Validation of an Algorithm to Identify Venous Thromboembolism in Health Insurance Claims Data Among Patients with Rheumatoid Arthritis

Figures & data

Table 1 Summary Description of Study Cohorts

	ORD Cohort 1	ORD Cohort 2	MC Cohort
Data sources	ORD linked patient-level charts	ORD linked patient-level charts	MC database matched-EHR provider clinical notes
Insurance type	Commercial and Medicare	Commercial and Medicare	Commercial, Medicare, and Medicaid
Identification window	July 1, 2016, to October 30, 2019	May 1, 2016, to November 30, 2020	May 1, 2016, to August 31, 2020
Index RA therapy	TNFi^a or JAKi^b	TNFi^a, JAKi^b, or non-TNFi^c	TNFi^a, JAKi^b, or non-TNFi^c
Specific exclusion criteria	Not new RA therapy user^d Not eligible for re-identification^e	Not eligible for re-identification^e	No matched-EHR provider notes^f
VTE algorithm applied	Primary, secondary	Primary	Primary
Study objective	Primary^g, secondary	Primary^g	Exploratory

Notes: ^aTNFi therapy included adalimumab, certolizumab pegol, etanercept, golimumab, or infliximab. ^bJAKi therapy included baricitinib, tofacitinib, or upadacitinib. ^cNon-TNFi therapy included abatacept, anakinra, tocilizumab, or sarilumab. ^dPatients with ≥1 claim for a TNFi therapy or a JAKi therapy during the 6-month CAW were excluded only from ORD Cohort 1. ^ePatients without patient and provider identifiable information available to support medical chart identification and abstraction. ^fPatients with provider(s) who had contributed free-text clinical notes within the 3-month CAW in the Optum-matched EHR database that could be extracted and redacted for abstraction. ^gPrimary study objective was applied to patients identified from ORD Cohort 1 and ORD Cohort 2 who had a presumptive VTE defined using the primary case definition (primary objective sample).

Abbreviations: CAW, covariate assessment window; EHR, electronic health record; JAKi, Janus kinase inhibitor; MC, Market Clarity; ORD, Optum Research Database; RA, rheumatoid arthritis; TNFi, tumor necrosis factor inhibitor; VTE, venous thromboembolism.

Figure 1 Study design.

Notes: Exclusion criteria are applied to each patient and arranged in event time anchored at study entry [Day 0] and qualifying event [Day ED]. ^aThe index date was the date of either the first observed claim for RA therapy or the first claim for a JAKi (even if a claim for a TNFi was observed first), prioritizing baricitinib, followed by tofacitinib and then upadacitinib. ^bExclusion of patients with ≥1 claim for a TNFi or a JAKi during the 6-month exclusion window was only applied to Optum Research Database Cohort 1. ^cExclusion of patients with ≥1 claim for AC therapy prescribed before the VTE qualifying event and with a runout date that overlaps with the VTE qualifying ED. ^dCensor occurred at either disenrollment, date of presumptive VTE [Day ED], or end of study period (December 31, 2020), leaving the last 31 days for evaluation of claims for AC therapy (part of presumptive VTE identification).

Abbreviations: AC, anticoagulant; ED, event date; JAKi, Janus kinase inhibitor; RA, rheumatoid arthritis; TNFi, tumor necrosis factor inhibitor; VTE, venous thromboembolism.

Table 2 VTE Case Definition Criteria

Primary Algorithm	Secondary Algorithm
PE diagnosis Hospital or ER 1. Primary position 2. Secondary position + AC therapy^a	PE diagnosisHospital or ER 1. Primary position 2. Secondary positionOutpatient 3. Any position + AC therapy^a
DVT diagnosis Hospital or ER 1. Lower in primary position 2. Upper in primary position + AC therapy^a 3. Lower in secondary position + AC therapy^aOutpatient 4. Lower in any position + AC therapy^a 5. Upper in any position + AC therapy^a	DVT diagnosis Hospital or ER 1. Lower in primary position 2. Lower in secondary position 3. Upper in primary position 4. Upper in secondary position + AC therapy^aOutpatient 5. Lower in any position^a 6. Upper in any position + AC therapy^a
Other VT Any setting Any position + AC therapy^a	Other VT Hospital or ER 1. Primary position 2. Secondary position + AC therapy^a Outpatient  Any position + AC therapy^a

Notes: ^aAnticoagulant therapy included was defined as having a claim for apixaban, dabigatran, dalteparin, edoxaban, enoxaparin, fondaparinux, rivaroxaban, tinzaparin, or warfarin within 31 days of the presumptive VTE.

Abbreviations: AC, anticoagulant; DVT, deep vein thromboembolism; ER, emergency room; PE, pulmonary embolism; VT, venous thrombosis; VTE, venous thromboembolism.

Table 3 Patient Demographic and Clinical Characteristics by Study Sample

Sample, Cohort Source	Primary Objective, ORD Cohort 1+2 (n = 155)	Secondary Objective, ORD Cohort 1 (n = 76)	Exploratory Objective, MC Cohort (n = 64)
Claims-based variables
Age, y	66.4 (10.7)	64.6 (11.4)	60.2 (12.5)
Age ≥65 years	97 (62.6)	38 (50.0)	24 (37.5)
Female	114 (73.5)	52 (68.4)	48 (75.0)
Non-commercial insurance	125 (80.6)	52 (68.4)	32 (50.0)^a,b
Charlson comorbidity score	2.5 (1.7)	2.6 (2.1)	1.9 (1.2)
Follow-up time, d	912.0 (524.7)	588.3 (341.6)	1134.1 (466.0)
Time to qualifying VTE, d	481.8 (417.6)	279.1 (259.1)	545.9 (454.7)
RA therapy during follow-up
TNFi: adalimumab	47 (30.3)	26 (34.2)	17 (26.6)
Certolizumab	7 (4.5)	6 (7.9)	2 (3.1)
Etanercept	29 (18.7)	11 (14.5)	18 (28.1)
Golimumab SC	2 (1.3)	2 (2.6)	2 (3.1)
Golimumab IV	11 (7.1)	6 (7.9)	7 (10.9)
infliximab	20 (12.9)	8 (10.5)	14 (21.9)
JAKi: baricitinib	1 (0.6)	0	0
Tofacitinib	32 (20.6)	23 (30.3)	10 (15.6)
Upadacitinib	0	1 (1.3)	0
Non-TNFi: abatacept	20 (17.1)^c	NR	10 (15.6)
Anakinra	1 (0.9)^c		0
Sarilumab	3 (2.6)^c		0
Tocilizumab	28 (23.9)^c		9 (14.1)
Charts- or notes-based comorbidities
Evidence of prior VTE	44 (28.4)	21 (27.6)	18 (28.1)
Atrial fibrillation	25 (16.1)	10 (13.2)	6 (9.4)
COPD	42 (27.1)	21 (27.6)	9 (14.1)
Hypertension	112 (72.3)	51 (67.1)	37 (57.8)
Cancer	12 (7.7)	5 (6.6)	6 (9.4)
Obese (BMI ≥30 kg/m^2)^b	66 (46.8)	35 (50.0)	36 (67.9)
Ever smoker^b	82 (55.8)	36 (56.3)	29 (54.7)

Notes: Data represent mean (SD) for continuous variables and n (%) for categorical variables. ^aIncluded 20 (31.25%) covered under Medicare and 12 (18.75%) under Medicaid. ^bDenominator varies accounting for missing data. ^cClaims for non-TNFi were not reported in ORD Cohort 1, so the denominator is 117.

Abbreviations: BMI, body mass index; COPD, chronic obstructive pulmonary disease; IV, intravenous; JAKi, Janus kinase inhibitor; MC, Market Clarity; NR, not reported; ORD, Optum Research Database; RA, rheumatoid arthritis; SC, subcutaneous; SD, standard deviation; TNFi, tumor necrosis factor inhibitor; VTE, venous thromboembolism.

Table 4 Performance of VTE Case Definitions

Sample	Primary Objective			Secondary Objective	Exploratory Objective
Cohort Source	ORD Cohort 1+2 (n = 155)	ORD Cohort 1 (n = 38)	ORD Cohort 2 (n = 117)	ORD Cohort 1 (n = 76)	MC Cohort (n = 64)
False positive, n (%)	38 (24.5)	9 (23.7)	29 (24.8)	36 (47.4)	28 (43.8)
True positive, n (%)	117 (75.5)	29 (76.3)	88 (75.2)	40 (52.6)	36 (56.3)
PPV, % (95% CI)	75.5 (68.7, 82.3)	76.3 (62.8, 89.8)	75.2 (67.4, 83.0)	52.6 (41.4, 63.9)	56.3 (44.1, 68.4)

Abbreviations: CI, confidence interval; MC, Market Clarity; ORD, Optum Research Database; PPV, positive predictive value; VTE, venous thromboembolism.

Table 5 Characteristics of Confirmed VTEs

Sample, Cohort Source	Primary Objective, ORD Cohort 1+2 (n = 155)	Secondary Objective, ORD Cohort 1 (n = 76)	Exploratory Objective, MC Cohort (n = 64)
Total number of patients with confirmed VTEs	117	40	36
Patients with 1 VTE, n (%)	87 (56.1)	26 (34.2)	29 (45.3)
Patients with 2 VTE, n (%)	30 (19.4)	14 (18.4)	7 (10.9)
VTE type, n (%)
PE	67 (43.2)	27 (35.5)	17 (26.6)
Lower extremity DVT	75 (48.4)	27 (35.5)	19 (29.7)
Upper extremity DVT	1 (0.6)	0	3 (4.7)
Other VT	4 (2.6)	0	4 (6.3)
Both PE and lower extremity DVT	NR	14 (18.4)	NR

Abbreviations: DVT, deep vein thromboembolism; MC, Market Clarity; NR, not reported; ORD, Optum Research Database; PE, pulmonary embolism; VT, venous thrombosis; VTE, venous thromboembolism.