Table 1.

Summary of five studies included in the meta-analysis

First author, year (country)	Data source	Sample size (met/total)	Study design	Inclusion period	Age (year)	Diagnostic criteria	Adjusted effect sizes (95% CI)	Adjustment variables
Blitzer et al. 2021 (USA) [34]	In people with diabetes
	IBM MarketScan commercial and medicare supplemental databases	76,968/160,759	Retrospective	January 2006–December 2017	≥ 55	ICD-9 and ICD-10	0.95 (0.93–0.97)	Age, sex, region, hypertension, hyperlipidaemia, obesity, CCI score, diabetes, smoking, diabetic medications, statins, and DR
	In people with and without diabetes
	IBM MarketScan commercial and medicare supplemental databases	80,811/624,780	Retrospective	January 2006–December 2017	≥ 55	ICD-9 and ICD-10	0.94 (0.92–0.96)	Age, sex, region, hypertension, hyperlipidaemia, obesity, CCI score, smoking, diabetic medications, statins, and DR
Stewart et al. 2020 (USA) [33]	University of California San Francisco electronic medical record of diabetes patients	1807/3120	Retrospective	April 2012–August 2019	≥ 60	ICD-9 and/or ICD-10	0.70 (0.55–0.88)	Age, smoking status, metabolic syndrome, and socioeconomic status (insurance status, gender, and race)
Brown et al. 2019 (USA) [32]	In people with diabetes
	Integrated data repository from University of Florida health clinic	660/4947	Retrospective	June 2011–June 2017	≥ 55	ICD-9	0.70 (0.49–0.98)	Age, sex, race, insurance status, BMI, CCI score, hypertension, ocular comorbidities, and medications used
	In people with and without diabetes
	Integrated data repository from University of Florida health clinic	695/7788	Retrospective	June 2011–June 2017	≥ 55	ICD-9	0.58 (0.43–0.79)	Age, sex, race, insurance status, BMI, CCI score, diabetes, hypertension, ocular comorbidities, and medications used
Chen et al. 2019 (Taiwan) [35]	Taiwan National Health Insurance Research Database of newly diagnosed T2DM patient	45,524/68,205	Retrospective	January 2001–31 December 2013	≥ 43	ICD-9	0.54 (0.50–0.58)	Age, sex, hypertension, hyperlipidaemia, coronary artery disease, obesity, DR, chronic kidney disease, and insulin treatment
Lee et al. 2019 (South Korea) [36]	Korean National Health Insurance Service Database Record of patients with diabetes or cardiovascular diseases	1173/25,608	Retrospective	January 2012–December 2015	≥ 65	ICD-10	1.15 (0.91–1.45)	Income level, CCI, number of prescriptions, cerebrovascular disease history, complicated or uncomplicated diabetes, hyperlipidaemia, hypertension and peripheral vascular disease, and medications used. Cases and controls are matched by age, sex, cohort entry date, and follow-up duration

BMI body mass index, CCI Charlson Comorbidity Index, CI confidence interval, DR diabetic retinopathy, ICD International Classification of Diseases, T2DM type 2 diabetes mellitus, USA United States of America