Quality of oral anticoagulation with vitamin K antagonists in ‘real-world’ patients with atrial fibrillation: a report from the prospective multicentre FANTASIIA registry Free

Abstract

Aims

The efficacy and safety of oral anticoagulation (OAC) using the vitamin K antagonists (VKA) are closely associated with the quality of anticoagulation, reflected by time in therapeutic range (TTR). The SAMe-TT₂R₂ is a risk score developed to predict the quality of anticoagulation control among VKA users. To analyse the quality of anticoagulation and its clinical determinants based on different methods in a prospective cohort of atrial fibrillation patients on VKA treatment participating in the multicentre Spanish observational registry FANTASIIA.

Methods and results

Estimated TTR was calculated from Rosendaal, direct method, international normalized ratio variability, and NICE criteria. Time in therapeutic range values were compared for those patients with a SAMe-TT₂R₂ score 0–2 and >2. One thousand four hundred and seventy patients were analysed (56.4% male, mean age 74.1 ± 9.5 years). Mean TTR was 61.5 ± 25.1 with Rosendaal and 64.7 ± 24.2 with direct method. There was a high correlation between both methods (ρ = 0.805). The prevalence of poor anticoagulation control was 55%. Diabetes mellitus [odds ratio (OR) 1.38; P = 0.008], peripheral artery disease (PAD, OR 1.62; P = 0.048), and HAS-BLED (OR 1.13; P = 0.022) were independently associated with TTR < 70%. SAMe-TT₂R₂ score 0–2 had a higher mean TTR than patients with SAMe-TT₂R₂ >2 (P = 0.044), with a specificity of > 90% for predicting TTR < 70%. Patients with TTR < 70% had higher risk of events (21.7 vs. 16.8%; P = 0.021).

Conclusion

In a multicentre prospective registry, 55% of AF patients had poor anticoagulation control with diabetes mellitus, PAD, and HAS-BLED being independently associated with TTR < 70%. A high SAMe-TT₂R₂ scores had a high specificity for predicting a TTR < 70% as an indicator of poor quality anticoagulation.

Introduction

The efficacy and safety of oral anticoagulation (OAC) using the vitamin K antagonists (VKA) are closely associated with the quality of anticoagulation control, as reflected by the average percentage of the time in therapeutic range (TTR). Indeed, therapeutic anticoagulant effect of VKAs has high inter- and intra-patient variability.¹ Many clinical factors influence TTR, and the more common ones have been used to formulate the SAMe-TT₂R₂ score [sex female, age < 60 years, medical history (more than two co-morbidities), treatment (interacting drugs), tobacco use (doubled), and race (doubled)], which is a simple clinical risk score developed to predict the quality of anticoagulation control among VKA users.²

Several methods have been proposed to define how good is the quality of anticoagulation. The most commonly used method is the linear interpolation method proposed by Rosendaal, but this is not simple to calculate.³ Another simpler proposed method was the percentage of international normalized ratio (INR) in the therapeutic range (PINRR). The latter is an intuitive method but is not the most widely accepted.⁴ International normalized ratio variability calculated as the standard deviation around the mean INR value by Fihn et al.’s⁵ modified method is also another method to assess the quality of anticoagulation therapy. Various ways to evaluate the quality of anticoagulation treatment have also been proposed by National Institute for Health and Care Excellence (NICE) guidelines.⁶

The recent European Consensus document and European Society of Cardiology (ESC) clinical guidelines recommended an individual average of TTR of 70% to maximize the efficacy and safety of VKA treatment. Atrial fibrillation (AF) patients with an individual TTR of 70% or greater having a significantly lower incidence of complications overall compared with these patients with lower TTRs.⁷

Although the non-VKA oral anticoagulants (NOACs) are increasingly used as the first-line treatment option in patients with AF, many health care systems are currently unenthusiastic to implement a first-line strategy with NOACs due to the higher costs.⁸ Indeed, recent studies⁹ have shown that VKAs with high TTR could be as efficacious as NOACs, given that the main benefits of NOACs compared with warfarin may be only marginal in those patients with high TTR, although a reduction in intracranial haemorrhage is still evident.¹⁰

The aim of our study was to analyse the clinical determinants of quality of anticoagulation and to explore the predictive role of anticoagulation control based on different definitions of ‘good anticoagulation control’, in a prospective cohort of AF patients on VKA treatment participating in the multicentre Spanish observational registry FANTASIIA.

Methods

FANTASIIA (Spanish Acronym for ‘Fibrilación Auricular: influencia del Nivel y Tipo de Anticoagulación Sobre la Incidencia de Ictus y Accidentes hemorrágicos’) registry is an observational, prospective, national, and multicentre study of clinical and demographical characteristics of Spanish AF patients. The study design of FANTASIIA registry has been described previously.¹¹ In brief, the main objective is to assess the incidence of thromboembolic and bleeding events in an unselected population of patients with AF; specifically, the type of oral anticoagulant (VKA or NOACs) and the quality of anticoagulation with VKAs.

Study population

We prospectively included all outpatients with a confirmed diagnosis of AF. Patients were excluded if they were younger than 18 years old, had history of any heart valve disease (rheumatic valve disease, moderate–severe valve disease, prosthesis valve, or mitral valve repair surgery), hospitalization at the moment of inclusion, or were included in another clinical trial. All subjects provided signed informed consent. The study was conducted according to the ethics principles of Declaration of Helsinki and Good Clinical Practice Guidelines and was approved by the Clinical Research Ethics Committee at Hospital Universitario de San Juan (Spain) with the approval number 12/220 and by the Spanish Agency of Medicine and Health Products as a prospective follow-up post-authorization study with the approval number SEC-ACO-2012-01.

Data collection

Clinical and demographic data were collected from all AF patients included. We included demographic information, data of cardiovascular factors, and co-morbidities. Stroke risk was calculated using the Congestive heart failure, Hypertension, Age ≥ 75 years, Diabetes mellitus, Stroke or transient ischaemic attack (CHADS₂ score) and Congestive heart failure, Hypertension, Age ≥ 75 years, Diabetes mellitus, Stroke or transient ischaemic attack, Vascular disease, Age 65–74 years and Sex category (female) (CHA₂DS₂-VASc score). Bleeding risk was calculated using the Hypertension (uncontrolled systolic blood pressure > 160 mmHg), Abnormal renal and/or liver function, previous Stroke, Bleeding history or predisposition (anaemia), Labile INR (only applies to a VKA user; not applicable for a non-VKA user), Elderly (age ≥ 65 years), and concomitant Drugs (antiplatelet or non-steroidal anti-inflammatory drugs) and/or alcohol excess (HAS-BLED score). The control of anticoagulation therapy was performed by two different organizational systems: general practitioners (GPs) or haematologists. In the first case, the control of anticoagulation therapy with VKAs was developed completely in the primary health centres by GPs. In the second case, the haematologist was the main responsible for anticoagulation therapy in the anticoagulation clinics.

Quality of anticoagulation

For patients treated with VKA, coagulation status was determined by monthly INR values 6 months before the study inclusion, 6 months after baseline, and at 1 year of follow-up. The quality of anticoagulation control was assessed by four methods: direct method (PINRR), Rosendaal method, variability of INR, and National Institute for Health and Care Excellence (NICE) clinical guidelines anticoagulation quality. Direct method or percentage of INRs within therapeutic range (PINRR)⁴ evaluates the percent of visits in the range looks at how many visits had INR result in the therapeutic range (for AF patients INR between 2.0 and 3.0) and divides this by the total number of visits. With the Rosenddal linear method,³ TTR is calculated by incorporating INR measurement frequency and values, assuming changes between consecutive INR measurements are linear, but interpolation of values not measured may create gaps in credibility. Another method is the variability of INR measured by Fihn’s method modified. The last method reflects the degree of variation (variance growth rate) from the target INR over a period of time.⁵ One more method is the one proposed by NICE guidelines. On the basis of the NICE guidelines,⁶ poor anticoagulation control is defined if AF patient shows any of the following values: 2 INR values higher than 5 or 1 INR value higher than 8 within the past 6 months and 2 INR values less than 1.5 within the past 6 months or TTR < 65%. However, for ESC guidelines, poor anticoagulation control was defined as an estimated time spent in the therapeutic INR range below 70%.

To predict poor INR control with VKA, SAMe-TT₂R₂ score was also calculated.² Results from both direct and Rosendaal methods were determined for each value of SAMe-TT₂R₂.

Follow-up and clinical outcomes

Follow-up started the day of the inclusion for 12 months. Thromboembolic events were classified as central nervous system (CNS) or non-CNS embolic events. Death was classified as a cardiovascular event (acute coronary syndrome, heart failure, lethal arrhythmia or sudden death, and artery aneurysm rupture or stroke) or another non-vascular death. Major bleeding events were defined according to the 2005 International Society of Thrombosis and Haemostasis criteria: fatal bleeding or symptomatic bleeding in a critical anatomical site (intracranial, intraspinal, intraocular, retroperitoneal, intraarticular, pericardial, or intramuscular with compartment syndrome) and/or bleeding causing a fall in haemoglobin ≥ 20 g/L, or transfusion of ≥2 units of packed red blood cells. Clinically significant events included the composite of stroke, major bleeding, all-cause mortality, acute coronary syndrome, and acute heart failure.

Statistical analyses

Quantitative variables were described using the mean ± standard deviation or median (interquartile range). Categorical variables are expressed as percentages. Normal distribution of continuous variables was tested with the Kolmogorov–Smirnov method. For between-group comparisons, we used the Student’s t-test for continuous variables and the χ² test for qualitative variables. Logistic regression analyses were employed for univariate analyses and for multivariate adjustment. Multivariate models were performed including variables with recognized clinical relevance with VKA control and those with a P-value <0.15 in the univariate analysis. Results are presented as odds ratios (ORs) and 95% confidence intervals (95% CIs). The specificity and sensitivity of SAMe-TT₂R₂ score > 2 for predicting a TTR < 70% were determined. Statistical significance was defined as P < 0.05. Statistical analyses were performed with SPSS statistical package version 22.0 for Windows (SPSS Inc, Chicago, IL, USA) and STATA.

Results

We enrolled 2178 patients in FANTASIIA registry (1648 patients treated with VKAs and 530 patients treated with NOACs). One thousand nine hundred and thirty-eight patients (89% of the total sample) completed 1 year of follow-up (1470 patients with VKAs and 468 patients with NOACs). For the present analysis, we selected all patients treated with VKAs with complete follow-up, i.e. 1470 patients (75.9%; mean age 74.1 ± 9.5 and 56.4% male). Clinical and demographic baseline characteristics classified according to the quality of anticoagulation control (poor quality defined as TTR < 70%) are summarized in Table 1. Patients with TTR < 70% had more prevalent cardiovascular risk factors, e.g. diabetes, chronic kidney disease (CKD), or peripheral artery disease (PAD) (all P < 0.05) as well as a low level of education (P = 0.019). Patients with TTR < 70% had higher mean CHA₂DS₂-VASc (3.8 ± 1.6 vs. 3.6 ± 1.6; P = 0.014) and HAS-BLED scores (2.1 ± 1.1 vs. 1.9 ± 1.0; P = 0.002), respectively.

Quality of anticoagulation control

All patients had INR determinations 6 months previously to the inclusion visit. By the Rosendaal method, the mean TTR was 61.5 ± 25.1%. Using the direct method (PINRR), the mean TTR was higher, 64.7 ± 24.2%. There was a high correlation between Rosendaal and direct method [ρ = 0.805, 95% CI (0.786–0.822); P < 0.05].

With regard to the quality of anticoagulation control, 59.1% of the population following ESC criteria (TTR < 70%) and the 52.5% of the population following NICE criteria had poor anticoagulation control. The distribution of quality of anticoagulation control according different methods is shown in Table 2. There was a progressive and significant decline in mean TTR with different AF risk scores (CHADS₂, CHA₂DS₂-VASc, HAS-BLED, or SAMe-TT₂R₂) as shown in Supplementary material online, Table S1.

Clinical factors related with poor anticoagulation control (TTR < 70%) on univariate analysis are shown in Supplementary material online, Table S2. After multivariate adjustment, variables independently associated with poor anticoagulation control were diabetes mellitus [OR 1.38 (95% CI 1.09–1.75; P = 0.008)], PAD [OR 1.62 (95% CI 1.00–2.63; P = 0.048)], and HAS-BLED score [OR 1.13 (95% CI 1.02–1.25; P = 0.022)].

SAMe-TT₂R₂ and time in therapeutic range levels

A progressive reduction of TTR was found with increasing SAMe-TT₂R₂ scores (Table 3). There was a significant reduction in mean TTR values for patients with scores 0–2 and >2 with both methods: Rosendaal (61.9 ± 25.3 vs. 57.8 ± 22.9%; P = 0.044) and direct method (65.2 ± 24.3 vs. 60.6 ± 22.5%; P = 0.007).

Linear regression analyses show that SAMe-TT₂R₂ was inversely associated with TTR [standardized beta: −2.22 (95% CI −3.55 to  − 0.90); P = 0.001] with an associated linear relationship between SAMe-TT₂R₂ and TTR. A SAMe-TT₂R₂ score >2 was associated with a specificity of >90% (90.03%) for predicting a TTR < 70% and a sensitivity of 11.6%. We observed an increase in positive predictive value (PPV) to poor anticoagulation control with increasing SAMe-TT₂R₂ scores (see Supplementary material online, Table S3).

Clinical outcomes

After a median follow-up period of 369 (363–385) days, we analysed clinical outcomes related with poor anticoagulation control. During the follow-up, 13 patients had stroke (0.9% per year), 50 patients had major bleeding events (3.4% per year), and 69 patients died (4.7% per year). Patients with poor TTR had higher risk of the composite of clinically significant events (21.7 vs. 16.8%; P = 0.021) (Tables 4 and 5).

Discussion

In a multicentre prospective FANTASIIA registry, approximately 55% of AF patients had poor anticoagulation control with diabetes, PAD, and HAS-BLED being independently related with poor anticoagulation control. There was a high correlation between Rosendaal and direct method to assess anticoagulation quality. Indeed, we observed a correlation between SAMe-TT₂R₂ score and TTR. A high SAMe-TT₂R₂ score (>2) had a high specificity for predicting TTR < 70% as an indicator of poor quality anticoagulation in patients on VKA.

In patients treated with VKAs is imperative to assess the quality of anticoagulation control. Different clinical trials have shown a strong correlation between high TTR and low thromboembolic and bleeding outcomes. In a large-scale international observational AF patient registries such as GARFIELD-AF,¹² the mean of TTR was 55.0% and 58.9% of the patients had poor anticoagulation control. Similar findings reported in PREFER-AF^13,14 registry, where 72.1% of the patients had adequate INR control. In our results, we observed high rate of poor anticoagulation control, with 59.2% of AF patients with TTR < 70% assessed by Rosendaal method and a median of 61.5 ± 25.1% and a median of 64.7 ± 24.2% by direct method. The PINNR (i.e. direct method) is intuitive but is not the most widely accepted in previous registries, because the TTR measured by Rosendaal and PINNR are not interchangeable even if they correlate.⁴ In our registry, we observed very good correlation between both methods (ρ = 0.805), so both methods could be used indistinctly in daily practice.

The cut-off definition of ‘good anticoagulation control’ varies between 65% of NICE guidelines and 70% in European guidelines.⁷ Hylek¹⁵ showed that AF patients with poor control defined as TTR < 60% had higher rates of mortality and major bleeding compared with the control group (P < 0.001). Björck et al.⁹ analysed a total of 40 449 Swedish patients and compared the rates of clinical outcomes in AF patients well managed with TTR > 70% or below, with low risk of complications in a well-managed group. We also reported a significant difference in composite adverse outcomes with high events in patients with TTR < 70% (21.7 vs. 16.8%; P = 0.021). In our registry, an increase of 6.6% of poor anticoagulated patients was observed when the cut-off point of TTR changed from 65% to 70%. Recently, ESC guidelines⁷ change the cut-off point of good anticoagulation control with VKAs from 65% to 70% and treatment with VKA with high TTR (≥70%) could be as effective as NOACs in preventing adverse outcomes except for the rate of intracranial bleeding.^9,16

Several factors are related with poor anticoagulation control. In our study, diabetes mellitus, PAD, and HAS-BLED score were independently related with poor anticoagulation control. Classical factors such as age, hypertension, use of antiplatelet agents, or low education level are related with poor TTR and adverse outcomes.¹⁵ In the FANTASIIA registry, 55% of AF population treated with VKA had poor anticoagulation control (with TTR < 70%). Many reasons had been proposed for lack of adherence and difficulties for maintaining an optimal INR: discontinuation of treatment during admissions or concomitant diseases, lack of patients’ awareness of the AF-associated stroke risk; perception of aspirin as being protective against AF-related stroke, low cognitive function, or dementia leading to inappropriate dose administration; or underdosing for fear of bleeding.¹⁷ Physicians should identify patients with VKAs treatment at risk of poor TTR and develop intensive measures to correct modifiable risk factors related with variability of TTR (i.e. uncontrolled hypertension) and to improve patient education. Indeed, the evaluation of potential pharmacological interactions and the implementation of active interventions to improve anticoagulation control and to recognize the lack of adherence to VKAs is essential. If a TTR ≥ 70% is not achieved despite all these improvement measures, treatment with NOACs should be started to avoid ischaemic and bleeding events in AF patients.

Perhaps the most interesting practical issue is to identity which patients could be well with VKAs therapy and achieve good TTR to avoid the trial of VKA or ‘warfarin stress test’ in patients who could be not well with VKAs. Some practice a ‘trial of warfarin or warfarin stress test’ to see whether a patient can achieve a high TTR or not prior to considering the treatment with NOAC. But this ‘trial with warfarin’ may put these AF patients at risk of thromboembolic or bleeding evens during the initial period of treatment.¹⁸ In many countries where economic considerations are an important factor, it is mandatory to start anticoagulation therapy of AF patients with VKA and only if they do not have good TTR after 6 months of treatment, then it is possible to switch to NOACs.¹⁹ Time in therapeutic range measured during the first 6 months of initiation of VKA treatment is lower than TTR measured after the warfarin inception period. A large study based in the Veterans Health Administration reported a TTR of 48% in the first 6 months of therapy compared to 61% for the warfarin experienced cohort²⁰ and patients initiating warfarin had two-fold increased risk of ischaemic stroke in the first 30 days of use. Indeed, this risk was highest in the first week of warfarin initiation.²¹ To avoid the trial of VKAs and to predict which patients do well with VKAs, Apostolakis et al.² developed SAMe-TT₂R₂ score. This simple clinical score has been validated in several prospective and retrospective AF cohorts.

To our knowledge,¹⁷ this is the first time that SAMe-TT₂R₂ score is validated prospectively collected multicentre cohort of patients treated with acenocumarol. We observed a significantly decrease of TTR related with high values of SAMe-TT₂R₂ (P = 0.005). SAMe-TT₂R₂ score 0–2 had a higher mean TTR than patients with SAMe-TT₂R₂ >2 (P = 0.044). The latter was associated with a specificity of >90% (90.03%) for predicting a TTR < 70%, highlighting the importance to identify patients with SAMe-TT₂R₂ >2 as an indicator of poor quality anticoagulation. The linear regression analysis showed that SAMe-TT₂R₂ was inversely associated with TTR. For that reason, SAMe-TT₂R₂ score could be a useful daily clinical tool that complements previous ischaemic and bleeding scores. Nonetheless, risk stratification is not a static on–off process but a dynamic process, and all patients should be regularly reviewed to check if their risk has changed and to correct the modifiable factors.

In this regard, the decision-making process for stroke prevention in AF patients in daily clinical practice needs to be simple for clinicians. The ‘Birmingham 3-step’ management pathway developed by Lip²² emphasizes that ‘simplicity is the best’. The ‘first step’ is to identify low-risk patients with CHA₂DS₂-VASc score (0 in male and 1 in female) to avoid antithrombotic therapy in them. The ‘second step’ is to consider stroke prevention with oral anticoagulants in the rest of patients and try to correct modifiable risk factors in those with HAS-BLED >3. And the ‘third step’ is to choose between NOAC or VKA treatment with TTR > 70%. For this purpose, SAMe-TT₂R₂ is imperative: patients with SAMe-TT₂R₂ score 0–2 are likely to do well on VKA treatment, whereas those with SAMe-TT₂R₂ score > 2 are less unlikely to achieve TTR > 70% so it could be necessary additional measures to improve INR control or select an NOAC as anticoagulation treatment.

Limitations

The first are the inherent limitations of observational studies. The second is related with the anticoagulation treatment. The most common VKA used in Spain is acenocoumarol and shows a shorter half-life than warfarin (10 h vs. 36 h) but without differences on the time on therapeutic range. It could be a limitation with the comparison of other validation studies of the SAMe-TT₂R₂ score with warfarin treatment. Patients are representative of a Spanish population and the results might not be extrapolated to other countries.

Conclusions

In a multicentre prospective registry, approximately 55% of AF patients had poor anticoagulation control with diabetes mellitus, PAD, and HAS-BLED being independently associated with poor anticoagulation control. There was a high correlation between Rosendaal and direct method. A high SAMe-TT₂R₂ scores had a high specificity for predicting a TTR < 70% as an indicator of poor quality anticoagulation.

Supplementary material

Supplementary material is available at Europace online.

Funding

The FANTASIIA registry was funded by an unconditional grant from Pfizer/Bristol-Myers-Squibb and by grants from the Instituto de Salud Carlos III (Madrid)-FEDER (RD12/0042/0068, RD12/0042/0010, RD12/0042/0069 and RD12/0042/0063).

The authors are supported by RD12/0042/0049 (RETICS) from ISCIII and PI13/00513/FEDER from ISCIII. Fundación Séneca (19245/PI/14), Instituto Murciano de Investigación Biosanitaria (IMIB16/AP/01/06). José Miguel Rivera-Caravaca has received a grant from Sociedad Española de Trombosis y Hemostasia (SETH; grant for short international training stays 2016).

Conflict of interest: none declared.

References

Author notes