Loss to Follow-up Among Patients With Neovascular Age-Related Macular Degeneration Who Received Intravitreal Anti–Vascular Endothelial Growth Factor Injections

This cohort study investigates rates of risk factors associated with loss to follow-up after receipt of intravitreal injections of anti–vascular endothelial growth factor (anti-VEGF) injections among patients with neovascular age-related macular degeneration.

Key Points

Question

What percentage of patients with neovascular age-related macular degeneration are lost to follow-up after receiving intravitreal anti–vascular endothelial growth factor injections?

Findings

This cohort study of data from 9007 patients with neovascular age-related macular degeneration reported that the percentage of patients lost to follow-up after an anti–vascular endothelial growth factor injection was greater than 20% when choosing at least 12 months as the interval with no subsequent follow-up visit. Older age, race, lower regional adjusted gross income, greater distance to clinic, and unilateral eye disease were risk factors associated with loss to follow-up.

Meaning

These results suggest that a sizable number of patients with neovascular age-related macular degeneration are lost to follow-up after anti–vascular endothelial growth factor injections and that several risk factors could help to identify patients who are at high risk of being lost to follow-up.

Abstract

Importance

Loss to follow-up (LTFU) after anti–vascular endothelial growth factor (anti-VEGF) injections increases the risk of vision loss among patients with neovascular age-related macular degeneration (nAMD).

Objective

To report rates of LTFU among patients with nAMD after anti-VEGF injections and to identify risk factors associated with LTFU in this population.

Design, Setting, and Participants

This retrospective cohort study of data from 9007 patients who received anti-VEGF injections for treatment of nAMD was performed at an urban, private retina practice with multiple locations from April 1, 2012, to January 12, 2016.

Main Outcomes and Measures

Rates of LTFU after anti-VEGF injections. Loss to follow-up was defined as receipt of 1 or more injections with no subsequent follow-up visit within 12 months.

Results

Among the 9007 patients (mean [SD] age, 81.2 [8.8] years; 5917 [65.7%] female; 7905 [87.8%] white), 2003 (22.2%) were LTFU. Odds of LTFU were greater among patients 81 to 85 years of age (odds ratio [OR], 1.58; 95% CI, 1.38-1.82; P < .001), 86 to 90 years of age (OR, 2.29; 95% CI, 2.00-2.62; P < .001), and more than 90 years of age (OR, 3.31; 95% CI, 2.83-3.86; P < .001) compared with patients 80 years of age and younger. Odds of LTFU among African American patients (OR, 1.47; 95% CI, 1.00-2.16; P = .05), Asian patients (OR, 2.63; 95% CI, 1.71-4.03; P < .001), patients of other race (OR, 3.07; 95% CI, 1.38-6.82; P = .006), and patients of unreported race (OR, 2.29; 95% CI, 1.96-2.68; P < .001) were greater than odds of LTFU among white patients. Odds of LTFU were greater among patients with regional adjusted gross income of $50 000 or less (OR, 1.52; 95% CI, 1.30-1.79; P < .001), $51 000 to $75 000 (OR, 1.35; 95% CI, 1.17-1.56; P < .001), and $76 000 to $100 000 (OR, 1.28; 95% CI, 1.08-1.50; P = .004) compared with patients with incomes greater than $100 000. Odds of LTFU for patients living 21 to 30 miles (OR, 1.33; 95% CI, 1.05-1.69; P = .02) and more than 30 miles (OR, 1.55; 95% CI, 1.28-1.88; P < .001) from clinic were greater compared with patients who lived 10 miles or less from the clinic. Odds of LTFU were greater among patients who received unilateral injections (OR, 1.44; 95% CI, 1.28-1.61; P < .001) than among patients who received bilateral injections.

Conclusions and Relevance

We found a high rate of LTFU after anti-VEGF injections among patients with nAMD and identified multiple risk factors associated with LTFU among this population. Although our results may not be generalizable, data on LTFU in a clinical practice setting are needed to understand the scope of the problem so that interventions may be designed to improve outcomes.

Introduction

Intravitreal injection (IVI) of anti–vascular endothelial growth factor (VEGF) agents has become the mainstay of therapy for patients with neovascular age-related macular degeneration (nAMD).^1,2,3,4 Although different protocols for treatment have been used, such as treat and extend, pro re nata, and monthly injections,⁵ follow-up visits after treatment are imperative to optimize visual outcomes.^6,7 One study has shown that patients with nAMD receive fewer injections annually than recommended, which raises questions about adherence to the recommended guidelines.⁸ This outcome is concerning given that studies have repeatedly shown the importance of frequent injections for the preservation of vision.^9,10 There have been 2 earlier studies involving European-based practices that have attempted to evaluate nonadherence and loss to follow-up (LTFU) among patients with nAMD receiving IVIs.^11,12 However, these studies used small and homogenous samples, which limits the generalizability of their findings. Therefore, our study aimed to determine LTFU rates among patients with nAMD who received IVIs from an urban, private retina practice with 13 operational sites; investigate potential risk factors associated with LTFU; and create a geospatial distribution of LTFU rates to identify regions associated with a high risk of LTFU.

Methods

Study Population

The study was conducted with approval from the Wills Eye Hospital Institutional Review Board, which waived the need for informed consent, and was compliant with the Health Insurance Portability and Accountability Act. All patients with nAMD who received an IVI with anti-VEGF from April 1, 2012, to January 12, 2016, were identified from the electronic medical record database of the Wills Eye Hospital outpatient department and the Mid Atlantic Retina clinics using International Classification of Diseases, Ninth Revision; International Statistical Classification of Diseases and Related Health Problems, Tenth Revision Clinical Modification; and Current Procedural Terminology billing codes. The data were deidentified. Mid Atlantic Retina is an urban, private retina practice with 13 operational sites in the tristate region of Pennsylvania, New Jersey, and Delaware. Intervals between each IVI and the immediate subsequent follow-up visit were measured for each patient. For patients with multiple injections, the interval of longest duration was selected to assess for LTFU. To account for reasons that might have influenced the frequency of follow-up visits, patients were excluded from the final analysis if they met any 1 of the following criteria: history of diabetic retinopathy, retinal vein occlusion, myopic degeneration, angioid streaks, idiopathic choroidal neovascularization, or central serous retinopathy; enrollment in a prospective clinical trial; first injection after January 12, 2016; or deceased.

Definition of LTFU

Loss to follow-up was defined as at least 1 interval exceeding 12 months for any patient. The duration threshold was chosen to account for the variation in physicians’ designation of the follow-up visit, provide adequate time for patients who might have missed their appointment for a variety of health or personal reasons to return, and compensate for patients who might travel to other states for a portion of the year. To allow a sufficient amount of time for all patients to experience at least 1 year of follow-up after their last injection, a period of observation was used.¹³

Patient Characteristics

Race was self-reported and identified using the retina practice identification sheet. Patients who did not identify their race/ethnicity were categorized as having unreported race/ethnicity. Patients who identified their race/ethnicity as Hispanic, Native American, or Pacific Islander were grouped into the other race/ethnicity category because they were a small cohort of 28 patients. Regional average adjusted gross income (AGI) was defined as the average AGI of the patient’s zip code area and was determined using the Internal Revenue Service’s AGI database.¹⁴ Patients who received injections in only 1 eye for treatment of nAMD were classified as having unilateral active disease, whereas patients who received injections in both eyes for treatment of nAMD were considered to have bilateral active disease.

Visual Acuity

Only a subgroup of the cohort had recorded visual acuities that could be used in the final analysis. This occurrence was secondary to a recent transition in storing patient information in an electronic medical record system. The majority of patient history and clinical data from before the electronic medical records system was adopted were placed in storage and rendered inaccessible. Data on uncorrected, corrected, and pinhole vision were obtained for each patient when available. The best available Snellen visual acuity from the 3 measurements was then used in the final analysis. Measurements were obtained at both the first and the final injections. For patients who received bilateral injections, the eye with the better visual acuity was used. The visual acuity was converted to the logMAR for analysis. Visual acuity measurements were also stratified by minimal (≥20/40), moderate (20/50 to 20/200), and severe (<20/200) vision loss. For patients who underwent 2 or more procedures, change in logMAR visual acuity was calculated by subtracting the logMAR visual acuity measured at the first procedure from the logMAR visual acuity measured at the final procedure.

Distance to Clinic and Geospatial Mapping

Patients’ residential addresses were translated into a coordinate format, and the spherical distance from patient household to the retina clinic where the patient was being seen was calculated using a Haversine formula.^15,16 Location was then plotted by latitude and longitude on a geographic layout using Excel Bing maps extension (Microsoft). Loss to follow-up was then translated into a heat signal that represented the aggregate mean LTFU rate within a given geographic region.

Statistical Analysis

Statistical analyses were performed using SAS, version 9.4 (SAS Institute Inc). The number and percentage of patients who were LTFU was determined. The LTFU rates were stratified by patient characteristics and assessed for disparities using a χ² test. Continuous variables were categorized on the basis of either distribution or clinical relevance. Potential risk factors for LTFU were then evaluated using a univariate logistic regression. Risk factors with P < .1 on univariate regression were used in a multivariate logistic regression model to adjust for potential confounders. Two-tailed P values were used in the analyses.

Results

Study Population and LTFU Rates

A total of 12 582 patients who received IVI for treatment of nAMD were initially identified, and 9007 patients were eligible for analysis (Figure 1). A total of 115 964 IVIs with anti-VEGF were given at a total of 136 527 visits. Median (interquartile range [IQR]) patient age and mean (SD) patient age were 82 years (IQR, 76-87 years) and 81.2 (8.8) years, respectively. Median (IQR) distance to clinic was 6 miles (4-12 miles). Median (IQR) regional average AGI was $69 000 ($53 000–$90 000). Baseline characteristics of patients who were and were not LTFU are summarized in Table 1. Among 9007 patients, 2003 (22.2%) were LTFU and 7004 (77.8%) had a follow-up visit within 12 months. Among the patients who were LTFU, 1854 (92.6%) had no follow-up visit and 149 (7.4%) had a follow-up visit after 12 months. There were 1980 patients (98.9%) with only 1 LTFU episode and 23 (1.1%) with more than 1 LTFU episodes, of whom 21 had no follow-up visit after an IVI. Patients who were not LTFU had a mean (SD) of 14.3 (10.8) injections and 17.1 (10.7) visits, whereas patients who were LTFU had a mean (SD) of 7.8 (6.8) injections and 8.5 (7.2) visits. Mean (SD) duration of follow-up for patients who were not LTFU was 2.4 (1.2) years, whereas the mean (SD) duration of follow-up for patients LTFU was 1.2 (1.1) years.

Figure 1. Flowchart of Patients With Neovascular Age-Related Macular Degeneration (nAMD) Included in the Final Analysis.

Table 1. Characteristics of Patients With Neovascular Age-Related Macular Degeneration Who Received Intravitreal Anti–Vascular Endothelial Growth Factor Injections by Follow-up Status.

Variable	Follow-up Status, No. (%)		P Valuea
	Not Lost to Follow-up (n = 7004)	Lost to Follow-up (n = 2003)
Age, y
≤80	3104 (44.3)	583 (29.1)	<.001
81-85	1713 (24.5)	458 (22.9)
86-90	1475 (21.1)	567 (28.3)
>90	712 (10.2)	395 (19.7)
Sex
Male	2411 (34.4)	679 (33.9)	.66
Female	4593 (65.6)	1324 (66.1)
Race/ethnicity
White	6299 (89.9)	1606 (80.2)	<.001
African American	104 (1.5)	39 (1.9)
Asian	67 (1.0)	34 (1.7)
Otherb	15 (0.2)	11 (0.5)
Unreported	519 (7.4)	313 (15.6)
Regional average AGI, $
>100 000	1610 (23.0)	362 (18.1)	<.001
76 000-100 000	1418 (20.2)	392 (19.6)
51 000-75 000	2587 (36.9)	765 (38.2)
≤50 000	1389 (19.8)	484 (24.2)
Distance from clinic, miles
≤10	4892 (69.8)	1355 (67.6)	.004
11-20	1354 (19.3)	374 (18.7)
21-30	307 (4.4)	103 (5.1)
>30	451 (6.4)	171 (8.5)
Eye involvement
Bilateral	2353 (33.6)	518 (25.9)	<.001
Unilateral	4651 (66.4)	1485 (74.1)

Abbreviation: AGI, adjusted gross income (rounded to the nearest $1000).

^aP values obtained by χ² test.

^bOther includes Hispanic, Native American, and Pacific Islander.

Risk Factors for LTFU

The rate of LTFU increased with increasing age, with rates of 15.8% among patients 80 years of age or younger, 21.1% among patients 81 to 85 years of age, 27.8% among patients 86 to 90 years of age, and 35.7% among patients greater than 90 years of age (P < .001, by χ² test). Rate of LTFU also differed by race/ethnicity, with rates ranging of 20.3% among white patients, 27.3% among African American patients, 33.7% among Asian patients, 42.3% among patients of other race, and 37.6% among patients of unreported race (P < .001, by χ² test). Rate of LTFU increased as regional average AGI decreased, with rates of 18.4%, 21.7%, 22.8%, and 25.8% for incomes greater than $100 000, $76 000 to $100 000, $51 000 to $75 000, and $50 000 or less, respectively (P < .001, by χ² test). Distance also had an influence on LTFU rates, with rates of 21.7% among patients living 10 miles or less from the clinic, 21.6% among those living 11 to 20 miles from the clinic, 25.1% among those living 21 to 30 miles from the clinic, and 27.5% among patients living more than 30 miles from the clinic (P = .004, by χ² test). Patients who received bilateral injections had an LTFU rate of 18.0%, whereas patients who received unilateral injections had an LTFU rate of 24.2% (P < .001, by χ² test). No differences in LTFU rates were noted between male patients (22.0%) and female patients (22.4%) (P = .66, by χ² test). Multivariate analysis showed age, race/ethnicity, regional average AGI, distance to clinic, and unilateral disease as independently associated with LTFU (Table 2).

Table 2. Univariate and Multivariate Logistic Regression Analyses of Potential Risk Factors and the Association With Loss to Follow-up Among Patients With Neovascular Age-Related Macular Degeneration Who Received Intravitreal Anti–Vascular Endothelial Growth Factor Injections.

Variable	Lost to Follow-up, No. (%)	Univariate Model		Multivariate Modela
		Odds Ratio (95% CI)	P Value	Odds Ratio (95% CI)	P Value
Age, y
≤80	583 (15.8)	1 [Reference]		1 [Reference]
81-85	458 (21.1)	1.42 (1.24-1.63)	<.001	1.58 (1.38-1.82)	<.001
86-90	567 (27.8)	2.05 (1.80-2.33)	<.001	2.29 (2.00-2.62)	<.001
>90	395 (35.7)	2.95 (2.54-3.44)	<.001	3.31 (2.83-3.86)	<.001
Sex
Male	679 (22.0)	1 [Reference]
Female	1324 (22.4)	1.02 (0.92-1.14)	.66	ND	ND
Race/ethnicity
White	1606 (20.3)	1 [Reference]		1 [Reference]
African American	39 (27.3)	1.47 (1.01-2.13)	.04	1.47 (1.00-2.16)	.05
Asian	34 (33.7)	1.99 (1.31-3.02)	.001	2.63 (1.71-4.03)	<.001
Otherb	11 (42.3)	2.88 (1.32-6.27)	.008	3.07 (1.38-6.82)	.006
Unreported	313 (37.6)	2.37 (2.04-2.75)	<.001	2.29 (1.96-2.68)	<.001
Regional average AGI, $
>100 000	362 (18.4)	1 [Reference]		1 [Reference]
76 000-100 000	392 (21.7)	1.23 (1.05-1.44)	.01	1.28 (1.08-1.50)	.004
51 000-75 000	765 (22.8)	1.32 (1.14-1.51)	<.001	1.35 (1.17-1.56)	<.001
≤50 000	484 (25.8)	1.55 (1.33-1.81)	<.001	1.52 (1.30-1.79)	<.001
Distance, miles
≤10	1355 (21.7)	1 [Reference]		1 [Reference]
11-20	374 (21.6)	1.00 (0.88-1.14)	.97	1.05 (0.91-1.20)	.51
21-30	103 (25.1)	1.21 (0.96-1.53)	.10	1.33 (1.05-1.69)	.02
>30	171 (27.5)	1.37 (1.14-1.65)	.001	1.55 (1.28-1.88)	<.001
Eye involvement
Bilateral	518 (18.0)	1 [Reference]		1 [Reference]
Unilateral	1485 (24.2)	1.45 (1.30-1.62)	<.001	1.44 (1.28-1.61)	<.001

Abbreviation: AGI, adjusted gross income (rounded to the nearest $1000); ND, no data.

^aMultivariate analysis adjusted for age, race/ethnicity, regional average AGI, distance, and eye involvement.

^bOther includes Hispanic, Native American, and Pacific Islander.

Sensitivity Analysis

Although the study was not designed to assess noncompliance with physician follow-up recommendations at intervals less than 12 months, when a sensitivity analysis was performed (eTable in the Supplement), the results for each of the different thresholds (12 months, 6 months, and 3 months) were similar. The results suggest that the independent variables are risk factors regardless of the specified threshold for the outcome. The independent variables seemed to be more strongly associated with LTFU at the 12-month interval.

LTFU by Visual Acuity

A total of 5307 patients had visual acuity measurements at their final injection, with visual acuity for the follow-up group recorded after a mean of 12 months of additional follow-up. There was an increase in LTFU rates as vision worsened from minimal to moderate to severe vision loss at the first and final injections (P < .001, by χ² test) (Table 3). Rate of LTFU increased as the fellow eye visual acuity at final injection decreased (n = 5300) (Table 3). Only patients who had no change in visual acuity had higher odds of LTFU than patients who gained more than 2 lines of visual acuity (n = 1956) (Table 3).

Table 3. Loss to Follow-up by Visual Acuity of Case Eye, Visual Acuity of Fellow Eye Stratified by Case Eye, and Change in Visual Acuity Among Patients With Neovascular Age-Related Macular Degeneration Who Received Intravitreal Injections With Anti–Vascular Endothelial Growth Factor.

Visual Acuity	Lost to Follow-up, No. (%)	Univariate Model		Multivariate Model
		Odds Ratio (95% CI)	P Value	Odds Ratio (95% CI)a	P Value
Case eye visual acuity
First injection (n = 2201)
≥20/40	68 (11.9)	1 [Reference]		1 [Reference]
20/50 to 20/200	186 (17.0)	1.51 (1.12-2.04)	.007	1.31 (0.96-1.79)	.09
<20/200	115 (21.5)	2.03 (1.46-2.82)	<.001	1.53 (1.09-2.16)	.02
Final injection (n = 5307)
≥20/40	203 (10.3)	1 [Reference]		1 [Reference]
20/50 to 20/200	372 (16.0)	1.67 (1.39-2.00)	<.001	1.38 (1.14-1.66)	.001
<20/200	196 (19.5)	2.11 (1.71-2.61)	<.001	1.45 (1.16-1.82)	.001
Fellow eye visual acuity
By minimal visual acuity loss in case eye (final injection, n = 1975)b
≥20/40	108 (9.4)	1 [Reference]		1 [Reference]
20/50 to 20/200	39 (10.2)	1.10 (0.75-1.62)	.62	1.30 (0.85-1.98)	.22
<20/200	56 (12.8)	1.42 (1.01-2.00)	.045	1.21 (0.84-1.75)	.32
By moderate visual acuity loss in case eye (final injection, n = 2319)c
≥20/40	96 (12.0)	1 [Reference]		1 [Reference]
20/50 to 20/200	116 (16.4)	1.44 (1.07-1.92)	.02	1.40 (1.03-1.92)	.04
<20/200	157 (19.4)	1.76 (1.34-2.32)	<.001	1.50 (1.11-2.02)	.008
By severe visual acuity loss in case eye (final injection, n = 1006)d
≥20/40	49 (13.1)	1 [Reference]		1 [Reference]
20/50 to 20/200	65 (23.5)	2.04 (1.36-3.07)	.001	1.79 (1.16-2.77)	.008
<20/200	82 (23.2)	2.01 (1.36-2.96)	<.001	1.57 (1.02-2.42)	.04
Change in visual acuity (n = 1956)e
>2 Lines improvement	67 (12.5)	1 [Reference]		1 [Reference]
≤2 Lines improvement	41 (12.2)	0.97 (0.64-1.47)	.89	0.95 (0.61-1.46)	.81
No change	110 (18.8)	1.62 (1.16-2.25)	.004	1.50 (1.06-2.11)	.02
≤2 Lines worsening	31 (13.9)	1.13 (0.71-1.78)	.61	1.22 (0.76-1.97)	.42
>2 Lines worsening	52 (18.8)	1.61 (1.09-2.40)	.02	1.49 (0.98-2.26)	.06

^aMultivariate analysis adjusted for age, race/ethnicity, regional average adjusted gross income, distance, and eye involvement.

^bMinimal visual acuity loss was defined as Snellen visual acuity ≥20/40.

^cModerate visual acuity loss was defined as Snellen visual acuity between 20/50 and 20/200.

^dSevere visual acuity loss was defined as Snellen visual acuity <20/200.

^eChange in visual acuity was calculated by subtracting the logarithm of the minimal angle of resolution at the first injection from the final injection for the case eyes that received at least 2 intravitreal injections with anti–vascular endothelial growth factor.

Geospatial Mapping

Heat mapping yielded signals corresponding to areas of increased rates of LTFU as shown in Figure 2A and B. A heat map using the risk factors from the multivariate model to predict LTFU showed multiple regions of correlation between the observed LTFU and probability of LTFU (Figure 2C).

Figure 2. Geospatial Mapping of Loss to Follow-up Rates Across Pennsylvania and New Jersey.

Geospatial heat map of loss to follow-up rates in southeastern Pennsylvania and southern New Jersey (A). Black arrowheads point to regions with high rates of loss to follow-up after receipt of intravitreal injection for neovascular age-related macular degeneration. Location markers indicate retina clinics associated with the practice. Scale bars indicate approximately 3 miles. B, Observed rates of loss to follow-up. C, Map of predicted rates of loss to follow-up determined with use of a predictive score from the multivariate model. B and C, Numerous correlated areas with high rates of patients who returned for follow-up (white arrowheads) and patients who were lost to follow-up (black arrowheads) are shown. Location markers represent 2 of the retina clinics associated with the practice. Scale bars indicates approximately 1 mile. The total area studied was approximately 19 000 square miles.

Discussion

Our study analyzed LTFU rates after receipt of IVI in a cohort of patients with nAMD. We found the rate of LTFU to be 22.2%, with most patients who were LTFU having no subsequent follow-up visit after 12 months (92.6%). The relatively high LTFU rates parallel findings of some earlier studies that assessed noncompliance with follow-up and LTFU among patients receiving IVI for treatment of nAMD. In a cross-sectional study by Droege et al¹² conducted in Germany, 95 patients receiving ranibizumab for nAMD had a noncompliance rate of 18.9%. Boulanger-Scemama et al¹¹ demonstrated that, over a 5-year period, 115 of 201 patients with nAMD (57%) were LTFU after receiving their first injection of ranibizumab.¹¹ These numbers, together with our findings, demonstrate that there is a sizable proportion of patients who potentially do not go back to their physician after receiving treatment.

A notable finding in our study was the disparity in LTFU rates among different races/ethnicities, with white patients having the lowest rate of LTFU. To date, there has been mixed evidence as to whether race is associated with any differences in noncompliance with eye care recommendations. African American patients with nAMD have been shown to be 23% less likely to receive anti-VEGF treatment¹⁷ and 18% less likely to have regular eye examinations for age-related macular degeneration compared with white patients.¹⁸ Patients of Hispanic and Asian race also have exhibited high noncompliance to recommended eye care.^19,20 Chinese Americans over the age of 50 years have demonstrated rates of noncompliance with regular eye examinations that parallel those among African American patients.²⁰ Studies have identified visit costs, insurance, and lack of time as potential causes of noncompliance among African American patients,²¹ whereas language barriers and having lived less than 10 years in the United States were other risk factors associated with noncompliance among Chinese Americans.²⁰ However, because of the complexity of the subject and the number of factors that might influence LTFU, a definitive explanation regarding why these rates actually differ remains unclear.

Another interesting finding in our analysis was the high rate of LTFU among patients of unreported race/ethnicity, which is concerning because this group included a sizable portion of the patient cohort. A study showed that patients of unreported race/ethnicity received fewer injections annually for nAMD compared with white patients.¹⁷ This observed phenomenon might be associated with a psychosocial component and possibly reflects a distrust in the health care system.

Increasing age was also associated with increasing LTFU rates. Earlier studies have shown that noncompliant patients and patients who are LTFU are, on average, 6 years older than other patients.^11,12 A possible explanation for this occurrence is the greater number of comorbidities among older patients. Patients older than 85 years have a multimorbidity (≥2 chronic diseases) prevalence of 81.5%, a 20% increase compared with individuals between the ages of 65 and 74 years.²² Moreover, the total number of comorbid conditions has also been shown to increase with age,²² and comorbid conditions can severely limit an individual’s ability to operate independently. Consequently, patients between 70 and 79 years of age and those older than 80 years have been shown to be 1.4 and 3.6 times, respectively, more likely to require assistance in performing activities of daily living compared with patients between the ages of 60 and 69 years.²³ Such impairments can be devastating to patients who are already receiving treatment for numerous chronic diseases.

Our results also showed an increase in LTFU as distance to clinic increased. The influence of distance became most notable at distances greater than 20 miles. In the study by Boulanger-Scemama et al,¹¹ distance also appeared to play a significant role in determining LTFU, with 51.7% of the patients who were LTFU identifying distance as the major cause of their noncompliance. Medicare beneficiaries who lived more than 20 miles from their ophthalmologist were 33% less likely to receive regular eye examinations within 4 consecutive 15-month periods than were patients who lived less than 20 miles from the ophthalmologist.²⁴ However, distance did not seem to influence the likelihood of receiving at least 2 examinations within the same follow-up period.²⁴ This result could imply that patients who live farther from a clinic might have trouble maintaining compliance with more rigorous follow-up schedules that entail shorter intervals and a greater number of visits.

Increased regional average AGI was also shown to be inversely associated with LTFU. Although regional average AGI is not necessarily reflective of an individual’s AGI, neighborhood socioeconomic status has been shown to influence patient behavior and health. Noncompliance with medical recommendations was found to be greater than 60% in low-income populations.²⁵ Moreover, individuals who come from neighborhoods of low socioeconomic status exhibit higher all-cause mortality and morbidity rates with influences that persist even after adjustment for individual household incomes.²⁶ It is postulated that low-income environments could potentially impede the social framework, which may in turn create a feeling of hopelessness.²⁶ Such feelings could be devastating to patients with nAMD, who may have to cope with several comorbidities and ailments given their age.

Our results also suggest an association between higher rates of LTFU and worse vision in both the case eye and the fellow eye. The evaluation of change in visual acuity also showed that only patients who did not see any appreciable difference were susceptible to LTFU, whereas any form of improvement was associated with lower rates of LTFU. However, the role of vision in determining whether a patient is susceptible to noncompliance or LTFU has not been sufficiently explored, and earlier studies have yielded equivocal results.^11,12,27 In addition, our study suggests that patients who receive bilateral injections have lower LTFU rates, which may be related to an association between more severe disease and stricter follow-up.

The geospatial mapping of LTFU created a more readily interpretable visualization of LTFU rates across the area covered by the retina clinics. The mapping of such data are potentially important for several reasons. Models assessing patients who are LTFU cannot account for all the observed variation given that behavior is often influenced by a variety of factors. Therefore, geospatial mapping might offer a more practical approach to identify populations most susceptible to LTFU. Geospatial maps have been used previously in public health studies. Coburn et al²⁸ constructed a map of HIV infection prevalence in Lesotho that provided insight into density distribution, allowing for potential treatment strategy implementation. Advanced software could potentially provide insight into trends associated with LTFU as well as identify regions associated with high susceptibility to LTFU for selective targeting. This approach could prove to be essential for the growing field of telemedicine. Finally, identified risk factors could be used to map the predicted LTFU rates of similar environments, which in turn would allow for future primary intervention strategies (Figure 2C).

Limitations

The study has several limitations. One is the definition of LTFU. Patients were considered to be LTFU whether they had only 1 or multiple LTFU intervals. This definition did not allow us to distinguish patients with multiple LTFU episodes from patients with only 1 LFTU episode. However, only 23 of the 2003 patients who were LTFU had more than 1 interval exceeding 12 months, with most patients never returning for a visit. Our study also did not assess for noncompliance with physician follow-up recommendations at shorter intervals. A study has shown that a significant decrease in visual acuity is observed in as little as 6 months after discontinuation of therapy.²⁹ Instead, our study aimed to identify those who were LTFU on the basis of the assumption that no treating physician in the practice would administer an injection to a patient and instruct them to return more than 12 months later. We chose 12 months rather than a shorter interval to improve the specificity of the observed findings but recognize that this timeline in turn may compromise sensitivity. Thus, patients who returned within 12 months may also have been susceptible to disease sequelae if they were not adherent to medical recommendations. Future evaluations of noncompliance based on physician-recommended follow-up intervals will be needed to enhance our understanding of that particular patient population. Another limitation is the use of billing codes, which may not be consistently accurate. Although limited evidence on coding accuracy exists for nAMD, a study showed that these codes were accurate when evaluating patients with diabetic retinopathy.³⁰ Additional limitations arise from our visual acuity analysis because a portion of the cohort had missing data and because of our reliance on Snellen visual acuity. However, these are issues that other practices might face when evaluating risk factors for LTFU. Finally, the results presented may not be representative of patients with nAMD in other practices or regions, thus necessitating future studies to confirm the validity of our findings.

Conclusions

Our study revealed that high rates of LTFU among patients with nAMD receiving anti-VEGF injections at a multilocation private retina practice were associated with several potential risk factors. With use of a novel geospatial mapping technique, we were able to graphically identify regions associated with high rates of LTFU. Although our results may not be generalizable, future studies in different regions can use a combination of both approaches that may allow for the identification of patients at highest risk for LTFU on a clinical and public health level. Additional real-world data on LTFU are needed to understand the scope of the problem so that interventions may be designed to improve patient outcomes.

Supplement.

eTable. Sensitivity Analysis Comparing Different Thresholds of Loss to Follow-Up Among Patients Receiving Anti-Vascular Endothelial Growth Factor for Neovascular Age Related Macular Degeneration