https://orcid.org/0000-0001-7591-4104
Figures
Abstract
Background
Patients with atrial fibrillation (AF) have a high risk for recurrent clinical events after an ischemic stroke. Direct oral anticoagulants (DOAC) are prescribed for secondary prevention. Adherence to DOAC is crucial mainly because of their short elimination half-life. Non-adherence to DOAC can negatively impact patients’ outcomes. The relationship between (non-)adherence and recurrent clinical events is unknown in AF patients after initial stroke. We investigated adherence to DOAC in stroke survivors with AF who were included in the MAAESTRO study at the University Hospital Basel, Switzerland, between 2008 and 2022.
Methods
This study is a secondary analysis of data from MAAESTRO with a matched nested case-control design and 1:2 ratio. DOAC intake was measured with a small electronic device (Time4MedTM). We defined two arbitrary intervals of 17 days and 95 days as the longest time spans with electronic monitoring data per patient to maximize the number of participants with adequate amount of observation time available for analysis. Taking and timing adherence were calculated retrospectively i.e., prior to the recurrent event for cases. Trendline analysis of adherence over 95 days was calculated. Linear regression analysis was performed after adjusting for the co-variables age and daily pill burden. Sensitivity analysis was performed with controls for intervals in the reverse direction (prospectively).
Results
We analyzed 11 cases and 22 matched controls (mean age: 75.9 ± 9.2 years vs. 73.1 ± 8.4 years; n.s.) with similar stroke characteristics (NIHSS, mRS, MoCA) and 36.4% women in each group. Mean adherence values were high and similar between cases and controls (95 days taking: 87.0 ± 18.9% (cases) vs. 90.8 ± 9.8% (controls), n.s.; similar values for timing adherence). Six hemorrhagic and five ischemic events had occurred. Compared to controls, a significantly higher 95 days taking adherence was observed for hemorrhagic events (96.0 ± 5.0% (cases) vs. 88.1 ± 11.5% (controls); p<0.01) and a significantly lower 95 days taking adherence was observed for ischemic events (75.7 ± 24.8% (cases) vs. 94.2 ± 6.2% (controls), p = 0.024). Values for timing adherence were similar. A non-significant downward linear trend of adherence was observed over 95 days independently of the clinical events. The sensitivity analysis showed that the direction of the interval had negligible impact on the 95 days adherence.
Conclusion
Because recurrent ischemic events after an AF-related stroke were associated with low adherence to DOAC <76%, adherence enhancing interventions seem crucial in anticoagulated AF-patients. However, AF-patients with high adherence might benefit from a regular re-assessment of the bleeding risk as hemorrhagic complications were associated with adherence to DOAC >96%.
Citation: Rekk K, Arnet I, Dietrich F, Polymeris AA, Lyrer PA, Engelter ST, et al. (2024) Relationship between electronically monitored adherence to direct oral anticoagulants and ischemic or hemorrhagic events after an initial ischemic stroke—A case control study. PLoS ONE 19(4): e0301421. https://doi.org/10.1371/journal.pone.0301421
Editor: Giulio Francesco Romiti, Sapienza University of Rome: Universita degli Studi di Roma La Sapienza, ITALY
Received: November 3, 2023; Accepted: March 15, 2024; Published: April 25, 2024
Copyright: © 2024 Rekk et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are available in the Supporting Information files.
Funding: The MAAESTRO study on which the present article is based was funded by the University Hospital Basel, Neurology Fund and Scientific Fund, and a grant from the Swiss Academy of Medical Sciences and the Bangerter-Foundation (YTCR 31/17). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Stroke is the second leading cause of death worldwide and the third leading cause of disability [1–3]. Stroke prevalence reaches 7% in the elderly [4] and is sex specific due to societal and biological gender-related factors [5]. Atrial fibrillation (AF) increases the risk of stroke. In patients with AF, the CHA2DS2-VASc score is used for risk stratification of ischemic stroke and thromboembolism [6, 7], while the HAS-BLED score indicates the risk of major hemorrhagic event [8].
Oral therapy with direct oral anticoagulants (DOAC) is essential for the secondary prevention of stroke in AF-patients [9]. Because of their short elimination half-lives and the lack of standard monitoring methods [9, 10], DOAC require a high level of adherence, that is, patients should take medications very thoroughly according to the recommendation of their health care providers [11]. There are different manners to measure adherence that generate data of various quality [12]. The electronic monitoring of medication intakes delivers objective data that enable to calculate specific adherence metrics [12, 13]. With electronic records, adherence can be estimated by taking the correct amount of the prescribed medicine (taking adherence) and at the appropriate time (timing adherence) [11].
Prevalence of recurrent events after an initial stroke is estimated at 1% after 30 days and raises to 19% within the first five days [14–17]. Reasons for recurrence are various and mostly non-modifiable such as age, previous stroke, congestive heart failure, among others [18]. However, non-adherence to anticoagulation therapy has been mentioned by researchers [19] and was even associated with negative clinical outcomes such as recurrent thromboembolic events [20]. Only few studies investigated the extent of non-adherence to anticoagulation therapy that leads to recurrent events after an initial stroke in AF-patients. A Spanish study included 132 AF-patients with electronic monitoring of the DOAC intake over 4 months [21]. Taking adherence was very high with a mean of 95.8% and thromboembolic events were more often observed in patients who failed to take >5% of the DOAC doses compared to patients who missed intakes only between 0 and 5% [21].
The MAAESTRO study was conducted at the Stroke Unit of the University Hospital Basel, Switzerland [22] (trial registration: ClinicalTrials.gov NCT03344146). Recruitment period started 01 January 2018 and ended 08 September 2021. All participants provided written informed consent. The MAAESTRO study aimed at evaluating the effect of an intake reminder on the adherence to DOAC in stroke patients with AF [23]. In brief, 130 stroke survivors were included who self-administered a DOAC after discharge [22]. In addition, patients recorded the DOAC intake with a small electronic monitoring device (Time4Med TM Adherence Innovations, Hong Kong, China) [13]. The study consisted of an observational phase of 3–6 months to collect baseline adherence data, followed by randomization to a group with reminder and a group without reminder for 3 months, and crossover for 3 more months [22]. During the 12-month MAAESTRO study, a total of 21 recurrent ischemic or bleeding events were observed in 16 patients (12.3%). These patients are the subject of the present study.
Aims
To investigate the relationship between adherence to DOAC and clinical events in secondary stroke prevention in patients with AF. The hypothesis is that patients with a recurrent ischemic/hemorrhagic event will show a lower/higher level of adherence to DOAC in secondary prevention compared to patients with no recurrent clinical event.
Materials and methods
Study design and original data access
This study is a secondary analysis of data from the MAAESTRO study that was approved by the Ethics Committee of Northwest/Central Switzerland (EKNZ 2017–01552) and registered at ClinicalTrials.gov (NCT03344146). All participants gave written informed consent; study participation was voluntary. Original data were accessed on 08 January 2023. Authors had no access to information that could identify individual participants during or after data collection. We used a matched nested case-control design [24] that allows to study time-dependent exposures on rare outcomes where the use of randomized controlled trials (RCT) would be unethical [24]. This design allowed us to match a patient with a recurrent clinical event at a given point in time (case) to a patient with similar characteristics who had not experienced a recurrent clinical event (control). We used the STROBE case-control reporting guidelines [25].
Case definition and cohort time axis
A patient was defined as case if they had developed a clinical event during the MAAESTRO study. A clinical event was defined as ischemic stroke, myocardial infarction, transient ischemic attack (TIA), intracranial hemorrhage, major extracranial hemorrhage or death [22]. For each case, we considered only the first recurrent event; subsequent recurrent events were not considered. In addition, adherence data had to be available prior to the clinical event. Patients with no adherence data were excluded.
Selection of matching factors
Patient-related matching factors were sex, age, CHA2DS2-VASc score and HAS-BLED score as they are associated with a risk for clinical events. Allocation to MAAESTRO study phase (observational phase; interventional phase with or without reminder use) was incorporated as matching factor imposed by the RCT design of the MAAESTRO-study.
Selection and number of controls
Patients who did not develop a recurrent event (i.e., controls) were randomly selected from the list of MAAESTRO patients who finished the study. Sampling was performed without replacement and with excluding the patient from the set. Once the list of all candidate controls was obtained, the final controls were selected with a randomly generated list of numbers. We selected a case-to-control ratio of 1:2 because it provides gains in statistical power [26].
Stroke characteristics
Stroke severity was assessed on the NIHSS (National Institute of Health Stroke Scale; minor stroke, 1–4; moderate stroke, 5–15; moderate to severe stroke, 16–20; severe stroke, 21–24 [27]). Post-stroke functional status was assessed with the mRS (modified Ranking Scale; mild disability, 1; moderate disability, 2–4; extreme disability, 5 [28]). Cognitive screening was done with the MoCA (Montreal Cognitive Assessment; mild impairment, <26; severe impairment, <19 [29, 30]).
Electronic monitoring data and adherence metrics
The electronic monitoring data were processed as described previously [31]. There were no missing data as lack of electronic data was defined as missed intake. Days of hospitalization, planned treatment pauses, and device dysfunctions were excluded from the calculation [31]. We selected two interval lengths of 17 days and 95 days to calculate adherence to DOAC. This arbitrary choice permitted to select the largest number of cases from the MAAESTRO study with the longest time span of electronic monitoring data available per patient, that is, 11 cases over 17 days and 9 cases over 95 days. We selected the interval prior to the recurrent event for cases, and prior to the last day of the MAAESTRO study i.e., observation or intervention phase for controls (i.e., retrospectively). We performed a sensitivity analysis and defined for controls the inverse direction of the interval (i.e., prospectively) that is, starting from the first day of the MAAESTRO observation or intervention phase.
Taking and timing adherence were calculated as percentage of the number of recorded doses divided by the number of prescribed doses during the predefined interval length. For timing adherence, we defined a grace period for the intake to occur within 25% of the median intake time i.e., ±3 hours for a twice daily regimen and ±6 hours for a once daily regimen [11, 22, 31].
Statistical analysis
Microsoft® Excel Version 16.54 (Microsoft Corporation, Redmond, WA, USA) was used for adherence calculations. SPSS statistics version 29 (IBM Corp., Armonk, New York, USA) was used for patient matching and statistical analysis. Numbers (with percentages) are presented for binary and categorical variables; means (± standard deviations, SD) or medians (with min-max) for continuous variables. Box-plots are presented with median (central line) separating the first and the third quartile; upper and lower whisker limits, and outliers (dots). Not normally distributed continuous variables (i.e., taking and timing adherence; characteristics of patients in the matched control groups) were evaluated using the Mann-Whitney U test. A p-value <0.05 was considered statistically significant with a two-sided approach for all patients and one-sided for specific events. Cases were subdivided into hemorrhagic and ischemic events and subgroup analysis was performed.
Weekly taking and timing adherence over the interval of 95 days were calculated for all patients. Trend-lines with simple linear regression (Y = aX+b) were calculated, with a negative slope (a) indicating downward trend of adherence over time. R-squared was calculated and averaged for cases and controls separately. Means with SD of the slope values were calculated for cases and controls separately. Linear regression analysis was performed after adjusting for the co-variables age and daily pill burden. We did not control for additional confounders because of much smaller groups than in the overall study. Retrospective data over the 95 days interval were used to calculate taking and timing adherence. Each model was fitted once in all pairs and once stratified by type of event.
During the medical visits that took place after the observation phase and after the intervention phase of the MAAESTRO study, patients were given the possibility to comment on their past intakes with a graphical visualization of the electronic monitoring data. Field notes were recorded by the investigator during the medical visits. Patients’ statements about difficulties with the DOAC intake are reported.
Results
From the 16 eligible patients from the MAAESTRO study, five were excluded from the analysis because of missing adherence data following discharge (2), death (2) and technical issues (1). From the remaining 11 cases, six had developed hemorrhagic events (4 with major extracranial hemorrhage; 2 with intracranial hemorrhage) and five had developed ischemic events (4 with recurrent stroke or TIA; 1 with myocardial infarction). Five cases were observed during the observational phase (45.5%) and six during the interventional phase, of whom two used a reminder (18.2%) and four used no reminder (36.4%). The 11 cases were matched with 22 controls.
Cohort characteristics
Mean age, sex distribution, mean risk scores and lifestyle factors were similar between cases and controls (Table 1). Stroke characteristics (NIHSS, mRS, MoCA) were comparable between cases and controls. All patients had at least two other medicines. Cases had more often hyperlipidemia compared to controls (100% vs 63.5%; p = 0.021). The use of anticoagulants (i.e., DOAC or vitamin K antagonists) prior to the initial stroke event was more often observed in cases compared to controls (72.7% vs 36.4%; p = 0.051). All patients were discharged with a DOAC, apparently with no preferred agent (Table 1).
Statistically significant values are highlighted in bold.
Regarding the anticoagulant agents used at the moment of the recurrent event, one switch from apixaban to rivaroxaban had occurred after hospital discharge without known reasons. From the 11 cases, five had a recurrent event under apixaban (45.5%), and three each under edoxaban and rivaroxaban (27.3%). No recurrent event was observed with dabigatran.
The median onset time for a recurrent ischemic event was 155 days (range: 47–214 days) for recurrent stroke; the myocardial infarction occurred after 231 days. The median onset time for a recurrent hemorrhagic event was 110 days (range: 32–182 days) for major extracranial hemorrhage and 266 days (range: 265–266) for intracranial hemorrhage. For one patient, the DOAC was discontinued two days before a planned surgical procedure, in accordance with guidelines [32], and hemorrhagic complications occurred during surgery. A recurrent event occurred after the re-initiation of the DOAC.
Adherence
For two cases, the recurrent events occurred 17 days and 30 days after entry in the MAAESTRO study. These cases were excluded from the 95 days calculation because monitoring data were not enough. Taking and timing adherence values were similar for cases and controls over the interval of 17 and 95 days, respectively, and ranged from 80.1 ± 27.7% to 90.8 ± 9.8% (Table 2, upper panel). Differences were observed when ischemic and hemorrhagic events were analyzed separately. For hemorrhagic events, taking and timing adherence values over 95 days were significantly higher for cases compared to controls (taking adherence: 96.0 ± 5.0% vs 86.2 ± 11.1%; U = 6.0; p<0.01; Table 2, middle panel; Fig 1). For ischemic events, taking and timing adherence values were significantly lower for cases compared to controls independently of the interval length with values ranging between 67.1 ± 36.3% for cases and 95.8 ± 3.5% for controls (U = 40.5; p<0.05; Table 2, lower panel; Fig 2).
See Table 2, middle panel for the detailed values. Statistical significance is marked with an asterisk.
Box plots of five cases with ischemic events (dotted) and 10 matched controls (plain) with taking adherence (upper panel C) and timing adherence (lower panel D) over the intervals of 17 days (white) and 95 days (grey). See Table 2, lower panel for detailed values. Statistical significance is marked with an asterisk.
Statistically significant values are highlighted in bold.
The linear regression analysis showed that neither age nor pill burden had significant effects on taking and timing adherence of controls compared to cases after adjusting for the co-variables. The difference in adherence between cases and controls was similar with approximately +15%-point difference for ischemic events and approximately -10%-point difference for hemorrhagic events (Tables 2 and 3).
Adherence change prior to the recurrent event
For 9 cases and 18 controls, mean weekly adherence (taking and timing) over 95 days prior to the recurrent event ranged from 100% to 14.3% (see illustrative example Fig 3). A downward sloping trend was observed in 5 (55%) cases and 8 (44%) controls for taking adherence, and in 5 cases and 10 controls (55% each) for timing adherence. The average slopes of the trendlines were similar in cases and controls with more pronounced negative slopes in cases (taking adherence: -0.7 ± 1.6% vs -0.2 ± 1.9%, n.s.; timing adherence: -0.7 ± 1.5% vs -0.4 ± 2.2%, n.s.; Table 3). Values were similar for hemorrhagic and ischemic events separately (Table 4). Mean R-squared were similar for taking and timing adherence and varied between 0.1903 ± 0.17 for cases (range: 0.0099–0.5088) and 0.163 ± 0.17 (range: 0–0.5023) for controls.
Sensitivity analysis
Mean time span of adherence monitoring for controls was 126 ± 37 days (range: 95–187 days). Adherence values of controls did not differ if the direction of the intervals was analog to the cases i.e., retrospective (starting the last day of the study phase) or inverse to the cases i.e., prospective (starting the first day of the study phase). When mean adherence values of cases were compared to controls, the results were identical independently if controls were taken prospectively (Table 5) or retrospectively (Table 2), with the exception of 95 days timing adherence of the ischemic events that lost significant difference with prospective controls (Table 5).
Patients’ statements
Three cases reported difficulties with the DOAC use prior to the recurrent event. One patient was taking a twice daily medication only once a day without knowing his error prior to a recurrent myocardial infarction. One patient mentioned 1–2 missed doses per week; his taking and timing adherence were 93.8% over 17 days and 98.9% over 95 days prior to a recurrent intracranial hemorrhage. One patient reported psycho-social problems because of the death of a family member; her adherence was 100% 17 days prior to a major extracranial hemorrhage.
Discussion
This study provides a detailed insight into the adherence of patients with AF-related ischemic stroke who developed a recurrent event despite being appropriately anticoagulated with DOAC. Adherence to DOAC over 95 days prior to the event was high with mean taking adherence values of 87.0 ± 18.9% for cases and 90.8 ± 9.8% for controls. These results appear to be in line with those from previous studies with electronic measure of adherence to DOAC. As an example, taking adherence of 90.8% and 89.2% at 6 and 12 months were observed after a mixed intervention [33, 34]. However, in our study, adherence prior to the recurrent event was significantly higher in patients with recurrent hemorrhagic events, and lower in patients with recurrent ischemic events compared to matched controls. In other terms, a mean taking adherence above 96% was associated with hemorrhagic events, while a value below 76% was associated with ischemic events. In addition, the weekly adherence showed a moderate downward sloping trend in patients with a recurrent event, albeit without statistical significance. To our knowledge, this is the first time that levels of non-adherence to DOAC are directly associated to recurrent events in stroke survivors with AF with a distinction between hemorrhagic and ischemic events.
Study design
We selected a nested case-control design for several reasons. The case-control is an observational epidemiological method used to evaluate factors when the outcome is infrequent [35] or occurs after a long latent period [36]. Even if an initial stroke increases the risk for a second stroke [15, 17], recurrent events after a stroke while on appropriate anticoagulation are uncommon with an estimated prevalence of 4.6% within 3 months after the index stroke, calculated from 2,082 patients in 11 centers in Switzerland, Germany and the USA [19]. Incidences from population-based studies range from 5.1% in Korea [37] to 15.5% in Italy [38]. Consequently, we have extrapolated that between 6 and 20 patients with recurrent events would be best identified from the 130 patients recruited in the MAAESTRO study. Further, the nested design enables to perform a study within a cohort and compare patients who developed the event of interest with randomly selected patients among those who did not develop the event. One of the challenges in case-control studies is the selection of the matching factors to obtain a similar distribution of patients’ characteristics. The intervention in the MAAESTRO study was aimed at increasing adherence to DOAC and was divided in two different phases i.e., with and without reminder use. Previous findings showed that reminder use increases medication adherence [34]. Thus, the allocation to the phase of MAAESTRO (observational phase; interventional phase with or without reminder use) was incorporated as matching factor to reduce confounding effect. By doing so, we diminished the risk of selection bias and obtained a higher level of evidence. Finally, it seemed unethical to develop a randomized controlled study on (non-)adherence to secondary prevention medicine with the endpoint being a debilitating recurrent event such as a myocardial infarction or a severe hemorrhage.
Recurrent events
Among the 130 patients who participated in the 12-month MAAESTRO study after an AF-related ischemic stroke, a total of 21 recurrent events were observed in 16 patients (12.3%). This is similar to the cumulative incidence of 8.3% AF-patients from 8 centers in the USA who developed ischemic events (6.7%) and intracranial hemorrhage (1.6%) within 90 days of a cardioembolic stroke [39]. In our cohort study, six cases (4.6%) developed hemorrhagic event, which are expected complications in DOAC-treated patients [40]. However, when compared to vitamin K antagonists, the risk of intracranial hemorrhage is reduced [41] and associated with better outcomes [40]. Further, when balancing between thromboembolic risk, clinical outcome, and hemorrhagic risk, the sentence “fear the clot not the bleed” [42] places the emphasis clearly on avoiding a recurring thromboembolic event. In our study, cases with hemorrhagic events had taking adherence values of 91% over 17 days and 96% over 95 days, which are much higher than the usual cutoff of 80% used in adherence studies [43]. The use of the 80% cutoff may explain why no difference for hemorrhagic events was observed between adherent or non-adherent patients in a systematic review [44]. Thus, interventions to optimize chronic DOAC treatment in AF-patients should not be limited to adherence enhancing actions in view of avoiding ischemic events. The regular re-assessment of bleeding risk might be equally important because hemorrhagic complications were associated with high adherence to DOAC >96%.
In our study, all cases and 64% of controls presented with dyslipidemia at study entry. This corresponds to a well-known risk factor for stroke, as a history of hyperlipidemia has been statistically associated with recurrent ischemic events [37, 39], among the complex interplay of stroke risk factors. Thus, our patients’ characteristics coincide with the usual stroke risk factors and our results are generalizable.
Electronic monitoring
Among the different methods to measure adherence, we selected the electronic method because it is the gold standard in adherence research [45]. One persistent concern with any indirect method of adherence measure is that the recording does no guarantee the ingestion of the medicine. Moreover, a Hawthorne effect (i.e., an “aware or unconscious complex behavior change in a study environment” [46]) cannot be excluded when individuals know that they are observed [31, 47]. Although all medical research with observational design is prone to the Hawthorne effect, its significance disappears in well-designed RCT [46] because they are designed to reduce the risk of bias. Moreover, it can be assumed that all groups are equally expose to the Hawthorne effect so that its impact on the main outcome may be reduced [48]. Finally, in clinical studies where adherence measurement is unblinded and part of the intervention, it is unlikely that patients register the intakes without ingesting the medication.
Interval lengths and direction
The length of 3 months is the most commonly used period of use for electronic bottles in adherence studies, with a range between one week to 24 months [49]. Reasons in favor of 3 months are that this length of time is needed to develop a reliable medication habit [50] and it gives sufficient information about the patient’s adherence [51]. For the data collection of the controls, we selected the same direction as for the cases, that is, retrospectively prior to the last day of the MAAESTRO study phase (observation or intervention phase). The sensitivity analysis showed that the reversing of the direction had a negligible effect on the adherence values, with a difference of approximately 2% for all patients and a partial loss of the significance only for the 95 days timing adherence of the ischemic events. Because of the low number of 4 cases for this analysis, this result must be interpreted with caution. Nevertheless, taking adherence values and the interval length of 95 days seem representative for long-term adherence calculation, and deliver robust results.
Stroke characteristics
We used three different scores to capture the functional disabilities after an acute stroke, according to medical practice. The NIHSS estimates stroke severity, the mRS functional status, and MoCA cognitive impairment. In our study, all cases and controls had minor stroke (mean NIHSS <2.2) and low functional and cognitive deficits (mean mRS <2 and mean MoCA <25). This is likely to be the consequence of one specific inclusion criteria of the MAAESTRO study, namely that patients had to be able to manage their medication by themselves, which require cognitive and functional capacities. The severity of stroke, as measured by NIHSS, can influence the ability of patients to attain high levels of adherence to medication. Similarly, cognitive impairment, even mild, can contribute to poor adherence [52, 53]. Thus, it was not surprising to observe high adherence values in our cohort of patients who had little post-stroke impairment. Nevertheless, cases with a recurrent event had a significantly suboptimal adherence compared to controls, which emphasizes the deep individual nature of medication adherence. Finally, post-stroke impairment has been reported to affect multiple cognitive domains [54], of which some may be related to medication adherence.
Trendline analysis of adherence
A slight downward sloping trend of the adherence prior to the recurrent event was observed. This trend fits common sense for ischemic events, but might be counterintuitive for hemorrhagic events in view of mean 95 days adherence values that were lower for the former and higher for the latter. In fact, the trendlines of the cases and controls were moderate and mean R-squared values were below 0.2, indicating that the association is weak and that a linear relationship between adherence fluctuation and time is unlikely. Furthermore, the illustrative case presented is exceptional with an overall taking adherence of 40% over 95 days, and a drop to 14.3% the week prior to the recurrent event, which equals to the intake of two tablets per week. More data are needed to investigate adherence trajectories over time.
Strengths and limitations
We acknowledge several strengths. Firstly, we used a nested design and matched cases to controls from the same study population. Thus, we were able to identify infrequent outcomes (i.e., recurrent events in AF-related ischemic stroke patients) in a setting where randomization would be unethical. Secondly, we selected matching factors that are associated with the outcome of interest (i.e., recurrent event) and thus, are appropriate. Thus, overmatching was avoided and the statistical efficiency of the analysis was guaranteed [24]. Thirdly, we generated highly detailed information about adherence before a recurrent event through the use of electronic monitoring. This degree of insight and knowledge is unprecedented in the literature. Consequently, the association between the main results (i.e., adherence) and the outcome (i.e., clinical events) is of good evidence. Fourthly, the original MAAESTRO study has generated data with high quality that already permitted another secondary data analysis [55]. Fifthly, we used a small electronic device to monitor the medication intake, Time4MedTM, that was well accepted by the patients. This method delivers objective real-life data and gives credibility to the adherence metrics and thus, to our overall results.
We acknowledge some limitations. Firstly, our sample size with 11 survivors of AF-related ischemic stroke is relatively small. Thus, our results need to be interpreted with caution. Nevertheless, the methodological rigor of our design gives robustness to our results. To confirm our findings, a larger sample size might be necessary. Secondly, the results are limited to an elderly Swiss population from the University Hospital Basel, Switzerland. Thus, the generalizability of our findings is limited. For further analysis, the sample needs to be more heterogeneous.
Conclusion
Our study showed that recurrent events occurred in AF-stroke patients under DOAC treatment, with ischemic events being associated with low adherence to DOAC and hemorrhagic complications being associated with high adherence to DOAC. Thus, even if adherence-enhancing interventions seem crucial in anticoagulated AF-patients, a regular re-assessment of the hemorrhagic risk might be equally appropriate. Thus, interventions to optimize chronic DOAC treatment in AF-patients should include a regular re-evaluation of the adherence and a re-assessment of the hemorrhagic risk. Our results might influence the counseling of DOAC patients after a stroke.
Supporting information
S1 File. Raw adherence values (taking and timing) with prospective or retrospective interval for cases and controls over 95 days and 17 days.
https://doi.org/10.1371/journal.pone.0301421.s001
(XLSX)
S2 File. Raw variables including age and daily pill burden for 11 cases and 22 matched controls for the linear model calculation.
https://doi.org/10.1371/journal.pone.0301421.s002
(XLSX)
S3 File. Raw weekly adherence values (taking and timing) for 9 cases and 18 matched controls including date of clinical event and start date of calculation.
https://doi.org/10.1371/journal.pone.0301421.s003
(XLSX)
References
- 1. World Health Organization. The top 10 causes of death 2020 [13.04.23]. Available from: https://www.who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death.
- 2. World Health Organization. Global Health Estimates: Life expectancy and leading causes of death and disability 2019 [14.04.23]. Available from: https://www.who.int/data/gho/data/themes/mortality-and-global-health-estimates.
- 3. Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Anderson CS. Long-term disability after first-ever stroke and related prognostic factors in the Perth Community Stroke Study, 1989–1990. Stroke. 2002;33(4):1034–40. pmid:11935057.
- 4. Fatemeh R, Mojgan R, Roumina R, Mohsen K. Prevalence of stroke in the elderly: a systematic review and meta-analysis. Interdiscip Neurosurg. 2023;32:101746. https://doi.org/10.1016/j.inat.2023.101746.
- 5. Ospel J, Singh N, Ganesh A, Goyal M. Sex and gender differences in stroke and their practical implications in acute care. J Stroke. 2023;25(1):16–25. Epub 2023/02/07. pmid:36746379; PubMed Central PMCID: PMC9911850.
- 6. Lane DA, Skjoth F, Lip GYH, Larsen TB, Kotecha D. Temporal trends in incidence, prevalence, and mortality of atrial fibrillation in primary care. J Am Heart Assoc. 2017;6(5). Epub 20170428. pmid:28455344; PubMed Central PMCID: PMC5524079.
- 7. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137(2):263–72. Epub 20090917. pmid:19762550.
- 8. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest. 2010;138(5):1093–100. Epub 20100318. pmid:20299623.
- 9. Hindricks G, Potpara T, Dagres N, Arbelo E, Bax JJ, Blomström-Lundqvist C, et al. 2020 ESC Guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS): The Task Force for the diagnosis and management of atrial fibrillation of the European Society of Cardiology (ESC) Developed with the special contribution of the European Heart Rhythm Association (EHRA) of the ESC. Eur Heart J. 2020;42(5):373–498. pmid:32860505
- 10. Siedler G, Macha K, Stoll S, Plechschmidt J, Wang R, Gerner ST, et al. Monitoring of direct oral anticoagulants plasma levels for secondary stroke prevention. J Thromb Haemost. 2022;20(5):1138–45. Epub 20220312. pmid:35171533.
- 11. Vrijens B, De Geest S, Hughes DA, Przemyslaw K, Demonceau J, Ruppar T, et al. A new taxonomy for describing and defining adherence to medications. Br J Clin Pharmacol. 2012;73(5):691–705. pmid:22486599; PubMed Central PMCID: PMC3403197.
- 12. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med. 2005;353(5):487–97. pmid:16079372
- 13. Arnet I, Rothen JP, Hersberger KE. Validation of a novel electronic device for medication adherence monitoring of ambulatory patients. Pharmacy (Basel). 2019;7(4). Epub 2019/11/24. pmid:31756904; PubMed Central PMCID: PMC6958479.
- 14. Hardie K, Hankey GJ, Jamrozik K, Broadhurst RJ, Anderson C. Ten-year risk of first recurrent stroke and disability after first-ever stroke in the Perth Community Stroke Study. Stroke. 2004;35(3):731–5. Epub 20040205. pmid:14764929.
- 15. Hillen T, Coshall C, Tilling K, Rudd AG, McGovern R, Wolfe CD, South London Stroke R. Cause of stroke recurrence is multifactorial: patterns, risk factors, and outcomes of stroke recurrence in the South London Stroke Register. Stroke. 2003;34(6):1457–63. Epub 20030515. pmid:12750544.
- 16. Hankey GJ. Secondary stroke prevention. Lancet Neurol. 2014;13(2):178–94. Epub 20131220. pmid:24361114.
- 17. Stahmeyer JT, Stubenrauch S, Geyer S, Weissenborn K, Eberhard S. The frequency and timing of recurrent stroke: an analysis of routine health insurance data. Dtsch Arztebl Int. 2019;116(42):711–7. pmid:31711561; PubMed Central PMCID: PMC6891883.
- 18. Paciaroni M, Agnelli G, Caso V, Silvestrelli G, Seiffge DJ, Engelter S, et al. Causes and risk factors of cerebral Ischemic events in patients with atrial fibrillation treated with non-vitamin K antagonist oral anticoagulants for stroke prevention. Stroke. 2019;50(8):2168–74. Epub 20190625. pmid:31234756.
- 19. Polymeris AA, Meinel TR, Oehler H, Holscher K, Zietz A, Scheitz JF, et al. Aetiology, secondary prevention strategies and outcomes of ischaemic stroke despite oral anticoagulant therapy in patients with atrial fibrillation. J Neurol Neurosurg Psychiatry. 2022;93(6):588–98. Epub 20220408. pmid:35396339; PubMed Central PMCID: PMC9148984.
- 20. Seiffge DJ, De Marchis GM, Koga M, Paciaroni M, Wilson D, Cappellari M, et al. Ischemic stroke despite oral anticoagulant therapy in patients with atrial fibrillation. Ann Neurol. 2020;87(5):677–87. Epub 20200212. pmid:32052481; PubMed Central PMCID: PMC7383617.
- 21. Solla-Ruiz I, Villanueva-Benito I, Paredes-Galán E, Salterain-González N, Oria-González G, De La Cuesta-Arzamendi F, et al. Differences between patient-driven adherence to vitamin K antagonists and direct oral anticoagulants. Do few missed doses matter? ACO-MEMS Study. Thromb Res. 2019;179:20–7. Epub 20190424. pmid:31075697.
- 22. Polymeris AA, Albert V, Hersberger KE, Engelter ST, Schaedelin S, Arnet I, et al. Protocol for MAAESTRO: electronic Monitoring and improvement of Adherence to direct oral Anticoagulant treatment-a randomized crossover study of an Educational and reminder-based intervention in ischemic STROke patients under polypharmacy. Front Neurol. 2018;9. PubMed WOS:000454099000001. pmid:30622509
- 23. Dietrich F, Polymeris AA, Albert V, Engelter ST, Hersberger KE, Schaedelin S, et al. Intake reminders are effective in enhancing adherence to direct oral anticoagulants in stroke patients: a randomised cross-over trial (MAAESTRO study). J Neurol. 2023. Epub 2023/10/13. pmid:37831125.
- 24. Partlett C, Hall NJ, Leaf A, Juszczak E, Linsell L. Application of the matched nested case-control design to the secondary analysis of trial data. BMC Med Res Methodol. 2020;20(1):117. Epub 20200514. pmid:32410578; PubMed Central PMCID: PMC7227268.
- 25. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. Int J Surg. 2014;12(12):1495–9. Epub 20140718. pmid:25046131.
- 26. Cologne JB, Sharp GB, Neriishi K, Verkasalo PK, Land CE, Nakachi K. Improving the efficiency of nested case-control studies of interaction by selecting controls using counter matching on exposure. Int J Epidemiol. 2004;33(3):485–92. Epub 20040422. pmid:15105408.
- 27. Adams HP Jr, Davis PH, Leira EC, Chang KC, Bendixen BH, Clarke WR, et al. Baseline NIH Stroke Scale score strongly predicts outcome after stroke: A report of the Trial of Org 10172 in Acute Stroke Treatment (TOAST). Neurology. 1999;53(1):126–31. pmid:10408548.
- 28. Banks JL, Marotta CA. Outcomes validity and reliability of the modified Rankin scale: implications for stroke clinical trials: a literature review and synthesis. Stroke. 2007;38(3):1091–6. Epub 20070201. pmid:17272767.
- 29. Nasreddine ZS, Phillips NA, Bédirian V, Charbonneau S, Whitehead V, Collin I, et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc. 2005;53(4):695–9. pmid:15817019.
- 30. Dautzenberg G, Lijmer J, Beekman A. Diagnostic accuracy of the Montreal Cognitive Assessment (MoCA) for cognitive screening in old age psychiatry: Determining cutoff scores in clinical practice. Avoiding spectrum bias caused by healthy controls. Int J Geriatr Psychiatry. 2020;35(3):261–9. Epub 20191227. pmid:31650623; PubMed Central PMCID: PMC7028034.
- 31. Albert V, Polymeris AA, Dietrich F, Engelter ST, Hersberger KE, Schaedelin S, et al. Insights into direct oral anticoagulant therapy implementation of stroke survivors with atrial fbrillation in an ambulatory setting. J Stroke Cerebrovasc Dis. 2021;30(2):105530. Epub 2020/12/18. pmid:33333334.
- 32. McIlmoyle K, Tran H. Perioperative management of oral anticoagulation. BJA Educ. 2018;18(9):259–64. Epub 20180730. pmid:33456842; PubMed Central PMCID: PMC7808108.
- 33. Kolmanova E, Dvorackova S, Voriskova E, Sisakova M, Brabec M, Kala P, et al. Electronic and self-reported adherence to direct oral anticoagulants in patients with atrial fibrillation. Int J Clin Pharm. 2022;44(6):1547.
- 34. Márquez-Contreras E, Martell-Claros N, Márquez-Rivero S, Hermida-Campa E, Gracia-Diez C, Sanchez-López E, et al. Strategies for improving dabigatran adherence for stroke prevention in patients with non-valvular atrial fibrillation: education and drug intake reminders (FACILITA study). Curr Med Res Opin. 2018;34(7):1301–8. Epub 20180221. pmid:29384410.
- 35. Soto A, Cvetkovic-Vega A. Case-control studies. Rev Fac Med Hum. 2020;20(1):138–43.
- 36. Pandis N. Case-control studies: Part 1. Am J Orthod Dentofacial Orthop. 2014;146(2):266–7. pmid:25085311.
- 37. Kim BJ, Hwang YH, Park MS, Kim JT, Choi KH, Jung JM, et al. Atrial fibrillation related and unrelated stroke recurrence among ischemic stroke patients with atrial fibrillation. Front Neurol. 2021;12:744607. Epub 20211021. pmid:34744981; PubMed Central PMCID: PMC8566747.
- 38. Paciaroni M, Caso V, Agnelli G, Mosconi MG, Giustozzi M, Seiffge DJ, et al. Recurrent ischemic stroke and bleeding in patients with atrial fibrillation who suffered an acute stroke while on treatment with nonvitamin K antagonist oral anticoagulants: the RENO-EXTEND study. Stroke. 2022;53(8):2620–7. Epub 20220511. pmid:35543133.
- 39. Yaghi S, Henninger N, Scher E, Giles J, Liu A, Nagy M, et al. Early ischaemic and haemorrhagic complications after atrial fibrillation-related ischaemic stroke: analysis of the IAC study. J Neurol Neurosurg Psychiatry. 2020;91(7):750–5. Epub 20200513. pmid:32404380; PubMed Central PMCID: PMC8179007.
- 40. Steffel J, Collins R, Antz M, Cornu P, Desteghe L, Haeusler KG, et al. 2021 European Heart Rhythm Association practical guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation. Europace. 2021;23(10):1612–76. pmid:33895845.
- 41. Seiffge DJ, Paciaroni M, Wilson D, Koga M, Macha K, Cappellari M, et al. Direct oral anticoagulants versus vitamin K antagonists after recent ischemic stroke in patients with atrial fibrillation. Ann Neurol. 2019;85(6):823–34. Epub 20190430. pmid:30980560; PubMed Central PMCID: PMC6563449.
- 42. Hagerty T, Rich MW. Fall risk and anticoagulation for atrial fibrillation in the elderly: A delicate balance. Cleve Clin J Med. 2017;84(1):35–40. pmid:28084982.
- 43. Baumgartner PC, Haynes RB, Hersberger KE, Arnet I. A systematic review of medication adherence thresholds dependent of clinical outcomes. Front Pharmacol. 2018;9:1290. Epub 20181120. pmid:30524276; PubMed Central PMCID: PMC6256123.
- 44. Ozaki AF, Choi AS, Le QT, Ko DT, Han JK, Park SS, et al. Real-world adherence and persistence to direct oral anticoagulants in patients with atrial fibrillation: a systematic review and meta-analysis. Circ Cardiovasc Qual Outcomes. 2020;13(3):e005969. Epub 20200309. pmid:32148102.
- 45. Wetzels G, Nelemans P, Schouten J, van Wijk B, Prins M. All that glisters is not gold: a comparison of electronic monitoring versus filled prescriptions—an observational study. BMC Health Serv Res. 2006;6(1):8. pmid:16472388
- 46. Berkhout C, Berbra O, Favre J, Collins C, Calafiore M, Peremans L, et al. Defining and evaluating the Hawthorne effect in primary care, a systematic review and meta-analysis. Front Med (Lausanne). 2022;9:1033486. Epub 20221108. pmid:36425097; PubMed Central PMCID: PMC9679018.
- 47. Sedgwick P, Greenwood N. Understanding the Hawthorne effect. Br Med J. 2015;351:h4672. Epub 20150904. pmid:26341898.
- 48. McCambridge J, Kypri K, Elbourne D. In randomization we trust? There are overlooked problems in experimenting with people in behavioral intervention trials. J Clin Epidemiol. 2014;67(3):247–53. Epub 20131204. pmid:24314401; PubMed Central PMCID: PMC3969092.
- 49. Hartman L, Lems WF, Boers M. Outcome measures for adherence data from a medication event monitoring system: A literature review. J Clin Pharm Ther. 2019;44(1):1–5. Epub 2018/09/02. pmid:30171815.
- 50. Stawarz K, Cox A, Blandford A. Understanding the role of contextual cues in supporting the formation of medication-taking habits. Front Public Health (Conference Abstract: 2nd Behaviour Change Conference: Digital Health and Wellbeing). 2016.
- 51. Jochmann A, Artusio L, Usemann J, Jamalzadeh A, Frey U, Bush A, et al. A three months period of electronic monitoring is sufficient to assess adherence and improve asthma control. Eur Respir J. 2019;54(suppl 63):PA5429.
- 52. Smith D, Lovell J, Weller C, Kennedy B, Winbolt M, Young C, et al. A systematic review of medication non-adherence in persons with dementia or cognitive impairment. PLOS ONE. 2017;12(2):e0170651. pmid:28166234
- 53. Hawkins L, Kilian S, Firek A, Kashner TM, Firek CJ, Silvet H. Cognitive impairment and medication adherence in outpatients with heart failure. Heart Lung. 2012;41(6):572–82. pmid:22784869
- 54. Jokinen H, Melkas S, Ylikoski R, Pohjasvaara T, Kaste M, Erkinjuntti T, et al. Post-stroke cognitive impairment is common even after successful clinical recovery. Eur J Neurol. 2015;22(9):1288–94. Epub 20150604. pmid:26040251.
- 55. Dietrich F, Polymeris AA, Verbeek M, Engelter ST, Hersberger KE, Schaedelin S, et al. Impact of the COVID-19 lockdown on the adherence of stroke patients to direct oral anticoagulants: a secondary analysis from the MAAESTRO study. J Neurol. 2022;269(1):19–25. Epub 20210603. pmid:34081196; PubMed Central PMCID: PMC8173508.