Key Points
Question
What are the outcomes of glaucoma referrals in adults aged 18 to 40 years, and how can referral efficiency be improved?
Findings
In this cohort study of 292 453 adults aged 18 to 40 years, only 8% of those referred for glaucoma evaluation were diagnosed with glaucoma within 2 years. A simple 3-variable model based on age, intraocular pressure, and cup-disc ratio stratified patients with low risk (3%) and high risk (14%) of being diagnosed with glaucoma.
Meaning
The findings of this study suggest that the diagnostic yield of glaucoma evaluations in adults aged 18 to 40 years is low but that standardized referral guidelines can help risk stratify patients to ensure equitable eye care and efficient resource allocation.
This cohort study assesses the outcomes of glaucoma referrals in adults aged 18 to 40 years and the efficiency of models to stratify patients at risk of developing glaucoma.
Abstract
Importance
While early detection of glaucoma is vital to prevent irreversible vision loss, there are sparse data on the effectiveness of glaucoma referrals and methods to establish evidence-based referral guidelines in large, diverse populations.
Objective
To assess the prevalence and risk factors for diagnosed glaucoma and loss to follow-up among adult patients aged 18 to 40 years after a new diagnosis of referable glaucoma (ie, with glaucoma or suspected glaucoma).
Design, Setting, and Participants
This retrospective cohort study included patients aged 18 to 40 years presenting to Kaiser Permanente Southern California, a large managed health care system, for first-time eye examinations between January 1, 2013, and December 31, 2018. Data analysis occurred between September 2022 and August 2024.
Main Outcomes and Measures
The primary outcome was glaucoma diagnosed within 2 years of the first eye examination. The secondary outcome was the loss to follow-up, defined as failure to receive a glaucoma evaluation with visual field or optical coherence tomography testing within 2 years.
Results
The cohort included 292 453 patients aged 18 to 40 years who underwent first-time eye evaluations (mean [SD] age, 29.8 [6.4] years). Among 12 050 identified patients with referable glaucoma (52.3% female), 6827 (56.7%) completed glaucoma evaluations, of whom 563 (8.2%) were diagnosed with glaucoma (344 [61.1%] with open angle, 28 [5.0%] with angle closure, 84 [14.9%] with secondary glaucoma, and 107 [19.0%] with unspecified glaucoma). On multivariable analysis, male sex (odds ratio [OR], 1.55 [95% CI, 1.07-2.27]), higher intraocular pressure (IOP) (OR, 1.19 [95% CI, 1.15-1.23] per 1 mm Hg), and greater cup-disc ratio (CDR) (OR, 1.53 [95% CI, 1.34-1.75] per 0.1 unit) were associated with greater odds of glaucoma. Dichotomized age, IOP, and CDR models stratified 51 of 1613 patients (3.2%) into the low-risk group and 202 of 1477 patients (13.7%) into the high-risk group. Being younger than 32 years and having an IOP less than 18 mm Hg and a CDR less than 0.7 yielded a negative predictive value of 98.2% for a glaucoma diagnosis.
Conclusions and Relevance
In this cohort study, the diagnostic yield of glaucoma referrals was low among adults aged 18 to 40 years with first-time eye examinations. A simple risk-stratification strategy could help identify individuals with low and high risks of developing glaucoma, and adoption of evidence-based risk stratification and referral guidelines by health care systems and clinicians could improve equity of glaucoma care and use of eye-care resources.
Introduction
Glaucoma is a major public health issue and the leading cause of irreversible vision loss worldwide.1 The burden of glaucoma on patients, clinicians, and health care systems is expected to increase rapidly as its global prevalence rises to 112 million people by 2040.1 Individuals are identified as being at risk for glaucoma (or who have suspected glaucoma) if they exhibit risk factors, such as an enlarged cup-disc ratio (CDR) or elevated intraocular pressure (IOP), on routine eye examination.2,3 Due to the irreversible nature of glaucomatous vision loss, the American Academy of Ophthalmology and the World Glaucoma Association emphasize the importance of referring individuals with suspected glaucoma for in-office glaucoma evaluation and diagnostic testing by an ophthalmologist to establish the presence or absence of the disease.4,5 However, there is a paucity of clear guidelines about referral criteria and evidence supporting the effectiveness of current referral practice patterns.
Earlier studies investigating the outcomes of glaucoma referrals were mostly conducted prior to the popularization of research using data from electronic health records (EHRs) and are limited by small sample sizes and homogeneous study populations.6,7,8 Therefore, further study of large, diverse patient populations is needed to establish methods for risk stratification and guidelines for glaucoma referrals. The importance of standardized referral guidelines is highlighted by glaring disparities in glaucoma detection and outcomes.9 For example, Black and Hispanic Americans are less likely to be evaluated or identified as at risk for glaucoma but are more likely to require surgery or develop blindness from glaucoma.10,11,12 Improved allocation of eye-care resources is another potential benefit of standardized referral guidelines. Glaucoma care is costly; direct glaucoma-related expenditures exceed $3 billion annually in the US alone.1,13 Therefore, a systemic approach for delivering care to individuals with manifest glaucoma could aid in the long-term sustainability of glaucoma services, especially as demand for care rises, and the number of ophthalmologists declines.14,15
In this study, we aimed to evaluate the outcomes of glaucoma referrals by using data from a large cohort of adults aged 18 to 40 years in the Kaiser Permanente Southern California (KPSC) managed health care system. Kaiser Permanente is the largest integrated managed care consortium in the US; KPSC comprises a network of 13 medical centers serving 4.7 million members as of 2023.16 Although KPSC prioritizes preventative care, it lacks standardized evidence-based guidelines for glaucoma referrals. Therefore, there is an opportunity to improve the current referral system, especially for adults aged 18 to 40 years who are generally at low risk of glaucoma compared with adults older than 60 years.17 There is also an opportunity to identify systemic deficiencies, such as loss to follow-up of individuals with suspected glaucoma who are at high risk after referral, which could contribute to disparities in glaucoma outcomes.
Methods
This cohort study was approved by the KPSC Institutional Review Board with a waiver of informed consent because the research was conducted with deidentified data and posed minimal risk to the participants. The study followed the tenets of the Declaration of Helsinki18 and complied with HIPAA (Health Insurance Portability and Accountability Act). We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
KPSC members can freely schedule comprehensive eye examinations with an optometrist or receive a referral from a primary care physician to see an ophthalmologist. All KPSC patients undergoing comprehensive eye examinations are evaluated for glaucoma risk factors, such as enlarged CDR or elevated IOP. IOP is checked using a range of contact and noncontact methods and devices, including Goldmann Applanation Tonometry, iCare (Revenio Group), Tono-Pen (Reichert), and Marco Tonoref II M3 (Nidek). Patients with a first-time glaucoma suspect or glaucoma diagnosis are referred for glaucoma evaluation with visual field (VF) and/or optical coherence tomography (OCT) testing. Referrals are made based on clinical suspicion by individual clinicians rather than standardized guidelines issued by KPSC.
Available sociodemographic variables included sex, age at diagnosis, and self-reported race, which was combined with self-reported ethnicity at the time of entry into the EHR. Race and ethnicity data were collected because there are known racial and ethnic differences in glaucoma prevalence and risk. Based on available data, the primary race and ethnicity categories were Asian, Black, Hispanic, White, and other (including American Indian or Alaska Native and Native Hawaiian or Other Pacific Islander). Those who declined to self-report their race and ethnicity were combined under unknown. Available clinical data included glaucoma subtype diagnosis, date of diagnosis, dates of VF and OCT testing, number of visits with eye-care clinicians, IOP, CDR, and refractive error. Disease diagnoses were provided by a wide range of KPSC clinicians using broad definitions rather than standardized study definitions.
Study Definitions and Data
All KPSC members aged 18 to 40 years who received first-time eye examinations with a KPSC optometrist or KPSC ophthalmologist between January 1, 2013, and December 31, 2018, were identified based on encounter date and internal clinician and specialty codes (eFigure in Supplement 1). Referable glaucoma was defined as a composite diagnosis of glaucoma suspect or any glaucoma subtype based on International Classification of Diseases, Ninth Revision (ICD-9) and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes (eTable 1 in Supplement 1) recorded in the Kaiser EHR system (Epic Systems). The index date was defined as the date of the first eye examination for all patients except those with a diagnosis of referable glaucoma that occurred later in the study period. For those patients, the index date was defined as the date of the first diagnosis of referable glaucoma. Inclusion in the study required continuous observability (enrollment with KPSC) of at least 2 years after the index date. Data analysis occurred between September 2022 and August 2024.
The referred group was defined as the subset of patients diagnosed with referable glaucoma by an optometrist or ophthalmologist. The evaluated group was defined as the subset of referred patients who successfully completed VF or OCT testing based on ICD-9 and ICD-10 codes within 2 years of the index date (eTable 1 in Supplement 1). The lost-to-follow-up group was defined as the subset of referred patients who did not complete VF or OCT testing within 2 years of the index date. A 2-year window was selected to capture the majority (94.7%) of first VF or OCT testing without excluding patients who may have experienced temporary barriers to care.
The primary outcome measured was diagnosed glaucoma, defined as the presence of any glaucoma subtype diagnosis based on ICD-9 and ICD-10 codes (eTable 1 in Supplement 1) at the last visit within 2 years of the index date, occurring on or after the date of VF or OCT testing. In the absence of a glaucoma diagnosis after evaluation, patients remained categorized as those with suspected glaucoma. A secondary outcome measure was loss to follow-up, defined as failing to obtain VF or OCT testing within 2 years of the index date.
CDR, IOP, and refractive error data were derived from the first ophthalmologic or optometric visit in the Kaiser EHR database within the study period. One eye was selected for analysis based on disease severity to avoid mislabeling and analysis of the nonglaucomatous eye due to patient-level diagnosis codes. The eye with a greater CDR or, if a CDR was unavailable, a higher IOP was selected for analysis. If CDR and IOP data were both unavailable, the study eye was chosen at random. The method of tonometry was not considered in this study. Spherical equivalent was calculated as the sum of the sphere power and one-half of the cylinder power and categorized as hyperopia (spherical equivalent > 0.5 diopters [D]), emmetropia (−0.5 < spherical equivalent ≤ 0.5 D), low myopia (−3.0 < spherical equivalent ≤ −0.5 D), and moderate and high myopia (spherical equivalent ≤ −3.0 D).19
Statistical Analysis
The proportion of referred, evaluated, and diagnosed patients within the overall cohort was calculated and stratified by age, sex, and race and ethnicity. Continuous data were expressed as means and SDs and compared using the Wilcoxon rank sum test for 2 groups or Kruskal-Wallis and post hoc Dunn tests for 3 or more groups. Categorical data were expressed as proportions and percentages and were compared using the χ2 test. Univariable and multivariable logistic regression analyses were performed for the primary outcome of diagnosed glaucoma (among evaluated patients) and for the secondary outcome of lost to follow-up (among referred patients), with age, sex, race and ethnicity, CDR, IOP, and spherical equivalent as covariates.
A secondary logistic regression analysis was performed by dichotomizing age, IOP, and CDR into high-risk and low-risk categories with diagnosed glaucoma as the outcome measure. Dichotomization was based on means to create groups of relatively equal sizes for easier interpretation and implementation. High-risk age, IOP, and CDR were defined as being at or above the mean, while low-risk age, IOP, and CDR were defined as being below the mean. Four groups were created based on combinations of dichotomized age and IOP (low-risk age and IOP, low-risk age and high-risk IOP, high-risk age and low-risk IOP, and high-risk age and IOP). The probability of detected glaucoma per a 0.1-unit increase in the CDR was plotted for each group, and the negative predictive value and positive predictive value of each group were calculated. The proportion of evaluated patients expected to be under 2.5%, 5.0%, and 10.0% risk of glaucoma based on their age and IOP risk group and CDR was calculated along with the actual proportion in these low-risk categories that met the glaucoma outcome. A manual review of clinic notes from patients diagnosed with glaucoma who were very low risk was conducted to assess the validity of ICD-9 and ICD-10 codes. Univariable and logistic regression analyses were conducted with dichotomized age, IOP, and CDR. Clinically significant effects were defined as odds ratio (OR) greater than 1.10 or less than 0.90 and a 1-sided P < .05. Statistical analysis was performed using R, version 4.3.1 (R Project for Statistical Computing).
Results
A total of 292 453 KPSC patients between 18 and 40 years of age met criteria for inclusion (mean [SD] age, 29.8 [6.4] years) (eFigure in Supplement 1). Among 12 050 identified patients with referable glaucoma (4.1%), 52.3% were female, 47.7% were male, 17.7% were Asian, 9.4% were Black, 43.4% were Hispanic, 19.2% were White, and 7.6% were of other race and ethnicity (Table 1). Of these referred patients, 6827 (56.7%) were evaluated with either VF or OCT testing within 2 years. Evaluated patients had greater mean (SD) numbers of ophthalmologic visits (2.2 [4.9]) compared with the general population (0.5 [1.6]) and the referred group (1.7 [4.4]) (P < .001). However, they had a similar mean (SD) number of optometry visits (1.5 [1.7]) compared with the general population (1.4 [1.2]; P = .44) and the referred group (1.5 [1.6]; P = .11). 2344 of 12 050 patients with referable glaucoma (19.5%), 1045 of 6827 patients with evaluated glaucoma (15.3%), and 187 of 563 patients with diagnosed glaucoma (33.2%) had an IOP greater than 21 mm Hg.
Table 1. Referred, Evaluated, and Diagnosed Patients With Glaucoma Stratified by Age and Sex.
| Age range, y | Female | Male | Overall | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Referred, No. | Evaluated, No. | Diagnosed, No. | With glaucoma, % | Referred, No. | Evaluated, No. | Diagnosed, No. | With glaucoma, % | Referred, No. | Evaluated, No. | Diagnosed, No. | With glaucoma, % | |
| 18-19 | 181 | 91 | 2 | 2.2 | 196 | 96 | 5 | 5.2 | 377 | 187 | 7 | 3.7 |
| 20-29 | 2206 | 1154 | 48 | 4.2 | 1853 | 974 | 69 | 7.1 | 4059 | 2128 | 117 | 5.5 |
| 30-40 | 3914 | 2297 | 173 | 7.5 | 3700 | 2215 | 266 | 12.0 | 7614 | 4512 | 439 | 9.7 |
| Overall | 6301 | 3542 | 223 | 6.3 | 5749 | 3285 | 340 | 10.4 | 12 050 | 6827 | 563 | 8.2 |
Among the 6827 evaluated patients, there were 1439 (21.1%) who were Asian, 595 (8.7%) who were Black, 2827 (41.4%) who were Hispanic, 1257 (18.4%) who were White, and 530 (7.8%) who were of other race and ethnicity (eTable 2 in Supplement 1). There were a total of 563 patients (8.2%) who were diagnosed with glaucoma (344 [61.1%] with open angle, 28 [5.0%] with angle closure, 84 [14.9%] with secondary glaucoma, and 107 [19.0%] with unspecified glaucoma). Patients with diagnosed glaucoma had a greater prevalence of moderate and high myopia (40.0% vs 29.0%; P < .001), a higher mean (SD) IOP (22.3 [9.8] mm Hg vs 17.2 [4.7] mm Hg; P < .001), and a larger mean (SD) CDR (0.7 [0.2] vs 0.6 [0.2]; P < .001). On multivariable analysis (Table 2), male sex (OR, 1.55 [95% CI, 1.07-2.27]; P = .02), moderate and high myopia (OR, 2.24 [95% CI, 1.29-4.05]; P = .005), an elevated IOP (OR, 1.19 [95%CI, 1.15-1.23] per 1 mm Hg; P < .001), and a greater CDR (OR, 1.53 [95% CI, 1.34-1.75] per 0.1 unit; P < .001) were associated with greater odds of glaucoma (area under the receiver operating characteristic curve, 0.77 [95% CI, 0.73-0.82]; P ≤ .02).
Table 2. Univariable and Multivariable Logistic Regression Analyses of Baseline Factors Associated With Glaucoma Detection.
| Characteristic | Total evaluated patients, No. | Evaluated patients diagnosed with glaucoma, No. (%) | Analysis | |||
|---|---|---|---|---|---|---|
| Univariable | Multivariable | |||||
| OR (95% CI) | P value | OR (95% CI) | P value | |||
| Age, y | NA | NA | 1.06 (1.04-1.08) | <.001 | 1.06 (1.03-1.10) | <.001 |
| Sex | ||||||
| Female | 3542 | 223 (6.3) | 1 [Reference] | NA | 1 [Reference] | NA |
| Male | 3285 | 340 (10.4) | 1.72 (1.44-2.05) | <.001 | 1.55 (1.07-2.27) | .02 |
| Race and ethnicity | ||||||
| Asian | 1439 | 124 (8.6) | 1.05 (0.80-1.37) | .75 | 1.41 (0.73-2.81) | .31 |
| Black | 595 | 87 (14.6) | 1.90 (1.40-2.57) | <.001 | 1.79 (0.88-3.72) | .11 |
| Hispanic | 2827 | 187 (6.6) | 0.79 (0.61-1.01) | .06 | 1.02 (0.56-1.94) | .96 |
| White | 1257 | 104 (8.3) | 1 [Reference] | NA | 1 [Reference] | NA |
| Othera | 530 | 38 (7.2) | 0.86 (0.58-1.25) | .43 | 0.92 (0.39-2.13) | .86 |
| Unknown | 179 | 23 (12.8) | 1.63 (0.99-2.60) | .05 | 2.04 (0.68-5.45) | .17 |
| Spherical equivalent | ||||||
| Emmetropia | 878 | 32 (4.1) | 1 [Reference] | NA | 1 [Reference] | NA |
| Hyperopia | 254 | 11 (4.3) | 1.31 (0.63-2.56) | .44 | 1.20 (0.43-2.96) | .71 |
| Low myopia | 1512 | 91 (6.0) | 1.71 (1.12-2.68) | .02 | 1.36 (0.78-2.44) | .29 |
| Moderate and high myopia | 1041 | 88 (8.5) | 2.45 (1.61-3.85) | <.001 | 2.24 (1.29-4.05) | .005 |
| IOP, mm Hg | NA | NA | 1.12 (1.11-1.14) | <.001 | 1.19 (1.15-1.23) | <.001 |
| CDR, per 0.1 unit | NA | NA | 1.41 (1.30-1.53) | <.001 | 1.53 (1.34-1.75) | <.001 |
Abbreviations: CDR, cup-disc ratio; IOP, intraocular pressure; NA, not applicable; OR, odds ratio.
Includes but is not limited to American Indian or Alaska Native and Native Hawaiian or Other Pacific Islander.
Dichotomization cutoffs were based on the mean (SD) of age (32.0 [6.0] years), IOP (17.6 [5.5] mm Hg), and CDR (0.7 [0.2]). There were 16.0% to 28.4% of the evaluated group among each of the 4 risk groups created using age and IOP. Only 51 of 1613 patients (3.2%) in the low-risk age and IOP group were diagnosed with glaucoma compared with 202 of 1477 (13.7%) in the high-risk age and IOP group. CDR-specific risk of detected glaucoma remained lower than the overall risk (8.2%), with the CDR less than 0.8 for all risk groups except the high-risk age and IOP group (Figure). The multivariable analysis with dichotomized age, IOP, and CDR was similar to the undichotomized model, except Black race remained significantly associated with glaucoma diagnosis (eTable 3 in Supplement 1). Among 309 patients, aged 32 years or older with a CDR of 0.7 or higher and an IOP of 18 mm Hg or higher, there was a 23.3% positive predictive value of glaucoma detection compared with a 2.0% positive predictive value of glaucoma detection among 768 patients in the lowest risk for age, CDR, and IOP (Table 3).
Figure. Proportion of Evaluated Patients Detected to Have Glaucoma Within 2 Years After Referable Glaucoma Diagnosis, Stratified by Dichotomized Age and Intraocular Pressure (IOP) Risk Group.
The overall risk of glaucoma (8.2%) is indicated by the horizontal dotted black line. Low-risk age includes patients younger than 32 years; low-risk IOP, less than 18 mm Hg; high-risk age, 32 years or older; high-risk IOP, 18 mm Hg or more. Patients with a cup-disc ratio less than 0.7 were considered low risk; 0.7 or more was considered high risk.
Table 3. Negative and Positive Predictive Values of Age, IOP, and CDR for Glaucoma Detection.
| Outcome measure | TN or TP, No. | FN or FP, No. | Predictive value, %a |
|---|---|---|---|
| Negative results | |||
| Age <32 y | 2824 | 182 | 93.9 |
| IOP <18 mm Hg | 3279 | 185 | 94.7 |
| CDR <0.7 | 3049 | 141 | 95.6 |
| Age <32 y and IOP <18 mm Hg | 1562 | 51 | 96.8 |
| Age <32 y and CDR <0.7 | 1325 | 38 | 97.2 |
| IOP <18 mm Hg and CDR <0.7 | 1813 | 46 | 97.5 |
| Age <32 y and IOP <18 mm Hg and CDR <0.7 | 754 | 14 | 98.2 |
| Positive results | |||
| Age ≥32 y | 391 | 3440 | 10.2 |
| IOP ≥18 mm Hg | 300 | 2267 | 11.7 |
| CDR ≥0.7 | 192 | 1505 | 11.3 |
| Age ≥32 y and IOP ≥18 mm Hg | 202 | 1275 | 13.7 |
| Age ≥32 y and CDR ≥0.7 | 128 | 781 | 14.1 |
| IOP ≥18 mm Hg and CDR ≥0.7 | 114 | 445 | 20.4 |
| Age ≥32 y and IOP ≥18 mm Hg and CDR ≥0.7 | 72 | 237 | 23.3 |
Abbreviations: CDR, cup-disc ratio; FN, false negative or number of patients who had an incorrect negative prediction; FP, false positive or number of patients who had an incorrect positive prediction; IOP, intraocular pressure; TN, true negative or number of patients who had a correct negative prediction; TP, true positive or number of patients who had a correct positive prediction.
The formula for a negative predictive value is TN/(TN + FN); the formula for a positive predictive value is TP/(TP + FP).
Among the 6827 evaluated patients, 28.2% had a predicted risk below 2.5%, 56.4% had a predicted risk below 5.0%, and 91.1% had a predicted risk below 10.0%, based on dichotomous age and IOP and continuous CDR (Figure). Among 325 diagnosed patients with age, IOP, and CDR data, 24 patients (7.4%) had a predicted risk below 2.5%. Upon manual review of clinic notes from the 24 patients with very low risk, 9 (37.5%) received IOP-lowering treatments, 3 (12.5%) were diagnosed with normal tension glaucoma, 3 (12.5%) were diagnosed with secondary or angle-closure glaucoma, 2 (8.3%) were patients with large discrepancies in CDR compared with later visits, and 1 (4.2%) had suspected glaucoma that was miscoded as open-angle glaucoma.
Among the 6827 who received testing within 2 years, 1025 (15.0%) had VF testing only, 1544 (22.6%) had OCT testing only, and 4258 (62.4%) had both VF and OCT. Among 12 050 patients with referable glaucoma, 5223 (43.3%) were lost to follow-up and did not receive VF testing or OCT within 2 years. Although Black race was associated with higher odds (OR, 1.90 [95% CI, 1.40-2.57]; P < .001) of being diagnosed with glaucoma compared with White race, on multivariable analysis (Table 4), those with higher odds of being lost to follow-up (area under the receiver operating characteristic curve, 0.68 [95% CI, 0.67-0.70]) included patients who were Black (OR, 1.45 [95% CI, 1.16-1.82]; P = .001) and Hispanic (OR, 1.19 [95% CI, 1.01-1.39]; P = .04) and those who had a smaller CDR (OR, 1.52 [95% CI, 1.47-1.59] per 0.1 decrease; P < .001) and low myopia (OR, 1.15 [95% CI, 1.00-1.33]; P = .05).
Table 4. Univariable and Multivariable Logistic Regression Analyses of Baseline Factors Associated With Loss to Follow-Up.
| Characteristic | Total referred patients, No. | Referred patients lost to follow-up, No. (%) | Univariable analysis | Multivariable analysis | ||
|---|---|---|---|---|---|---|
| OR (95% CI) | P value | OR (95% CI) | P value | |||
| Age, y | NA | NA | 0.97 (0.97-0.98) | <.001 | 0.97 (0.96-0.98) | <.001 |
| Sex | ||||||
| Female | 6301 | 2759 (43.8) | 1 [Reference] | NA | 1 [Reference] | NA |
| Male | 5749 | 2464 (42.9) | 0.96 (0.89-1.03) | .31 | 1.11 (0.99-1.25) | .07 |
| Race and ethnicity | ||||||
| Asian | 2130 | 691 (32.4) | 0.57 (0.50-0.64) | <.001 | 0.83 (0.68-1.02) | .07 |
| Black | 1128 | 533 (47.3) | 1.06 (0.92-1.22) | .41 | 1.45 (1.16-1.82) | .001 |
| Hispanic | 5227 | 2400 (45.9) | 1.01 (0.91-1.11) | .91 | 1.19 (1.01-1.39) | .04 |
| White | 2318 | 1061 (45.8) | 1 [Reference] | NA | 1 [Reference] | NA |
| Othera | 327 | 148 (45.3) | 0.87 (0.75-1.02) | .08 | 1.05 (0.83-1.35) | .64 |
| Unknown | 920 | 390 (42.4) | 0.98 (0.78-1.23) | .86 | 1.08 (0.75-1.54) | .70 |
| Spherical equivalent | ||||||
| Emmetropia | 1615 | 737 (45.6) | 1 [Reference] | NA | 1 [Reference] | NA |
| Hyperopia | 532 | 278 (52.3) | 1.11 (0.91-1.35) | .31 | 1.18 (0.93-1.49) | .19 |
| Low myopia | 3095 | 1583 (51.1) | 1.09 (0.96-1.23) | .19 | 1.15 (1.00-1.33) | .05 |
| Moderate and high myopia | 1863 | 822 (44.1) | 0.82 (0.71-0.93) | .004 | 0.93 (0.80-1.10) | .45 |
| IOP, mm Hg | NA | NA | 1.00 (0.99-1.00) | .33 | 0.94 (0.93-0.96) | <.001 |
| CDR, per 0.1 unit | NA | NA | 0.72 (0.70-0.74) | <.001 | 0.66 (0.63-0.68) | <.001 |
Abbreviations: CDR, cup-disc ratio; IOP, intraocular pressure; NA, not applicable; OR, odds ratio.
Includes American Indian or Alaska Native and Native Hawaiian or Other Pacific Islander.
Discussion
In this retrospective cohort study of adults aged 18 to 40 years from a large managed health care system, only 8.2% of patients who were referred and received a glaucoma evaluation were diagnosed with glaucoma. Black and Hispanic patients were at higher risk for not receiving glaucoma evaluation after referral. A simple 3-variable model based on age, IOP, and CDR effectively stratified patients with low risk (3.2%) and high risk (13.7%) of being diagnosed with glaucoma. Our model demonstrated that implementing a 2.5% risk threshold for glaucoma referral in adults aged 18 to 40 years could reduce such referrals by 28.2%. This study provided a framework for evaluating the outcomes of glaucoma referrals and establishing standardized risk-stratification guidelines that may improve the equity of eye care and efficiency of resource utilization.
Our findings raise questions about the effectiveness of glaucoma referrals among adults aged 18 to 40 years in the absence of standardized referral guidelines. Previous studies of older adults over 40 years of age, including those in Los Angeles County, reported higher diagnosis rates of up to 25.0% following referral for specialist evaluation.6,20 One obvious explanation for the low yield of glaucoma evaluations in our study cohort is the low overall prevalence of glaucoma among these adults aged 18 to 40 years, ranging between 0.16% and 0.4% in previous studies.21,22 Additionally, tools developed to risk stratify older adults may not generalize to adults aged 18 to 40 years, leaving clinicians with few options available to discern which patients are at sufficient risk to warrant referral.3,23
Given that the KPSC glaucoma-referral process is typically initiated by optometrists, we elected to model only factors that are routinely assessed in general eye-care settings. The lowest-risk group, which satisfied all 3 low-risk criteria (aged <32 years, IOP <18 mm Hg, and CDR <0.7), comprised 11.2% of evaluated patients (n = 768 of 6827) and carried a negative predictive value of 98.2%. These results demonstrate that risk stratification of adults aged 18 to 40 years in the KPSC system is feasible using readily accessible data. Assessing glaucoma risk factors that require specialized equipment, such as visual fields, refractive error, and central corneal thickness, could improve the accuracy of diagnosis but may be time or cost prohibitive in a screening setting.7,24,25 Therefore, the marginal diagnostic benefit compared with the cost of assessing each risk factor should be thoroughly considered in the context of available health care resources prior to implementation.
We intentionally avoided ascribing clinical significance to specific glaucoma risk thresholds, as establishing such thresholds is a complex issue that varies by patient population and health care system. Value-based approaches, in which the benefit of a diagnostic test is quantified, using standardized measurements such as quality-adjusted life-year or an incremental cost-effectiveness ratio, have been applied to other complex clinical decisions.26 Risk tolerance must also be discussed between the clinician and patient, and decisions to refer should be made on an individualized basis while considering the complete clinical picture. However, implementing evidence-based referral thresholds could have significant resource-conserving benefits. For example, implementing a 5.0% risk threshold could reduce the existing number of KPSC glaucoma referrals among adults aged 18 to 40 years by 56.4%. While KPSC may have sufficient eye-care resources to accommodate a more stringent risk threshold, there are resource-constrained safety-net health care systems that must prioritize the care of a smaller number of individuals with higher glaucoma risk.
Our data suggest that the increased workload of evaluating referable glaucoma falls primarily on ophthalmologists rather than optometrists, a concerning finding given the low yield of evaluations and worsening ophthalmologic workforce shortage.27 Patients evaluated for glaucoma had the same number of optometry visits during the study period but a greater number of ophthalmologic visits compared with the adults aged 18 to 40 years in our cohort. This increased use of ophthalmologic appointments may represent a suboptimal allocation of clinician resources, given that 91.8% of all evaluated patients were not diagnosed with glaucoma. Given the significant decline in ophthalmologist density per 100 000 people from 1995 to 2017 in the US, enhancements to glaucoma-referral pathways that enable ophthalmologists to focus on higher-level patient care compared with low-yield evaluations should be considered.15 For example, KPSC runs a low-resource teleophthalmologic monitoring program that follows individuals with suspected glaucoma with low risk after initial evaluation and detection.28
While there is a need to improve glaucoma-referral practices by decreasing the frequency of evaluations for patients with low risk, some patients with high risk are not receiving evaluations at all. Just under half (43.3%) of the patients in our study (n = 5223 of 12 050) did not return for a glaucoma evaluation after referral, representing a glaring lost opportunity to detect glaucoma. Black and Hispanic patients were at higher risk of being lost to follow-up, even though Black race was associated with higher odds (OR, 1.90 [95% CI, 1.40-2.57]) of being diagnosed with glaucoma compared with White race. These findings are consistent with previous epidemiologic studies and are concerning given disparities in glaucoma detection, and outcomes are widely recognized.29,30,31,32 We provide evidence that racial and ethnic disparities may negatively impact the early trajectory of these patients during the critical period preceding their diagnosis, thus delaying potential detection and treatment.33 Finally, our findings highlight that health insurance coverage is only 1 of many social determinants of health, including health literacy and cultural beliefs, that may contribute to disparities in eye-care utilization and glaucoma outcomes. Further development of interventions that increase adherence to glaucoma referrals, such as patient education and appointment notifications, is urgently needed.34
Limitations
Our study has several limitations. First, glaucoma diagnoses in our study were derived from ICD-9 and ICD-10 codes that may have been affected by misdiagnosis or miscoding errors. This led us to perform an audit of clinic notes from patients diagnosed with glaucoma who were predicted to be at low risk by our model (most likely to be misdiagnosed or miscoded), which revealed only 1 patient with suspected glaucoma that was miscoded as primary open-angle glaucoma. The majority of the remaining patients were previously diagnosed and were receiving IOP-lowering treatment, which is not factored into our models, or experienced normal tension glaucoma, which is rare among adults aged 18 to 40 years.35 Overall, these findings are consistent with prior studies supporting the validity of using ICD codes to identify patients with glaucoma.36,37 Second, ICD-9 and ICD-10 codes may not provide accurate information about the laterality of a referable glaucoma diagnosis. Therefore, we opted to include the eye with more severe CDR and/or IOP in our analysis, which may have introduced sampling error. However, 66.1% of the patients in our study with glaucoma were diagnosed with either primary angle-closure glaucoma or primary open-angle glaucoma, which are typically bilateral diseases.38 Third, clinical measurements were acquired from free text entries in the EHRs, which sometimes lacked specificity (eg, vertical or horizontal CDR).4 This uncertainty is a limitation of current EHR systems and highlights the need for standardized data concepts to represent these measurements.39 Fourth, tonometry was performed using a range of different methods and devices in KPSC, which could have contributed to false positives or negatives compared with a single approach.40 However, the benefit of standardizing tonometric procedures in the risk-stratification process requires further study prior to widespread implementation. Finally, our window for glaucoma diagnosis spanned only 2 years after the first referable glaucoma diagnosis; therefore, patients with longer intervals before receiving a glaucoma evaluation could have been missed. Similarly, the care of patients whose 2-year windows extended into 2020 could have been disrupted by the COVID-19 pandemic.
Conclusions
The findings of this cohort study provide insight into the low yield of glaucoma evaluations in an adult population aged 18 to 40 years in the absence of standardized evidence-based referral guidelines. Further work is needed using longitudinal data to assess the optimal frequency of reevaluation for referred and evaluated patients. In addition, efforts to improve glaucoma care must continue to be evaluated for fairness and bias to ensure racial and ethnic minority groups receive equitable care. This study provides a crucial step toward standardizing glaucoma-referral protocols, reducing low-yield eye-care expenditures, and ensuring reproducible and equitable care in diverse patient populations.
eFigure. Attrition Diagram and Study Group Definitions
eTable 1. Referable Glaucoma Diagnosis Codes and Procedural Codes
eTable 2. Glaucoma Diagnoses at the Last Visit Stratified by Race and Subtype
eTable 3. Univariable and Multivariable Analysis of Risk Factors for Detection of Glaucoma in Which Predictors Age, Intraocular Pressure (IOP), and Cup-to-Disc Ratio (CDR) Are Dichotomized at the Mean
Data Sharing Statement
References
- 1.Tham YC, Li X, Wong TY, Quigley HA, Aung T, Cheng CY. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmology. 2014;121(11):2081-2090. doi: 10.1016/j.ophtha.2014.05.013 [DOI] [PubMed] [Google Scholar]
- 2.Chang RT, Singh K. Glaucoma suspect: diagnosis and management. Asia Pac J Ophthalmol (Phila). 2016;5(1):32-37. doi: 10.1097/APO.0000000000000173 [DOI] [PubMed] [Google Scholar]
- 3.Gordon MO, Beiser JA, Brandt JD, et al. The Ocular Hypertension Treatment Study: baseline factors that predict the onset of primary open-angle glaucoma. Arch Ophthalmol. 2002;120(6):714-720. doi: 10.1001/archopht.120.6.714 [DOI] [PubMed] [Google Scholar]
- 4.Gedde SJ, Lind JT, Wright MM, et al. ; American Academy of Ophthalmology Preferred Practice Pattern Glaucoma Panel . Primary Open-Angle Glaucoma Suspect Preferred Practice Pattern®. Ophthalmology. 2021;128(1):151-P192. doi: 10.1016/j.ophtha.2020.10.023 [DOI] [PubMed] [Google Scholar]
- 5.Jayaram H, Kolko M, Friedman DS, Gazzard G. Glaucoma: now and beyond. Lancet. 2023;402(10414):1788-1801. doi: 10.1016/S0140-6736(23)01289-8 [DOI] [PubMed] [Google Scholar]
- 6.Bowling B, Chen SDM, Salmon JF. Outcomes of referrals by community optometrists to a hospital glaucoma service. Br J Ophthalmol. 2005;89(9):1102-1104. doi: 10.1136/bjo.2004.064378 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Founti P, Topouzis F, Holló G, et al. Prospective study of glaucoma referrals across Europe: are we using resources wisely? Br J Ophthalmol. 2018;102(3):329-337. doi: 10.1136/bjophthalmol-2017-310249 [DOI] [PubMed] [Google Scholar]
- 8.Lockwood AJ, Kirwan JF, Ashleigh Z. Optometrists referrals for glaucoma assessment: a prospective survey of clinical data and outcomes. Eye (Lond). 2010;24(9):1515-1519. doi: 10.1038/eye.2010.77 [DOI] [PubMed] [Google Scholar]
- 9.Davuluru SS, Jess AT, Kim JSB, Yoo K, Nguyen V, Xu BY. Identifying, understanding, and addressing disparities in glaucoma care in the United States. Transl Vis Sci Technol. 2023;12(10):18. doi: 10.1167/tvst.12.10.18 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Halawa OA, Kolli A, Oh G, et al. Racial and socioeconomic differences in eye care utilization among Medicare beneficiaries with glaucoma. Ophthalmology. 2022;129(4):397-405. doi: 10.1016/j.ophtha.2021.09.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Apolo G, Bohner A, Pardeshi A, et al. Racial and sociodemographic disparities in the detection of narrow angles before detection of primary angle-closure glaucoma in the United States. Ophthalmol Glaucoma. 2022;5(4):388-395. doi: 10.1016/j.ogla.2022.01.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Shah SN, Zhou S, Sanvicente C, et al. Prevalence and risk factors of blindness among primary angle closure glaucoma patients in the United States: an IRIS Registry analysis. Am J Ophthalmol. 2024;259:131-140. doi: 10.1016/j.ajo.2023.11.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rein DB, Zhang P, Wirth KE, et al. The economic burden of major adult visual disorders in the United States. Arch Ophthalmol. 2006;124(12):1754-1760. doi: 10.1001/archopht.124.12.1754 [DOI] [PubMed] [Google Scholar]
- 14.Leshno A, Liebmann JM. The glaucoma suspect problem: ways forward. Asia Pac J Ophthalmol (Phila). 2022;11(6):503-504. doi: 10.1097/APO.0000000000000564 [DOI] [PubMed] [Google Scholar]
- 15.Feng PW, Ahluwalia A, Feng H, Adelman RA. National trends in the United States eye care workforce from 1995 to 2017. Am J Ophthalmol. 2020;218:128-135. doi: 10.1016/j.ajo.2020.05.018 [DOI] [PubMed] [Google Scholar]
- 16.Southern California Research Program. Kaiser Permanente . December 31, 2023. Accessed October 28, 2024. https://www.kp-scalresearch.org/aboutus/fast-facts/
- 17.Allison K, Patel D, Alabi O. Epidemiology of glaucoma: the past, present, and predictions for the future. Cureus. 2020;12(11):e11686. doi: 10.7759/cureus.11686 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.World Medical Association . World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 2013;310(20):2191-2194. doi: 10.1001/jama.2013.281053 [DOI] [PubMed] [Google Scholar]
- 19.Flitcroft DI, He M, Jonas JB, et al. IMI—defining and classifying myopia: a proposed set of standards for clinical and epidemiologic studies. Invest Ophthalmol Vis Sci. 2019;60(3):M20-M30. doi: 10.1167/iovs.18-25957 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yuen J, Xu B, Song BJ, Daskivich LP, Rodman J, Wong BJ. Effectiveness of glaucoma screening and factors associated with follow-up adherence among glaucoma suspects in a safety-net teleretinal screening program. Ophthalmol Glaucoma. 2023;6(3):247-254. doi: 10.1016/j.ogla.2022.10.007 [DOI] [PubMed] [Google Scholar]
- 21.Marx-Gross S, Laubert-Reh D, Schneider A, et al. The prevalence of glaucoma in young people. Dtsch Arztebl Int. 2017;114(12):204-210. doi: 10.3238/arztebl.2017.0204 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Werner DL, Sherman J. Prevalence of glaucoma in optometric patients under 40 years of age. Am J Optom Physiol Opt. 1977;54(11):767-770. doi: 10.1097/00006324-197711000-00004 [DOI] [PubMed] [Google Scholar]
- 23.Heijl A, Bengtsson B, Oskarsdottir SE. Prevalence and severity of undetected manifest glaucoma: results from the early manifest glaucoma trial screening. Ophthalmology. 2013;120(8):1541-1545. doi: 10.1016/j.ophtha.2013.01.043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Gedde SJ, Vinod K, Wright MM, et al. ; American Academy of Ophthalmology Preferred Practice Pattern Glaucoma Panel . Primary Open-Angle Glaucoma Preferred Practice Pattern®. Ophthalmology. 2021;128(1):71-P150. doi: 10.1016/j.ophtha.2020.10.022 [DOI] [PubMed] [Google Scholar]
- 25.Gedde SJ, Chen PP, Muir KW, et al. ; American Academy of Ophthalmology Preferred Practice Pattern Glaucoma Panel . Primary Angle-Closure Disease Preferred Practice Pattern®. Ophthalmology. 2021;128(1):30-P70. doi: 10.1016/j.ophtha.2020.10.021 [DOI] [PubMed] [Google Scholar]
- 26.Newman-Toker DE, McDonald KM, Meltzer DO. How much diagnostic safety can we afford, and how should we decide? A health economics perspective. BMJ Qual Saf. 2013;22(suppl 2)(suppl 2):ii11-ii20. doi: 10.1136/bmjqs-2012-001616 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Resnikoff S, Felch W, Gauthier TM, Spivey B. The number of ophthalmologists in practice and training worldwide: a growing gap despite more than 200,000 practitioners. Br J Ophthalmol. 2012;96(6):783-787. doi: 10.1136/bjophthalmol-2011-301378 [DOI] [PubMed] [Google Scholar]
- 28.Modjtahedi BS, Chu K, Luong TQ, et al. Two-year outcomes of a pilot glaucoma suspect telemedicine monitoring program. Clin Ophthalmol. 2018;12:2095-2102. doi: 10.2147/OPTH.S171263 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Williams AM, Huang W, Muir KW, Stinnett SS, Stone JS, Rosdahl JA. Identifying risk factors for blindness from primary open-angle glaucoma by race: a case-control study. Clin Ophthalmol. 2018;12:377-383. doi: 10.2147/OPTH.S143417 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Murakami Y, Lee BW, Duncan M, et al. Racial and ethnic disparities in adherence to glaucoma follow-up visits in a county hospital population. Arch Ophthalmol. 2011;129(7):872-878. doi: 10.1001/archophthalmol.2011.163 [DOI] [PubMed] [Google Scholar]
- 31.Varma R, Ying-Lai M, Francis BA, et al. ; Los Angeles Latino Eye Study Group . Prevalence of open-angle glaucoma and ocular hypertension in Latinos: the Los Angeles Latino Eye Study. Ophthalmology. 2004;111(8):1439-1448. doi: 10.1016/j.ophtha.2004.01.025 [DOI] [PubMed] [Google Scholar]
- 32.Tielsch JM, Sommer A, Katz J, Royall RM, Quigley HA, Javitt J. Racial variations in the prevalence of primary open-angle glaucoma. The Baltimore Eye Survey. JAMA. 1991;266(3):369-374. doi: 10.1001/jama.1991.03470030069026 [DOI] [PubMed] [Google Scholar]
- 33.Paz SH, Varma R, Klein R, Wu J, Azen SP; Los Angeles Latino Eye Study Group . Noncompliance with vision care guidelines in Latinos with type 2 diabetes mellitus: the Los Angeles Latino Eye Study. Ophthalmology. 2006;113(8):1372-1377. doi: 10.1016/j.ophtha.2006.04.018 [DOI] [PubMed] [Google Scholar]
- 34.Kim YK, Jeoung JW, Park KH. Understanding the reasons for loss to follow-up in patients with glaucoma at a tertiary referral teaching hospital in Korea. Br J Ophthalmol. 2017;101(8):1059-1065. doi: 10.1136/bjophthalmol-2016-309103 [DOI] [PubMed] [Google Scholar]
- 35.Razeghinejad MR, Lee D. Managing normal tension glaucoma by lowering the intraocular pressure. Surv Ophthalmol. 2019;64(1):111-116. doi: 10.1016/j.survophthal.2018.07.003 [DOI] [PubMed] [Google Scholar]
- 36.Holt J, Warsy A, Wright P. Medical decision making: guide to improved CPT coding. South Med J. 2010;103(4):316-322. doi: 10.1097/SMJ.0b013e3181d2f19b [DOI] [PubMed] [Google Scholar]
- 37.Muir KW, Gupta C, Gill P, Stein JD. Accuracy of international classification of diseases, ninth revision, clinical modification billing codes for common ophthalmic conditions. JAMA Ophthalmol. 2013;131(1):119-120. doi: 10.1001/jamaophthalmol.2013.577 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Broman AT, Quigley HA, West SK, et al. Estimating the rate of progressive visual field damage in those with open-angle glaucoma, from cross-sectional data. Invest Ophthalmol Vis Sci. 2008;49(1):66-76. doi: 10.1167/iovs.07-0866 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Hallaj S, Khawaja AP, Rodrigues IAS, et al. Gap analysis of glaucoma examination concept representations within standard Systemized Nomenclature of Medicine–Clinical Terms. Ophthalmol Glaucoma. Published online August 13, 2024. doi: 10.1016/j.ogla.2024.08.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Güler M, Bilak Ş, Bilgin B, Şimşek A, Çapkin M, Hakim Reyhan A. Comparison of intraocular pressure measurements obtained by iCare PRO rebound tonometer, Tomey FT-1000 noncontact tonometer, and Goldmann applanation tonometer in healthy subjects. J Glaucoma. 2015;24(8):613-618. doi: 10.1097/IJG.0000000000000132 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure. Attrition Diagram and Study Group Definitions
eTable 1. Referable Glaucoma Diagnosis Codes and Procedural Codes
eTable 2. Glaucoma Diagnoses at the Last Visit Stratified by Race and Subtype
eTable 3. Univariable and Multivariable Analysis of Risk Factors for Detection of Glaucoma in Which Predictors Age, Intraocular Pressure (IOP), and Cup-to-Disc Ratio (CDR) Are Dichotomized at the Mean
Data Sharing Statement