Ophthalmology's Injection Revenue Grew Faster Than Its Patient Base
Medicare paid ophthalmologists $2,591,256,755 for aflibercept injections in 2021. In 2013, that number was $784,651,938. The patient population receiving intravitreal injections grew 39% over roughly the same period. The drug revenue grew 230%.
Those two numbers don't fit together unless something other than patient volume is doing the work.
Drug Revenue Tripled While the Patient Count Grew Modestly
Between 2013 and 2023, the number of Medicare beneficiaries receiving intravitreal injections (billed under procedure code 67028) grew from 577,617 to 800,982, a 39% increase. Over the same period, total services grew from 2,493,548 to 3,473,739, a 39% increase as well. Patient volume and injection volume moved in near-perfect lockstep.
What didn't move in lockstep was the money. Total Medicare payments for aflibercept (J0178) went from $784,651,938 in 2013 to $2,591,256,755 in 2021, more than tripling in eight years. The drug revenue curve and the patient volume curve are not the same shape.
The administration fee for each injection (code 67028) tells a different story: payments rose from $197,199,503 in 2013 to $320,189,502 in 2023, a 62% increase that tracks reasonably close to volume growth. That's what a procedure-driven cost curve looks like. The aflibercept curve looks like something else entirely.
Injection Frequency Per Patient Has Been Essentially Flat
One obvious explanation for rising costs would be that patients are receiving more injections per year. The data doesn't support that. The average number of services per beneficiary for code 67028 was 4.32 in 2013, peaked at 4.46 in 2021, and settled at 4.34 in 2023. A range of 4.18 to 4.46 across an entire decade is, for practical purposes, flat.
This matters because it rules out a straightforward utilization story. Ophthalmologists aren't injecting existing patients more frequently. The number of distinct providers billing 67028 grew from 3,391 in 2013 to 4,224 in 2023, a 25% increase, which also doesn't come close to explaining a 230% revenue increase on the drug side.
What the data does show is that aflibercept services grew from 1,022,900 in 2013 to 3,537,196 in 2021, a 246% increase in volume for that specific drug. Aflibercept's share of the intravitreal injection market expanded dramatically, even as the total injection market grew at a much slower pace. More patients who were already getting injections shifted to aflibercept, or aflibercept captured a larger share of new patients entering the market.
The Administration Fee Tells a Different Story Than the Drug Fee
| Year | Beneficiaries (67028) | Avg Services/Bene | J0178 Medicare Payments |
|---|---|---|---|
| 2013 | 577,617 | 4.32 | $784,651,938 |
| 2015 | 649,895 | 4.18 | $1,329,942,855 |
| 2017 | 718,498 | 4.29 | $1,813,644,720 |
| 2019 | 782,156 | 4.39 | $2,171,356,124 |
| 2021 | 766,987 | 4.46 | $2,591,256,755 |
The average payment per service for the injection administration code rose from $79.21 in 2013 to $91.50 in 2021, before settling at $89.00 in 2023. That's a 12% increase over a decade, roughly in line with medical inflation. The administration side of this equation behaved like a normal Medicare fee schedule.
The drug side did not. Anti-VEGF agents like aflibercept are billed under Medicare Part B as physician-administered drugs, reimbursed at average sales price plus a percentage. When the drug's price rises, Medicare's payment rises automatically, and the volume of that drug's use determines the total exposure. A drug that captures a larger share of a growing market, while its price also rises, produces exactly the revenue curve visible in this data.
For Medicare's budget, the distinction between "more patients" and "more expensive drug per patient" is consequential. A patient-volume increase is bounded by disease prevalence. A drug-pricing and market-share dynamic has no such natural ceiling, and the decade-long trajectory of aflibercept payments in ophthalmology makes that difference visible in stark terms.
Ask a follow-up
Run one of these against the same data. The question is filled in for you.
Explore the data yourself
Run your own queries against 240M+ rows of federal health data using natural language — powered by AI.
Start analyzing