Impact of iStent Microbypass Shunt on Medicare Part B Glaucoma Surgical Expenditure

Abstract

Purpose:

To examine the impact of iStent on the recent glaucoma surgical expenditure for Medicare Part B beneficiaries.

Design:

Retrospective, observational, population-based analysis.

Subjects and Participants:

All applicable cases in the Centers for Medicare and Medicaid Services (CMS) Part B Summary Data Files.

Methods:

The Part B National Summary Data Files from 2007 to 2017 were obtained through CMS. Glaucoma surgical procedures, including trabeculectomy, glaucoma drainage implants (GDIs), and select minimally invasive glaucoma surgeries (MIGS) including iStent were queried from the database using Current Procedural Terminology codes. We used Chow’s test to confirm significant changes in expenditures trajectories. We built a mixed-effects regression model to examine the effect of demographic factors on each state’s iStent adoption speed.

Main Outcome Measures:

Proportion of iStent in total glaucoma surgical spending for individual states for each year.

Results:

Total Medicare part B payment for the selected glaucoma procedures rose from $52.0 million in 2007 to $179.9 million in 2017. The percentage for trabeculectomy and GDIs decreased from 92.3% to 21.2%. Conversely, the iStent, approved by FDA in 2012, rose to represent 57.9% of total payment by 2017. There were significant changes in the slope of glaucoma surgical Medicare payment (P<.00001) and iStent payment (P<.0001) trajectories in 2012. Mixed-effect regression analysis showed a wide range among the states’ rates of increase in iStent proportion between 2012–2017 (Range: 5.12% – 14.54% per year). Higher male proportion in the population was associated with faster increases in iStent proportions [12.4% per 5% increase in male proportion, 95% interval: 4.3%−20.5%, P=0.003]. Higher median age of the population was associated with slower increases [−3.6% per 1 year increase in median age, 95% interval: −0.4% to −6.8%, P=0.026].

Conclusion:

Rising payment for iStent represents the majority of the increase in the glaucoma surgical spending in the recent decade. Male gender and age significantly affects the state-wise speed of adoption for iStent. The impact of iStent on the comprehensive glaucoma Medicare expenditure in the same time period warrants further study.

Précis:

The iStent microbypass shunt is an increasingly significant driver of glaucoma medicare expenditure. Its rates of adoption among individual states vary widely and appear to be affected by demographic factors.

Introduction:

Glaucoma is a leading cause of irreversible blindness in United States.^1,2 The costs for the treatment of glaucoma are projected to increase in an accelerating manner,³ and may be primarily due to the shift in demographics, as the population ages and people have longer life spans. As glaucoma is largely a disease of the elderly, the incidence and prevalence of glaucoma is projected to increase.^4,5 Much of the glaucoma healthcare is covered under the Medicare program,⁶ therefore, it is important to understand the factors that influence glaucoma healthcare costs.

An important driver of glaucoma treatment costs is surgery.⁷ The traditional glaucoma surgical interventions have been trabeculectomy and glaucoma drainage implants (GDI) and recently, micro invasive glaucoma surgeries (MIGS) have been gaining in popularity. There are many MIGS devices on the market, but the most widely-used among them is the iStent (Glaukos, San Clemente, CA), a trabecular micro bypass shunt approved by the FDA in 2012. iStent is indicated for use in conjunction with cataract surgery in adult patients with mild to moderate open-angle glaucoma, which constitutes a large number of patients.

While the rate of iStent usage has rapidly increased since its introduction, there are few studies describing its economic impact and the regional differences in the speed of adoption.^8,9 This study examines the recent trends in glaucoma surgical procedure payment for Medicare Part B beneficiaries and how novel MIGS procedures, especially iStent, have affected them at the national and state level. It also aims to elucidate potential drivers of the iStent adoption at the state-level using census-derived demographic data.

Methods:

Data Source:

The Part B National Summary Data Files from 2007 to 2017 were obtained through Centers for Medicare and Medicaid Services. These years were chosen in order to be able to visualize the general trend of glaucoma procedures both before and after the release of iStent in the year 2012. No explicit patient identifiable information is contained in the database and our study was exempted by the Yale Institutional Review Board. The present study also adhered to the tenets of declaration of Helsinki. Glaucoma surgical procedures, including trabeculectomy, GDIs, and select MIGS were queried from the databases using current procedural terminology (CPT) codes. The CPT codes used were 66170/66172 for trabeculectomy, 66180/66179 for GDI, 65820 for goniotomy, 66183 for ab externo drainage device, 66174 for transluminal dilation, 66711 for Endocyclophotocoagulation (ECP), 0191T for iStent, 0449T for Xen Gel Stent (Allergan Inc, Irvine, CA), and 0474T for CyPass (Alcon, Fort Worth, TX). For each procedure in each year, the number of allowed services (number of procedures payed for by Medicare Part B) and the payment amount were collected. While the data files broke down the procedures with common modifiers that may have been used with the CPT code, the total payment, including all modifiers, were used for each procedure for the purposes of this study. Payment figures were adjusted for inflation to 2017 equivalents using the inflation calculator from U.S. Bureau of Labor Statistics that can be found at https://data.bls.gov/cgi-bin/cpicalc.pl.

State level census data including median age, gender, ethnicity, and proportion of population below the poverty level for individuals greater or equal to 65 years of age were obtained through American Community Survey from United States Census Bureau for each year under analysis.

Study Design:

The trends in total number of procedures and payments were analyzed. Chow’s test compares the coefficients in two linear regression on different parts of a curve to test for the presence of a structural break and was used to detect significant changes in the spending curves of each procedure over the years. Mixed-effect regression analysis was used to examine the state-level trends in the proportion of total glaucoma surgical procedure spending due to iStent procedures. State-level covariates, including median age, gender, poverty, and ethnicity were included in the model to explain state-level variability in the iStent spending trends across time. Additionally, state-level slope random effects were included to account for unexplained spatial variability in iStent spending trends across time. The state-level intercepts for the model were fixed at zero and the slopes were centered at 2011 to reflect the fact that spending prior to 2012 was virtually non-existent (i.e. proportions centered around zero in 2011). Alaska and New Mexico were excluded from the model due to incomplete data.

The primary outcome variable was defined as the proportion of total glaucoma procedure spending due to iStent procedures in a state during a given year. Model fitting was carried out using R statistical software (R Core Team (2014). R: A language and environment for statistical computing. R Foundation for statistical Computing, Vienna, Austria) version 3.6.1 with lme4 and Lmertest packages. The statistical model was defined as follows:

$$\begin{gathered} \hspace{1em}\hspace{1em}\hspace{1em}\hspace{0.17em}\hspace{1em}{Y}_{it};i=1,\hspace{0.17em}\dots \hspace{0.17em},n;\hspace{0.17em}t=1,\hspace{0.17em}\dots \hspace{0.17em},q \\ {Y}_{it}={\beta }_{1i}(t-2011)+{\mathbf{x}}_{it}^{\text{T}}\mathbf{\gamma }+{ϵ}_{it};\hspace{0.17em}{ϵ}_{it}|{\sigma }^2~N(0,{\sigma }^2) \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{0.17em}\hspace{1em}\hspace{1em}\hspace{1em}{\beta }_{1i}={\beta }_1+b_{1i} \\ \hspace{0.17em}\hspace{0.17em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{b}_{1i}~\text{N}(0,{\sigma }_b^2),i=1,\hspace{0.17em}\dots \hspace{0.17em},n \end{gathered}$$

where:

Y_it: Proportion spent on iStent procedures in state/area i during year t.

n: Total number of states/areas included in the analysis; currently equal to 51.

q: Total number of years in the analysis; currently equal to 9.

x_it: Vector of covariates specific to state/area i and year t.

x_it1: Median age of those aged 65 and older.

x_it2: Percent of the population aged 65 and older that is male.

x_it3: Percent of the population aged 65 and older that is White.

x_it4: Percent of the population aged 65 and older that is living under 100 % federal poverty level.

b_1i: State/area i random effect accounting for state-specific variability in the slopes.

${\sigma }_b^2$: Random effect variance parameter.

Results:

Trends in the Glaucoma Procedure Numbers and Payment

From 2007 to 2017, the number of selected glaucoma surgical procedures increased from 76,827 to 177,422. The number of trabeculectomy cases decreased from 45,463 to 23,177 while GDI’s increased from 16,014 to 20,321, peaking at 21,323 in 2015. Among MIGS, the number of iStent procedures rose from 783 in 2012 to 84,454 in 2017. In the same time period, the number of goniotomies increased from 110 to 11,904 and transluminal dilation procedures rose from 916 to 4,155. The number of endocyclophotocoagulations fluctuated over the years, but overall increased from 15,254 in 2007 to 17,722 in 2017, peaking in 2016 at 21,023. Ab Externo shunts rose and fell over this time period, beginning in 2008 at 830 and ending in 2017 at 5,778 while peaking in 2012 at 9,949. (Table 1)

Table 1.

Number of Glaucoma Surgical Procedures under Medicare Part B, 2007–2017

Procedure	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	2017
Trabeculectomy	45,463	43,152	39,776	35,961	32,145	30,218	29,784	28,188	28,497	25,905	23,177
GDI	16,014	16,451	15,932	17,544	17,848	19,612	20997	21219	21323	21218	20321
Total	61,477	59,603	55,708	53,505	49,993	49,830	50,781	49,407	49,820	47,123	43,498
iStent	0	5	5	34	42	783	14301	33093	52106	75794	84454
ECP	15,254	17073	20058	20617	19675	20361	20174	19368	20230	21023	17722
Goniotomy	96	147	189	163	95	110	223	159	249	4986	11904
CyPass	0	0	0	0	0	0	0	0	0	0	6252
Ab externo shunts	0	830	4686	7490	9947	9949	9561	9645	8404	7083	5778
Transluminal Dilation	0	0	0	0	399	916	754	576	1578	2125	4155
XenGel	0	0	0	0	0	0	0	0	0	0	2539
Total	15,350	18,055	24,938	28,304	30,158	32,119	45,013	62,841	82,567	111,011	132,804
Total	76,827	77,658	80,646	81,809	80,155	81,984	95,794	112,248	132,387	158,265	177,422

GDI= Glaucoma Drainage implant

ECP= Endocyclophotocoagulation

In terms of payments, the total inflation-adjusted Medicare part B expenditure for these selected glaucoma procedures rose from $52.0 million in 2007 to $179.9 million in 2017. (Table 2) In 2017, iStent represented 57.9% of the total payment. Furthermore, the $103.3 million increase in iStent payment between 2012 and 2017 represented 92.2% of the total $112 million increase in the Medicare Part B expenditure. (Figure 1A) During this period, the proportion of the total payment contributed by trabeculectomy and GDIs decreased from 70.3% to 21.2%. (Figure 1B)

Table 2.

Total Payments ($) for Glaucoma Surgical Procedures under Medicare Part B, 2007–2017

Procedure Name	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	2017
Trabeculectomy	35,622,930.00	32,831,434.88	31,398,160.15	30,727,851.83	28,929,186.81	26,846,354.92	26,794,521.44	25,444,009.45	25,236,046.35	19,441,751.73	17,784,661.07
GDI	12,390,395.00	12,962,026.82	14,011,623.05	17,404,646.20	19,404,332.45	20,874,059.77	22,471,505.54	22,311,195.32	22,004,204.02	22,151,858.11	20,277,437.52
Total ($)	48,013,325.00	45,793,461.70	45,409,783.20	48,132,498.03	48,333,519.26	47,720,414.69	49,266,026.98	47,755,204.77	47,240,250.37	41,593,609.84	38,062,098.59
ECP	3,960,694.11	4,334,042.90	5,399,952.97	5,544,164.72	5,873,701.74	6,878,332.71	6,912,146.25	6,659,377.90	8,358,105.76	8,985,093.22	7,418,693.33
iStent	0.00	3,087.83	3,250.09	29,498.38	58,211.15	891,336.29	14,435,464.78	32,813,993.92	52,543,920.73	79,419,527.49	104,160,817.30
XenGel	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	2,893,096.40
CyPass	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	6,325,285.27
Goniotomy	48,391.29	56,767.79	68,865.90	69,801.42	59,277.58	64,677.50	147,738.38	92,085.55	133,389.00	4,626,268.17	10,087,763.19
Transluminal Dilation	0.00	0.00	0.00	0.00	451,858.39	985,075.12	793,647.39	590,256.70	1,649,320.25	1,295,635.07	4,033,813.18
Ab-Externo Aqeuous Drainage Device	0.00	676,022.15	5,511,850.09	8,828,059.07	11,556,713.26	11,271,953.27	10,363,213.02	10,157,616.91	8,803,221.91	7,438,163.70	5,749,719.85
Total ($)	4,009,085.40	5,069,920.67	10,983,919.06	14,471,523.59	17,999,762.12	20,091,374.88	32,652,209.82	50,313,330.98	71,487,957.65	101,764,687.65	140,669,188.52
Overall Total ($)	52,022,410.00	50,863,382.37	56,393,702.26	62,604,021.62	66,336,284.36	67,851,590.60	81,922,985.97	98,077,051.44	118,735,894.60	143,497,070.88	179,854,398.22

GDI= Glaucoma Drainage implant

ECP= Endocyclophotocoagulation

Figure 1A.

Medicare Part B Payments for Select Glaucoma Surgical Procedures, 2007–2017

Figure 1B.

Medicare Part B Payments for Select Glaucoma Surgical Procedures except iStent, 2007–2017

Chow’s test was performed to examine the payment trends over years for each procedure as well as the total payment. The curves for the total overall Medicare payment (P<0.001) and iStent payment (P<0.001) both had significant structural break in 2012 when the slopes became steeper. (Figure 1A) The payment curves for GDI and ab externo devices also had significant break points in 2012 (all P<0.001), but the curve plateaued and turned downward, respectively, unlike the iStent and total payment curves. There were no significant change points for other procedures including trabeculectomy.

State Variations in the iStent proportions and speed of adoption

The yearly trend in the proportion of iStent over total payment was broken down by states from 2012 to 2017. There was a large variation in terms of iStent usage among the states. In 2017, the 5 states with the highest iStent usages were South Carolina (76.6%), Vermont (75.3%), Iowa (74.4%), New Jersey (73.9%) and Nebraska (73.2%). The 5 states with the lowest iStent proportions were Maryland (23.4%), Arkansas (26.2%), Rhode Island (31.7%), Connecticut (33.1%), and Michigan (36.7%).

A mixed-effect regression model was built to examine and quantify the association between iStent proportion and state-level covariates, including median age, male population proportion, ethnicity, and income level. The covariate-controlled analysis shows widespread increases in iStent use over the time period, but the rate of adoption varied widely between states (Table 3). These slopes of the per-year iStent proportion increase, adjusted for the aforementioned covariates, were plotted as a choropleth over the map of the United States (Figure 2). The 5 states with the highest trend slopes, or the fastest to adopt iStent, were New Jersey (14.5%, 95% Confidence Interval 12.3 – 15.6%), South Carolina (14.0%, 12.6% – 15.9%), Iowa (13.2%, 11.0% – 14.3%), Missouri (12.6%, 10.7% – 14.0%), and Kentucky (12.3%, 10.8% – 14.1%). Conversely, the 5 states with the lowest slopes, or the slowest states to adopt iStent, were West Virginia (5.1%, 3.5% - 6.9%), Maryland (5.1%, 3.0% - 6.4%), Arkansas (5.3%, 3.5% – 6.9%), Michigan (6.2%, 4.3% – 7.6%), and Wisconsin (6.3%, 4.4% – 7.7%).

Table 3.

States and Yearly Regression Coefficients (iStent Adoption Rates), 2012–2017

STATE	Estimated Slope (%/Year) [95% CI]	STATE	Estimated Slope (%/Year) [95% CI]
WEST VIRGINIA	5.1 [3.5 – 6.9]	N. CAROLINA	9.9 [8.1 – 11.5]
MARYLAND	5.1 [3.0 – 6.4]	NEBRASKA	10.0 [8.0 – 11.3]
ARKANSAS	5.3 [3.5 – 6.9]	HAWAII	10.0 [8.3 – 11.3]
MICHIGAN	6.2 [4.3 – 7.6]	VERMONT	10.3 [9.0 – 12.3]
WISCONSIN	6.3 [4.4 – 7.7]	OHIO	10.4 [8.4 – 11.7]
MONTANA	6.5 [5.2 – 8.5]	ARIZONA	10.6 [9.0 – 12.3]
LOUISIANA	6.8 [5.0 – 8.4]	NEVADA	10.6 [9.8 – 13.1]
CONNECTICUT	7.2 [4.9 – 8.3]	TENNESSEE	10.7 [9.1 – 12.5]
RHODE ISLAND	7.3 [5.4 – 8.7]	DELAWARE	10.8 [9.1 – 12.5]
NEW HAMPSHIRE	7.3 [5.9 – 9.2]	WYOMING	10.8 [9.7 – 13.1]
ALABAMA	7.7 [5.8 – 9.1]	PENNSYLVANIA	10.8 [8.6 – 12.0]
INDIANA	8.0 [6.1 – 9.5]	ILLINOIS	10.9 [8.8 – 12.1]
FLORIDA	8.0 [6.0 – 9.3]	GEORGIA	11.1 [9.5 – 12.8]
MASSACHUSETTS	8.2 [6.1 – 9.5]	COLORADO	11.1 [9.9 – 13.2]
TEXAS	8.2 [6.6 – 10.0]	S. DAKOTA	11.2 [9.5 – 12.8]
NEW YORK	8.6 [6.4 – 9.7]	N. DAKOTA	11.6 [9.5 – 12.9]
KANSAS	8.6 [6.6 – 10.0]	CALIFORNIA	11.9 [10.3 – 13.6]
WASHINGTON	9.0 [7.5 – 10.9]	IDAHO	12.0 [10.8 – 14.2]
VIRGINIA	9.0 [8.4 – 11.7]	UTAH	12.2 [10.8 – 14.1]
OREGON	9.1 [7.7 – 11.1]	KENTUCKY	12.3 [10.8 – 14.1]
MISSISSIPPI	9.1 [7.3 – 10.6]	MISSOURI	12.6 [10.7 – 14.0]
OKLAHOMA	9.6 [7.7 – 11.0]	IOWA	13.2 [11.0 – 14.3]
MINNESOTA	9.7 [7.8 – 11.1]	S. CAROLINA	14.0 [12.6 – 15.9]
MAINE	9.8 [8.7 – 12.0]	NEW JERSEY	14.5 [12.3 – 15.6]

^*New Mexico and Alaska were excluded due to incomplete data.

Figure 2.

Map of State-Level iStent Adoption Rates, 2012–2017

Potential Drivers of the Variation

The regression analysis shows that the male proportion of the population over 65 and median age were statistically significant covariates (P=0.003 and 0.026 respectively). A 5 percent increase in the male population percentage was associated with an 12.4% (95% confidence interval: 4.3% to 20.5%) increase in the state iStent proportion. In contrast, a 1 year increase in the median age of the population above 65 was associated with −3.6% (95% confidence interval: −6.8% to −0.4%) change in the state’s iStent proportion. Ethnicity, estimated by percent of state population who is Caucasian, and proportion of state population living under federal poverty line was not significantly associated with yearly changes in iStent usage. (Table 4)

Table 4.

Mixed Effect Regression Analysis on Proportion of Medicare Part B Glaucoma Spending Due to iStent.

Variable	Parameter Estimate	95% Confidence Interval	P Values
Median Agea,c	−0.036	(−0.068, −0.004)	0.026
% Maleb,c	0.124	(0.043, 0.205)	0.003
% Caucasianb,c	−0.003	(−0.011, 0.005)	0.517
% Povertyb,c	−0.002	(−0.048, 0.044)	0.930
Yeara	0.095	(0.087, 0.102)	<0.0001

^aPer 1 year increase.

^bPer 5% increase.

^cAmong those aged 65 years and older.

Discussion:

The present study focused on the glaucoma surgical expenditure for Medicare Part B beneficiaries between the years 2007 and 2017. We identified significant, ongoing shifts in the surgical treatment of glaucoma in the US through comparing the usage of trabeculectomy and GDI with MIGS procedures, especially iStent. The geographic distribution of iStent usage, and the potential causes that underlie the rapid adoption of this relatively new modality have not been previously described.

There have been previous studies focusing on the trends of glaucoma healthcare delivery for Medicare beneficiaries. Quigley et al. used the Medicare billing information to study the cost of glaucoma in Medicare Part A and B beneficiaries and concluded that glaucoma care costs per person has decreased between 2002–2009.⁶ Arora et al. examined the use of glaucoma surgeries and laser procedures by Medicare Part B beneficiaries in the years 1994 to 2012, describing a decreasing number of glaucoma procedures despite rising number of beneficiaries and a trend away from trabeculectomy.¹³ Our results corroborate the decline of trabeculectomy, which appears to have steadily continued.^7,10 As for GDI’s, the number of cases increased gradually between 2007–2013. This is consistent with the previously reported shift in practice^10,11 that may have been influenced by the Tube versus Trabeculectomy Study published in 2005.¹² However, our results also show that GDI usage also seems to be plateauing in the recent years. Not surprisingly, the total Medicare payment for trabeculectomy and GDI have also decreased.

The economic impact of these changes is highlighted by the increasing popularity and usage of MIGS, especially of iStent. In the period following its approval by the FDA, from 2012 to 2017, the increase in the Medicare part B spending on iStent represents the large majority (92.2%) of the total increase in glaucoma surgical expenditure. iStent has been adopted to a wide extent in nearly every individual state. While the rates of MIGS adoption vary, roughly half of the states saw increases in iStent proportion of more than 10% a year throughout the 6-year period. The large numberl of patients with mild to moderate glaucoma for whom iStent is indicated may have contributed significantly to its increased usage. While the IOP-lowering effect of iStent is modest compared to trabeculectomy and GDI, studies have established an excellent safety profile and ability to reduce long-term topical medication burden to the patients.^9,14 It adds minimal risk and operating time to a cataract extraction procedure, which may make it appealing to surgeons.^15,16 In addition, unlike trabeculectomy and GDI surgeries, which are typically performed by fellowship-trained glaucoma subspecialists, iStent may be performed by general ophthalmologists without the need of subspecialty training. Additionally, the decreasing Medicare payment for stand-alone cataract surgery may also have boosted the utilization rates for iStent.^17,18

The dramatic increase in the Medicare expenditure for iStent brings to attention the need to study its cost-effectiveness. To date, much of the cost analysis for iStent have occurred outside of United States. Ordonez et al. in Colombia,¹⁹ Tan et al. in United Kingdom,²⁰ Ngan et al. in New Zealand²¹ and Patel et al. in Canada²² have separately assessed the cost-effectiveness of iStent. These studies generally agreed that the iStent procedure is largely cost-effective due to its ability to reduce the number of glaucoma medication and its safety profile, which limited the number of postoperative complications. In the United States, Berdahl et al. performed a cost-comparison study between selective laser trabeculoplasty versus medications versus two iStents and similarly concluded that while the two iStents had higher costs upfront, they offered the lowest annual costs thereafter and resulted in savings after 5 years.²³

The present study examined only the Medicare payments for glaucoma surgical procedures. It is possible that the rise in the Medicare spending on glaucoma surgery, due to increase in the number of iStent procedures, may be offset by the decrease in the number of medications prescribed in the same population via Medicare part D or in the number of laser procedures commonly considered in the mild-to-moderate glaucoma patients for whom iStent is indicated. While previous studies suggest that rising iStent usage may lead to overall net negative in terms of glaucoma treatment expenditure, their findings do not necessarily indicate that iStent represents the most cost-effective surgical modality. Other MIGS modalities have also entered the market,²⁴ and as illustrated in Figure 1, 2017 saw several MIGS procedures begin to rise in their own payment curves, including Xen gel stent and procedures under the umbrella of goniotomy. Future comparisons in the cost-effectiveness between iStent and these upcoming procedures along with the analysis of glaucoma medication and laser procedure trends could elucidate more regarding the general effect of MIGS on overall glaucoma expenditures.

Our regression analysis suggests higher male proportion in the population and lower median age as the two potential drivers for adoption of iStent. One possible explanation for why the male gender leads to higher utilization of the iStent is the difference in adherence with medical management. Several studies have associated males with decreased adherence with medical glaucoma therapy.^25–28 Decreased need for medical therapy is one of the major benefits of iStent,²⁹ and thus ophthalmologists may be offering iStent more to male patients in the setting of difficulty with adhering to topical medication regimen. As for the lower iStent usage with increasing age, in the relatively older Medicare beneficiary population, the severity of existing glaucoma may call for modalities that offer higher therapeutic impact such as trabeculectomy and GDI.

This study has several limitations. Medicare Part B Summary is a database that provides only the number of services and total payment per procedure type per year by state. The database does not include patients with private insurance, which could skew the portrayal of national trends. The treatment outcomes are not linked to individual cases and the clinical benefit or deficits associated with the increasing Medicare expenditure may not be ascertained. The lack of patient context for the procedures, in terms of the glaucoma severity and prior procedures, is also a limitation. Similarly, the clinical context such as gender of the surgeon and the surgical setting of the procedures could also be factors affecting iStent usage but is lacking in the database.

There are many factors outside those considered in this study that could affect the use of iStent at the state-level, such as number of ophthalmologists, number of glaucoma specialists, and the changing financial incentives for glaucoma procedures. Considered together with those, the significance of the demographic factors may change. Finally, the demographic data used to build the regression analysis model were derived from the census data which raises the concern of ecological fallacy, that an inference is made about an individual based on aggregate group data. We sought to address this issue by extracting only the most pertinent data for the Medicare part B beneficiaries by using the census responses from a population over the age of 65. The regression analysis was constructed using such a subset in order to decrease the risk for error when estimating the effect of demographic factors on the iStent utilization trends.

Overall, this study utilized a large database to capture select glaucoma procedures across the United States. The breakdown of the number of allowed services and total payment at the State level enabled examination of potential factors that may be influencing the ongoing shifts in glaucoma surgical practice. MIGS, especially iStent, have played a major role in the rising Medicare glaucoma surgical procedure expenditure as seen in this analysis of trends from 2007 to 2017. The iStent was responsible for over 90% of the increase in the select glaucoma procedural expenditure and has quickly surpassed trabeculectomy and GDI in payment since its entrance to market in 2012. The adoption of iStent occurred universally throughout individual states with wide range in the speed of adoption and demographic differences may be driving those differences. There are several other factors that could also be contributing to the ongoing increase in the iStent usage. This along with the impact of iStent on the comprehensive glaucoma Medicare spending warrant future research.