Table 2.

Characteristics and findings of cost-effectiveness studies investigating different DR screening intervals

Author (year)	Type of economic evaluation and model	Population studied	Comparators	Methods (perspective, time horizon and discount rate)	Methods (costs, outcomes, ICER and sensitivity analyses)	Results and main conclusions
Brailsford et al (2007)41	EE: CEA Model: Discrete event simulation embedded in an optimisation model using POST	Hypothetical population of 100 000 people with T2DM	Two screening policies using different strategies vs no screening: 1. optometrist funduscopy 2. diabetologist ophthalmoscopy 3. GP ophthalmoscopy 4. mobile camera 5. mydriatic 7 field photography by ophthalmologist (gold standard).Screening interval was varied between 6 months and 36 months.	Study perspective: Not stated Time horizon: 100 years Discount rate: 0%, 3% 5% Currency/price year: UK £—year not stated	Outcomes: Total number of years of sight saved Costs: Direct costs of screening and treatment, outpatient visits ICER: Incremental cost per year of sight saved Sensitivity analyses: Not stated	Most cost-effective screening policy is where the optometrist carries out both screens (policy 2) and if screen 2 is positive this is confirmed by the gold standard test. Screening should be carried out at 30 month intervals.
Chalk et al (2012)42	EE: CEA Model: Simulation model using POST	Hypothetical population of 5000 people with T2DM without DR	Annual (or 6-monthly) screening vs a 2-year screening programme	Study perspective: Not stated Time horizon: 15 years Discount rate: Not stated Currency/price year: UK £—year not stated	Outcomes: Proportion of patients with diabetes with vision loss Costs: Screening test, ophthalmology visits and laser treatment ICER: None stated Sensitivity analyses: One-way	The 2-year screening costs were £1 360 516 and annual screening costs were £1 834 060, which represents a 25.8% reduction in screening costs. A screening test every 2 years was a safe and cost-effective strategy.
Dasbach et al (1991)43	EE: CEA Model: Simulation model using a Markov process	Hypothetical groups of a 1000 patients with onset diabetes: 1. younger patients; 2. older patients taking insulin; and 3. older patients, not taking insulin	Seven screening strategies: (1) no care (2) and (3) annual or biannual visits to a community healthcare professional (4) and (5) annual or biannual non-mydriatic camera screening (6) and (7) annual or biannual mydriatic camera screening	Study perspective: Societal Time horizon: 10 years and 60 years Discount rate: 5% (varied between 0% and 10%) Currency/price year: US$ in 1989 prices	Outcomes: Sight years saved Costs: Screening and clinic visits, treatments and rehabilitation ICER: None stated Sensitivity analyses: One-way	60-year results: annual examination with mydriatic fundus photography for groups 1, 2 and 3 might save from 303 to 319, from 58 to 62 and from 19 to 21 sight years, respectively. The results suggest that screening annually compared with 6 monthly was favoured.
Davies et al (2002)44	EE: CEA Model: Discrete event simulation model using POST.	Hypothetical population of 500 000 people with T1DM or T2DM who could develop DR	Each scenario compared with no screening. Screening done by a mobile camera, diabetologist, optometrist or GP. Policy 1, screening every 12 months and a 6-month interval between visits once DR detected. Policy 2, screening every 12 months, even after the detection of background retinopathy, until treatable retinopathy is detected (every 6 months). Mydriatic seven-field photography by an ophthalmologist, screening every 6 months, with visits every 3 months after DR had been detected.	Study perspective: Not stated Time horizon: 25 years Discount rate: Not undertaken Currency/price year: UK £—year not stated	Outcomes: Average years of sight saved Costs: Screener, ophthalmology outpatient visits, treatment and mobile camera (including set-up costs). ICER: Costs per year of sight saved Sensitivity analyses: One-way	For both types of patients, the mobile camera (Policy 2) had the lowest costs at £449 200 per year and a cost per sight year saved of £2842. Policy 2 was more cost-effective than policy 1 as long as the screening sensitivity and compliance were relatively high. Results suggested there is little difference in the number of sight years saved between the different modes of screening when screening intervals are ≤1 year and compliance is high.
Javitt et al (1990)45	EE: CEA Model: Monte Carlo Simulation model using PROPHET	Hypothetical cohort of patients with T1DM	Five screening strategies all have dilated ophthalmoscopy: 1. every 2 years 2. annually 3. annually for patients with no DR and examination every 6 months for those with DR 4. full fundus photographs annually 5. full fundus photographs annually for patients with no DR and examination every 6 months for those with DR	Study perspective: Government Time horizon: Lifetime Discount rate: 5% Currency/price year: US$ in 1986 prices	Outcomes: Person years of sight saved Costs: Screening (eye examination, angiography) and treatment (laser pan retinal or focal) ICER: None stated Sensitivity analyses: One-way	All strategies resulted in cost savings. There is an economic advantage in adding semiannual visits under strategy 3. Although it was slightly less cost-saving than annual examination alone, more years of sight are saved than less frequent examination.
Javitt et al (1994)6	EE: CEA Model: Simulation model using PROPHET	Hypothetical cohort of patients with T2DM with DR	Eight screening strategies: (1) and (2) screening every 2 years. Patients with background or more advanced DR seen semiannually under strategy 1 or annually under strategy 2. (3), (4) and (5) screening every 3 years. Patients with background DR scheduled every 6 months, 12 months or 18 months, respectively (6), (7) and (8) screening every 4 years Patients with background DR scheduled every 6 months, 12 months or 24 months, respectively	Study perspective: Government Time horizon: Lifetime Discount rate: 5% (varied between 2.5% and 10%) Currency/price year: US$ in 1990 prices	Outcomes: Person years of sight saved Costs: Screening and treatment and cost of blindness ICER: None stated Sensitivity analyses: One-way	Changing the frequency of screening for patients with no or mild background DR from 1 year to 2 years has no detrimental effect on years of sight saved while reducing costs. Once patients develop moderate non-proliferative or more advanced DR, savings in sight-years are sensitive to the screening interval.
Rein et al (2011)46	EE: CUA Model: Monte Carlo simulation	Hypothetical 10 million patients with T2DM with no or early DR	Four screening methods: 1. patient self-referral following visual symptoms 2. annual eye evaluation, 3. biennial eye evaluation 4. annual telemedicine screening in primary care settings	Study perspective: Societal Time horizon: Lifetime Discount rate: 3% Currency/price year: US$ in 2010 prices	Outcomes: QALYs Costs: Intervention (including telemedicine) and treatment costs and productivity losses ICER: Cost per QALY gained Sensitivity analyses: Probabilistic	Current annual eye evaluation was costly compared with either treatment alternative. Self-referral offered the lowest costs and QALYs, followed by telemedicine, biennial evaluation and annual evaluation.
Tung et al (2008)8	EE: CEA and CUA Model: Markov-decision type model	Community-based patients with T2DM	Five screening strategies compared with no screening: 1. annual screening 2. biennial screening 3. 3-year screening 4. 4-year screening 5-year screening	Study perspective: Not stated Time horizon: 10 years Discount rate: 5% Currency/price year: New Taiwan (NT) $ in 2004 prices	Outcomes: Sight years saved and QALYs Costs: Direct costs of screening, drugs and treatment (laser photocoagulation and surgery) ICER: Cost per sight year saved and cost per QALY gained Sensitivity analyses: One-way	Annual screening should be conducted.
Vijan et al (2000)47	EE: CUA Model: Markov model.	Hypothetical T2DM patients	Four screening strategies compared with no screening: 1. annual screening 2. biennial screening 3. 3-year screening 4. 5-year screening	Study perspective: Third party payer (government and societal used in sensitivity analyses) Time horizon: Lifetime Discount rate: 3% Currency/price year: US$—year not stated	Outcomes: QALYs Costs: Screening, ophthalmology visits, laser treatment and angiogram ICER: Cost per QALY gained Sensitivity analyses: One-way and multivariate	Screening every other year maybe the most cost-effective option. with the option of tailoring screening to the needs of different individuals.

CEA, cost-effectiveness analysis; CUA, cost-utility analysis; DR, diabetic retinopathy; EE, economic evaluation; GP, General Practitioner; ICER, incremental cost-effectiveness ratio; POST, Patient Orientated Simulation Technique; PROPHET, PROspective Population Health Event Tabulation; QALY, quality-adjusted life year; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.