Incremental Health Care Expenditures Associated With Glaucoma in the United States: A Propensity Score–matched Analysis

Supplemental Digital Content is available in the text. Précis: Adjusting for sociodemographics and comorbidities, patients with glaucoma incur an annual incremental economic burden of $1863.17, translating to $9.2 billion nationally. When analyzed by the health care service sector, prescription medication expenditures were higher for glaucoma patients. Purpose: The purpose of this study was to estimate the incremental health care burden, defined as attributable costs solely due to a diagnosis, of patients with diagnosed glaucoma, controlling for comorbidities, and sociodemographics. Design: A retrospective cross-sectional analysis of Medical Expenditure Panel Survey (MEPS) participants (age above 18 y) between 2016 and 2018. Methods: A cross-validated 2-part generalized linear regression model estimated the incremental glaucoma expenditures in aggregate and by sociodemographic subgroups and health care service sector [inpatient, outpatient (including surgical procedures), emergency room, home health, and medications] after 1:3 propensity matching. Results: After 1:3 propensity matching for sociodemographics and the Charlson Comorbidity Index, this study analyzed 1521 glaucoma patients (mean expenditures: $13,585.68±1367.03) and 4563 patients without glaucoma (mean expenditures: $12,048.92±782.49). A higher proportion of glaucoma patients are female, elderly, publicly insured (Medicare/Medicaid), college educated, identify ethnically as non-Hispanic, reside in the Northeast, and have more comorbidities (P<0.001). There were no differences in health care burden based on sex, income, insurance status, education, and year of care received for patients with glaucoma. Controlling for comorbidities and socioeconomic factors, propensity-matched glaucoma patients incur an annual incremental health care burden of $1863.17 (95% confidence interval, 393.44-3117.23, P=0.013), translating into an additional $9.2 billion in population-level US health care expenditures. By health care service sector, the expenditure ratio for health care expenditures was higher for prescription medications (expenditure ratio=1.20, 95% confidence interval, 1.02-1.42, P=0.031). Conclusions: Glaucoma patients have a substantial incremental economic health care burden after accounting for demographics and comorbidities, largely secondary to prescription medications. There is a need to continue identifying and studying treatment options for patients with glaucoma to maintain vision while minimizing health care expenditures.

G laucoma is a major source of morbidity and mortality worldwide and a major contributor to blindness both domestically and internationally. 1 The national burden of glaucoma in the United States for the entire population is estimated to be $5.8 billion and is expected to increase due to the aging American population. 2 For adults over 40 years of age in the United States, the estimated prevalence of diagnosed glaucoma is 2.1%, 3 and the prevalence of undiagnosed glaucoma is as high as 2.9%. 4 The prevalence of both diagnosed and undiagnosed glaucoma is expected to increase in the future with an aging population and other demographic changes. 5 There have been many studies analyzing the economic burden of glaucoma and the drivers behind those costs. [6][7][8][9][10][11][12] However, some of these studies were based on European populations and were not applicable to the general American population. 10,12 Some of these studies were not representative of the national population and were limited in their sample sizes. 6,7 In addition, these studies are 9 or more years old, and there is a need for more recent studies focusing on glaucoma's economic burden more currently. In recent years, though there has been a shift towards surgical procedures such as microinvasive glaucoma surgeries (MIGS), sometimes used as first-line therapy in lieu of the traditional medical management, literature regarding expenditure profiles for various treatment options continues to be updated with recent research reporting that conventional glaucoma surgeries and selective laser trabeculoplasty surgeries are more cost-efficient surgical methods to reduce intraocular pressure (IOP) when compared with MIGS. [13][14][15][16][17] To develop strategies for more affordable, efficient, and effective health care, there needs to be additional evidence on the direct incremental economic burden seen in patients treated for glaucoma. The incremental economic burden is defined as the health care expenditures for a patient solely due to the diagnosis of a condition and independent of their sociodemographic characteristics and other comorbidities.
In the present study, 3 years of data from a nationally representative database, the Medical Expenditure Panel Survey (MEPS), was utilized to calculate the estimated economic health care burden for patients with glaucoma, both by aggregate and by class of health care intervention [inpatient, outpatient (including surgical procedures), emergency room, home health, and prescription medications]. An incremental health care burden was calculated for these patients after adjusting costs to be independent of the effects of comorbidities and sociodemographic factors.
This study aims to estimate and offer further insights into this incremental health care burden and to elucidate drivers behind these health care expenditures.

Study Population
The MEPS is a set of large-scale surveys of families and individuals, their medical providers, and employers across the United States and was used for this retrospective study. 18 The MEPS survey is cosponsored by the Agency for Health care Research and Quality (AHRQ) and the National Center for Health Statistics (NCHS). It is a nationally representative survey of the noninstitutionalized American civilian population based on a subsample of households that participate in the National Health Interview Survey (conducted by the National Center for Health Statistics) and collects patient data such as medical expenses, demographic characteristics, health conditions, and access to care using questionnaires fielded to individual household members and their medical providers. More details about the MEPS designs and methods are available elsewhere. 18 Institutional review board approval was not needed to use this deidentified and publicly available database, and waiver of approval was obtained for this study. This study is a retrospective analysis of 3 years of data from the MEPS database (2016-2018), which was used to assess health care system expenditures associated with patients with glaucoma.
To analyze expenditures with demographic characteristics and medical diagnoses, the household component of the MEPS database and medical condition files were linked. Patients over the age of 18 were then stratified based on the presence of glaucoma based on the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes (H40.xx for glaucoma) that can be found on the MEPS Web site. 18 The complete ICD-10-CM code is not included in the MEPS database, and it is not possible to ascertain glaucoma severity.

Study Variables and Statistical Analysis
For each patient, total expenditures, which included inpatient, outpatient, emergency room, home health visits, medications, and all payments by third-party payers and out-of-pocket costs, were recorded. Mean expenditures were calculated for the entire cohort and the cohort with and without a recorded medical diagnosis of glaucoma from 2016 to 2018.
All expenditures were adjusted to 2018 US dollars using the urban Consumer Price Index of the US Bureau of Labor Statistics. 19 This study accounted for the complex survey design (sampling weights, clustering, and stratification) to more accurately estimate aggregate health care expenditures using the R survey package. With this data, we investigated the per-capita system health care expenditures based on whether patients had a diagnosis of glaucoma and secondarily stratified based on certain sociodemographic characteristics and the presence of comorbidities.
Self-identified race/ethnicity was categorized as non-Hispanic white, Hispanic, black, American Indian, or Asian. Age was categorized as a continuous variable. Educational level was categorized as no degree obtained, high school diploma or equivalent, or a college degree or higher. Income was calculated in relation to the 2018 consumer price index. Individuals were categorized as poor [ < 100% of federal poverty limit (FPL)], near poor (100% to 124% of FPL), low income (125% to 199% of FPL), middle income (200% to 399% of FPL), or high income ( > 399% of FPL). Insurance status was categorized as private (includes patients with concurrent Medicare coverage), public (Medicare or Medicaid patients with no private insurance), or uninsured. For regional analysis, patients were categorized as living in the Northeast, South, Midwest, or West based on US census classifications. The Charlson Comorbidity Index (CCI), which was used to quantify comorbidities, 20 is calculated as a sum of various comorbidities of an individual such as congestive heart failure, myocardial infarction, and diabetes since these comorbidities independently contribute to costs and health outcomes and was analyzed as a continuous variable (Supplemental Table 1, Supplemental Digital Content 1, http://links.lww.com/IJG/A585). 21 Demographic characteristics of the cohorts with and without glaucoma were compared using Rao-Scott χ 2 tests. A P-value of 0.05 was used to determine statistical significance. To account for additional confounders that may contribute to increased costs outside of the specified covariates above, patients with glaucoma were matched to controls (patients without glaucoma) based on propensity scores to minimize the effect of confounders between these 2 groups. A propensity score is assigned to each patient. It is calculated as the probability for each patient based on their baseline sociodemographic characteristics and comorbidities (age, sex, race, income, education, region, comorbidity index, and year of treatment) to have glaucoma (the experimental group). Each patient in the glaucoma cohort was matched to 3 patients in the patients without glaucoma cohort with a similar propensity score to minimize the confounding effect of baseline characteristics on analysis, which has been repeated in prior studies. 22 The remainder of the analysis on these 2 groups was performed on the group that was matched based on propensity scoring.
Health care expenditure data such as this has a high concentration of zero expenditures and positive skewing of expenditures. 23 A 2-part regression model was used to compare the patient cohorts with and without glaucoma. This 2-part model consists of a logistic regression model to predict the likelihood of nonzero expenditure and a generalized linear model with gamma distribution with log-link function to address the positive skewness of expenditure data. 24,25 A gamma-distribution model is more accurate than log ordinary least squares models due to reduced normality and homoscedasticity assumptions. 25 This model was then used to estimate expenditure ratios (ERs), which represents the multiplicative effect of a variable when compared withthe reference variable, and accounted for the effects of demographic characteristics and the CCI. 95% confidence intervals (CIs) were calculated for these ERs. For patients with glaucoma, each individual sociodemographic variable was analyzed with respect to a reference variable after adjusting for all others. Age and CCI were analyzed as continuous variables. The reference variables for each category were female (sex), non-Hispanic white (race/ethnicity), Northeast (geographic region), poor income (income), no high school degree (educational attainment), and being privately insured (insurance status). For example, an ER of 3 for males relative to females suggests that males with glaucoma have 3 times the incremental burden of glaucoma compared with females independent of comorbidities and all other sociodemographic characteristics. The 2-part model was also used to compare costs both in aggregate and by each health care service sector for patients with and without glaucoma. For example, an ER of 2 when comparing glaucoma to a reference individual without glaucoma would suggest would indicate twice the health care expenditure for a glaucoma patient when compared with an individual without glaucoma if they had identical sociodemographics and comorbidity index. All of the tables in the study summarize annual averages from 2016 to 2018, which were derived using person-level weights provided by the AHRQ, and was extrapolated to the civilian noninstitutionalized US population.

RESULTS
In total, our sample included 70,824 MEPS respondents over the age of 18 from 2016 to 2018 (Table 1) A higher proportion of glaucoma patients are female, older, college educated, identify ethnically as non-Hispanic, live in the Northeast, have public insurance (Medicare/ Medicaid), and have a higher CCI (P = 0.002 for education, P < 0.001 otherwise) before propensity matching (Table 2). After 1:3 propensity matching, there was no difference in baseline characteristics between the 2 cohorts for the subsequent analysis. Within the glaucoma patient cohort, independent ERs for each recorded sociodemographic variable and CCI with respect to a reference variable was analyzed (Table 3). Age was significantly directly associated with higher costs (ER = 1.01, 95% CI, 1.00-1.02, P = 0.013). Glaucoma patients in the Midwest faced lower costs (ER = 0.70, 95% CI, 0.51-0.96, P = 0.027) when compared with patients in the Northeast. Self-identified Asians reported less expenditures (ER = 0.61, 95% CI, 0.46-0.81, P < 0.001) when compared with non-Hispanic white glaucoma patients. CCI was directly associated with increased costs (ER = 1.26, 95% CI, 1.17-1.35, P < 0.001). There were no differences in health care burden based on sex, income, insurance status, education, and year of care received.
When comparing patients with and without glaucoma, patients with glaucoma had higher overall and prescription medication expenditures after adjusting for demographics and comorbidities (

DISCUSSION
This study reports that a higher proportion of patients with glaucoma are female, elderly (average age: 70 ± 12), college educated, identify ethnically as non-Hispanic, reside in the Northeast, utilize public insurance (Medicare/Medicaid), and have more comorbidities when compared with patients without glaucoma in the MEPS database. These results mirror findings reported by Gupta et al 3 utilizing optic fundus photographs of individuals in the National Health and Nutrition Examination Survey (NHANES) database, in which the prevalence of glaucoma for individuals over 40 years of age was 2.1%. In addition, they report that the prevalence of glaucoma was highest in non-Hispanic blacks followed by non-Hispanic whites and lowest in Mexican Americans, and lower for patients without insurance, likely due to the higher proportion of patients with glaucoma utilizing Medicare. After extrapolating based on current glaucoma prevalence in the MEPS database and calculated health care burden, the estimated annual incremental expenditures for glaucoma, independent of demographic factors and comorbidities, was roughly $9.2 billion annually for glaucoma patients between 2016 and 2018 using the 2018 US population (325.9 million individuals, glaucoma prevalence of 2.16% for those above 18 years of age in the entire MEPS database) in 2018 dollars. This burden does not account for functional limitations faced by patients with glaucoma with such as unemployment, lost productivity, residential care, difficulties in activities of daily living, and premature deaths, as well as patients with undiagnosed glaucoma. 12,13,26 When analyzed by the health care service sector, patients with glaucoma had higher prescription medication expenditures after adjusting for sociodemographics and comorbidities.
Findings in the study supplement existing literature regarding the additional economic burden of glaucoma in patients when controlling for their respective socioeconomic factors and comorbidities. Stein et al, 27    Bold values indicate statistical significance (P < 0.05). *Statistical tests performed: χ 2 test of independence; t test. P-value indicates if the distribution of demographic characteristics is statistically significantly different between the 2 groups. The before matching table compares the demographic characteristics of patients with and without glaucoma for the entire Medical Expenditure Panel Survey (MEPS) database. Propensity matching was utilized to match the baseline characteristics of patients in a 3:1 ratio in the after matching table, and these patients were utilized for subsequent analysis.
FPL indicates federal poverty limit.
With respect to health care service sectors, glaucoma patients only had higher prescription medication expenditures, contributing largely to the increased total expenditures. Increased medication expenditures reported for patients with glaucoma in this study may have a multifactorial etiology.
Many large multicenter trials report that close management of IOPs is important to prevent glaucoma disease progression and is often achieved with regular intraocular medication drops. [30][31][32] Consequently, there is a trend towards more frequent clinic visits in patients with glaucoma to optimize IOP management to prevent worsening vision. 33 Increased medication expenditures for these patients are likely directed towards long-term daily glaucoma medications used to achieve and maintain the target IOP. De Natale et al, 34 in a retrospective cross-sectional study with 2228 patients in a single center in Europe between 1997 and 2002, reported an increase in the use of prostaglandin derivatives and carbonic anhydrase inhibitors and a trend towards medication addition rather than substitution, leading to an increase in the number of medications per patient over time. Lam et al,35 in a retrospective cross-sectional study using MEPS data from 2001 to 2006 on glaucoma medication expenditures, reported significant decreases in beta-blocker expenditures and a significant increase in expenditures on prostaglandin analogs and alpha-agonists. It has been reported that prostaglandin analogs, alpha-agonists, and carbon anhydrase inhibitors have higher costs than beta-blockers in the medical management of glaucoma. [36][37][38] Increased utilization of more expensive medications and total number of medications to mitigate glaucoma progression over the course of the disease contribute to prescription medication expenditures for these patients.
Although this study accounted for comorbidities such as diabetes, human immunodeficiency virus, cardiovascular disease, and renal disease using the CCI, there may be yet other comorbidities, especially ocular comorbidities, 27,39,40 in patients with glaucoma that could be contributing to higher expenditures in these patients. Griffith and Goldberg 39 reported that nearly 15% of patients with glaucoma had retinal comorbidities, and these patients had worse visual outcomes. Stein et al 27 reported that ocular comorbidities such as age-related macular degeneration and having recent cataract surgery contributed to higher overall expenditures for patients with recently diagnosed OAG. There is likely is a screening bias as glaucoma patients undergo ophthalmic exams more frequently, which allow for the detection of other ophthalmologic conditions.
The optimal management of glaucoma and, consequently, the associated expenditures depends on the specific type of glaucoma, including OAG and angle-closure glaucoma. 41 Angle-closure glaucoma is more prevalent among Asian Americans. 42 The reduced health care burden reported by the Asian population in this study may be due to the difference in management and subsequent health care expenditures associated with the subtypes of glaucoma. However, it is not possible to ascertain glaucoma subtype from the MEPS database. Further studies are needed to elucidate the decreased expenditure among Asians in this group.  There are limitations in this study. As mentioned previously, the MEPS database does not include disease subtype and disease severity, both of which would influence expenditure. Furthermore, the MEPS database does not represent individuals in institutionalized settings such as prisons or nursing homes, potentially leading to underrepresented expenditure due to the exclusion of patients with glaucoma in nursing homes. This work is based on observational, crosssectional data leading to bias. Recall bias may also be a confounder, although most patient data is verified by the AHRQ with the clinicians. MEPS clinical diagnoses are limited to the 3-digit ICD-10-CM codes, which affects comorbidities recorded for patients, leading to over and underrepresentation of certain comorbidities and also limits in recording glaucoma disease subtype and severity in analysis. H40 was used as the International Classification of Diseases, 10th Revision (ICD-10) code for glaucoma suspects and glaucoma patients with varying degrees of severity. Management differs significantly based on severity, especially when comparing glaucoma suspect patients to patients with confirmed glaucoma, and it was not possible to account for these differences in this analysis. In calculating health care service sector expenditures, all health care sector visits in which glaucoma was listed as an ICD-10 code was included in the analysis; however, it was not possible to delineate visits in which glaucoma was not the primary reason for the visit leading to potential overestimations of expenditures specifically in the inpatient and emergency room settings. It was not possible to discern public insurance between Medicare and Medicaid. In addition, it was not possible to account for individuals with glaucoma that have not yet been diagnosed since this analysis was done using ICD-10-CM codes. Many of these limitations are inherent to the MEPS database. Despite these limitations, MEPS has been widely used by previous researchers due to its large sample sizes, follow-up period, and recorded cost data for the estimation of disease-attributable expenses in various fields including ophthalmology. 2,22,29,[43][44][45][46][47] This study's findings present updated estimates for expenditures for glaucoma patients accounting for demographics and comorbidities and analyze cost based on various sociodemographic and health care service sectors. Overall, these results suggest prescription medication expenditures among glaucoma patients were the largest contributors to total expenditures. There is a need for further research comparing expenditures and visual outcomes for patients undergoing lasers and MIGS to understand the economic benefits of these procedures relative to medications. There continues to be a health care economic burden for patients with glaucoma independent of demographics and comorbidities, and further research is needed to understand and maximize the costeffectiveness of various glaucoma therapies.