Background
Oral anticoagulation (OAC) is a lifelong treatment that involves roughly 2 % of the western population [
1]. Specifically, OAC is mainly indicated in non-valvular atrial fibrillation to prevent the incidence of ischemic stroke, as well as in other medical conditions that are associated with thromboembolic complications, such as the deep vein thrombosis [
2]. An inadequate anticoagulation control is also associated with adverse events, which are severe bleeding, thromboembolic events, and mortality [
3]. The most frequent adverse event is severe bleeding, showing an incidence ranging from 2 to 5% yearly, without significant differences between countries [
4]. Thromboembolic complications associated with inadequate anticoagulation control show a yearly incidence of roughly 2% [
5]. Mortality in anticoagulated patients is mainly associated with fatal bleeding, showing an incidence ranging from 0.5 to 1% yearly [
3,
4].
Vitamin K antagonists (VKAs) have been the cornerstone of OAC for more than five decades [
6]. Thus far, VKAs are the most widely used drugs for OAC for selected patients when the use of direct oral anticoagulants (DOACs) is not recommended—e.g., patients with rheumatic mitral valve disease in atrial fibrillation and/or a mechanical heart valve prosthesis [
7]. The main challenges of OAC with VKAs are given by the requirements to maintain an adequate anticoagulation control, as several changes in lifestyle habits are required, such as dietary restriction, drug interactions and frequent blood monitoring [
1,
8].
Literature shows two main possible strategies for OAC management: Physician-managed (i.e., usual care approach) and self-monitoring models [
9]. Self-monitoring models, for their part, encompass patient self-testing and self-management: In self-testing, patients can have their test result managed by their healthcare provider; conversely, in self-management, patients can interpret their international normalized ratio (INR) result, and adjust their own dose of anticoagulant accordingly [
10]. Despite self-monitoring models are a safe option for suitable patients of all ages [
9], physician-managed model is still the most used approach for managing VKAs [
11]. However, empirical evidence and meta-analysis synthesis showed that physician-managed model has been associated with lower levels of health-related quality of life (HRQoL) than the ones reported for self-monitoring [
9,
12‐
14]. The interpretation of the observed differences referred to HRQoL has not found a consensus yet, supporting the hypothesis that suitable patients to self-monitoring receive more benefits when they are able to express their feeling of empowerment in managing their therapy [
9].
HRQoL is an important outcome for anticoagulated patients, acknowledging that higher levels of HRQoL are associated with higher adjustment to treatment [
15]. HRQoL in anticoagulated patients is generally adequate, even if there is still room for improvement [
14,
16]. The individual characteristics that can be changed to achieve an improvement of HRQoL are defined as modifiable determinants, while fixed factors of an individual (e.g. genetic, age) are defined as unmodifiable determinants [
17]. Literature has already shown that patients’ treatment-specific knowledge, health literacy, and treatment beliefs are the most described modifiable determinants of clinical outcomes and treatment adherence [
18‐
20].
Treatment-specific knowledge was described as the level of patients’ awareness about risks and benefits of OAC [
21] being also reported as a positive determinant of anticoagulation control [
22] and being associated with better treatment adherence [
19]. Furthermore, treatment-specific knowledge and adherence to OAC were described to be associated also with treatment beliefs [
23]. Specifically, beliefs are defined as patients’ concerns for treatment, and their understanding of the necessity of being treated with OAC [
23]. Lastly, health literacy was defined as an individual’s capacity to obtain, understand, and function the basic information and services to best manage his or her health, and engage in proper decision-making [
24]. Previous research reported that anticoagulated patients with low health literacy were less adherent to OAC, exhibiting an increased risk of complication and disease-related mortality [
25].
Thus far, it is not clear whether treatment-specific knowledge, health literacy, and treatment beliefs maintain their effects also in modulating the levels of HRQoL in anticoagulated patients, beyond their role in determining clinical outcomes [
26]. In fact, previous research described knowledge, health literacy, and beliefs as potential modifiable determinants of HRQoL [
26]. Research aimed at describing the modifiable determinants of HRQoL is pivotal for anticoagulated patients, as those determinants could be susceptible of improvement by evidence-grounded educational strategies [
14,
26]. However, the role of those determinants in modulation HRQoL still remains less described than the effects of non-modifiable determinants in influencing HRQoL of anticoagulated patients. This gap is particularly important for physician-managed anticoagulated patients using VKAs, as they seem to report lower level of HRQoL than patients managed using self-monitoring [
12‐
14]. Therefore, we aimed to identify the determinants of HRQoL in physician-managed anticoagulated patients using VKAs, highlighting the specific contribution of each non-modifiable and modifiable variable in modulating HRQoL.
Methods
An observational study was conducted involving two anticoagulation clinics (ACs) in the north of Italy from January 2019 to July 2019. The ‘STrengthening the Reporting of OBservational studies in Epidemiology’ (STROBE) checklist was used as a guide for the study reporting (see Additional file
1).
A convenience and consecutive sampling was used to enroll adult patients with a steady coagulation profile and able to complete self-reported questionnaires through a cross-sectional data collection. A good general rule of thumb for determining desirable sample size for studies involving self-report questionnaires is to consider a minimum of 50 participants per domain to allow an adequate variation of the measurements [
27]. In this study, the theoretical construct with higher number of domains was given by health literacy, as it encompasses three different sub-measurements for functional, communicative, and critical health literacy. For this reason, we calculated the desirable sample size considering the number of domains included in the Health Literacy Questionnaire (HLQ), which is described in the measurements’ paragraph [(n
domains = 3)*50 patients =150 patients]. According with previous research with a similar data collection approach [
28], we expected a hypothetical response rate of 70%. Thus, we invited to participate in this study a total of 195 adult anticoagulated patients.
In accordance with previous research about HRQoL in real-world anticoagulated population [
29], the inclusion criteria for this study were: Patients aged ≥18 years, in long-term treatment with VKAs (for at least 3 months) using physician-managed model, speaking fluently Italian. The exclusion criteria were: Patients short-term treated (< 6 months), patients who had to interrupt the treatment for a surgery in the last 3 months, with serious comorbidities (Charlson Comorbility Index [CCI] > 4), and with cognitive impairment (Six Item Screener test [SIS] < 4). Data collection was performed using paper-based case report form (CRF) in the outpatient settings of each involved hospital, and the questionnaires’ compilation required roughly 10 min.
Measurements
Socio-demographic and clinical data (i.e., non-modifiable determinants) were detected from electronic medical records in accordance with the current guidelines, and previous research aimed at describing the anticoagulated population [
7,
30]. Socio-demographics were age, sex, marital status, educational level, and occupation. Clinical variables were time in OAC, clinical indication for OAC, time in therapeutic range (TTR) computed according to Rosendaal method, anamnesis of thromboembolic or bleeding complications in the last 3 months. Modifiable determinants (i.e., knowledge, beliefs and health literacy) and HRQoL were assessed through valid and reliable self-reported questionnaires available in Italian, which were Italian Anticoagulation Knowledge Tool (I-AKT) [
31], Health Literacy Questionnaire (HLQ) [
32], Beliefs about Medicine Questionnaire (BMQ) [
33], and Short Form survey (SF-12) [
34]. These questionnaires provided a measurement of patients’ perceptions about their clinical condition and beliefs related to OAC [
35].
Specifically, the Italian Anticoagulation Knowledge Tool (I-AKT) was used to measure the treatment-specific knowledge [
21,
31]. It encompasses two section: the first section (
n = 20 item) assesses the general anticoagulation knowledge for any kind of anticoagulants, while the second one (
n = 8 item) is specific for the treatment with VKAs. The percentage rate of correct answers was computed into a final score ranging from 0 to 100%. In this study internal consistency reliability was satisfactory, as Cronbach’s α was equal to 0.738 for the overall scale.
Health Literacy Questionnaire (HLQ) is a multidimensional tool composed by 44 items, which could be categorized in three main domains in accordance with Nutbeam’s theory: functional, communicative, and critical literacy [
32,
36]. These domains measure: (a) The ability of patients in being able to use basic skills of daily life functioning, such as writing and reading (i.e., functional health literacy), (b) the ability to obtain and understand health information (i.e., communicative health literacy), and (c) to perform a proper health decision making (i.e., critical health literacy). Each domain has a maximum score on 5-point Likert scale, and higher scores indicate better levels of health literacy [
32]. In this study internal consistency reliability was adequate for each domain, exhibiting the following Cronbach’s α for each domain: functional health literacy = 0.773; communicative health literacy =0.945; critical health literacy = 0.900.
Beliefs about Medicines Questionnaire (BMQ) encompasses two sections related to concerns (
n = 6 item) and necessity referred to the drug prescription (
n = 5 item) from the patients’ perspective. The first domain assesses patients’ anxiety for VKAs, while the second domain assesses the therapeutic need as recognized by patients. The final score was computed on 5-point Likert scale for each domain [
33]. In this study internal consistency reliability was adequate for each domain, showing Cronbach’s α equal to 0.799 for the domain of concerns, and 0.794 for the domain of necessity.
Finally, the short form survey (SF-12) was used to assess HRQoL. It encompasses physical component summary (PCS12), and mental component summary (MCS12) using a standardized score ranging from 0 to 100, wherein higher value indicates a more favorable health perception. According with previous research [
37], it is possible to dichotomize the scores through an adequate versus an inadequate HRQoL if we consider the median split strategy [
34] acknowledging that the median scores of general Italian population, clustered by different age ranges, were previously published [
38]. In this study, both components showed adequate internal consistency with Cronbach’s α equal to 0.699 for PCS12 and 0.687 for MCS12.
Ethical considerations
The institutional review board of reference of each involved centers approved the study (n. 88/INT/2018). The research methodology was performed in accordance to the ethical values of declaration of Helsinki, good clinical practice (GCP) recommendations, and European law on data privacy (GDPR 2016/679 for non-interventional studies). A written informed consent form was requested to each eligible patients before the inclusion into the study. All participants were full informed about the aims and the method of the study, as well as about the confidentiality of their responses.
Statistical analysis
The collected variables were preliminary checked for possible missing data, outliers or errors through an analysis of frequency distribution. Qualitative variables were synthetized using frequency and percentage. Quantitative variables were initially assessed for normality using the analysis of skewness and kurtosis, followed by Shapiro-Wilk test. We employed mean ± standard deviation (SD) for the normally distributed quantitative variables. The determinants of adequate HRQoL were assessed through two logistic regression (LR) models within the framework of generalized linear models. A median split approach was used to dichotomize PCS12 and MCS12, by considering the median values previously described among Italians with an age ≥ 65 years [
38]. Accordingly, PCS12 scores lower than 44.1 were considered as indicating inadequate physical health, while MCS12 scores lower than 46.4 were considered as indicating inadequate mental health. LR models were run considering (a) the correlational analysis interpretations in the first comparisons between the dichotomized PCS12, MCS12 with the collected individual-level variables (i.e., socio-demographic, clinical, knowledge, health literacy, and beliefs), and (b) the potential role of each independent variable considering previous studies [
29,
39,
40]. Variables exhibiting point biserial correlation coefficient (r
pb) with
p-value ≤0.15 were included into the LR models as independent variables [
41]. Subsequently, PCS12 and MCS12 were employed as dichotomous outcomes in the logistic regression (LR) models, after the assessment for possible collinearity between the identified independent variable by checking the strength of their bivariate associations, which should not exceed 0.45 [
42]. The goodness of fit for each LR model was determined using the Hosmer-Lemeshow test and Nagelkerke’s pseudo-
R2. The independent variables were entered simultaneously into the models for examining each variable’s relatively unique contribution to HRQoL. Statistical analysis was run through Statistical Package for the Social Sciences version 22 (IBM Corporation), with a two-tailed significance value set at 0.05.
Discussion
This study provided an overall description of the role of modifiable and non-modifiable determinants of HRQoL in physician-managed anticoagulated patients using VKAs. Among these, health literacy showed interesting effects in determining adequate PCS12 and MCS12. Broadly, promoting adequate levels of health literacy is a public health goal for ensuring equity of care, reduction of health costs, and better achievement of outcomes [
24,
43], acknowledging that health literacy is a key indicator for evaluating the quality of care delivery [
44,
45].
However, in our study, health literacy showed a paradoxical effect, as critical health literacy decreased the likelihood of achieving adequate mental health perception. In accordance with previous research, critical health literacy should predict more positive outcome, such as self-care and HRQoL [
24,
46,
47]. In our study, we reported that patients with higher critical thinking abilities about their chronic treatment (higher levels of critical health literacy) reported lower levels of mental health perception. To interpret this paradoxical result, we hypothesized that the physician-managed model in managing OAC could influence the relationship between critical health literacy and mental health perception. In fact, patients with higher levels of critical health literacy are often those exhibiting higher desire to take control of their health management, including treatment [
48,
49]. In other words, patients with higher critical health literacy are generally inclined for self-management, as they would like to take control of their own chronic condition [
50]. This suggests, possibly explaining this result, that patients with high critical health literacy can experience more frustration and distress when they are managed using traditional models for OAC, inasmuch they are not actively involved in the therapeutic decision-making [
50,
51]. Notwithstanding, it should be empirically tested whether the patients with higher critical health literacy, managed using self-monitoring models, exhibit higher levels of mental health perception.
In this study, the levels of knowledge, health literacy, and recognition of the therapeutic need to be adherent with OAC are generally low. In accordance with previous evidence, more attention should thus be paid in improving the modifiable determinants of health through educational interventions, especially identifying patients that could have benefits from self-monitoring [
19,
20,
23,
52,
53]. In other words, the default use of the traditional OAC management model—as it happens in the real-world clinical practice in many countries—seems to meet poorly the requirements for optimizing patients’ HRQoL through enhancing their knowledge, health literacy, and correct beliefs: One size (traditional OAC management) cannot fit all!
Thus far, the main recognized criteria to identify suitable patients for self-monitoring models were given by (a) anamnesis of adequate anticoagulation control, (b) motivation to be engaged in managing OAC through self-monitoring, (c) adequate treatment-specific knowledge, and (d) adequate cognitive functioning [
7,
53]. That said, this study contributes to further understand which criteria could be useful in practice to identify suitable patients for self-monitoring, as per the role of health literacy in determining HRQoL. Health literacy was previously presented as a public health goal, explained within an integrated framework that encompasses knowledge, motivation, and cognitive functioning as personal determinants of each health literacy level (i.e., functional, communicative, and critical) [
54]. This implies that the individual’s health literacy levels are shaped as the result of the relationships between personal determinants and situational factors (e.g., healthcare and social contexts) [
54]. For this reason, the assessment of health literacy as a pivotal criteria for identifying suitable patients for self-monitoring could be a strategic practical implication for the international healthcare providers’ community, due to patients with higher ability to perform proper health decision making (i.e., critical health literacy) could be easily identified as suitable for self-monitoring models. Furthermore, the tools aimed at assessing health literacy levels are available in many languages, being also validated in different settings [
55].
We found almost no significant relationships between HRQoL and sociodemographic and clinical characteristics, while previous evidence showed that worsen HRQoL was associated with being young, female, having lower education, and with higher comorbidity, being in short-term treatment (lower than 1 year), and with previous haemorrhagic complications [
10,
12]. In our study, the age was the only sociodemographic determinant of physical health perception. This result is consistent with previous research [
16,
56]. Hence, further research should in-depth clarify how socio-demographics and clinical variables modulate health perception in anticoagulated patients over time, using longitudinally collected data.
Study limitations
This study has some limitations. Firstly, the adopted cross-sectional data collection using a limited sample did not provide information on the trajectory of HRQoL and its determinants over time, and possibly the associations between independent variables and outcomes were under-estimated. Secondly, the convenience sampling suggested caution in generalizing the results. Thirdly, the comparison between different OAC management models was not included in the study design, as only physician-managed anticoagulated patients were enrolled, acknowledging that self-monitoring models report a limited use in practice in the context of this study; however, future research should compare the effects of health literacy levels in modulating HRQoL between different OAC models. Furthermore, the role of the modifiable determinants of HRQoL could be context-specific, as some cultural and social characteristics could influence how they interact with HRQoL. For this reason, cross-national research will be useful to overcome this limit. Finally, variables related to stress perception were not collected in this study: To better understand the paradoxical result given by the association between high levels of critical health literacy and lower levels of perceived mental health, future investigations should also keep into account the assessment of stress perception.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.