Self-testing of anticoagulation improves outcomes, but is expensive. One might assume it is more helpful for patients living farther from care, but the authors disprove this assumption.
Objectives: Patient self-testing (PST) improves anticoagulation control and patient satisfaction. It is unknown whether these effects are more pronounced when the patient lives farther from the anticoagulation clinic (ACC). If the benefits of PST are limited to a subset of patients (those living farther from care), selectively providing PST to that subset could enhance cost-effectiveness.
Study Design: This is a secondary analysis of a randomized trial of PST versus usual ACC care, which involved 2922 patients of the Veterans Health Administration (VHA).
Methods: Our 3 outcomes were the primary composite clinical end point (stroke, major hemorrhage, or death), anticoagulation control (percent time in therapeutic range), and satisfaction with anticoagulation care. We measured the driving distance between the patient’s residence and the nearest VHA facility. We divided patients into quartiles by distance and looked for evidence of an interaction between distance and the effect of the intervention on the 3 outcomes.
Results: The median driving distance was 12 miles (interquartile range = 6-21). Patients living in the farthest quartile had higher rates of the primary composite clinical end point in both groups compared with patients living in the nearest quartile. For PST, the hazard ratio (HR) was 1.77 (95% CI, 1.18-2.64), and for usual care, the HR was 1.81 (95% CI, 1.19-2.75). Interaction terms did not suggest that distance to care modified the effect of the intervention on any outcome.
Conclusions: The benefits of PST were not enhanced among patients living farther from care. Restricting PST to patients living more than a certain distance from the ACC is not likely to improve its cost-effectiveness.
Am J Manag Care. 2016;22(1):65-71
Self-testing of anticoagulation improves outcomes, but is expensive. Because its main impact is to enable frequent testing, it could have greater benefit for patients living farther from care.
Oral anticoagulants are received by millions of patients each year to treat or prevent thromboembolic disease.1 Despite the introduction of novel anticoagulants, warfarin is likely to remain in widespread use for years to come, in part due to concerns about the cost-effectiveness of the novel agents2 and their safety and efficacy in real-world settings. The effective use of warfarin, however, presents several important challenges. First, excellent anticoagulation control can improve patient outcomes,3-5 but it can be difficult to achieve6,7; therefore, there is a great need to find effective strategies to improve anticoagulation control.8-10 Second, the burden and cost of frequent clinic visits for monitoring the International Normalized Ratio (INR) can fall heavily on patients and their caregivers.11-13
Patient self-testing (PST), the use of a point-of-care device to monitor INR at home, has the potential to address both challenges. Several meta-analyses have suggested that PST generally reduces rates of adverse events (defined here as stroke, major hemorrhage, and all-cause mortality) and improves percent time in therapeutic range (TTR), although the effects were relatively small.14-16 In one large study of PST, The Home INR Study (THINRS), PST was associated with small but significant improvements in TTR and satisfaction with anticoagulation care; the difference in adverse event rates seemed to favor PST, but did not reach statistical significance.17 It is generally assumed that the causal pathway for these effects involves test frequency and the ease of testing.18 Patients may find it difficult to test INR frequently under usual care and, in fact, may test less frequently than would be optimal. Since PST makes it easier to test more frequently, patients are less likely to resist requests to test more frequently when their INR has been unstable. Indeed, the general practice with PST is to test weekly, regardless of the stability of INR, because the burden associated with frequent testing is minimal. Because it reduces the burden of testing, PST should improve satisfaction with care and contribute to improved anticoagulation control,18 which in turn would prevent adverse events.3-5
This implies that the benefits of PST should be magnified among patients who have the greatest difficulty visiting the anticoagulation clinic (ACC). These patients would be the most likely to delay needed INR testing, to the detriment of their anticoagulation control and outcomes.19 Also, these patients would tend to have low baseline levels of satisfaction, due to the burden of frequent visits to the ACC. This raises the possibility that limiting PST to patients with the greatest difficulty accessing care might realize a disproportionate share of its benefits at a fraction of the cost of providing it for all patients. This conjectured causal pathway has not been empirically examined.
In this study, we used data from THINRS to examine whether PST would be more effective among patients living farther from the nearest Veterans Health Administration (VHA) facility. We expected to find, among the usual ACC group, that patients living farther from care would have less frequent INR testing, lower satisfaction with care, lower TTR, and higher rates of adverse events. Among the PST group, however, we expected all these parameters would be unaffected by distance to care. We consequently expected to find that PST would have a greater impact on these outcomes among patients living farther from care.
The Home INR Study
THINRS was a randomized trial of PST versus high-quality usual ACC care, funded by the Veterans Affairs (VA) Cooperative Studies Program (CSP 481). The methods and main results of THINRS have been discussed elsewhere.17,18 Briefly, THINRS recruited VHA patients with atrial fibrillation and/or mechanical heart valves who required chronic warfarin therapy. Those deemed competent to perform PST were randomized in a 1:1 ratio to usual ACC care (with testing once every 4 weeks) or PST (with most patients testing once a week). Follow-up visits were scheduled approximately every 3 months after randomization to collect information about medical events and other data, and to check whether PST patients were still competent to use the meter. The primary clinical end point was time to first major event (stroke, major hemorrhage, or death). The study was approved by the institutional review boards of all VHA medical centers where patients were enrolled or research was conducted.
Driving Distance to Nearest VHA Facility
We used the VHA’s centralized file that contains the driving distance for each patient to the nearest VHA facility. These driving distances are calculated using ESRI StreetMap Premium for ArcGIS (ESRI Corporation, Redlands, California) and we used these distances for THINRS patients as a proxy measure for their likely burden of transportation to each ACC visit. We linked information for each of the 2922 patients randomized in the THINRS study to these records and found 2903 with distance information (of the 19 nonmatches, most were due to a missing address). We classified these 2903 patients into quartiles based on driving distance for addresses at the time of randomization. We excluded data for 89 patients who moved to a different distance quartile during the 2-year follow-up period, leaving 2814 patients. Of these, 2755 (1360 in the usual ACC group and 1395 in the PST group) had INR values during the 2-year follow-up period and comprised the analytic population for this study.
Patient-level outcomes included the composite primary clinical end point, anticoagulation control, and satisfaction with care. All outcomes were measured during the 2-year period following randomization. The composite primary clinical end point included stroke, major hemorrhage, or death. These outcomes were confirmed by chart review and adjudicated by an independent committee blinded to treatment assignment.17
Anticoagulation control was measured using percent TTR, computed according to the method of Rosendaal.20 TTR summarizes anticoagulation control over time by using linear interpolation to assign an INR value to each day between successive observed INR values. After interpolation, the percentage of time during which the interpolated INR values lie within the patient’s target INR range (from 0%-100%) is calculated.20
Satisfaction with anticoagulation care was measured using the Duke Anticoagulation Satisfaction Scale,21 a validated instrument for assessing health-related quality of life (HRQoL) specifically related to long-term oral anticoagulation. For this study, we dichotomized patients into those who were “highly satisfied” with anticoagulation care (the highest tertile of satisfaction) versus all others.
The frequency of INR testing was also examined as a possible link in the causal pathway between PST and improved TTR. For each patient, we calculated the number of INR tests per patient-year. We hypothesized that test frequency would be lower with increased distance to care in the usual care group, but unaffected by distance in the PST group.
We compared baseline characteristics between PST and usual ACC patients, including distance to care. We compared PST and usual ACC patients regarding our main outcomes of interest (primary composite clinical end point, TTR, and satisfaction with care) overall, and by quartiles of distance to care. We performed tests of increasing or decreasing trends by distance within each treatment group, using the Cochran-Armitage test for categorical outcomes22,23 and the Jonckheere-Terpstra test for continuous outcomes.24,25 For comparisons of study outcomes, all 2755 patients were involved in the analyses; however, only 1977 provided data for satisfaction with anticoagulation care at 2 years of follow-up.
Finally, we looked for evidence of a statistical interaction between group assignment and distance quartile for each outcome. For these formal tests of interaction, we used linear or logistic regression, as appropriate, and structured the distance quartiles as a class variable, rather than forcing its levels into a single linear function. Analyses were performed using SAS version 9.1 (SAS Institute, Cary, North Carolina).
Among the 2755 patients included in our study, 1395 received PST and 1360 received usual ACC care. Characteristics were generally balanced between groups (). Enrollees were overwhelmingly male (98%) and white (92%), with an average age of 67 years. The mean distance to care was 16 miles (SD = 17); the median was 12 miles (interquartile range = 6-21). Ten percent of patients lived more than 38 miles from the nearest VHA facility. As with other patient characteristics, distance to care was balanced between groups.
Satisfaction with Anticoagulation Care
eAppendix Table 1
For the entire study, the PST group had a nonsignificantly higher proportion of patients reporting that they were “highly satisfied” with their anticoagulation care (32% vs 28%; OR, 1.21; 95% CI, 0.99-1.47; P = .06). We did not find evidence of an increasing or decreasing trend for satisfaction in the usual ACC group based on distance to care (see [eAppendices available at www.ajmc.com]), although the interaction term between distance quartile and treatment group was close to being statistically significant (P = .06). If anything, distance seemed to impact satisfaction more in the PST group (test for trend, P = .028) than in the usual ACC group (P = .41), which was not what we had hypothesized.
Using the sample for the present study, the PST intervention was associated with a small, but statistically significant overall improvement in TTR (65.7% vs 63.0%; 95% CI for difference, 1.5%-3.8%; P <.001).We did not find evidence of an increasing or decreasing trend in TTR by distance to care within either group (). The interaction term between distance to care and group assignment was not statistically significant for the outcome of TTR (P = .26).
INR Test Frequency
By design, the PST intervention was associated with an almost 3-fold increase in INR test frequency (Table 2) (47.2 vs 16.5 tests/year; P <.001). We did not find evidence of a trend for test frequency in the usual ACC group by distance to care, and the interaction term between distance and group assignment was not significant (P = .42). Thus, we did not have any evidence that patients who live farther from care were postponing needed INR tests under usual ACC care.
Primary Composite Clinical End Point
eAppendix Table 2
is the Kaplan-Meier curve comparing time to first event within the first 2 years of follow-up for the usual ACC group by distance quartile, and Figure 2 is a similar curve for the PST group. The hazard ratio, its 95% CI, and the P value for the treatment comparison within each distance quartile are in . Tabular presentation of the data underlying Figures 1 and can be found in .
Within each treatment group, the comparison across distance quartiles was statistically significant (indicating that the hazard for the composite outcome was not the same across all 4 distance quartiles; see Figures 1 and 2). Specifically, there was an increased hazard for the composite outcome in the group living farthest from care compared with the group living closest. For the comparison between usual ACC and PST groups within each distance quartile, the only statistically significant difference found was for the second closest distance group (Table 3). The interaction term between distance quartile and treatment group using the Cox proportional hazards regression model did not reach statistical significance (P = .07).
We also divided distance to care into deciles rather than quartiles to ensure that patients living great distances from care (≥38 miles in the farthest decile) did not show effects that were obscured within larger groups. Findings were similar to those seen with distance quartiles and are not shown.
Patient self-testing has some proven benefits for patients, particularly in terms of improved anticoagulation control and improved satisfaction with care.17 However, PST is costly and it would be logical to think that selectively providing it to patients living farther from the ACC could target a smaller group who are more likely to benefit. In this study, we examined this logical, but unproven, supposition. We had expected to find that among the usual ACC group—who did not receive PST—patients living farther from care would have less frequent INR testing because patients would tend to resist suggestions to follow up sooner. Consequently, we also expected to find that patients living farther from the ACC would have worse TTR as a result of testing their INR less frequently than recommended.19 However, we found no empirical evidence to support any step in our hypothesized causal pathway; namely, that patients living father from care would have decreased frequency of INR testing, leading to poor anticoagulation control19 and thus to an increased incidence of adverse events.5
Based on our findings with regard to test frequency, TTR, and the primary combined clinical end point, it would not be possible to argue that limiting PST to patients living farther from the ACC would enhance its cost-effectiveness. Somewhat surprisingly, we also did not find that patients living farther from the ACC had lower anticoagulation-specific HRQoL under usual ACC care, nor did we find that PST had a particular benefit for HRQoL among such patients. Therefore, we also cannot argue for selective use of PST with patients living farther from care based on considerations of patient satisfaction or HRQoL.
Perhaps our most striking finding was that patients living farther from the nearest VHA site of care had higher rates of the combined primary clinical end point of stroke, major hemorrhage, or death—a finding observed in both the PST and the usual ACC groups. Clearly, distance from care is important, but was not operating as we had anticipated. Little has been written about the impact of distance to care for patients receiving warfarin; the one study of which we are aware showed that patients living farthest from care had a small decrement in TTR (approximately 1%), but only during the first 6 months of warfarin therapy.26 A difference of this magnitude is unlikely to explain a meaningfully higher rate of adverse events.
The findings of the present study appear to be novel and deserving of further investigation. Although this finding remains to be replicated in a second study, a dose-response gradient—as was seen here with distance to care—argues fairly strongly that the finding is real and not merely an artifact. The mechanism of this finding is not immediately clear, although one possibility is that patients living farther from the nearest hospital may have some degree of hesitancy in seeking emergency or inpatient care,27 possibly delaying the onset of treatment when serious adverse events occur.
Strengths and Limitations
Our study has considerable strengths. In particular, nesting this analysis within the setting of a well-conducted randomized trial ensures a balance of both measured and unmeasured confounders. In addition, the ascertainment of adverse events was extremely rigorous. However, our study also has some noteworthy limitations, such as our use of driving distance as a proxy for the burden of transportation to the ACC, because driving distance may not always reflect actual travel time, although studies have shown that they are highly correlated.28 Nevertheless, for urban patients, driving distance may fail to capture added travel time associated with using public transportation.11 In addition, VHA patients are overwhelmingly male and mostly Caucasian, which may impact generalizability to the general population.
We did not find any evidence that patients living farther from the ACC receive a disproportionate benefit from PST in terms of satisfaction with anticoagulation care, anticoagulation control, or preventing adverse events. Therefore, our study does not support the notion that limiting PST to patients living farther from care would enhance its cost-effectiveness.
Author Affiliations: Center for Healthcare Organization and Implementation Research, Bedford VA Medical Center (AJR), Bedford, MA; Department of Medicine, Section of General Internal Medicine, Boston University School of Medicine (AJR), Boston, MA; Health Economics Resource Center (CSP, PS) and Cooperative Studies Program Coordinating Center (LU, RE, M-CS), VA Palo Alto Health Care System, Palo Alto, CA; Department of Pediatrics and Center for Primary Care and Outcomes Research, Stanford University School of Medicine (CSP), Stanford CA; Department of Health Research and Policy, Division of Biostatistics, Stanford University (M-CS), Stanford, CA; Jerry L. Pettis VA Medical Center, Research and Development Service (AJ), Loma Linda, CA; Department of Internal Medicine, School of Medicine, Loma Linda University (AJ), Loma Linda, CA; Durham VA Medical Center (DBM), Durham, NC; Division of General Medicine, Department of Medicine and Center for Clinical Health Policy Research, Duke University Medical Center (DBM), Durham, NC; Health Services and Systems Research Program, Duke-National University of Singapore Graduate Medical School (DBM), Singapore.
Source of Funding: The Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Clinical Sciences Research and Development Service, Cooperative Studies Program. The opinions expressed in this manuscript are those of the authors and do not necessarily represent the official views or policies of the US Department of Veterans Affairs.
Author Disclosures: Mr Su and Ms Uyeda are employees of the US Department of Veterans Affairs, which sponsored the main study, “The Home INR Study (THINRS),” which was the source of the data for this manuscript. Dr Jacobson reports being an advisory board member for the Anticoagulation Forum and the Loma Linda Veterans Association for Research; receiving consulting fees from Biosite, Boehringer Ingelheim, Daiichi Sankyo, Farallon Medical, Hemo-Sense, Inverness Medical, Pfizer Medical, Quality Assured Services, Roche Diagnostics, Sanofi-Aventis, and Tapestry Medical; receiving grants or has grants pending from Biosite, Boehringer Ingelheim, Farallon Medical, HemoSense, Inverness Medical, and Sanofi-Aventis; receiving honoraria from GlaxoSmithKline and Tapestry Medical; receiving fees for the development of educational presentations and for serving on speakers’ bureaus from Boehringer Ingelheim and GlaxoSmithKline; and receiving travel support from Biosite, Boehringer Ingelheim, Daiichi Sankyo, GlaxoSmithKline, and Sanofi-Aventis. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (AJR, CSP, AJ); acquisition of data (CSP, RE, AJ, DBM); analysis and interpretation of data (AJR, CSP, RE, M-CS, DBM, LU, PS); drafting of the manuscript (AJR, RE, M-CS, LU); critical revision of the manuscript for important intellectual content (AJR, CSP, RE, M-CS, AJ, DBM, PS); statistical analysis (RE, M-CS, DBM, PS, LU); provision of patients or study materials (AJ, DBM); obtaining funding (AJ, DBM); administrative, technical, or logistic support (DBM); and supervision (AJ).
Address correspondence to: Adam Rose, MD, MSc, Bedford VA Medical Center, 200 Springs Rd, Bedford, MA 01730. E-mail: firstname.lastname@example.org
1. Wysowski DK, Nourjah P, Swartz L. Bleeding complications with warfarin use: a prevalent adverse effect resulting in regulatory action. Arch Intern Med. 2007;167(13):1414-1419.
2. Shah SV, Gage BF. Cost-effectiveness of dabigatran for stroke prophylaxis in atrial fibrillation. Circulation. 2011;123(22):2562-2570.
3. Connolly SJ, Pogue J, Eikelboom J, et al; ACTIVE W Investigators. Benefit of oral anticoagulant over antiplatelet therapy in atrial fibrillation depends on the quality of international normalized ratio control achieved by centers and countries as measured by time in therapeutic range. Circulation. 2008;118(20):2029-2037.
4. Wallentin L, Yusuf S, Ezekowitz MD, et al; RE-LY Investigators. Efficacy and safety of dabigatran compared with warfarin at different levels of international normalised ratio control for stroke prevention in atrial fibrillation: an analysis of the RE-LY trial. Lancet. 2010;376(9745):975-983.
5. White HD, Gruber M, Feyzi J, et al. Comparison of outcomes among patients randomized to warfarin therapy according to anticoagulant control: results from SPORTIF III and V. Arch Intern Med. 2007;167(3):239-245.
6. Rose AJ, Hylek EM, Ozonoff A, Ash AS, Reisman JI, Berlowitz DR. Risk-adjusted percent time in therapeutic range as a quality indicator for outpatient oral anticoagulation: results of the Veterans Affairs Study to Improve Anticoagulation (VARIA). Circ Cardiovasc Qual Outcomes. 2011;4(1):22-29.
7. van Walraven C, Jennings A, Oake N, Fergusson D, Forster AJ. Effect of study setting on anticoagulation control: a systematic review and metaregression. Chest. 2006;129(5):1155-1166.
8. Garcia DA, Witt DM, Hylek E, et al; Anticoagulation Forum. Delivery of optimized anticoagulant therapy: consensus statement from the Anticoagulation Forum. Ann Pharmacother. 2008;42(7):979-988.
9. Rose AJ, Berlowitz DR, Frayne SM, Hylek EM. Measuring quality of oral anticoagulation care: extending quality measurement to a new field. Jt Comm J Qual Patient Saf. 2009;35(3):146-155.
10. Holbrook A, Schulman S, Witt DM, et al; American College of Chest Physicians. Evidence-based management of anticoagulant therapy: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest. 2012;141(suppl 2):e152S-e184S.
11. Hwang JM, Clemente J, Sharma KP, Taylor TN, Garwood CL. Transportation cost of anticoagulation clinic visits in an urban setting. J Manag Care Pharm. 2011;17(8):635-640.
12. Jowett S, Bryan S, Mahé I, et al. A multinational investigation of time and traveling costs in attending anticoagulation clinics. Value Health. 2008;11(2):207-212.
13. Schulman S, Anderson DR, Bungard TJ, et al. Direct and indirect costs of management of long-term warfarin therapy in Canada. J Thromb Haemost. 2010;8(10):2192-2200.
14. Garcia-Alamino JM, Ward AM, Alonso-Coello P, et al. Self-monitoring and self-management of oral anticoagulation. Cochrane Database Syst Rev. 2010;(4):CD003839.
15. Bloomfield HE, Krause A, Greer N, et al. Meta-analysis: effect of patient self-testing and self-management of long-term anticoagulation on major clinical outcomes. Ann Intern Med. 2011;154(7):472-482.
16. Heneghan C, Ward A, Perera R, et al; Self-Monitoring Trialist Collaboration. Self-monitoring of oral anticoagulation: systematic review and meta-analysis of individual patient data. Lancet. 2012;379(9813):322-334.
17. Matchar DB, Jacobson A, Dolor R, et al; THINRS Executive Committee and Site Investigators. Effect of home testing of international normalized ratio on clinical events. N Engl J Med. 2010;363(17):1608-1620.
18. Matchar DB, Jacobson AK, Edson RG, et al. The impact of patient self-testing of prothrombin time for managing anticoagulation: rationale and design of VA Cooperative Study #481--The Home INR Study (THINRS). J Thromb Thrombolysis. 2005;19(3):163-172.
19. Rose AJ, Hylek EM, Berlowitz DR, Ash AS, Reisman JI, Ozonoff A. Prompt repeat testing after out-of-range INR values: a quality indicator for anticoagulation care. Circ Cardiovasc Qual Outcomes. 2011;4(3):276-282.
20. Rosendaal FR, Cannegieter SC, van der Meer FJ, Briët E. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost. 1993;69(3):236-239.
21. Samsa G, Matchar DB, Dolor RJ, et al. A new instrument for measuring anticoagulation-related quality of life: development and preliminary validation. Health Qual Life Outcomes. 2004;2:22.
22. Cochran WG. Some methods of strengthening the common chi-squared tests. Biometrics. 1954;10:417-451.
23. Armitage P. Tests for linear trends in proportions and frequencies. Biometrics. 1955;11(3):375-386.
24. Jonckheere AR. A distribution-free k-sample test against ordered alternatives. Biometrika. 1954;41(1-2):133-145.
25. Terpstra TJ. The asymptotic normality and consistency of Kendall’s test against trend, when ties are present in one ranking. Indigationes Mathematicae. 1952;14:327-333.
26. Rose AJ, Hylek EM, Ozonoff A, Ash AS, Reisman JI, Berlowitz DR. Patient characteristics associated with oral anticoagulation control: results of the Veterans AffaiRs Study to Improve Anticoagulation (VARIA). J Thromb Haemost. 2010;8(10):2182-2191.
27. Goodman DC, Fisher E, Stukel TA, Chang C. The distance to community medical care and the likelihood of hospitalization: is closer always better? Am J Publ Health. 1997;87(7):1144-1150.
28. Jones SG, Ashby AJ, Momin SR, Naidoo A. Spatial implications associated with using euclidean distance measurements and geographic centroid imputation in health care research. Health Serv Res. 2010;45(1):316-327.