Cost Evaluation of Early Vitrectomy versus Panretinal Photocoagulation and Intravitreal Ranibizumab for Proliferative Diabetic Retinopathy

Abstract

Purpose:

To evaluate costs and cost-utility of early vitrectomy (pars plana vitrectomy [PPV]) compared with panretinal photocoagulation (PRP) and intravitreal ranibizumab (IVR) for proliferative diabetic retinopathy (PDR) without diabetic macular edema.

Design:

A decision analysis model of cost-utility.

Participants:

There were no participants.

Methods:

A decision analysis was based on results from the Diabetic Retinopathy Clinical Research Network Protocol S comparing treatment of PRP with IVR (0.3 mg) in PDR without incident macular edema to model the total 2-year costs and outcomes for each treatment scenario. These values were compared with the 2-year hypothetical costs of early PPV for PDR. Centers for Medicare and Medicaid Services data were used to calculate associated modeled costs in a hospital/facility-based and nonfacility setting. Cost-utility was calculated on the basis of the preserved visual utility and estimated life years remaining. In addition, costs for lifetime treatment were modeled for all scenarios and used to calculate lifetime quality-adjusted life years (QALY) costs for each scenario. Sensitivity analyses were performed to evaluate the impact of the model’s assumptions.

Main Outcome Measures:

Cost of treatment, utility, and cost per QALY.

Results:

The modeled cost per QALY of treatment for PDR for 2 years of utility in the facility (nonfacility) setting was $163 988 ($102 559) in the PRP group, $436 992 ($326 424) in the IVR group, and $181 144 ($107 965) in the PPV group. Sensitivity analysis showed that both IVR and PPV groups would have equivalent costs per QALY over the first 2 years if 78% (facility) and 80% (nonfacility) of patients in the PPV group required additional treatment with IVR (at the dose of 10.1 injections as in Protocol S). Beyond 2 years, the cost per QALY in the facility (nonfacility) setting was calculated as $61 695 ($21 752) in the PRP group, $338 348 ($239 741) in the IVR group, and $63 942 ($22 261) in the PPV group.

Conclusions:

Early PPV as a strategy for treatment of PDR without macular edema demonstrates cost-utility similar to management with PRP and more favorable cost-utility compared with IVR in the short term. This advantage over IVR continues when lifetime costs are factored.

Diabetic retinopathy is a leading cause of global blindness; approximately one third of the estimated 285 million people with diabetes mellitus worldwide have signs of diabetic retinopathy, with one third of these having vision-threatening diabetic retinopathy.¹ The development of proliferative diabetic retinopathy (PDR) can lead to varying degrees of vision loss, including blindness from vitreous hemorrhage, neovascular glaucoma, or tractional retinal detachment. Treatment for PDR has historically relied on panretinal photocoagulation (PRP) to prevent future progression and to effect regression of proliferative disease. The Diabetic Retinopathy Study (DRS) established PRP as the standard of care approximately 40 years ago,² when it demonstrated that PRP reduces the rate of severe visual loss (defined as visual acuity <5/200 at ≥2 consecutively completed visits at 4-month intervals) by 50%.³

Panretinal photocoagulation’s principal mechanism of action seems to involve a reduction of vascular endothelial growth factor (VEGF).⁴ Intravitreal anti-VEGF therapy has now been recognized as an alternative to treatment of PDR, and with US Food and Drug Administration approval has been increasingly advocated as such.^5,6 In particular, intravitreal ranibizumab (IVR) (Lucentis; Genentech, Inc., South San Francisco, CA) has been shown to have a synergistic effect with PRP and to be noninferior to PRP in regard to improvement in visual acuity.^7,8 The CLARITY trial⁹ has recently reported similar results with intravitreal aflibercept (Eylea; Bayer Pharma AG; Berlin, Germany) in a cohort without incident diabetic macular edema (DME); in fact, a 4-letter superior visual outcome over PRP was reported.

Anti-VEGF agents have the dual effect of decreasing DME and non-PDR, as well as PDR.^10–12 Their clinical superiority in treating DME is undeniable, an important consideration given the substantial prevalence of coexisting DME among eyes with concurrent PDR—71% in one study.¹³ Chronic anti-VEGF therapy has the potential of prompting the same paradigm shift in PDR treatment as it has for DME. However, anti-VEGF therapy is expensive, offers marginal clinical benefit over alternative therapies, requires a substantial patient treatment burden (threatening compliance), and carries a finite risk of causing deleterious side effects, such as endophthalmitis or cardiovascular side effects.

These disadvantages raise the question of whether the option of early pars plana vitrectomy (PPV) for PDR treatment is a feasible alternative. Although generally considered for diabetic patients with more advanced complications, such as at least several months of nonclearing vitreous hemorrhage,^14–17 the advent of small-gauge vitrectomy (and perhaps the use of preoperative anti-VEGF agents) seems to have reduced surgical complications and improved clinical results; thus, there seems to have been a de facto expansion of the indications for earlier stages of complications in PDR.^18–26 Relatedly, PPV has also been widely reported, even advocated, for DME treatment.^27–29

The purpose of this study was to extend a previously published cost-utility model comparing initial treatment of PDR with PRP versus IVR,³⁰ to consider the cost-utility ramifications of using early PPV for PDR.

Methods

Institutional review board approval was not required because there were no research participants or medical records reviewed. The current cost-utility analysis model uses published resource use and outcomes data from the Diabetic Retinopathy Clinical Research Network Protocol S, a randomized clinical trial comparing PRP with IVR as the primary treatment for PDR, as a basis for cost of treatment.⁸ Only data from the subset of patients in Protocol S without baseline DME were included in this current study. Because there are no published index studies to quantitate use parameters for early PPV in the setting of PDR, the cost-utility values for PPV are derived from these investigators’ estimates based on the clinical courses reported for IVR and PRP, tempered by clinical experience.

Medicare fee data from 2017 were calculated using the Centers for Medicare and Medicaid Services schedules to ascertain the allowable cost (in 2017 US dollars) associated with each procedure, injection, study, or office visit.^31–33 Costs were calculated for both a hospital-based facility (with any surgery performed in a hospital operating room) and a nonfacility (i.e., office-based clinical services with any surgery performed in an ambulatory surgery center) to define the range of potential reimbursement settings. Professional fees (including the use of an anesthesiologist in the operating room for the fraction needing PPV or subsequent cataract extraction) and facility fees were included in the calculations.

The dollars conversion factor per relative value unit (RVU) used was $35.89, the established rate for 2017.³¹ The equation used to calculate the cost for a given provider service included work RVUs in the form of professional fees, practice expense RVUs, and malpractice RVUs. Each of these factors is subject to geographic modifiers; the rates for Miami, Florida, were used in this analysis.³¹

Anesthesia fees were calculated by multiplying the base units, time units, and conversion factor for the specific procedure (in this study PPV and cataract surgery). The conversion factor for Miami, Florida, in 2017, is $24.24.³¹Anesthesia for vitreoretinal surgery (Current Procedural Terminology [CPT] code 00145) is weighed as 6 base units, and anesthesia for cataract surgery (CPT 00142) is weighed as 4 base units. One time unit is 15 minutes, and consistent with our previous analyses the assumed case length was 1 hour for vitreoretinal surgery and 30 minutes for cataract surgery.^34,35

A decision analysis model was formulated to generate a total cost and cost per quality-adjusted life year (QALY) (cost/QUALY) for each treatment scenario based on these assumptions.^36,37 Three scenarios were modeled as the primary treatment for PDR: PRP, IVR and PPV. Intravitreal ranibizumab and PRP data were drawn from the Protocol S report using methodology based on the previous study published by these same authors, with the modification of only patients without concurrent DME modeled.^8,30 Data regarding the stratification according to baseline DME of patients needing PPV in Protocol S were not available to the authors. The primary analysis incorporated the costs (and utilities) over 2 years and was extended to lifetime based on the same utility gained and life expectancy years obtained from the actuarial tables of the Social Security Administration (weighted by gender) as previously published.^30,38

Costs for patients who received secondary IVR in the PRP and PPV scenarios were estimated, as well as those who received secondary PRP in the NR scenario based on Protocol S data for the former and estimates for the latter. A sensitivity analysis was performed in the PPV group while varying the number and frequency of IVR (putatively for coincident DME or management of recurrent PDR) to give the expected cost-utility ranges in which treatments for the PPV group would be expected to be as efficacious as treatments in the IVR group for 2 years and extended over a lifetime.

The CPT codes used for the procedures were as follows: 67228 for PRP, 67028 for intravitreal injection, 67210 for focal laser, 67040 for PPV with laser, and 66984 for cataract surgery (Table 1). A comprehensive eye code (CPT 92004) was included for the initial visit, with subsequent follow-up visits calculated using the intermediate eye code (CPT 92012). The Healthcare Common Procedure Coding System code used for ranibizumab was J2778. The reimbursement schedules for procedures were based on the Centers for Medicare and Medicaid Services terminology for procedures performed in the facility and nonfacility settings.

Table 1.

Medicare Allowable Costs for Panretinal Photocoagulation/Intravitreal Ranibizumab/Pars Plana Vitrectomy and Associated Treatments

Procedure	Facility					Nonfacility
	CPT Code	Professional Fee	Facility Fee	Anesthesia	Total	Professional Fee	Facility Fee	Anesthesia	Total
PRP	67228	$335	$470	$0	$805	$349	$0	$0	$349
IVR	67028	$110	$279	$0	$389	$115	$0	$0	$115
Ranibizumab	J2778 (HCPCS code)				$1134 (stocking fee $49)
OCT	92134	$43	$55	$0	$98	$35	$0	$0	$35
Fundus photography	92250	$69	$100	$0	$169	$31	$0	$0	$31
Focal laser	67210	$543	$470	$0	$1013	$505	$0	$0	$505
PPV with PRP	67040	$1135	$3419	$242	$4796	$1152	$1750	$242	$3145
Cataract extraction	66984	$694	$1824	$145	$2664	$677	$978	$145	$1800
New comprehensive eye exam	92004	$106	$107	$0	$213	$139	$0	$0	$139
Intermediate follow-up exam	92012	$57	$107	$0	$163	$71	$0	$0	$71

CPT = Current Procedural Terminology; HCPCS = Healthcare Common Procedure Coding System; IVR = intravitreal injection of ranibizumab; PPV = pars plana vitrectomy; PRP = panretinal photocoagulation.

This cost-analysis used the Food and Drug Administration–approved IVR dose of 0.3 mg for diabetic retinopathy, which was different from the 0.5 mg dose used in Protocol S.⁸ For comparison, the authors modeled costs using IVR dose of 0.5 mg as well. Reimbursement per 0.1-mg unit of ranibizumab is $378 according to the 2017 Centers for Medicare and Medicaid Services reimbursement rates³⁹ A drug maintenance cost of 4.3% of the medication cost was applied to the overall injection cost in each group. As used in this study, a dose of ranibizumab 0.3 mg would cost $1134, and the 4.3% stocking fee would be $49. In the facility setting, the cost model included both the professional and hospital fees; in the nonfacility setting, there were no facility charges for laser or injections. It was assumed that all injections were done in the office setting, so no anesthesia professional fees were applied.

The DRS defined severe visual loss in untreated eyes as best-corrected visual acuity of 5/200 corresponding to a utility of 0.52 and a mean PRP treatment success of 20/100, corresponding to a utility of 0.67.^2,40 Therefore, the amount of utility saved by pre-serving 20/100 instead of 5/200 would be 0.67–0.52=0.15 utility units. Data from the DRS showed that there was a 15% absolute risk reduction in severe vision loss (5/200 or worse) with PRP treatment compared with observation in high-risk PDR.² This model assumed all patients with PDR would have visual outcomes (immediately) equivalent to the high-risk characteristics subgroup, so a 15% absolute risk reduction was modeled, yielding a saved utility of 15%×0.1 5=0.023 utility units, all in the first year. A parallel analysis of maintaining 20/50 best-corrected visual acuity, which might be more realistically applicable to current practice (or a cohort undergoing even earlier surgical intervention), was also calculated (utility value of 0.77). If we assumed the better outcome (from earlier intervention), the amount of utility saved would be 0.77–0.52=0.25 units; compounding that with the 15% absolute risk reduction yielded l5%×0.25=0.0375 utility units saved. In all cases, the visual acuity in the treated eye was assumed to be equal to or worse than that in the fellow eye, so that “second eye” utility values were used.

Beyond 2 years, calculations assumed a 5% repeat PRP rate per year in years 3 to 4 (PRP treatment group) but none after year 4, extrapolating from Protocol S data. The frequency of injections for subsequent years was modeled as 2 per year for the lifetime (IVR group). The PPV rates were assumed to be 5% per year in the PRP group and 2.5% per year in the IVR group in years 3 and 4. In the PPV group, there were no additional lifetime treatment costs (i.e., no IVR or PRP, but a sensitivity analysis of hypothetical IVR use was performed) other than an assumed 5% reoperation rate within the first 2 years. After the first 2 years, an assumption of 3 yearly follow-up visits in the IVR group, l yearly follow-up visit in the PRP group, and 1 yearly follow-up visit in the PPY group was made. To consider the cost differential, a model that factored in the American Academy of Ophthalmology recommendations for 1 annual visit by subtracting its costs was made as wel1.⁴¹ All future costs/QALY factored in a standard 3% inflation rate per year. Calculations and analyses were performed using Microsoft Excel (Microsoft Corp., Seattle, WA).

Results

The assumed unit costs, including the facility and professional fees (Table 1), were applied to the estimated frequency of resource use based on the average number of treatments for 2 years in the PRP, IVR, and PPV scenarios (Table 2). These results yielded cost-utility measures (Table 3).

Table 2.

Estimated Use of Resources Based on Initial Procedure over 2 Years

Procedure	CPT Code	Scenario 1, PRP	Scenario 2, IVR	Scenario 3, PPV
New comprehensive eye exam	92004	1	1	1
Intermediate follow-up exam	92012	5.5	15.25	8
PRP	67228	1.5	0.05	0
Intravitreal injection	67028	2.2	10.1	0.5
OCT	92134	3	3	3
Fundus photography	92250	3	3	3
Focal laser	67210	0.08	0.05	0
PPV with PRP	67040	0.15	0.04	1 (0.05 reoperations)
Cataract extraction	66984	0	0	0.1

CPT = Current Procedural Terminology; IVR = intravitreal ranibizumab; PPV = pars plana vitrectomy; PRP = panretinal photocoagulation.

Table 3.

Total Cost and Cost/Quality-Adjusted Life Years at 2 Years and Lifetime (Intravitreal Ranibizumab 0.3 mg)

Primary Procedure	Utility	Years Remaining (Average)	Facility Billing				Nonfacility Billing
			Total Cost (2 yrs)	Cost/QALY (2 yrs)	Lifetime Cost	Cost/QALY (Lifetime)	Total Cost (2 yrs)	Cost/QALY (2 yrs)	Lifetime Cost	Cost/QALY (Lifetime)
PRP	0.023	30.3	$7379	$163 988	$42 182	$61 695	$4615	$102 559	$14 872	$21 752
IVR	0.023	29.3	$19 665	$436 992	$244 192	$338 348	$14 689	$326 424	$158 855	$239 741
PPV	0.023	29.3	$8151	$181 144	$42 369	$63 942	$4858	$107 965	$14 750	$22 261

IVR = intravitreal ranibizumab; PPV = pars plana vitrectomy; PRP = panretinal photocoagulation; QALY = quality-adjusted life years.

Scenario 1: Panretinal Photocoagulation as Primary Treatment

The imputed 2-year cost for facility (nonfacility) treatment was $7379 ($4615), yielding a cost/QALY of $163 988 ($102 559) for those initial 2 years of utility. The modeled lifetime cost would be $42 182 ($14 872) and the lifetime cost/QALY would be an estimated $61 695 ($21 752) (Fig 1).

Figure 1.

Estimated cumulative lifetime cost of additional therapy. IVR = intravitreal ranibizumab; PPV = pars plana vitrectomy; PRP = panretinal photocoagulation; QALY = quality-adjusted life year.

Scenario 2: Intravitreal Ranibizumab as Primary Treatment

Primary treatment in the facility (nonfacility) setting with IVR yielded an imputed 2-year cost of $ 19 665 ($14 689) and a cost/QALY of $436 992 ($326 424) for those initial 2 years of utility. Lifetime cost would be $244 192 ($158 855), and the lifetime cost/QALY for a maintained visual acuity benefit would be $338 348 ($239 741) (Fig 1).

The cost analysis using the 0.5-mg dose was calculated as well for comparison (Table 4), assuming it would cost $1890, plus the stocking fee of 4.3%. By using these numbers, the increase over the 2 years in cost using 0.5 mg instead of 0.3 mg would be $38 549/QALY in the PRP treatment scenario (24% increase in the facility setting and 38% increase in the nonfacility setting) and an increase of $176 976/QALY in the IVR treatment scenario (40% increase in the facility setting and 54% increase in the nonfacility setting).

Table 4.

Total Cost and Cost/Quality-Adjusted Life Years at 2 Years and Lifetime (Intravitreal Ranibizumab 0.5 mg)

Primary Procedure	Utility	Years Remaining (Average)	Facility Billing				Nonfacility Billing
			Total Cost (2 yrs)	Cost/QALY (2 yrs)	Lifetime Cost	Cost/QALY (Lifetime)	Total Cost (2 yrs)	Cost/QALY (2 yrs)	Lifetime Cost	Cost/QALY (Lifetime)
PRP	0.023	30.3	$9114	$202 537	$43 917	$64 232	$6350	$141 108	$16 607	$24 289
IVR	0.023	29.3	$27 629	$613 968	$309 434	$446 994	$22 653	$503 400	$244 097	$368 387
PPV	0.023	29.3	$8550	$189 993	$42 767	$64 543	$5257	$116 814	$15 148	$22 861

IVR = intravitreal ranibizumab; PPV = pars plana vitrectomy; PRP = panretinal photocoagulation; QALY = quality-adjusted life years.

Scenario 3: Pars Plana Vitrectomy as Primary Treatment

The imputed 2-year cost for treatment was $8151 ($4858) in the facility (nonfacility) setting, and the cost/QALY was $181 144 ($107 965) for the initial 2 years of utility. Lifetime cost would be $42 369 ($14 750), and cost/QALY over a lifetime would be $63 942 ($22 261) (Fig 1). If assuming a 10% rate of cataract extraction in patients who underwent PPV, the cost/QALY over 2 years would be $187 064 ($111 965), a 3% to 4% increase. If considering the 0.5-mg IVR dose, there would be an increase to $8849/QALY (5% increase in the facility setting and 8% increase in the nonfacility setting) in the first 2 years (Table 4).

A sensitivity analysis was performed, and in the facility (nonfacility) setting 78% (80%) of those in the group would need to be treated with 10.1 injections of IVR (the frequency used in Protocol S) on top of the assumptions of the PPV group for cost/QALY to be even with that of the IVR group in the 2-year treatment period. Calculated alternatively as the average number of injections for the entire group for cost equivalence, this would equate to an approximate 8 additional IVR injections over the 2 years. Varying the annual rate of IVR injections for the PPV group after year 2, lifetime cost/QALY would equilibrate with the IVR group with 2 to 4 additional injections, but still be favorable from a cost-utility perspective for <2 hypothetical annual IVR injections (Table 5). If the assumed vision saved with early PPV were 20/50 from 5/200, cost/QALY over a lifetime would be approximately 40% less across all treatments (Table 6).

Table 5.

Lifetime Cost/Quality-Adjusted Life Years for Pars Plana Vitrectomy with Various Rates of Annual Intravitreal Ranibizumab Injections, Sensitivity Analysis

	Rate of Annual IVR Injections
	0	0.5	1	2	4
PPV Cost/QALY
Facility	$63 942	$122 069	$180 196	$296 450	$528 958
Nonfacility	$22 261	$70 246	$118 231	$214 202	$406 143

IVR = intravitreal ranibizumab; PPV = pars plana vitrectomy; QALY = quality-adjusted life years.

Table 6.

Cost/Quality-Adjusted Life Years over 2 Years and Lifetime (Assuming Maintaining 20/50 Best-Corrected Visual Acuity)

Primary Procedure	Utility	Years Remaining (Average)	Facility Billing		Nonfacility Billing
			Cost/QALY (2 yrs)	Cost/QALY (Lifetime)	Cost/QALY (2 yrs)	Cost/QALY (Lifetime)
PRP	0.0375	30.3	$98 393	$37 017	$61 535	$13 051
IVR	0.0375	29.3	$262 195	$203 009	$195 854	$143 845
PPV	0.0375	29.3	$108 686	$38 365	$64 779	$13 356

IVR = intravitreal ranibizumab; PPV = pars plana vitrectomy; PRP = panretinal photocoagulation; QALY = quality-adjusted life years.

Although this study assumed 3 annual visits in the IVR group and 1 annual visit in the PRP/PPV groups, if 1 fewer visit was modeled according to the American Academy of Ophthalmology guidelines, this would yield a lifetime cost/QALY of $49 996 ($16 685) for the PRP group, $326 276 ($234 514) for the IVR group, and $51 870 ($17 033) for the PPV group.

Discussion

This decision analysis model calculated the costs and cost-utility of primary treatment of PDR without DME for 3 scenarios: PRP, IVR, and PPV. The cost/QALY for the first 2 years of treatment was lowest for PRP—early PPV was 1.1 times that of PRP, whereas IVR was 2.7 to 3.2 times that of PRP and 2.4 to 3 times that of early PPV. Early PPV maintained a cost utility similar to that of PRP during the first 2 years of treatment as well as over a lifetime of treatment. However, over a lifetime the gap between IVR and early PPV continued to increase, with cost/QALY 5.3 to10.8 times higher, because the IVR group was presumed to need more annual injections than the PPV group. If only considering cost utility for 2 years, all treatment scenarios were above the acceptable range of $50 000 to $100 000/QALY.³⁷ However, if it is assumed that patients would, based on their average life expectancy, be living beyond 2 years, both PRP and early PPV would fall within the acceptable lifetime cost/QALY range, whereas IVR still exceeded this limit. As the sensitivity analysis demonstrates, the greatest accruing cost is that of IVR. Although it is the authors’ experience that new-onset DME is uncommon after diabetic PPV, there are not good data on incidence; if one assumes even an annual average of 0.5 injections in the PPV group, the cost/QALY would essentially double in the facility setting and triple in the nonfacility setting (Table 5).

The presumed economic advantage of early PPV is that (like PRP) it has the theoretical potential of being a temporally definitive treatment for PDR, preventing further visual loss with few or no ongoing expenses. This would seem particularly attractive from the cost-utility perspective for younger patients. Treatment of PDR with early PPV has been found to be beneficial in the context of vitreous hemorrhage, as established by the Diabetic Retinopathy Vitrectomy Study Group.^15,17 Although 27% of the IVR group and 34% of the PRP group in Protocol S developed vitreous hemorrhage, only 4% and 15% in each group, respectively, underwent PPV.⁸ Although the overall rates of PPV were low in Protocol S, studies have shown that in the treatment of vitreous hemorrhage for PDR, up to one third of patients may need PPV despite treatment with IVR.⁴² If the rate of progression to PPV is substantial, early PPV would be particularly beneficial from a cost perspective by forgoing the additional interval expenses of preoperative and postoperative treatments with IVR. Additionally, because early PPV includes endolaser, this might have the theoretical benefit of precluding the need for further PRP; moreover, it is these authors’ clinical impression that the PRP done in conjunction with PPV is often less extensive than PRP in a nonsurgical patient. Although preservation of visual field may represent a theoretical reason to choose IVR over PRP/PPV with endolaser, there is evidence to suggest from patient-centered outcomes of PRP (without a PPV cohort) that peripheral visual field loss might not actually be different between the treatments.⁴³

This study did not include patients from Protocol S with baseline DME; IVR has the advantage of concurrently treating DME and has been shown to have an improved incremental cost-effectiveness compared with PRP in this subset of patients, but still at a relatively high cost.⁴⁴ The lack of such data in the patient after PPV would prohibit a fair analysis for that subset. However, data are available making feasible a comparison of PRP with IVR subsets; although it is pure conjecture, the PPV subgroup would likely resemble the PRP group more than the IVR group. Hutton et al⁴⁴ did not include extrapolated maintenance costs to preserve utility beyond 2 years. That study calculated marginal differences in the utility of IVR over PRP as standard care using the incremental cost-effectiveness ratio, so although it did not include the cost-utility involved in getting to the state studied, it showed the extremely high cost necessary for the relatively small utility IVR might offer. The current study attempted to calculate the absolute utility values directly, albeit compared with hypothetical untreated natural history drawn from DRS control data. Like IVR, early PPV may be helpful in the treatment of concurrent DME, at 2 years and beyond, an effect presumably lacking with PRP.^27–29 Some studies have shown benefit by inducing a posterior vitreous detachment during PPV in treating or preventing DME recurrence, further increasing the cost benefit of this treatment,⁴⁵ but a prospective, randomized study did not confirm this benefit.²⁷

An admittedly substantial limitation of this study is its reliance on estimates of outcomes and resource use, especially beyond an initial 2-year treatment course. Because there are no existing clinical trials that have examined the benefits of early PPV over IVR or PRP, there is not a firm basis to explore the implications of its potential ongoing costs. Historically, indications for PPV in PDR have involved more severe processes, so relying on DRS or Data Reporting and Analytics Solutions data is also fundamentally flawed. Although with the increased safety of PPV and its earlier use in treating complications of diabetic retinopathy, PPV use as modeled here may currently not be considered clinically indicated because many such patients would be considered treatable only with PRP or IVR.²² In fact, that is why this study does have some contemporary merit. Moreover, this model assumes PPV to be noninferior to PRP and IVR for PDR, but this has not been established by a definitive study.

There are several areas where one might debate the use assumptions in this study model that might affect the results, some more or less than might be intuitively expected. For example, if 1 follow-up visit were subtracted, figuring an annual baseline visit in accordance with the American Academy of Ophthalmology guidelines, there would be a lifetime reduction of 19% to 23% of cost/QALY in the PRP/PPV groups, while only a 2% to 4% reduction in the IVR group. Other factors were also varied to gain a sense of their impact on cost-utility. The 10% cataract extraction rate was based on the generally lower rates seen in diabetic post-vitrectomized eyes consistent with other studies.^46,47 The 5% reoperation rate in the PPV group is an estimate but is based on data for reoperations for vitreous hemorrhage that have been reported to be at 13%, so if this were applied to the 27% to 34% rate of vitreous hemorrhage in Protocol S, the reoperation rate would be an estimated 3% to 5%.^8,48 The authors considered that PPV for cases of PDR that would otherwise be enrolled in chronic IVR treatment would likely not have as high rebleed/reoperation rates as those reports; thus, the 5% assumed rate did not seem low by the authors’ estimates. Even if a sensitivity analysis were conducted where 100% of patients in the PPV group needed a single reoperation, IVR would still have a 79% to 89% higher lifetime cost/QALY, which further highlights the cost benefit of PPV. Because preoperative anti-VEGF injection is now administered in some cases, the impact of a single preoperative IVR would be a 4% increase in lifetime cost/QALY in the facility setting (9% increase in the nonfacility setting).

Other costs (but also utilities/benefits) related to perioperative optimization of medical covariates (which might otherwise be neglected but beneficial), such as optimizing blood glucose and blood pressure, and managing anti-coagulation to avoid excessive intraoperative bleeding, were not modeled. Although the frequency of other complications is low, costs/QALY including a 2% risk of postoperative rhegmatogenous retinal detachment requiring scleral buckling or vitrectomy (modeling with costs according to CPT 67108) would be increased only by approximately 1% over the initial 2 years. Likewise, the risks of endophthalmitis for PPV and intravitreal injections are similar, and even lower (~ 0.04% each), considering multiple injections as extrapolated from the Protocol S injection rates and as reported after PPV.^8,49 The incidence of endophthalmitis was found to have a negligible effect on increased lifetime cost/QALY (<0.1% in the IVR group and 0.2%–0.4% in the PPV group). Additionally, the theoretical increased risk for adverse events with IVR as defined by the Antiplatelet Trialists’ Collaboration, including arterial thrombotic events, myocardial infarction, stroke, transient ischemic attack, vascular deaths, and major vascular events, was not calculated (but noted to be low and not different between PRP and IVR groups in Protocol S), because the varied theoretical costs for these events would be beyond the predictive value of this study.^{8, 50} Thus, the true cost-utility of PDR treatment is only roughly estimated by this study. Given the differences in costs of PDR treatment in other countries, it would be difficult to apply this analysis fully and specifically in the international context. Nevertheless, this study demonstrates a more favorable cost-utility for PPV as an initial management for PDR.

The duration of effectiveness of anti-VEGF after PPV has been shown to be shortened given the reduced half-life of the drug, which would mean that more injections might be necessary in those requiring it.⁵¹ Despite the decreased duration of injections, the authors could only speculate that it would be unlikely for eyes to require injections after PPV, at least for PDR. As was performed in the previous study,³⁰ substituting intravitreal bevacizumab assuming a cost of $100/1.25-mg dose yielded a 2-year cost/QALY that was 65% to 74% less than primary treatment with IVR. Although this would decrease the cost-differential between primary treatment with anti-VEGF compared with early PPV, the lifetime cost/QALY of the PPV group would still be 60% to 65% lower.

The management of diabetic patients continues to be complex, not only because of the broad spectrum of involvement and outcomes but also because of the multiplicity of management options. Although PPV may not be indicated as initial treatment in all patients with PDR, this cost analysis demonstrates the value of considering it earlier in the course of treatment, especially compared with chronic treatment with IVR or other branded anti-VEGF agents.

Abbreviations and Acronyms:

CPT

Current Procedural Terminology

DME

diabetic macular edema

DRS

Diabetic Retinopathy Study

IVR

intravitreal ranibizumab

PDR

proliferative diabetic retinopathy

PPV

pars plana vitrectomy

PRP

panretinal photocoagulation

QALY

quality-adjusted life years

RVU

relative value unit

VEGF

vascular endothelial growth factor