Patient Experience Survey of Anti-Vascular Endothelial Growth Factor Treatment for Neovascular Age-Related Macular Degeneration and Diabetic Macular Edema

Diabetic macular edema (DME) has a debilitating effect on visual acuity (VA) and is a leading cause of vision loss among the working population [1‒3]. Given the increasing prevalence of type 2 diabetes, the incidence of DME is projected to rise accordingly [4]. Neovascular age-related macular degeneration (nAMD) accounts for most cases of AMD‐related severe vision loss, and its incidence is projected to increase with the aging population [5‒7].

Besides potential deterioration in VA and blindness if DME and nAMD are left untreated [1, 2, 5], patients also experience reduced functional status and quality of life (QoL), requiring ongoing treatment with regular follow-up and monitoring to control disease activity [8‒10]. Intravitreal anti-vascular endothelial growth factor (anti-VEGF) therapies (e.g., ranibizumab, aflibercept) have been successful in improving and preserving vision and QoL for patients [11, 12]. However, real-world treatment outcomes for patients with DME or nAMD repeatedly lag behind those from clinical trials [11, 12]. Insufficient management, attributed to the treatment burden on patients, caregivers, and healthcare systems, has been reported to lead to patients receiving fewer anti-VEGF injections and less frequent monitoring, and being unable to follow their treatment plans [11, 12]. Recent surveys highlight the importance of continuous medical and mental support, targeting risk factors for nonadherence (e.g., long treatment period, low VA, and patients’ general health), in helping patients follow their treatment plan [13, 14]. Despite the potential benefits of anti-VEGF treatments, innovations are needed to improve patients’ ability and/or willingness to follow their treatment regimen and improve real-world outcomes. The Patient Experience and Preference (PEP) study in DME and nAMD aimed to understand current treatment experiences with anti-VEGF injections and barriers to adherence from the patient perspective.

This was a multicenter, noninterventional, cross-sectional, observational study of patient treatment experience in nAMD and DME with administration of a patient survey and retrospective chart review. A survey for each condition was used to collect quantitative data from the patient perspective.

Patients were recruited using standardized materials (see online suppl. material 1; for all online suppl. material, see https://doi.org/10.1159/000538975, for description) via clinical sites in the USA, the UK, Canada, France, Italy, and Spain from January to December 2021. The population recruited was a convenience sample, and reasons for patients declining to participate were not collected. Eligible patients were ≥18 years of age, with a physician-confirmed diagnosis of either DME or nAMD, had initiated anti-VEGF injection treatment ≥6 months before March 2020 (to capture treatment experience before coronavirus disease [COVID-19] outbreak), had their most recent anti-VEGF injection within ≤6 months of screening, and had ≥18 months of medical record data available. Patients were excluded if they had a serious psychiatric disorder or other condition that impaired cognitive function to the extent of providing consent or completing questionnaires, were enrolled in clinical trials for any condition, current or historical retinal vein occlusion or myopic choroidal neovascularization, were treated with anti-VEGF therapy in a clinical trial for retinal diseases, current or historical DME or diabetic retinopathy treated with intravitreal anti-VEGF in patients with nAMD, or any current history of nAMD in patients with DME.

The study complied with local confidentiality regulations and legislation (EU General Data Protection Regulation, US Health Insurance Portability and Accountability Act, Canadian Personal Information Protection and Electronic Documents Act). Institutional Review Board (IRB)/Ethics Committee approval was obtained. Responses were de-identified and participants assigned unique identification numbers kept securely and separate from the surveys.

Surveys were developed using a targeted literature review and qualitative research involving interviews with DME and nAMD patients, clinicians, and caregivers, as described in-depth elsewhere [15]. Patient advocacy organizations invited members living with nAMD or DME to review the draft surveys to ensure relevance and clarity.

The surveys included clinical history and treatment experience; a validated questionnaire (Macular Disease Treatment Satisfaction Questionnaire [MacTSQ] for nAMD patients; Retinopathy Treatment Satisfaction Questionnaire [RetTSQ] for DME patients) to measure treatment satisfaction; the validated 25-item National Eye Institute Visual Functioning Questionnaire (NEI-VFQ-25) to measure the patient’s vision-related functioning; and sociodemographic characteristics. Copies of the full surveys are available on request.

To accommodate potential challenges due to visual impairment, surveys were available via paper, online, or telephone interview formats. Patients completed the survey at home using their preferred mode of administration. Methods for managing nonresponse and for minimizing data errors are described in online supplementary Material 1. Medical records were reviewed by trained staff and information on clinical history (e.g., treatments, visits, VA, and comorbidities) was entered into electronic case report forms (eCRF).

The MacTSQ and RetTSQ are 14-item validated and reliable questionnaires assessing treatment satisfaction among nAMD and DME patients, respectively (MacTSQ score range 0–72; RetTSQ score range 0–78). Higher scores denote greater satisfaction [16, 17]. The NEI-VFQ-25 is a validated and reliable 25-item tool measuring the influence of visual impairment on QoL across 11 subscales (plus a stand-alone question on general health) [18‒20]. Higher scores (range 0–100) indicate better visual functioning. Linguistically validated non-English language versions of the surveys, and all included measures, were available and used as appropriate.

Patients were defined as nonadherent if they met either of the following criteria within the previous 12 months: failure to attend or cancellation of ≥2 scheduled monitoring or treatment visits without returning for a subsequent visit within 4 weeks, or rescheduling ≥2 monitoring or treatment visits to >4 weeks after their intended date.

The definition of nonadherence was informed by discussions with payers, representatives from patient advocacy organizations, qualitative interviews with physicians, literature review, and discussion with therapy area experts. The relatively low stringency of this definition was to account for potential impact of the COVID-19 pandemic.

The VA collected in the eCRF was recorded as measured per routine clinical practice (e.g., Snellen). VA was converted to approximate Early Treatment Diabetic Retinopathy Study (approxETDRS) letters for analysis using the method from Gregori and colleagues [21].

A sample size of 1,600 patients (100 per country per condition in the UK, France, Italy, and Spain, 200 per country per condition in the USA and Canada) was prespecified (online suppl. Material 2). Descriptive statistics were used to summarize survey data. Analyses were conducted separately by condition. The association between treatment satisfaction and elements of treatment burden was explored using descriptive statistics and p values generated using analysis of varianc for between-group comparisons.

Exploratory univariable linear or logistic regression models (as appropriate) regressed VA or NEI-VFQ-25 composite scores on variables of treatment adherence, the number of missed visits, or number of anti-VEGF injections received. VA was either modeled continuously or as a binary variable (≤69 or ≥70 approxETDRS letters). The cut point of 70 approxETDRS letters corresponds to a 20/40 Snellen equivalent, which was used as a measure of driving vision.

Handling of missing data is described in online supplementary Material 2. p values were not adjusted for multiple comparisons due to the exploratory nature of the analysis. All programming was conducted using SAS v9.4 (SAS Institute, Inc., Cary, NC, USA).

Of 1,003 patients with eCRF data (358 DME patients; 645 nAMD), 574 (57.2%: 183 DME patients; 391 nAMD) completed the survey (details on mode of administration in online suppl. Table 1). Characteristics of patients with eCRF data (online suppl. Tables 2–5) and with complete survey data (Table 1) were similar. DME patients tended to be younger (mean = 65.9, standard deviation [SD] = 11.5 years) than nAMD patients (mean = 80.1, SD = 7.7 years). Most were female in nAMD (64.5%) and male in DME (62.8%); 63.2% of nAMD and 68.9% of DME were white (Table 1). More patients with nAMD (74.7%) than DME (49.7%) were retired.

More DME (n = 146, 79.8%) than nAMD patients (n = 219, 56.0%) had bilateral disease. At index, mean VA was slightly worse for nAMD (56.2 approxETDRS letters, SD = 22.9) than DME patients (64.3, SD = 18.1). At enrollment, DME and nAMD patients had been treated with anti-VEGF for means of 4.9 and 5.9 years, respectively. For most patients (69.4% DME; 85.2% nAMD), anti-VEGF treatment was the first treatment for their condition. Within 12 months, nAMD patients attended more visits where anti-VEGF injections were administered (nAMD: mean = 7.74, SD = 3.63; DME: mean = 6.54, SD = 3.40). Less than a quarter of patients switched anti-VEGF therapies during the study period (12.0% DME; 14.8% nAMD). One-third of DME (33.9%) and one-quarter of nAMD patients (25.8%) were on a fixed treatment schedule, with 73.6% (n = 120 of 163) receiving injections every ≤8 weeks. About one-third of nAMD (38.1%) and one-quarter of DME (26.8%) patients were on a treat-and-extend (T&E) regimen, with 69.7% (n = 138 of 198) receiving injections every ≤8 weeks. A smaller proportion of patients were treated pro re nata (PRN) (10.4% DME; 13.3% nAMD) (Table 2).

Patient-reported outcome scores showed moderately high vision-related functioning with no substantive differences between DME (NEI-VFQ-25 composite score: mean = 77.8, SD = 24.5) and nAMD (mean = 73.4, SD = 24.7). DME patients scored slightly higher (i.e., better visual functioning) across multiple domains. Patients were generally satisfied with their current treatment (MacTSQ, mean total score = 59.2, SD = 9.6; RetTSQ, mean total score = 61.3, SD = 12.8).

Treatment satisfaction scores in DME patients were generally lower among those who had longer typical appointment durations (p = 0.027) or recovery time after injections (p = 0.013), reported a negative impact on employment (p < 0.001), and expected their vision to remain the same (p = 0.035). For nAMD patients, scores were generally lower among those who had longer clinic wait times (p < 0.001) or recovery time after injections (p < 0.001), experienced a higher level/extent of financial impact (p < 0.001), had a caregiver (p = 0.018), and reported a negative impact on time with family/friends (p < 0.001), traveling/other leisure activities (p < 0.001), and activities of daily living (ADL; p < 0.001) (online suppl. Table 6).

Per protocol definition, 16.5% (59 of 358) of DME and 7.0% (45 of 645) of nAMD patients in the eCRF population and 16.4% (30 of 183) of DME and 5.9% (23 of 391) of nAMD patients in the survey population were nonadherent. Based on available patient self-report, 14.5% (23 of 159) of DME and 4.2% (15 of 360) of nAMD patients missed ≥1 injection visit in the past 12 months.

Despite these relatively high adherence levels, patients still reported barriers and burden linked to the management of their eye conditions (online suppl. Tables 7, 8). Indeed, 66.1% (121/183) of DME patients and 49.2% (192/390) of nAMD patients reported ≥1 barrier preventing them from receiving treatment or attending visits. Among barriers reported at least once, the most frequent were related to treatment (n = 201, 35.0%; e.g., pain/discomfort during/after anti-VEGF injection), clinic (n = 187, 32.6%; e.g., no-one to accompany them to appointments) and COVID-19 factors (n = 121, 21.1%; e.g., clinic canceled/rescheduled appointments) (Fig. 1).

Around a quarter of patients (n = 139, 24.2%) needed ≥1 day to recover from their injections. Overall, 41.3% (n = 237) reported having ≥1 h of travel time to their appointment and 29.4% (n = 169) had ≥1 h of waiting time. Approximately equal proportions of patients experienced a combined (injection and examination) appointment duration of <1 h (29.4%; n = 169) and ≥1 h (32.9%; n = 198).

Nearly half of patients (n = 258, 44.9%) reported some level of impairment in their ADL due to treatment. Among working patients (n = 42 DME; n = 24 nAMD), 62.1% experienced some level of absenteeism at work.

Tables 3 and 4 summarize findings of exploratory regression analyses. In unadjusted univariable analyses, treatment adherence was related to a higher most recent VA (β = 8.98 letters; CI, 1.34–16.62) and lower odds of VA below driving vision (≤69 letters vs. ≥ 70 letters) (OR = 0.50; CI, 0.25–1.00) among nAMD patients. Similar findings of treatment adherence related to lower odds of VA below driving vision were observed in the DME population in unadjusted analyses (OR = 0.51; CI, 0.29–0.91).

Number of anti-VEGF injections was associated with a higher most recent VA (DME: β = 0.77 letters; CI, 0.24–1.30; nAMD: β = 1.60 letters; CI, 1.14–2.05), lower odds of VA below driving vision (DME: OR = 0.94; CI, 0.89–0.99; nAMD: OR = 0.91; CI, 0.87–0.94), in unadjusted univariable analyses. Finally, nAMD patients who received more injections had better self-reported NEI-VFQ-25 composite (β = 1.33 letters; CI, 0.76–1.89) scores in unadjusted analyses.

This real-world multinational study sheds light on the experience of DME and nAMD patients with anti-VEGF treatment. The results show that patients devote substantial time to their treatment management plans. Nearly half of patients reported impacts of treatment on their ADL, and over half reported barriers to treatment related to treatment, clinic, or COVID-19. Treatment satisfaction scores were worse with higher time-related impacts of treatment (both nAMD and DME), higher impacts on finances and daily life (nAMD), and negative impacts on employment and lower expectations for treatment effectiveness (DME). Treatment adherence was related to higher most recent VA (nAMD) and lower odds of VA below driving vision (nAMD and DME). Number of injections was associated with higher most recent VA and lower odds of VA below driving vision in both indications. Number of missed visits was associated with lower most recent VA and lower odds of VA below driving vision in nAMD patients. nAMD patients who received more injections had better self-reported NEI-VFQ-25 composite scores in unadjusted analyses.

Despite the relatively high VA observed (which was expected given the high adherence), patient-reported treatment burden and barriers indicated room for improvement in treatment management. Patients with both conditions reported burdens related to impairment in ADL, time traveling to appointments, clinic wait time, appointment durations, recovery from injections, finances, and other personal/work-related factors. For instance, nearly half reported impairment to their ADL due to treatment, and over half of working patients reported treatment impacts on absenteeism. Moreover, patients reporting a higher treatment burden (e.g., longer wait times) had worse treatment satisfaction scores.

Patients reported various barriers that may hinder their ability to continue managing anti-VEGF therapy, including treatment- and clinic-related factors, such as the high frequency of treatment or eye examinations and the amount of time the visits take, including travel and waiting times. In addition, there were COVID-19-related barriers, for example, having had to miss or reschedule appointments due to having COVID-19 or being in a high-risk population. This suggests that reducing treatment/visit frequency and time spent waiting at and attending visits may improve patient experience overall and reduce treatment burden.

Our results provide evidence that missed visits and treatment nonadherence were associated with worse visual outcomes. Missed visits were associated with lower VA, and treatment nonadherence was associated with lower VA and higher odds of VA below driving vision (≤69 letters) among nAMD patients. The smaller sample of DME patients (compared with the nAMD population) may have had weaker statistical power for detecting such differences and correlations. However, in both nAMD and DME, treatment nonadherence was associated with higher odds of VA below driving vision, suggesting nonadherence could impact patients’ self-sufficiency. Relatedly, more anti-VEGF injections were associated with better visual outcomes, consistent with previous studies [22, 23].

Our results are in line with findings from prior studies that reported an association between nonadherence and worse visual outcomes. In a secondary analysis of the Comparison of Age-Related Macular Degeneration Treatment Trial (CATT) it was shown that nAMD patients who did not adhere to clinic visit schedules had worse VA [24]. In a retrospective, single-center database study of DME and AMD patients from Germany, Weiss and colleagues found that a lower proportion of patients with DME were on schedule compared with those with nAMD, and treatment break-off was associated with worse outcomes in patients with DME [25]. Another retrospective, observational single-center study from Germany also reported lower adherence amongst patients with DME compared with those with nAMD and that nonadherence was negatively associated with vision gain in nAMD and positively associated with clinically significant VA loss in DME [26].

Overall, the results of this study are in line with previous findings that suggest practical and clinical barriers may negatively impact a patients’ ability to follow their treatment plan [15, 27‒29]. Moreover, nAMD and DME treatments can be onerous and present significant burden to patients. It is thus important to target modifiable barriers and burden, including those related to treatment (e.g., frequency of treatment) and the clinic (e.g., long/frequent appointments) to improve patient experience with anti-VEGF therapy. More durable treatments that lead to reduced frequency of injections and a reduced number of clinic visits could help alleviate practical and clinical barriers to treatment. Fewer injections and visits could also reduce the treatment burden on patients’ daily lives. Finally, our results suggest that nonadherence, missed visits, and fewer injections were associated with worse visual outcomes. Thus, treatment modalities that reduce treatment burden and barriers, such as those with longer durability, may improve adherence and treatment satisfaction, ultimately leading to improved visual outcomes.

The survey is part of a multinational study combining qualitative and quantitative approaches to gather the perspectives of patients with nAMD and DME and their caregivers. Information was collected from a relatively large sample of patients through the surveys and objective information from medical chart data extractions with source data verification. The survey was available in multiple formats to accommodate potential challenges due to visual impairment and optimize survey participation.

One such format was a telephone interview. Though this format permitted interviewers to ask questions directly to facilitate survey completion, this may have also introduced response bias, for example, if patients did not respond truthfully to the interviewer or if they felt pressured to complete the survey [30]. Furthermore, patient responses may have also been subject to recall and information bias.

Despite efforts to enrich the sample with nonadherent patients, these patients remained hard to find and enroll in the study. Additionally, due to recruitment limitations during the COVID-19 pandemic period, the planned sample sizes were not achieved. This limited the exploration of treatment barriers and burden in that specific group and precludes the comparison of nonadherence in our sample to other studies. However, the observed trend of better adherence in nAMD than DME aligns with previously published findings [25, 26].

Despite efforts to recruit a diverse sample, there was limited diversity in patient ethnicity/race, and selection bias could have occurred at sites. The multinational design of this study and choice of survey formats enabled findings that are primarily generalizable to patients across North America and Europe. However, it is important to note that patient needs and barriers may differ at a country level, and the pooled nature of the data presented here does not permit interpretation of these differences. In addition, the univariable analysis does not account for potential confounding factors, such as baseline VA and duration of anti-VEGF treatment. Finally, additional research is needed to fully understand the exploratory findings of the study.

In summary, patients with nAMD and DME reported important impairments of ADL, work absenteeism, barriers, and burden related to anti-VEGF therapy. More durable treatment leading to reduced injection frequency (and thus clinic visits) could potentially reduce treatment barriers and the burden on patients’ daily lives. Reduced barriers and burden may potentially improve treatment adherence, patient satisfaction, and visual outcomes.

We thank the patients who took part in the survey. We thank Jennifer Eriksson (ICON plc) for the development of the CRF design and Valentin Barbier, Ankit Pahwa, and Shravan Kumar Adepu (ICON plc) for their support in the quantitative analysis. We would also like to thank Andrea De Palma and Muna Tahir (ICON plc) for their support in medical writing and manuscript editing. Additional editorial support was provided by Christopher A. Lamb, PhD, of Envision Pharma Group. Finally, we would like to thank all participating sites and the associated principal investigators for their support to this study. A full list is available in online supplementary Table 9.

The study protocol and study materials were reviewed and approved by the local authorities and ethical review boards as follows: Canada: Advarra IRB (Aurora, ON, Canada; no reference number provided); France: Comité de Protection des Personnes Sud-Méditerranée III (reference 2021.02.07 bis_ 20.11.18.48450), CNIL Data Protection Authority (reference CNIL 921172_MR41928 [EudraCT 2020-A02937-32] Roche); Italy: Comitato Etico Centrale Fondazione G.B. Bietti IRCCS Lazio (reference N.112/21/FB), Comitato Etico delle Province di Chieti e Pescara (reference MR41928-ITA003), Comitato Etico Milano Area 2 (reference 742_2021), Comitato Etico Area Vasta Emilia Centro (reference 2028/2021); Spain: CEIm Grupo Hospitalario Quirónsalud-Catalunya (reference 2021/19-OFT-HUGC), Comitè Ètic de la Investigació, Hospital Universitari de Bellvitge (reference EPA027/20), Comité de Ética de la Investigación con Medicamentos (CEIm) Área de Salud Valladolid Oeste (reference 21-EO017), CElm de la Comunidad Foral de Navarra (reference 0165/0596 Roche Spain); the UK: Health Research Authority Research Ethics Committees (references EC Ref: 19/EM/0259 and IRAS ID: 267010); and the USA: Advarra IRB (Aurora, ON, Canada, reference MOD00824561). Written informed consent was obtained from participants prior to participating in the survey.

J.L. and H.B.L. were employees of ICON plc at the time the study was conducted. G.C.C., M.M., and B.G. were employees of Roche at the time the study was conducted. N.H.: consulting fee for AGTC, Allergan, Annexon, Apellis, Bayer, Cardinal, Clearside Biosciences, EyePoint Pharmaceuticals, Gemini, Genentech, Gyroscope, Nacuity, NGM, Notal Vision, Novartis, Ocuphire, Outlook Therapeutics, Regeneron, Laboratoires Théa, Stealth Biosciences; Speakers Bureau for Allergan, Apellis, Genentech, Regeneron, Spark; Contracted Research for Gemini, Genentech, Gyroscope, Notal Vision; Intellectual Property/Patent for Katalyst Surgical. She is a current employee of Hoffmann-La Roche. A.G.-A. is a consultant for AbbVie, Alcon, Bayer, Horus, Novartis, Roche, and Théa.

Roche/Genentech funded and participated in the survey design, implementation, and analysis of this study, which was conducted by ICON plc, and provided funding for editorial support.

Study design: G.C.C., M.M., B.G., H.B.L., and J.L.; data acquisition and quantitative analysis: H.L. and J.L.; data review and interpretation: G.C.C., M.M., B.G., H.B.L., J.L., N.H., A.G.-A., A.G.-L., T.P., F.V., P.J.K., and A.K.; critically reviewed the manuscript and approved the final version prior to its submission: G.C.C., M.M., B.G., H.B.L., J.L., N.H., A.G.-A., A.G.-L., T.P., F.V., P.J.K., and A.K.

For up-to-date details on Roche’s Global Policy on the sharing of clinical information and how to request access to related clinical study documents, see here: https://go.roche.com/data_sharing. Anonymized records for individual patients across more than one data source external to Roche cannot, and should not, be linked due to a potential increase in risk of patient re-identification. Requests for the data underlying this publication require a detailed, hypothesis-driven statistical analysis plan that is collaboratively developed by the requestor and company subject matter experts. Direct such requests to Roche for consideration.