Summary of Review Findings

Table 10 summarizes the evidence reviewed for this update. The prior screening CER² that informed the 2012 USPSTF recommendation included no studies comparing glaucoma screening with no screening. We identified one trial of frail elderly persons not included in the prior screening CER that found no difference when comparing vision screening by an optometrist (including components relevant for diagnosis of glaucoma) with no screening on vision outcomes or vision-related quality of life.¹³² In this trial, vision screening was not specific for glaucoma, imaging was not utilized as part of screening, and the proportion of patients referred for glaucoma management was small (5.5% of those judged to need treatment). In addition, most patients in the no screening arm had seen an eye care professional in the prior year, which could have attenuated potential benefits of screening. Unexpectedly, the trial found screening associated with an increased risk of falls (number needed to screen 6.7 for 1 additional person falling), with a non-statistically significant increased risk of fractures. The reason for the increase in falls was unclear, but could be related to difficulty adapting to large corrections in vision or use of multifocal lenses. No study evaluated outcomes comparing referral to an eye health provider with no referral. Although one new study comparing screening with no screening was published prior to finalization of this report, it did not meet inclusion criteria because it was observational and had serious methodological limitations (control group was nonparticipants/nonresponders, and study did not perform statistical adjustment for potential confounders). In addition, the proportion of glaucoma patients with exfoliation glaucoma was very high (>50%), with uncertain applicability to primary care screening.⁵⁰

Table 10

Summary of Evidence – Glaucoma.

For diagnostic accuracy, our review found spectral domain-OCT and visual field assessment using the Humphrey Automated Field Analyzer to be associated with reasonable accuracy for identifying persons with glaucoma compared with a comprehensive eye exam. Although visual field assessment generally requires referral to an eye specialty setting, OCT could be ordered from a primary care clinic and has potential as a standalone screening test. Swept source-OCT, a newer OCT technology with increased scan speed and resolution compared with spectral domain-OCT, appears to offer improved visualization of ocular structures, but evidence on diagnostic accuracy for glaucoma is currently limited.¹⁷⁶ Tonometry for measurement of IOP was associated with high specificity but low sensitivity, indicating that it is insufficient as a standalone screening test. The low sensitivity of tonometry was consistent with data indicating that a significant proportion of patients with glaucoma have normal IOP. The gold standard for tonometry is the Goldmann Applanation Tonometer, which is not widely used in primary care settings. Evidence on other screening tests, including swept source-OCT, optic disc photography, ophthalmoscopy and biomicroscopy, and pachymetry was more limited. Our review differs from and expands upon the prior screening CER,² which included case-control studies, included screening tests no longer in use, focused on head-to-head comparisons of screening modalities, did not perform meta-analysis, and did not include more recent studies on spectral domain-OCT. A persistent challenge in interpreting studies on diagnostic accuracy of screening is the variability and lack of standardization in the reference standard.

An ophthalmic telemedicine screening program was associated with inconsistent sensitivity but high specificity for identifying persons with glaucoma compared with a face-to-face evaluation.^109–111 Although telemedicine screening was performed in a Veterans Affairs primary care setting, potential barriers to more widespread implementation include the need to install specialized instruments and utilize trained technicians. One study evaluated the accuracy of a risk assessment instrument for identifying persons with glaucoma, but reported low sensitivity and was conducted over 30 years ago, with no subsequent validation.¹¹⁷ The risk assessment instrument also may be of limited applicability to screening because one of the primary risk factors was a previous glaucoma diagnosis.

Our findings on the effectiveness of medications for early OAG and ocular hypertension are generally consistent with the prior treatment CER,³ which found moderate evidence that medical and surgical treatments can lower IOP and reduce risk of progression by visual field and optic nerve criteria, but no studies on effects of treatments on visual impairment or patient reported outcomes. Our review differs from the prior treatment CER by focusing on comparisons involving treatment versus placebo or no treatment and conducting meta-analysis. We included several new trials conducted since the prior treatment CER that evaluated effects on visual acuity and vision-related function or quality of life. Most notable were the addition of the large UKGTS,³⁵ which compared latanoprost with placebo in persons with untreated glaucoma, and LiGHT,^37,82 which compared SLT with topical medications. Our meta-analysis found topical medications associated with decreased risk of glaucoma progression, defined by visual field or optic disc changes, with a number needed to treat of 20.8 to prevent one case of progression over 2 to 10 years. Results were similar when the analysis was restricted to trials that defined glaucoma progression based on visual field changes (excluding optic disc changes). However, these results are difficult to interpret because methods for defining visual field changes varied and were based on the development of focal visual field deficits or any visual field loss; no trial evaluated the proportion of patients who met a minimum clinically important different threshold¹⁴⁹ for overall visual field loss. In the UKGTS, there were no differences between medications and placebo in mean visual field loss or visual acuity.³⁵

Medications were also associated with greater reduction in IOP compared with placebo or no medication, with a pooled difference of about 3 mm Hg. Although there was statistical heterogeneity in the IOP meta-analysis, the inconsistency was in the magnitude rather than direction of effects, as results favored therapy in all studies and for all specific classes and medications. Across individual studies and in various stratified analyses, effects of medications on mean IOP generally ranged from 2 to 3 mm Hg. Data on harms of topical medical therapies was limited, but did not indicate an increased risk of serious adverse events, though non-serious ocular adverse events (e.g., redness, irritation, itching, burning, tearing) were more common. Newly-approved topical medications for glaucoma (netarsudil and latanaprost bunod) were associated with similar or greater IOP reducing effects compared with older medications, but increased risk of adverse events. New data on effects of medical treatment for glaucoma on quality of life was available from the UKGTS,⁹⁰ which found no differences between latanoprost and placebo in general or vision-related quality of life at 2 years. For SLT compared with medical therapy, LiGHT found effects of SLT and medical therapy associated with similar effects on IOP, visual acuity, visual field, and quality of life, with no differences in serious adverse events or ocular adverse events.^37,82 Results of smaller trials comparing SLT and medical therapy were consistent with LiGHT. Our findings regarding treatment are most applicable to patients with ocular hypertension or early, untreated OAG, the populations typically enrolled in the trials.

As in the prior treatment CER, interpretation of effects of treatment on IOP and glaucoma progression is a challenge because of a lack of evidence on the association between improvements in these and other intermediate outcomes (e.g., optic nerve damage) following treatment for OAG or ocular hypertension and improvement in visual impairment, quality of life, or function. Although cross-sectional studies indicated an association between more severe visual field loss and greater visual impairment and patient reported outcomes, such studies cannot demonstrate causality and do not evaluated the association between changes in IOP following treatment and subsequent outcomes.

Limitations

Our evidence review has some limitations. First, we excluded non-English language studies, which could introduce language bias. Although we identified one RCT of medical therapy published in Japanese, it was small (n=16) and would not impact conclusions.¹⁷⁷ Second, there was statistical heterogeneity in pooled analyses on effects of medical therapy compared with placebo or no treatment on IOP. However, as described above, inconsistency was in the magnitude but not direction of effect, which favored medical therapy across studies. In addition, differences in IOP lowering effects were small, generally ranging from 1 to 2 mm Hg. Because of anticipated heterogeneity, we utilized a random effects model for pooling. Third, statistical heterogeneity was present in pooled analyses of sensitivity and specificity. However, standard methods for measuring statistical heterogeneity do not account for the variability in estimates related to threshold effects. Despite the statistical heterogeneity, results were robust in stratified and sensitivity analyses. Fourth, direct evidence on benefits and harms comparing screening with no screening and comparing effects of treatment with no treatment for ocular hypertension or early OAG on visual impairment, quality of life, and function remains very limited, though the UKGTS study found no effects on quality of life or function. Fifth, we excluded case-control studies of diagnostic accuracy, which reduced the evidence available for evaluating screening tests, but reduced potential spectrum bias. Sixth, evaluations of publication bias through graphical or statistical methods was limited by small numbers of studies or statistical heterogeneity. However, we did not identify unpublished studies likely to impact findings. Seventh, unlike the prior screening and treatment CERs,^2,3 we excluded most head-to-head comparisons, which might provide indirect evidence regarding the diagnostic accuracy of screening tests and outcomes of treatment. However, we included trials comparing newly FDA-approved medications with older medications and comparing SLT with medical therapy, because the new medications and SLT are considered first-line treatments and placebo- and sham-controlled trials are not available.

Emerging Issues/Next Steps

Latanoprostene bunod (a nitric oxide-donating medication) was FDA-approved in 2017 and netarsudil (a Rho kinase inhibitor) was approved in 2019. These medications are the first in their respective classes for treatment of glaucoma and decrease IOP primarily by increasing outflow (rather than reducing aqueous production). Although some trials comparing latanoprostene bunod with netarsudil are available, additional studies with longer-term followup are needed to verify benefits and harms. The development of newer minimally-invasive surgical procedures for treatment of OAG is ongoing,⁴⁵ including angle-based surgeries (Kahook dual blade and gonioscopy-assisted transluminal trabeculotomy), micro-shunting surgeries (Xen gel stent), and micropulse laser therapy.

With regard to screening tests, OCT technology has evolved rapidly with respect to scanning speed and resolution, which may lead to improvements in diagnostic accuracy (e.g., with use of swept source-OCT).¹⁷⁸ Although OCT may have some potential as a standalone screening test for glaucoma,¹⁷⁹ evidence on the effects of screening with OCT are not available, and evidence on effects of treatments in persons diagnosed using OCT is limited. An area of high interest and a rapidly expanding evidence base is the use of artificial intelligence to analyze and categorize data from OCT and other screening tests,⁴⁸ potentially improving diagnostic accuracy and facilitating implementation of screening. However, we did not identify studies utilizing artificial intelligence that were eligible for our review, because they utilized (non-clinical) imaging databanks, used a case-control design, lacked validation testing, or did not evaluate algorithms available for widespread use. The FDA has published a proposed regulatory framework to evaluate artificial intelligence and machine learning as a medical device;¹⁸⁰ as yet no artificial intelligence technologies have been approved for diagnosis of glaucoma.

Relevance for Priority Populations

Glaucoma disproportionately impacts Black persons, and to a lesser extent, Latino persons, relative to non-Latino White persons. Black persons have the highest prevalence of glaucoma, a higher rate of glaucoma progression and blindness, and earlier presentation of glaucoma.^17,181 Race-related disparities have been reported in glaucoma management and adherence to care,^182–184 and disparities exist with regard to access to care. Evidence on how race or ethnicity impacts effectiveness of treatment is limited. In the OHTS, in which 25 percent of participants were Black, there was no interaction between race and effects of medical treatment compared with placebo on IOP, likelihood of progression from ocular hypertension to OAG, or adverse events.¹⁴⁸ Although Black participants were at increased risk of progression to OAG in univariate analysis, race was not a predictor when analyses adjusted for other demographic factors, markers of glaucoma severity, and comorbidities. In LiGHT, in which 22 percent of participant were Black, 6.5 percent were Asian, and 68 percent were White, race/ethnicity was not a predictor of response to SLT.³⁶ In UKGTS, approximately 5 percent of participants were Black and 3.1 percent were Asian; no analysis was performed on the interaction between race/ethnicity and effects of latanoprost compared with placebo.³⁵ Two trials that did not meet inclusion criteria because they evaluated surgery (one trial also enrolled persons with advanced glaucoma) found that Black participants had worse outcomes than White participants who had surgery first.^185,186

Future Research

Important gaps remain in the evidence on screening for glaucoma. Additional trials comparing screening with no screening that utilize contemporary screening and diagnostic modalities (e.g., spectral domain-OCT or swept source-OCT) and include vision-related outcome, function, and quality of life would provide direct evidence on effects of screening. Research is also needed to determine optimal screening approaches, such as strategies that target higher-risk populations compared with screening of all adults. Research on the accuracy of instruments for identifying persons at increased risk of glaucoma would be useful for informing screening strategies. Studies are needed to verify the diagnostic accuracy of current screening tests when applied to screened populations. Studies are needed to better understand the utility of artificial intelligence to aid in the analysis and interpretation of screening tests, using validated algorithms in clinical cohorts of patients that are available for use in clinical practice (ideally, FDA-approved). Telehealth approaches to screening that can be implemented in primary care settings could potentially facilitate access and are particularly relevant in the post-COVID-19 era. Studies on the effects of referral to glaucoma screening from primary care compared with no referral are lacking and would help clarify outcomes associated with referral. Research is needed to better understand the long-term effects of treatment on visual impairment, quality of life, and function; to understand how effects of treatment vary by race/ethnicity; and to verify that benefits of treatment are retained in persons diagnosed with OAG using newer imaging methods. Longer-term studies of the recently approved medications netarsudil and latanoprostene bunod would help clarify benefits and harms relative to older first-line therapies.

Conclusions

Direct evidence comparing glaucoma screening with no screening is limited and showed no benefits on vision-related quality of life or function, and increased risk of falls. Screening tests (OCT, visual field assessment) can identify persons with OAG with reasonable accuracy. Treatment for ocular hypertension or untreated OAG is associated with reduction in IOP and reduced risk of glaucoma progression based on visual fields or optic nerve changes, but limited evidence on the association with visual outcome, quality of life, and function indicates no clear effects.