Out-of-range international normalized ratio values and healthcare cost among new warfarin patients with non-valvular atrial fibrillation

Abstract

Background:

Patients with out-of-range international normalized ratio (INR) values <2.0 and >3.0 have been associated with increased risk of thromboembolic and bleeding events. INR monitoring is costly, because of associated physician and nurse time, laboratory resource use, and dose adjustments.

Objectives:

This study assessed the healthcare cost burden associated with out-of-range INR among warfarin initiator patients diagnosed with non-valvular atrial fibrillation (NVAF) in the US Veterans Health Administration (VHA) population.

Methods:

Adult NVAF patients (≥18 years) initiating warfarin were selected from the VHA dataset for the study period October 1, 2007–September 30, 2012. Only valid INR measurements (0.5 ≤ INR ≤ 20) were examined for the follow-up period, from the index date (warfarin initiation date) until the end of warfarin exposure or death. All-cause healthcare costs within 30 days were measured starting from the second month (31 days post-index date) to the end of the study period. Costs for inpatient stays, emergency room, outpatient facility, physician office visits, and other services were computed separately. Multiple regression was performed using the generalized linear model for overall cost analysis.

Results:

In total, 29,463 patients were included in the study sample. Mean costs for out-of-range INR ranged from $3419 to $5126. Inpatient, outpatient, outpatient pharmacy, and total costs were significantly higher after patients experienced out-of-range results (INR < 2, INR > 3), compared with in-range INR (2 ≤ INR ≤ 3). When exposed to out-of-range INR, patients also incurred higher mean total costs within 2–6 months ($3840–$5820) than after the first 6 months ($2789–$3503) of warfarin therapy.

Conclusion:

In the VHA population, INR measures outside of the 2–3 range were associated with significantly higher healthcare costs. Increased costs were especially apparent when INR values were below 2, although INR measures above 3 were also associated with higher costs relative to in-range values.

Keywords::

• Atrial fibrillation
• Healthcare cost
• International normalized ratio
• Warfarin

Introduction

Atrial fibrillation (AF) is the most common heart dysrhythmia in the US, affecting an estimated 2–3 million patients1^,2. When the dysrhythmia is not associated with rheumatic mitral valve disease or prosthetic heart valves, the disease is referred to as non-valvular (NV) AF3. AF is associated with significant clinical morbidity and is an independent risk factor for mortality4.

AF prevalence is expected to increase, which will increase healthcare costs. The total cost of AF care in the US is estimated at $6.65 billion per year4. In a 2002 study, direct costs among AF patients were $15,553 per year, $12,349 higher than costs per year for enrollees without AF4. In another retrospective study, a total of 35,255 primary AF patient claims were examined, and the resulting inpatient costs were $11,307, whereas outpatient costs were $2827 (total $14,133) per patient over 12 months5. Costs were primarily due to inpatient expenses. However, the frequent need for re-admission contributes to hospital costs among AF patients. In a 2013 study of 6439 hospitalized NVAF patients, 30-day re-admission rates from the index hospital stay were 18%6.

AF treatment and associated complications increase healthcare resource use and contribute to greater healthcare costs4. Anti-coagulant medications are the current preventative therapy for high-risk patients diagnosed with NVAF, and warfarin is commonly prescribed7^,8. Several large-scale clinical trials have demonstrated the effectiveness of warfarin to control AF-related risk factors9–11. However, to maximize therapy benefits and minimize complications (e.g., bleeding), warfarin therapy should be closely monitored and adjusted within the narrow therapeutic index, using the international normalized ratio (INR) test12. Current guidelines recommend that INR values for AF patients should be maintained between 2.0 and 3.013. If the INR is below range, the patient will be at increased risk of thromboembolism, and, if the INR is above range, the patient will be at increased risk of experiencing a hemorrhagic event14.

Studies have documented the substantial costs associated with monitoring warfarin therapy among patients with chronic AF15. Warfarin costs are low per se, but monitoring the INR incurs substantial costs, mostly because the process involves physicians’ and nurses’ time as well as the use of laboratory resources and testing devices8^,14. Furthermore, dose adjustments and complications may increase costs considerably. Although the above-mentioned analyses have been performed, to our knowledge, no studies have focused on the healthcare costs associated with out-of-range INR values among warfarin initiators. The objective of this study was to assess the healthcare cost burden associated with out-of-range INR test results among NVAF patients initiating warfarin treatment in the US Veterans Health Administration (VHA) population.

Methods

This study report was written in compliance with the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) statement16.

Data source

To evaluate costs among warfarin initiator patients, the VHA dataset was used to assess medical, pharmacy, laboratory, and enrollment information. The VHA is the largest integrated healthcare system in the US, providing care for veterans nationwide. According to the US Department of Veterans Affairs (VA), the system includes: 162 VA hospitals, 137 nursing homes, 43 domiciliaries, more than 850 community and facility-based clinics, 14,800 doctors, 61,000 nurses, and 5 million patients17.

Electronic health data collected within the VA’s national Medical SAS Dataset and Decision Support System (DSS) were evaluated18^,19. These data include hospital and outpatient diagnoses (recorded using International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] codes), procedures (recorded using ICD-9 and Current Procedural Terminology [CPT] codes), laboratory results, and dispensed medications. The ICD-9 and CPT codes are provided in the Appendix. The VHA DSS data files comprise a longitudinal, secondary relational database combining selected clinical data and fiscal (cost) data.

Study population

The study population consisted of patients aged ≥18 years with at least one pharmacy claim for warfarin (National Drug Code) during the identification period from October 1, 2008, through September 30, 2011. The first such claim date was designated as the index date. Patients who had at least one medical claim for AF (ICD-9-CM code 427.31) during the period beginning 30 days before and extending through the index date were included in the study. These patients had at least one valid INR test result value (0.5 ≤ INR ≤ 20) within 14 days after the index date, at least three valid INR measurements following the index date, and continuous health plan enrollment for the 12-month baseline period.

The population was further restricted to patients diagnosed with AF but without valvular heart disease (NVAF). Therefore, patients with at least one medical claim indicating the presence of transient perioperative AF disease, mitral or aortic valve repair or replacement, or at least two medical claims for hyperthyroidism at any time during the study period from October 1, 2007 through September 30, 2012, were excluded. To ensure patients had newly initiated warfarin, those with at least one pharmacy claim for warfarin during the baseline period were excluded. In addition, patients with a record indicating pregnancy or delivery during the entire study period were also excluded.

Study measures

After evaluating the distribution, three patient cost groups were constructed corresponding to the 30-day period after exposure to: (1) INR values <2; (2) INR values >3; and (3) in-range INR values (2–3). For patients with multiple INR results, these INR measurements were required to occur at least 30 days apart. Thirty-day healthcare costs were determined by randomly selecting one INR value in each group per patient and summing the direct medical costs during the 30-day period after that INR test. Therefore, each patient contributed at most three 30-day costs, corresponding to randomly selected INR values that were <2, 2–3, and >3. INR values from the first month of the study period were excluded to avoid bias caused by fluctuation of INR values and costs. A sensitivity analysis was also conducted among a subset of patients who contributed to all three cost groups.

Costs for inpatient stays and ER, outpatient facility and physician office visits, pharmacy, and total costs (inpatient + outpatient + pharmacy) were calculated. Costs were adjusted to 2012 US dollars using the annual medical care component of the Consumer Price Index to reflect inflation between 2007 and 2012.

Data analyses

Baseline demographic and clinical characteristics for warfarin initiator patients with NVAF were summarized. Descriptive statistics were employed to report categorical variables as counts and percentages, and quantitative variables as means and standard deviations (SDs).

The overall costs and sub-categories of cost (inpatient, outpatient, and pharmacy) associated with out-of-range INR within a 30-day period were assessed. In addition, because more resources may be expended when INR is stabilizing during the early months after warfarin initiation, the cost comparison analyses were conducted in two separate intervals: the first 2–6 months and after 6 months, to evaluate the variation in the cost burden. Since some patients did not contribute to the three cost values, multiple regression was conducted using a generalized linear model. The Park Test was applied to choose the right family function20. Four generalized linear models with log-link function and gamma variance were applied to estimate each dependent variable: inpatient, outpatient, pharmacy, and total healthcare costs. The regression model was adjusted for patient demographic and clinical characteristics. Statistical significance was assessed for p-values <0.05. In addition to p-values, standardized differences were calculated for each variable. All analyses were conducted using SAS statistical software (Version 9.3, SAS Institute, Cary, NC).

Results

The study population included 29,463 NVAF patients who initiated warfarin. Among those, 23,588 patients had at least one 30-day period with INR results at <2; 21,142 had in-range (2 ≤ INR ≤ 3) results; and 16,529 had INR test values at >3 during warfarin exposure. Figure 1 shows the sample selection process. The same patient may have had periods of varying INR results. Patients were subdivided into three cost categories, and some did not contribute to all three 30-day cost categories. Demographic and baseline characteristics for the 29,463 patients who contributed to any of the three cost groups are provided in Table 1.

Figure 1. Patient selection. AF, atrial fibrillation; INR, international normalized ratio; VHA, Veterans Health Administration.

Table 1. Baseline demographic and clinical characteristics for warfarin patients with NVAF (patients contribute to three cost groups).

	Warfarin patients with NVAF (n = 29,463)
	n/Mean	%/SD
Age
Age (mean)	70.77	10.09
Age 18–64	9553	32.42%
Age 65–74	8381	28.45%
Age 75+	11,529	39.13%
Gender
Male	28,988	98.39%
Female	475	1.61%
US geographic region
Northeast	4255	14.44%
Midwest	7378	25.04%
South	10,255	34.81%
West	6049	20.53%
Other region	1526	5.18%
Race
Black	2238	7.60%
Hispanic	957	3.25%
White	20,556	69.77%
Other race	5712	19.39%
Baseline comorbid conditions
Indices
Charlson comorbidity index score	1.93	1.98
Chronic disease score	7.66	4.43
CHADS2 score
Low (0)	3080	10.45%
Moderate (1)	7928	26.91%
High (>1)	18,455	62.64%
CHADS2 VASc score
Low (0)	1678	5.70%
Moderate (1)	3605	12.24%
High (>1)	24,180	82.07%
Atria bleeding risk score
Low (0–3)	23,656	80.29%
Moderate (4)	2099	7.12%
High (5–10)	3708	12.59%
Comorbidities
Congestive heart failure	5762	19.56%
Diabetes mellitus	11,758	39.91%
Hypertension	22,563	76.58%
Ischemic stroke	1386	4.70%
Transient ischemic attack	710	2.41%
Vascular disease	11,460	38.90%
End-stage renal disease	290	0.98%
Hospitalized hemorrhagic stroke	0	0.00%
Chronic kidney disease	3634	12.33%
Chronic obstructive pulmonary disease	1356	4.60%
Anemia	3783	12.84%
Hospitalized major bleeding	121	0.41%
Bleeding disorder	472	1.60%
Baseline medications
Use of rate control medications	19,415	65.90%
Use of rhythm control medications	2266	7.69%
Use of aspirin	8711	29.57%

NVAF, non-valvular atrial fibrillation; SD, standard deviation.

CHADS₂ score: Congestive heart failure (1 point), Hypertension (1 point), Age ≥75 years (1 point), Diabetes mellitus (1 point), prior Stroke or transient ischemic attack (2 points).

CHADS₂ VASc score27: Congestive heart failure (1 point), Hypertension (1 point), Age ≥75 years (2 points), Diabetes mellitus (1 point), prior Stroke or transient ischemic attack (2 points); Vascular disease (1 point), Age 65–74 years (1 point), female category (1 point).

Atria bleeding risk score: anemia = hemoglobin <13 g/dL (male), <12 g/dL (female) (3 points); severe renal disease = glomerular filtration rate <30 mL/min or dialysis-dependent (3 points); age ≥75 (2 points); any prior hemorrhage diagnosis (1 point); hypertension history (1 point).

Follow-up healthcare costs associated with out-of-range INR, within a 30-day period

The mean (±SD) of the randomly selected INR that started the 30-day period for the three groups are as follows: 1.58 (±0.28) for <2, 2.40 (±0.29) for 2–3, and 3.77 (±1.08) for >3. Figure 2 presents the healthcare costs incurred within a 30-day period after exposure to an out-of-range INR result, compared with those incurred during 30 days of in-range INR results. Inpatient, outpatient, pharmacy, and total costs were significantly higher after INR values were <2 and >3, compared with in-range (2 ≤ INR ≤ 3). Exposure to INR < 2 was associated with higher costs than exposure to INR > 3.

Figure 2. Follow-up healthcare costs associated with out-of-range INR, within a 30-day period. Costs and utilization within a 30-day period were assessed starting from Month 2 (31 days post index date) through the end of the study period. In-range INR (2–3): randomly selected INR test occurrence (INR 2–3 result occurred at least 30 days apart and no other INR range type [<2 or >3] occurred within 30 days). Out-of-range INR (<2 or >3): randomly selected INR test occurrence (INR tests occurred at least 30 days apart; different range types occurring within 30 days were allowed). INR, international normalized ratio. *p-Values are based on multiple regression using a generalized linear model.

Follow-up healthcare costs in a 30-day period, within the first 2–6 months of an INR test

Figure 3 further stratifies the cost data according to INR range for two periods: within 2–6 months and after 6 months of warfarin therapy. During both periods, total 30-day costs after exposure to out-of-range INR results were significantly higher than those during in-range INR exposure. Similar to the main study findings, follow-up healthcare costs were higher after exposure to INR < 2 compared with INR > 3. The study results show that the cost burden of out-of-range INR was less intensive after the first 6 months of warfarin therapy.

Figure 3. Follow-up healthcare resource use within a 30-day period, for the first 2–6 months and after 6 months of warfarin exposure. Patients with INR 2–3: randomly selected INR test for the 30-day period were collected from the second month through the end of warfarin exposure. Patients with INR <2 or >3: randomly selected INR test occurrence during the first 2–6 months and after 6 months of warfarin treatment. INR, international normalized ratio. *p-Values are based on multiple regression using a generalized linear model.

Sensitivity analysis

After limiting patients who contributed to all three cost groups, the smaller sample of 10,770 patients became more homogenous, and the differences in 30-day total costs between out-of-range INR and in-range INR remained significant and smaller. Follow-up healthcare costs for INR < 2, 2 ≤ INR ≤ 3, and INR > 3 were $3426, $2105, and $2807, respectively (INR < 2 vs 2 ≤ INR ≤ 3: p < 0.0001; INR > 3 vs 2 ≤ INR ≤ 3: p < 0.0001).

Discussion

Among NVAF patients with newly initiated warfarin therapy, most of the patients in our study were exposed to out-of-range values (80% with at least one INR value of <2; 56% with at least one INR value of >3). Our results demonstrated that the total cost after INR values of <2 were 2.18-times higher than that when INR was in range. Similarly, total costs after INR values of >3 were 1.45-times higher than that during in-range INR values. The annual costs for INR monitoring have been reported in other studies, but to our knowledge, our study is the first to report costs specifically associated with out-of-range INR.

A 1996–1998 study by Menzin et al.21 involving NVAF patients reported the mean per-patient costs of monitoring warfarin therapy for 1 year ranged from $216 to $339. In a recent systematic review of 29 studies evaluating the costs associated with INR monitoring during oral anti-coagulation with vitamin K antagonists, the costs of one INR test were reported to range from $6.19 to $145.7014. These studies did not explore the potential cost burden associated with out-of-range INR values. A review published between 1995 and 2012 included 28 economic models involving warfarin; however, the impact of warfarin was not assessed in 40% of the models22. When the impact of warfarin control was involved, the models only accounted for costs associated with stroke and bleeding events based on TTR. The current analysis acknowledges that most patients have warfarin fluctuations, which account for the overall cost burden of out-of-range INRs and have contributed a new aspect in cost evaluation in warfarin therapy.

According to existing literature, the average patient prescribed warfarin spends a distressingly high proportion of time either over or under the guideline-recommended INR range. In a meta-analysis that assessed eight studies involving 22,237 warfarin-treated AF patients, the overall AF patients spent a mean 55% (95% CI = 51–58%) of their time within the therapeutic INR range23. Patients in the out-of-range INR category have an increased risk of complications, leading to higher healthcare costs and utilization24.

In another study, anti-coagulation services run by nurses led to an average yearly net cost savings of $241,400 per 100 patients25. The current study results suggest that out-of-range INR values could be a marker of higher cost burden. Future studies are necessary to determine whether the cost burden is modifiable by better INR management or alternative therapeutic options. Recent studies have shown that patients who use the new oral anti-coagulants are at lower risk of developing intracerebral hemorrhage26. In comparison to warfarin, they have fewer drug–drug and drug–food interactions, resulting in fewer INR fluctuations. These factors may offset the higher drug costs of the newer agents.

The findings from the current study also demonstrate that 30-day healthcare costs were higher when out-of-range INR test results occurred within the first 2–6 months of warfarin therapy compared with after 6 months. This could be due to patients’ and clinicians’ familiarity with warfarin therapy, contributing to less urgent action after an out-of-range INR result. The impact of this phenomenon is unknown, because less action may indicate either a lower quality of care or more judicious use of healthcare resources. Another reason could be survival bias; healthier patients remained in the patient population pool longer, leading to lower costs as follow-up periods were extended.

The current study was subject to certain limitations. First, although claims data are extremely valuable for the efficient and effective examination of healthcare outcomes, treatment patterns, and costs, claims data are collected for the purpose of payment and not research. Second, the presence of a claim for a filled prescription does not indicate that the medication was consumed or that it was taken as prescribed. In the current analysis, the presence of INR testing and refill data of warfarin increased the certainty that warfarin was indeed consumed. Third, diagnostic codes in claims data may contain inaccuracies or omissions. These inaccuracies are expected to occur randomly and are unlikely to have significantly impacted the findings. Finally, a general limitation of claims data analysis is that only observable factors were used, and there may be residual confounding because of unmeasured clinical and disease-specific parameters.

The results obtained using the VHA claims database may not be generalizable to other populations with NVAF. In particular, the VHA population contains very little information regarding female patients, and veterans may have additional or different risk factors compared with the general population. Conclusions drawn from the current study should be cautiously interpreted, and additional research with data from other populations is needed.

Conclusions

Exposure to out-of-range INR values was associated with higher follow-up healthcare costs. Costs associated with sub-therapeutic INR (INR < 2) were especially high. Previous literature has suggested that negative events are more likely to occur when INR values are below or above the normal INR target range; this work suggests that costs are also significantly higher as INR values exit the target range. As INR fluctuation among warfarin patients is common, these findings highlight the substantial need to evaluate the long-term quality of INR management and to take action when markers of cost burden emerge. Furthermore, although such action is needed when INR values are out-of-range, it may be especially justified when INR values are sub-therapeutic.

Transparency

Declaration of funding

The current study was sponsored by Janssen Scientific Affairs, LLC.

Declaration of financial/other relationships

LW and OB are employees of STATinMED Research, a paid consultant to Janssen Scientific Affairs, LLC.

Acknowledgments

The authors thank Sulena Shrestha, MPH, Sara Cosgrove, BA, and Elizabeth M. Moran of STATinMED Research for assistance with the writing and editing of this manuscript. The authors would like to acknowledge Lisa Grauer, MSc, of Chameleon Communications International, who provided editorial support, with funds provided by Janssen Scientific Affairs, LLC.

Appendix

Codes used to identify warfarin initiator and comorbidities.

Diseases/condition	Codes (ICD-9 or CPT)
Transient perioperative AF diseases	36.10–36.19, 37.10–37.12, 37.31–37.33 or 37.40, 35.00–35.04, 35.31–35.39, 35.41–35.42, 35.50–35.56, 35.60–35.63 or 35.70–35.73
Mitral or aortic valve repair or replacement	394.0, 394.2, 396.0, 396.1, 396.8, V43.3, V42.2, 35.10, 35.14, 35.20, 35.28
Hyperthyroidism	242.0–242.9
Pregnancy	630–648
Cesarean delivery	74.xx
Vaginal delivery	72.xx, 73.xx, 75.xx, v27, 650–659
Congestive heart failure	428.xx
Diabetes mellitus	250.xx, 249.xx
Hypertension	401.xx, 405.xx
Ischemic stroke	433.xx, 434.01, 434.11, 434.91, and 437.1
Transient ischemic attack	435.xx
Vascular disease	410.xx, 411.1x, 411.8x, 414.xx, 443.9
End-stage renal disease	ICD-9-CM procedure: V42.0, V56.0, V56.8, 39.95, 54.98, and 55.69 CPT: 90921, 90925, 90935–90947, 99512, 99559, 50360, 50365
Hospitalized hemorrhagic stroke	430.xx, 431.xx and 432.xx, 852.xx, 853.xx, 854.xx accompanied by 438.xx during the same hospitalization
Chronic kidney disease	403.xx, 404.xx, 585.xx, 586.xx, 588.xx
Chronic obstructive pulmonary disease	491.2x, 492.xx, 518.1, 518.2
Anemia	280.xx, 281.xx, 282.xx, 283.xx, 284.xx, 285.xx