Population base eye screening in Scotland. Challenges of equity and coverage Graham Leese University of Dundee Edinburgh July 16 th 2018
NUMBER OF PATIENTS RECEIVING LASER IN TAYSIDE 300 250 200 Number of patients receiving laser 62% reduction 150 100 50 Number of patients with diabetes (x100) 60% increase 0 2001 2002 2003 2004 2005 2006 Vallance et al Diab Care 2008
PERCENTAGE OF PATIENTS RECEIVING LASER 2.5 2 % of patients receiving laser 1.5 1 2.5 fold reduction for both 0.5 0 % of patients receiving incident laser 2001 2002 2003 2004 2005 2006 Vallance et al Diab Care 2008
Prevalence of Blindness in Scotland due to Diabetes Rate per 10,000 80 70 60 50 40 30 20 10 0 2008 2009 2010 2011 2014 2015 2016 Scottish Diabetes Survey Figures
A comparison of the causes of blindness certifications in England and Wales in working age adults (16 64 years), 1999 2000 with 2009 2010 Liew et al. BMJ Open. 2014; 4(2):
Visual Outcomes One episode of missing eye screening: 3.1x increased risk of laser Leese et al Diab Care 2008 From 1990-1995 16/17 Diabetes related blindness was due to poor attendance Rhatigan et al Eye 1999 From 1990-1999 the majority of blindness due to diabetes related to poor attendance Cormack et al BJO 2001
Factors associated with nonattendance at retinal screening BACKGROUND Tayside: community screening from 1990 (digital 2000) and comprehensive from 2002 Study years 2004-2006 15,150 patients, 32,621 screening episodes Average age 63 years Mean diabetes duration 7.3yrs 54% male 12% DNA rate at screening Leese et al Diab Care 2008
Clinical Risk Factors Associated with Nonattendance Young age Long diabetes duration High HbA1c High BP Smoker
Geography and Non Attendance GIS: looking at distance and time Average distance to screening 3.3 miles Average time 11.7 min (0-87.2min) Distance or Time NOT associated with attendance
Deprivation as a risk of Non-Attendance 2.5 * * p<0.01 2 Relative Risk * 1.5 1 0.5 0 1 2 3 4 5 SIMD Deprivation Category Leese et al 2008
Reducing DNA in General Practice Patients given code to record, made to make verbal or written reminders. Poster with frequency of attendees (not DNAs). 5 0-5 -10-15 -20-25 -30-35 Code Verbal Written Verbal, Written, Poster Change in rate of DNA Martin et al JRSoc Med 2012
Are we looking in the right direction for solutions?...
Summary: What might help? Telephone and Text reminders Integrate with other diabetes appointments Patient to give verbal or written confirmation Evening appointments Opportunistic (IP/ OP/ Transition) Local Provision (Mobile unit/community Hall) Help with Language barriers
Prevalence of diabetes in Scotland 000s 350 300 4.1% 5.4% 250 200 150 100 50 0 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Are there low risk patients who could be screened less often without harm?
% Annual Progression to Referable Retinopathy TYPE-2 DIABETES No baseline retinopathy UKPDS newly diagnosed (n=2316), LES all comers (n=9890) Other studies all: No retinopathy at baseline (Sweden HbA1c 46mmol/mol) Sweden n=1322. Wales n=57,199. England n=16,444. Scotland n=101,539
Progression to Referable Retinopathy TYPE-1 DIABETES LES all comers (n=501). Scotland no baseline retinopathy (n=7869)
Risk Groups Risk level First screen Second screen 1 R1 both eyes R1 both eyes 2 R1 one eye R1 both eyes 3 R0 both eyes R1 both eyes 4 R1 both eyes R1 one eye 5 R1 one eye R1 one eye 6 R0 both eyes R1 one eye 7 R1 both eyes R0 both eyes 8 R1 one eye R0 both eyes 9 R0 both eyes R0 both eyes Stratton et al 2013 19
Time to STDR from background or no DR e.g. Programme 1 Incidence of STDR yield 5 % 1 highest risk group Note: step rises reflects annual screening 2.5 % 9 lowest risk group 1 % Stratton et al 2013 20
DRS National Scotland Data Type-2 diabetes 2y 2y 1y 1y Looker et al 2014
DRS National Scotland Data Type-1 diabetes 2y 2y 1y 1y Looker et al 2014
UK Four Nations Retinal Screening Research Group Rates of Progression (%) in 354,549 patients from 7 UK centres No Baseline retinopathy BR in one eye REFERABLE RETINOPATHY PROLIFERATIVE RETINOPATHY Leese GP, Stratton IM et al Diab Care 2015
2015 ADA: March 2017 (Solomon et al Diab Care 2017)
Screening Interval Determined by Risk Two baselines with no retinopathy - exclude people with previous background reverting to normal - exclude people who do not attend regularly ie absence of baseline May be possible to further refine risk: - Type of diabetes (T2?) - Duration of diabetes - HbA1c - Blood Pressure - Proteomics and Genetics??
Hazards Ratio for Risk Factors - HTA SBP 1.03 DBP 0.95 Sight threatening Retinopathy Maculopathy T2 vs T1* 0.45 (0.38-0.58) 0.54 0.39 Smoking 0.99 Female 1.10 HbA1c (per 10mmol) 1.31 (1.26-1.36) 1.29 1.47 Diabetes Duration 1.2 (1.17-1.24) 1.18 1.26 BMI 0.99 Creatinine 1.02 Cholesterol 1.14 (1.07-1.22) 1.16 1.15 Urine albumin 1.01 (1-1.02) 1.01 1.01 *T1 vs T2 (Duration 17 vs 2 yr; HbA1c 64 vs 50 mmol/mol) Pre-/Proliferative disease
Cox-proportional Hazards - HTA Criterion Sight threatening Retinopathy Background retinopathy (Both eyes) 7.1 (5.8-8.7) Background retinopathy (One eye) 2.6 (2-3.2) HbA1c (per 10mmol/mol) 1.28 (1.23-1.34) Diabetes Duration (per 5years) 1.20 (1.16-1.24) Cholesterol (per 1mmol/l) 1.12 (1.05-1.19) Creatinine (per 10umol/l) 1.04 (1.01-1.07)
Individualized Clinical Risk Score Baseline Retinopathy Glycaemic control Duration of diabetes Type of Diabetes Renal function Cholesterol Blood pressure Age?
Risk of transition to Referable retinopathy from second year of screening, using the variables: DR grades age Sex diabetes duration BMI HbA1c Blood pressure Blood lipids egfr visual acuity smoking status statin drug use hypertensive drug use CVD status SDRN epi group 2018
T2D Predictive Models Model AUROC and log likelihood on the test data Model AUROC LogLikelihood (1) Grade only 0.803-42735.4 2) Grade + Age + Sex + Duration 0.812-42697.2 (3) Full Model 0.824-42411.0 (4) Backward Selection 0.824-42410.1 (5) Grade + Age + Sex + Duration + HbA1c + Tchol (6) Grade + Duration + HbA1c + systolic blood pressure (replication of Aspelund et al) (7) Grade + HbA1c (replication of DCCT/EDIC) 0.823-42432.8 0.756-44827.1 0.797-43165.8
T2D Screening Policy Number of patients screened per year Current Policy Observed Risk of transition to referable DR in the current system(%) Number of Screenings per year required in the current system Prior DR grade Current Screening Intervals None 173477 12 0.49 173477 Mild 44530 12 3.76 44530 Moderate 2267 6 17.3 4533 Overall 1.6 222540 Personalised Interval Prior DR grade Accepted Risk Threshold for transition to referable DR Number of Screening Ratio Screenings required Current:Personalised per year policy None 1.6 59130 4.4:1 Mild 1.6 59294 0.8:1 Moderate 1.6 672 6.7:1 Overall 1.6 119096 1.9:1 Prior DR grade Stratum specific/averaged Interval Accepted Risk Threshold for Number of transition to Screenings required referable DR per year Screening Ratio Current:Stratum specific policy None 1.6 59478 2.9:1 Mild 1.6 133590 0.3:1 Moderate 1.6 27200 0.2:1 Overall 1.6 220268 1:1
Impact of Increased Screening Intervals on Attendance One episode of missing eye screening: 3.1x increased risk of laser Leese et al 2008
AUTOMATED GRADING
Automated Grading EQA results Q3 2012 Centres Autograder Sensitivity 88.6-95.5% 95.5% Specificity 87.0-97.8% 34.8% Supports data from Philip S et al. Br J Ophthalmology 2007;91:1512-1517
Autograder Grading Level 1 grader No abnormality Level 2 Mild to moderate retinopathy Rescreen 6 or 12 months Level 3 Refer ophthalmology Internal quality assurance
External Quality Assurance Source: Scottish DRS EQA System, Dr K.A. Goatman, University of Aberdeen
Maculopathy and Macular Oedema
Intra-vitreal VEGF therapy Ranibizumab (Lucentis): Licenced Aflibercet (Eylea): Licenced Bevacizumab (Avastin): Cheap
Use of Optical Coherence Tomography (OCT)
Macular Disease 80% of referrals to Ophthalmology are for maculopathy 80%+ of these do not require treatment (at time of referral) Therefore at least 65% of referrals unnecessary Eye Clinics overloaded
Retinal Screening Retinal Screening M2 result M2 Result 100% OCT screening 20% Eye Clinic Eye Clinic Incorporation of OCT step within screening programme
Double Benefit a) Fewer referrals b) Can discharge more from Eye Clinic
The Retina a Source of Imaging Biomarkers for Global Vascular Health
Vascular Assessment and Measurement Platform for Images of the Retina Tom McGillivray Dundee Prof Trucco Dundee Machine learning supported measurement of retinal microvascular parameters CRAE CRVE AVR Arterial Tortuosity Venular Tortuosity Optic Disc Radius Fractal Dimensions etc Calibre Tortuosity Fractal
Retinal Risk Score = Increased Venular Tortuosity Plus Reduced Venular Fractal Dimension
Improved Technology and Screening Automated Grading Efficiency Integrated OCT screening 2 yearly screening for low risk patients - individualised screening intervals
Improved Patient Factors Text reminders More convenient and novel screening opportunities Reaching out to socially disadvantaged Greater attention (rather than less) to those who do not attend
V&A Dundee THANK YOU FOR LISTENING
Use of dilating drops % not requiring dilating drops 85+ 75-84 65-74 55-64 45-54 35-44 % not requiring dilating drops 25-34 15-24 12-14 0-11 0 10 20 30 40 50 60 70 80 90 100