Association Between Preserved Glaucoma Eye Drops Exposure and Glaucoma Surgery Occurrence: A Nationwide 10-Year Outcome Study

Abstract

Introduction

Preserved eye drops induce ocular toxicity, which may impair the management of glaucoma. This study aimed at assessing the association between the level of exposure to preserved glaucoma eye drops and the occurrence of glaucoma surgery.

Methods

This retrospective population-based cohort study included all patients ≥ 18 years old with a first glaucoma eye drops claim in the French national health insurance database between 1 January 2009 and 31 December 2017. A look-back period of 3 years before inclusion was used to select incident cases. Included patients were followed up until 31 December 2018. Patients were grouped according to the level of preservative exposure during follow-up: 0% (only preservative-free [PF] eye drops), “low-mixed” (both preserved and PF eye drops, with < 50% preserved), “high-mixed” (both preserved and PF eye drops, with ≥ 50% preserved), and 100% (preserved eye drops only) exposure. The occurrence of glaucoma surgery was analyzed for each group, and its association with preservative exposure estimated using multivariate Cox regression models, adjusted on age, sex, history of high blood pressure and cardio-neurovascular disease, glaucoma disease severity, and year of patient inclusion.

Results

We included 664,494 patients (57.0% women; median age 66.0 [interquartile range 57.0–75.0] years) with a first glaucoma eye drops claim and a median follow-up of 4.8 (2.8–7.1) years. Among them, 8,467 (1.3%) had glaucoma surgery during follow-up. The hazard ratio of glaucoma surgery increased with the level of exposure to preservatives from 2.12 (95% confidence interval 1.81–2.48) in the low-mixed group to 4.57 (3.94–5.29) in the 100% preserved group compared with the 0% preserved exposure group.

Conclusions

The risk of glaucoma surgery was associated with the level of exposure to preserved glaucoma eye drops in this nationwide cohort study. Large-scale, randomized studies would be beneficial to investigate whether PF eye drops could decrease or delay glaucoma surgery occurrence.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40123-025-01203-1.

Key Summary Points

Why carry out this study?

Preserved ocular hypotensive eye drops such as benzalkonium chloride (BAK)-containing formulations cause ocular toxicity, which may negatively affect the long-term management of glaucoma.

Based on the French national health insurance database (SNDS, Système National des Données de Santé), which covers 99% of the French population, the present study investigated the association between the level of exposure to preserved glaucoma eye drops and the occurrence of glaucoma surgery in France during a 10-year period.

What was learned from the study?

In this nationwide cohort study including 664,494 patients, the risk of glaucoma surgery was found to increase with the level of exposure to preserved ocular hypotensive eye drops.

The findings of this study support an association between the risk of glaucoma surgery and the level of exposure to preserved glaucoma eye drops and therefore suggest that switching to preservative-free formulations may translate to a decreased or delayed need for surgery.

Introduction

Excessively high intraocular pressure (IOP) is a major risk factor for glaucoma onset and progression for patients with ocular hypertension (OHT) or open-angle glaucoma (OAG), and usually requires long-term therapy. Despite advances in laser and surgical treatments, lowering of IOP with topical drugs remains the initial treatment of choice for most patients with OAG or OHT [1, 2]. However, monotherapy fails to achieve a satisfactory IOP reduction in 40–75% of patients with glaucoma after > 2 years of therapy, and the use of add-on therapy may be required to lower the patient’s IOP to a desirable target pressure [3].

Patients with OAG or OHT have a higher prevalence of ocular surface disease (OSD) such as dry eye, meibomian gland dysfunction, and chronic allergy compared to the general population [2, 4]. Such adverse effects can be due to the active pharmaceutical ingredient, the drop vehicle, or the preservatives added to formulations because of their antimicrobial or antifungal properties. The cytotoxic effects of preservatives, especially benzalkonium chloride (BAK), on the ocular surface and trabecular meshwork have been extensively documented [5]. Signs and symptoms of OSD decrease when BAK-preserved drops are substituted with preservative-free (PF) drops [6–9]. This may be a source of nonadherence to medications, ultimately leading to poor IOP control and the need for glaucoma surgery [10, 11]. The benefit of PF formulations, whose development has notably increased in the last 15 years, has been acknowledged by the European Medicines Agency [12]. An increasing number of experimental and observational studies also support the use of such PF formulations [13, 14]. The present study is based on the whole French national health insurance database [15], which covers 99% of the French population (> 66 million people), and aims to further investigate the association between the level of exposure to preserved glaucoma eye drops and the occurrence of glaucoma surgery in France during a 10-year period.

Methods

Data Source

Nationwide data were extracted in March 2020 from the French national health data system “Système National des Données de Santé” (SNDS). The SNDS contains comprehensive, individualized, and anonymized data on health spending reimbursements for most individuals living in France, i.e., 99% of the French population (> 66 million people) [15, 16]. This includes sociodemographic data (such as age, sex, area of residence, date of death) and medical characteristics of beneficiaries with all medical reimbursed claims, including private clinic and public hospital care from the hospital discharge database “Programme de Médicalisation des Systèmes d’Information” (PMSI), office medicine, paramedical intervention, laboratory tests (without the results), medical devices, outpatient medication prescriptions and deliveries, and costly innovative drugs in hospital settings. Data also include medical procedures coded according to the Common Classification of Medical Procedures and the date and duration of hospitalizations with the main, related, and associated diagnoses coded with the International Classification of Diseases, 10th revision (ICD-10).

Study Design and Population

This retrospective population-based cohort study included all patients aged ≥ 18 years who had a first glaucoma eye drops claim recorded in the SNDS database between 1 January 2009 and 31 December 2017 and with a look-back period of 3 years. The index date was defined as the first glaucoma eye drops being delivered during the study period. This included only new users who had no prior prescriptions for glaucoma eye drops within the look-back period of 3 years preceding the first delivery. During the first year of follow-up, glaucoma surgeries were identified. If glaucoma surgery had occurred within this first year, patients were included in the study if they had received at least 3 consecutive months of glaucoma eye drop treatment. In the absence of glaucoma surgery during the first year, patients needed to have at least 6 consecutive months of treatment to be included in the study. In addition, patients with a history of medical or surgical glaucoma treatment or surgery at risk for ocular hypertension (based on codes for trabeculectomy, deep sclerectomy with or without viscocanaloplasty, trabeculotomy, cyclodestruction, laser trabeculoplasty, ab interno or ab externo vitreoretinal surgery, and/or keratoplasty; see Table S1 in the Electronic Supplementary Material for details) were excluded. Patients were followed up from the index date until 31 December 2018, or until death occurrence, absence of healthcare reimbursement for at least 12 months or occurrence of glaucoma surgery, whichever occurred first. The study design is further presented in Fig. 1.

Fig. 1.

Study design

This study was submitted to the French Health Data Hub (HDH) in June 2019 and obtained a favorable opinion from CESREES (Ethical and Scientific Committee for Research, Studies and Evaluation in Public Health) in July 2019. The study was approved by the CNIL (French Data protection authority) on August 2, 2019 (DR-2019-217/No 919284). Permission to access and use data from the database utilized in this study was obtained from the CNIL. As this study was a retrospective analysis of data from the SNDS, no informed consent was required.

Exposure to Preservatives

All glaucoma eye drops (see Table S2 for the list of products considered in the analysis) dispensed and reimbursed to the patients during the follow-up period (from inclusion to surgery or end of follow-up) were extracted. Eye drop exposure was assumed to start at the date of delivery. The number of eye drops (with or without preservatives) delivered to each patient was calculated based on 1-month treatment supplies (e.g., for a 3-month supply, 3 eye drops claims were counted). For each year of follow-up, patients were then divided in four groups according to the rate of exposure to preserved eye drops (calculated as the ratio of the number of preserved eye drops received to the total number of eye drops received during the year): “0% preservatives” (rate = 0%; patients received PF eye drops exclusively); “Low-mixed” (patients received both preserved and PF eye drops with < 50% preservatives); “High-mixed” (patients received both preserved and PF eye drops with ≥ 50% of preservatives); and “100% preserved” (rate = 100%; patients received preserved eye drops exclusively).

Study Outcomes

The main study outcome was the occurrence of a surgical procedure for glaucoma identified by any hospitalization (as main or related diagnosis of open-angle (ICD-10: H40.1) or unspecified (H40.9) glaucoma) with one of the following surgical codes: BEFA008 for trabeculectomy, BGFA014 for deep sclerectomy, BEPA003 for trabeculotomy performed via a sclerotomy, and BGFA900 for deep sclerectomy with visco-canaloplasty. Microinvasive glaucoma surgery was not considered since a specific code was not available during the study period and few procedures were performed in France at that time [17]. Additionally, patients who underwent surgery for angle-closure, congenital, post-traumatic, or uveitic glaucoma were not included in the analysis.

The number of ocular hypotensive drugs (molecules) dispensed simultaneously was used as an indicator of disease severity.

Statistical Analysis

All statistical analyses were conducted using SAS^® software, version 9.4 (SAS Institute, Cary, NC, USA). These consisted of descriptive statistics for the quantitative variables (described presenting the median and interquartile range [IQR]) and frequency distribution for the categorical variables (described presenting the count and percentage of each modality).

For analysis of the SNDS database, results below or equal to a threshold of 10 patients were reported as “N ≤ 10” to comply with the European General Data Protection Regulation (GDPR) and French data privacy protection guidelines [18, 19].

Besides the summary statistics, multivariate Cox regression models were used to evaluate the association between preservative exposure and occurrence of glaucoma surgery at different time periods from inclusion (2, 3, 5, 7, and 10 years post-inclusion). Models were adjusted for age, sex, history of systemic hypertension, and cardio-neurovascular disease (see Table S3 in the Electronic Supplementary Material for details), number of ocular hypotensive drugs (molecules), dispensed simultaneously (proxy of glaucoma disease severity), and year of patient inclusion. Patients who did not undergo surgery were censored at the end of follow-up or date of death. Adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) of glaucoma surgery were estimated according to preservative exposure, with the “0% preserved” group as the reference. Statistical significance was defined as P < 0.05.

Results

Study Sample

A total of 2,352,253 patients had at least one glaucoma eye drops claim between January 1, 2009, and December 31, 2017 (study inclusion period; Fig. 2). Among them, 1,010,965 aged ≥ 18 years had a minimum 3-year delay without any claim before this first dispensed treatment, and 978,224 did not have exclusion criteria (glaucoma surgery, vitreoretinal surgery, and/or keratoplasty) within the 3-year look-back period. Finally, 664,494 patients who received eye drops for at least 3 months if an event (such as glaucoma surgery) occurred within the first year or 6 months otherwise were included in the study and considered for analysis.

Fig. 2.

Flow chart of study participants

For these 664,494 patients, a total of 38,127,364 1-month treatments were delivered during the study period, with a median per patient of 46.0 (IQR 25.0–78.0) treatments. Between 10.1% and 11.9% of patients were included each year from 2009 to 2017. The median follow-up was 4.8 (IQR 2.8–7.1) years.

Patient Characteristics and Treatment Groups

The main patient characteristics are reported in Table 1. Median age at inclusion was 66.0 (IQR 57.0–75.0) years, and 57.0% were women. At inclusion, most patients received a preserved treatment (N = 403,682, 60.8%) and a monotherapy (N = 578,110, 87.0%), and most received prostaglandin analogs (N = 423,831, 63.8%), followed by beta-blockers (N = 223,600, 33.6%).

Table 1.

Patients’ characteristics of the study population (N = 664,494) according to preservative exposure group

Characteristic	Overall N = 664,494	Preservative exposure group during follow-up				P-valuea
		0% preserved N = 121,600	Low mixed N = 139,305	High mixed N = 184,439	100% preserved N = 219,150
Age at inclusion (years)
Mean (SD)	65.5 (12.9)	64.1 (12.6)	64.9 (12.4)	65.3 (12.7)	66.7 (13.3)	< 0.0001b
Median (IQR)	66.0 (57.0–75.0)	65.0 (56.0–73.0)	65.0 (57.0–74.0)	66.0 (57.0–75.0)	67.0 (58.0–77.0)
By class, n (%)						< 0.0001c
< 45	38,422 (5.8)	7941 (6.5)	8091 (5.8)	10,679 (5.8)	11,711 (5.3)
[45; 65[	264,563 (39.8)	51,867 (42.7)	56,799 (40.8)	73,808 (40.0)	82,089 (37.5)
[65; 75[	189,923 (28.6)	36,259 (29.8)	42,618 (30.6)	52,969 (28.7)	58,077 (26.5)
≥ 75	171,586 (25.8)	25,533 (21.0)	31,797 (22.8)	46,983 (25.5)	67,273 (30.7)
Sex						< 0.0001c
Male, n (%)	286,034 (43.0)	51,277 (42.2)	56,771 (40.8)	77,478 (42.0)	100,508 (45.9)
Female, n (%)	378,460 (57.0)	70,323 (57.8)	82,534 (59.2)	106,961 (58.0)	118,642 (54.1)
Comorbidities
High blood pressure, n (%)	308,603 (46.4)	51,061 (42.0)	59,851 (43.0)	85,889 (46.6)	111,802 (51.0)	< 0.0001c
Cardio-neurovascular disease, n (%)	80,740 (12.2)	12,152 (10.0)	14,329 (10.3)	21,606 (11.7)	32,653 (14.9)	< 0.0001c
Type of therapy usedd						< 0.0001e
Monotherapy, n (%)	555,750 (83.6)	118,975 (97.8)	118,205 (84.9)	127,441 (69.1)	191,129 (87.2)
Bitherapy, n (%)	100,327 (15.1)	NC	NC	51,517 (27.9)	25,802 (11.8)
Tritherapy or more, n (%)	8417 (1.3)	N ≤ 10f	NC	5481 (3.0)	2219 (1.0)

IQR interquartile range, NC not communicated (because of GDPR rules; see below^f), SD standard deviation

^aComparison of patient characteristics according to preservative exposure groups

^bANOVA (analysis of variance) test

^cChi-squared test

^dUsed as a proxy for the severity of glaucoma. This was estimated based on two steps: (1) therapy with the longest mean duration each year and (2) maximum of the number of ocular hypotensive drugs (molecules) received during the study follow-up

^eFisher’s exact test

^fResults with ≤ 10 patients are not presented to comply with French individual data protection policy (GDPR rules)

During the study follow-up, 18.3% of patients received only PF eye drops (0% preserved group), 48.7% both preserved and PF eye drops, and 33.0% exclusively preserved eye drops (100% preserved group). Of all patients aged ≥ 75 years and male, with high blood pressure and cardio-neurovascular disease at inclusion, most belonged to the 100% preserved group. Regarding the type of therapy (used as a proxy for disease severity), the percentage of patients that received a bi- or tritherapy (or more) during the study follow-up was higher in the high-mixed group (30.9%) than in the other groups (from 2.2% in the 0% preserved group to 15.1% in the low-mixed group; Table 1).

The use of PF formulations increased during the study follow-up period, particularly for prostaglandin analogs (few patients [< 0.1%] received PF prostaglandin analog eye drops during the 2009–2013 period vs 62.9% during the 2014–2017 period; data not shown).

Occurrence of Glaucoma Surgery

Overall, 8467 patients (1.3%) underwent glaucoma surgery during the follow-up period (Table 2). For these patients, the median time between inclusion and glaucoma surgery was 2.4 (IQR 1.1–4.4) years. The percentage of patients undergoing glaucoma surgery was 0.2% in the 0% preserved group, 0.7% in the low-mixed group, 2.3% in the high-mixed group, and 1.4% in the 100% preserved group. Most patients undergoing glaucoma surgery during follow-up were on bi- (50.6%) or tritherapy or more (11.4%).

Table 2.

Glaucoma surgery occurrence in the study population (N = 664,494) over the follow-up period (2009–2018) according to preservative exposure group

	Overall N = 664,494	Preservative exposure group				P-valuea
		0% preserved N = 121,600	Low mixed N = 139,305	High mixed N = 184,439	100% preserved N = 219,150
Surgery occurrence, n (%)	8467 (1.3)	196 (0.2)	919 (0.7)	4261 (2.3)	3091 (1.4)
Surgery occurrence by type of therapy used
Monotherapy	n = 555,750	n = 118,975	n = 118,205	n = 127,441	n = 191,129
Patients with surgery, n (%)	3217 (0.6)	158 (0.1)	443 (0.4)	1256 (1.0)	1360 (0.7)	< 0.0001b
Bitherapy	n = 100,327	NC	NC	n = 51,517	n = 25,802
Patients with surgery, n (%)	4288 (4.3)	38 (1.4)	412 (2.0)	2423 (4.7)	1415 (5.5)	< 0.0001b
Tritherapy or more	n = 8417	N ≤ 10d	NC	n = 5481	n = 2219
Patients with surgery, n (%)	962 (11.4)	0 (0)	64 (8.9)	582 (10.6)	316 (14.2)	< 0.0001c

NC not communicated (because of GDPR rules; see below^d)

^aComparison of the occurrence of surgery (yes/no) according to preservative exposure groups

^bChi-squared test

^cFisher’s exact test

^dResults with ≤ 10 patients are not presented to comply with French individual data protection policy (GDPR rules)

The percentage of glaucoma surgery occurrence increased with preservative exposure (Table 2): from 0.1% in the 0% preserved group to 0.7% in the 100% preserved group for patients on monotherapy, from 1.4% in the 0% preserved group to 5.5% in the 100% preserved group for patients on bitherapy, and from 0% in the 0% preserved group to 14.2% in the 100% preserved group for patients on tritherapy or more. Univariate outcomes were confirmed by multivariate Cox model with the risk of glaucoma surgery (HR) at 10 years post-inclusion increasing with the level of exposure to preservatives from 2.12 (95% CI 1.81–2.48) in the low-mixed group to 4.57 (95% CI 3.94–5.29) in the 100% preserved group compared with the 0% preserved exposure group (Fig. 3).

Fig. 3.

Forest plot of multivariate Cox regression analysis of glaucoma surgery occurrence at 10 years post-inclusion. Models performed in patients with ≤ 10 years of follow-up and adjusted for all variables are presented in Table S4 (see the Electronic Supplementary Material for details), i.e., age, sex, history of high blood pressure and cardio-neurovascular disease, the number of ocular hypotensive drugs (molecules) dispensed simultaneously (used as a proxy for disease severity), and year of patient inclusion. The 0% preserved group was set as reference. 95% CI 95% confidence interval

Discussion

The present study was performed at a nationwide level, with data analyzed from 664,494 patients requiring glaucoma eye drops. An association was observed between the level of exposure to preserved glaucoma eye drops and the occurrence of glaucoma surgery. These results confirm our previous findings performed on a smaller sample of this database (12,454 patients) [20]. Of note, the relative risk of surgery was 3.94 (95% CI 1.54–10.05) for patients who received both PF and preserved glaucoma eye drops (“mixed” exposure group) and 7.97 (95% CI 3.07–20.67) for patients under preserved treatments only (100% exposure group) compared with patients on PF formulations only. The present analysis is nationwide, based on a more recent study period and refined preservative exposure groups (with the addition of low-mixed/high-mixed categories). It allows for a better estimate of the risk of glaucoma surgery based on preservative exposure groups and therefore further highlights the association between the level of exposure to preserved glaucoma eye drops and the risk of glaucoma surgery.

Patient Characteristics and Comparison with Other Studies

Baseline characteristics of subjects included in the present study were consistent with existing epidemiological data on patients with primary OAG (POAG) [21]. High blood pressure and older age (≥ 75 years) were associated in the current study with a lower risk of glaucoma surgery (HR 0.83 and 0.74, respectively; P < 0.0001 each), whereas male sex was associated with a higher risk (HR 1.17; P < 0.0001) (Table S4). Similar findings were reported in another nationwide cohort study conducted in Japan, where having hypertension, age ≥ 80, and female sex were associated with a decreased risk of trabeculectomy surgery [22]. Although systemic hypertension is postulated to be a risk factor for developing POAG, this association remains unclear, with various epidemiological studies reporting conflicting results [23]. A recent meta-analysis of 11 population-based cohort studies confirmed that the use of systemic β-blockers was associated with a significantly lower IOP [24], suggesting that patients with glaucoma on such antihypertensive treatments benefit from better IOP control. Regarding gender, conflicting results were found in the literature. While our findings were consistent with the prior Japanese study [22], they contrast with those of a large Korean cohort study in adults with newly diagnosed OAG [25], where women had a significantly increased risk of requiring surgery within 5 years of diagnosis (odds ratio 1.46; 95% CI 1.10–1.94). Several potential explanations have been proposed for this observation, including sex-based differences in the acceptance of surgery or an increased risk of glaucoma progression in women, particularly in postmenopausal women. However, the reasons for discrepancy between studies remain unclear.

Regarding exposure to preserved treatments, results were consistent with previous reports, with a large majority of patients (> 80%) exposed to preserved glaucoma eye drops treatment during the study [26, 27]. The relatively low proportion of patients in the 0% preserved group reflects the limited availability of PF topical formulations during the study period, with a much larger choice of preserved medications. More recently, PF formulations have replaced preserved eye drops as the most commonly prescribed hypotensive eye drop medication, a trend observed across Europe [27, 28].

The median delay between the first prescription of glaucoma eye drops and glaucoma surgery was 2.4 years (2.3 years in the EGB trial) [20]. Limited epidemiological data are currently available on the duration of treatment before referral for first glaucoma surgery. In the ReF-GS study analyzing this delay in Europe, the duration of treatment prior to referral was reported to greatly vary between countries, with a mean of 7.7 years in France [26]. However, this finding was derived from the collection of data from only 12 questionnaires in France, at the time of patients’ referral for glaucoma surgery. The shorter delay found in our analysis may be due to the study design (open cohort with incident cases), leading to an under-representation of individuals with long enough follow-up since glaucoma diagnosis.

Association Between Glaucoma Surgery and Preservative Exposure

A clear association is shown in this study between the level of exposure to preservatives and occurrence of glaucoma surgery at 10 years post-inclusion. Considering that stage of the disease is most likely related to more medications (and therefore to more preserved medications), and glaucoma eyes with more advanced disease are more prone to undergo surgery, patients were subdivided by type of therapy used (i.e., monotherapy, bitherapy, tritherapy, or more), which was considered an indicator of disease severity. Descriptive statistical analyses based on each “group of severity” confirmed that the percentage of patients undergoing glaucoma surgery increased with the level of exposure to preservatives.

This association between type of eye drops used (preserved vs PF) and occurrence of glaucoma surgery is supported by a substantial body of evidence in the literature regarding the detrimental effects of preservatives, especially BAK, on the eye. Historically, most topical glaucoma eye drops have been preserved with BAK, an effective antimicrobial with broad activity on bacteria and fungi. However, it is now recognized that the inherent detergent properties of BAK also pose significant risks, causing ocular surface damage, particularly following repeated long-term exposure and subsequent accumulation [14, 29–31]. Toxic effects include conjunctival hyperemia, squamous metaplasia and apoptosis, decreased tear production, tear film instability, superficial punctate keratitis, and disruption of the corneal epithelial barrier. BAK is also associated with a loss of goblet cells, which are particularly susceptible to toxic insults, leading to reduced mucin production and further tear film instability [32]. Together, these effects may result in multiple symptoms of ocular discomfort due to dry eye and inflammatory irritation, adversely impacting patients’ quality of life [33, 34]. This is particularly important in a chronic disease like glaucoma, where it is likely that once diagnosed, the patients will be on ocular hypotensive medication for the rest of their lives. The presence of OSD in patients with glaucoma may translate into nonadherence to medications, which is one of the major problems in clinical practice, in controlling IOP, and therefore in treating glaucoma [11, 35]. BAK-induced pro-inflammatory responses could also spread into deeper ocular tissues such as the anterior chamber and trabecular meshwork, adding to the trabecular dysfunction observed in glaucoma [5, 36, 37]. Over time, accumulation of BAK in the trabecular meshwork triggers chronic inflammation (releasing inflammatory cytokines, immune cell infiltration, and trabecular meshwork cell apoptosis) and contributes to the apparent progressive loss of efficacy of topical IOP-lowering drugs [5, 37–39].

Although some authors have reported that preservatives, and notably BAK, are necessary to provide penetration of drugs into the eye, a growing body of evidence shows that developed PF formulations improve the ocular tolerability, while maintaining the IOP-lowering efficacy, compared to preserved formulations [40]. Dubrulle and coworkers showed that treating ocular surface inflammation, including switching to PF glaucoma eye drops, may improve IOP control in patients with medically uncontrolled glaucoma associated with OSD [10]. Hence, switching from preserved to PF formulations could improve the ocular surface, preserve deeper ocular tissues such as the trabecular meshwork, and ultimately improve compliance and IOP control, therefore decreasing or delaying the need for surgery. Our results are coherent with the recent recommendations of a group of experts, which highlight that eliminating preserved medications (especially BAK-preserved eye drops) is effective in limiting ocular surface inflammation, restoring ocular health, and improving the management of patients with glaucoma [41].

Strengths and Limitations

To our knowledge, the present study is the largest nationwide cohort study, including > 660,000 patients initiating glaucoma treatment, with a 9-year inclusion period and up to 10 years of follow-up. Therefore, the main strengths of our study were the large number of patients included and the duration of follow-up. As our data were based on daily clinical practices and are representative of the medico-administrative database, which includes almost 99% of the French population, our results do not need any extrapolation for our country. Also, the assessment of exposure was based on glaucoma treatment delivery and not on prescriptions, possibly limiting the inclusion of nonadherent patients.

However, using the national healthcare database has some limitations. First, the database does not contain some specific clinical information regarding, for instance, disease stage and rate of progression, clinical findings at diagnosis, or the eye laterality concerned by the medical treatment or surgery. As more severe glaucoma requires multiple ocular hypotensive drugs and thus is more at risk of preservative exposure, a proxy of the disease severity, defined as the number of drugs (molecules) used simultaneously (monotherapy, bitherapy, tritherapy, or more), was used to bypass this issue. Notably, although we cannot entirely rule out the presence of collinearity between the proxy used to define disease severity and the preservative exposure groups, this variable was introduced into the model for adjustment because of its clinical importance. The risk of bias by omitting this variable (not adjusting for severity) may be more problematic than the risk of collinearity [42]. In addition, since the present study was conducted using the SNDS database with a large sample of included patients, the impact of multicollinearity may be limited by the statistical power of the study [43]. Also, indications for surgery were unknown and can differ between patients (e.g., target IOP not met, intolerance to medications, compliance issues, surgeon expertise, and prescription habits). Second, the present study investigated the association between the level of exposure to preservatives and glaucoma surgery occurrence based on the rate rather than the actual duration of exposure. Considering that the effects of preservatives are mainly associated with their chronic usage, this may have impacted the present results. However, this was addressed by introducing the majority treatment group over the follow-up into the Cox model and adjusting this model according to the year of patient inclusion in the study (proxy of treatment duration). Finally, since we focused our analysis mainly on POAG, the most frequent glaucoma etiology, our results do not allow reaching conclusions regarding other types of glaucoma. Of note, ICD-10 code H40.9 was considered in the present study because in France the practice coding in the SNDS medico-administrative database may use this code in association with surgical procedures (automated coding) instead of H40.1 for coding POAG surgery.

Conclusion

The risk of glaucoma surgery was associated with the level of exposure to preserved glaucoma eye drops in this nationwide cohort study. Large-scale, prospective, randomized studies would be beneficial to investigate whether PF eye drops can decrease glaucoma surgery occurrence. Switching from preserved to PF formulations may improve the ocular surface, preserve the trabecular meshwork, and improve compliance, therefore decreasing or delaying the need for surgery.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

All named authors meet the International Committee of Medical Journal (ICMJE) criteria for authorship for this article, taking responsibility for the integrity of the work as a whole, and have given their approval for this version to be published. Chloé Chamard, Alain Bron, Barbara Roux, Cédric Collin, Max Villain, Eloi Debourdeau, Christophe Baudouin and Vincent Daien contributed to the study conception and design, and to the interpretation of the results. Barbara Roux and Cédric Collin performed the data collection and statistical analysis. Chloé Chamard, Vincent Daien, Barbara Roux and Cédric Collin prepared the first draft of the manuscript. All authors reviewed the manuscript for important intellectual content and approved the final manuscript.

Funding

This work was funded by Laboratoires THEA, including editorial assistance and payment of the journal’s Rapid Service Fee. The sponsor had no role in the design and conduct of the study.

Data Availability

Barbara Roux and Cédric Collin had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. This study used anonymized patient data that are not publicly available due to confidentiality considerations.

Declarations

Conflict of Interest

Chloé Chamard, Alain Bron, Barbara Roux, Cédric Collin, Max Villain, Eloi Debourdeau, Christophe Baudouin and Vincent Daien have no competing interests to disclose.

Ethical Approval

This Study was submitted to the French Health Data Hub (HDH) in June 2019 and obtained a favorable opinion from CESREES (Ethical and Scientific Committee for Research, Studies and Evaluation in Public Health) in July 2019. The study was approved by the CNIL (French Data protection Authority) on August 2, 2019 (DR-2019-217/No 919284). Permission to access and use data from the database utilized in this study was obtained from the CNIL. As this study was a retrospective analysis of data from the SNDS, no informed consent was required.