Warfarin is the oral anticoagulant that is most widely distributed by the Unified Health System (SUS, acronym in Portuguese) in Brazil, and it is still widely used worldwide.¹ Different subgroups of patients can benefit from the use of this medication, particularly individuals with heart disease with atrial fibrillation (AF)² and/or mechanical valve prosthesis,³ Chagas disease,⁴ and rheumatic mitral valve diseases, as well as non-cardiac patients with indication for anticoagulation, respecting specificities and individualities.³

As warfarin has a narrow therapeutic range and interacts with medications and foods, its use must be monitored. This medication can cause undesirable events, such as hemorrhagic transformation associated with stroke,^5,6 which increases when patients use the medication incorrectly. It is recommended to use with caution, especially in individuals with low socioeconomic and cultural conditions.⁶

Monitoring of warfarin treatment is carried out using the international normalized ratio (INR),^3,7 a test calculated based on prothrombin activity. The therapeutic range for most indications recommends target INR between 2.00 and 3.00, and the clinical response to treatment and the assessment of coagulation stability are also influenced by drug interactions, patient genetic variability, interaction with diet, and associated comorbidities.^2,8 When the INR is below 2.00, the benefits resulting from the use of the medication may be compromised, for example, preventing the occurrence of ischemic stroke, whereas INR above 3.00 is associated with hemorrhagic events.⁹ These characteristics are challenges for the safety of this pharmacotherapy, making adherence to therapy a necessary condition to improve the effectiveness of the treatment.¹⁰

The assessment of the quality of oral anticoagulation control can be carried out by calculating the time in therapeutic range (TTR), which identifies the proportion of time in which the INR results of a given patient remained in the target therapeutic range.^9,11 The TTR is calculated from a linear interpolation of a historical series of INR results and can be used as a method for estimating the quality of anticoagulation in health systems.¹²

It is recommended that the TTR value be above 60% in order for anticoagulant therapy to be superior to isolated therapy with antiplatelet agents.⁹ Mean TTR values below 60% suggest the need to implement strategies that encourage adherence, understanding, and other actions related to rational use for groups of patients who present TTR values below expectations.⁹ By recognizing the importance of TTR for the effectiveness of anticoagulant therapy with warfarin, the present study aimed to identify factors associated with TTR variation.

During the study period, 889 patients from the H2 outpatient clinic were identified with a mean age of 61 years, and 482 of them were female (54%). In the H1 outpatient clinic, 468 patients were identified with a mean age of 60 years, and 270 of them were female (58%). Specification of the mean TTR and INR values in relation to the total number of monitored patients treated in the period (1,357) are displayed in Table 1, and the characterization of patients with low TTR is displayed in Table 2.

Table 1

Variable specification	H1			H2
	Mean	Standard deviation	CI	Mean	Standard deviation	CI	P value
TTR	62.60	27.90	60.0 - 55.20	57.5	33.46	55.23 - 59.70	<0.001
INR	2.60	0.90	2.6 - 2.70	2.5	1.08	2.45 - 2.52	<0.001

H1: university hospital 1; H2: university hospital 2; INR: international normalized ratio; TTR: time in therapeutic range. Source: Prepared by the authors; CI: confidence interval.

Table 2

Variable specification	H1	H2	Inadequate total TTR N(%)	Percentage in relation to the total number of patients monitored (N = 1357)
	N (%)	N (%)
Female	139 (36.58)	241 (63.42)	380 (100.00)	28.00
Male	62 (23.66)	200 (76.34)	262 (100.00)	46.13
Residents in Belo Horizonte	113 (23.54)	367 (76.46)	480 (100.00)	35.37
Residents in the countryside of Minas Gerais	10 (76.92)	3 (23.08)	13 (100.00)	0.86
Residents in the metropolitan region	67 (48.55)	71 (51.45)	138 (100.00)	10,17
Report of incorrect use of warfarin	17 (58.62)	12 (41.38)	29 (100.00)	2.14
Changes in other medications in use	26 (65.00)	14 (35.00)	40 (100.00)	2.94
Change in eating habits	14 (40.00)	21 (60.00)	35 (100.00)	2.57
Indication for anticoagulation
Idiopathic stroke	22(48.89)	23 (51.11)	45 (100.00)	3.32
Valvular AF	36 (66.67)	18 (33.33)	54 (100.00)	3.98
Non-valvular AF/flutter	58 (18.77)	251 (81.23)	309 (100.00)	22.77
Metal prosthesis	36 (65.45)	19 (34.55)	55 (100.00)	4.05
Thromboembolism	2 (1.77)	111 (98.23)	113 (100.00)	8.32
Others	4 (18.18)	18 (81.82)	22 (100.00)	1.62

AF: atrial fibrillation; H1: university hospital 1; H2: university hospital 2; TTR: time in therapeutic range.

In binary logistic regression, incorrect use of warfarin, indication for anticoagulation, and changes in eating habits were associated with inadequate TTR, as seen in Table 3.

Table 3

Variable specification	Reference variable	OR	CI	P value
Male	Female	0.89	0.71 - 1.11	0.30
Age over 60	Less than 60	1.05	0.71 - 1.53	0.82
Incorrect use of warfarin	Correct use	2.81	1.42 - 5.56	0.03
Change in other medications being used	No changes	0.89	0.56 - 1.40	0.61
H2 anticoagulation site	H1	6.4	0.23 - 175.44	0.27
Indication for anticoagulation due to valvular AF + non-valvular AF/flutter	Stroke	1.70	1.12 - 2.54	0.01
Indication for anticoagulation due to metallic prosthesis	Stroke	1.44	0.84 - 2.45	0.19
Indication for anticoagulation due to thromboembolism	Stroke	1.27	0.81 - 2.01	0.30
Indication for anticoagulation - others	Stroke	0.98	0.51 - 1.89	0.95
Change in eating habits	No change	1.85	1.06 - 3.25	0.03

AF: atrial fibrillation; H1: university hospital 1; H2: university hospital 2; OR: odds ratio; TTR: time in therapeutic range. Source: Prepared by the authors; CI: confidence interval.

In the multiple regression, as shown in Table 4, male sex was presented as a protective factor against inadequate TTR, while incorrect use of warfarin was associated with a 3.5 times greater chance of inadequate TTR; and indication for anticoagulation (higher chance of TTR unsuitable for patients with idiopathic stroke and anticoagulation indication due to other reasons). The Central Illustration represents the article’s main results.

Table 4

Variable specification	Reference variable	OR	CI	P value
Male	Women	0.42	0.26 - 0.68	0.00
Age over 60 years	Less than 60	1.08	0.67 - 1.76	0.74
Resident in the metropolitan region of Belo Horizonte	Resident in BH	0.89	0.55 - 1.46	0.66
Resident in the countryside of Minas Gerais	Resident in BH	0.64	0.24 - 1.75	0.39
Incorrect use of warfarin	Correct use	3.55	1.31 - 9.63	0.01
Change in other medications being used	No changes	0.67	0.35 - 1.29	0.23
Change in eating habits	No change	1.45	0.57 - 3.71	0.44
Indication for anticoagulation due to idiopathic stroke	AF	0.65	0.34 - 1.25	0.19
Indication for anticoagulation using a metal prosthesis	AF	1.01	0.55 - 1.88	0.97
Indication for anticoagulation due to thromboembolism	Stroke	0.38	0.08 - 1.86	0.23
Indication for anticoagulation - others	Stroke	0.28	0.10 - 0.86	0.03
Change in eating habits	No change	1.45	0.57 - 3.71	0.44

AF: atrial fibrillation; BH: Belo Horizonte; OR: odds ratio. Source: Prepared by the authors; CI: confidence interval.

Anticoagulation clinics, which are generally multidisciplinary, monitor patients using warfarin, measure the INR, and adjust the dose of the medication. The mean TTR identified at the H2 clinic was below the minimum percentage recommended, which indicates the need to implement strategies that improve the care process. A multicenter study with outpatients undergoing oral anticoagulation control identified that the implementation of anticoagulation clinics contributed to better care for patients using this medication.^14,15 Another study identified that patients treated at anticoagulation reference centers tend to present better treatment results.¹⁶

Therefore, the identification of a mean TTR value that was lower than expected at the H2 outpatient clinic suggests an underutilization of this clinic’s potential. Furthermore, this value encourages the identification of the mean TTR of patients monitored in medical clinics and basic health units, where professionals are considered less prepared to manage warfarin pharmacotherapy, with less infrastructure for the monitoring this patient profile.

Although mean TTR values lower than the desired percentages have been identified in developing countries, as a consequence of different access to health services and cultural and educational disparities that influence treatment,^2,17 the mean TTR at the H1 outpatient clinic was above this value. This indicates that mean TTR at percentages considered desirable is achievable in these countries. A large variation in INR control, measured by TTR, has been identified between different countries, with a population mean model predicting that a TTR of 60% would be necessary for patients to benefit from anticoagulant therapy.⁹

However, despite the benefits of specialized clinics in the control of oral anticoagulation, complicating factors, such as limited time for care, language barriers, advanced age, low educational level, and cognitive difficulties presented by some patients,^2,6 suggest the need for additional educational strategies that consider the specificities and individualities of patients.^18,19 Better understanding the specificities of clinical patients could contribute to a better quality of care offered.

A study involving a multiple logistic regression model showed that a 10% increase in the time outside the INR range was associated with an increased risk of mortality (OR: 1.29; p = 0.001), occurrence of stroke (OR: 1.10; p = 0.006), and other thromboembolic events (OR: 1.12; p < 0.001). The hospitalization rate in the study population was also higher when the INR was outside the target range. This emphasizes the association between suboptimal anticoagulation and undesirable clinical outcomes.⁹

Although the difference between the mean TTR of the clinics showed a statistically significant difference in the descriptive calculations, the anticoagulation clinic was not associated with inadequate TTR in the binary and multiple regressions. It is noteworthy that the present study is based on a real-world context, and although the mean TTR values in the two groups are close to lower TTR threshold, the mean value of 33.46% at H2 in relation to 62.60% at H1 suggests the potential benefits for H1 patients in maintaining the INR within the target therapeutic range for a longer period of time. Furthermore, it suggests that other factors were greater determinants of inadequate TTR than the clinic itself; that is, patient characteristics were more important.

Multiple regression identified that the variable male sex was a protective factor in relation to TTR < 60% (OR: 0.42; 95% CI: 0.26 to 0.68). A multicenter study that considered the global impact of geographic regions on TTR also identified a lower mean TTR value for women when compared to men (women: 53.3 ± 21.3; men: 56.4 ± 21.2; p < 0.001).²⁰ This result points to a greater need to identify possible differences and understand how to carry out treatment between male and female patients, in addition to the implementation of measures that contribute to pharmacotherapeutic success in female patients. It is noteworthy that, even with a higher percentage of female patients (58%), the H1 outpatient clinic had a higher mean TTR than that of H2.

The association between incorrect use of warfarin and inadequate TTR was already expected, considering that adherence is a fundamental factor for treatment success. This finding reinforces the need to invest in educational actions that contribute to better knowledge, empowerment, and adherence of patients who use this medication.^21-24

Regarding the indication for anticoagulation, a greater chance of presenting TTR at lower levels was identified in patients with valvular or non-valvular AF in the binary logistic regression, as well as indication due to other reasons in the multiple regression. No studies were identified that associate variations in TTR with indications for anticoagulation, and additional studies with this focus are recommended.

Although some studies use the measurement of mean INR value to evaluate the quality of anticoagulation, the literature points to the benefits of calculating the TTR for a better projection of achieving the target INR over time.²⁵

The present study has the limitation of having collected data from medical records, which restricted the choice of variables analyzed to those common to the outpatient clinics under study. Furthermore, although the results contribute to a better understanding of factors associated with low quality of anticoagulation, the design and sample used do not allow extrapolation of the data.

The present study aimed to identify the factors associated with TTR variation, considering the importance of TTR for the effectiveness of anticoagulant therapy with a vitamin K antagonist. The outpatient clinics presented different values in the mean TTR; one met the expected percentage, while the other did not. The identification of variables associated with inadequate TTR contributes to identifying weaknesses in the care process and implementing improvement actions.