Poor anticoagulation control in patients taking warfarin at a tertiary and district-level prothrombin clinic in Cape Town, South Africa.
Warfarin is the most widely used oral anticoagulant for primary and secondary stroke prevention  and is the only vitamin K antagonist (VKA) available in SA.  Alternative oral anticoagulants are not routinely available in public sector healthcare facilities owing to their high cost.  Aspirin is a poor alternative in patients with AF, as it is much less effective at preventing cardioembolic events. [7,10] Warfarin has unpredictable pharmacokinetics and dynamics, and requires individualised dosing to achieve optimal anticoagulation. Warfarin has a narrow therapeutic range, placing patients at risk of bleeding if the target is exceeded and at risk of thromboembolic complications if subtherapeutic. [5,10] Warfarin is a leading cause of adverse drug reaction (ADR)-related medical admissions in SA. 
Time in the therapeutic range (TTR) is a calculation that reflects the duration of time in which a patient's international normalised ratio (INR) values were within the desired range and is used to evaluate the effectiveness of warfarin therapy. For effective anticoagulation, patients on warfarin should achieve an INR in the therapeutic range for =65% of the follow-up time.  Patients with TTR <65% have reduced warfarin efficacy.  Low TTR is also associated with an increased risk of both bleeding and thromboembolic complications.  Two multinational multicentre clinical trials (ACTIVE W and RE-LY) that included patients from SA found poor INR control in SA patients receiving warfarin. [10,12]
We quantified anticoagulation control over a 6-month period for patients on long-term warfarin managed at two dedicated INR clinics in Cape Town, SA. Specific objectives were to determine the proportion of patients achieving a TTR =65%, and to identify predictors of adequate control (TTR =65%).
We reviewed folders of patients attending the warfarin anticoagulation monitoring clinics at Mitchell's Plain Community Health Centre (MPC) and Groote Schuur Hospital (GSH) in Cape Town. These outpatient clinics assess the INR at patient visits and adjust the warfarin dose in response to the INR result. At GSH, INR results are available within hours and dose modification occurs immediately. At MPC, INR results are available the next day and dose adjustment instructions are given to patients telephonically. Both facilities use the same, unvalidated dosing algorithm to guide dose adjustments.
We identified patients who were on long-term warfarin therapy, which we defined as at least 2 years on warfarin, after an initial 3-month dose-stabilisation period to achieve the required INR. The INR monitoring results were reviewed, and we excluded patients with <6 months of consecutive INR data and those who visited both or other sites and had >25% of follow-up visits at an alternative site. We also excluded patients who had >70 days between INR quantification.  Patient follow-up at clinics occurs at least every 2 months (56 days) and we allowed for patients to be 14 days late for routine follow-up. We excluded patients with larger gaps between INRs as this may affect the accuracy of TTR quantification and under- or overestimate the TTR. 
We powered the study to detect a 15% difference in the mean TTR between the two sites (with [alpha]=0.05 and 80% power) and calculated that we would need 173 patients from each site for a total sample size of 346.
We calculated the TTR using the widely accepted Rosendaal method.  The calculation assumes a linear relationship (increase or decrease) between consecutive INR values to determine the proportion of time within the therapeutic range.  We calculated TTR over 6 months (180 days) and excluded the first 90 days of regular INR monitoring data from the TTR calculation, to allow patients newly initiated on warfarin to achieve stability. TTR was therefore calculated starting from the first INR result after the 90-day window. The target INR range for patients taking warfarin for AF and most other indications is 2.0-3.0. The target INR range for VHD is 2.5-3.5.  The therapeutic range for the patient's clinical condition was used as the target for that patient in the TTR calculation. In patients with an unknown indication for warfarin therapy, we assumed a target INR range of 2.0-3.0. We defined poor control as TTR <65% over the 180-day period, and good control as TTR =65%. 
We summarised continuous variables as means (standard deviation (SD)) if normally distributed and as medians with interquartile ranges (IQRs) if abnormally distributed, and used the Wilcoxon rank-sum test for between-group comparisons of continuous variables. Univariate associations between categorical variables were explored using the .2 test. We constructed a multivariate logistic regression model of associations with good anticoagulation (TTR =65%), and included age, sex, site and target INR in the model based on an a priori decision. Age was split into two groups at the median and included in the model as a binary variable. A p-value <0.05 was considered statistically significant. Stata SE 13.1 (StataCorp, USA) was used for the analyses.
The study was approved by the Human Research Ethics Committee at the University of Cape Town (ref. no. 658/2014) and the Western Cape Health Research Committee (ref. no. WC_2015RP8_111). Permission to conduct the research was given by hospital management of MPC and GSH. The study was conducted in accordance with the Declaration of Helsinki (last updated 2013)  and the Guideline for Good Clinical Practice. 
We screened the clinical records of 949 patients, of whom 586 were excluded and 363 were included in the analysis (Fig. 1). Screening took place in September 2014. INR data were reviewed between 2009 and 2014, and all participants with more than 27 months of INR data were included in the study.
Patient characteristics are set out in Table 1. The most common indications for warfarin therapy were AF (25.1%) and VHD (45.7%). Other indications were pulmonary embolus, venous thromboembolism, systemic lupus erythematosus, hypercoagulable states and atrial flutter. An indication was not documented in 10.5% of patients.
The mean (SD) TTR was 47.0% (24.0%) and did not differ significantly between sites (Fig. 2). In 272/363 patients (74.9%) the
TTR was <65%. Poorly controlled patients had more frequent INR monitoring than those with good control, with a median of 8 readings (IQR 6-10) and 6 readings (IQR 5.0-8.0) over the 6-month period, respectively (rank-sum p<0.0001). Patients aged <55 years had a mean TTR of 43% (95% confidence interval (CI) 40-47) compared with a mean of 50% for those aged =55 years (95% CI 46-53) (p=0.0105). In the multivariate analysis, age =55 years was associated with better INR control (Table 2).
The majority of patients in our study had poor INR control--only 25.1% had a TTR =65%. Older age was associated with better control, but even in patients aged =55 years, only 30.6% achieved a TTR of =65% (Table 2). This is a concerning finding. Despite regular follow-up, most patients did not achieve adequate INR control for efficacy and are at risk of warfarin adverse effects.
Even under clinical trial conditions, anticoagulation control of South Africans taking warfarin was poor. SA participants with AF randomised to warfarin in the international multicentre ACTIVE-W trial (dual antiplatelet therapy v. warfarin anticoagulation) had a mean TTR of 46%.  In the RELY trial (randomised evaluation of long-term anticoagulation therapy), which randomised participants with AF to dabigatran or warfarin, the mean TTR of SA participants was 58%. 
In keeping with our findings, in an earlier study  only 50% of patients in Cape Town receiving warfarin for rheumatic heart disease had good anticoagulation control as measured by TTR calculated by the Rosendaal method, but only three INR readings were used in this study. A cross-sectional study at Victoria Hospital in Cape Town found that only 49% of patients had a therapeutic INR.  There are few African studies outside SA. In an Ethiopian cross-sectional study, only 30.3% of patients had a therapeutic INR.  These cross-sectional studies are limited in their design by the fact that they only provide a snapshot of INR control at a single time point. The association between older age and improved control that we observed is in keeping with findings from other studies. [18,20] The Victoria Hospital study in Cape Town found that patients aged =60 years were more likely than younger patients to have a therapeutic INR. A Swedish study also found a correlation between improved TTR and older age. [18,20] We found no association between INR control and sex, similar to the study at Victoria Hospital.  INR control did not differ between the two clinics in our study.
Patients with poor anticoagulation control (TTR <65%) had more frequent INR monitoring than those with good control, but despite regular monitoring to guide dose adjustments, INR control was poor in these patients. This may reflect flaws in the unvalidated algorithm used to guide warfarin dose adjustment at our study sites.
Poor INR control may result in serious clinical consequences. A recent SA survey found that warfarin was the fourth most commonly implicated drug in ADR-related admissions and the most commonly implicated drug in preventable ADR-related admissions.  A median hospital stay of 6 days (IQR 4-8.5) for all bleeds related to warfarin and non-steroidal anti-inflammatory drugs was recorded.  In an American study, INR instability was associated with higher healthcare utilisation, driven by increased length of hospital stay. 
Direct oral anticoagulants (DOACs) are an alternative to warfarin. DOACs have a number of advantages, but are expensive when the direct cost is considered and therefore currently not routinely available in the SA public sector. DOACs have more rapid onset than warfarin, do not require heparin bridging, have fixed doses, are less susceptible to food and drug-drug interactions and do not require routine anticoagulation monitoring. [9,22-24] A meta-analysis of stroke prevention in AF patients comparing three DOACs (rivaroxaban, dabigatran and apixaban) with VKAs, published in 2017, found a significantly lower risk of intracranial haemorrhage with DOACs, with similar efficacy to that of DOACs.  However, warfarin therapy is likely to remain an important anticoagulant option in our setting, as DOACs are currently not easily accessed in the public healthcare sector. In addition, poor TTR control may be compounded by suboptimal warfarin dose adjustment practice. In SA, healthcare workers may prescribe half-tablet dosages in order to achieve a warfarin dose less than the only available 5 mg strength. A study comparing warfarin measured half-tablet drug content against target drug content found that a third of half-tablets fell outside of the proxy United States Pharmacopeia specification.  These findings suggest that warfarin may not always be uniformly distributed within the tablets, which may contribute to the variability and difficulty in achieving effective warfarin dose titration. Individualised small incremental dose adjustments may therefore not always be possible.
Our study has several limitations. We did not have data on some covariates such as concomitant comorbid disease, including hepatic or cardiac dysfunction, interacting medications and diet, which could affect anticoagulation control. Data on warfarin adherence, dosing recommendations or adherence to dose adjustment recommendations were not available. Patients with >70-day gaps between INRs were excluded, in line with recommendations for the Rosendaal method.  This may bias towards inclusion of more adherent patients, as patients missing clinic visits would be excluded. Despite this potential for bias, TTR was low in our included patients. Patients with <27 months of follow-up were not included, and this exclusion may also have introduced some bias, as patients with an indication for long-term anticoagulation who died before 27 months were not eligible for inclusion. For patients with 'unknown indication' we assumed a lower INR target level. This may have biased TTR control positively if the higher target (INR 2.5-3.5) was required. Although our total sample from the two sites exceeded the 346 patients required based on our power calculation, we were only able to include 152 patients who were eligible for inclusion from the MPC site. We therefore did not meet the power requirement to detect a 15% TTR difference between the two sites. The Rosendaal method to calculate TTR, which is the widely recommended method for describing anticoagulation control, also has certain limitations, particularly when individual INR values are far outside the recommended therapeutic range, as it assumes that the change in INR over time is linear between each time-point, which may not always be true. Our TTR target of =65% is derived from studies in AF, and may be too conservative for patients with VHD.
Despite these limitations, this study provides good evidence in clinical practice of inadequate anticoagulation control in patients attending two high-volume INR clinics located in an urban SA community healthcare centre and at a tertiary academic hospital.
We found poor levels of anticoagulation control in our longitudinal study at two large urban dedicated SA INR clinics. Poor INR control places patients at risk of complications due to toxicity or lack of efficacy of therapy. Further research identifying predictors of poor control is required. The impact of poor anticoagulation control on clinical outcomes and healthcare costs in SA patients requires quantification. Warfarin therapy is likely to remain an important anticoagulant option in our setting, until the affordability and hence accessibility of alternative agents in SA conditions have been formally assessed. Evidence-based, locally validated algorithms to guide warfarin dose adjustment are urgently needed. The only warfarin formulation currently available in the public sector is the 5 mg tablet. Lower warfarin dosage formulations may optimise individual dose titration, and advocacy for these formulations is advised.
Acknowledgements. The authors gratefully acknowledge the study patients and the nursing staff at the two prothrombin clinics, MPC and GSH. Special acknowledgement goes to Dr J P Mouton and Mr William Msemburi for their support in statistics and extraction of the data. We also thank Annemie Stewart for creating the data base, Nicky Kramer for her contribution during the protocol development and Dawn Rossiter for editing. Lastly, we thank the National Health Laboratory Service for providing patient INR data.
Author contributions. Study concept and design: IE, MB and AB. Drafting of the manuscript: IE, MB, AB and KC. Statistical analysis and statistical support: IE, JPM, WM and KC. Interpretation of the data: all authors. Critical revision of the manuscript for important intellectual content: all authors.
Conflicts of interest. None.
[1.] Singer D, Albers G, James D, et al. Antithrombotic therapy in atrial fibrillation. Chest 2008; 133(6):546S-592S. https://doi.org/10.1378/chest.08-0678
[2.] South African Medicine Research Council. South African Burden of Disease report 2000. www.mrc. ac.za/bod (accessed 16 January 2016).
[3.] Horton R. Global disease burden GBD 2010: Understanding disease, injury, and risk. Lancet 2012; 380(9859):2053-2054. https://doi.org/10.1016/S0140-6736(12)62133-3
[4.] Whayne TF. A review of the role of anticoagulation in the treatment of peripheral arterial disease. Int J Angiol 2012; 21(4):187-194. https://doi.org/10.1055/s-0032-1330232
[5.] Hylek EM. Vitamin K antagonists and time in the therapeutic range: Implications, challenges, and strategies for improvement. J Thromb Thrombolysis 2013; 35(3):333-335. https://doi.org/10.1007/s11239013-0900-5
[6.] Cannegieter S, Rosendaal F, Briet E. Thromboembolic and bleeding complications in patients with mechanical heart valve prostheses. Circulation 1994; 89:635-641. https://doi.org/10.1161/01.CIR.89.2.635
[7.] Hart RG, Benavente O, McBride R, et al. Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: A meta-analysis. Ann Intern Med 1999; 131(7):492-501. https://doi.org/10.7326/0003-4819131-7-199910050- 00042
[8.] Bryer A. New antithrombotic drugs? A revolution in stroke management (Editorial). Cardiovasc J Afr 2012; 23(2):61-62. http://www.cvja.co.za/onlinejournal/vol23/vol23_issue2/files/assets/basic- html/index. html#7 (accessed 24 April 2018).
[9.] Sardar P, Chatterjee S, Chaudhari S, et al. New oral anticoagulants in elderly adults: Evidence from a metaanalysis of randomized trials. J Am Geriatr Soc 2014; 62(5):857-864. https://doi.org/10.1111/jgs.12799
[10.] Connolly SJ, Pogue J, Eikelboom J, et al. Benefit of oral anticoagulant over antiplatelet therapy in atrial fibrillation depends on the quality of international normalized ratio control achieved by centers and countries as measured by time in therapeutic range. Circulation 2008; 118:2029-2037. http://doi. org/10.1161/CIRCULATIONAHA.107.750000
[11.] Mouton JP, Njuguna C, Kramer N, et al. Adverse drug reactions causing admission to medical wards. Medicine (Baltimore) 2016; 95(19):1-10. https://doi.org/10.1097/MD.0000000000003437
[12.] Wallentin L, Yusuf S, Ezekowitz MD, et al. Efficacy and safety of dabigatran compared with warfarin at different levels of international normalised ratio control for stroke prevention in atrial fibrillation: An analysis of the RE-LY trial. Lancet 2010; 376(9745):975-983. https://doi.org/10.1016/S01406736(10)61194-4
[13.] Rosendaal FR, Cannegieter SC, van der Meer FJ, et al. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost 1993; 69(3):236-239.
[14.] Wan Y, Heneghan C, Perera R, et al. Anticoagulation control and prediction of adverse events in patients with atrial fibrillation: A systematic review. Circ Cardiovasc Qual Outcomes 2008; 1:84-91. https://doi. org/10.1161/CIRCOUTCOMES.108.796185
[15.] World Medical Association. World Medical Association Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Subjects. JAMA 2013; 310(20):2191-2194. https://doi.org/10.1001/ jama.2013.281053
[16.] ICH harmonized tripartite guideline: Guideline for Good Clinical Practice. J Postgrad Med 2001; 47(1):45-50. http://www.jpgmonline.com/text.asp?2001/47/1/45/235 (accessed 3 August 2017).
[17.] Barth DD, Zuhlke LJ, Joachim A, et al. Effect of distance to health facility on the maintenance of INR therapeutic ranges in rheumatic heart disease patients from Cape Town? No evidence for an association. BMC Health Serv Res 2015; 15(219):1-7. https://doi.org/10.1186/s12913-015-0890-4
[18.] Sonuga BO, Hellenberg DA, Cupido CS, Jaeger C. Profile and anticoagulation outcomes of patients on warfarin therapy in an urban hospital in Cape Town, South Africa. Afr J Primary Health Care Fam Med 2016; 8(1):1-8. https://doi.org/10.4102/phcfm.v8i1.1032
[19.] Teklay G, Shiferaw N, Legesse B, Bekele ML. Drug-drug interactions and risk of bleeding among inpatients on warfarin therapy? A prospective observational study. Thromb J 2014; 12:20. https://doi. org/10.1186/1477-9560-12-20
[20.] Wieloch M, Sjlander A, Frykman V, Rosenqvist M, Eriksson N, Svensson PJ. Anticoagulation control in Sweden: Reports of time in therapeutic range, major bleeding, and thrombo- embolic complications from the national quality registry AuriculA. Eur Heart J 2011; 32(18):2282-2289. https://doi. org/10.1093/eurheartj/ehr134
[21.] Laliberte F, Pilon D, Raut MK, et al. Is rivaroxaban associated with lower inpatient costs compared to warfarin among patients with non-valvular atrial fibrillation? Curr Med Res Opin 2014; 30(8):1521-1528. https://doi.org/10.1185/03007995.2014.916159
[22.] Gomez-Outes A, Terleira-Fernandez AI, Calvo-Rojas G, Suarez-Gea ML, Vargas-Castrillon E. Dabigatran, rivaroxaban, or apixaban versus warfarin in patients with nonvalvular atrial fibrillation? A systematic review and meta-analysis of subgroups. Thrombosis 2013 (2013), Article ID 640723. https://doi.org/10.1155/2013/640723
[23.] Ymer M, Agon M, Shkelzen D, et al. New oral anticoagulants? Their advantages and disadvantages compared with vitamin K antagonists in the prevention and treatment of patients with thromboembolic events. Ther Clin Risk Manag 2015; 11:967-977. https://doi.org/10.2147/TCRM.S84210
[24.] Verheugt FA. The new oral anticoagulants. Neth Heart J 2010; 18(6):314- 318. https://doi.org/10.1182/ blood-2009-09-241851
[25.] Ntaios G, Papavasileiou V, Makaritsis K, Vemmos K, Michel P, Lip G. Real- world setting comparison of nonvitamin-K antagonist oral anticoagulants versus vitamin-K antagonists for stroke prevention in atrial fibrillation: A systematic review and meta-analysis. Stroke 2017; 48(9):2494-2503. https://doi. org/10.1161/STROKEAHA.117.017549
[26.] Hill S, Varker AS, Karlage K, Myrdal PB. Analysis of drug content and weight uniformity for half-tablets of 6 commonly split medications. J Manag Care Spec Pharm 2009; 15(3):253-261. https://doi. org/10.18553/jmcp.2009.153.253
Accepted 2 February 2018.
I Ebrahim, (1) MB ChB, FC Clin Pharmacol (SA); A Bryer, (2) MB BCh, FCP (SA), MMed (Neurology), FC Neurology (SA), PhD; K Cohen, (1) MB ChB, MMed (Clin Pharmacol), FCFP (SA), MSc (Epidemiol), Dip HIV Man, Dip Obst; J P Mouton, (1) MB ChB; W Msemburi, (3) BSc Statistics, BSc Hons, MPhil (Demography); M Blockman, (1) MB ChB, BPharm, PG Dip Int Res Ethics, MMed (Clin Pharmacol)
(1) Division of Clinical Pharmacology, Groote Schuur Hospital and Faculty of Health Sciences, University of Cape Town, South Africa
(2) Division of Neurology, Department of Medicine, Groote Schuur Hospital and Faculty of Health Sciences, University of Cape Town, South Africa
(3) Clinical Research Institute, Groote Schuur Hospital and Faculty of Health Sciences, University of Cape Town, South Africa
Caption: Fig. 1. Selection of patients included in the analysis. (INR = international normalised ratio; MPC = Mitchell's Plain Community Health Centre; GSH = Groote Schuur Hospital.)
Caption: Fig. 2. Scatter plot of the percentage of time the INR was in the therapeutic range over 6 months of warfarin therapy. Solid bars represent means and standard deviations. The dotted line is at the TTR target (INR in therapeutic range for 65% of the time). TTR below the target indicates inadequate INR control. (INR = international normalised ratio; TTR = time in therapeutic range; MPC = Mitchell's Plain Community Health Centre; GSH = Groote Schuur Hospital.)
Table 1. Characteristics of included patients (N=363) MPC GSH Patients, n (%) 151 (41.3) 212 (58.4) Female sex 100 (66.2) 138 (65.1) Age (years), median (IQR) 57 (48-66) 53 (41-62) Target INR range, n (%) 2.0-3.0 * 121 (80.1) 76 (35.9) 2.5-3.5 ([dagger]) 30 (19.9) 136 (64.2) MPC = Mitchell's Plain Community Health Centre; GSH = Groote Schuur Hospital; IQR = interquartile range; INR = international normalised ratio. * Target INR range of 2.0-3.0 for patients anticoagulated with warfarin because of atrial fibrillation, pulmonary embolus, venous thromboembolism, systemic lupus erythematosus, hypercoagulable states, atrial flutter or undocumented indication. ([dagger]) Target INR range of 2.5-3.5 for patients anticoagulated with warfarin because they had valvular heart disease. Table 2. Multivariate logistic regression model of associations with good anticoagulation control (TTR [greater than or equal to] 65%) over 6 months (363 patients included in the model) Variable Category TTR [greater than or equal to] 65%, n (%) Age <55 years 34/177 (19.2) [greater than 57/186 (30.6) or equal to] 55 years Sex Female 56/238 (23.5) Male 35/125 (28.0) Site MPC 44/151 (29.1) GSH 47/212 (22.2) INR 2-3 * 57/197 (28.9) target 2.5-3.5 34/166 (20.5) ([dagger]) Variable Category Univariate analysis OR (95% CI) Wald test p-value Age <55 years ref -- [greater than 1.86 (1.14-3.02) 0.013 or equal to] 55 years Sex Female ref -- Male 1.26 (0.77-2.07) 0.351 Site MPC ref -- GSH 0.69 (0.43-1.12) 0.132 INR 2-3 * ref -- target 2.5-3.5 0.63 (0.39-1.03) 0.065 ([dagger]) Variable Category Multivariate analysis Adjusted OR (95% CI) Wald test p-value Age <55 years ref -- [greater than 1.69 (1.03-2.79) 0.039 or equal to] 55 years Sex Female ref -- Male 1.21 (0.73-1.99) 0.451 Site MPC ref -- GSH 0.85 (0.50-1.46) 0.560 INR 2-3 * ref -- target 2.5-3.5 0.75 (0.43-1.30) 0.312 ([dagger]) TTR = time in therapeutic range; INR = international normalised ratio; OR = odds ratio; CI = confidence interval; MPC = Mitchell's Plain Community Health Centre; GSH = Groote Schuur Hospital. *INR target 2-3: atrial fibrillation, pulmonary embolus, venous thromboembolism, systemic lupus erythematosus, hypercoagulable states, atrial flutter. ([dagger]) INR target 2.5-3.5: valvular heart disease.
|Printer friendly Cite/link Email Feedback|
|Author:||Ebrahim, I.; Bryer, A.; Cohen, K.; Mouton, J.P.; Msemburi, W.; Blockman, M.|
|Publication:||SAMJ South African Medical Journal|
|Date:||Jun 1, 2018|
|Previous Article:||Medical students' perspectives on euthanasia and physician-assisted suicide and their views on legalising these practices in South Africa.|
|Next Article:||Phenotypic and genotypic correlation of carbapenememase-producing Enterobacteriaceae and problems experienced in routine screening.|