The Diabetic Retinopathy Clinical Research Network Effect of Diabetes Education During Retinal Ophthalmology Visits on Diabetes Control (Protocol M) 11

Rationale for Diabetes Education During Retina Examination 2  Ocular complications often can be observed before substantial vision loss and may be a highly motivating finding for an individual to be vigilant about glycemic control.  Educational intervention at the ophthalmology office may have additional impact beyond the current standard of diabetes education at other medical offices.

3  Primary Objective Determine if glycemic control in individuals with type 1 or type 2 diabetes can be improved with subject-specific diabetes education and risk assessment during retina examinations  Compare mean change in HbA1c from baseline to 1 year in the intervention vs. control groups; within 2 cohorts 1.More Frequent Cohort- seen for standard care more frequently than every 12 months 2.Annual Cohort- for standard care every 12 months Protocol M: Primary Objective

Study Design 4 Participants meeting all of the following criteria*: At least 18 years old with Type 1 or type 2 diabetes Cluster Randomized Trial (N = 1875) Primary Outcome: Mean Change in HbA1c from baseline to 1 year † Primary Outcome: Mean Change in HbA1c from baseline to 1 year † *Only participants with a central laboratory measured HbA1c ≥ 6.0% will be included in the primary analysis †Parallel analyses were performed on all outcomes for the data collected at 3 months and 2 years. Results were similar to 1 year and are not included. Secondary Outcomes: Diabetes care knowledge assessments PAID and SCI-2, BMI, and BP Secondary Outcomes: Diabetes care knowledge assessments PAID and SCI-2, BMI, and BP Control Intervention Control Intervention Annual More Frequent More Frequent

 42 participating sites  34 randomized by site (Cluster Unit = Site)  8 randomized by investigator (Cluster Unit = subset of investigators at site)  50 Clusters  25 Assigned to Intervention  25 Assigned to Control  Recruitment Goal  900 in each cohort  Participants were included in the more frequent cohort if at least one retinal visit occurred between baseline and 1 year, otherwise they were included in the annual cohort. 5 Randomization

Enrollment and Follow-up 6 More Frequent N = 990 More Frequent N = 990 Control N = 502 Control N = 502 Intervention N = 488 Intervention N = 488 Dropped = 62 Deaths = 6 Dropped = 62 Deaths = 6 Completed = 434 Completed = 434 Dropped = 62 Deaths = 6 Dropped = 62 Deaths = 6 Completed = 434 Completed = Month Visit

Enrollment and Follow-up 7 Annual N = 756 Annual N = 756 Control N = 368 Control N = 368 Intervention N = 388 Intervention N = 388 Dropped = 59 Deaths = 4 Dropped = 59 Deaths = 4 Completed = 300 Completed = 300 Dropped = 57 Deaths = 11 Dropped = 57 Deaths = 11 Completed = 320 Completed = Month Visit

Treatment 8  Usual Care  HbA1c- immediate feedback  Personalized risk assessment including retinopathy level, HbA2c, and BP  Feedback to PCP  Supplemental diabetes management materials for home use  Testing with supplemental education by investigator as needed  follow-up with link to view personalized risk assessment reports

Study Sites 9

Results 10

Participant Baseline Characteristics 11 More FrequentAnnual ControlInterventionControlIntervention Gender: Women50%55%58% Median Age Race/Ethnicity: White 63%58%63%55% Type 1 Diabetes12%11%16%13% Type 2 Diabetes88%86%82%84% Mean HbA1c

Ocular Baseline Characteristics 12 More FrequentAnnual ControlInterventionControlIntervention Visual Acuity 20/20 or better (in the better seeing eye) 32%31%51%52% DME on clinical exam (worse of 2 eyes) 40%34%5%

Patient Knowledge of HbA1c (N = 631)* 13 *Among those who had a last known date within the last 6 months

In Office vs Central Lab HbA1c (N = 867) 14

More Frequent Cohort 15

More Frequent Cohort Number of Follow Up Retina Visits Prior to 12 Month Visit 16 Control † N = 424 Intervention † N = or more visits*42%32% Median # of visits*23 *Case report forms captures this information for clinical center visits at the DRCR.net site that enrolled the participant. † Includes only participants who completed the 12 month primary outcome visit and had a non-missing central lab value (i.e., participants in the primary analysis

More Frequent Cohort Number of Follow Up Interventions Prior to 12 Month Visit 17 # Follow up interventions* Intervention N = % 142% 226% 311% *Based on follow up visits where in office HbA1c was recorded.

More Frequent Cohort 12 Month Primary Analysis 18 Central Lab HbA1c Control N = 424 Intervention N = 423 Mean at Baseline Mean at 12M Mean change from baseline to 12M Mean difference (Cont-Int)* % Confidence Interval*-0.29, P-value*0.35 *From a repeated measures model adjusting for baseline HbA1c and fitting a correlation between patients within site, assuming the correlation is the same for any pair of patients at a given site, and the within site correlation is the same at all sites. P-value from adjusted model (race, education level, income) = 0.29

More Frequent Cohort Primary Analysis of HbA1c at 1 Year 19 Control N = 424 Intervention N = 423 % with HbA1c < 7.0% 23% % with HbA1c > 10.0% 11%14% % with relative decrease in HbA1c ≥ 10% 21%25% % with absolute decrease in HbA1c ≥ 0.5% 31%36% % with absolute increase in HbA1c ≥ 0.5% 26%

More Frequent Cohort 12 Month Secondary Outcomes 20 Control N = 430 Intervention N = 418 Mean Change in Mean Arterial BP (mmHg) Mean Change in Body Mass Index (kg/m 2 )

Survey Data at 1 Year 21 Control N = 432 Intervention N = 422 PAID Mean Change from baseline to 1 year Estimated difference in change, intervention minus control (P Value) +0.5 (0.72) ACI-2 Mean Change from baseline to 1 year Estimated difference in change, intervention minus control (P Value) +1.0 (0.43)

Annual Cohort Limited to participants with a baseline central lab HbA1c ≥6.0% 22

Annual Cohort 12 Month Primary Analysis 23 Central Lab HbA1c Control N = 289 Intervention N = 297 Mean at Baseline Mean at 12M Mean change from baseline to 12M Mean difference (Cont-Int)* % Confidence Interval*-0.18, P-value*0.68 *From a repeated measures model adjusting for baseline HbA1c and fitting a correlation between patients within site, assuming the correlation is the same for any pair of patients at a given site, and the within site correlation is the same at all sites. P-value from adjusted model (race, education level, income) = 0.52

Annual Cohort 12 Month Secondary Outcomes 24 Control N = 321 Intervention N = 340 Mean Change in Mean Arterial BP (mmHg) 0 Mean Change in Body Mass Index (kg/m 2 )

Survey Data at 1 Year 25 Control N = 432 Intervention N = 422 PAID Mean Change from baseline to 1 year Estimated difference in change, intervention minus control (P Value) +0.9 (0.60) ACI-2 Mean Change from baseline to 1 year Estimated difference in change, intervention minus control (P Value) +1.2 (0.28)

Sensitivity Analysis  Results were similar when annual cohort was pooled with more frequent cohort.  Sensitivity analyses were consistent with the primary analysis results, including using all available measurements (POC or PC) when central lab value not available excluding N = 48 (3%) cases where abnormal hemoglobin variant observed imputing for missing 1-year data evaluating separately by cluster type (site-level or investigator-level) 26

Sensitivity Analysis  Results were also consistent when stratified by predefined baseline subgroups of interest A1c, last known A1c, annual income, diabetes type, DR severity, VA, and DME  An interaction between site and treatment group was not identified 27

28 Discussion

 Individualized assessment of diabetes complication risks during ophthalmology visits did not alter glycemic control, compared with usual care.  The lack of intervention effect in this study could reflect the standard care given by this specialized investigator group which is highly attuned to evidence-based retinal care for individuals with diabetes, and possibly already providing patient education at a level where the prescribed intervention would not add incremental benefit. 29 Discussion

 The similarity of A1c between intervention and control groups at one year emphasizes the challenge and need for additional approaches other than those employed in this study, including : 1.Personalized retinal imaging 2.Displaying images of severe diabetic complications 3.Exposure to individuals who have experienced these debilitating complications 4.More frequent educational interaction 5.Additional communication with primary diabetes care providers. 30

Conclusion  The addition of personalized education and risk assessment during ophthalmology visits in this study did not improve glycemic control,  long-term optimization of glycemic control is still clearly a cornerstone of diabetes care.  These results suggest that optimizing glycemic control remains a substantive challenge requiring more extensive interventional paradigms than examined in this study and further research into new technologies and models of behavioral change. 31

Conclusion  Until more successful approaches are found, ophthalmologists and all other diabetes care providers should continue their efforts to maximize education, assessment, systemic control and treatment of complications for patients with diabetes. 32