Abstract
Background
The prevention of thromboembolism events remains challenging in cases of poor medication adherence. Unfortunately, clinical prediction of future adherence has been suboptimal.
Objective
To examine the correlation between two measures of real-time, self-reported adherence and anticoagulation control.
Methods
The IN-RANGE2 Cohort recruited patients initiating warfarin therapy in 3 urban anticoagulation clinics. At each study visit, participants reported adherence using a 100-point Visual Analogue Scale (VAS, marking % of pills taken since prior visit on a linear scale) and 7-day recall of pill taking behavior. anticoagulation control was measured by between visit percent time in INR range (BVTR), dichotomized at the cohort median. The longitudinal association between adherence and anticoagulation control was estimated using generalized estimating equations, controlling for clinical and demographic characteristics, prior BVTR, and warfarin dose changes.
Results
Among 598 participants with 3204 (median 4) visits, the median BVTR was 36.8% (Interquartile range 0–73.9%). Participants reported ≤80% adherence in 182 (5.7%) visits and missed pills in the past 7 days in 377 (11.8%) visits. Multivariable regression analysis found poorer anticoagulation control (BVTR<36.8%) in those with a VAS ≤80% (odds ratio [OR] 1.89, 95% confidence interval [CI] 1.12–3.18, p=0.02) and self-reported change in adherence since last visit (OR 1.55, 95% CI 1.20–2.01, p=0.001).
Conclusion
Self-reported VAS medication adherence at a clinic visit and changes in reported adherence since the last visit are independently associated with BVTR. Clinicians may gain additional insight into patients’ medication adherence by incorporating this information into patient management.
Keywords: Medication adherence, warfarin, visual analogue scale, self-report, anticoagulation
Introduction
Warfarin is the oldest and most commonly used oral anticoagulant (OAC) for the prevention of thromboembolic events. Despite extensive clinical experience with this medication, appropriate dosing remains challenging due to the influence of numerous patient factors, including medication non-adherence, on achieving and maintaining a therapeutic level.
Previous research using electronic pill monitoring has shown that up to 36% of patients on warfarin miss more than 20% of their doses, and every 10% increase in non-adherence is associated with a 10% increase in the odds of having a non-therapeutic International Normalized Ratio (INR).(1) INR is the strongest and most robust predictor of the risk of thromboembolic and hemorrhagic events.(2, 3) While the introduction of direct oral anticoagulants (DOAC) has offered patients new OAC therapeutic options that solve many of the challenges of warfarin use, adherence to DOACs has not been found to differ significantly from warfarin in practice.(4) Given the absence of clinically useful and accurate ways for clinicians to monitor adherence, medication non-adherence for patients on anticoagulation regimens remains a serious public health problem.
While physicians understand the importance of discussing adherence with patients, many report difficulty engaging in these conversations, with time representing the greatest barrier.(5) Most importantly, patient reports and physician assessments of patient adherence in routine clinical practice do not reflect objective measures of adherence using electronic pill monitoring.(6) Unfortunately, electronic pill monitoring is impractical for routine clinical practice. Formal prediction models to predict future adherence have suboptimal performance with poor validation.(7) Therefore, there is a need for accurate self-reported adherence measures that can be obtained quickly and efficiently (see conceptual framework, Figure 1).
Figure 1. Conceptual Framework.
Poor medication adherence is strongly associated with poor anticoagulation control which places patients at increased risk of hemorrhagic and thromboembolic events. A challenge in addressing poor medication adherence has been to readily identify patients with poor adherence in order to provide them with additional resources and targeted interventions. Previous studies have found prediction models of future adherence to be suboptimal, and objective measurement, through electronic pill monitoring to be accurate but impractical in routine clinical practice. Our study’s aim (dashed box) was to analyze the association between two quick and easily implemented self-reported adherence measurements, a visual analogue scale and change in 7-day pill recall, an anticoagulation control.
This study sought to characterize the association of two adherence measurements, the traditional 7-day recall and a Visual Analogue Scale (VAS), with anticoagulation control in patients starting warfarin therapy at three urban anticoagulation clinics. The VAS has been found to correlate with objective adherence measurements(8, 9, 10) and clinical outcomes in other patient populations,(8, 10, 11) but it has never been adequately tested in an anticoagulation population.
Methods
Study design
The INR Adherence and Genetics 2 (IN-RANGE2) cohort is an extension of the IN-RANGE cohort previously described.(1) Briefly, this is a large, multicenter, prospective cohort of patients initiating warfarin therapy between 2009 and 2013 at three urban anticoagulation Clinics: the Hospital of the University of Pennsylvania, the Corporal Michael J. Crescenz Veterans Affairs Medical Center, and the Johns Hopkins Medical Institutions. Self-identified Caucasians and African Americans were included in the study, with minimal exclusion criteria including age<21 years, inability to give consent, or abnormal INR before initiating therapy. For this analysis, additional exclusion criteria included having <2 in-person visits or <2 adherence measurements.
Data collection
Information on participant demographic and clinical characteristics as well as factors that can influence warfarin response, warfarin dose, and medication adherence were collected prospectively through in-person interviews by trained research nurses at baseline and at subsequent visits using standardized questionnaires and data collection forms. Data collection occurred before the current INR was revealed to participants or interviewers. Clinicians were blinded to interview responses.
Adherence and outcome measures
Medication adherence was assessed at each study visit using two measurements: VAS and 7-day recall. The VAS tool presented participants with a continuous line anchored by 0% and 100% with 10% intervals and asked them to mark the line at their best guess about their adherence since their previous visit (see eFigure 1). The 7-day recall tool asked participants whether they had skipped or taken any extra pills in the past 7 days and, if so, how many pills. INR was measured at each visit according to the clinics’ standard procedures, using either point of care fingerstick or venous samples from phlebotomy. The primary outcome was between visit time in therapeutic INR range (BVTR). Thus, for each visit with a patient-reported adherence measure, the BVTR was calculated using all INRs collected between the prior study visit and the current study visit, based on the Rosendaal linear interpolation method.(12) Patients could have multiple BVTRs throughout the study period, each corresponding to the interval in which they reported an adherence measurement.
Data analysis
Descriptive statistics were calculated for the cohort using mean, median, standard deviation, and interquartile range. The VAS score was analyzed a priori as dichotomized into greater than 80% vs ≤80%, a commonly used threshold for poor adherence.(1, 13, 14) Secondary analyses dichotomized the VAS score as 100% vs. <100%. Patient-reported number of pills taken correctly was converted into a continuous variable of percentage pill adherence. The correlation between the two adherence measurements was assessed using Pearson correlation coefficient and kappa statistic. BVTR was dichotomized at the median value to maximize power.
Three measurements of adherence were analyzed for each tool: adherence at current visit, adherence at prior visit, and change in adherence between visits. Univariable analysis was used to calculate the association of each adherence measure and each covariate with BVTR utilizing generalized estimating equations, based on an independent correlation matrix to account for longitudinal observations within participant. Adherence measurements which were not found to have a statistically significant association with BVTR were not included in subsequent adjusted models.
Three types of potential confounders were assessed in our adjusted analysis: demographics, clinical factors, and baseline medication taking practices (see eTable 1 for specific questions). These covariates were considered to be confounders and included in multivariate models if they were associated with BVTR with a univariate p-value <0.20. The final models were adjusted for these confounders, BVTR during prior interval, and having a warfarin dose change since the prior visit.
All statistics were performed with SAS (version 9.1, SAS Institute, Cary, NC, USA) and STATA (version 13.1, StataCorp, College Station, TX, USA). The Institutional Review Boards at all participating hospitals approved the study and all participants provided informed, signed consent.
Results
The IN-RANGE2 cohort comprised 687 participants, with 89 participants excluded from this analysis for having <2 in-person visits or adherence measurements, leaving a final population of 598 (87%) participants and 3,204 total in-person visits, with a median follow up of 4 visits (Interquartile Range [IQR] 2–7). Of these, 447 (75%) reached maintenance dose, 78 (13%) stopped warfarin before reaching the primary end point, 57 (9.6%) were lost to follow up, 15 (2.5%) did not reach the primary end point before the end of the study, and 1 (0.2%) withdrew consent. Baseline characteristics of our study population are shown in Table 1. Participants excluded from this analysis were more likely to be Caucasian, current smokers, to have been hospitalized in the past 12 months, and to score higher on the Cognitive Capacity Screening Exam (data not shown).
Table 1.
Participant Characteristics*
Total | |
---|---|
Participants | 598 |
Demographics | |
Age, y, mean (SD) | 55.5 (14.9) |
Gender, No. (%) | |
Male | 368 (61.5%) |
Female | 228 (38.1%) |
Race, No. (%) | |
African-American | 429 (71.7%) |
Caucasian | 157 (26.3%) |
Education, No. (%) | |
High School or less | 262 (43.8%) |
More than High School | 335 (56.0%) |
Marital Status, No. (%) | |
Married | 174 (29.1%) |
Separated | 159 (26.6%) |
Widowed | 59 (9.9%) |
Not Married | 196 (32.8%) |
Insurance Status, No. (%) | |
Medicare | 209 (34.9%) |
Medicaid | 58 (9.7%) |
Private | 186 (31.1%) |
VA | 80 (13.4%) |
Other | 33 (5.4%) |
None | 28 (4.7%) |
Site, No. (%) | |
Hospital of the University of Pennsylvania | 241 (40.3%) |
Corporal Michael J. Crescenz Veternas Affairs Medical Center | 173 (28.9%) |
Johns Hopkins Medical Institutions | 184 (30.8%) |
Baseline Clinical Characteristics | |
Indication, No. (%) | |
Atrial Fibrillation/Flutter | 189 (31.6%) |
Venous Thromboembolism | 311 (52.0%) |
Other | 95 (15.9%) |
History Prior warfarin use, No. (%) | 184 (30.8%) |
Doctor Visits in past 12 months, No. (%) | |
0–3 Visits | 110 (18.4%) |
4–12 Visits | 259 (43.3%) |
13+ Visits | 225 (37.6%) |
Alcohol Use with Warfarin, No. (%) | |
Yes | 211 (35.3%) |
No | 380 (63.5%) |
Smoking Status, No. (%) | |
Ever Smoker | 339 (56.7%) |
Never Smoker | 256 (42.8%) |
Poor Kidney Function, No. (%) | |
GFR<30 | 36 (6.0%) |
30<GFR<60 | 107 (17.9%) |
GFR>60 | 411 (68.7%) |
CHADS 2 Score, No. (%) | |
0 | 133 (22.2%) |
1 | 172 (28.8%) |
2+ | 282 (47.2%) |
Standardized General Health Perception, mean (SD) | 54.8 (23.3) |
Statins at Baseline, No. (%) | |
Yes | 263 (44.0%) |
No | 334 (55.9%) |
Amiodarone at Baseline, No. (%) | |
Yes | 35 (5.9%) |
No | 563 (94.1%) |
Some percentages are based on fewer than 598 participants due to missing data
Abbreviations: No. = Number; SD = Standard deviation,
The mean (median) number of INR measurements per BVTR measurement was 3.0 (2.0), with a median BVTR of 36.8% (IQR 0–73.9%). The median number of visits to reach maintenance dose was 5 (IQR 2–7).
Participants had a median of 4 (IQR 3–7) VAS and 7-day recall measurements. Mean adherence by VAS was 96.6% (Standard Deviation [SD] 5.8%), with participants reporting less than 100% adherence in 729 (28%) visits and less than or equal to 80% adherence in 182 (5.7%) visits. Mean adherence by 7-day recall was 97.4% (SD 10.4%), with participants reporting incorrect pill taking in 408 (13%) visits. The two adherence measures were moderately correlated with a Pearson’s correlation coefficient of 0.62 (p<0.001). When dichotomized, they had fair agreement with a kappa statistic of 0.40 (p<0.001).
Adherence and anticoagulation control
Univariable analysis demonstrated a VAS score of ≤80% at the current visit to be associated with 2.36 (95% Confidence Interval [CI] 1.71–3.25) times greater odds of poor anticoagulation control (Table 2). This association remained significant when modeling the VAS score continuously, but not when dichotomized at 100%. Incorrect pill taking by 7-day recall was associated with a 63% (95% CI 1.32–2.02) increase in the odds of poor anticoagulation control, with the association remaining when modeling percentage of pill adherence continuously. Adherence measurements at the prior visit, change in reported adherence, BVTR at the prior visit, and dosage changes since the last visit were found to have a significant association with anticoagulation control at the current visit (Table 2).
Table 2.
Univariable Analysis
Poor AC Control OR (95% CI) | p-value | |
---|---|---|
VAS | ||
>80% | Ref | |
≤80% | 2.36 (1.71, 3.25) | <0.001 |
100% | Ref | |
<100% | 1.15 (0.98, 1.34) | 0.08 |
Prior VAS | ||
>80% | Ref | |
≤80% | 2.20 (1.59, 3.05) | <0.001 |
100% | Ref | |
<100% | 1.15 (0.97, 1.35) | 0.1 |
Change in VAS | ||
No | Ref | |
Yes | 1.63 (1.42, 1.88) | <0.001 |
7-day recall | ||
No Incorrect pill | Ref | |
Incorrect pill | 1.63 (1.32, 2.02) | <0.001 |
Prior 7-day recall | ||
No Incorrect pill | Ref | |
Incorrect pill | 2.09 (1.66, 2.63) | <0.001 |
Change in 7-day recall | ||
No | Ref | |
Yes | 1.38 (1.14, 1.66) | <0.001 |
Prior BVTR | ||
10% decrease | 1.23 (1.20, 1.26) | <0.001 |
Dosage Change | ||
No | Ref | |
Yes | 3.52 (3.04, 4.07) | <0.001 |
Abbreviations: AC = anticoagulation, CI = confidence interval; OR = odds ratio; VAS = visual analogue scale
Covariates and anticoagulation control
Demographic covariates including age, sex, race, education, marital status, insurance status, and anticoagulation clinic site were found to be associated with BVTR with a p-value <0.20, and were included in subsequent models. Clinical factors including anticoagulation indication, varying dosing regimen, general health self-assessment, number of doctor visits in the past 12 months, smoking status, alcohol use, poor kidney function, use of statins at baseline, use of amiodarone at baseline, health care encounter since prior visit, and having warfarin stopped since last visit were also associated with BVTR with a p-value <0.20. Of 17 medication taking practice questions in the survey, 9 of them were associated with BVTR with a p-value <0.20 (see eTable 1).
Multivariable models
VAS scores at both current and prior visits were independently associated with anticoagulation control, when including only VAS adherence measurements, after adjustment for the above confounders, prior BVTR, and dose changes since the prior visit (Table 3). Models including only 7-day recall measurements found change in adherence since last visit to have a significant association with anticoagulation control, while the current 7-day recall value was not significant. Models using both VAS and 7-day recall measurements found an independent association between poor anticoagulation control and both a VAS score ≤80% at the current visit and a reported change in adherence since last visit using 7-day recall. Importantly, no other self-reported measurements of adherence were found to be independently associated with anticoagulation control. The association remained when using VAS score as a continuous measurement with a 10% decrease in adherence being associated with a 14% increase in the odds of having poor anticoagulation control (OR 1.14; 95% CI 1.00–1.29; p=0.04).
Table 3.
Multivariable Analysis
Model | VAS only model* | 7-day recall only model* | Final Combined Model* | |||
---|---|---|---|---|---|---|
OR (95% CI) | p-value | OR (95% CI) | p-value | OR (95% CI) | p-value | |
Current VAS Score (≤80%) | 1.88 (1.12, 3.18) | 0.02 | Not Included | 1.90 (1.13, 3.20) | 0.02 | |
Prior VAS Score (≤80%) | 1.94 (1.27 ,2.97) | 0.002 | Not Included | † | ||
Change in 7-day recall (Yes) | Not Included | 1.61 (1.24 , 2.08) | <0.001 | 1.56 (1.20 , 2.01) | 0.001 | |
Prior BVTR (10% decrease) | 1.18 (1.15, 1.22) | <0.001 | 1.19 (1.15, 1.22) | <0.001 | 1.19 (1.15, 1.22) | <0.001 |
Dosage Change since Prior Visit (Yes) | 4.11 (3.36, 5.04) | <0.001 | 4.09 (3.34, 5.02) | <0.001 | 4.10 (3.35, 5.03) | <0.001 |
Covariates included in models: gender, age, race, education, insurance status, marital status, AC clinic site, AC indication, varying dosing regimen, general health perception, number of doctor visits in past 12 months, smoking status, alcohol use, poor kidney function, use of statins at baseline, use of amiodarone at baseline, health encounter since prior visit, warfarin stopped since last visit, medication taking practice (questions 2, 3, 4, 7, 12, 13, 16, 17, 18).
Prior VAS score was not significant in combined models and was removed from final combined model.
Abbreviations: AC = anticoagulation, CI = confidence interval; OR = odds ratio; BVTR = between visit percent time in therapeutic INR range; VAS = visual analogue scale
Discussion
In this prospective study, we found that patient self-reported adherence using two quick and simple tools for the assessment of adherence at each patient visit – a visual analogue scale and changes in adherence using 7-day pill recall – were independently associated with anticoagulation control. Our results are consistent with previous studies in HIV patients (8, 10) and women on aromatase inhibitors.(11) The only previous study comparing VAS to anticoagulation control in warfarin treated patients found no association; however, it was a small cross-sectional study using a convenience sample.(15)
We found only a moderate correlation between 7-day recall and VAS, despite the fact that they were administered in succession. We found that participants reported imperfect adherence more often with the VAS than with the 7-day recall. While this might reflect that the tools ask about adherence over different periods of time, it is possible that the VAS is more accurate and less prone to desirability bias as patients can report imperfect adherence across a range of adherence on a scale without having to report their incorrect pill taking directly to a clinician.
Association between self-reported adherence and anticoagulation control
In our final combined model, the only measurements that had a significant association with anticoagulation control were adherence at the current visit reported using VAS and a change in adherence from the previous visit reported with 7-day recall. Interestingly, our models did not show a change in VAS score from the previous visit, or 7-day recall at the current visit to be significantly associated with anticoagulation control. As noted above, it is likely that 7-day recall measurements are less accurate and therefore correlated poorly with anticoagulation control. This suggests that more comprehensive adherence information may be uncovered by combining these tools and considering not only adherence reported at a current visit, but also assessing changes in reported behavior over time. However, given the logistical challenges of implementing two distinct adherence measurements in clinical practice, given that the VAS is easier to measure and has a stronger association with BVTR, clinicians should choose the VAS over 7-day recall if they are able to measure only one parameter
These findings are particularly relevant after the introduction of DOACs, as anticoagulation management is moving away from specialty anticoagulation clinics that are designed to spend time discussing adherence with patients to general practice clinics. At the same time, poor adherence to DOACs cannot be inferred as it can with warfarin because of the absence of a laboratory test (e.g., INR) for monitoring DOAC response. Further, the short half-lives of DOACs (compared with the very long half-life of warfarin) place patients at increased risk of a thromboembolic event after missing just 1 or 2 doses.(16) This study identifies the VAS as a promising tool that might help identify poor adherence to DOACs, given that its association with anticoagulation control was found to be independent of both changes in warfarin dosing and knowledge of a patient’s current INR. The simplicity of this tool would allow it to be administered during patient visits, flagging patients who report ≤80% adherence as being at risk of poor anticoagulation control. In these patients, clinicians should consider using visit time to discuss barriers to adherence, as well as consideration of treatment changes.
Study limitations
Our study had several limitations. First, we had no objective adherence measurement to validate patient self-reported measurements. While studies have found the VAS to correlate with objective measurements of medication adherence—including electronic monitoring,(9) pill counts,(8) and claims-based data(17)— it is possible that participants in our study were more likely to overestimate their adherence due to recall and social desirability bias. This might be reflected in the high VAS score (which is slightly higher, yet consistent, with those reported in the literature).(8, 9, 11) This would bias our results towards the null. Second, VAS was measured before 7-day recall, and it is possible that the order affected participants’ recall; however, the measurements could be easily implemented in this order in a clinical setting to replicate our findings. Third, we defined anticoagulation control using BVTR, a short-term measure specifically designed to detect effects between visit questionnaires; this measure should not be compared to the traditionally reported time in therapeutic INR range (TTR) which is a single measure over a patient’s entire course of therapy. Fourth, while our study identified the VAS as a promising tool for use in patients on DOACs, additional research is necessary to ensure that our findings apply in populations on anticoagulants other than warfarin. Fifth, the study was underpowered to detect clinical events such as thromboembolisms and bleeding; however, anticoagulation control is a well-established predictor of these outcomes. (2, 3) Fifth, warfarin metabolism and, subsequently, BVTR are known to be affected by a wide variety of factors besides medication adherence; although we adjusted for many of these factors, there may be unmeasured confounding from other important factors, such as variability in dietary vitamin K. Finally, this cohort only included participants initiating anticoagulation therapy, which may limit generalizability to patients already receiving stable warfarin dosing.
Conclusion
Medication non-adherence is an important public health issue, especially for patients on OACs. Clinicians are currently ill-equipped to address these challenges, with few simple tools to accurately identify patients with poor adherence. The VAS is a promising tool to help clinicians assess patient adherence, that is quick, inexpensive, and easily implemented. Future studies are needed to validate our findings and determine if self-reported medication adherence can predict outcomes for patients on anticoagulation and improve their safety.
Supplementary Material
Clinical Significance.
Clinicians may gain additional insight into patients’ medication adherence by incorporating information from the visual analogue scale and changes in 7-day recall into clinical decision making.
The visual analogue scale represents a promising tool to monitor adherence to other medications, and might be useful in patients on DOACs because the association is independent of knowing prior anticoagulation control.
Acknowledgments
Funding Sources: I confirm that I have mentioned all organizations that funded my research in the Acknowledgements section of my submission, including grant numbers where appropriate.
This work was supported by the National Institutes of Health (Grant #: R01HL066176; PI: Stephen E Kimmel) and by the American Heart Association (2016 Student scholarship in Cardiovascular Disease; PI: Stephen E Kimmel).
Footnotes
All authors had access to the data and a role in writing the manuscript.
Conflict of interest disclosures:
J. Sevilla Cazes: None.
B.S. Finkleman: None.
J. Chen: None.
C.M. Brensinger: None.
A. E. Epstein: None.
M. B. Streiff: Consultant; Bio2Medical, Janssen Healthcare, CSL Behring. Research support; Boehringer Ingelheim, Janssen Healthcare, Portola, Roche
S.E. Kimmel: Expert Witness; Bayer, Pfizer.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Kimmel SE, Chen Z, Price M, et al. The influence of patient adherence on anticoagulation control with warfarin: results from the International Normalized Ratio Adherence and Genetics (IN-RANGE) Study. Arch Intern Med. 2007;167:229–235. doi: 10.1001/archinte.167.3.229. [DOI] [PubMed] [Google Scholar]
- 2.Fihn SD, Callahan CM, Martin DC, McDonell MB, Henikoff JG, White RH. The risk for and severity of bleeding complications in elderly patients treated with warfarin. Ann Intern Med. 1996;124:970–979. doi: 10.7326/0003-4819-124-11-199606010-00004. [DOI] [PubMed] [Google Scholar]
- 3.Morgan CL, McEwan P, Tukiendorf A, Robinson PA, Clemens A, Plumb JM. Warfarin treatment in patients with atrial fibrillation: Observing outcomes associated with varying levels of INR control. Thromb Res. 2009;124:37–41. doi: 10.1016/j.thromres.2008.09.016. [DOI] [PubMed] [Google Scholar]
- 4.Luger S, Hohmann C, Niemann D, et al. Adherence to oral anticoagulant therapy in secondary stroke prevention – Impact of the novel oral anticoagulants. Patient Preference Adherence. 2015;9:695–1705. doi: 10.2147/PPA.S88994. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hines R, Stone NJ. Patients and Physicians Beliefs and Practices Regarding Adherence to Cardiovascular Medication. JAMA Cardiol. 2016;1:470–473. doi: 10.1001/jamacardio.2016.0634. [DOI] [PubMed] [Google Scholar]
- 6.Parker CS, Chen Z, Price M, et al. Adherence to warfarin assessed by electronic pill caps, clinician assessment, and patient reports: Results from the IN-RANGE study. J Gen Intern Med. 2007;22:1254–1259. doi: 10.1007/s11606-007-0233-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Platt AB, Localio AR, Brensinger CM, et al. Can we predict daily adherence to warfarin?: Results from the International Normalized Ratio Adherence and Genetics (IN-RANGE) Study. Chest. 2010;137:883–889. doi: 10.1378/chest.09-0039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Giordano TP, Guzman D, Clark R, Charlebois ED, Bangsberg DR. Measuring adherence to antiretroviral therapy in a diverse population using a visual analogue scale. HIV Clin Trials. 2004;5:74–79. doi: 10.1310/JFXH-G3X2-EYM6-D6UG. [DOI] [PubMed] [Google Scholar]
- 9.Gallagher BD, Muntner P, Moise N, Lin JJ, Kronish IM. Are two commonly used self-report questionnaires useful for identifying antihypertensive medication nonadherence? J Hypertens. 2015;33:1108–1113. doi: 10.1097/HJH.0000000000000503. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Walsh JC, Mandalia S, Gazzard BG. Responses to a 1 month self-report on adherence to antiretroviral therapy are consistent with electronic data and virological treatment outcome. AIDS. 2002;16:269–277. doi: 10.1097/00002030-200201250-00017. [DOI] [PubMed] [Google Scholar]
- 11.Brier MJ, Chambless D, Gross R, Su HI, DeMichele A, Mao JJ. Association between self-report adherence measures and oestrogen suppression among breast cancer survivors on aromatase inhibitors. Eur J Cancer. 2015;51:1890–1896. doi: 10.1016/j.ejca.2015.06.113. [DOI] [PubMed] [Google Scholar]
- 12.Rosendaal FR, Cannegieter SC, Van der Meer FJM, Briet E. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost. 1993;69:236–239. [PubMed] [Google Scholar]
- 13.Karve S, Cleves MA, Helm M, Hudson TJ, West DS, Martin BC. Good and poor adherence: Optimal cut-point for adherence measures using administrative claims data. Curr Med Res Opin. 2009;25:2303–2310. doi: 10.1185/03007990903126833. [DOI] [PubMed] [Google Scholar]
- 14.Casciano JP, Dotiwala ZJ, Martin BC, Kwong WJ. The costs of warfarin underuse and nonadherence in patients with atrial fibrillation: A commercial insurer perspective. J Managed Care Pharm. 2013;19:302–316. doi: 10.18553/jmcp.2013.19.4.302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wang Y, Kong MC, Ko Y. Comparison of three medication adherence measures in patients taking warfarin. J Thromb Thrombolysis. 2013;36:416–421. doi: 10.1007/s11239-013-0872-5. [DOI] [PubMed] [Google Scholar]
- 16.Barnes GD, Nallamothu BK, Sales AE, Froehlich JB. Reimagining Anticoagulation Clinics in the Era of Direct Oral Anticoagulants. Circ Cardiovasc Qual Outcomes. 2016;9:182–185. doi: 10.1161/CIRCOUTCOMES.115.002366. OUTCOMES.115.002366. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nau DP, Steinke DT, Williams LK, et al. Adherence analysis using visual analog scale versus claims-based estimation. Ann Pharmacother. 2007;41:1792–1797. doi: 10.1345/aph.1K264. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.