Background: Warfarin-associated adverse drug events are dangerous, common, and costly. Anticoagulation safety is a national priority. While multiple tools exist for warfarin management in the outpatient setting, there is a dearth of guidance with regard to inpatient management. This study aims to: 1) describe a health system’s inpatient chronic warfarin quality metrics, defined by International normalized ratio (INR) control, 2) explore associations between inpatient warfarin control and clinical outcomes, and 3) identify intrinsic and extrinsic factors associated with poor control.

Methods: We conducted a retrospective chart review extracting data from electronic health records of patients 65 years and older on chronic warfarin admitted between January 1, 2014 and June 30, 2016. We defined poor inpatient warfarin control as supra-therapeutic INRs (≥5) and limited analysis to INRs after 48 hours of hospitalization. We used logistic regression modelling to determine risk factors for poor control, and additional multivariate analysis to associate poor control with length of stay, bleeding, and mortality.

Results: There were 12,110 patients on chronic warfarin. Most patients (75.7%) were over age 75, female (51.2%), and white (70%). The most common comorbid medical conditions were congestive heart failure (46%), diabetes without chronic complications (31.4%), cerebrovascular disease (24%), chronic obstructive pulmonary disease (22.9%), moderate/severe chronic kidney disease (22.2%), and malignancy (22.1%). While 1,333 (11.0%) of patients reached an INR≥5 during the admission, 449 (33.7%) of these reached this maximum INR after 48 hours. When stratified by category, (poor control: INR≥5 after initial 48 hours versus non-poor control: INR<5 after initial 48 hours), length of stay more than doubled in the poor control group (15.6 days vs 6.8 days, <0.0001). The poor control group had a significantly higher bleed rate (27.4% vs 8.3%, adjusted OR 6.2, p<0.0001). While the overall mortality rate for the group was small (0.4%), there was a significantly higher mortality rate (3.12% vs 0.28%, adjusted OR 8.6, p<0.0001) in the group with poor control. Black race and weight were found to be protective against poor control after the first 48 hours; conversely, history of heart failure and antibiotic or amiodarone exposure were predictive of poor control. Moderate or severe CKD trended towards predicting poor control, but did not reach statistical significance (p<.068).

Conclusions: This is the largest study to date to examine warfarin quality metrics for older adults in the inpatient setting. We found that poor INR control is prevalent in the inpatient setting, is associated with poor outcomes, and is independently associated with low weight, heart failure, non-black race and antibiotic or amiodarone exposure. Our model may serve as the basis for identifying high risk patients on warfarin and for developing interventions for inpatient warfarin dosing strategies.