Risk Factors for Glaucoma Drainage Device Revision or Removal Using the IRIS Registry


To elucidate risk factors for revision or removal of glaucoma drainage devices (GDD) in glaucoma patients in the United States.


Retrospective cohort study.


IRIS Registry (Intelligent Research in Sight) patients who underwent GDD insertion between January 1, 2013 and December 31, 2018 were included. Various demographic and clinical factors were collected. Kaplan-Meier survival plots, Cox proportional-hazard models utilizing Firth’s Penalized Likelihood, and multivariate linear regression models were used. The main outcome measures were hazard ratios (HR) and beta coefficient (β) estimates.


A total of 44,330 distinct patients underwent at least 1 GDD implantation, and 3354 of these underwent subsequent GDD revision or removal surgery. With failure defined as GDD revision/removal, factors significantly associated with decreased failure included unknown race (HR = 0.83; P = .004) and unknown ethnicity (HR = 0.68; P < .001). Factors associated with increased risk of GDD revision/removal surgery included presence of chronic angle-closure glaucoma (HR = 1.32; P < .001) and dry eye disease (HR = 1.30; P = .007). Additionally, factors associated with a decreased average time (in days) to GDD revision/removal included male sex (β = -25.96; P = .044), unknown race (β = -55.28; P = .013), and right-eye laterality (β = -38.67; P = .026). Factors associated with an increased average time to GDD revision/removal included having a history of a past eye procedure (β = 104.83; P < .001) and being an active smoker (β = 38.15; P = .024).


The size and scope of the IRIS Registry allows for detection of subtle associations between risk factors and GDD revision or removal surgery. The aforementioned demographic and clinical factors may all have an impact on GDD longevity and can inform the treatment options available for glaucoma patients.


G laucoma drainage devices (GDDs) reduce intraocular pressure (IOP) by creating an alternate drainage route for aqueous humor to bypass diseased trabecular meshwork. Over the past few decades and since results from the Tube Versus Trabeculectomy (TVT) study suggested that GDD implantation had comparable IOP-lowering ability and long-term effectiveness to trabeculectomy, with potentially fewer perioperative complications, GDDs have increasingly been used in the management of refractory glaucoma. , Similarly, a review of Medicare claims data between 1995 and 2004 demonstrated a 53% decrease in the number of trabeculectomies performed in eyes without scarring and a 184% increase in the number of aqueous shunting devices placed in eyes with and without scarring over the study period. Interestingly, in the primary TVT study, rates of surgical failure of trabeculectomies and GDD implantation surgery in eyes without prior surgery were comparable at 3 years postoperatively, although trabeculectomies may result in a slightly lower IOP and medication burden compared with GDD implantation surgery.

Despite the promising benefits of GDD surgery compared with trabeculectomy, GDDs are not without complications that may necessitate repeat surgery. For example, GDD erosion through the conjunctiva causes tube exposure, which poses a major risk of endophthalmitis and subsequent vision loss. Prior studies have reported rates of GDD revision or removal surgery between 2% to 12%, , , , and rates of tube exposure between 1% to 8.9%. , , , ,

As the use of GDDs continues to rise, it is important to elucidate risk factors for complications requiring GDD revision or removal surgery. Given that GDD tube exposures almost always require surgical correction and that risk factors for GDD tube exposure have previously been explored in the literature, this study evaluated the findings in the context of the current literature for GDD tube exposures. In addition, some studies have noted that tube exposure is one of the most common causes for repeat GDD surgery. , Since GDD revision or removal surgery may also be performed for a reason other than tube exposure—such as tube malpositioning or occlusion, conjunctival erosion, IOP elevation due to bleb fibrosis, corneal decompensation, chronic inflammation, iris-tube touch, and cornea-tube touch— the incidence of GDD revision or removal surgery should be inclusive of and greater than the expected incidence of GDD exposure. Prior studies have suggested that factors such as sex, age, Hispanic race, history of smoking, prior ocular surgeries, and topical steroid use are associated with GDD exposure specifically; , , , , the current study hypothesized that similar factors may increase the risk of GDD revision or removal surgery among those who have a revision or removal. However, there is no current consensus in the literature regarding risk factors for GDD exposure ( Tables 1 and 2 ).


Selection of Studies Evaluating Risk Factors for Glaucoma Drainage Device Tube Exposures.

Publication Author, Year Present study Al-Beishri et al., 2019 Zhou et al., 2018 Chaku et al., 2016 Trubnik et al., 2015
Journal J Glaucoma. J Ophthal Vis Res. Clin Ophthal. J Glaucoma.
Study type R R R R R
Follow-up (months) Mean 39.8 (erosion group), 30.1 (non-erosion group) Mean 27.6 Mean 24.1 (erosion group), 20.3 (non-erosion group)
Total number of eyes 44,330 (with GDDs) 836 (53-53 matched case-control) 306 64 (32-32 matched case-control) 339
Number of GDD tube exposures or removals 3354 (7.57%) over 6 years 53 (6.3%) over 7 years 23 (7.52%) over 10 years 28 (8.3%) over 5 years
Risk factors associated with GDD exposure or repeat GDD surgery Factors associated with risk of revision/ removal GDD surgery
– Dry eye disease
– Unknown race/ ethnicity (protective)

Factors associated with time to revision/removal surgery
– Smoking status
– Prior eye procedure
– Unknown race
– Male sex (shorter time)
– Right-eye laterality (shorter time)
– Older age
– Female sex
– Baerveldt GDD compared with Ahmed GDD
– History of ocular surgeries
– Aphakia
– Uveitic glaucoma
– Longer use of postoperative steroids
– Younger age
– Ocular inflammation
– Pediatric and uveitic glaucoma
– Concomitant surgical procedures at time of GDD implantation
Factors not associated with GDD exposure or repeat GDD surgery – Asian race
– Black race
– Other race
– Age
– Diabetes mellitus
– Cardiac disease
– Scleral patch graft (protective)
– GDD location
– Suture material used to secure the tube or plate
– Prior history of intraoperative MMC/5-FU
– Preoperative use of alpha-agonists
– Preoperative use of CAI
– Preoperative use of cholinergic drugs
– Preoperative use of beta-blockers
– Age
– Sex
– Race
– Hypertension
– Diabetes mellitus
– Prior glaucoma laser procedure
– Prior glaucoma surgery
– Type of patch graft
– Implant location (superior versus inferior)
– Concomitant surgical procedures
– Preoperative BCVA, IOP, or number of meds
– Preoperative or postoperative prostaglandin use
– Diabetes mellitus (Protective)
– GDD type
– GDD location
– Sex
– Ethnicity/race
– Eye laterality
– Lens status
– Number of prior surgical procedures
– Hypertension
– Immune system abnormalities
– Prior surgeries
– Post-GDD surgeries
– Smoking
– Seasonal allergies
– Dry eye syndrome
– Hypertension
– Diabetes
– Rheumatological conditions
– Oral steroids/immunomodulators
– MMC/Avastin/Triescence
– Type of glaucoma
– Contact lens use
– Number of preoperative topical medications
– GDD type
– GDD location
– GDD position in eye (ie, AC, PC, PP)
– Patch graft material

Abbreviations: 5-FU = 5-fluorouracil; AC = anterior chamber; BCVA = best corrected visual acuity; CACG = chronic angle-closure glaucoma; CAI = carbonic anhydrase inhibitor; GDD = glaucoma drainage device; IOP = intraocular pressure; MMC = Mitomycin C; PC = posterior chamber; PP = pars plana; R = retrospective.


Additional Selection of Studies Evaluating Risk Factors for Glaucoma Drainage Device Tube Exposures.

Publication, Author, Year Geffen et al., 2014 Muir et al., 2014 Koval et al., 2013 Stewart et al., 2010 Byun et al., 2009
Journal J Glaucoma. BMJ Open. J Ophthal. Eur J Ophthal. Jpn J Ophthal.
Study type R R R R (meta-analysis) R
Follow-up (months) Mean 43.5 (range 12-103) Mean 41 (range 12-84) Mean 56.6 (range 0.3-161) (control group) Mean 26.1
Total number of eyes 158 1073 121 (41-80 matched case-control) 3255 266 (11-44 matched case-control)
Number of GDD tube exposures 14 (8.9%) over 8 years 67 (6.24%) over 6 years 64 (2.0%) (0.09% per month) 11 (4.1%) over 4 years
Risk factors associated with GDD exposure – Higher number of preoperative hypotensive medications – Female sex
– White race compared to Black race
– Older age for men
– Hispanic ethnicity
– Neovascular glaucoma
– Previous trabeculectomy
– Combined surgery
– Length of follow-up – Number of prior ocular surgeries
Factors not associated with GDD exposure – Age
– Sex
– Type of glaucoma
– Prior glaucoma procedures
– Prior ocular trauma
– Prevalence of ocular inflammatory and surface diseases requiring treatment
– Postoperative use of topical steroids
– Incision type (ie, fornix versus limbal base)
– Inferotemporal valve position (protective)
– Age in females
– Diabetes mellitus
– GDD location (superior versus inferior)
– Patch graft material
– Pre-existing GDD prior to surgery
– GDD type
– Age
– Sex
– Diabetes mellitus
– Hypertension
– Preoperative IOP
– Number of glaucoma medications
– Prior GDD implantation
– Prior laser procedure
– Prior EL/PRP, PK/PKP, or vitrectomy
– GDD type
– GDD location
– Patch graft type
– GDD type
– Study length
– Age
– Sex
– GDD location
– Diabetes mellitus
– Hypertension
– Type of glaucoma

Abbreviations: GDD = glaucoma drainage device; IOP = intraocular pressure; R = retrospective.

Given that prior studies may have been limited by the availability of patient data at a single institution, the American Academy of Ophthalmology (AAO) Intelligent Research in Sight

(IRIS) Registry provides a unique opportunity to identify subtle biomarkers and risk factors related to ocular diseases in patients across multiple practices throughout the United States. As of September 2019, the Registry contained 252.95 million patient encounters comprising 60.78 million unique patients. Using the big-data characteristics of the IRIS Registry, this study aimed to examine characteristics of patients undergoing GDD insertion, determine the incidence of GDD revision and removal surgery among these patients, and to elucidate potential risk factors associated with the need for repeat surgery and the time at which surgery is needed.


The methods of data collection of the IRIS Registry have previously been described. The AAO provided access to a limited version of the IRIS Registry for academic research to 4 different academic institutions in the spring of 2019. The version of the IRIS Registry used in this study was frozen on October 23, 2020 and accessed on November 12, 2020. This study was conducted in accordance with the Declaration of Helsinki. Given the use of de-identified data, the study was exempt from approval by the Mass General Brigham Institutional Review Board.

All patients with a documented GDD insertion between January 1, 2013 and December 31, 2018, from the IRIS Registry were included in this study. Glaucoma drainage device insertions were identified by Current Procedure Terminology (CPT) (Fourth edition) codes 66179 and 66180. Patients with repeat GDD surgery were identified using CPT codes for GDD revision surgery (66184 or 66185) or GDD removal surgery (67120). If a patient had more than 1 GDD insertion, only the first (earliest recorded) instance was used. Revision and removal cases were further identified by including only instances where GDD revision or removal surgery occurred after the date of GDD insertion, and where the laterality aligned with that of the GDD insertion. If there were multiple instances of GDD revision or removal, the instance recorded in closest proximity to the date of GDD insertion was included.

The following demographic and clinical factors were extracted: age; sex; race; ethnicity; smoking history; glaucoma diagnosis based on International Classification of Disease (ICD) (9th and 10th editions) codes; history of trabeculectomy based on CPT codes; and several systemic disease diagnoses based on ICD codes, including rheumatoid arthritis, diabetes mellitus (type 1 and type 2), systemic lupus erythematosus, hypertension, chronic obstructive pulmonary disease, and asthma. Presence or absence of precipitating diagnoses to revision/removal—such as tube shunt exposure, conjunctival erosion, tube occlusion/malposition, IOP elevation due to bleb fibrosis, corneal decompensation, chronic inflammation, iris-tube touch, cornea-tube touch, or other mechanical complications of ocular prosthetic devices, implants, and grafts—were also extracted using ICD codes. Race was categorized as ‘Other’ for individuals who chose not to self-identify as Caucasian, Asian, or Black or African American, and as ‘Unknown’ if individuals chose not to self-identify at all. Similarly, ethnicity was also categorized as ‘Unknown’ if individuals chose not to self-identify in this regard. Individuals who self-identified as ‘Non-Hispanic or Latino’ are those that identified as neither Hispanic nor Latino and will henceforth be referred to as ‘Non-Hispanic/Latino’. A full list of ICD and CPT codes used in this study is presented in Table 3 .


ICD 9 th and 10 th Edition Codes, and CPT Codes.

Diagnosis/Procedure ICD 10 ICD 9 CPT
Ocular hypertension H40.04 365.04
Unspecified open-angle H40.10 365.10
Primary open-angle H40.11 365.11
Low-tension H40.12 365.12
Pigmentary H40.13 365.13
Pseudoexfoliation H40.14 365.52
Residual stage H40.15 365.15
Unspecified primary open-angle H40.20 365.20
Acute angle-closure attack H40.21 365.22
Chronic angle-closure H40.22 365.23
Intermittent angle-closure H40.24 365.24
Secondary to trauma H40.3 365.65
Secondary to inflammation H40.4 365.62
Secondary to other eye disorders H40.5 365.60
Secondary to drugs H40.6 365.3
Increased episcleral venous pressure H40.81 365.82
Hypersecretion H40.82 365.81
Congenital Q15.0 365.4
Rheumatoid arthritis M06.9 714.0
Systemic lupus erythematosus M32.9 710.0
Diabetes mellitus type I E10 250.01
Diabetes mellitus type II E11 250.00
Hypertension I10, I11, I12, I13, I15, I16 401.0, 401.1, 401.9
Asthma J45.0-J45.998 493.00-494.99
Chronic obstructive pulmonary disease J40, J41.0, J41.1, J41.8, J42, J43.0, J43.1, J43.2, J43.8, J43.9, J44.0, J44.1, J44.9, J47 491.0, 491.1, 491.2, 491.21, 491.22, 491.8, 491.9, 492.0, 492.8, 494.0, 494.1, 496, 506.4
Dry eye H04.121, H04.123, H04.129 375.15
Tube shunt exposure Y77.2
Conjunctival erosion H11.10 372.50
Tube occlusion/malposition T85.39
IOP elevation due to bleb fibrosis H59.4 379.60
Corneal decompensation H18.899 371.89
Implant extrusion T85.39 996.59
Chronic inflammation H20.9 364.3
Iris-tube touch Y77.2
Cornea-tube touch Y77.2
Wound dehiscence T81.31 998.32
Other mechanical complication of other ocular prosthetic devices, implants, and grafts T85.39 996.59
GDD insertion 66179, 66180
GDD revision 66184, 66185
GDD removal 67120
Trabeculectomy 66170

Abbreviations: CPT = Current Procedural Terminology; GDD = glaucoma drainage device; ICD = International Classification of Disease.

Descriptive statistics and frequencies were calculated for all relevant variables. Survival analysis was performed with failure defined as having GDD revision or removal surgery. Kaplan-Meier (KM) plots of survival were constructed and stratified by baseline and clinical characteristics. Survival curves between groups were compared using log-rank tests. Despite the small overall incidence of repeat GDD surgery relative to the patient cohort, the log-rank test results were able to offer insight into differences in the pattern of survival over time between each stratum of interest, even though the curves themselves failed to clearly visualize these differences due to the non-parametric nature of the test and given that it is robust to censoring, assuming that the censoring is non-informative.

To appropriately account for the small number of events in this sample, particularly relative to the number of censored observations, Cox regression models utilizing Firth’s Penalized Likelihood correction (CRFPL) were fit to the data to reduce bias in the estimates. , Hazard ratios (HRs) and 95% confidence intervals (CIs) were derived from the CRPFL models using baseline characteristics. Univariate and multivariate linear regression models were also fit to model the average days from GDD insertion to revision or removal against baseline characteristics. Fully saturated models using all baseline covariates were fit first for both model types. Final multivariate linear regression models were derived from the full model via automated stepwise selection utilizing Akaike’s Information Criterion. Akaike’s Information Criterion is used as a quality metric to evaluate how well a model fits the data, relative to other models, using a weighted log-likelihood formula. The automated stepwise selection package used on the multivariate linear regression model was not compatible with the CRFPL model; therefore, the final CRFPL model was derived using backward selection with a P -value selection threshold of P = .05. All models and selection procedures were run on the full cohort of patients with a GDD insertion without stratifying on repeat surgery type, and were also separately run on GDD revision patients only and GDD removal patients only to determine if there were different risk factors between these two types of repeat surgeries. All analyses were conducted in R statistical programming software (version 3.5.1).


Between January 1, 2013 and December 31, 2018, 44,330 distinct patients in the IRIS registry underwent at least 1 GDD implantation. Of these, 3354 patients underwent at least 1 subsequent GDD revision or removal surgery in the same eye, hereafter known as “repeat GDD surgery.” The incidence of repeat GDD surgery was 3354 of 44,330 patients (7.57%) over the course of 6 years. Of these patients, 2933 of 3354 (87.45%) had a GDD revision, while 421 (12.55%) had a GDD removal. The demographic and clinical characteristics of these patients are listed in Table 4 . Some of the most common characteristics (> 75% of total cohort) among patients included in the total cohort were: Non-Hispanic/Latino ethnicity (77.97%), aged ≥ 55 years (82.21%), and no history of prior eye procedures (94.90%).


Demographics and Clinical Characteristics of Patients With Glaucoma Drainage Device Insertion Surgery.

Variable N (%) Variable N (%)
Number of patients 44,330 Age at GDD insertion
Sex 0-18 488 (1.10)
Female 20,706 (46.71) 19-25 408 (0.92)
Male 23,624 (53.29) 26-54 6990 (15.77)
Race 55+ 36,444 (82.21)
Asian 1332 (3.00) Smoking status
Black or African American 11,678 (26.35) Active 7637 (17.23)
Caucasian 24,604 (55.50) Former 4147 (9.35)
Other 1321 (2.98) Never 32,546 (73.42)
Unknown 5395 (12.17) Laterality
Ethnicity Right 11,894 (26.83)
Hispanic or Latino 5438 (12.27) Left 11,496 (25.93)
Not Hispanic or Latino 34,565 (77.97) Unknown 20,940 (47.24)
Unknown 4327 (9.76) Prior eye procedures
Severity Yes 2259 (5.10)
Mild 5646 (12.7) No 42,071 (94.90)
Moderate 5470 (12.3) Had revision/removal surgery
Severe 53 (0.2) Yes 3354 (7.57)
Unknown 33,161 (74.80) No 40,976 (92.43)
Revision/removal surgery type 1
Revision 2933 (87.45)
Removal 421 (12.55)
Comorbid systemic diseases 2
Asthma 50 (0.11) Dry eye disease 1214 (2.74)
COPD 25 (0.06) Systemic hypertension 789 (1.78)
Diabetes mellitus, type I 276 (0.62) Rheumatoid arthritis 26 (0.06)
Diabetes mellitus, type II 2833 (6.39) Systemic lupus erythematosus 8 (0.02)
Glaucoma types 2
Primary open-angle 13,117 (29.59) Increased episcleral venous pressure 28 (0.06)
Angle-closure 2260 (5.10) Low-tension 242 (0.55)
Congenital 112 (0.25) Ocular hypertension 1575 (5.3)
Medication-induced 379 (0.85) Pigmentary 372 (0.84)
Secondary to eye trauma 774 (1.75) Pseudoexfoliation 939 (2.12)
Quantitative Characteristics
Variable Mean (SD) Range (Min-Max) Median
Days to GDD revision/removal 1 293.82 (373.45) (1.00-2133.00) 133.00
Days of total follow-up 3 1744.75 (709.35) (1.00-3173.00) 1689.00
Age at GDD insertion (in years) 65.42 (14.16) (0.00-86.00) 68.00

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Sep 11, 2022 | Posted by in OPHTHALMOLOGY | Comments Off on Risk Factors for Glaucoma Drainage Device Revision or Removal Using the IRIS Registry

Full access? Get Clinical Tree

Get Clinical Tree app for offline access