Abstract
Context
Veterans Health Administration (VHA) maintained a registry of identified and verified cases of US Veterans with spinal cord injuries and disorders (SCI/D) since 1994: VHA SCI/D Registry (VHA SCIDR). Data elements, capture, and storage methods varied over time.
Objective
Describe the consolidation and harmonization of historical VHA SCIDR data spanning three decades during its evolution to an automated platform and report population characteristics.
Methods
The VHA SCIDR captured data using four distinct acquisition methods over 28 years, including cases of Veterans with SCI/D receiving SCI/D System of Care services, via 25 SCI/D Centers and 122 Spoke Sites throughout the VHA healthcare system. Foundational elements of VHA SCIDR data capture methods, harmonization of data elements with the current automated algorithm, access protocol, and governance structure are described.
Results
From Fiscal Years (FYs) 1994 to 2022, VHA SCIDR identified 52,407 Veterans with traumatic or non-traumatic SCI/D, and 96.95% were male, 56.09% White, 16.57% were Black, 1.23% Asian and Pacific Islander, 0.75% Native American, and 25.36% unknown. Traumatic etiology comprised 53.39% of the sample, while 31.75% were non-traumatic, with 14.87% missing etiology classification. Injury category proportions were 5.19% high tetraplegia, 5.83% low tetraplegia, 5.85% high paraplegia, 7.53% low paraplegia, and 23.35% AIS D, with 52.25% missing or unable to be calculated.
Conclusions
VHA SCIDR is one of the three largest SCI/D registries in North America and is the case-identification platform for VHA SCI/D operations, program evaluation, and research studies. VHA SCIDR is connected to each Veteran’s VHA healthcare data, facilitating big data research.
KEYWORDS: Spinal cord injuries, Registries, Veterans, United States, Population health
Introduction
The United States (US) Department of Veterans Affairs (VA), Veterans Health Administration (VHA) has maintained an updated list of verified cases of US Veterans with spinal cord injuries and disorders (SCI/D) since 1994. Over time, the VHA SCI/D Registry (VHA SCIDR) data capture methods have evolved to a recently modernized algorithm and data platform.1 The current VHA SCIDR data capture product is automated, connected with the VHA Enterprise Corporate Data Warehouse (CDW), updated monthly, and facilitates big data analytics, system improvement, and VHA clinical care.1 This manuscript includes an overview of the evolution of VHA SCIDR, a description of curating and harmonization efforts, and highlighted features and functionality useful to program evaluators and researchers.
The VHA SCIDR algorithm identifies new cases of Veterans with SCI/D by SCI/D diagnoses and their receipt of SCI/D-specific VHA healthcare.1 SCI/D-specific care is provided through the VHA SCI/D System of Care, a comprehensive, lifelong healthcare service and coordination system spanning the US with a SCI/D Hub and Spoke framework.2 By linking individual cases to the CDW, discrete common data elements, as well as a Veteran’s electronic medical record (EMR), VHA SCIDR is an integrating source linking all data created by a Veteran’s use of VHA care. With the volume of data in VHA SCIDR and being linked to data of various sources and types, VHA SCIDR is a facilitator of big data that can be used to support a variety of population characteristics, health process measures, and outcomes.3
VHA SCIDR was incepted in fiscal year (FY) 1994, fully implemented by FY1996, and evolved through several data capture iterations (Fig. 1).1,4,5 Initially, data were manually entered into an EMR-auxiliary application termed the Spinal Cord Dysfunction (SCD) Registry, and locally updated by individual VHA SCI/D Centers.1,4,5 SCD protocols resulted in nonstandardized entry, varied degrees of entry completeness, and duplicative entries between sites when Veterans visited more than one SCI/D Hub or Spoke.1,4,5 To replace SCD and facilitate consistent VHA SCIDR data entry, the VHA SCI/D National Program Office implemented the Spinal Cord Injuries and Disorders Outcomes (SCIDO) platform and trained informatics staff [Management of Information and Outcomes (MIO) Coordinators].1,4 SCIDO supported clinical care and validated a Veteran’s identity through the CDW, an administrative and clinical database framework termed Veterans Health Information Systems and Technology Architecture (VISTA), further fostering multi-level and multi-type registry data.4,5
Figure 1.
Evolution of the VHA SCI/D registry by data capture methods. Time of use of different VHA SCIDR data capture methods (A), and resulting case additions by Veteran qualifying fiscal year (B). Initial data capture was via SCD, involving manual input from a specialized data entry form used by SCI/D clinicians, case managers, SCI/D social workers, and from surveys sent to Veterans identified through SCI/D ICD-9 and administrative SCI/D bed section codes. Note that large numbers of cases added by Veteran qualifying FYs 1995 &1996 (B) are attributed to many cases having SCI/D onset years prior to registry initiation. Subsequent predominant data capture was via SCIDO, supported by dedicated & trained informatics staff at SCI/D Centers performing chart review and secondary entry of registry cases and outcomes data in standardized Excel spreadsheets. With the decommission of SCIDO on 9/30/2015, informatics staff continued chart reviews and spreadsheet maintenance, with data periodically aggregated by the SCI/D National Office. Testing of a new automated algorithm began in FY 2018 using algorithm-based extraction sets from the EMR and VSSC data compared to Excel spreadsheet (chart reviewed) data. In FY 2019, the current VSSC algorithm was implemented VHA-wide retrospectively back to FY 2013, capturing cases at earlier timepoints (or cases otherwise missed) than by previously used capture methods (B). Note the overlapping periods of use of different data capture methods (A). All years are in FYs. Abbreviations: EMR: Electronic Medical Record, FY: fiscal year, SCD: Spinal Cord Dysfunction, SCIDO: Spinal Cord Injury and Disorders Outcomes, VSSC: VHA Service Support Center.
SCIDO performed limited automated data extraction from the EMR, but because SCIDO was not directly integrated into the EMR, data still required chart review and secondary manual data entry.1,4 Additionally, between-site differences in VHA SCIDR case identification persisted due to differences in data entry (influenced by clinical processes and local operational needs), use of separate local data spreadsheets, and technical issues preventing direct data aggregation from different servers at the national level.1,4 SCIDO was decommissioned on 9/30/2015 and replaced with standardized Microsoft Excel spreadsheets locally-maintained at SCI/D Centers.
Development of a new EMR-linked and CDW-integrated algorithm began in April of 2018 as a partnered effort between the VHA SCI/D National Program Office and the VHA Service Support Center (VSSC).1 The new algorithm was designed with input from VHA SCI/D subject matter experts to capture Veterans qualifying for inclusion based on diagnoses and use of SCI/D services starting Oct 1, 2012. (This start date was chosen due to earlier start dates returning less reliable data elements.) Integrating the algorithm with the CDW further enhanced automation, national standardization, and data validity. In 2018, this VSSC algorithm was tested by MIO Coordinators and programmers for validity against aggregated VHA SCIDR Microsoft Excel spreadsheets maintained by individual SCI/D Centers. VHA SCIDR’s modernized platform was implemented in 2019 as the current modernized data capture system.1
VHA SCIDR inclusion criteria has been consistent with Veteran eligibility for the VHA Spinal Cord Injuries and Disorders System of Care (outlined in Appendix A of VHA Directive 1176).2 Any Veteran eligible for VHA SCI/D specialty services throughout the years, according to these criteria, has also been eligible for VHA SCIDR inclusion. From its inception, VHA SCIDR has captured patient identifiers (name, US social security number, date of birth), etiology (inclusive of both traumatic and non-traumatic conditions), neurological level of injury (NLI), impairment, and onset/injury date. Through VHA SCIDR’s evolution using different data capture methods (SCD, SCIDO, Microsoft Excel, VSSC), core data elements have conceptually remained the same. This paper describes consolidation and harmonization of VHA SCIDR historical data, integration of fragmented archival data with new data elements from the automated algorithm, and the current VHA SCIDR status as a big-data case-identification platform. For the first time, 28 years of comprehensive epidemiological VHA SCIDR data are presented. Foundational elements of data capture methods, consolidation steps, harmonization, access protocol, and governance structure are described.
Methods
The VHA SCIDR historical registry consolidation effort identified, standardized, and merged data for Veterans with SCI/D from multiple sources of archived data from past data capture methods with the current primary registry source, the VSSC SCI/D Registry algorithm platform.1 The key data points included in this consolidation were patient identifiers, SCI/D etiology, SCI/D date of onset, and injury category [categorized from NLI and severity by the American Spinal Injury Association (ASIA) Impairment Score (AIS)]. As early as 2011, injury category has been shared by the United States SCI Model Systems (US SCIMS) and International SCI Data Set (ISCIDS),6–8 and employed in recent utilization9 and outcomes10 VHA evaluations. Registry cases include Veterans engaged with VHA SCI/D System of Care Services that live with traumatic spinal cord injuries or non-traumatic spinal cord disorders, which include multiple sclerosis with spinal cord involvement and motor neuron disorders with spinal cord involvement. Data with missing or invalid record dates were excluded from the data set. The consolidated data tables included information merged from each historical data source (SCD, SCIDO, Microsoft Excel, and VSSC), as shown in Fig. 1. Examples of consolidated data tables and data definitions are available in Appendix A.
There are many cases in the consolidated data set where the Veteran had multiple entries. We constructed this data set so that the end-user can decide, for Veterans with more than one record for any of the data points, whether to use the earliest entered value, the most recently entered value, the most frequently entered value, or all values. Some Veterans only had one value entered for a particular data point. In order to establish a one-Veteran, one-record data set for the population epidemiological data, a best fit logic and resulting Structured Query Language (SQL) code algorithm was established. The algorithm prioritized selections based on completeness of the record, date of the record entry, and frequency (single or dominant entry over time).
After the one-Veteran, one-record data set was established, additional data for each Veteran were extracted from the VHA CDW: race, ethnicity, and service era. Age at SCI/D onset was calculated using the Veteran’s date of birth and SCI/D onset date. Injury category was determined using the Veteran’s NLI and AIS (definitions for injury categories are displayed in Table 1). In cases where more than one NLI was documented (e.g. from a change in NLI over time), the highest NLI was used for injury categorization. Traumatic or non-traumatic etiology status was categorized from detailed etiology data (see VHA SCIDR Etiology table in Appendix A).
Table 1.
VHA SCIDR SCI/D* Demographics and characteristics: living cohort snapshots and all cases FY1994 to FY2022.
| Characteristic | 1996 Living Cohort (n = 12,874) Percent (frequency) |
2008 Living Cohort (n = 19,140) Percent (frequency) |
2022 Living Cohort (n = 21,287) Percent (frequency) |
1994–2022 Total SCI/D* Cases (n = 52,407) Percent (frequency) |
|---|---|---|---|---|
| Sex | ||||
| Male | 98.00% (12,613) |
96.74% (18,516) |
94.96% (20,214) |
96.95% (50,806) |
| Female | 2.03% (261) |
3.26% (624) |
5.04% (1,073) |
3.05% (1,600) |
| Unknown | 0% (1) |
0% (1) |
0% (1) |
0% (1) |
| Race | ||||
| Asian | 0.20% (26) |
0.47% (90) |
0.83% (177) |
0.49% (255) |
| Black/African American | 10.48% (1,349) |
17.32% (3,315) |
21.64% (4,607) |
16.57% (8,685) |
| Native American | 0.49% (63) |
0.83% (158) |
0.99% (211) |
0.75% (392) |
| Native Hawaiian/Pacific Islander | 0.44% (57) |
0.89% (170) |
1.08% (230) |
0.74% (390) |
| White | 44.65% (5,748) |
67.94% (13,004) |
67.00% (14,263) |
56.09% (29,393) |
| Missng/Declined | 43.74% (5,631) |
12.55% (2,403) |
8.45% (1,799) |
25.36% (13,292) |
| Ethnicity | ||||
| Hispanic | 2.61% (36) |
4.47% (856) |
5.93% (1,263) |
4.17% (2,185) |
| Not Hispanic | 52.03% (6,698) |
85.31% (16,328) |
87.53% (18,362) |
70.71% (37,058) |
| Missing | 45.36% (5,840) |
10.22% (1,956) |
6.54% (1,392) |
25.12% (13,164) |
| Average SCI/D Onset Age | 39.55 years SD = 17.46 (12,847 cases used) |
42.16 years SD = 17.44 (19,086 cases used) |
46.25 years SD = 17.86 (19,984 cases used) |
49.30 years SD = 18.87 (49,527 cases used) |
| Average SCI/D Onset Age Missing | 0.002% (n = 27) |
0.003% (n = 54) |
0.061% (n = 1,303) |
0.055% (n = 2,880) |
| Injury Category | ||||
| Tetraplegia High (C1-C4 + AIS A-C) |
4.15% (534) |
6.08% (1,163) |
6.53% (1,389) |
5.19% (2,721) |
| Tetraplegia Low (C5-C8 + AIS A-C) |
8.22% (1,058) |
9.00% (1,722) |
6.84% (1,456) |
5.83% (3,055) |
| Paraplegia High (T1-T6 + AIS A-C) |
6.76% (870) |
8.57% (1,641) |
7.94% (1,690) |
5.85% (3,065) |
| Paraplegia Low (T7-S3 + AIS A-C) |
8.02% (1,032) |
11.09% (2,123) |
11.41% (2,428) |
7.53% (3,948) |
| AIS D (AIS D + any NLI) |
5.95% (766) |
19.42% (3,717) |
40.12% (8,540) |
23.35% (12,237) |
| Missing | 66.91% (8,614) |
45.84% (8,774) |
27.17% (5,784) |
52.25% (27,381) |
| Etiology | ||||
| Non-traumatic | 13.87% (1,785) |
24.99% (4,783) |
36.60% (7,790) |
31.75% (16,637) |
| Arthritic Disease/Columnar Degeneration | 2.92% (376) |
7.41% (1,419) |
18.70% (3,980) |
11.93% (6,254) |
| Autoimmune or Inflamatory | 0.13% (17) |
0.47% (90) |
1.34% (285) |
0.64% (336) |
| Genetic | 0.00% (0) |
0.10% (20) |
0.33% (70) |
0.16% (86) |
| Infection/Abscess | 1.15% (148) |
2.33% (446) |
2.92% (621) |
2.82% (1,476) |
| Tumor | 1.38% (178) |
1.95% (373) |
2.59% (552) |
2.50% (1,310) |
| Vascular Change | 0.18% (23) |
29.11% (5,572) |
2.28% (485) |
1.52% (796) |
| Syringomyelia | 0.54% (69) |
0.77% (147) |
0.93% (199) |
0.70% (369) |
| Other Non-traumatic | 7.57% (947) |
11.01% (2,108) |
7.51% (1,598) |
11.47% (6,010) |
| Traumatic | 72.84% (9,377) |
66.24% (12,678) |
55.17% (11,745) |
53.39% (27,979) |
| Fall | 13.09% (1.679) |
14.48% (2,772) |
15.89% (3,382) |
15.05% (7,889) |
| Sports | 5.64% (726) |
4.72% (903) |
3.93% (836) |
3.29% (1,726) |
| Vehicular | 31.52% (4,058) |
29.11% (5,572) |
22.84% (4,863) |
20.93% (10,970) |
| Violence | 11.52% (1,483) |
8.60% (1,646) |
5.56% (1,184) |
6.18% (3,237) |
| Other Traumatic | 11.07% (1,425) |
9.33% (1,785) |
6.95% (1,480) |
7.93% (4,157) |
| Missing | 13.3% (1,712) |
8.77% (1,879) |
8.23% (1,752) |
14.87% (7,791) |
| Period of Service | ||||
| Persian Gulf | 4.68% (603) |
13.23% (2,532) |
29.32% (6,241) |
14.03% (7,353) |
| Post-Vietnam | 11.51% (1,482) |
14.50% (2,776) |
20.20% (4,299) |
12.31% (6,453) |
| Vietnam Era | 43.32% (5,577) |
51.46% (9,849) |
45.90% (9,771) |
46.31% (24,272) |
| Post-Korean | 5.46% (703) |
4.23% (809) |
1.34% (286) |
4.32% (2,263) |
| Korean | 15.03% 1,935 |
9.34% (1,788) |
2.02% (429) |
10.46% (5,482) |
| Pre-Korean | 0.30% (39) |
0.15% (28) |
0.01% (2) |
0.23% (118) |
| World War II | 19.05% (2,452) |
6.39% (1,224) |
0.23% (50) |
11.49% (6,022) |
| World War I | 0.02% (3) |
0.00% (0) |
0.00% (0) |
0.02% (11) |
| Spanish American | 0.01% (1) |
0.01% (2) |
0.00% (0) |
0.00% (2) |
| Non-Veteran | 0.60% (77) |
0.61% (117) |
0.78% (167) |
0.70% (369) |
| Missing | 0.02% (2) |
0.08% (15) |
0.20% (42) |
0.12% (62) |
To describe the demographic and SCI/D characteristics of Veterans included in VHA SCIDR, we used frequencies and percentages. Living cohorts (marked in time by FY) were determined based on Veterans living one or more days during the FY.
Results
From fiscal years (FYs) 1994 to 2022 (October 1, 1993 to September 30, 2022), VHA SCIDR identified 73,947 total Veterans. This included 14,430 Veterans with multiple sclerosis (MS) with spinal cord involvement, 6,441 Veterans with motor neuron disease (MN) with spinal cord involvement, and 52,407 Veterans with traumatic or non-traumatic SCI/D (excluding the etiologies of MS and MN). While some Veterans with MS and MN are included in VHA SCIDR due to having spinal cord involvement in their disease process and receiving SCI/D System of Care services, they are excluded in the present results. We use the term SCI/D* to designate Veterans with traumatic and non-traumatic SCI/D excluding MS and MN etiologies. There were 669 additional cases that were presumed SCI/D*, but not used in the reporting of SCI/D characteristics because the elements of etiology, onset date, and injury category data were missing. Excluding the volume of new cases identified during seeding efforts in 1995–1996, Fig. 1(B) shows there was an average of 2,162 new cases per fiscal year (range: 948 to 2805) from 1994 to 2022.
Living cohort changes over time
Figure 2 shows the VHA SCIDR living cohort volume and average age over time by fiscal year (FY) for Veterans with SCI/D*. The living cohorts increased each FY until FY2019, when there was a slight decrease from more Veterans passing away than new cases added. Beginning in FY2013 when VSSC living cohorts can be identified from retroactive implementation of the VSSC algorithm, the yellow portions in Fig. 2 show VSSC-identified living cases, and the remaining blue portions show cases that were not identified by the VSSC algorithm but had been identified through SCD, SCIDO, or Microsoft Excel. Of note, the number of non-VSSC-identified living cases from FY2013 to FY2022 has decreased over time, and the average age of all living cohorts has increased over time.
Figure 2.
VHA SCIDR living SCI/D* cohort (excluding MS and MN) and living cohort average age over time. The VSSC algorithm was retroactively implemented back to FY2013. Living Veterans with SCI/D* identified by the VSSC capture method are shown in yellow, and Veterans with SCI/D* not identified by the VSSC capture method are in blue. Note that remaining living cases in SCIDR not identified by the VSSC algorithm decreased over time. Also note that the average age of living Veterans in SCIDR with SCI/D* increased over time (green line).
Demographics
Of the 52,407 Veterans with SCI/D*, overall 96.95% were male (Table 1). There were no missing sex data. While only 2.03% of the living cohort was female in 1996, the percentage of female Veterans with SCI/D increased to 3.26% in 2008 and 5.04% in 2022. Overall, 56.09% were White, 16.37% Black or African American, 1.23% Asian and Pacific Islander, 0.75% Native American, and 25.37% unknown or declined to answer. The percentage of Veterans who were Black or African American increased from 10.48% in 1996 to 21.64% in 2022; some of this increase could be accounted for by the substantial decrease in missing race data in VHA SCIDR from 1996–2022 (43.74% to 8.45%). This correlated to overall VHA (including non-SCI/D) missing race data improving from 41.3% in FY2005 to 8% in FY2021.11 The majority of Veterans were not Hispanic (70.71%), 4.17% were Hispanic, and 25.12% of Veterans had missing ethnicity data or declined to answer. Over time, the percentage of Hispanic Veterans increased to 5.93% in 2022, while the percentage with missing ethnicity data decreased from 45.36% in 1996 to 6.54% in 2022. Of all VHA SCIDR cases since FY1994, 46.31% are classified as Vietnam Era Service Members. For the 2022 living cohort, 45.90% Veterans served during the Vietnam Era (Table 1). Essentially all cases are Veterans, with only 369/52,407 (0.70%) being non-Veteran cases that could include eligible spouses, dependents, or active duty service members.
SCI/D characteristics
The overall group of VHA SCIDR cases with SCI/D* showed 56.72% percent were of traumatic etiology, 35% non-traumatic, and 8% were missing etiology classification. The percentage of Veterans with missing etiology data decreased from 13.30% to 8.23% from 1994 to 2022. Figure 3 shows snapshots of the distributions of SCI/D* etiologies for Veterans in SCIDR living in FY 1996, 2008, 2022, and the overall composite distribution for FYs 1994–2022 living cohorts. Non-traumatic SCI/D* cases increased over time. Injury category proportions were 5.13% high tetraplegia, 5.76% low tetraplegia, 5.78% high paraplegia, 7.44% low paraplegia, and 23.07% AIS D. NLI and AIS information were insufficient to calculate injury category for 52.83% of cases overall, but since automation of these data from the electronic medical record (EMR) in May of 2020, newer cases were much more likely to have these data. While 66.91% of Veterans were missing NLI and AIS information in 1996, only 27.17% of Veterans were missing information required for injury category in 2022. The overall mean age at onset of injury/disorder was 49.30 (SD = 18.87).
Figure 3.
Etiology category snapshots for living Veterans with SCI/D* in VHA SCIDR (excluding MS and MN) for: 1996, 2008, 2022, and composite (living and deceased, n = 52,407) 1994–2022.
Discussion
From 1994 to 2022, VHA SCIDR identified 52,407 Veterans with traumatic or non-traumatic SCI/D. Living cohorts grew until FY2019, when Veterans passing away started exceeding new cases added. VHA SCIDR living cohort average age and proportion of non-traumatic cases have continually increased over time. The proportions of female, black, and Hispanic Veterans in VHA SCIDR living cohorts have also increased over time, though some of the race and ethnicity changes could be attributed to reductions in overall missing VHA race and ethnicity data over time. The amount of missing etiology and injury category data has decreased over time, with injury category missing for 66.91% of living cases in 1996 and only 27.17% of living cases in 2022.
The VHA SCIDR has evolved to a data capture method that is automated, nationally-standardized, and integrated into the EMR and CDW. It consolidates and harmonizes new, automated data with historic, archived data and facilitates big-data research. VHA SCIDR data provides meaningful information to guide operations and policy, enable research collaborations, and develop analytic partnerships across both Veteran and non-Veteran populations. Compiling this valid, reliable dataset is a major step toward analytic models that can be used to enhance access, evaluate programs and services, and improve the overall quality of life and outcomes for Veterans with SCI/D.
VHA SCIDR changes over time
The increases in VHA SCIDR living cohort volume slowed over time and have begun to show signs of decline, consistent with the cohort’s overall increasing age. Cases of non-traumatic SCI/D* have been increasing over time and these cases tend to be older, likely contributing to the VHA SCIDR living cohorts’ increasing average ages. Referral and enrollment of non-traumatic SCI/D* cases may not have occurred as frequently as for traumatic SCI/D* cases in the remote past, and may have been uneven across different etiologies. Notably, prevalence in the VHA SCIDR population differs from incidence, since life expectancy is shorter with non-traumatic etiologies, even after controlling for age of onset.12 Missing injury category data for living cohorts has decreased over time, particularly with the automation of NLI and AIS within electronic health record clinical templates in May of 2020. With clinician uptake of utilizing the template for Veterans living with SCI/D between then and September 30, 2022, the amount of missing injury category and other critical SCI/D data significantly decreased.
VHA SCIDR and North American data sets
VHA SCIDR is one of the three largest SCI/D registries in North America.1,13–16 Compared with non-Veteran SCI/D databases and registries, VHA SCIDR has several key distinguishing characteristics (Table 2). (1) VHA SCIDR includes non-traumatic SCI/D, including multiple sclerosis affecting the spinal cord, and motor neuron disease with spinal cord involvement. This differs from the US Spinal Cord Injury Model Systems (SCIMS) database, which has exclusively focused on traumatic SCI4 and the Canadian Rick Hansen SCI Registry,13,15,16 which only began including non-traumatic SCI/D in 2020. VHA SCIDR also includes diagnostic category and etiology variables that permit users to isolate desired SCI/D subpopulations. (2) VHA SCIDR case capture is independent of where and when individuals received acute post-SCI rehabilitation. VHA SCIDR cases are added through healthcare data specifying SCI/D diagnoses and VA SCI/D-specific utilization, even if Veterans previously sought non-VA healthcare (including acute inpatient rehabilitation). Thus, a Veteran can be added at any point in their lifetime after acquiring SCI/D–even decades later. (3) Updating VHA SCIDR is continuous, automated, and involves varied data types. Updates includes demographics changes, EMR-embedded diagnoses, and VHA SCI/D specialty care utilization over the Veteran’s lifetime. This contrasts to other North American registries’ periodic acquisition (usually by phone) of rigorously standardized limited follow-up data. (4) Pre-SCI/D healthcare data are often available. Some Veterans have documented VA care prior to their SCI/D. This scenario contrasts to non-Veteran SCI/D datasets utilizing acute inpatient SCI/D rehabilitation as the registry/database standardized data initiation point. (5) Essentially all cases are US Veterans. Non-Veteran cases (less than 1%) include active military members and dependents or spouses of Veterans who receive VA healthcare; these can be identified for exclusion by cross-reference with relevant CDW data.
Table 2.
Distinguishing comparisons of three large North American SCI/D data sets: VHA SCIDR, US SCIMS, and RHSCIR.
| Registry/Database | SCI/D Etiology | Case Capture Timing and Source | Updating | Size (in 2022) |
Features |
|---|---|---|---|---|---|
| Veterans Health Administration Spinal Cord Injuries and Disorders Registry (VHA SCIDR) | Traumatic, non-traumatic | Acute or chronic; Veterans receiving 2 instances of SCI/D-specific care | Automated, continuous – monthly for new cases, nightly for VHA healthcare data | 73,956 total; 52,461 SCI/D-only, excluding MN and MS |
|
| United States Spinal Cord Injury Model Systems (SCIMS) National SCI Database | Traumatic | New SCI and admission to a SCIMS acute rehabilitation center | 1 and 5 year post-injury phone call / questionnaire, then every 5 years | 51,2277 | |
| Rick Hansen Spinal Cord Injury Registry (RHSCIR) | Traumatic, non-traumatic as of 20198,10 | New SCI and admission to a RHSCIR acute trauma center / rehabilitation facility | 18 months, 5 year post-injury phone call / questionnaire, then every 5 years (commenced 2020) | >10,60011 |
VHA SCIDR strengths
VHA SCIDR has robust case identification with inclusion criteria that have remained consistent for three decades. A Veteran presenting to the SCI/D System of Care and determined to have a SCI/D condition as specified by VHA Directive 11762 would be enrolled in SCI/D services, receive specialty SCI/D care, and be eligible for VHA SCIDR entry. The VSSC algorithm mimics the condition and SCI/D specialty services utilization criteria by searching the CDW for ICD diagnoses and SCI/D specialty care access data points. VHA SCIDR epidemiologic data can be matched with other data, such as healthcare utilization and EMR clinical and health elements. Although VHA SCIDR data collection and management methods changed over time, data entry and tracking has consistently included primary data from EMRs for Veterans whose eligibility for registry inclusion was established using standard assessments of injury and function (NLI and AIS). Harmonization of VHA SCIDR data over time strengthens the validity of the dataset by using redundancies and oversampling of data from multiple sources and exact patient matching.17,18 The consistency and uniformity of VHA SCIDR data will ideally facilitate collaboration and matching with community SCI/D databases to increase the generalizability of results from VHA studies to non-Veteran US and international populations. VHA SCIDR can be leveraged to evaluate outcomes, inform policy, and ultimately improve the care and health for Veterans with SCI/D.19
VHA SCIDR limitations
One limitation of VHA SCIDR is its exclusive focus on Veterans, who make up a significant but declining percentage of the US population.20 The focus on US Veterans limits the number of individuals enrolled in VHA SCIDR each year and restricts availability of the data to only VHA staff and investigators. To balance these constraints, VHA leaders and researchers have established collaborations with the US SCIMS to design joint analyses of data and align VHA SCIDR with other world SCI databases and research. Over 90% of Veterans in VHA SCIDR are also male, which could limit generalizability of results to other SCI/D populations. This trend may represent restricted ranges for some variables and affect the robustness of statistical analyses or the types of inferences that can be drawn from the data. Additionally, VHA SCIDR identifies only Veterans who have received VHA SCI/D System of Care services. While specific NLI and AIS data are available for individual cases, VHA SCIDR does not capture Subaxial Cervical Spine Injury Classification21 or Thoracolumbar Injury Classification and Severity Scores.21,22
Future directions
Future research is needed to determine if Veterans with SCI/D in VHA SCIDR (a) differ in significant ways from non-Veterans within or outside the US or (b) differ substantially from Veterans with SCI/D who have not accessed VHA services (most recently identified from FY2013 forward distinguishing between VSSC and non-VSSC identified VHA SCIDR cases). Based on the use case, future researchers could consider using VSSC cases [October 1, 2012 (FY2013) to present] versus all historical cases as early as FY1994. Historical cases that contain more critical data points and less missing data provide teams more assurance of true SCI/D cases. VSSC use is recommended for end-users valuing specificity, and all historical cases are recommended for end-users placing higher value on sensitivity. Plans are in place to address missing data (especially injury category) using natural language processing, and to conduct analyses to examine healthcare utilization and outcomes.
Access and governance
VHA SCIDR was developed and is maintained by the VHA SCI/D National Program Office. VHA researchers and program evaluators request VHA SCIDR access through the VA Information Resource Center (VIReC), the VA Informatics and Computing Infrastructure (VINCI) and the VA Data Access Request Tracker (DART). VA and VHA national policies restrict Veteran-level data access to key VHA personnel with appropriate clearances. Non-VHA employees who collaborate with VHA leaders and researchers on projects and initiatives are able to access VHA SCIDR data after they obtain proper security and privacy background checks for VHA data access. To increase access, national VHA policies and procedures are currently being examined to determine the feasibility of making de-identified, aggregate data sets publicly available to non-VHA researchers and stakeholders. These changes are under consideration in order to support future research using the VHA SCIDR to collaborate and harmonize data for comparisons of outcomes and access among individuals who are not receiving services in the VHA SCI/D System of Care. Future partnerships could include using VHA SCIDR data to unify and combine VHA and US Department of Defense SCI/D registries consistent with the VA Electronic Health Record Modernization initiative.23
Conclusions
The consolidation and harmonization of VHA SCIDR data and evolution of methods described in this paper will support increased consistency in the reporting of VHA SCIDR data. VHA SCIDR data from 1994 to present is valid and accessible to VHA employees and other key personnel. The VHA SCIDR resource can now be more easily leveraged for program evaluation, research, and partnered analyses with other world SCI data sets.
Supplementary Material
Funding Statement
Data to Improve Veterans’ Outcomes (DIVO) in SCI/D. Research to Impact for VeteRans (RIVRs) Program (RVR 19-474). Veterans Health Administration (VHA) Health Services Research & Development (HSR&D). Direct Costs: $500,000. 7/1/2019 to 6/30/2024. U.S. Department of Veterans Affairs.
Supplemental data
Supplemental data for this article can be accessed online at https://doi.org/10.1080/10790268.2024.2434305.
Disclaimer statements
Contributors None.
Declaration of interest None.
Conflicts of interest Authors have no conflict of interests to declare.
References
- 1.Sippel JL, Daly JE, Poggensee L, Ristau KD, Eberhart AC, Tam K, Evans CT, Lancaster B, Wickremasinghe IM, Burns SP, et al. Modernization of a large spinal cord injuries and disorders registry: the Veterans Administration experience. Arch Rehabil Res Clin Transl. 2022 Oct 17;4(4):100237. doi: 10.1016/j.arrct.2022.100237. PMID: 36545529; PMCID: PMC9761267. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.VHA Directive 1176(2): Spinal Cord Injuries and Disorders System of Care. United States Department of Veterans Affairs; 2020 [accessed 2024 May 29]. https://www.va.gov/vhapublications/ViewPublication.asp?pub_ID=8523. [Google Scholar]
- 3.Gandomi A, Haider M.. Beyond the hype: big data concepts, methods, and analytics. Int J Inf Manag. 2015 Apr;35(2):137–144. ISSN 0268-4012. doi: 10.1016/j.ijinfomgt.2014.10.007. [DOI] [Google Scholar]
- 4.Charlifue S, Tate D, Biering-Sorensen F, Burns S, Chen Y, Chun S, Jakeman LB, Kowalski RG, Noonan VK, Ullrich P.. Harmonization of databases: a step for advancing the knowledge about spinal cord injury. Arch Phys Med Rehabil. 2016 Oct;97(10):1805–18. doi: 10.1016/j.apmr.2016.03.030. Epub 2016 Apr 30. PMID: 27137095. [DOI] [PubMed] [Google Scholar]
- 5.Smith BM, Evans CT, Ullrich P, Burns S, Guihan M, Miskevics S, LaVela SL, Rajan S, Weaver FM.. Using VA data for research in persons with spinal cord injuries and disorders: lessons from SCI QUERI. J Rehabil Res Dev. 2010;47(8):679–88. doi: 10.1682/jrrd.2009.08.0117. PMID: 21110243. [DOI] [PubMed] [Google Scholar]
- 6.Chen Y, Wen H, Baidwan NK, DeVivo MJ.. Demographic and health profiles of people living with traumatic spinal cord injury in the United States during 2015–2019: findings from the Spinal Cord Injury Model Systems database. Arch Phys Med Rehabil. 2022 Apr;103(4):622–633. doi: 10.1016/j.apmr.2021.11.001. Epub 2021 Nov 20. Erratum in: Arch Phys Med Rehabil. 2023 Apr;104(4):710. doi:10.1016/j.apmr.2022.12.193. PMID: 34808122. [DOI] [PubMed] [Google Scholar]
- 7.Biering-Sørensen F, DeVivo MJ, Charlifue S, Chen Y, New PW, Noonan V, Post MWM, Vogel L.. International Spinal Cord Injury Core Data Set (version 2.0)-including standardization of reporting. Spinal Cord. 2017 Aug;55(8):759–764. doi: 10.1038/sc.2017.59. Epub 2017 May 30. PMID: 28555665. [DOI] [PubMed] [Google Scholar]
- 8.DeVivo MJ, Biering-Sørensen F, New P, Chen Y.. International Spinal Cord Injury Data Set. Standardization of data analysis and reporting of results from the International Spinal Cord Injury Core Data Set. Spinal Cord. 2011 May;49(5):596–9. doi: 10.1038/sc.2010.172. Epub 2010 Dec 7. PMID: 21135863. [DOI] [PubMed] [Google Scholar]
- 9.Sippel J, Etingen B, Stroupe K, Evans C, Huo Z, Frazier B, Wickremasinghe M, Smith B.. United States Veterans’ utilization of spinal cord injuries and disorders annual evaluation services. Arch Phys Med Rehabil. 2024 Sep 27: S0003-9993(24)01259-0. doi: 10.1016/j.apmr.2024.09.008. Epub ahead of print. PMID: 39343049. [DOI] [PubMed] [Google Scholar]
- 10.Burns SP, Eberhart AC, Sippel JL, Wilson GM, Evans CT.. Case-fatality with coronavirus disease 2019 (COVID-19) in United States Veterans with spinal cord injuries and disorders. Spinal Cord. 2020 Sep;58(9):1040–1041. doi: 10.1038/s41393-020-0529-0. Epub 2020 Jul 27. PMID: 32719529; PMCID: PMC7383660. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Mor M. Assessing race and ethnicity in VA data. VA Information Resource Center (VIReC) Database and Methods Seminar. April 3, 2023. Virtual national presentation. Publicly available recording link: https://www.hsrd.research.va.gov/for_researchers/cyber_seminars/archives/video_archive.cfm?SessionID=5271&Seriesid=22.
- 12.Hatch BB, Wood-Wentz CM, Therneau TM, Walker MG, Payne JM, Reeves RK.. Factors predictive of survival and estimated years of life lost in the decade following nontraumatic and traumatic spinal cord injury. Spinal Cord. 2017 Jun;55(6):540–544. doi: 10.1038/sc.2016.182. Epub 2017 Feb 7. PMID: 28169294. [DOI] [PubMed] [Google Scholar]
- 13.Chen Y, Mehta T, Sippel JL, Burns S, Noonan VK, Pierce D.. Advancing spinal cord injury research through data harmonization and integrative analysis across large data sets. 2022 ACRM annual conference abstracts. Arch Phys Med Rehabil. 2022;103:e45. [Google Scholar]
- 14.Chen Y, DeVivo MJ, Richards JS, SanAgustin TB.. Spinal Cord Injury Model Systems: review of program and national database from 1970 to 2015. Arch Phys Med Rehabil. 2016 Oct;97(10):1797–804. doi: 10.1016/j.apmr.2016.02.027. PMID: 27671806. [DOI] [PubMed] [Google Scholar]
- 15.Praxis Spinal Cord Institute . Rick Hansen Spinal Cord Injury Registry – a look at spinal cord injury in Canada in 2020. Vancouver (BC: ): Praxis Spinal Cord Institute; 2022; praxisinstitute.org/wp-content/uploads/2022/11/RHSCIR-2020-Report-final.pdf. [Google Scholar]
- 16.Kelly-Hedrik M, Abd-El-Barr MM, Aarabi B, Curt A, Howley SP, Harrop JS, Kirshblum S, Neal CJ, Noonan V, Park C, et al. Importance of prospective registries and clinical research networks in the evolution of spinal cord injury care. J Neurotrauma. 2023 Sep;40(17-18):1834–1848. doi: 10.1089/neu.2022.0450. PMID: 36576020. [DOI] [PubMed] [Google Scholar]
- 17.Wilkinson MD, Dumontier M, Jan Aalbersberg I, Appleton G, Axton M, Baak A, Blomberg N, Boiten JW, da Silva Santos LB, Bourne PE, et al. Addendum: the FAIR guiding principles for scientific data management and stewardship. Sci Data. 2019 Mar 19;6(1):6. doi: 10.1038/s41597-019-0009-6. Erratum for: Sci Data. 2016 Mar 15;3:160018. doi:10.1038/sdata.2016.18. PMID: 30890711; PMCID: PMC6427092. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Corley DA, Feigelson HS, Lieu TA, McGlynn EA.. Building data infrastructure to evaluate and improve quality: PCORnet. J Oncol Pract. 2015 May;11(3):204–6. doi: 10.1200/JOP.2014.003194. PMID: 25980016; PMCID: PMC4438109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Nelson EC, Dixon-Woods M, Batalden PB, Homa K, Van Citters AD, Morgan TS, Eftimovska E, Fisher ES, Ovretveit J, Harrison W, et al. Patient focused registries can improve health, care, and science. BMJ. 2016 Jul 1;354:i3319. doi: 10.1136/bmj.i3319. PMID: 27370543; PMCID: PMC5367618. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Vespa JE. Those who served: America’s veterans from World War II to the war on terror. ACS-43, American Community Survey Reports, U.S. Census Bureau, Washington (DC), 2020. https://www.census.gov/content/dam/Census/library/publications/2020.
- 21.Vaccaro AR, Hulbert RJ, Patel AA, Fisher C, Dvorak M, Lehman R Jr, Anderson P, Harrop J, Oner FC, Arnold P, et al. The subaxial cervical spine injury classification system: a novel approach to recognize the importance of morphology, neurology, and integrity of the disco-ligamentous complex. Spine (Phila Pa 1976). 2007 Oct 1;32(21):2365–74. doi: 10.1097/BRS.0b013e3181557b92. PMID: 17906580. [DOI] [PubMed] [Google Scholar]
- 22.Lee JY, Vaccaro AR, Lim MR, Oner FC, Hulbert RJ, Hedlund R, Fehlings MG, Arnold P, Harrop J, Bono CM, et al. Thoracolumbar injury classification and severity score: a new paradigm for the treatment of thoracolumbar spine trauma. J Orthop Sci. 2005 Nov;10(6):671–5. doi: 10.1007/s00776-005-0956-y. PMID: 16307197; PMCID: PMC2779435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Cogan AM, Rinne ST, Weiner M, Simon S, Davila J, Yano EM.. Using research to transform electronic health record modernization: advancing a VA partnered research agenda to increase research impacts. J Gen Intern Med. 2023 Oct;38(Suppl 4):965–973. doi: 10.1007/s11606-023-08289-y. Epub 2023 Oct 5. PMID: 37798575; PMCID: PMC10593706. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.