Highlights
- •
Half of the patients at an academic uveitis practice had a disease for which no Standardization of Uveitis Nomenclature (SUN) classification criteria existed.
- •
If a patient’s disease was listed by SUN, most patients could be classified.
- •
SUN classification generally agreed with the clinical diagnosis.
Purpose
To evaluate the new Standardization of Uveitis Nomenclature (SUN) classification criteria for uveitides by applying them to patients in an academic uveitis practice.
Design
Evaluation of classification criteria.
Methods
The charts of all patients attending the uveitis service at the University of Colorado Hospital between January 1, 2013, and December 31, 2020, were reviewed. Patients with scleritis, ocular cicatricial pemphigoid, and peripheral ulcerative keratitis were excluded. We attempted to classify each patient’s uveitis using the SUN classification criteria. Classification attempts were made within the relevant anatomical or infectious categories for their pathology but did not necessarily have to match their clinical diagnosis by a uveitis specialist. We recorded whether classification was possible as well as their clinical diagnosis by a uveitis specialist.
Results
All patients attending the uveitis clinic at our academic institution between January 1, 2013, and December 31, 2020, were reviewed. Of the 1143 patients with uveitis, 572 (50.0%) had a disease that was not listed in the SUN classification system, and so no attempt to classify these patients was possible. Of the remaining 571 patients, 522 (91.4%) were able to be classified by SUN and in 492 (94.3%) of the 522 cases, their SUN classification matched their clinical diagnosis by a uveitis specialist.
Conclusions
Half of the patients at an academic uveitis practice had a disease for which no SUN classification criteria existed. In cases where classification by SUN could be attempted, the system performed well and generally agreed with their clinical diagnosis.
T he ophthalmic subspecialty of uveitis encompasses diverse pathology that may affect any part of the ocular anatomy, and in some cases, include systemic manifestations. Given the breadth and heterogeneity of pathology, there has been until recently little standardization in the approach to defining and classifying the uveitides. Although these diseases account for approximately 2.8% to 10% of all blindness in the United States, they are still relatively uncommon. Standard terminology to define and organize uveitic diseases allows for improved patient care, better comparability of research studies, and pooling of data from different institutions, which is invaluable for these relatively uncommon diseases.
The barriers to combining data are not only limited to a lack of standardization among clinicians and scientists. For example, there is also substantial imprecision within the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, coding of the uveitides, with additional variation among different electronic medical records. , To allow for more meaningful research and enhanced clinical care, standard terminology to define and classify the uveitides must be employed.
The Standardization of Uveitis Nomenclature (SUN) working group was formed in 2005 with the goal of standardizing grading of inflammation, diagnostic terminology, and outcome measures in uveitis. The first SUN working group produced a grading scheme for anterior chamber and vitreous inflammation and began the process of creating standard terms for reporting clinical data. In 2009, the SUN working group initiated an informatics-based process to produce consensus terminologies for the 25 major uveitic clinical entities.
Finalization of these classification criteria began in 2019, and the new SUN disease classification criteria were published in 2021. SUN developed these criteria in 4 phases: informatics, case collection, case selection, and machine learning. , A total of 5766 cases were collected from 76 investigators. These cases were submitted to a panel, which used Delphi criteria to include or exclude cases for each disease. The final database of 4046 cases was randomly separated into training and validation sets, using an 85%-15% split. Machine learning was then used to determine the most relevant differentiating criteria, to create mutually exclusive criteria for the 25 uveitic diseases.
An important semantic distinction must be made between classification criteria and diagnostic criteria. Representing 2 ends of a continuum, classification criteria refers to standard definitions that are primarily used in clinical research studies, whereas diagnostic criteria are more usefully applied toward routine clinical care of patients. In some cases, often when a criterion standard test is available, diagnostic and classification criteria may overlap considerably. In general, however, classification for research purposes aims to create fairly homogenous groups that can be compared across different studies, whereas diagnostic criteria aim to encompass all the possible clinical presentations and severity that may exist for a disease. Classification criteria strive for higher specificity and lower sensitivity.
The criteria used in this report can be considered classification criteria, aiming to standardize the terms used to describe uveitides in clinical research. The purpose of this study is to apply the SUN classification criteria to a cohort of patients at an academic uveitis practice and evaluate the utility of the criteria in classifying this cohort.
METHODS
All charts of patients examined by the uveitis service at the University of Colorado Hospital between January 1, 2013, and December 31, 2020, were reviewed. Patients who had scleritis, peripheral keratitis, and/or ocular cicatricial pemphigoid without intraocular inflammation were excluded. Patients were also excluded if they had never been seen with active intraocular inflammation by our clinic. The SUN classification system was attempted to be applied to each patient using data from all uveitis visits, including history, examination, imaging, and laboratory results. A general laboratory and imaging workup was performed on nearly all patients in our study. If no diagnosis is suggested by patient history, our practice pattern is to obtain serologic testing for syphilis and a chest radiograph.
Bearing in mind the possible SUN classification diagnoses, each case was evaluated using both the clinical diagnosis and clinical features. If a case had no possible relevant SUN diagnosis, it was deemed ineligible for classification. Classification was attempted in eligible cases to determine whether the case fulfilled the criteria set forth by SUN. We attempted to classify each patient within the relevant anatomical or infectious categories for his or her pathology. This did not necessarily have to match the patient’s clinical diagnosis by a uveitis specialist.
For example, if a patient had only clinical anterior uveitis, we only attempted classification for the anterior uveitides listed by SUN. Additionally, if a patient had been given a diagnosis of multifocal choroiditis by his or her treating uveitis specialist, classification could be attempted for serpiginous choroiditis, punctate inner choroiditis, acute posterior multifocal placoid pigment epitheliopathy, multiple evanescent white dot syndrome, and multifocal choroiditis. An attempt to classify patients with panuveitis was made in both posterior uveitis and panuveitis SUN categories.
Two authors (L.M. and A.P.) reviewed each chart separately: A.P. is a fellowship-trained uveitis specialist and L.M. is a senior ophthalmology resident. Any discrepancies in classification were adjudicated by a third independent grader, A.R., who is also a fellowship-trained uveitis specialist. For each patient, we recorded the clinical diagnosis by their treating uveitis specialist. When classification could not be attempted because there was no appropriate SUN category, we recorded the patient as “unable to be classified.” When classification was attempted, the patient was noted to either have met the SUN classification criteria or to have not met the criteria for that diagnosis. Patient demographics including age, gender, race, and ethnicity were recorded, as well as length of follow-up.
Descriptive statistics were used to summarize the data, with the main outcome being the proportion of patients who were able to be classified. Secondary outcomes included concordance between their SUN diagnosis and clinical diagnosis, as well as the uveitic diseases that were unable to be classified. The Wilcoxon rank sum test was used to compare median length of time in our service for patients able to be classified vs patients not able to be classified. Data were analyzed using Stata, version 17.0 (StataCorp). The study received Colorado Multiple Institutional Review Board (COMIRB) exempt approval and was conducted in compliance with the Health Insurance Portability and Accountability Act regulations and the tenets set forth by the Declaration of Helsinki.
RESULTS
A total of 1143 patients were included in this study. The mean age at first visit was 47 years (SD 20.1 years). Fifty-eight percent were female, and the majority of patients were non-Hispanic white. The median length of time patients had been known to the uveitis service was 374 days. For patients who were able to be classified, the median length of time known to our service was 453 days, compared with 272 days for those patients who were unable to be classified ( P < .0001). In total, 117 charts required adjudication by the third, independent grader. Classification was attempted in 571 of 1143 (49.96%) patients, of whom 522 were able to be classified into 1 or more SUN categories ( Table 1 ).
| Uveitis Cohort | |
|---|---|
| Age at first visit, y | |
| Mean (SD) | 47.38 (20.15) |
| Range | 3-95 |
| Female sex, n (%) | 660 (57.74) |
| Race, n (%) | |
| Native American or Alaskan Native | 3 (0.26) |
| Asian | 48 (4.20) |
| Black or African American | 179 (15.66) |
| Native Hawaiian or Other Pacific Islander | 4 (0.35) |
| White | 788 (68.94) |
| More than 1 race | 13 (1.14) |
| Unknown/not reported | 108 (9.45) |
| Ethnicity, n (%) | |
| Hispanic or Latino | 132 (11.55) |
| Not Hispanic or Latino | 984 (86.09) |
| Unknown or not reported | 27 (2.36) |
| Length of follow-up, d | |
| Mean (SD) | 640 (727) |
| Median (range) | 374 (0-3511) |
| Patients where SUN classification was attempted, n (%) | 571 (49.96) |
| Number of patients who were able to be classified by SUN, n (%) | 522 (45.67) |
For 5 patients, SUN classification was attempted for 2 different diseases. One patient met criteria for both diseases, specifically, multifocal choroiditis in one eye and punctate inner choroidopathy in the contralateral eye. This was due to differences in lesion size between the 2 eyes. Three patients met criteria for only 1 of 2 attempted classifications: human leukocyte antigen (HLA)-B27 associated anterior uveitis and sarcoid uveitis (met criteria for sarcoid uveitis only), juvenile idiopathic arthritis, and HLA-B27–associated anterior uveitis (met criteria for HLA-B27–associated anterior uveitis only), and multiple sclerosis–associated intermediate uveitis and varicella zoster virus–associated anterior uveitis (met criteria for multiple sclerosis–associated intermediate uveitis only). One patient did not meet criteria for either of the 2 attempted classifications (herpes simplex virus [HSV] or varicella zoster virus anterior uveitis).
Table 2 shows the total number of attempts to apply SUN classification and includes these 5 patients where classification was attempted for 2 diseases: in total, there were 576 attempts of 571 patients to apply SUN classification criteria. SUN criteria was met by 522 patients, with 1 patient meeting criteria for 2 disease; thus, 523 of 576 classification attempts (90.80%) met SUN criteria for the disease, and in 493 of 523 cases (94.26%), the SUN classification agreed with the patient’s clinical diagnosis. For 9 of the 25 diseases, SUN criteria were met in all cases where classification was attempted, and for 8 of these 9 diseases, all cases agreed with their clinical diagnosis.
| SUN Category | Attempted Classification a , n (%) | Successful Classification (Met SUN Criteria) b , n (%) | Successful Classification Agreed With Clinical Diagnosis c , n (%) |
|---|---|---|---|
| CMV anterior uveitis | 6 (1.0) | 6 (100) | 6 (100) |
| HSV anterior uveitis | 12 (2.1) | 7 (58.3) | 7 (100) |
| VZV anterior uveitis | 24 (4.2) | 18 (75.0) | 16 (88.9) |
| Fuchs Uveitis Syndrome | 10 (1.7) | 8 (80.0) | 8 (100) |
| JIA associated uveitis | 27 (4.7) | 22 (81.5) | 22 (100) |
| Spondylarthritis/HLA-B27 associated uveitis | 132 (22.9) | 126 (95.5) | 120 (95.2) |
| TINU | 3 (0.5) | 3 (100) | 2 (66.7) |
| Pars planitis | 87 (15.1) | 86 (98.9) | 85 (98.8) |
| Intermediate uveitis, non-pars planitis | 77 (13.4) | 70 (90.01) | 55 (78.57) |
| MS-associated intermediate uveitis | 6 (1.0) | 6 (100) | 4 (66.7) |
| ARN | 11 (1.9) | 11 (100) | 11 (100) |
| CMV Retinitis | 17 (3.0) | 17 (100) | 17 (100) |
| Syphilitic uveitis | 7 (1.2) | 7 (100) | 7 (100) |
| Tuberculosis uveitis | 4 (0.7) | 3 (75.0) | 2 (66.7) |
| toxoplasmosis retinitis | 26 (4.5) | 19 (73.1) | 19 (100) |
| Behçet disease | 10 (1.7) | 8 (80.0) | 7 (87.5) |
| Sarcoid uveitis | 42 (7.3) | 39 (93.9) | 37 (94.9) |
| Sympathetic ophthalmia | 3 (0.5) | 2 (66.7) | 2 (100) |
| Early-stage VKH | 4 (0.7) | 3 (75.0) | 3 (100) |
| Late-stage VKH | 6 (1.0) | 3 (50.0) | 3 (100) |
| APMPPE | 4 (0.7) | 3 (75.0) | 2 (66.7) |
| Birdshot chorioretinitis | 20 (3.5) | 20 (100) | 20 (100) |
| MEWDS | 2 (0.3) | 1 (50.0) | 1 (100) |
| MFC with panuveitis | 19 (3.3) | 19 (100) | 19 (100) |
| PIC | 7 (1.2) | 7 (100) | 7 (100) |
| Serpiginous choroiditis | 10 (1.7) | 9 (90) | 9 (100) |
| Total d | 576 | 523 | 493 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
