Age-related macular degeneration (AMD) is the leading cause of blindness in the ≥60-year-old population in the US [1]. Eye Diseases Prevalence Research Group [2] data suggest advanced AMD (central geographic atrophy and/or neovascular AMD) affected 1.87 million people in 2012 [3]. Among advanced AMD patients, 70 % had neovascular AMD (NVAMD) in one or both eyes [2]. Approximately 171,350 patients developed NVAMD annually [2, 3]. Among US citizens reaching age 65 annually, 944,400 have Category 3 drusen [2, 3].

Several models predict the conversion of atrophic AMD to NVAMD, each using the Age Related Eye Diseases Study (AREDS) classification of AMD [48]. The more complex model in AREDS Report No. 17 [6], evaluated the eyes of 3212 participants, utilizing drusen severity and pigmentary abnormalities at baseline. The authors proposed a 9-step severity scale that combined a 6-step drusen area scale with a 5-step pigmentary abnormality scale. The 5-year risk of progression to advanced AMD varied from <1 % in Steps 1 and 2 to 43.5 % in Step 9 [6]. Nonetheless, the scale more accurately predicts central geographic atrophy (43.5 % in Step 9) than NVAMD (4.8 % in Step 9, but 21.1 % in the less severe Step 8) [6]. A simplified severity scale in AREDS Report No. 18 used large drusen and pigmentary changes in a 0–4 scale and demonstrated, when both were present bilaterally, the 5-year incidence of advanced AMD in at least one eye was 47.3 % [7]. While a top score of 4 identified 67.8 % of 5-year progressors to advanced AMD, it identified only 36.5 % progressing to NVAMD [7].

Blue Mountains Eye Study investigators [8], using an AREDS simplified severity scale, found a generalized estimating equation model showed 61.5 % of patients with bilateral drusen ≥125 μm and bilateral retinal pigment epithelial changes progressed to advanced AMD over 10 years. Assuming 70 % of those advanced AMD cases were neovascular [2], approximately 43 % of NVAMD cases would be identified. This converts to (57.0 % × 161,754 cases=) 92,200 new NVAMD patients aged ≥65 in the US not identified by phenotypic markers annually [13].

Multiple advances in AMD genetics have been made over the last decade [9-17]. A report by Seddon et al. [12], created models predicting advanced AMD development. They approach include age, gender, education, smoking, body mass index, single nucleotide polymorphisms in the CFH, ARMS2/HTRA1, C3, C2, and CFB genes, as well as important markers discovered through two large genome-wide association studies [4, 1316]. With the inclusion of cholesterol metabolic markers CETP, LIPC and ABCA1, the progression prediction to advanced disease within 10 years achieved unparalleled accuracy (C = 0.90) [4]. Using this model for a dichotomous risk score (“risk” vs “non risk”), Yu et al. [4], showed the 5-year progression of Category 3 AMD cases to advanced AMD could be predicted with both sensitivity and specificity over 80 %; sensitivity over 10 years was ≥90 %. There was a 10-year 20 % progression to neovascular AMD (Table 1) using phenotypic and genotypic markers, but they identified 90 % of people progressing to neovascular AMD [4].

Table 1 Progression for 2560 AREDS AMD (Category 1–3 at Baseline) patients over 10.3 years (Yu et al. [4])

Neovascular AMD therapy

Intravitreal ranibizumab therapy for NVAMD is among the greatest medical advances over the decade [1824]. Earlier therapy has a better visual prognosis than later therapy [20]. Thus, it is hoped greater risk awareness will allow patients to seek earlier care.

Considering the importance of earlier ranibizumab therapy, the authors undertook a Value-Based Medicine® (VBM) [2224], societal, cost-utility analysis to assess the patient preference-based, comparative effectiveness and cost-effectiveness (cost-utility) of genetic testing for NVAMD.


Features associated with genetic testing for NVAMD, and the economic modeling assumptions used are listed in Table 2. Comparative effectiveness quantified the incremental patient value gain (improvement in quality-of-life and/or length-of-life), though length-of-life change was not included, since better vision has not been well shown to lengthen life. Outcomes were measured in: (1) percent value gain, and (2) quality-adjusted life-year (QALY) gain [18, 20, 2225]. QALY gain was calculated by multiplying: (utility gain) × (years of interventional benefit). Financial metrics include: (1) cost-utility ratio, $/QALY, or dollars expended per QALY gained, associated with genetic testing-enabled, early-treatment ranibizumab for NVAMD, and (2) societal costs.

Table 2 Genetic testing for neovascular AMD. Cost-utility analysis model parameters and assumptions

AMD demographics

The AREDS Research Group [57] defined four categories of AMD and showed oral supplements decrease the progression of Category 3 AMD (macular drusen ≥125 µm) to NVAMD, though not from Category 3 to central geographic atrophy. Yu and colleagues [4] refined the AREDS four-category model to a five-category model, separating AREDS [57] advanced AMD cases into central geographic atrophy (Category 4) and NVAMD (Category 5) in their genotypic/phenotypic study of AMD. Approximately 1.54 million people had NVAMD in the US in 2012 [2, 3]. Central geographic atrophy (AREDS Category 4) in at least one eye was present in 1.24 million, and 8.34 million had drusen ≥125 µm (AREDS Category 3). The 944,000 people aged 65 years with Category 3 drusen (≥125 µm) annually were those who were theoretically screened in our cost-utility model [2, 3].

Fifteen-year, incidence data from the Beaver Dam Study [26] suggest approximately 171,350 new cases of NVAMD develop annually in the US. Among these, 5.6 % in the Minimally classic/occult trial of the Anti-VEGF antibody Ranibizumab In the treatment of Neovascular AMD (MARINA) Study [18, 20] presented before age 65. We excluded that percentage from our analysis since our model assumed genetic testing at age 65, leaving 161,754 annual new cases.

Yu et al. [4], noted 20 % of Category 3 patients progressed to NVAMD (Category 5 AMD) over 10 years (Table 1). They also found 22.5 % (343/1527) of Category 3 patients had a high-risk genetic profile (homozygous for genetic loci on relevant alleles). Among progressors to NVAMD, 90 % had a high-risk genetic profile [4]. Thus, genetic testing for Category 3 AMD patients identified 90 % of progressors to NVAMD, 47 % higher than the 43.0 % identified over 10 years in the Blue Mountains Study [8].

Utility analysis

The quality-of-life associated with AMD has been quantified using time tradeoff utility analysis [2737]. Utility anchors are 1.00 (normal bilateral vision permanently) and 0.00 (death). Vision utilities correlate most highly with acuity in the better-seeing eye, rather than the underlying disease [28]. As vision in the better-seeing eye decreases, the associated utility decreases [2735]. Vision of no light perception bilaterally has an associated utility of 0.26 [30].

Ophthalmic utilities are valid [32], reliable [33], and negligibly affected by systemic comorbidities [34]. The Wills Eye Institute Institutional Review Board approved utility acquisition.

Ophthalmologists underestimated the quality-of-life associated AMD levels by 96–750 % compared to AMD patients, with community utility estimates even more disparate [31]. Thus, VBM cost-utility analyses use patient utilities [22]. Utilities herein were derived from a >1100 direct interview, vision utility database from the Center for Value-Based Medicine® [2837].

Value-Based Medicine®

Value-Based Medicine® (VBM) integrates the highest level, evidence-based, clinical trial data with standardized inputs, including: (1) time tradeoff utilities, (2) patient utility respondents, (3) a national Medicare Fee Schedule, and (4) societal and 3rd party insurer cost perspectives [2224]. VBM has been used extensively in ophthalmology, especially for AMD interventions [2224, 36, 37].

Originated at the Center for Value-Based Medicine® and based upon primary, ophthalmic patient data, the first-eye model assumes vision loss occurs in one eye, while the fellow eye has good vision [2224, 36]. In this instance, full patient value gain is not accrued until the fellow eye also develops NVAMD. Utility data from Center for Value-Based Medicine files demonstrate a mean utility difference of 0.0398 between unilateral good vision and 20/40–20/80 vision in the second eye, versus unilateral good vision and ≤20/160 vision in the second eye. This utility gain was weighted to appropriate first-eye model instances herein.

The second-eye model assumes first-eye vision has been lost and the second eye is affected [2224, 36]. Thus, greater patient value gain occurs with ranibizumab therapy. The combined-eye model used here integrates weighted first-eye model and second-eye models [22, 24]. Extrapolation of data from Barbazetto and colleagues [39] with Markov modeling (TreeAgePro for Healthcare 2012, Williamstown, MA, USA) showed 82 % of patients with unilateral NVAMD in the MARINA/ANCHOR trials developed bilateral NVAMD within 5 years, rising to 96 % by 12 years, the model timeline (Table 3).

Table 3 Fellow eye conversion to neovascular AMD in the MARINA/ANCHOR trials, Barbazetto et al. [39]

Patient value gain

Shah and DelPriore [25] modeled the natural course of untreated NVAMD using control cohorts from six randomized, NVAMD trials. In a meta-analysis using Lineweaver–Burke plots, they demonstrated mean vision loss to 20/640 over 8–9 years, after which vision stabilized. Increasing time since NVAMD highly correlated with increasing vision loss.

A double-blind, randomized, clinical trial, MARINA participants had 20/80 baseline vision in both ranibizumab-treatment and sham-treatment cohorts. The 24-month, mean, ranibizumab-treatment cohort vision was 20/63, while mean sham-treatment vision was 20/160−2 [18, 20].

Shah and DelPriore data [25] were employed to model MARINA 25–144 month, non-randomized sham results since many sham-treatment patients were allocated to ranibizumab therapy after month 24. Treatment cohort data for months 25–144 were modeled in a LOCF (last observation carried forward) fashion. Twelve years was selected as the model length since this was the average life expectancy of the average NVAMD patient.

An analysis of baseline vision in 50 consecutive Pennsylvania/New Jersey/Delaware patients presenting with first-eye NVAMD in the vitreoretinal practice of author GCB since 2010 was undertaken. The mean, baseline, first-eye vision was 20/182, while that in the fellow eye was 20/47. Overall, 78 % (39/50) presented with vision ≤20/200 in the first eye. Only 18 % (9/50) had first-eye vision ≥20/80 vision at presentation (Table 4).

Table 4 Category 3 AMD cases undergoing genetic testing: baseline vision in neovascular AMD eyes in clinical practice integrated with MARINA trial presenting percentages

An analysis of 98 consecutive patients presenting with second-eye NVAMD revealed a mean vision of 20/94. Overall, 62.2 % had vision ≤20/160. The vision was ≥20/80 in 37 % of eyes (Table 4).

Critical to our analysis, the excellent report by Boyer et al. [20] demonstrated NVAMD early-treatment eyes (baseline vision 20/40-20/80) had better long-term vision (mean 20/40−1) than late-treatment eyes (baseline vision ≤20/160 and mean long-term vision of 20/160+2) [20]. Total patient value gains (Table 4), to account for those eyes presenting with ≤20/160 vision and other parameters, had an overall multiplier of 78 % (1st eyes presenting with vision ≤20/160) × 62.2 % (2nd eyes presenting with ≤20/160) × 85.7 % (combined-eye multiplier to account for NVAMD conversion to both eyes) × 90 % (genetic testing sensitivity) = 37.4 %. Adverse events disutilities were subtracted in early-treatment and late-treatment cohorts for months 1–144 [23].


The mean, incremental, 12-year, direct, ophthalmic medical costs included genetic testing/monitoring. Ranibizumab therapy costs were excluded since they were assumed similar for early-treatment and late-treatment cohorts, though differences were analyzed in the sensitivity analysis.

Based upon work by Yu et al. [4], genetic testing costs are shown in Table 5. Compounded from $1461 at the time of genetic testing (age 65) to base-case age 75 at the initiation of ranibizumab therapy, they totaled $1906. A $299 cost for an extra, annual, ophthalmic examination and optical coherence tomogram (three/year rather than two/year) was included in the genetic costs for the 22.5 % [4] of genetic-screened high-risk patients progressing to NVAMD. The total cost for genetic testing/monitoring of Stage 3 AMD patients was $2205 per capita.

Table 5 Genetic testing costs for genes as per Yu et al. [4]

The base-case scenario shows that 30.3 Category 3 AMD patients required screening/monitoring to facilitate one early-treatment. The total cost of screening/monitoring for each early-treatment patient was therefore $66,873 (30.3 × $2205).

Twelve-year negative costs accrued against genetic testing/monitoring costs [4, 20, 3941]. Costs saved by early-treatment, vs. late-treatment (Table 6), are addressed below [3941].

Table 6 Societal costs (2012 US real dollars) per early-treatment patient from genetic testing/monitoring for neovascular AMD enabling incremental early-treatment (vs. late-treatment) ranibizumab therapy

Javitt and colleagues [39] demonstrated increased direct, non-ophthalmic, medical costs for depression, trauma, Skilled Nursing Facilities, nursing homes and unidentified entities associated with vision loss (Table 6). This 12-year cost gained by improving vision per early-treatment patient is (−$40,914) (Table 6).

Schmier and associates [40] reported increasing caregiver costs associated with decreasing levels of vision. Early-treatment ranibizumab therapy resulted in a 12-year, (−$172,443) caregiver cost saving vs. late-treatment therapy (Table 6).

Vision loss decreases employment by 45.6 % and hourly wage loss by 32.5 % referent to age-matched normals, resulting in 36.7 % of normal earnings (Table 6) [41]. Integrating age-related, US employment levels, early-treatment ranibizumab therapy accrued a 12-year employment cost gain of (−$14,098) [41].


Patient value gain

The mean MARINA, early-treatment, 20/40−1 vision outcome correlated with a 0.789 utility, while late-treatment 20/160+2 vision correlated with a 0.658 utility (Table 7) [20, 29, 30]. Additionally, 17.9 % of early-treatment, MARINA eyes achieved vision ≥20/25 bilaterally [20], (utility = 0.97 for bilateral 20/20–20/25 and 0.89 for unilateral 20/20–20/25). This conferred a 12-year, additional 0.121 QALY gain for early-treatment, a 2.0 % quality-of-life gain. First-eye therapy conferred 0.0884 QALY gain, a 1.5 % quality-of-life gain.

Table 7 Patient value gain from genetic screening-enabled, early-treatment ranibizumab therapy for neovascular AMD, integrating first-eye and second-eye MARINA Study models

The 12-year, combined-eye model, QALY accrual for each case of screening-facilitated, early-treatment, ranibizumab therapy was 0.845 QALY, a mean, a 14.1 % quality-of-life gain over sham therapy. Late-treatment accrued 0.250 QALY, a 4.2 % quality-of-life gain over sham-therapy. Thus, early-treatment conferred a, 10.0 %, absolute, quality-of-life gain, a 0.595 QALY gain per early-treatment patient. The QALY gain per genetic screened/monitored patient was 0.0177, a 0.33 % quality-of-life gain. Bilateral good vision QALY gain and first-eye model QALY gain were weighted appropriately


Direct ophthalmic medical costs for early-treatment vs. late-treatment ranibizumab therapy were assumed the same, thus excluded in the base case analysis, but addressed in the sensitivity analysis.

Assuming one early-treatment case per 30.3 genetically screened cases, the base-case genetic testing/monitoring cost for each early-treatment case was $66,873 (Table 6).

The incremental negative cost for each early-treatment ranibizumab patient was (−$227,455) (Table 6). With genetic testing/monitoring costs for Category 3 patients screened of $66,873 for each incremental early-treatment case, the overall societal cost per early-treatment case was (−$160,582) (Table 6).

The national, direct ophthalmic cost for genetic testing/monitoring for an annual cohort of 944,400 Category 3 AMD patients million was $2.082 billion. Total negative costs were $7.083 billion. This resulted in a 12-year, financial return-on-investment (ROI) of 240 % referent to genetic testing/monitoring costs. An incremental $260 million societal saving occurred for each 1 % of patients undergoing early-treatment ranibizumab therapy. When genetic testing facilitated an incremental 12,965 (8.0 %) of the 161,754, annual NVAMD patients in the US to undergo early-treatment ranibizumab therapy, an overall, net financial gain for society accrues at the rate of $160,582 per additional early-treatment patient.

The financial ROI distribution, assuming that genetic testing for 30.3 Category 3 AMD cases resulted in one incremental early-treatment case is shown in Table 8. The negative cost for each patient screened (cost of $2205) was (−$7500), an overall societal cost of (−$5295), also a 240 %, 12-year societal ROI. The ROI offset Medicare screening costs by 35 %, Medicaid costs by 63 % and commercial insurer costs by 22–24 %. Patients had the greatest ROI for out-of-pocket genetic testing costs, a net $6725. This converted to a 12-year 16,945 % ROI.

Table 8 12-year return-on-investment (ROI) for early-treatment (per patient screened) for ranibizumab therapy for neovascular AMD

Cost-utility ratio (CUR)

$144,000/QALY For a $144,000/QALY CUR, the upper limit of cost-effectiveness in the US according to World Health Organization criteria [43], an incremental 4.1 % of annual NVAMD patients were required to undergo genetic testing-enabled, early-treatment, ranibizumab therapy. This 4.1 % converts to 6634 patients among the 161,754 annual cohort of new NVAMD patients, 92,200 of whom are not identified as high risk to develop NVAMD by phenotypic features alone (Table 9) [13, 9]. It also equates to 1 per 142 of the 944,000 Category 3 AMD patients screened at age 65 annually. For a 3rd party insurer CUR of $144,000/QALY, an increment of 10.1 % of all annual NVAMD patients had to undergo early-treatment for cost-effectiveness (Table 10).

Table 9 Societal, cost-utility ratios for category 3 [5] age-related macular degeneration (AMD) patients undergoing genetic testing to predict progression to neovascular AMD
Table 10 3rd party insurer, cost-utility ratios for Category 3 [5] age-related macular degeneration patients undergoing genetic testing for progression to neovascular AMD

For a $100,000/QALY CUR, the upper limit of cost-effectiveness commonly utilized in the US [22], an incremental 4.5 % of annual NVAMD patients were required to undergo early-treatment, ranibizumab therapy for cost-effectiveness (Table 9). For a 3rd party insurer CUR of $100,000/QALY, an incremental 13.2 % of all annual NVAMD patients had to undergo early-treatment for cost-effectiveness (Table 10).

Sensitivity analysis

Sensitivity analysis (Table 11) showed decreasing age made genetic testing considerably less cost-effective (age 40 societal CUR = $1,088,253/QALY) and 3rd party insurer CUR = $231,137/QALY). Deleting extra ophthalmic monitoring costs had negligible effect, while decreasing genetic testing price improved cost-effectiveness. If the cost of ranibizumab therapy for early-treatment cases is twice that of the ranibizumab cost for late-treatment cases, an increment of approximately 6.0 % (9700) of NVAMD cases undergoing early-treatment ranibizumab therapy is required for genetic testing to be cost-effective using WHO criteria.

Table 11 Sensitivity analysis


Our analysis demonstrated, using MARINA Trial data [20], that each incremental, early-treatment, ranibizumab case of NVAMD conferred a combined-eye model, 14.1 % quality-of-life improvement, versus a 4.2 % quality-of-life improvement with late-treatment (baseline vision of 20/160–20/320) (Table 7), a 10.0 %, incremental, patient value gain. The net, societal, financial ROI was $160,582 per early-treatment case, a net $5295 ROI per patient screened.

While the US has no formal, cost-effectiveness upper limits, interventions costing ≤$100,000/QALY are generally believed cost-effective [22]. Formal World Health Organization standards indicate interventions costing ≤3× GDP per capita (~US $144,000) per DALY (disability-adjusted life-year), a metric similar to the QALY, are cost-effective (Tables 9, 10) [42]. Screening is still cost-effective if only 4.1 % overall NVAMD cases, or 7.2 % of cases not forecast phenotypically, receive genetic testing-facilitated early-treatment ranibizumab therapy

The presence of AMD, even with good vision, can decrease a patient’s quality-of-life [31]. We believe low-risk/medium-risk genetic profiles for progression to advanced AMD, likely allay patient fears and improve patient quality-of-life. We are uncertain how a positive, high-risk genetic profile affects patient quality-of-life, but since only 22.5 % of screened patients have a high-risk profile [4], the overall quality-of-life gain in the low-risk and medium-risk genetic profile cohorts could possibly outweigh total quality-of-life loss in the high-risk profile cohort.

Wealth of the nation

The data herein support the work of Nordhaus [43], the Yale economist who estimated 50 % of the wealth of the United States created during the twentieth century occurred from healthcare advances. While secondary to patient value gain, the increase in national wealth associated with medical interventions is an important factor. If 12,965 (8.0 %) of the 62,279 NVAMD cohort patients not predicted phenotypically to develop NVAMD are recruited to early-treatment due to genetic testing, the cost of testing is at a breakeven point. Each patient undergoing early-treatment in addition adds $160,582 to societal wealth.

When to genetically screen

We selected the age of 65 years as the most cost-effective age for genetic screening since it encompasses 94.4 % of those who develop NVAMD [18]. As sensitivity analysis shows, the earlier genetic testing is performed, the greater the expense of testing, since analyses must account for the time value of money.

Patients with NVAMD may not come in promptly due to numerous reasons, including: (1) unawareness of NVAMD symptoms [44], (2) not realizing vision is decreased in one eye, (3) denial, (4) absence of pain, (5) believing refraction or cataract is their problem, (6) cognitive difficulties and (7) others. In non-ophthalmic specialties, well-defined follow-up plans [45], good relationships with healthcare providers [46], and focused programs [46] result in greater patient adherence. While we cannot be certain, we are hopeful that high-risk phenotypic/genotypic profile patients will be especially aware to present promptly when they develop NVAMD symptoms.

To our knowledge, convincing data that show more frequent ophthalmic screening allows earlier NVAMD treatment are lacking. Nonetheless, we included the extra costs of screening herein to be conservative in our economic analysis.

Some might argue genetic testing of Category 3 AMD cases is of no benefit over phenotypic progression parameters. While the case for Category 4 atrophic AMD centrally or unilateral NVAMD, phenotypic progression parameters for Category 3 AMD are less reliable [6, 47]. Furthermore, only a small increment (4.1 %) in genetic testing-enabled, early-treatment, ranibizumab therapy cases is necessary for screening cost-effectiveness. As genetic testing likely increases in accuracy and decreases in price, the patient and financial value gains will be more pronounced. Though the study was performed in US dollars, WHO criteria are very similar for many developed countries globally [48].


In summary, genetic screening for NVAMD is cost-effective by the WHO standard of $144,000/QALY if it facilitates early-treatment with ranibizumab for an incremental 4.1 % of annual NVAMD cases. If genetic testing allows earlier ranibizumab therapy for an incremental one NVAMD case per 142 Category 3 AMD patients screened, it remains cost-effective by WHO standards. Using the often accepted US upper limit of cost-effectiveness of $100,000/QALY, genetic testing for NVAMD is cost-effective if it facilitates an incremental 4.5 % of annual NVAMD cases to undergo early-treatment ranibizumab therapy. This information can be used by clinicians to decide whether genetic testing for neovascular age-related macular degeneration is appropriate for their patients.