Anemia in Inflammatory Bowel Disease: Practice Patterns and Clinical Outcomes in a Large National Veterans Cohort

Abstract

Background

Anemia is a common complication of inflammatory bowel disease (IBD), impacting patient quality of life and clinical outcomes. This study examines anemia management practices and clinical outcomes in a national cohort of veterans with IBD.

Methods

A retrospective cohort study of 89 687 veterans with IBD from 2000 to 2017 was conducted using Veterans Health Administration data. We assessed anemia screening, iron store assessments, and iron therapy. Primary outcomes included anemia resolution within 12 months. Secondary outcomes included IBD-related hospitalizations, steroid use, and surgery within five years. Multivariate logistic regression models assessed associations between patient characteristics and clinical outcomes.

Results

Among 89 687 VA patients with IBD, 56% experienced anemia, but only 36% received iron testing. Only 23% of patients with iron deficiency received iron therapy within 6 months. Hemoglobin normalization occurred in 19% within a year. IBD patients with comorbidities had lower resolution rates (1–2 comorbidities: OR 0.89). Over five years, 30% were hospitalized, with lower rates for nonsmokers (OR 0.81) and higher for those with comorbidities (OR 1.10) or on biologics (OR 1.91). Steroid use was 17.5%, higher in those with comorbidities (OR 1.20) or on biologics (OR 4.11), and lower in Black patients (OR 0.78). Only 2% had surgery, less common among Black patients (OR 0.69) and non-smokers (OR 0.72).

Conclusions

Anemia is underdiagnosed and undertreated in veterans with IBD, leading to poor clinical outcomes. Enhanced screening and treatment protocols are essential to improve anemia resolution rates and reduce IBD-related complications.

Graphical Abstract

Graphical Abstract.

Introduction

Anemia is one of the most prevalent complications affecting patients with inflammatory bowel disease (IBD), which includes Crohn’s disease (CD) and ulcerative colitis (UC).¹ Anemia frequently exacerbates the clinical course of IBD patients, exerting a profound influence on both the management and prognosis of the disease. Given its significant impact on health outcomes, a thorough understanding of its prevalence, management strategies, and clinical implications in diverse patient populations is essential for optimizing care and improving patient quality of life.^2,3

Recent studies report anemia prevalence rates in IBD range from 29% to 52%.^4,5 Both CD and UC patients are vulnerable to iron deficiency anemia and anemia of chronic disease, both of which significantly impair quality of life through reduced work capacity, cognitive function, exercise tolerance, and increased fatigue.⁶

Until 2017, comprehensive guidelines for anemia management in IBD, such as the Anemia Care Pathway issued by the Crohn’s and Colitis Foundation, were lacking; however, significant gaps in practical implementation still persist.⁷ These discrepancies in adherence often result in suboptimal patient outcomes. Reasons for this include a lack of awareness or prioritization of anemia in IBD patients among healthcare providers, inconsistent application of screening protocols, and variations in treatment approaches. Additionally, economic, and systemic barriers may hinder the effective implementation of these guidelines, further complicating the clinical management of these patients. Such gaps not only exacerbate the burden of disease but also elevate the overall healthcare costs associated with IBD, underscoring the critical need for targeted research to bridge these divides in care.⁸

This study aims to address these gaps by investigating the real-world practice patterns of anemia management in IBD patients within the Veterans Health Administration (VHA). We sought to further understand anemia screening, iron store assessment, and iron therapy in IBD patients with anemia in addition to rates of anemia resolution. Secondly, we investigated various demographic and clinical variables associated with the outcomes of hospitalization, steroid prescription, and surgery in IBD patients with anemia.

Materials and Methods

Study Design and Patient Population

We performed a retrospective cohort study on veterans diagnosed with IBD from 2000 to 2017 receiving care in the Veterans Affairs (VA) health system. IBD diagnosis was identified using International Classification of Diseases (ICD-9/10) codes for CD (K50) and UC (K51) with >/= 2 inpatient or outpatient encounters in the VA, at least one being an outpatient visit.⁹We utilized data from the Veterans Health Administration Corporate Data Warehouse (CDW). Patients with codes for both CD and UC were classified as indeterminate colitis. We relied on previously validated search strategies, which show an 83% positive predictive value for CD and 89% for UC.⁹ The study was approved by the Institutional Review Board.

Covariates

We included demographic and clinical variables including age, gender, race/ethnicity, smoking status, type of IBD (UC/CD), Charlson Comorbidity Index, US geographic region, and IBD medication class. Additionally, we considered VA priority levels as potential modifiers. The VA Priority Groups classify veterans into eight levels based on factors like service-connected disabilities, income, and special circumstances, determining their access to healthcare benefits and potential costs, with lower-numbered groups receiving the highest benefits and priority for care.

Study Outcomes

Anemia screening, presence of anemia, iron stores, and treatment of anemia in IBD patients were assessed. Screening of anemia was defined by having a hemoglobin assessed. Presence of anemia was defined as count of anemia cases where hemoglobin levels were below 13 g/dL for males and 12 g/dL for females.⁷ Iron stores were defined as assessment of ferritin and iron saturation within 6 months of anemia hemoglobin value. Patients were defined as having inadequate iron stores if the ferritin was less than 100 ng/ml or total iron saturation was less than 20% within months of confirmed anemia, as defined by the Crohn’s and Colitis Anemia Care Pathway^9,7.13. Iron treatment included patients who received a prescription for oral iron supplementation (ferrous sulfate or ferrous gluconate) or IV iron (ferric gluconate, iron dextran, or iron sucrose) within 6 months of either Anemia confirmation (lab) or Iron Store assessment (lab).

The primary outcome included resolution of anemia within 12 months of anemia diagnosis. Secondary outcomes included IBD-related hospitalizations, steroid use, or IBD-related surgery within 5 years of anemia diagnosis. Steroid prescription included both budesonide and prednisone. Surgery included the operations of a total or hemi-colectomy with end ileostomy, ileal pouch-anal anastomosis, or segmental bowel resection.

Statistical Analysis

Descriptive statistics were presented as proportions for categorical variables and means with standard deviation (SD) of continuous and ordinal variables. Anemia screening, iron deficiency, iron stores, and iron therapy were reported. Univariate and multiple logistic models were fit for the primary outcome resolution of anemia, and secondary outcomes hospitalization, steroid use, and surgery within a 5-year period of index. Data analysis was conducted using SAS^® (9.4) statistical software.

Results

IBD Anemia Practice Patterns

A total of 89 687 VA patients with IBD were identified from 2000 to 2017. 50 781 (56%) of these VA IBD patients were found to have anemia at least once during the study period (Table 1). Only 36% of anemic patients had iron labs within 6 months of the index anemia date. The prevalence rate of anemia screening was approximately 80% throughout the study period. Of those with iron labs, 89% of anemic patients had iron deficient anemia, with only 23% of patients with iron deficiency received iron therapy within 6 months of iron labs or iron store assessment (Figure 1). The majority of IBD patients with anemia had a hemoglobin of 10 or greater (Figure 2).

Table 1.

Cohort characteristics for IBD patients with anemia.

Patient characteristics	Frequency (%)
Age (years) Mean (±SD)	71.4 (±14.1)
Gender Male Female	47 247 (93.0) 3534(7.0)
Race White African American Others Unknown	38 032 (74.9) 5482 (10.8) 750 (1.5) 6517 (12.8)
Ethnicity Non-Hispanic/Latino Hispanic/Latino Unknown	43 804 (86.3) 1687 (3.3) 5290 (10.4)
Smoking status Current Former Never Unknown	8019 (15.8) 13 177 (26.0) 12 011 (23.6) 17 574 (34.6)
Comorbidity No comorbidity 1-2 >=3	26 124(51.4) 16 506(32.5) 8151(16.1)
Encounters*2 Mean (±SD)	26.8(±30.0)
Priority group 7–8 1–6	11 658 (23.0) 39 123 (77.0)
Region Pacific Continental Midwest North Atlantic Southeast	8084 (15.9) 7302 (14.4) 12 059 (23.8) 13 464 (26.5) 9872 (19.4)
Academic affiliation No Yes	2860 (5.6) 47 921 (94.4)
IBD type Crohn’s disease Ulcerative Colitis Indeterminate	22,185 (43.7) 27 439 (54.0) 1157 (2.3)
Concomitant IBD medication * 1
ASA alone Biologics + immunomodulator All others	14 622(28.8) 2509(4.9) 33650(66.3)

*

¹±3 months of Index Anemia;

*

²Within 1 year of Index IBD.

Figure 1.

Presence of anemia: Defined as male Hb <13 g/dL, female Hb <12 g/dL.

Figure 2.

Number of IBD patient with anemia by hemoglobin level.

Resolution of Anemia

Within one year of the index anemia date, only 19% of anemic IBD patients had their hemoglobin return to the normal range. Patients with higher hemoglobin were more likely to see anemia resolutions and patients with more severe anemia were more likely to receive supplementation (Figure 3). In comparison to active smokers, both former smokers and never-smokers had a decreased likelihood of resolution of anemia (adjusted OR 0.90 (0.84–0.97) and 0.93 (0.86–0.99), respectively). When patients with no comorbidities were compared to those with 1–2 comorbidities, the adjusted odds ratio (OR) was 0.89 (0.85–0.94). For patients with three or more comorbidities, the adjusted OR was 0.60 (0.56–0.65) (unadjusted-Table 2 / adjusted-Table 3). Patients with UC were more likely to have resolution of anemia when compared with those with CD (adjusted OR 1.13 (1.08–1.18), When compared with patients receiving 5-ASA alone, those receiving biologics ± immunomodulators had decreased likelihood of anemia resolution.

Figure 3.

Percent of Patients that received Iron supplementation and anemia resolution by hemoglobin level.

Table 2.

Unadjusted bivariate models for IBD patients with anemia.

	Unadjusted logistic models: OR (95%CI)
Patient characteristics	Resolution of anemia	Hospitalization within 5 years	Steroid within 5 years	Surgery within 5 years
Age (years) Mean (±SD)	0.99(0.99–0.99)	0.97(0.97–0.97)	0.97(0.97–0.97)	0.97(0.97–0.97)
Gender Male Female	Ref 1.31(1.21–1.43)	Ref 1.23(1.15–1.33)	Ref 1.52(1.40–1.65)	Ref 0.81(0.60–1.09)
Race White African American Others Unknown	Ref 0.84(0.78–0.91) 0.89(0.74–1.08) 0.75(0.69–0.81)	Ref 1.23(1.16–1.30) 1.08(0.93–1.26) 0.70(0.65–0.74)	Ref 0.98(0.91–1.05) 1.19(0.99–1.42) 0.40(0.37–0.44)	Ref 0.83(0.65–1.05) 0.69(0.35–1.33) 0.48(0.37–0.63)
Ethnicity Non–Hispanic/Latino Hispanic/Latino Unknown	Ref 1.00(0.88–1.13) 0.72(0.66–0.78)	Ref 1.16(1.05–1.29) 0.71(0.67–0.76)	Ref 1.32(1.18–1.49) 0.31(0.27–0.34)	1.22(0.86–1.73) 0.49(0.36–0.66)
Smoking status Current Former Never Unknown	Ref 0.96(0.89–1.03) 1.05(0.97–1.12) 0.86(0.80–0.92)	Ref 0.53(0.50–0.56) 0.60(0.56–0.63) 0.82(0.77–0.86)	Ref 0.78(0.73–0.83) 0.80(0.75–0.86) 0.53(0.50–0.57)	Ref 0.54(0.44–0.66) 0.55(0.45–0.68) 0.62(0.52–0.75)
Comorbidity No Comorbidity 1–2 >=3	Ref 0.81(0.77–0.85) 0.48(0.45–0.52)	Ref 1.05(1.01–1.10) 1.16(1.10–1.22)	Ref 1.07(1.02–1.13) 1.07(1.01–1.14)	Ref 0.89(0.76–1.04) 1.08(0.89–1.30)
Encounters*2 Mean(±SD)	0.99(0.99–0.99)	1.01(1.01–1.02)	1.01(1.01–1.01)	1.01(1.01–1.01)
Priority group 7–8 1–6	Ref 0.93(0.89–0.98)	Ref 2.08(1.98–2.19)	Ref 1.55(1.46–1.65)	Ref 1.68(1.38–2.04)
Region Pacific Continental Midwest North Atlantic Southeast	Ref 1.12(1.04–1.22) 0.97(0.90–1.04) 1.00(0.93–1.07) 1.05(0.97–1.13)	Ref 0.81(0.76–0.87) 0.81(0.77–0.86) 0.69(0.65–0.73) 0.78(0.73–0.83)	Ref 0.94(0.87–1.02) 1.00(0.93–1.08) 0.85(0.79–0.91) 0.90(0.83–0.97)	Ref 1.06(0.84–1.33) 0.95(0.77–1.17) 0.74(0.59–0.91) 0.56(0.44–0.72)
Academic affiliation No Yes	Ref 0.84(0.77–0.92)	Ref 1.74(1.58–1.91)	Ref 1.10(0.99–1.21)	Ref 1.28(0.91–1.79)
IBD type Crohn’s disease Ulcerative colitis Indeterminate	Ref 1.13(1.08–1.18) 0.85(0.72–0.99)	Ref 0.54(0.52–0.56) 1.03(0.91–1.17)	Ref 0.81(0.77–0.85) 1.05(0.90–1.21)	Ref 0.84(0.73–0.96) 1.42(0.96–2.09)
Concomitant IBD medication*1
ASA alone Biologics + immunomodulator All others	Ref 0.83(0.75–0.93) 0.83(0.79–0.88)	Ref 3.64(3.34–3.98) 1.14(1.09–1.19)	Ref 6.74(6.12–7.41) 2.04(1.92–2.16)	Ref 5.01(3.86–6.49) 1.93(1.59–2.34)

*

¹±3 months of Index Anemia;

*

²Within 1 year of Index IBD.

Table 3:

Adjusted multivariate logistic models for IBD patients with anemia

	Adjusted Logistic Models: OR (95%CI)
Patient Characteristics	Resolution of Anemia	Hospitalization within 5 years	Steroid within 5 years	Surgery within 5 years
Age (years) Mean (±SD) Median(IQR)	0.99(0.99–0.99)	0.98(0.97–0.98)	0.98(0.98–0.98)	0.98(0.97–0.98)
Gender Male Female	Ref 1.34(1.23–1.46)	Ref 0.79(0.73–0.85)	Ref 1.04(0.95–1.13)	Ref 0.54(0.40–0.73)
Race White African American Others Unknown	Ref 0.82(0.76–0.89) 0.87(0.72–1.06) 0.84(0.76–0.92)	Ref 0.93(0.88–0.99) 0.91(0.77–1.07) 0.79(0.72–0.86)	Ref 0.78(0.72–0.84) 1.03(0.85–1.23) 0.71(0.63–0.79)	Ref 0.69(0.53–0.85) 0.57(0.29–1.11) 0.61(0.47–0.95)
Ethnicity Hispanic/Latino Non–Hispanic/Latino Unknown	Ref 1.03(0.91–1.16) 0.79(0.71–0.88)	Ref 0.98(0.88–1.09) 0.92(0.84–1.01)	Ref 1.11(0.98–1.26) 0.46(0.40–0.52)	Ref 1.02(0.72–1.47) 0.75(0.51–1.11)
Smoking Status Current Former Never Unknown	Ref 0.90(0.84–0.97) 0.93(0.86–0.99) 0.81(0.76–0.87)	Ref 0.83(0.78–0.89) 0.81(0.76–0.86) 1.19(1.12–1.26)	Ref 1.13(1.05–1.22) 1.01(0.94–1.09) 0.73(0.68–0.78)	Ref 0.77(0.62–0.96) 0.72(0.58–0.89) 0.86(0.71–1.04)
Comorbidity No Comorbidity 1–2 >=3	Ref 0.89(0.85–0.94) 0.60(0.56–0.65)	Ref 1.10(1.05–1.16) 1.06(0.99–1.13)	Ref 1.20(1.13–1.27) 1.07(0.99–1.15)	Ref 0.90(0.76–1.07) 0.88(0.71–1.09)
Encounters*2 Mean(±SD)	0.99(0.99–0.99)	1.01(1.01–1.01)	1.01(1.01–1.01)	1.01(1.01–1.01)
Priority Group 7–8 1–6	Ref 1.05(0.99–1.11)	Ref 1.47(1.39–1.55)	Ref 1.08(1.01–1.15)	Ref 1.08(0.89–1.32)
Region Pacific Continental Midwest North Atlantic Southeast	Ref 1.07(0.99–1.16) 0.92(0.85–0.99) 0.95(0.89–1.02) 1.01(0.94–1.10)	Ref 0.79(0.74–0.85) 0.89(0.84–0.95) 0.76(0.71–0.81) 0.80(0.75–0.85)	Ref 0.94(0.86–1.02) 1.05(0.97–1.14) 0.90(0.83–0.97) 0.96(0.88–1.04)	Ref 1.09(0.86–1.38) 1.03(0.83–1.27) 0.84(0.67–1.04) 0.61(0.47–0.79)
Academic Affiliation No Yes	Ref 0.88(0.80–0.96)	Ref 1.59(1.44–1.75)	Ref 1.02(0.92–1.04)	Ref 1.21(0.86–1.71)
IBD type Crohn’s disease Ulcerative Colitis Indeterminate	Ref 1.13(1.08–1.18) 0.97(0.82–1.14)	Ref 0.61(0.58–0.63) 0.92(0.81–1.04)	Ref 0.95(0.90–0.99) 1.00(0.86–1.17)	Ref 1.04(0.90–1.21) 1.30(0.87–1.92)
Concomitant IBD medication*1
ASA alone Biologics +/–Immunomodulator All others	Ref 0.84(0.75–0.94) 0.87(0.83–0.92)	Ref 1.91(1.73–2.09) 0.89(0.85–0.93)	Ref 4.11(3.72–4.55) 1.77(1.66–1.88)	Ref 2.76(2.10–3.63) 1.56(1.28–1.90)

*

¹±3 months of Index Anemia;

*

²Within 1 year of Index IBD.

Hospitalization Within 5 Years of Index Anemia

Within five years of the index anemia date, 30% of patients with anemia were hospitalized, with 60% of these hospitalizations occurring within the first year. Table 3 demonstrates adjusted models. Both former smokers and never-smokers showed a decreased hospitalization risk compared to current smokers, with adjusted ORs of 0.83(0.78–0.89) and 0.81(0.7–0.86), respectively. When compared with patients with no-comorbidity, those with 1–2 comorbidities had higher odds of hospitalization (1.10(1.05–1.16)). UC patients had a significantly lower rate of hospitalization than those with CD, with an adjusted OR of 0.61 (0.58–0.63). Patients on both biologics ± immunomodulators had a higher likelihood of hospitalization compared to those on 5-ASA alone, with an adjusted OR of 1.91 (1.73–2.09).

Steroid Within 5 Years of Index Anemia

Within five years of anemia diagnosis, 17.5% of all IBD patients with anemia received a steroid prescription, either prednisone or budesonide. Black patients were less likely to receive a steroid prescription within this timeframe, with an adjusted odds ratio (OR) of 0.78 (0.72–0.84). Patients with 1–2 comorbidities faced a significantly higher probability of being prescribed steroids compared to those with no comorbidities, with an adjusted OR of 1.20 (1.13–1.27). Patients who were on biologics ± immunomodulators had a significantly higher odds of requiring a steroid prescription compared to those on 5-ASA alone, with an adjusted OR of 4.11 (3.72–4.55).

Surgery Within 5 Years of Index anemia

Within five years of the index anemia date, only 2% of patients underwent IBD-associated surgery. Black patients were less likely to have surgery within this period, with an adjusted odds ratio (OR) of 0.69 (0.53–0.85). Both former smokers and never-smokers had a lower likelihood of needing surgery compared to current smokers, with adjusted ORs of 0.77 (0.62–0.96) and 0.72 (0.58–0.89), respectively. Patients who were on biologics ± immunomodulators had an adjusted OR of 2.76 (2.10–3.63) for surgery compared to those on 5-ASA alone.

Discussion

Our study identified 89 687 VA patients with inflammatory bowel disease (IBD) from 2000 to 2017, with 56% experiencing anemia at least once. Anemia screening was robust at 80%, yet only 36% of those diagnosed with anemia underwent labs for assessment of iron stores within six months of detection. Among those, the majority (89%) were found to have iron deficiency anemia, but only 23% received iron therapy within the subsequent six months. Anemia resolution within a year was low at 19% and inversely related to the number of comorbidities and interestingly higher in smokers. Hospitalization within 5 years post-anemia diagnosis occurred in 30% of patients, highly associated with smoking status, VA priority group, IBD medication class, and racial demographics. Additionally, within the study period, 17% of IBD patients with anemia required steroids, and only 2% underwent IBD-related surgery, with both interventions showing notable variations based on smoking status, comorbidities, and medication regimen.

In our study, 56% of IBD patients were found to be anemic; however, only a minority (36%) of these patients underwent subsequent testing for iron stores. This prevalence of anemia and the breakdown between iron deficiency anemia (IDA) and anemia of chronic disease (ACD) aligns with findings from the literature.¹⁰ A study involving 1394 Greek IBD patients showed robust iron testing; in fact, serum ferritin levels were measured in 60% of patients, and 78.9% of those diagnosed with anemia received treatment.¹¹ Although these treatment rates were higher than what we observed, a smaller study of VA IBD patients reported that 37% received oral iron, while only 2.8% were treated with IV iron, reflecting more similar treatment patterns to our findings.¹² The low rate of anemia resolution in our study was likely driven by the inadequate treatment of IDA at only 23%.

We found that Black patients had lower rates of anemia resolution, steroid prescriptions, and hospitalization risk compared to White patients. Given the reported increased severity of IBD in Black, it is plausible that racial disparities contributed significantly to these findings.^13,14 Although to a modest degree, we found that smokers had higher odds of anemia resolution than non-smokers. Interestingly, this is consistent with prior data showing that smoking is inversely associated with anemia risk.¹⁵ Perhaps this can be ascribed to secondary polycythemia and increased erythropoietin production in smokers.¹⁶ However, IBD patients who smoked with anemia had significantly higher rates of hospitalization and surgery within 5 years of index anemia date, consistent with the literature associating smoking with a more severe Crohn’s disease phenotype¹⁷ Charlson Comorbidity index of 1–2 was associated with lower odds of resolution of anemia and higher odds of hospitalization. The strong association noted between biologic therapy with anemia resolution and clinical outcomes compared to 5-ASA was likely driven by the confounder of IBD disease activity and severity, which is not fully captured within the retrospective VA database.

Our study faces several limitations that impact the interpretation and applicability of the results. The demographic composition of the Veterans Health Administration (VHA) population, predominantly male (93%), may reduce the generalizability of our findings to female IBD patients. Additionally, the retrospective cohort design is subject to inherent risks of selection and confounder biases. Capturing disease activity and severity within this dataset poses challenges, potentially confounding many of our secondary outcomes. In particular, the effect of iron supplementation on anemia resolution is difficult to isolate, as anemia improvement is likely driven by overall disease control. This limits our ability to assess the independent impact of supplementation on anemia outcomes. Additionally, our study may underrepresent iron supplementation as over the counter oral supplementation is not commonly reported in the electronic medical record. Despite these challenges, our study also has significant strengths. It leverages the comprehensive national VHA database, which provides a robust dataset from electronic medical records across all VHA facilities nationwide. This extensive access enables a large study population enhancing the generalizability and statistical power of our findings.

Anemia remains a significantly underdiagnosed and undertreated condition in patients with Crohn’s disease (CD) and ulcerative colitis (UC), substantially affecting quality of life and associated with an increased risk of adverse IBD-related outcomes. This study highlights critical gaps in anemia management within the VHA, suggesting that enhanced clinical protocols and targeted policy reforms are necessary to standardize care, improve anemia resolution rates, and reduce the substantial healthcare burden associated with inflammatory bowel disease. By addressing these gaps, healthcare providers can improve management strategies and ultimately enhance health outcomes for all IBD patients.

Author Contributions

J.S.: conceptualization, writing-original draft, writing-review & editing. C.B.: conceptualization, writing-original draft, writing-review & editing. S.S.: methodology, formal analysis, writing-review & editing. S.S.: writing-review & editing. R.S.: writing-review & editing. J.H.: conceptualization, methodology, formal analysis, writing-review & editing, funding acquisition.

Funding

This project was supported with a grant from American Regent. American Regent was not directly involved with the analyses, interpretation of data, or writing of this manuscript. JH is supported by a grant from VA Health Services Research and Development (IIR 19-045). JH is also supported in part with resources at the VA HSR&D Center for Innovations in Quality, Effectiveness and Safety (#CIN 13-413), at the Michael E. DeBakey VA Medical Center, Houston, TX.

Conflicts of Interest

J.S.: none. C.B.: none. S.S.: none. S.S.ub: none. R.S. none. J.H. has received research funding from Redhill Biosciences, Janssen, AbbVie, Celgene, Genentech, Bristol Myer Squib, Eli-Lilly, Lycera, and Pfizer Inc.

Data Availability

The data underlying this article cannot be shared publicly due to VA policy. The data will be shared on reasonable request to the corresponding author.