation inhibitors; ASA = acetylsalicylic acid.
Data are displayed as median and (range).OAT exposure: time from first dose of OAT to end of OAT (temporary or definitive).Adjustment phase: time from first INR at start/restart of OAT to the 2nd INR value within target range.OAT maintenance: Time from 2nd INR value in target range after start/restart to end of OAT (temporary or definitive) with normalisation of INR. % of INR measurements in target range: Number of measurements within the clinical defined target range divided by number of performed measurements.a Time under OAT (days)/number of measurements; b number of episodes [new beginnings + restart after intended withdrawal].

Table 4 :
INR target range and oral anticoagulation therapy (OAT) control during OAT maintenance*.

Table 6 :
Number of bleeding events stratified by platelet aggregation inhibitors (PAI) and oral anticoagulation therapy (OAT) status.
#Number of bleeding events/ total number of exposed patients and percent (); # Fisher's exact Test.ASA = acetylsalicylic acid; °p-value of difference of number of bleedings between OAT and no-OAT group; a two bleeding events in the same patient.Swiss Medical Weekly • PDF of the online version • www.smw.ch