Cost-effectiveness of Different Diabetic Retinopathy Screening Modalities

Abstract

Current screening strategies aimed at detection of diabetic retinopathy (DR) historically have poor compliance, but advancements in technology can enable improved access to care. Nearly 80% of all persons with diabetes live in low- and middle-income countries (LMICs), highlighting the importance of a cost effective screening program. Establishing mechanisms to reach populations with geographic and financial barriers to access is essential to prevent visual disability. Teleretinal programs leverage technology to improve access and reduce cost. The quality of currently employed screening modalities depends on many variables including the instrument used, use of pupillary mydriasis, number of photographic fields, and the qualifications of the photographer and image interpreter. Recent telemedicine and newer technological approaches have been introduced, but data for these technologies is yet limited. We present results of a systematic review of studies evaluating cost-effectiveness of DR screening, and discuss potential relevance for LMICs.

Recent estimates indicate that globally, ~382 million people have diabetes.¹ Diabetic retinopathy (DR), the most frequently occurring microvascular complication of diabetes, affects approximately 28% of people with known diabetes and 11% of those newly diagnosed. It can affect nearly all patients with sufficient duration of the disease.^2,3 Sight threatening diabetic retinopathy can at least be delayed with good blood pressure and glycemic control.⁴ However, since the pathophysiological changes in the eye continue in the background and occur asymptomatically, actively screening persons with diabetes on a regular basis becomes necessary. Screening frequency varies by setting and guideline,^1,4 and has been reviewed previously.^5,6 Despite this knowledge, systematic implementation of diabetic retinopathy screening that reaches every person with diabetes is not common in many countries, especially low- and middle-income countries (LMICs).^4,7

Nearly 80% of all persons with diabetes live in LMICs, where primary healthcare facilities for managing diabetes and its complications are inadequate or nonexistent. In these settings, good-quality data on DR prevalence is also lacking.⁸ Even in upper-middle-income or high-income countries, healthcare disparities exist such that access to care may be limited by geography, race, culture, and/or finances.⁹ Furthermore, the low access groups have been linked to higher rates of DR.^10-12 Given that screening for DR can be expensive and logistically challenging, here we present data from a systematic review of economic studies of DR screening and discuss potential relevance for LMICs.

Methods

We systematically searched the PubMed, Embase, and Web of Science electronic databases for articles published between January 1990 and August 2015 using a combination of terms including “diabetic retinopathy,” “cost-effectiveness” or “cost-utility,” and “screening” as search terms. We excluded studies in languages other than English, narrative reviews, abstracts from scientific meetings not linked to peer reviewed publications, or studies evaluating screening intervals, as this has been evaluated in 2 recent systematic reviews.^5,6 Results were merged to identify duplicates. Full texts of relevant articles were assessed. The search yielded a total of 449 results from PubMed (130), Embase (190), and Web of Science (129). There were 352 results after removal of duplicates. The search was supplemented by reviewing the reference lists of relevant publications. Studies were grouped according to screening approach or teleophthalmology strategies.

Results

After excluding studies examining frequency intervals, we identified 18 studies assessing cost-effectiveness of different screening approaches (eg, who performs screening, opportunistic vs systematic screening), different delivery modalities (eg, clinic camera, telemedicine), and factors that influence success (Table 1).

Table 1.

Economic Studies on Diabetic Retinopathy.

Author, year, country	Population characteristics	Comparators	Screening modalities/personnel conducting screening	Screening outcomes	Economic outcomes
Screening approach
Lairson et al,43 1992, US	N = 352 diabetic patients	Primary care vs ophthalmologist	45-degree photographs (technician), direct and indirect ophthalmoscopy by ophthalmologists, and direct ophthalmoscopy by technicians with 7-field stereoscopic fundus photography	Cost per true-positive case detection	Cost was lower for the 45-degree camera with dilation ($295) vs nondilation ($378), standard examination ($390), and direct fundoscopy by a PA or NP ($794) for DR screening
Sculpher et al,44 1992, UK	N = 3423 diabetic patients	Evaluation of 13 screening options	GP, optician, hospital based camera, GP-visiting camera, combined GP and GP-visiting camera, combined optician and GP-visiting camera, selective screening modalities	Cost of different modalities, expected cost per true positive case detected	Cost savings can result with systematic screening during the same appointment as other routine health checks, compared to screening requiring additional visits
James et al,14 2000, England	N = 320, and systematic screening of 1363 diabetic patients	Systematic vs opportunistic screening	Systematic: 3-field, nonstereoscopic photography using mydriasis; opportunistic: direct ophthalmoscopy by GP, optometrists, and diabetologists	Sight-threatening eye disease	The CE was £209 and £289 for systematic and opportunistic screening, respectively, and incremental CE was £32 for each additional case; systematic screening remained more cost-effective than opportunistic screening
Facey et al,13 2002, Scotland	N = 2000 iterations through Crystal Ball	Systematic vs opportunistic screening	Conducted by optometrists, hospitals and GPs at any opportunity vs a systematic health authority program, primarily by digital camera (mydriatic and nonmydriatic screening)	Cost per QALY for the move from one screening program to another	The most cost-effective modality: combination of single staffed hospital units and mobile vans using nonmydriatic digital photography
Tu et al,45 2004, England	N = 769 optometric screen and N = 874 digital photography	Optometry vs digital photography screening	Topcon nonmydriatic model (professional medical photographer) vs slit-lamp biomicroscopy (optometrists)	Detection of sight-threatening DR	CE for optometry = total cost/true positives = £18 454/22 = £839; cost per patient screened = £25 599.30/874 = £29.29; CE for digital photography = £25 599/30 = £853; CE was poor in both models
Khan et al,26 2013, South Africa	N = 14 541, primary care, T2D	Systematic vs opportunistic screening	Mobile nonmydriatic digital camera (photographs taken by a trained technician with supervision by an ophthalmic nurse)	Cost per blindness case averted	Nonmydriatic fundus photography is cost-effective; the cost of DR screening was $22 per person; ICER was $1206 per blindness case averted
Lian et al,27 2013, Hong Kong	N = 2766 diabetic patients	Subjects randomized to free group (N = 1387) vs pay group (N = 1379)	Nonmydriatic fundus camera (optometrist); subsequently graded by optometrist and ophthalmologists	Uptake of screening and severity of DR detected	Lower screening (OR, 0.59; CI, 0.47-0.74) and a lower detection rate of DR (OR, 0.73; CI, 0.60-0.90) in the pay group
Prescott et al,34 2014, UK	N = 3170 diabetic patients	Surrogate photographic markers used to screen for ME in England and Scotland, and a hybrid and automated schemes	45° macula-centered color digital retinal photograph and OCT	Years free of moderate visual loss and QALYs; CE of the alternative grading schemes for triggering referral	The use of OCT in conjunction with photography within screening programs, for patients with surrogate markers of ME, is likely to be CE (OCT screening program: £32 vs referral to ophthalmology: £143
Kawasaki et al,46 2014, Japan	N = 50 000 hypothetical cohort	Systematic vs opportunistic screening	Incidental diagnosis, nonmydriatic 45° photograph in high risk people, annual fundus examinations; systematic screening by ophthalmologists using dilated fundus examination	Rate of detecting DR, preventing blindness, and costs of DR management	DR screening program in Japan is cost-effective compared to the no systematic screening; blindness reduction of ~16%; incremental cost of $64.6, and incremental effectiveness of 0.0054 QALYs per person screened; ICER was $11 857 per QALY
Teleophthalmology
Bjorvig et al,20 2002, Norway	N = 250, hypothetical cohort of 250 subjects. 42 diabetic patients	TO vs conventional screening	Conventional evaluation by ophthalmologist vs digital images transmitted via email	Cost comparisons depending on volume of screening	At higher workloads, telemedicine led to lower costs; at 200 patients per year, telemedicine cost $164 per patient and conventional examinations cost $243.5 per patient
Maberley et al,10 2003, Canada	N = 650, isolated communities	Specialist visit vs screening with a digital camera	Visits every 6 months by retina specialists vs photographic screening with a digital camera	Costs per sight-year saved and costs per QALY	The camera program was more cost-effective, and had the best cost-per-QALY ratio, at $15 000; the camera program would cost less than $5000 per year of vision saved if 65% or more of the population was screened
Aoki et al,17 2004, US	N = 10 000 inmates, a 40-year-old AA man as a reference case	TO vs conventional screening	Nonmydriatic retinal camera TO vs conventional evaluation by eye care provider	QALYs gained and costs generated	Average CE was $882 per QALY for TO and $947 for non-TO; in the TO strategy, 12.4% of patients reached blindness versus 20.5% in non-TO; ARR for blindness: 8.1%, NNS by TO to prevent a blindness case: 12.4%
Whited et al,19 2005, US	Large cohort from IHS, VA, and DoD data	Nonmydriatic digital TO compared with conventional screening	Clinic based ophthalmoscopy with pupil dilation vs JVN digital TO system (JVN)	Number of true positive cases of proliferative DR detected	Number of additional cases and savings with JVN: IHS: 148 cases and $525 690; VA: 96 PDR cases and $2 966 111; DoD: 165 and $129 046; JVN provides better outcomes at lower costs than clinic-based ophthalmoscopy in most scenarios
Li et al,21 2012, US	N = 611 diabetic patients	TO vs conventional screening	Nonmydriatic fundus camera vs conventional retinal examination	Prevalence of DR/cost comparison	Telemedicine-based DR screening cost less than conventional examinations ($49.95 vs $77.80)
Rachapelle et al,24 2013, India	N = 1000 hypothetical cohort, rural, 40 years, no previous screen, 25 years follow-up	TO vs no screening program	Mobile van, optometrist takes 4 dilated stereoscopic 45-degree fields digital retinal photographs with nonmydriatic camera	QALY gained from TO vs no screening, CU at different intervals	Rural TO was cost-effective ($1320 per QALY) compared with no screening; screening intervals of up to every 2 years also were cost-effective, but annual screening was not ($3183 per QALY)
Kirkizlar et al,18 2013, US	N = 900, T1D and T2D	TO vs no screening program or ophthalmologist	TO vs regular office visits and evaluation by ophthalmologist	DR, ME, blindness, and associated QALYs	TO is CE in most conditions; telemedicine screening is not CE in patients aged older than 80 years or in populations with more than 3500 patients
Phan et al,47 2014, US	N = 1793 diabetic patients	TO vs direct eye clinic visit	Topcon digital retinal cameras, nonmydriatic imaging	Cost of teleretinal screening	Teleretinal screening was associated with cost reduction to health plan payers (average cost reduction per screen of $24.38) and a decrease in eye clinic physician workload but failed to match the investment cost (53% gained back by study end)
Brady et al,48 2014, US	N = 99 (base case), N = 100 000 trials (Monte Carlo simulation)	Decision-tree analysis compared to no screening	3-field nonmydriatic fundus photography; images were transmitted to a remote expert reader	Estimation of costs of screening for PDR	TO screening for PDR resulted in savings of $36 per patient (base case), and a median of $48 in the simulation model

Abbreviations: AA, African American; ARR, absolute risk reduction; CE, cost effectiveness; CI, confidence interval; CU, cost-utility; DoD, Department of Defense; DR, diabetic retinopathy; GP, general practitioner; ICER, incremental cost-effectiveness ratio; IHS, Indian Health Service; JVN, Joslin Vision Network; ME, macular edema; NNS, number needed to screen; NP, nurse practitioner; OCT, optical coherence tomography; OR, odds ratio; PA, physician assistant; PDR, proliferative diabetic retinopathy; QALY, quality-adjusted life-year; T1D, type 1 diabetes; T2D, type 2 diabetes; TO, teleophthalmology; VA, Department of Veterans Affairs.

Systematically screening patients (through structured screening programs), at the population level is complex, but can outperform opportunistic screening, which usually covers a minority of patients with diabetes in developed countries (Table 1).^13,14 In contrast to clinical examination, telemedicine reduces the burden in the eye clinic and improves access in remote environments. Maberley et al¹⁰ reported that over 10 years, 67 versus 56 sight years were saved with telemedicine ($3900 vs $9800 per sight year and $15 000 vs $37 000 per QALY) compared to no screening. Photographer medical qualifications influence the specificity but not sensitivity of DR detection.¹⁵ The use of trained photographic graders in lieu of physicians, is especially valuable in low or middle income regions where the number of ophthalmologists per capita is often lower compared to developed countries.^13,16

Cost-effective telemedicine programs have been reported in a variety of settings, including the United States, Canada, United Kingdom, India, and Norway (Table 1).^10,17-21 Telemedicine programs are more cost-effective in people who derive more benefit. For example, populations have more benefit when screened at a younger age, using insulin, higher HbA1c, faster HbA1c change rates, or with high transportation costs.^17,18,22,23 In addition, population size and disease burden can determine the cost-effectiveness of a screening program such that screening a low number of individuals is not economically sound.^18,20,24

One common target is omission of pupillary mydriasis. This facilitates increased brevity of the screening encounter, comfort for the patient, and overall acceptance. Results from studies evaluating nonmydriatic approaches perform favorably and are cost-effective.^{10,13,19,21,25,26}

Interestingly, the need to pay for care seems to affect usage. Results from a recent randomized trial, evaluating the inverse care law in a DR screening program, showed that paying for the screening (US$8) resulted in a lower uptake of screening than being provided with free screening (OR, 0.59). Paying also resulted in a lower detection rate of DR (OR, 0.73) after adjustment for potential confounding factors. Subjects with higher income or living in better housing were more likely to be screened but less likely to have DR detected, suggesting that those in greatest need might be less able to access care.²⁷ Free systematic DR screening can be a cost-effective option if the health care system is willing to invest US$16 000 per QALY gained.²⁸

The effectiveness of DR screening intervals has been examined in recent systematic reviews.^5,6 Echouffo-Tcheugui et al concluded that in patients without DR, screening intervals could safely and effectively be extended to 2 years unless the individuals had poor glycemic control or uncontrolled hypertension.⁵ A similar systematic review by Taylor-Phillips et al found similar results but arrived to a different conclusion, suggesting that current evidence does not support a shift to screening intervals beyond 1 year, given the lack of experimental research design and heterogeneity in definition of those at low risk.⁶ Current recommendations by the American Diabetes Association suggest that if there is no evidence of DR for 1 or more evaluations, then screening every 2 years may be considered, however if DR is present subsequent examinations should be repeated annually or more frequently by an ophthalmologist or optometrist.²⁹

Advancements in technology could enable improved access to care, but data for recent telemedicine and newer technological approaches is yet limited. Alternative screening innovations such as optical coherence tomography, handheld fundoscopy, and other cell-phone-based techniques have been introduced.^30-35 The ubiquity and relative low cost of smartphones with cameras makes for an attractive platform for both image acquisition,²¹ interpretation,³³ and transmission.³⁰ Techniques using a handheld condensing lens paired with a smartphone camera can capture images at a relatively low cost.^31,32,35 Recently, Ryan et al reported a prospective comparative study of 3 modalities including: smartphone fundus photography, nonmydriatic fundus photography, and 7-field mydriatic fundus photography. The smartphone is able to detect DR and sight threatening disease, but at a lower sensitivity compared to nonmydriatic fundus photography.³⁶ The economic and clinical feasibility of newer technologies in teleophthalmology need to be further investigated.

Conclusions

Establishing mechanisms to reach populations with geographic and financial barriers to access is essential to prevent visual disability globally. Current screening strategies aimed at detection of DR have poor compliance.^37-39 Further compounding the challenge is that nearly 80% of all persons with diabetes live in LMICs.

Mydriatic 7-field photography or clinical fundus examination are considered to be the gold standard for DR screening.¹⁵ Screening modalities can vary according to instrument used (eg, film, Polaroid, scanning laser or digital photography; slit lamp, direct and indirect ophthalmoscope), mydriatic status, number of photographic fields, and qualifications of the photographer and interpreter. The sensitivity of detecting DR depends on the training of individual. In general, ophthalmic personnel outperform nonophthalmic personnel at accuracy of screening for DR.¹³ However, to improve access, teleophthalmology will likely be the cornerstone of most DR screening programs. This strategy has flaws. The need for photography depends on the equipment, which can be cost prohibitive for many systems.¹⁴ Skill is required for image acquisition and interpretation,^18,40 and action must then be taken to provide the definitive care when deemed necessary with appropriate referral to the ophthalmologist.

Mobile programs help solve the geographic access problem,⁴¹ but equipment cost remains prohibitive for routine providers and communities that are not supported by governments or foundations.

Recommendations

Cost-effective strategies and technology to provide wide coverage are necessary. This is particularly important for regions where the ratio of providers and distance to reach them is most prohibitive, and large gaps exist. To improve the status quo, several options can be considered: (1) economic viability may be improved by decreased screening frequency in low risk individuals;^5,29,42 (2) to improve geographic access, governments and health care payers may consider teleretinopathy screening programs; (3) modern nonmydriatic cameras should be considered when economically feasible; (4) to improve economic viability, more portable and less expensive equipment to detect diabetic retinopathy can be considered, recognizing the trade off in performance,^30,32,33,35 and supporting the acceleration of this research may have important economic and social benefits.

The ideal screening technology must be portable, noninvasive, reliable, and easy to use by relatively unskilled persons. Testing must be deployed in areas with sufficient volume of patients that resources spent on travel cover the cost reduction in preventing blinding disease.^18,20 The objective of screening programs is to identify individuals who will benefit from sight saving laser therapy. As such, the final obstacle to overcome for successful implementation of these screening programs is to partner with an ophthalmologist who can deliver timely laser treatment when indicated.