Abstract
Importance:
The American Academy of Neurology Axon Registry® provides real-world data for patients with multiple sclerosis and neuro-myelitis optica. However, some data are incomplete (e.g. demographics) and some relevant outcomes are not systematically captured in neurology documentation (e.g. visual acuity). The American Academy of Ophthalmology IRIS® Registry (Intelligent Research in Sight) contains demographic and visual function data that may complement Axon Registry-derived data to enhance understanding of real-world visual outcomes in neurological disease.
Objective:
To combine Axon Registry and IRIS Registry data to reduce missingness of demographic information and characterize visual outcomes in patients with multiple sclerosis and neuro-myelitis optica.
Design:
Cross-sectional study
Setting:
Outpatient neurology and ophthalmology clinical practices
Participants:
Patients participating in both registries between January 1, 2014 through December 10, 2021 were included if they had repeat ICD-9/10 codes for with multiple sclerosis or neuro-myelitis optica in the Axon registry.
Exposure:
Diagnosis (multiple sclerosis or neuro-myelitis optica)
Main outcome and measure:
Age, sex, race and ethnicity were assessed in the individual registries and classified as conflicting, missing, or not missing in the combined data set. The IRIS Registry contributed visual acuity data.
Results:
Among 60,316 patients with multiple sclerosis and 1,068 patients with neuro-myelitis optica in the Axon Registry, 14,085 and 252 had temporal overlap in the IRIS Registry. Combining data reduced missing or conflicting data for race and ethnicity by 15–19% (absolute reduction, all p≤0.0005), but not age (p=1.0) or gender (p=0.08). 10,907 patients with MS and 142 with NMO had visual acuity data in the IRIS Registry. Visual acuity averaged between eyes was worse in patients with NMO after adjusting for age and gender (0.17 logMAR, 95%CI 0.12,0.21, p<0.0005).
Conclusion and Relevance:
Using data from two registries reduced missing data for race and ethnicity and enabled examination of outcomes captured in the IRIS Registry for conditions that are diagnosed more frequently in the Axon Registry, demonstrating the utility of a multi-registry analysis.
Introduction
Real-world data derived from electronic health records (EHR) and contained in medical registries has been demonstrated to generate insights into patient care and outcomes beyond that seen in prospective studies.1 However, missing data, a challenge in registries, limits the evidence and insights that can be generated both due to limitations in what variables can be considered and bias generated by missing data.2,3
Medical specialty specific registries have been developed by multiple medical societies with the aims of quality improvement and research. The American Academy of Neurology Axon Registry®4,5 is a neurology-specific patient registry that collects, reports, and analyzes EHR data. As of August 2022, more than 17.9 million patient visits from 3 million unique patients exist in the database. The American Academy of Ophthalmology IRIS® Registry (Intelligent Registry in Sight)6 is the nation’s first and world’s largest comprehensive eye disease clinical database with over 70 million unique patients and over 12,000 contributing ophthalmologists. These specialty registries have proven applications in generating insights relevant to their specialty practice overall and for specific diseases.1,7–10 However, they are limited in their use towards studying interdisciplinary care and outcomes.
Multiple sclerosis (MS) and neuromyelitis optica (NMO) are neurological disorders causing visual morbidity,11 with multiple prospective studies demonstrating worse functional and structural visual outcomes in people with NMO than in people with MS.12,13 Because people with MS and NMO receive care in the neurology setting, analysis of Axon Registry data has the potential to generate real-world evidence and insights for these important diseases. However, neurology outpatient documentation does not routinely include quantitative visual function assessments such as visual acuity, hindering the study of visual outcomes in these neurological diseases. Conversely, IRIS Registry data includes visual outcomes, but likely contains incomplete records pertaining to neurological disease diagnosis, which potentially hinders its use in isolation to study visual outcomes in neurological and other non-ophthalmic diseases.
The objective of this study was to compare real-world visual acuity (VA) outcomes between people with MS and NMO by combining American Academy of Neurology Axon Registry data and American Academy of Ophthalmology IRIS Registry.
Methods
Study Design
This was a cross-sectional study of people with MS or NMO who received outpatient neurology care at a practice participating in the American Academy of Neurology Axon Registry within the study period of January 1, 2014 through December 10, 2021.
Subjects
Patients with MS were defined as those with ≥3 ICD9/10 codes for MS (340, G35) and no codes for NMO (341.0, G36.0) on separate dates in the Axon Registry14 on or after January 1, 2014. Patients with NMO were defined as those with ≥3 ICD9/10 codes for NMO on separate dates on or after January 1, 2014 and no code for MS after the most recent code for NMO in the Axon Registry. These criteria were applied to de-identified structured EHR derived data from the Axon Registry.
Within the MS and NMO groups, “overlapping patients” were defined as those concurrently in de-identified structured EHR-derived data from the IRIS Registry using person and time criteria (i.e. same person, same time) (Figure 1). This was achieved using de-identified globally unique identifiers at the patient-level, which are created by tokenizing a combination of upstream data such as patient name and date of birth. This enabled linking a patient’s record in one data set with a record for the same patient in a different set using only de-identified information. For the overlapping patients, study entry was the earliest time point during which a patient had data in both registries. Patients with no available data in the IRIS Registry were categorized as being only in the Axon Registry. For these patients, study entry was the point in time at which they entered the Axon Registry.
Figure 1: Definition of patient groups using the Axon Registry and Iris Registry.
Inclusion criteria were applied to Axon registry patients (vertical pattern and cross pattern) to identify patients with MS and NMO. Overlapping patients were defined as those who also contributed data to the IRIS Registry.
Standard Protocol Approvals, Registrations, and Patient Consents
This study complies with the tenets of the Declaration of Helsinki. Data stored within the Axon Registry and the IRIS Registry were de-identified and have been qualified as Health Insurance Portability and Accountability Act compliant. This study was reviewed and deemed exempt by the WCG Institutional Review Board (Puyallup, WA).
Outcomes
Diagnoses of MS or NMO identified within the IRIS Registry were assessed for overlapping patients by applying the same ICD criteria used within the Axon Registry to de-identified structured data from the ophthalmology specialty database.
Demographics (age, gender, race/ethnicity) were extracted at the patient level from the Axon Registry in all those eligible for inclusion in the study and from the IRIS Registry only for those that were also present in the IRIS patient registry (i.e. overlapping patients). Age at time study entry was used. Unknown demographic information was classified as missing.
For overlapping patients, VA in the right and left eyes was extracted in logMAR units from the de-identified structured data in the IRIS Registry. If a patient had a VA reading at time of study entry, then this was used as the initial VA. If they did not have a VA reading on the study entry date, the first reading after the study entry date was recorded. Uncorrected VA values were excluded. All chart distances were considered.VA was averaged between eyes for purposes of comparison.
Statistical Analysis
Demographics for overlapping patients with MS or NMO were compared to those only present in the Axon Registry using Chi- square tests to assess if there was any potential bias in the study population that was present in both patient registries (i.e. overlapping patients). The proportions of overlapping patients meeting disease inclusion criteria when leveraging diagnostic codes from the IRIS Registry were calculated for MS and NMO subjects to confirm the hypothesis that neurological diagnoses have a lower capture in the IRIS Registry.
For overlapping patients, demographic information at time of study entry across both specialty registries was combined. In instances where information within the Axon Registry was not available, values from the IRIS Registry were used if present. Conflicting information between registries was categorized as conflicting. For each demographic variable, the proportion of patients with missing or conflicting values derived only from the Axon Registry was compared to the proportion estimated after combining de-identified data from both the Axon Registry and the IRIS Registry data using McNemar’s statistic.
VA was compared between overlapping patients with MS and NMO using a linear regression model that included the covariates for age and gender.
All analyses were performed using PySpark software (Apache Software Foundation) version 3.2.3 and R Statistical Software (v4.2.1; R Core Team 2021) with p<0.05 as the threshold for statistical significance.
Results:
Comparison of Axon Registry only and Overlapping Axon Registry/IRIS Registry patients
Among 60,316 subjects meeting MS inclusion criteria in the Axon Registry, 14,085 (23%) had temporal overlap in the IRIS Registry. Overlapping patients were older, more likely to be female, more likely to be White, and less likely to be of Hispanic ethnicity (Table 1). Among 887 NMO subjects meeting inclusion criteria in the Axon Registry, 252 (28%) had temporal overlap in the IRIS Registry. Overlapping patients were older compared to Axon Registry only subjects (Table 2).
Table 1:
Demographics of patients with MS
| Axon Registry only (n=46 231) |
Axon Registry/IRIS Registry overlap (n=14 085) |
Comparison (Chi square) | Comparison* (McNemar) | ||
|---|---|---|---|---|---|
| Data source | Axon Registry A n (%) |
Axon Registry B n (%) |
Axon Registry + IRIS Registry C n (%) |
A vs. B | B. vs. C. |
| Age group | P < 0.0005 | P = 1 | |||
| ≤ 17 | 159 (0.3%) | 30 (0%) | 30 (0%) | ||
| 18 – 30 | 4,201 (9.1%) | 646 (4.6%) | 646 (4.6%) | ||
| 31 – 40 | 9,204 (20.0%) | 1,635 (11.6%) | 1635 (11.6%) | ||
| 41 – 50 | 12,341 (26.7%) | 2,896 (20.6%) | 2,896 (20.6%) | ||
| 51 – 60 | 11,857 (25.7%) | 3,957 (28.1%) | 3,957 (28.1%) | ||
| 61 – 70 | 6,541 (14.2%) | 3,565 (25.3%) | 3,565 (25.3%) | ||
| 71 – 80 | 1,529 (3.3%) | 1,236 (8. 8%) | 1,236 (8. 8%) | ||
| ≥ 81 | 166 (0.4%) | 120 (0.9%) | 120 (0.9%) | ||
| Missing | 233 (0.5%) | < 10 | < 10 | ||
| Conflicting | < 10 | ||||
| Gender | P < 0.0005 | P = 0.08 | |||
| Female | 34,401 (74.4%) | 11,316 (80.3%) | 11,307 (80.3%) | ||
| Male | 11,777 (25.5%) | 2,752 (19.5%) | 2,748 (19.5%) | ||
| Missing | 53 (0 %) | 17 (0%) | < 10 | ||
| Conflicting | 30 (0%) | ||||
| Race | P < 0.0005 | P < 0.0005 | |||
| White | 31,275 (67.7%) | 10,540 (74.8%) | 11,755 (83.5%) | ||
| Black | 5,418 (11.7%) | 1,417 (10.1%) | 1,526 (10.8%) | ||
| Asian | 249 (0.5%) | 55 (0%) | 53 (0%) | ||
| Native American or Alaska native | 96 (0%) | 31 (0%) | 18 (0%) | ||
| Native Hawaiian or Pacific Islander | 68 (0%) | 14 (0%) | 10 (0%) | ||
| Other | 196 (0.4%) | 23 (0%) | 49 (0%) | ||
| Missing | 8,929 (19.3%) | 2,005 (14.2%) | 477 (3.4%) | ||
| Conflicting | 197 (1.4%) | ||||
| Ethnicity | P < 0.0005 | P < 0.0005 | |||
| Hispanic | 2,283 (4.9%) | 545 (3.9%) | 511 (3.6%) | ||
| Non-Hispanic | 30,847 (66.7%) | 9,664 (68.6%) | 12,262 (87.1%) | ||
| Missing | 13,101 (28.3%) | 3,876 (27.5%) | 1,112 (7.9%) | ||
| Conflicting | 200 (1.4%) | ||||
missing and conflicting excluded
Table 2:
Demographics of subjects with NMO
| Axon Registry only (n=635) |
Axon Registry/IRIS Registry overlap (n=252) |
Comparison (chi square) | Comparison* (McNemar) | ||
|---|---|---|---|---|---|
| Data source | Axon Registry A n (%) |
Axon Registry B n (%) |
Axon + IRIS Registry C n (%) |
A vs. B | B. vs. C. |
| Age group | P < 0.0005 | P = 1 | |||
| ≤ 17 | 16 (3%) | < 10 | < 10 | ||
| 18 – 30 | 109 (17.2%) | 20 (8%) | 20 (7%) | ||
| 31 – 40 | 117 (18.4%) | 42 (17%) | 42 (17%) | ||
| 41 – 50 | 135 (21.3%) | 49 (19%) | 49 (20%) | ||
| 51 – 60 | 125 (19.7%) | 59 (23%) | 59 (23%) | ||
| 61 – 70 | 84 (13%) | 59 (23%) | 59 (23%) | ||
| 71 – 80 | 39 (6%) | 18 (7%) | 18 (7%) | ||
| ≥ 81 | < 10 | < 10 | < 10 | ||
| Missing | < 10 | < 10 | < 10 | ||
| Conflicting | < 10 | ||||
| Gender | P = 0.06 | P = 1 | |||
| Female | 501 (78.9%) | 206 (81.8%) | 207 (82.1%) | ||
| Male | 132 (20.8%) | 45 (18%) | 45 (18%) | ||
| Missing | < 10 | < 10 | < 10 | ||
| Conflicting | < 10 | ||||
| Race | P < 0.03 | P < 0.0005 | |||
| White | 300 (47.2%) | 126 (50.0%) | 143 (56.8%) | ||
| Black | 188 (29.6%) | 73 (29%) | 74 (29%) | ||
| Asian | 17 (3%) | < 10 | < 10 | ||
| Native American or Alaska Native | < 10 | < 10 | < 10 | ||
| Native Hawaiian or Pacific Islander | < 10 | < 10 | < 10 | ||
| Other | 10 (2%) | < 10 | < 10 | ||
| Missing | 119 (18.7%) | 43 (17%) | 13 (5%) | ||
| Conflicting | < 10 | ||||
| Ethnicity | P = 0.07 | P < 0.0005 | |||
| Hispanic | 71 (11%) | 28 (11%) | 31 (12%) | ||
| Non-Hispanic | 363 (57.2%) | 151 (59.9%) | 187 (74.2%) | ||
| Missing | 201 (31.7%) | 73 (29%) | 27 (11%) | ||
| Conflicting | < 10 | ||||
missing and conflicting excluded
Among the 14,085 patients meeting the MS inclusion criteria in the Axon Registry with temporal overlap within the IRIS Registry, 4,612 (33%) met the inclusion criteria for MS diagnosis using only IRIS Registry data. Among 252 patients meeting the NMO inclusion criteria in the Axon Registry with temporal overlap within the IRIS Registry, 52 (20%) met the ICD inclusion criteria for NMO using only IRIS Registry data.
Combining demographic data in Axon Registry/IRIS Registry overlap subjects
Among overlapping patients with MS, no patient had missing data for age in the two registries and one had a conflict in age year, but not category, between the two registries (Table 1). However,14.2% and 15.9% had missing data for race in the Axon Registry and the IRIS Registry, respectively, which reduced to 3.4% missing and 1.4% conflicting in the combined data set. For ethnicity, 27.5% and 24.4% had missing data in the Axon registry and IRIS registry, respectively, which reduced to 7.9% missing and 1.4% conflicting in the combined data set. Gender had 0.1% and 0.3% missing data in the Axon Registry and IRIS Registry, respectively, which reduced to 0% missing, but had 0.2% conflicting in the combined data set.
Among overlapping patients with NMO, age was completely available in either registry and there were no conflicts (Table 2). There was 17.1% and 18.7% missing data for race in the Axon Registry and IRIS Registry, respectively, which decreased to 5.2% missing and 3.6% conflicting in the combined data set. For ethnicity, 29.0% and 23.2% missing data was observed in the Axon Registry and IRIS Registry, respectively, which also had a decline in missingness to 10.7% and resulting in 2.8% conflicting in the combined data set. Gender had 0.4% and 0% missing data in the Axon Registry and IRIS Registry, respectively, with 0% conflicting in the combined data set.
Visual Acuity
For overlapping patients, 10,920 patients with MS and 142 patients with NMO had VA records in the IRIS Registry. VA was worse in patients with NMO when compared to patients with MS (NMO: median = 0.176 (IQR: 0.049, 0.398); MS: median = 0.097 (IQR: 0, 0.239)). When adjusting for age and gender in a linear regression model, VA averaged between eyes was still worse in patients with NMO as compared to those with MS (0.17 logMAR (95% CI: 0.12, 0.21); p<0.0005).
Discussion
In this study, a de-identified neurology EHR derived data set, the Axon Registry, was used to identify real-world patients with neurological diseases, specifically MS and NMO. Contemporary data on the same patients from a de-identified ophthalmology EHR derived data set, the IRIS Registry, when combined with data from the Axon Registry, enabled greater capture of missing race and ethnicity data and provided visual outcome data. The results are comparable to previously reported visual outcome differences between patients with MS and NMO15 and the sample size is larger than other studies comparing vision in MS and NMO.12,13 We used a unique data set combining two national registries from different medical specialties with broad geographic and practice types representing real-world data that was collected during the course of routine medical care.
While some diseases are cared for exclusively by single medical specialties, many receive interdisciplinary care and have relevant outcomes in different specialties from where most of the care is delivered. In the case of neuro-immunologic diseases, including MS and NMO, neurology practices are the sites of diagnosis and primary treatment management. While neurology records are likely to document visual symptoms, typical neurological examination documentation does not include quantitative VA. Thus, prior studies of visual outcomes in MS and NMO have been prospectively collected, typically in academic neurology practices, with selection bias impacting the study sample. While the Axon Registry and IRIS Registry participants are not a population sample, the nature of EHR-derived data lowers the barrier for participation by allowing passive contribution of medical data. Furthermore, geographic and practice type diversity represented in the participating practices increases diversity and potential generalizability. While there was some bias introduced by the overlap criterion, the minority of overlapping patients who fulfilled inclusion criteria using IRIS Registry data alone justified the use of the Axon Registry to identify patients with neurological disease.
Using the Axon Registry to identify MS and NMO subjects and the IRIS Registry to measure visual acuity, we found that patients with NMO have worse visual outcomes with average VA being 1–2 Snellen lines worse than subjects with MS subjects. This result is consistent with other previously published findings,12 which is an important proof-of-concept for the use of real-world EHR-derived data with applications for characterizing disability, planning interventions, and planning clinical trials.
A challenge with registries with EHR-derived data, like many retrospective data sets, is missingness of data collected during routine clinical care. Missing data can impact sample size for studies that leverage complete case analyses; when missingness is substantial, it can also limit consideration of variables as covariates within adjusted models. In both the Axon Registry and IRIS registry, age and gender had a low proportion of missingness while race and ethnicity had substantial missingness. Combining the data reduced the proportion of subjects with missing race and ethnicity and biased it towards a more female and older sample. These demographic shifts are consistent with the epidemiology of ophthalmic diseases which have increasing prevalence with age and a slight preference for females due to longer female life expectency. A small proportion of the combined data conflicted between the registries and was treated as equivalent to missing since which one is accurate could not be ascertained. These conflicts likely resulting from data entry errors into the EHRs. Although combining registries decreased the overall sample size, the remaining cohort remained large..
Limitations of this study include those inherent to registries with EHR-derived structured data sets. This was not a population sample, and availability of data points was contingent on what was documented by the clinician. Diagnosis of MS and NMO were based on ICD codes, though the definition used is consistent with what has been previously applied in the literature and internal validation work performed by a separate team suggests near 100% accuracy of this definition in registry participants with unstructured data available to validate the diagnosis (unpublished).The vision analysis did not account for co-morbid ophthalmic disease, but this is likely not to be different between groups. The vision analysis did not account for neurological disease features including duration, history of optic neuritis and treatment.
Conclusion
Combining de-identified registry data allowed inclusion of relevant outcomes not in the primary disease registry at the cost of decreasing sample size. Although the demographics of the patients in the combined data set shifted slightly from those in the Axon Registry alone (i.e., overlapping patients with MS were more likely to be female, more likely to be White, and less likely to be of Hispanic ethnicity, and those with NMO were more likely to be older), the reduction of missing race and ethnicity data was substantial. The reduction of missing data and the addition of important VA outcome data suggests that a linked data set can be used to reflect the different types of care patients may receive across various interactions with the healthcare system more comprehensively.
Highlights.
People with MS or NMO in the Axon registry were subjects
The Axon registry had missing data for race and ethnicity in >15% of subjects
>30% of subjects also contributed data to AAO’s IRIS registry
Combining demographic data from the two registries reduced missing data
Visual function (from the IRIS registry) was worse in people with NMO than MS
Acknowledgement
Funding was provided by Verana Health. Dr. Moss is a medical advisor to Verana Health. All other authors were employees of Verana Health at time this study was performed. Dr. Torres had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Role of Funding Source:
This study was supported by Verana Health. All authors were employees or consultants for Verana Health at the time of study performance.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declaration of interest:
All authors were employees or consultants for Verana Health at the time of study performance.
Disclaimer
The Axon Registry® is an initiative of the American Academy of Neurology Institute. There are limitations of the Axon Registry data.4,5 The views expressed in this manuscript represent those of the authors and do not necessarily represent the official views of the Axon Registry, American Academy of Neurology or the American Academy of Neurology Institute. The American Academy of Neurology (AAN) is not responsible for the claims made in this manuscript. The AAN does not endorse companies or their products.
References
- 1.Taipale H, Schneider-Thoma J, Pinzón-Espinosa J, et al. Representation and Outcomes of Individuals With Schizophrenia Seen in Everyday Practice Who Are Ineligible for Randomized Clinical Trials. JAMA Psychiatry. 2022;79(3):210–218. doi: 10.1001/jamapsychiatry.2021.3990 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Shivasabesan G, Mitra B, O’Reilly GM. Missing data in trauma registries: A systematic review. Injury. 2018;49(9):1641–1647. doi: 10.1016/j.injury.2018.03.035 [DOI] [PubMed] [Google Scholar]
- 3.Yang DX, Khera R, Miccio JA, et al. Prevalence of Missing Data in the National Cancer Database and Association With Overall Survival. JAMA Netw Open. 2021;4(3):e211793. doi: 10.1001/jamanetworkopen.2021.1793 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Baca CM, Benish S, Videnovic A, et al. Axon Registry® data validation: Accuracy assessment of data extraction and measure specification. Neurology. 2019;92(18):847–858. doi: 10.1212/WNL.0000000000007404 [DOI] [PubMed] [Google Scholar]
- 5.Sigsbee B, Goldenberg JN, Bever CT, Schierman B, Jones LK. Introducing the Axon Registry: An opportunity to improve quality of neurologic care. Neurology. 2016;87(21):2254–2258. doi: 10.1212/WNL.0000000000003264 [DOI] [PubMed] [Google Scholar]
- 6.Parke Ii DW, Lum F, Rich WL. The IRIS® Registry : Purpose and perspectives. Ophthalmologe. 2017;114(Suppl 1):1–6. doi: 10.1007/s00347-016-0265-1 [DOI] [PubMed] [Google Scholar]
- 7.Chiang MF, Sommer A, Rich WL, Lum F, Parke DW. The 2016 American Academy of Ophthalmology IRIS® Registry (Intelligent Research in Sight) Database: Characteristics and Methods. Ophthalmology. 2018;125(8):1143–1148. doi: 10.1016/j.ophtha.2017.12.001 [DOI] [PubMed] [Google Scholar]
- 8.MacCumber MW, Wykoff CC, Karcher H, et al. One-Year Brolucizumab Outcomes in Neovascular Age-Related Macular Degeneration from a Large United States Cohort in the IRIS® Registry. Ophthalmology. Published online April 21, 2023:S0161–6420(23)00277–4. doi: 10.1016/j.ophtha.2023.04.012 [DOI] [PubMed] [Google Scholar]
- 9.Lim HW, Pershing S, Moshfeghi DM, et al. Causes of Childhood Blindness in the United States using the IRIS® Registry (Intelligent Research in Sight). Ophthalmology. Published online April 8, 2023:S0161–6420(23)00245–2. doi: 10.1016/j.ophtha.2023.04.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Armbrust KR, Kopplin LJ. Characteristics and Outcomes of Patients with Scleritis in the IRIS® Registry (Intelligent Research in Sight) Database. Ophthalmol Sci. 2022;2(3):100178. doi: 10.1016/j.xops.2022.100178 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Salter AR, Tyry T, Vollmer T, Cutter GR, Marrie RA. “Seeing” in NARCOMS: a look at vision-related quality of life in the NARCOMS registry. Mult Scler. 2013;19(7):953–960. doi: 10.1177/1352458512469694 [DOI] [PubMed] [Google Scholar]
- 12.Shen T, You Y, Arunachalam S, et al. Differing Structural and Functional Patterns of Optic Nerve Damage in Multiple Sclerosis and Neuromyelitis Optica Spectrum Disorder. Ophthalmology. 2019;126(3):445–453. doi: 10.1016/j.ophtha.2018.06.022 [DOI] [PubMed] [Google Scholar]
- 13.Ratchford JN, Quigg ME, Conger A, et al. Optical coherence tomography helps differentiate neuromyelitis optica and MS optic neuropathies. Neurology. 2009;73(4):302–308. doi: 10.1212/WNL.0b013e3181af78b8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Culpepper WJ, Marrie RA, Langer-Gould A, et al. Validation of an algorithm for identifying MS cases in administrative health claims datasets. Neurology. 2019;92(10):e1016–e1028. doi: 10.1212/WNL.0000000000007043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Macaron G, Khoury J, Bena J, et al. Early age of onset predicts severity of visual impairment in patients with neuromyelitis optica spectrum disorder. Mult Scler. 2021;27(11):1749–1759. doi: 10.1177/1352458520981736 [DOI] [PubMed] [Google Scholar]