Impact of Demographics on Regional Visual Field Loss and Deterioration in Glaucoma

Abstract

Purpose: To elucidate the impact of demographics, including gender, race, ethnicity, and preferred language, on regional visual field (VF) loss and progression in glaucoma.

Methods: Multivariable linear mixed regressions were performed to determine the impact of race, ethnicity, and preferred language on regional VF loss with adjustment for age and gender. Regional VF loss was defined by pointwise total deviation values and VF loss patterns quantified by an unsupervised machine learning method termed archetypal analysis. All cross-sectional and longitudinal analyses were performed both without and with adjustment for VF mean deviation, which represented overall VF loss severity. P values were corrected for multiple comparisons.

Results: All results mentioned had corrected P values less than 0.05. Asian and Black patients showed worse pointwise VF loss than White patients with superior hemifield more affected. Patients with a preferred language other than English demonstrated worse pointwise VF loss than patients with English as their preferred language. Longitudinal analyses revealed Black patients showed worse VF loss/year compared to White patients. Patients with a preferred language other than English demonstrated worse VF loss/year compared to patients preferring English.

Conclusions: Blacks and non-English speakers have more severe VF loss, with superior hemifield being more affected and faster VF worsening.

Translational Relevance: This study furthered our understanding of racial, ethnic, and socioeconomic disparities in glaucoma outcomes. Understanding the VF loss burden in different racial, ethnic, and socioeconomic groups may guide more effective glaucoma screening and community outreach efforts. This research could help reduce vision loss and improve quality of life in disproportionately affected populations by guiding public health efforts to promote glaucoma awareness and access to care.

Introduction

Glaucoma is the second leading cause of blindness worldwide and a significant public health concern in the United States. Racial, ethnic, gender, and age-based disparities in the prevalence and progression are well documented: the prevalence of glaucoma is consistently higher among Black patients than among White patients, and primary open-angle glaucoma is more prevalent in Asian patients than among White patients.^1–5 Female patients and older patients experience a higher likelihood of being diagnosed with glaucoma.^6,7 Worse baseline severity of glaucoma also exists among patients with preferred languages other than English.^8,9 

Disparities also exist in the impact of glaucoma on progressive visual field loss among patients with different demographic characteristics. Black patients experience higher baseline visual field (VF) loss and faster VF deterioration than their White counterparts, yet also receive visual field testing less frequently.⁴ A separate cross-sectional study reported that Black individuals faced higher risks of having early central and advanced VF loss compared with non-Hispanic Whites.¹⁰ However, demographic-based disparities in the longitudinal effects of glaucoma are much less well understood than disparities in the mere prevalence of glaucoma. 

Although prior work has shown that specific VF regions are more susceptible to damage in certain clinical subtypes of glaucoma (e.g., primary open-angle glaucoma is often associated with a faster rate of VF loss in the superior hemifield), any differences in such associations between different demographic groups remain poorly characterized.¹¹ In addition, a previous study shows that in primary open-angle glaucoma, greater disc ovality, larger beta-peripapillary atrophy area, and thinner central cornea thickness are associated with initial visual field defects starting in the superior hemifield, suggesting potential differences in the glaucomatous damaging process between the superior and inferior halves of the optic disc.¹² Other studies provide insights into the functional differences between the upper and lower visual fields at the neural level, demonstrating that the superior colliculus, an important structure involved in visual processing and eye movements, exhibits enhanced sensitivity and representation for the upper visual field compared to the lower visual field.^13,14 Studying the impact of glaucoma on different visual field regions could provide insights into the interplay between the structural and functional changes associated with the disease progression. Loss of specific VF regions can also have varying effects on patients’ quality of life, as different regions are involved in tasks such as walking or reading.¹⁵ Identifying these regional differences can aid in subtype diagnosis, facilitating tailored screening procedures and targeted treatment strategies.¹⁶ 

This study aims to explore the impacts of various demographic parameters on glaucoma severity and longitudinal VF deterioration. Moreover, we look specifically at VF loss in different areas to investigate potential regional variations in glaucomatous damage progression. We employ linear mixed models to characterize demographic-based disparities in glaucomatous VF loss and deterioration. We also characterize regional differences in VF deterioration as they vary between demographics using two different outcome measures: pointwise total deviation values and 16 previously established VF loss patterns (Fig. 1) in glaucoma quantified by an unsupervised machine learning method termed archetypal analysis.^17–19 

Figure 1.

 

Sixteen VF loss patterns quantified by archetypal analysis in our prior works.

Methods

This retrospective cross-sectional and longitudinal study was conducted using 24-2 VFs from Massachusetts Eye and Ear (MEE). This study was approved by the MEE Institutional Review Board and adhered to the tenets of the Declaration of Helsinki and all federal and state laws, including the Health Insurance Portability and Accountability Act of 1996. 

Participants and Data

Reliable Swedish interactive thresholding algorithm standard 24-2 VFs with stimulus size III measured by the Humphrey Field Analyzer (Carl Zeiss Meditec, Dublin, CA, USA) were selected for our data analysis. The total deviation (TD) values from each of the 52 locations tested in the 24-2 pattern were extracted. Random eye was selected from all patients, but all VF results were analyzed in the right-eye format to standardize the data. Only reliable VF tests (false-positive rate ≤20%, false-negative rate ≤20%, and fixation loss rate ≤33%)^20–25 were finally included in our data analyses. 

Electronic medical records were used to obtain demographic characteristics for all study participants. All data were deidentified, thereby eliminating the need for direct patient consent. Patients were asked to fill out a form in the clinic documenting demographic characteristics, including race, ethnicity, gender, and preferred languages. Ethnicity data were obtained by a question for patients to identify as “Hispanic” or “non-Hispanic.” Race is a question asking patients to choose from “American Indian or Alaska Native,” “Asian,” “Black or African American,” “Latino,” “Native Hawaiian or Other Pacific Islander,” “Other,” and “White or Caucasian.” Categorical independent variables for this analysis were organized as follows: for the purposes of adequate sample size, only Asian, Black, and White racial categories were used; for similar reasons related to sample size, only subjects with Hispanic or non-Hispanic categories were included, and preferred language was limited to English, Spanish, and Other to represent all other preferred languages; gender was Male, Female, or Other/Unknown. Information on the patient’s age was also collected. 

For the cross-sectional analyses, the first reliable VF of each eye was included. For the longitudinal analyses, eyes with at least five reliable VFs across at least 4 years were included. 

Statistical Analyses

Analysis was conducted using multivariable mixed linear regression models. For the cross-sectional analysis, multivariable linear mixed analyses were used to associate age, gender, race, ethnicity, and preferred language with regional VF loss measured by pointwise TD values and archetypal VF pattern coefficients. Sixteen previously established VF loss (Fig. 1) archetypes were used to analyze VF loss in specific areas.^17,19 For the longitudinal analysis, multivariable linear mixed analyses were used to associate age, gender, race, ethnicity, and preferred language with regional VF worsening measured by TD slopes and archetypal VF pattern slopes. All results were adjusted with intraocular pressure (IOP) to ensure patients are in the same stage of glaucoma since IOP is a measure for glaucomatous progression.^26,27 All statistical analyses were conducted using R Statistical Software (R Project for Statistical Computing, Vienna, Austria). 

Results

A total of 200,810 Humphrey VF measurements collected from 100,405 patients were used for cross-sectional analyses (Table 1). Approximately half were female (58,479; 58.2%), and most self-identified as White/Caucasian (70,697; 70.4%). The mean deviation (MD) for Whites was –3.60 ± 5.29 dB, as compared to –5.46 ± 6.83 dB for Blacks, –3.92 ± 5.40 dB for Asians, and –4.52 ± 6.07 dB for the other races, which differed between racial groups significantly (P < 0.001). The vision loss MD for patients with English as the preferred language was –3.69 ± 5.38 dB, which was better (P < 0.001) than patients who preferred Spanish (–5.96 ± 6.74 dB). For the longitudinal analysis, 16,440 patients who were evaluated five or more times across at least 4 years are included. 

Table 1.

 

Data Characteristics

 

Cross-Sectional Study: Pointwise Analysis

In the cross-sectional analysis using the first VF of each eye (Fig. 2), male patients showed worse pointwise VF loss than female patients, up to –0.8 dB (P = 0.03), with the inferior hemifield more damaged. Compared with White patients, Asian patients showed worse pointwise VF loss, up to –0.4 dB (P = 0.02), with greater impairment in the superior hemifield; Black patients showed worse pointwise VF loss across all regions also with the superior hemifield more affected, as severe as –2.9 dB (P = 0.005) difference. Hispanic patients showed worse VF loss than their non-Hispanic counterparts, up to –0.3 dB (P = 0.003) in the superior peripheral region, and less VF loss, up to 0.3 dB (P = 0.01) in the inferior VF. Patients who preferred Spanish or other languages had worse VF loss than English speakers, across all regions, with differences in loss as severe as –3.4 and –2.6 dB, respectively (P = 0.01). This analysis was repeated and achieved similar results, with adjustment for MD (Supplementary Fig. S1), which allowed comparison between populations with the same general VF loss severity. 

Figure 2.

 

Demographic parameters’ impact on visual field loss in cross-sectional analysis (without MD adjustments).

Cross-Sectional Study: Archetype Analysis

As mentioned in the Methods section, 16 VF loss (Fig. 1) archetypes determined in our prior work were used to analyze regional VF loss patterns.^17,28 Note that any archetypal pattern coefficient is between 0% and 100%. As compared to female patients (Table 2), male patients showed the highest positive association with the near-total loss, inferior altitudinal defect, and temporal hemianopia patterns (archetypes 6, 13, and 12; P = 0.03, P = 0.001, and P < 0.001). Male patients exhibited the strongest negative association with normal VF, superior nasal step, and superior peripheral defect patterns (archetypes 1, 3, and 2; P < 0.001, P = 0.008, and P = 0.03). Compared to White patients, African American patients showed the highest positive association with the near-total loss, superior peripheral defect, and superior altitudinal defect patterns (archetypes 6, 2, and 8; P = 0.01, P = 0.04, and P = 0.002), and the strongest negative association for normal VF (archetype 1; P = 0.002). Asian patients showed the highest positive association with the temporal wedge defect and superior altitudinal defect patterns (archetypes 4 and 8; P = 0.04 and P < 0.001) and the strongest negative association for the inferior altitudinal defect, inferotemporal defect, and central scotoma patterns (archetypes 13, 9, and 7; P = 0.04, P = 0.03, and P = 0.002). 

Table 2.

 

Demographic Parameters’ Impact on Visual Field Loss With Archetypes in Cross-sectional Analysis (Without MD Adjustments)

 

Compared to non-Hispanic patients, Hispanic patients demonstrated the highest positive association with the superior peripheral defect pattern (archetype 2; P = 0.03) and the strongest negative association with the inferotemporal defect, inferior altitudinal defect, and superior paracentral defect patterns (archetypes 9, 13, and 14; P < 0.001, P = 0.01, and P = 0.006). Compared to patients with English as their preferred language, Spanish speakers showed the highest positive association with the near-total loss, superior peripheral defect, and concentric peripheral defect patterns (archetypes 6, 2, and 11; P = 0.03, P = 0.001, and P = 0.004) and the strongest negative association for normal VF (archetypes 1; P = 0.03). Archetypal analysis with adjustment for MD is shown in Supplementary Table S1. Our archetypal pattern analysis results were consistent with our pointwise TD analysis results. 

Longitudinal Study: Pointwise Analysis

As shown in Figure 3, males showed more VF loss deterioration up to –0.08 dB/y (P = 0.01). Compared to White patients, Asian patients showed less VF loss deterioration up to 0.12 dB/y (P = 0.02), and Black patients showed less VF loss deterioration in the nasal area up to –0.06 dB/y (P = 0.007) and more VF loss deterioration in the temporal area up to –0.05 dB/y (P = 0.04), respectively. Hispanics showed less VF loss deterioration everywhere up to 0.11 dB/y (P = 0.008). Compared to patients preferring English, patients preferring Spanish or other languages demonstrated greater VF loss deterioration up to –0.1 dB/y (P = 0.04) and –0.07 dB/y (P = 0.09), respectively.¹⁷ The results with MD adjustments are included in the supplemental materials (Supplementary Fig. S2). 

Figure 3.

 

Demographic parameters’ impact on visual field loss in longitudinal analysis (without MD adjustments).

Longitudinal Study: Archetypal Analysis

As compared to female patients (Table 3), male patients showed the highest positive association with increasing superonasal defects, temporal hemianopia, and inferonasal defect patterns (archetypes 3, 12, and 10; P = 0.02, P < 0.001, and P = 0.03). Compared to European-derived people, Asians exhibited increasing inferior paracentral loss (archetypes 16; P = 0.008), and Blacks demonstrated increasing superior paracentral defect (archetypes 14; P = 0.04). Noticeably, Spanish speakers were associated with increasing superior altitudinal and concentric peripheral defects (archetypes 8 and 11; P = 0.02, P < 0.001). The results with MD adjustments are included in the supplemental materials (Supplementary Table S2). 

Table 3.

 

Demographic Parameters’ Impact on Visual Field Loss With Archetypes in Longitudinal Analysis (Without MD Adjustments)

 

Discussion

Our results demonstrated significant differences in both VF loss patterns and worsening among patients with glaucoma of different demographics. Our cross-sectional results suggest that male patients experience more severe VF loss compared to females, Asian and Black patients experience more severe VF loss compared to White patients, and patients preferring Spanish and other languages group experience more severe VF loss compared to English speakers. Across groups, the superior hemisphere of the VF demonstrates more severe vision loss than its inferior counterpart. Our longitudinal analysis demonstrated similar results: compared to White patients, Asian patients had slower VF progression, while Black patients had faster VF progression. Patients with Spanish or other preferred languages experienced faster VF loss deterioration compared to English speakers, and male patients experienced faster VF loss deterioration than females. 

VF loss difference between genders could be attributed to factors such as hormonal differences, anatomic variations in optic nerve head structure, or variations in susceptibility to glaucomatous damage between genders. Regarding the longitudinal analysis, the slower VF progression observed in Asian patients compared to White patients could be related to genetic or anatomic protective factors, better adherence to treatment regimens, or differences in the underlying pathophysiology of glaucoma across these racial/ethnic groups.²⁹ On the other hand, the faster VF progression seen in Black patients compared to White patients may be due to increased susceptibility to glaucomatous damage, delayed diagnosis or treatment initiation, or the presence of comorbidities that exacerbate glaucoma progression in this population. Possible explanations for discrepancies between language groups include differences for those preferring non-English languages, such as a lack of access to health care³⁰ and reduced confidence to seek health care caused by negative past experiences.³¹ 

These results are again mirrored in our analyses examining regional VF loss over time: male patients experienced faster VF loss deterioration, mostly in the inferior hemisphere, than female patients. Asian patients experienced slower VF loss deterioration than White patients, while Black patients experienced faster VF loss deterioration, mostly in the right VF hemisphere. Patients preferring Spanish experienced slower central vision loss than patients preferring English, whereas patients with any other preferred language experienced faster central VF loss. 

Further research needs to be done to find out the cause behind these results, but possible factors that lead to the difference in the right VF hemisphere among different race groups can be explained by the right VF hemisphere (or right hemifield) corresponding to the visual field represented in the left side of the brain. There may be racial/ethnic differences in the underlying pathophysiology, risk factors, or comorbidities that preferentially affect or impact the left hemisphere/right VF in Black patients.²⁹ The pattern of VF loss initiating or advancing faster in the right hemifield may be more common or characteristic in Black patients due to anatomic, vascular, or other biological reasons specific to this population.³² This lateralized difference, with the right VF being disproportionately affected, implies that the progression of visual impairment and patterns of vision loss may follow a distinct clinical trajectory in different racial groups. 

Overall, this study demonstrates that there are significant differences in glaucomatous visual field loss between different gender, race, ethnicity, and preferred language groups. These differences exist not only in general visual field loss but also in different areas of the visual field, suggesting there may be a different clinical course or pattern in different groups for glaucoma. Understanding such racial disparities is crucial for tailoring screening, monitoring, and therapeutic approaches to different patient populations. 

These results are largely congruent with prior literature examining the role of race in determining the differential burden of disease in glaucoma.^4,33–35 It is crucial to consider the socioeconomic factors affecting the risk factors of race and ethnicity.^36–38 Patients from historically marginalized communities may experience exogenous contributing factors to worsening VF, including but not limited to systemic racism, language barriers, socioeconomic status, limited access to care, gaps in nutrition, health literacy, and education, and hesitancy in seeking care.^39–42 

Importantly, whether Black race is associated with faster VF deterioration has remained an unresolved question in the literature. Some prior studies have reported that Black patients experience faster VF progression than White patients, while other studies have not.^4,43 Our findings support the former and may help shape future work aimed at parsing out the factors driving this finding. The slower VF loss experienced by Asian patients in our cohort is interesting in the context of previous findings demonstrating that Asian patients experience higher rates of primary open-angle glaucoma compared with Whites; further work may aim to elucidate possible mechanisms underlying this observation.^1,39,44 Our findings also build on prior work demonstrating high rates of glaucoma prevalence among Hispanic populations.^45–47 By specifically demonstrating that Hispanic patients exhibit a higher rate of VF loss progression in the upper hemifield while maintaining an overall healthier VF compared to non-Hispanic patients, we provide an opportunity for enhanced clinical counseling of these patients with respect to patients’ quality of life and further study of causal factors. 

The patterns of VF loss that differ by demographic characteristics may aid clinicians in prognostication of the specific portions of their patients’ vision that may be most subject to glaucomatous deterioration.⁴⁸ Future work may consider an intersectional approach to add even further detail to our understanding of glaucomatous VF worsening (e.g., to understand whether race and gender intersect to produce specific patterns of VF loss). 

This study has multiple important benefits. The sample size was large, and the cohort study utilized data from patients over years of clinical visits. The large sample size and long time frame contribute to the reliability of the analysis. This study also employs a variety of analysis modalities, including pointwise and sectoral analysis in both the cross-sectional and longitudinal data, which provide comprehensive and granular information about differences between demographic groups. 

Limitations apply to this study. All data were collected from one medical center in Massachusetts, limiting the generalizability of our sample. Specifically, the proportion of White patients in our sample is particularly large. However, the demographics of our cohort are still not entirely different from those of the United States as a whole.⁴⁹ In addition, some patients did not answer race or language preference questions during clinical intake procedures, which may have introduced some selection bias in that our cohort only included patients who more thoroughly completed intake documents. Additionally, minor institution-based differences may exist in the administration of VF tests, although these differences are likely distributed evenly among different demographic groups. Most important, our data do not include information about socioeconomic status. Even though language preference has been shown to be related to different access to health care and willingness to get health care, it cannot be a perfect surrogate measure for socioeconomic status. It is possible that adjusting for socioeconomic status may mitigate the magnitude of the effects seen in our results. However, prior work has demonstrated that race-based differences persist even after adjustment for socioeconomic variables, highlighting the importance of ongoing studies to continue elucidating the relationship between demographic factors and VF loss in glaucoma.⁵⁰ 

This analysis used a large single-center cohort to explore demographic-based differences in glaucomatous VF loss and deterioration. Black and Asian patients have greater VF loss than White patients, but while Black patients have faster VF loss deterioration, Asian patients experience slower VF loss deterioration. Despite experiencing slower visual field loss progression in the lower hemifield, patients who prefer Spanish have an overall higher VF loss and average faster VF deterioration. Patients preferring languages other than Spanish or English have both overall high VF loss and overall faster VF deterioration. The mechanisms underlying these phenomena remain unknown. Further research is needed to explore how socioeconomic status might affect these findings and to elucidate the causal factors driving the associations seen in our data. 

Acknowledgments

Disclosure: Y. Pang, None; M. Tang, None; M. Shi, None; Y. Tian, None; Y. Luo, None; T. Elze, None; L.R. Pasquale, None; N. Zebardast, None; M.V. Boland, None; D.S. Friedman, None; L.Q. Shen, None; A. Lokhande, None; M. Wang, None