Table B5Studies Modeling Screening for Type 2 Diabetes (KQ1)

Author, Year
(in date order)
ObjectiveType of screening; PerspectiveType of modelPopulation;
Included costsDiscount rateBase case assumptionsTime horizonData sourcesSensitivity analysesInterventionOutcomesConclusionsQuality assessment
CDC Diabetes Cost-effectiveness Group, 199890To estimate the cost-effectiveness of early detection and treatment of DM2 compared to current practice (clinical diagnosis)One-time opportunistic screening during regular physician visit;
Single-payer health care system
Monte Carlo Computer simulationHypothetical cohort of 10,000 persons with newly-diagnosed DM2 from the general United States population >25 yDirect costs: screening, diagnostic tests, treatment3%; costs expressed in in 1995 US$Screening reduces the prediagnosis interval by 5y (from 10.5y to 5.5y); prevalence of undiagnosed DM2 is 3.2% (varied by age, sex, race per NHANES data); glycemic control relates to microvascular (but not macrovascular) complicationsLifetime or age 95yVarious epidemiologic data and treatment trials, including UKPDSA1c as screening test (decreases $/QALY), sensitivity and specificity of the screening test, prediagnosis interval (shorter interval, increased $/QALY); prevalence of DM2 (increased prevalence produces decreased $/QALY); intensive treatment for glycemic control (increases $/QALY)One-time screening intervention with FPG, OGTT for confirmation of positivesIncremental cost of screening is $236,449 per life-year gained and $56,649/QALY; more CE among younger persons (as more complication-free years and CHD not modeled) and among African AmericansScreening may produce cost/QALY within range of currently acceptable, especially for younger persons and African AmericansLimited sensitivity analyses
CVD not modeled; screening and treatment only influence microvascular complications
No information on how QALYs determined
No mention harms of screening
Lack of transparency of details of model
Used data from DM1 for microvascular disease risk reduction with treatment
Goyder et al, 200091To determine whether the potential benefits of screening are likely to outweigh the potential harms; explore which variables influence the balance of benefit and harm from screeningUniversal screening
Perspective: NA (does not involve cost)
Decision analysisCohort of 10,000, mainly Caucasian 45–60 y
United Kingdom
NA3% annual rate for QALYsPositive screening test is followed by a ‘gold standard’ diagnostic test before treatment; harms of negative or false positive test negligible; reduction in QALYs associated with early diagnosis proportional to time from diagnosis to when clinical diagnosis would have been made; optimal treatment is available from the time of clinical diagnosis; diabetes will be diagnosed at the time of or before symptomatic complications present; baseline risk of CVD complications is similar in diagnosed and undiagnosed DM2; sensitivity of screening test 90%; treatment for 1 CVD risk factor leads to a risk reduction of 1/3; extent to which BG is reduced during early treatment is 50% of that achieved after clinical diagnosis; clinical diagnosis 6y after onsetLifetimeUKPDS and other sourcesOne-way sensitivity analysis: benefits no longer outweigh harms if: baseline annual risk of CVD is <0.8%; RR CVD is reduced by <13% during earlier treatment; discount rate >7%Various interventions for hyperglycemia, HTN, lipidsQALYs gained by screening 10,000 persons: 10.5: 4 from postponed microvascular complications, 17 from avoided CVD complications and 11 lost from early diagnosisThe immediate disutility of earlier diagnosis and additional treatment may be greater than the potential long-term benefit from postponing microvascular complications; screening decisions should be based largely on CVD risk and interventions to reduce that riskUsed data from DM1 for microvascular disease risk reduction with treatment
Details and assumptions of the model not clear
Hofer et al, 200092To define the relative benefits of screening for DM2Universal and targeted screeningMarkov modelCohort of recent onset DM2 (<5y)NANAOnset of DM2 prior to diagnosis 5y; A1c increases at constant rate of 0.2%/y in diagnosed and undiagnosed; one-time drop in A1c of 10% at time of start of treatment; undiagnosed were diagnosed at rate of 5%/y up to A1c of 13%, beyond which were diagnosed at 50%/yLifetimeNHANES III, UKPDS for progression of glycemia, DCCT for benefits of tight glycemia control on ESRD and retinopathyDuration undiagnosed DM2, treatment effect, rate of case findingPerfect screening: diagnosis at time of onset
Improved treatment: A1c ≤ 9%
Number blind/1000 persons with diabetes, age 40y, A1c 12%:
Case finding: 141
Perfect screening: 133
Case finding, A1c <9%: 90
Screening, A1c <9%: 41
Screening produces 7% of the benefit of reduced number of cases of blindness; improved treatment alone is 65%

Targeted screening (with 2+ risk factors): achieved 75% of the benefits of universal screening
Largest impact of improving treatment and diagnosis is in younger person with high A1c; focus should first be on improving glycemic control of known diabetics with high A1c; if that is achieved then the benefits of screening will become more importantDoes not include benefits of HTN and lipid treatment
Only examines microvascular complications
Chen et al, 200143To evaluate the efficacy of screening for DM2 compared to no screening; to evaluate the inter-screening interval and age of start of screening on health outcomes; to examine the CE of screeningMass screening
Single payer health plan
Markov process Monte Carlo simulationOver age 30y, general community population; cohort of 30,000
Direct costs including costs of screening, treatment; indirect costs not included; costs in US$3% annual rateEarly diagnosis and treatment can control BG and reduce micro- and macrovascular complications30y or until deathTaiwan demographic data; transition parameters from a variety of sources including Framingham Heart Study, UKPDSNoneScreening program lasts for 10y; standard treatments such as that of UKPDS for persons with DM2Cumulative incidence rates of microvascular complications:
2y screening: Blindness: 3.06%; ESRD: 0.19%; LEA: 0.97%
5y screening: Blindness 3.13%; ESRD: 0.19%; LEA: 0.99%
Control (no screening): Blindness: 4.3%; ESRD: 0.54%; LEA: 1.43%
NSD between 2 and 5-y screening

Cost-effectiveness (cost/QALY): 2-y: $17,833; 5-y: $10,531
Incremental cost/QALY: lowest 40–49y group ($9,193), highest 70y+ ($36,467)
Mass screening is CE compared to opportunistic screening
Costs incurred with mass screenings are offset with life-years gained
Mass screening for DM2 is relatively CE compared to other screening interventions (e.g. cervical cancer or HTN)
Screening is more CE in younger than older patients
Lack of transparency for assumptions, data synthesis
No sensitivity analyses
Do not include CVD risk reduction in model
Do not include adverse effects of screening
Hoerger et al, 200487To estimate the incremental cost-effectiveness of two diabetes screening strategies: targeted to people with HTN and universal screeningOne-time opportunistic screening during regular physician visit
Targeted to persons with HTN
Single payer health care system
Not an economic study
Markov model with cohort simulation; is an update of the CDC model (CDC Diabetes Group 2002); considers 5 complications: nephropathy, neuropathy, retinopathy, coronary heart disease, strokeGeneral primary care population based on census United StatesDirect medical costs: screening, diagnostic tests, treatment3% annual rateIn the absence of screening, DM2 diagnosed on average 10y after onset; one-time screening makes diagnosis 5y after onset; with targeted screening only people with HTN are screened; with universal screening all persons are screened; 47% of people age 45–74 have HTN; intensive BP control adds as much benefit to DM2 as to prediabetes; RRR CHD events 51%; initially screening by capillary blood glucose with (+) followed by FPG which is repeated if (+); assume 100% sensitivity and specificity of FPG; intensive glycemic control after diagnosisLifetime; Cost/QALYUKPDS, HOT trial, US Census dataOne-way sensitivity analysis for age 55y, examining 129 critical parameters: findings were robust to treatment costs, screening costs, screening lead time, effect of HTN therapyTreatment of HTN to goal of DBP 80mm Hg (HOT); intensive glycemic control for diagnosed DM2 (UKPDS)Results per true diabetes case, compared to no screening, with intensive glycemic control and intensified HTN control after diagnosis:
Targeted screening for people with HTN only: QALYs gained per person screened (cost/QALY) ranged from 0.08 with screening at 35y ($87,096), to 0.23 for screening at 65y ($31,228)
Universal screening: QALYs gained per person screened (cost/QALY) ranged from 0.05 with screening at 35y ($126,238), to 0.11 for screening at 75y ($48,146)
Universal vs. targeted screening, incremental cost/QALY: 35y: $143,830; 75y $443,433

Universal vs targeted screening:
Relative to targeted screening, universal screening has high cost-effectiveness ratios which increase with age
Targeted screening to persons with HTN is more CE than universal screening at every age when each alternative is compared to no screening
Targeted and universal screening are more CE when take into account reduction in CHD events from earlier treatment of HTN for ages 55, 65, 75 than for 35 and 45y
The most CE approach to one-time screening: target people with HTN 55 to 75y
Benefit of screening comes mainly from reducing CHD events by control of HTN rather than from reducing microvascular complications
Did not include adverse effects of screening
Thorough sensitivity analyses
Includes submodels for CVD and stroke
Includes benefits for tight BP control, but not other CVD risk reduction interventions
Assumes 100% uptake and follow-up
Glumer et al, 200693To describe the uncertainties in estimates of the cost-effectiveness of screening for DM2 where the outcome is CHD riskNR; appears to be health care system perspectivePopulation-based simulation modelBased on community sample age 30–60y
Screening and treatment for DM2 and complications0Overall compliance rates from 30 to 75%; risk prediction for CHD events from the UKPDS; risk reductions in screened populations same as those in RCTs of various diabetes-related treatments; examine 2 extreme scenarios for assumption on how single CVD risk factor reductions combine when more than 1 factor is treated:combined therapy only as effective as most effective single agent and where risk reductions combine in a multiplicative manner5yUKPDS, Danish Inter99 study (population data), other RCTsModel not sensitive to decisions about which groups to screen nor to costs of screening or treatment; model strongly affected by assumptions about how treatments combine to reduce riskOptimal treatment of screen-detected persons; details not providedLeast conservative model (low costs and multiplicative risk reduction for combined treatments): CE ratio: 23,000 to 82,000 pounds; major contributors to uncertainty: risk reduction for hypertension treatment and UKPDS risk model interceptThere is considerable uncertainty about the cost-effectiveness of screening for DM2; the most important parameter is the effect of treatment and whether risk reductions are multiplicative or additiveModel combines effects of treatment of hyperglycemia, hypertension and dyslipidemia
Time horizon 5y
Waugh et al, 200713

Health Technology Assessment
To quantify the trade-off between the costs and benefits of screening and early treatmentPopulation screening

National Health Service
Markov modelUnited Kingdom general population 40–70 yScreening and treatment for DM2 and complications3.5% for costs and benefitsOnset of DM2 A1c is 5.9%; preclinical phase 11y; prevalence of undiagnosed DM2 1.4 to 4.4%; 14% CHD risk reduction per 1% fall in A1c (per UKPDS); prevalence of diagnosed CVD negligible (would have been screened)40yUKPDS CVD risk engine; other sourcesRate of A1c progression, risk reduction with glycemic control; various treatment regimes; costsScreening with A1c followed by OGTT if A1c > 5.7%
Various interventions for hyperglycemia, HTN, lipids
Cost reduction and QALYs gained from fewer CVD events, largely from statin treatment, as well as fewer microvascular complications

Incremental cost per QALY £2,266 for base case (40–70y)
CE greatest for 60–69y: cost per QALY £1,152
Screening is relatively cost effective for persons 40–70y of age; more cost-effective for the older group and for persons with hypertension or obesityIncludes macro and microvascular complications; relatively simple model

Abbreviations: BG, blood glucose; BP, blood pressure; CDC, Center for Disease Control; CE, cost effectiveness; CHD, coronary heart disease; CVD, cardiovascular disease; DBP, diastolic blood pressure; DCCT, Diabetes Control and Complications Trial; DM1, type 1 diabetes; DM2, type 2 diabetes; ESRD, end-stage renal disease; FPG, fasting plasma glucose; HOT, Hypertension Optimal Trial; HTN, hypertension; LEA, lower extremity amputation; NA, Not applicable; NHANES, National Health and Nutrition Examination Survey; NR, not reported; NSD, no significant difference; OGTT, oral glucose tolerance test; QALY, quality-adjusted life year; RCT, randomized controlled trial; RR, relative risk; RRR, relative risk reduction; UKPDS, United Kingdom Prospective Diabetes Study; US, United States; y, year.

From: Appendix B Evidence Tables

Cover of Screening for Type 2 Diabetes Mellitus: Update of 2003 Systematic Evidence Review for the U.S. Preventive Services Task Force
Screening for Type 2 Diabetes Mellitus: Update of 2003 Systematic Evidence Review for the U.S. Preventive Services Task Force [Internet].
Evidence Syntheses, No. 61.
Norris SL, Kansagara D, Bougatsos C, et al.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.