Healthcare Management in Ophthalmology: The Economics of Better Care

How leaders can align AI investment, service redesign and payment with patient benefit.

Imagine approving retinal-screening technology in primary care, only to discover that nobody has budgeted for the additional referrals. The equipment works. The patient pathway does not. This hypothetical example captures a practical problem: who will fund the assessment and treatment needs that screening brings into care?

My central argument is that an investment remains incomplete until the organisations expected to deliver the resulting care have agreed how it will be resourced. Patient benefit, value for money and provider viability must be assessed separately before they can be connected.

The World Health Organization estimates that at least 2.2 billion people have near or distance vision impairment, including at least 1 billion whose impairment could have been prevented or remains unaddressed. Refractive errors and cataracts are leading causes. [1] Investment should therefore start with unmet need, not a preferred technology.

Healthcare Management in Ophthalmology: Begin with the Decision

Before considering individual studies, I would ask an investment committee to answer five questions. Their purpose is to expose assumptions while decisions can still be changed.

Each appraisal should state its perspective, time horizon, price year and key assumptions. A payment between organisations is a transfer, not automatically a saving for the wider system.

An investment committee should also see what happens when uptake is lower, referral demand higher or treatment capacity tighter than forecast. A proposal that works only under its most optimistic assumptions is not yet a reliable operating plan.

Patient value is not the same as financial return

Value-based healthcare concerns patient outcomes relative to the costs of achieving them across the care cycle; it is not simply payment linked to outcomes. [2] Cost-effectiveness asks whether additional benefit justifies additional expenditure against an alternative. Budget-impact analysis addresses the consequences for a particular budget holder. [3]

A worthwhile intervention can increase total spending. A provider may fund coordination while patients save travel time and another payer avoids later expenditure. Conversely, an operating surplus does not demonstrate better care.

For cataract services, the International Consortium for Health Outcomes Measurement (ICHOM) offers an established outcome set incorporating patient-reported visual function. [4] Comparisons should examine baseline severity, case mix and missing follow-up. Increasing appropriate treatment for underserved patients may improve value; fewer procedures are not inherently better.

Evaluate the AI service, including its safeguards

Artificial intelligence (AI) should be evaluated for a defined task and population. The US Food and Drug Administration’s original 2018 authorisation of IDx-DR covered detection of more-than-mild diabetic retinopathy in adults using specified equipment, not comprehensive eye examination. [5]

In the Johns Hopkins ACCESS trial, 164 participants aged 8–21 with diabetes were randomised. Screening completion within six months was 81/81 with point-of-care AI versus 18/82 in the analysed referral-and-education group. Among 25 participants with an abnormal AI result, 16 completed follow-up with an eye-care provider. [6]

All images also received retina-specialist safety over-reading because the system was not labelled for that age group. The comparator was later referral, not equally accessible on-site human grading. The trial supports its tested delivery arrangement; it does not establish unattended paediatric deployment or long-term visual benefit. [6]

A Stanford study evaluated an AI–human hybrid workflow, principally through diagnostic-performance and image-gradability outcomes. Its observational design and retrospective analyses do not establish lower total costs or improved long-term vision. [7] Both AI studies disclose commercial interests; those disclosures do not invalidate their findings, but deserve scrutiny alongside design and applicability. [6,7]

For procurement, I would require budgets for repeat imaging, false-positive referrals, non-diagnostic results, treatment, training and ongoing assurance. The service also needs a defined response to possible missed disease, changing performance and system downtime. Local modelling should test utilisation and costs against realistic alternatives. More detected disease can justify more expenditure, but screening success alone cannot demonstrate economic value.

Distinguish capacity from cash

Time-driven activity-based costing links activity time to resource costs per unit of practical capacity. [8] Released staff time creates a potential capacity benefit; it becomes a cash saving only when paid hours, overtime, agency expenditure or another cash outflow falls. Whether that time can be usefully redeployed also needs testing.

A Michigan analysis of 16,092 cataract operations estimated US$877.85 in additional day-of-surgery costs for complex versus simple surgery against US$231.01 in additional Medicare reimbursement. External cost estimates were inflation-adjusted to 2021, and professional fees used 2021 local rates. [9] These historical, single-centre estimates show a relative payment mismatch—not an absolute loss on every complex operation.

The management implication is to negotiate funding that recognises complexity. Costing should inform the service’s design, not become a reason to exclude patients whose care takes longer.

Use comparative evidence without overstating it

Singapore’s National University Hospital cataract programme was associated with lower recorded costs while a ten-item quality score remained high. The interrupted time-series study lacked a concurrent control hospital, and the score was already near its maximum. It therefore cannot establish causal savings or clinical equivalence across all important outcomes. [10]

ECHoES modelled community versus hospital monitoring for treated, quiescent neovascular age-related macular degeneration. Its four-week evaluation used virtual clinical vignettes, not complete real-world pathways. The estimated cost difference was small and uncertain, with its confidence interval allowing either lower or higher community costs. [11] This is not a verdict on contemporary community care; it illustrates how downstream referrals and treatment assumptions can change an apparently straightforward comparison.

The LiGHT randomised trial provides a different example: selective laser trabeculoplasty versus drops as initial treatment for treatment-naïve open-angle glaucoma or ocular hypertension. Its three-year UK evaluation supported laser-first cost-effectiveness from an ophthalmology-cost perspective. [12] That finding is specific to the intervention, population and economic setting. It also reminds leaders to consider established treatment pathways, not only new digital products.

These examples illustrate why the comparator, outcome and costing boundary matter. Their findings inform the decision questions; they do not validate them.

Make payment and access part of the same agreement

A 2024 scoping review identified 12 articles, with implementation concentrated on selected elements of value-based healthcare rather than the complete agenda. [13] My recommendation is to choose organisational arrangements around local needs, not presume that an integrated practice unit is always preferable to a coordinated network.

Funding agreements should identify responsibility for assessment, treatment, complications and continuing care. Episode-based payment may suit some cataract pathways; chronic disease may require longer arrangements. These are proposals to test, with safeguards against avoiding complex patients or providing too little care. The agreement should name who receives referrals, monitors completion and resolves capacity shortfalls.

WHO’s integrated people-centred eye-care guidance links service development to universal health coverage and financial protection. [14] Access should consequently be reviewed alongside expenditure: who waits, who misses follow-up and who cannot afford to attend? Additional screening is not a sufficient answer when treatment remains inaccessible.

Keep the investment open to challenge

Lee and colleagues’ prospective observational study of 220 improvement ideas in 12 community-health teams associated brainstorming and reflection on team process with differing implementation patterns. [15] This is evidence about team improvement, not a direct test of executive investment decisions.

I would use staged investment with explicit review points, inviting staff and patients to identify problems and recording what changed. Early operational results should not be mistaken for durable clinical benefit. A smaller launch may expose a correctable bottleneck before expansion makes it expensive to resolve.

The alternative use of the money belongs in that discussion. Additional treatment capacity, accessible spectacles or a more reliable recall service may deserve priority. The test is not whether a proposal looks innovative, but whether it addresses the most important unresolved need within the resources available.

Conclusion

Clinical ambition, affordability and equitable access sometimes require difficult trade-offs. The executive responsibility is to make those choices visible—and ensure that a funded intervention becomes a service capable of completing the care patients need.