AUTHORS
Izat Mohammad khawajah Barzani, MD, Evan Draper, Pharm D2, Jan A Anderson, Pharm D3, Carlee O'Connor, D.N.P., R.N4, Ewa Wysokinska, MD5, Surbhi Shah, MBBS, MD6, Paul Daniels, MD7, Rajiv Pruthi, MBBS8, Harold Hellweg, MD9, Robert D McBane, MD10, Ana Isabel Casanegra, MD, MS11, Damon E Houghton, MD, MS12
AFFILIATIONS
1 Mayo Clinic, Rochester, Minnesota, United States, MayoClinic, Rochester, Minnesota, United States
2 Mayo Clinic, Rochester, Minnesota, United States, MayoClinic, Rochester, Minnesota, United States
3 MayoClinic, Rochester, Minnesota, United States
4 MayoClinic, Rochester, Minnesota, United States
5 MayoClinic, Rochester, Minnesota, United States
6 MayoClinic, Rochester, Minnesota, United States
7 MayoClinic, Rochester, Minnesota, United States
8 MayoClinic, Rochester, Minnesota, United States
9 MayoClinic, Rochester, Minnesota, United States, MayoClinic, Rochester, Minnesota, United States
10 MayoClinic, Rochester, Minnesota, United States
11 MayoClinic, Rochester, Minnesota, United States
12 MayoClinic, Rochester, Minnesota, United States
ABSTRACT
Background
Patients at increased risk of bleeding or thrombotic events may be assigned INR target ranges narrower than standard ranges during warfarin therapy to maintain efficacy while reducing harm. Because INR is inherently variable, it is unclear whether narrowing the assigned range meaningfully improves warfarin control. Time in therapeutic range (TTR) is widely used to assess warfarin management but is inherently influenced by target width and may be lower with narrow ranges. We compared conventional TTR with target-centered variability measures to determine whether narrow ranges were associated with more precise anticoagulation control in clinical practice.
Methods
We conducted a retrospective longitudinal study of patients managed in protocol-driven warfarin clinics across the Mayo Clinic Enterprise in Minnesota, Wisconsin, Iowa, Arizona, and Florida from 2019 through 2025. Standard INR target ranges were 2.0–3.0 or 2.5–3.5. Narrow targets were physician-assigned ranges 0.5 INR units wide that overlapped either standard range; the low-intensity INR target of 1.5–2.0 and other nonoverlapping ranges were excluded. Patients were required to contribute at least 90 eligible patient-days per target type; the first 14 days after warfarin initiation were excluded. Assigned-range TTR was calculated using Rosendaal-style patient-day interpolation. Because TTR is influenced by target width, target-centered analyses provide width-neutral comparisons. Mean absolute deviation (MAD) measured the average absolute distance from the assigned target center, whereas root mean squared error (RMSE) gave greater weight to larger deviations. We also assessed proportions within common ±0.25 and ±0.50 windows around the target center. Adjusted whole-cohort generalized estimating equation models provided population-averaged estimates while accounting for within-patient correlation. Monitoring frequency was summarized at the patient-target level using eligible INR measurements per 30 follow-up days and median intervals between consecutive measurements.
Results
The cohort included 19,120 unique patients; 18,852 contributed standard-target data and 524 narrow-target data, including 256 under both target types. Standard-target contributors included INR 2.0–3.0 (n=17,826) and 2.5–3.5 (n=1,523); narrow-target contributors included 1.8–2.3 (n=59), 2.0–2.5 (n=287), 2.5–3.0 (n=148), and 3.0–3.5 (n=52). Exact-range counts were nonexclusive. Target-centered metrics were derived from 1,406,197 eligible INR measurements (standard, 1,366,141; narrow, 40,056). Median eligible follow-up was 31.6 months (IQR, 12.3–64.8) and 20.2 months (IQR, 7.7–39.6), respectively. Mean assigned-range TTR was 69.2% with standard targets and 49.0% with narrow targets, corresponding to a 20.2-percentage-point lower TTR during narrow-target management. In adjusted whole-cohort target-centered analyses, narrow-target management was associated with lower MAD (difference, −0.11 INR units; 95% CI, −0.12 to −0.09), lower RMSE (difference, −0.15 INR units; 95% CI, −0.17 to −0.12), and a 16.50-percentage-point greater mean patient-level proportion within the common ±0.25 window (95% CI, 15.10–17.91). Mean patient-level proportions within the common ±0.50 window were 61.8% with standard targets and 66.2% with narrow targets, with an adjusted difference of 6.1 percentage points (95% CI, 4.69–7.45). Median monitoring rates were 1.9 INR measurements per 30 days (IQR, 1.3–2.8) with standard targets and 2.8 (IQR, 1.8–4.2) with narrow targets. Among patient-target periods with at least 2 INR measurements, median intermeasurement intervals were 11.5 days (IQR, 7.0–17.5) and 7.0 days (IQR, 7.0–14.0), respectively.
Conclusions
Lower TTR was observed during narrow-target management when calculated with traditional methods. However, target-center measures of variability were statistically improved demonstrating that for selected patients, a narrower range may help optimize anticoagulation control. Bleeding, thrombotic, and mortality outcomes were not evaluated, and therefore our results do not establish the clinical superiority of either strategy.
DISCLOSURES
Izat Mohammad khawajah Barzani, MD: No relevant conflicts of interest to declare.