Abstract
Background
Vitamin K antagonists (VKA) continue to be the principal anticoagulants for both the treatment and prevention of venous thromboembolism. The use of VKA often requires regular monitoring to avoid over-anticoagulation and prevent thromboembolic complications. The aim was to determine the indication for VKA use and factors associated with suboptimal anticoagulation control among patients in northern Tanzania.
Methods
This was a retrospective cohort study that examined the anticoagulation data of patients on long-term VKA from 1st January 2020 to 31st December 2022. Eligible participants were those on VKAs for at least 7 days and with at least 3 international normalized ratio (INR) results. The level of anticoagulation control was determined through the calculation of the time-in-therapeutic range (TiTR) using the Rosendaal and the percent of INR in therapeutic range methods.
Results
TiTR was found to be 17% using the direct method and 16% using the Rosendaal formula. 102 tests out of 365 were within the target range (28%). Absence of health insurance (aRR: 1.24, 95% CI: 1.06-1.44, P = .007), alcohol consumption (aRR: 1.37, 95% CI: 1.15-1.62, P < .001), and prolonged intervals between tests of 14 to 28 days (aRR: 1.34, 95% CI: 1.05-1.74, P = .018) showed association with INR being out of range.
Conclusion
Patients who achieved target therapeutic anticoagulation control were less than the acceptable 65%. Anticoagulation outcomes were better in patients with frequent INR monitoring and those with health insurance. Alcohol consumption carries a high risk of poor anticoagulation control. Further studies are needed to enforce better INR control.
Keywords
Introduction
Patients on vitamin K antagonists (VKA) are at constant risk of either under- or over-anticoagulation. Suboptimal anticoagulation control is associated with an increased risk of thromboembolism, while overtreatment is associated with major bleeding, leading to possible and deleterious outcomes. VKAs are commonly used to prevent thromboembolic evets in patients with atrial fibrillation (AF) and mechanical prosthetic heart valves, with an approximate risk reduction of 64% and a decrease in all-cause mortality by 26%. 1 The most common indications for VKAs in sub-Saharan Africa (SSA) are venous thromboembolism (VTE) (49%), AF (32%), and valvular heart disease at 13%. 2
VKA anticoagulation monitoring is achieved through a systematic international normalized ratio (INR) measurement. 3 The INR is used to standardize prothrombin time (PT) and to determine the degree to which VKA has successfully suppressed the patient's coagulation system. 4 The target INR recommended for most warfarin indications is 2.0 to 3.0 (venous thromboembolism and cardiac arrhythmia) and 2.5 to 3.5 for mechanical prosthetic valves. 4 Sustaining a patient in therapeutic ranges reduces the risk of VTE while minimizing the risk of bleeding attributable to excessive anticoagulation. On the other hand, more time spent with INR below and above the therapeutic range increases the risk of VTE and bleeding complications. 5 Consequently, the percentage of time spent in the therapeutic range (TiTR) is a way of summarizing anticoagulation control over time. The NICE Guidelines recommend a TiTR above 65% as an optimal anticoagulation level for the prevention of thromboembolism without excessive risk of bleeding from VKA. 5 However, a TiTR of less than 40% is not associated with any significant mortality benefit.5,6
Because of their narrow therapeutic window, considerable variability in dose response among patients, and multiple interactions with other drugs and diet, the use of VKA is generally complicated. 7 It is often challenging to balance the risk of bleeding and, at the same time, prevent clotting. 8 Anticoagulation monitoring and frequent VKA dose adjustments are required because the therapeutic or prophylactic benefits of VKA depend on their maintenance within therapeutic ranges. 3
Guidelines recommend that stable patients receiving warfarin undergo INR monitoring every 2 to 4 weeks when in good control. 8 Daily monitoring of INR is recommended for admitted patients. 9 Those patients initiated on warfarin or those transitioning to another anticoagulant need monitoring at least 3 times per week, and those with frequent dose adjustments or changes in clinical status are recommended to monitor their INR at least weekly.9,10
In this resource-limited setting, frequent monitoring of INR is not always achieved, and investigating this would better help understand the group of patients who are at higher risk for poor anticoagulation control and help direct the available resources toward better management for the patients. Therefore, this hospital-based retrospective cohort study aimed to examine the clinical indications, factors associated with optimal anticoagulation, and estimated levels of anticoagulation control for patients on warfarin.
Methods
Study Design and Setting
A hospital-based retrospective cohort study among adult patients on VKA was conducted at the outpatient medical clinics at Kilimanjaro Christian Medical Centre from 1st January 2020 to 31st December 2022.
In this study, data was accessed and collected from records of adult patients from 1st October 2022 to 31st March 2023. The study included all sampled patients who have attended as both inpatients and outpatients at Kilimanjaro Christian Medical Centre during the specified period. Eligible participants were all those on VKAs during the study period who took Warfarin for at least 7 days. Excluded patients were those with less than 3 INR readings and patients with liver failure.
Data Collection
Clinical information was gathered using a structured data form that retrieved data from both the hospital's electronic management system. Information was extracted on the indications, type, and duration of VKA use from patient's hospital electronic records. Sociodemographic factors, coexisting medical conditions, and concurrent use of drugs that interact with warfarin were documented. Other data included the final 3 consecutive INR values of patients between 1st January 2020 and 31st December 2022. In addition to INR values, the duration between INR tests and corresponding warfarin dosages was documented.
With a minimum sample size of 100 participants, the study was able to estimate a TiTR of 41% with a margin of error of 5% at a 2-sided alpha level of 0.05. The dependent variable was anticoagulation control. Independent variables were socio-demographic and clinical characteristics, which included the duration between INR measurement and INR values. The data was entered directly into an Excel spreadsheet and then into Statistical Package for Social Sciences for analysis.
The study followed the methods of Mwita et al 2022. 11 The percent of INR within the therapeutic range was calculated by both the direct method and the Rosendaal formula. The direct method calculates the percentage of INR in the therapeutic range for all patients divided by the overall number of INRs during the period. 3 Therapeutic INR was defined as 2.0 to 3.0 for VTE and AF and 2.5 to 3.5 for mechanical prosthetic valves. 4 A TiTR of less than 65% was considered suboptimal anticoagulation control. The TiTR by Rosendaal formula was calculated using the Excel software.
Statistical Analysis
Clean data was imported from an Excel spreadsheet and analyzed using Statistical Package for Social Sciences version 26. Categorical variables were presented as frequencies and percentages, and continuous variables as means (standard deviation) or medians (interquartile ranges). Patients were allocated into 2 categories based on their clinical indications to facilitate the comparison of variables. The study applied modified Poisson regression analysis with robust standard errors to estimate the magnitude of the association between socio-clinical factors and the level of anticoagulation control. The model considered the crude risk ratio (cRR) at ρ-value <0.05% and 95% CI with the robust standard estimator to be statistically significant. Characteristics found to have associations from univariate analysis were considered for multivariate analysis of adjusted risk ratio (aRR) to control for various confounding characteristics.
Results
Among the 178 patients who were screened, data was only analyzed from 122 patients, as 54 patients were not eligible due to having less than 3 INR readings (lost to follow-up) and 4 patients having a concurrent liver disease. In Table 1, 83 (68.0%) of the study patients were female, the median age distribution was 52 years (IQR 35-67 years), and 50 patients (41.0%) were aged between 41 and 65 years. More than half (58.2%) of the patients were employed, and 54.9% had health insurance coverage. Few were underweight or had a normal BMI, while 40 (40.4%) and 26 (26.3%) were obese and overweight, respectively. More than half (55.7%) were married. Seventy patients (57.4%) either consumed alcohol in conjunction with warfarin or had a documented history of alcohol consumption (Table 1). The majority of the patients were maintained at a warfarin dose of 5 mg (68.0%), followed by 2.5 mg (17.0%). Most patients had at least 1 co-morbidity (45.1%), whereas 23% had 2 or more (Table 2).
Social Demographic Characteristics of Patients on VKAs (n = 122).
Clinical Characteristics of Patients on VKAs (n = 122).
The most common co-morbid condition was heart failure (41.2%), followed by hypertension (34.1%) and diabetes mellitus (21.2%). The median duration of INR measurements was 28 days (IQR 7-60 days), but the majority of the patients (45.9%) would surpass 28 days to check their INR (Table 1). Out of the 122 patients, 39% were using warfarin for deep venous thrombosis (DVT), 19% for pulmonary embolism, 11% for AF, and 19% for valvular replacement, which required chronic anticoagulation.
Using the direct method, INR tests found to be within range were only 17.2%, 75.4% of patients’ INR was below range, and 7.4% was above the target range (Table 2). By using the Rosendaal formula, from a total number of 12,114 days, days within the range were found to be 1,935. Therefore, the percentage TiTR was established at 16%, which was similar to that obtained in the direct method. The total number of tests done was 365, where 102 tests (28%) were within the target range.
Alcohol consumption or having a history of alcohol intake increased the risk of having poor anticoagulation control by 37% (aRR: 1.37, 95% CI: 1.15-1.62, P < .001) even after adjusting to gender, marital status, employment status, insurance, alcohol, BMI, warfarin dose, and duration between INR tests (Table 3). Being obese showed an insignificant association with having suboptimal anticoagulation on VKAs (aRR: 1.12, 95% CI: 0.91-1.38, P = .280). A warfarin dose of 2.5 mg was associated with a 42% lower risk of being out of range, and a significant protective effect was observed (aRR: 0.58, 95% CI: 0.37-0.91, P = .016). Having co-morbidities did not show any significant association with INR being out of range (Table 3).
Factors Associated With Suboptimal Control of INR (n = 122).
WR = within range; OoR = out of range; cRR = crude risk ratio; aRR = adjusted risk ratio.
Discussion
This study reported poor anticoagulation control using both the direct method (17.0%) and the Rosendaal formula (28.0%). A multinational audit involving several African countries reported poor TiTR for patients on VKA ranging from 29% to 47%. 2 A higher TiTR is reported in studies performed in developed countries. A TiTR of 63% was observed in a large study involving 6,706 patients in the USA, 5 and in another study conducted using data from the RHD registry predominantly conducted in African countries, the TiTR was found to be 28%, and patients who were out of range were observed to be more likely to be subtherapeutic than being overtherapeutic. 12 In this study, this ratio was even bigger, whereby 92% of patients who were out of range were subtherapeutic, and only 8% of the patients were overtherapeutic. TiTR was only 20.0% in a cross-sectional study undertaken in a warfarin anticoagulation clinic in the national hospital of Tanzania involving 190 patients, 13 as these results were similar to this study despite the wider scope of patients involved, where inpatients and outpatients from different medical clinics were included.
The most common indication for warfarin in this study was DVT. Similarly, other studies in Africa also showed that the most common indication for warfarin was venous thromboembolism, particularly DVT.5,14 However, several studies show cardiac indications, such as the presence of mechanical heart valves, as the leading indication for VKAs. 15 A study done at another center in Tanzania in 2021 showed that mechanical heart valves (71.6%) accounted for the majority of patients in VKAs, followed by DVT (18.9%). 13 The service of cardiac valvular replacement is not yet available in the study center; the patients reviewed are mostly follow-up patients who had their surgeries done elsewhere but reside in the northern zone. This could explain the comparatively fewer patients with mechanical heart valves in this study.
This study revealed a positive association between alcohol consumption and an increased risk of having poor anticoagulation control (aRR: 1.37, 95% CI: 1.15-1.62, P = .000) even after adjusting for other socio-clinical characteristics. These results are similar to a study done in South Africa, where a significant association was found between alcohol consumption and poor anticoagulation outcomes (P < .022). 12 In a case-control study done in America, among 265 cases and 305 controls, patients who screened positive for moderate/severe alcohol misuse or heavy episodic drinking had an approximately 2-fold increased risk of major bleeding in adjusted analyses (RR: 2.10, 95% CI: 1.08-4.07; and RR: 2.36, 95% CI: 1.24-4.50, respectively). The same study also showed there was an increased alcohol-related major bleeding risk in patients on warfarin for ≥1 year. 16 Acute alcohol intake increases anticoagulation by decreasing warfarin metabolism; this excessive warfarin activity results from alcohol-related inhibition of warfarin metabolism by cytochrome P450 in the liver. 17 However, chronic alcohol ingestion has been shown to decrease anticoagulation by activating cytochrome P450 and, consequently, increasing warfarin metabolism. As a result, warfarin is broken down faster than normal, and higher warfarin doses are required to achieve the desired anticoagulant effect. 18 Therefore, acute and chronic alcohol intake predisposes a patient on warfarin to poor anticoagulation control. Alcohol screening questionnaires, as well as the improvement of dose adjustments, close monitoring, and counseling of patients with a history of alcohol consumption, are useful in managing patients on warfarin anticoagulation. 16
Lack of health insurance coverage was significantly associated with poor anticoagulation control in our study, even after adjustment to other socio-clinical characteristics (aRR: 1.24, 95% CI: 1.06-1.44, P = .007). Patients must pay approximately US$6 for each INR test; and more costs are incurred for transportation to and from the healthcare facility and purchasing monthly doses of warfarin. Most patients cannot afford such expenses monthly and therefore depend on the availability of health insurance. In a study done in Ugandan clinics to assess the quality of anticoagulation services, they found poor coagulation control in facilities where a patient had to pay approximately US$5 per INR test. 2
In a large study done in the USA, the mean follow-up interval varied from 25 to 38 days. As this interval became longer, the risk-adjusted TiTR was found to be worse (−0.51% per day, P = .004). In this study, the duration between INR tests of 2 to 4 weeks showed an independent association with INR being out of range. Considering that most of the patients in our study were poorly anticoagulated, they most likely required more frequent monitoring to improve their control. These results were contrary to those observed in South Africa, where patients with poor anticoagulation (TiTR <65%) had more frequent INR monitoring compared to those with good control. 15 Patients who have poor INR control have to be monitored more frequently than those with better INR control. 19 Therefore, it is expected that those with longer durations between tests have better anticoagulation control than those with shorter durations between INR tests. However, patients who do not monitor their INR regularly may unsuspectingly stay out of therapeutic range for longer periods.
The warfarin dose of 2.5 mg showed a significant protective effect in this study compared to patients on doses of 5 mg and 7.5 mg, although most of the study participants were maintained at a dose of 5 mg. Upon initiation of warfarin, the standard dose of 5 mg is generally recommended, and doses of 2.5 mg are preferred for frail, older (>70 years), malnourished, patients with liver or kidney disease or heart failure, or those receiving a medication known to increase warfarin sensitivity. 20 Loading doses of warfarin are discouraged as they may place the patient in a hypercoagulable state due to severe depletion of protein C. 21
Larger studies on the anticoagulation outcome at different initiation and maintenance doses of warfarin are recommended to further conclude the protective effect of a 2.5-mg dose against poor anticoagulation. If one is unable to monitor INR religiously or in facilities where INR services are not available, would it be safer for the patient to be prescribed a dose of 2.5 mg without concern of them being poorly anticoagulated? Considering that INR services are still unavailable in many healthcare facilities in Tanzania, particularly at the district and village levels, this hypothesis is worth investigating.
The study limitations included that this was a retrospective study, and therefore, it has some missing data in clinical records, which have been accounted for in the analysis of variables, particularly BMI. The study used a robust model as the approach for handling missing data. The study had no groups of cases with higher rates of missing data that could impact the generalizability of these findings. Another limitation is the absence of important patient particulars from the electronic medical records that might lead to poor anticoagulation, such as the patient's dietary patterns, medication adherence, nutrition status, and smoking history. Patients did not come for the recommended monthly visits, and they were not checking their INR outside the hospitals in facilities that were closer to their homes.
Conclusion
In this study, patients who achieved target therapeutic anticoagulation control were less than the recommended 65%. To achieve better control, it is advised to work on alcohol use, health insurance, and frequent INR monitoring. VKAs are mostly prescribed for deep vein thrombosis, pulmonary embolism, and patients with mechanical prosthetic valves. Anticoagulation outcomes were better in patients with frequent INR monitoring and those with health insurance. Being an alcoholic carries a high risk of poor anticoagulation control. Larger studies are needed to examine the protective effect of the warfarin dose of 2.5 mg in patients who are unable to monitor their INR regularly.
Footnotes
Availability of Data and Materials
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Consent for Publication
Not applicable
Declaration of Conflicting Interests
The authors declare that there is no conflict of interest.
Ethics Approval and Consent to Participate
Permission to conduct the study was sought from the Kilimanjaro Christian Medical University College Research Ethics and Review Committee (No. PG136/2022). Written consent was waived by the ethics committe due to the restrospective nature of the study.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
