Prompt Repeat Testing After Out-of-Range INR Values
A Quality Indicator for Anticoagulation Care
Background—Improved control of oral anticoagulation reduces adverse events. A program of quality measurement is needed for oral anticoagulation. The interval until the next test after an out-of-range International Normalized Ratio (INR) value (the “follow-up interval”) could serve as a process of care measure.
Methods and Results—We studied 104 451 patients cared for by 100 anticoagulation clinics in the Veterans Health Administration (VA). For each site, we computed the average follow-up interval after low (≤1.5) or high (≥4.0) INR. Our outcome was each site's average anticoagulation control, measured by percent time in therapeutic range (TTR); 59 837 patients (57%) contributed to the low INR analysis, 37 697 (36%) contributed to the high INR analysis, and all patients contributed to the dependent variable (mean site TTR). After a low INR, site mean follow-up interval ranged from 10 to 24 days. Longer follow-up intervals were associated with worse site-level control (1.04% lower for each additional day, P<0.001). After a high INR, site mean follow-up interval ranged from 6 to 18 days, with longer follow-up intervals associated with worse site-level control (1.12% lower for each additional day, P<0.001). These relationships were somewhat attenuated but still highly statistically significant when the proportion of INR values in-range was used as the dependent variable rather than TTR.
Conclusions—Prompt repeat testing after out-of-range INR values is associated with better anticoagulation control at the site level and could be an important part of a quality improvement effort for oral anticoagulation.
Although anticoagulation therapy with warfarin is potentially life-saving, it is also potentially dangerous. Warfarin has an extremely narrow therapeutic window, and fluctuations in the degree of anticoagulation can be difficult to anticipate or prevent.1 Meticulous control of anticoagulation, as measured by percent time in therapeutic range (TTR), has been shown to reduce the rate of adverse events in patients receiving anticoagulation, both at the level of the individual patient2–5 and at the site of care level.6,7 However, anticoagulation control is often suboptimal, leaving much room for improvement.8,9 Before we can improve the quality of anticoagulation care, we must be able to measure it. Therefore, we are in need of a program of quality measurement and quality improvement in the management of oral anticoagulation. Our group has proposed an outcome measure (anticoagulation control, as measured by risk-adjusted TTR)10 and has used it to profile 100 sites of care in the Veterans Health Administration (VA).11
TTR is an intermediate outcome measure that has been linked to definitive outcomes including stroke, venous thromboembolism, and major hemorrhage. Although outcome measures can be an excellent way to measure quality of care, they do not provide a prescription for action. Process measures are attractive in this regard because they do provide a prescription for action and because process can be measured at each clinical encounter.12,13 One attractive process measure for oral anticoagulation would be promptness of repeat testing after an out-of-range International Normalized Ratio (INR) value. Although both high and low INR values have been linked to patient harm,14–18 clinical guidelines do not give specific advice about the optimal follow-up interval after a high or low INR value.1,19 This is probably because there have been no studies of this important issue.
We therefore used a database of 100 sites of care and over 100 000 unique patients in the VA to address 2 questions. First, do sites of care differ regarding the interval until the next INR test after a low (≤1.5) or a high (≥4.0) INR value? Second, how do these differences relate to site-level performance, as defined by risk-adjusted TTR? By demonstrating both significant variations in practice and their relationship with intermediate outcomes of care, we sought to find support for promptness of follow-up after an out-of-range INR value as a process measure for oral anticoagulation care.
Episodes of excessive or insufficient anticoagulation increase the risks of bleeding and thromboembolism, respectively.
There have been no previous studies regarding the ideal follow-up interval after a high (≥4) or low (≤1.5) International Normalized Ratio (INR) value.
In the absence of empirical evidence, clinical guidelines make no specific recommendations regarding follow-up after an out-of-range INR value.
WHAT IS KNOWN
In our study, the mean interval until the next INR test after a high or low INR varied widely among 100 sites of care in an integrated health care system (from 6 to 18 days after a high INR and from 10 to 24 days after a low INR).
Sites with shorter mean follow-up intervals had better anticoagulation control. Risk-adjusted site mean percent time in range was approximately 1% lower for each additional day of the follow-up interval after either a high or low INR.
Follow-up within 1 week after a high or low INR appears to be ideal, based on our results; this has the potential to serve both as a performance measure and as a putative standard of care.
WHAT THE STUDY ADDS
The database for this study has also been described elsewhere.8,11 The Veterans AffaiRs Study to Improve Anticoagulation (VARIA) included all patients deemed to be receiving oral anticoagulation therapy (OAT) from the VA between October 1, 2006, to September 30, 2008, based on the criteria described below. The study was approved by the Institutional Review Board of the Bedford VA Medical Center.
We included all patients who received warfarin from the VA during the 2-year study period (ie, at least 30 days' worth dispensed by the pharmacy) and who had at least 2 valid intervals for calculating percent TTR.20 For this purpose, a valid interval consists of 2 INR values separated by 56 days or less, without an intervening hospitalization, as in the original study by Rosendaal et al.20
We excluded patients whose primary indication to receive warfarin was valvular heart disease. Many such patients have a target INR range of 2.5 to 3.5 rather than the more standard 2 to 3, but it is not possible to determine with certainty which patients have the higher target range. Without specific knowledge of the target range, we cannot calculate TTR. We also excluded patients who only recorded INR values 1.2 and lower, reasoning that most such patients received INR tests for reasons unrelated to warfarin management (eg, frequent emergency department visits).
Laboratory Values and Calculation of Percent TTR
We included INR values within the VA system that were obtained while patients were “on warfarin,” that is, when a patient was either (1) in possession of warfarin or (2) having INR tests at least every 42 days. This choice of a 42-day interval is based on previous work by Go et al21 as well as the current American College of Cardiology/American Heart Association/European Society of Cardiology guidelines for the maximum allowable follow-up interval for patients anticoagulated for atrial fibrillation.22 We defined the period of warfarin possession as the duration of the most recent VA prescription for warfarin, plus 30 days. We excluded INR tests measured while the patient was hospitalized within the VA system. Patients who are hospitalized may receive temporary parenteral anticoagulation (eg, with heparin) or no anticoagulation; therefore out-of-range INR values while hospitalized may be intentional and do not necessarily reflect poor quality of care. For this study, we also excluded patient INR data from the first 6 months of therapy with warfarin (the “inception period”). We have previously shown that TTR is lower during the inception period,8,11,23 and decisions regarding the follow-up interval may also differ during this period.
We calculated TTR using the Rosendaal method,20 which uses linear interpolation to assign an INR value to each day between successive observed INR values. Gaps of 56 days or more between INR values are not interpolated. After interpolation, the percentage of time during which the interpolated INR values lie between 2.0 and 3.0 (from 0% to 100%) is calculated.20
Sites of Care
We included 100 VA sites of care, each of which includes a hospital, an outpatient care center, and several outlying community-based clinics. Each site has a specialized anticoagulation clinic, which is usually run by clinical pharmacists under the supervision of a medical director.24 Therefore, essentially all patients whose anticoagulation is managed in the VA are treated by specialized anticoagulation clinics. Most patients only visited one site of care, and their INR data were assigned to that site. If a patient visited more than 1 site (3% of patients), we partitioned their data by site.
Risk Adjustment Model
We have previously described the derivation and validation of our risk adjustment model for TTR.8 We considered many potential variables that we thought were likely to affect TTR, including demographics, area-level poverty, driving distance to care, physical health conditions, mental health conditions, number of medications, and number of hospitalizations. Most variables were retained within the model, with the exception of several comorbid conditions that did not have appreciable effect sizes. The model was derived and validated according to customary procedures, which included considerations of maximizing predictive ability, clinical credibility, and ease of use and understanding.8 This patient-level risk adjustment model for TTR has an R2 of 13.3% when used with this dataset.11 Table 1 contains all the variables that were retained in the final model.
Dependent Variables: Site-Level Anticoagulation Control
For our dependent variable, we used 2 separate measures of site-level anticoagulation control. Our main dependent variable was mean site risk-adjusted TTR. We calculated risk-adjusted TTR for each patient who received anticoagulation management at our 100 sites of care, whether or not they recorded any out-of-range INR values. This is because we wanted to measure the results achieved by each site for all of its patients, as a measure of overall quality of care. Site risk-adjusted TTR was calculated using the following procedure. First, for each patient, we calculated the observed TTR (“O”) and applied the risk adjustment model to calculate the expected TTR (“E”). Then, an observed minus expected (O-E) score was calculated for each patient. The mean O for each site constituted its unadjusted TTR, whereas the mean O-E score for each site constituted its risk-adjusted TTR. A full explanation of how we calculated risk-adjusted TTR, as well as a comparison between adjusted and unadjusted TTR, can be found in our earlier reports.8,11
One possible criticism of using TTR as the outcome for this study is that TTR will necessarily reward sites for following up sooner after an out-of-range value, even if the date on which the patient's control improved was in fact the same, and all that changed was the promptness of measurement. We therefore examined the extent to which our results depended on this mathematical truism. We substituted site-level proportion of INR values in range for risk-adjusted TTR as an alternate dependent variable and reran our main analyses. This allowed us to determine whether we still saw the same effect even without the measurement property related to the calculation of TTR.
Independent Variable: Site Mean Interval Until Next INR Test
We characterized each site regarding the mean interval until the next test after an out-of-range INR value (the “follow-up interval”). We divided out-of-range INR values into 2 categories: ≤1.5 (“low”) and ≥4.0 (“high”). Both high and low INR values have clearly been linked to patient harm1,14–18 and should be addressed promptly to bring the patient back within the therapeutic range. In addition, we examined follow-up intervals after INR values that are only slightly out of range: “slightly low” (1.6 to 1.9) and “slightly high” (3.1 to 3.9). The evidence regarding harm from such values is less robust, so we wanted to characterize the extent to which sites are responding more promptly to the more extreme values. In addition, we wanted to explore the effect of the management of slightly out-of-range INR values on site-level TTR.
We located all patients who had at least 1 high or low INR value (the “index value”) followed by another INR within 56 days without intervening hospitalizations. For an out-of-range INR value, another INR is expected soon after; therefore, the following INR could be recorded as soon as the next day. If the patient was hospitalized between the index INR and the next value, we looked back to the next possible index value instead, because hospitalization also counts as prompt follow-up of the aberrant value. When a patient had multiple qualifying episodes of a high or low INR, we selected the last such episode, so that each patient was sampled no more than once for high and/or once for low. We also reran our analyses after selecting the first such episode or a random episode of high or low INR; the results did not change appreciably (data not shown). We averaged values from individual patients to calculate mean values for each site. We followed similar procedures to characterize each site regarding its response to slightly low (1.6 to 1.9) and slightly high (3.1 to 3.9) INR values.
We examined the baseline characteristics of patients in our source population as well as the characteristics of patients who were included in the subsamples to analyze follow-up intervals after a high or low INR value. We characterized each site by a mean follow-up interval after a low INR (≤1.5), after a high INR (≥4.0), and after an INR value that was slightly low (1.6 to 1.9) or slightly high (3.1 to 3.9). We modeled the site-level relationships between follow-up intervals and risk-adjusted TTR using simple correlation, linear regression, and ANOVA with the Tukey honestly significant differences test (after grouping sites into quintiles by follow-up intervals). We repeated these analyses using our alternate dependent variable (proportion of INR values in range by site). All analyses were conducted using SAS, version 9.1 (SAS Corporation). Dr Rose had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
We studied 104 451 unique patients who received anticoagulation during the experienced period (ie, >6 months' experience with anticoagulation). Baseline characteristics for the source population are described in Table 1. The sample was mostly male (98%) and had a median age of 72 years. Most patients (64%) were anticoagulated for atrial fibrillation, with the remainder anticoagulated for venous thromboembolism (27%) or other indications (9%; eg, mural thrombus, cardiomyopathy, pulmonary hypertension, etc). The population had a substantial burden of comorbidity. For example, 40% had diabetes mellitus, 33% had heart failure, 14% had chronic kidney disease, and 7% were newly diagnosed with (nonskin) cancer during the study period. The burden of mental illness and substance abuse was also considerable: 22% had major depression, 9% had a diagnosis of alcohol abuse, and 5% had dementia.
In general, differences between the source population and the samples used to study high and low INR were slight. Some characteristics that we have previously linked to lower patient-level TTR (such as cancer, alcohol abuse, and substance abuse) were slightly more common among the high and low INR samples; these differences attained statistical significance due to the large study size (for example, P<0.001 for these 3 variables). Mean TTR for the source population was 61%, compared with 54% in the high and low INR groups, respectively (P<0.001 for both comparisons).
Relationship Between Follow-Up Intervals and Site-Level Anticoagulation Control
There were 100 sites of care in the database. Site mean TTR ranged from 41% to 72% and site risk-adjusted performance (site O-E score) ranged from 19% below to 12% above expected. Performance was above expected at 46 sites and below expected at 54 sites; the median site O-E score was 0.6% below expected. Sites differed widely regarding mean follow-up intervals. After a low INR (≤1.5), site mean follow-up intervals ranged from 10 to 24 days (Figure 1). After a high INR (≥4.0), site mean follow-up intervals ranged from 6 to 18 days (Figure 2). Site mean follow-up intervals after slightly low and slightly high INR ranged from 12 to 32 and from 13 to 31 days, respectively (data not shown). Generally, the sites that pursued prompter follow-up after low INR also pursued prompter follow-up after high INR. For example, the correlation between site-level mean follow-up interval after low and high INR values was 0.72.
With regard to follow-up after a low INR (≤1.5), shorter site-level intervals correlated with improved site-level performance as measured by risk-adjusted TTR (Figure 1; r=−0.59, P<0.001). This result was unchanged when unadjusted site-level TTR was used as the outcome (r=−0.59). With regard to follow-up after a high INR (≥4.0), shorter site-level intervals correlated with improved site-level performance as measured by risk-adjusted TTR (Figure 2; r=−0.57, P<0.001). This result was unchanged when unadjusted site-level TTR was used as the outcome (r=−0.57).
We further examined the strength of these relationships. For each additional day of mean site-level follow-up interval after a low INR, site-level risk-adjusted TTR was 1.04% lower (95% confidence interval, 0.75 to 1.32%; P<0.001). We found similar results when we used a cutoff of ≤1.3 to define a low INR value (data not shown). For each additional day of mean site-level follow-up interval after a high INR, site performance was 1.12% lower (95% confidence interval, 0.80 to 1.43%; P<0.001). We found similar results when we used a cutoff of ≥5.0 to define a high INR value (data not shown).
We repeated these analyses in subpopulations defined by indication for anticoagulation, namely patients anticoagulated for atrial fibrillation (64% of the sample) and patients anticoagulated for venous thromboembolism (27% of the sample). The main findings of the study did not change. For example, for each additional day after a low INR value, site-level risk-adjusted TTR was 1.00% lower among atrial fibrillation patients, 1.01% lower among venous thromboembolism patients, and 1.04% lower among all patients (P<0.001 for all 3 findings).
We also divided sites into quintiles (20 sites per group), based on the intervals after a low or high INR value. Site performance was generally best in the quintile with the shortest follow-up and worst in the quintile with the longest follow-up (Table 2), although differences among the middle 3 quintiles were small and not statistically significant.
We also examined the correlation between follow-up after mildly out-of-range INR values and site risk-adjusted TTR. These correlations were slightly less than for more pronounced deviations from the target range, but were still considerable (Table 3). For slightly low values (1.6 to 1.9), the correlation was −0.53 (P<0.001), compared with −0.59 for low values (≤1.5). For slightly high values (3.1 to 3.9), the correlation was −0.45 (P<0.001), compared with −0.57 for high values (≥4.0).
Sensitivity Analysis: Substituting Proportion of INR Values in Range for Percent Time in Range
As discussed in the methods, we substituted the proportion of INR values in range at each site for the site O-E score (ie, risk-adjusted TTR) as our dependent variable. As shown in Table 3, substitution of proportion of INR values in range lessened the magnitude of these correlations, but they were still present and statistically significant. Effect sizes decreased much more with regard to follow-up after mildly out-of-range INR values than severely out-of-range values. For example, the correlation for follow-up after low INR (≤1.5) decreased from −0.59 to −0.42 and the correlation for follow-up after slightly low INR (1.6 to 1.9) decreased from −0.53 to −0.29.
Improving the quality of care in oral anticoagulation has the potential to save thousands of lives per year in the United States as well as preventing numerous nonfatal events that nevertheless lead to hospitalization or institutionalization.2–7,10 To improve quality of care in oral anticoagulation, we need valid quality measures. In particular, process measures may be useful because they can provide a ready prescription for action and remediation.10 In this study, we found that sites vary considerably with regard to promptness of repeat testing after an out-of-range INR value and that prompter repeat testing after out-of-range INR values (a process measure) is associated with improved site-level anticoagulation control (an intermediate outcome of care). This relationship was true both with and without risk adjustment for patient characteristics and both with and without linear interpolation between adjacent INR values.
Despite being part of an integrated health system, the 100 sites that we studied had a wide range of practice in this regard, possibly due to the relative lack of evidence and clear guideline recommendations. For example, after a high INR (≥4.0), site mean follow-up intervals ranged from 6 to 18 days. These variations had important consequences for anticoagulation control: Based on our regression models, a site with an average 6-day interval after a high INR would be expected to have a mean TTR 13% higher than a site with an average 18-day interval. This is a very large difference in anticoagulation control and one that has been linked to considerable differences in rates of adverse events.2–7 It is likely that clinicians practicing at each site arrive at a consensus about the ideal follow-up interval in certain situations, whether by written policy or unwritten common practice. These site-level tendencies can presumably be changed, and our results suggest that prompter follow-up could improve TTR considerably for most of the sites in our study.
In addition to its implications for quality measurement, our study also has implications for clinical practice guidelines in anticoagulation care. Our results suggest that anticoagulation control could be improved considerably by following up within 7 days after a high (≥4.0) or low (≤1.5) INR value and within 14 days after a mildly high (3.1 to 3.9) or mildly low (1.6 to 1.9) INR value. If all VA patients had been treated in this manner during our study, our results suggest that the VA might have recorded an overall TTR between 5% to 10% higher, a difference that has been associated with meaningful improvements in the rates of outcomes such as stroke, venous thromboembolism, major hemorrhage, and mortality.2–7
This study has several strengths. We used a large and powerful database, rich in clinical detail. We used 2 measures of anticoagulation control, both a simple one (proportion of values in range) and one that is the result of much development by our group and represents the state of the art in quality measurement for oral anticoagulation (risk-adjusted TTR).8,10,11 The consistency of our findings suggests that they are not attributable only to the measures used but represent a real and important finding that means exactly what one would think it means.
However, several limitations should be noted. First, we measured average follow-up intervals at the site level rather than at the level of the individual patient or the individual instance. Sites of care probably determine these follow-up intervals through written or unwritten policies, which can and should be changed to improve performance. However, follow-up for individual patients is dependent on many clinical considerations and is inherently variable. Therefore, we believe that it would be ill advised to attempt to measure quality of care for individual patients using these measures—although this issue could certainly be examined empirically. Second, we measured the actual interval that elapsed between INR values but did not measure the interval that the clinician requested. To some extent, nonadherence to recommendations on the part of the patient might have played a role in the correlation between longer follow-up intervals and poor control. However, our main outcome measure was risk-adjusted for the patient population at each site, which should have controlled for many of the patient-level factors that contribute to poor compliance and poor control. Third, this study only included patients with a target INR range of 2 to 3. Therefore, these results may not be generalizable to patients with other target ranges (most often 2.5 to 3.5). Finally, VA patients are mostly male and have a high burden of comorbidity. However, it is unclear how this fact would have altered the basic relationships that we showed between follow-up intervals and site-level anticoagulation control.
In summary, we found that there is a wide range of practice regarding the interval until a repeat test after out-of-range and mildly out-of-range INR values. Longer follow-up intervals were associated with worse anticoagulation control. We believe that we have truly identified a quality measure for oral anticoagulation care. Our study suggests that optimizing follow-up intervals after out-of-range INR values could greatly improve anticoagulation control and prevent thousands of fatal or morbid adverse events each year.
Sources of Funding
Dr Rose is supported by a career development award from the US Department of Veterans Affairs, Health Services Research, and Development Service (CDA-08-017). The sponsor had no role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation, review, and approval of the manuscript.
Dr Hylek received honoraria from Bayer and Bristol Myers Squibb and has served on advisory boards for Boehringer-Ingelheim, Bristol Myers Squibb, Merck, and Sanofi-aventis.
The opinions expressed in this manuscript do not necessarily represent the official views of the Department of Veterans Affairs.
- Received November 21, 2010.
- Accepted February 18, 2011.
- © 2011 American Heart Association, Inc.
- Veeger NJ,
- Piersma-Wichers M,
- Tijssen JG,
- Hillege HL,
- van der Meer J
- Connolly SJ,
- Pogue J,
- Eikelboom J,
- Flaker G,
- Commerford P,
- Franzosi MG,
- Healey JS,
- Yusuf S
- Wallentin L,
- Yusuf S,
- Ezekowitz MD,
- Alings M,
- Flather M,
- Franzosi MG,
- Pais P,
- Dans A,
- Eikelboom J,
- Oldgren J,
- Pogue J,
- Reilly PA,
- Yang S,
- Connolly SJ
- Rose AJ,
- Hylek EM,
- Ozonoff A,
- Ash AS,
- Reisman JI,
- Berlowitz DR
- Donabedian A
- Rose AJ,
- Ozonoff A,
- Grant RG,
- Henault LE,
- Hylek EM
- Ansell JE,
- Oertel LB,
- Wittkowsky AK
- Fuster V,
- Ryden LE,
- Cannom DS,
- Crijns HJ,
- Curtis AB,
- Ellenbogen KA,
- Halperin JL,
- Le Heuzey JY,
- Kay GN,
- Lowe JE,
- Olsson SB,
- Prystowsky EN,
- Tamargo JL,
- Wann S,
- Smith SC Jr.,
- Jacobs AK,
- Adams CD,
- Anderson JL,
- Antman EM,
- Halperin JL,
- Hunt SA,
- Nishimura R,
- Ornato JP,
- Page RL,
- Riegel B,
- Priori SG,
- Blanc JJ,
- Budaj A,
- Camm AJ,
- Dean V,
- Deckers JW,
- Despres C,
- Dickstein K,
- Lekakis J,
- McGregor K,
- Metra M,
- Morais J,
- Osterspey A,
- Tamargo JL,
- Zamorano JL
Department of Veterans Affairs: Veterans Health Administration. VHA Directive 2009-003: Anticoagulation Therapy Management. Washington, DC: February 2, 2009.