|Study||Sample||Design||Number||Baseline mean age||Hypoglycemia||Cognitive measures||Adjustment variables||Association with cognitive function|
|Aung et al.  (2012)||Patients with type 2 diabetes participating in the Edinburgh Type 2 Diabetes Study; Scotland||Cross-sectional, observational||1,066||Mean 68 ± 4 years||Baseline self-report of history of SH (defined as episode requiring assistance)||MMSE, composite score from seven cognitive tests, and estimate of pre-morbid ability||Age, sex, duration of diabetes, anti-diabetes medication, depression, alcohol, smoking, blood pressure, HbA1c, stroke, TIA, MI, angina, and retinopathy. Analyses of estimated lifetime decline additionally adjusted for estimated pre-morbid ability.||History of SH associated with lower cognitive function and steeper estimated lifetime decline (fully adjusted analyses). Linear negative relationship between number of episodes of SH in the year before cognitive testing and cognitive function (analysis controlling for age and sex).|
|Bruce et al.  (2008)||Patients with type 2 diabetes participating in the Fremantle Diabetes Study; Australia||8-year retrospective, observational||302||Mean 76 ± 5 years||Hypoglycemia resulting in coma or hospitalization, self-reported at baseline and 8 years earlier||Dementia and MCI identified from screening instruments/clinical interview||None||Cross-sectional analysis: increased prevalence of history of hypoglycemia in groups with poorer cognitive function. No findings in prospective analyses.|
|Bruce et al.  (2008)||Patients with type 2 diabetes participating in the Fremantle Diabetes Study; Australia||8-year retrospective, 2-year prospective, observational||205||Mean 75 ± 4 years||Hypoglycemia resulting in coma or hospitalization, self-reported 8 years prior to baseline cognitive assessment||Dementia and MCI identified from screening instruments/clinical interview at baseline and at 2-year follow-up. ‘Cognitive decline’ defined as downward conversion between ‘normal’, MCI, and dementia.||None||No association|
|Bruce et al.  (2009)||Patients with type 2 diabetes participating in the Fremantle Diabetes Study; Australia||5-year prospective, observational||302||Mean 76 ± 5 years||1. Self-reported medical assistance or unconsciousness or both.||Dementia and MCI identified from screening instruments/clinical interview. ‘All cognitive impairment’ summarizes groups of dementia and MCI.||Based on preliminary associations of covariates with cognition, the following adjustment variables were selected for: analyses of MCI: age, sex, education, and cardiovascular disease; for analyses of dementia: duration of diabetes and peripheral arterial disease; for analyses of all cognitive impairment: cardiovascular disease, peripheral arterial disease, and duration of diabetes.||
Cross-sectional analyses: presence of MCI and all cognitive impairment associated with history of all three measures of hypoglycemia (fully adjusted analyses). No finding for dementia.|
Prospective analyses: no association of baseline history of hypoglycemia and risk of ‘cognitive decline’ in patients free of cognitive impairment at baseline (unadjusted). Increased risk of first-ever incident HSH in group with dementia at baseline (adjusted for insulin use, body mass index, inability to self-manage medication, and history of severe hypoglycemia). No finding on risk of subsequent hypoglycemia in patients with MCI at baseline.
2. Episodes rated by medical staff as ‘doctor-verified’.|
3. Codes for ambulance or emergency treatment for hypoglycemia in hospital records (HSH).
|‘Cognitive decline’ defined as conversion between unimpaired, MCI, and dementia.|
|de Galan et al.  (2009)||Patients with type 2 diabetes participating in ADVANCE arm on glycemic control, receiving standard target versus target HbA1c ≤6.5 %; Australia||5-year trial on effects of intensified blood pressure control and intensified glycemic control||11,140||Mean 66 ± 6 years||Incident SH defined as blood glucose <2.8 mmol/L or symptoms consistent with hypoglycemia with absence of another cause and requiring external assistance. Incident mild hypoglycemia defined as self-treated episode.||At baseline and 2-year intervals: MMSE followed by clinical interview for patients with MMSE <24 or suspected dementia. ‘Normal’ cognitive function defined as MMSE ≥28; ‘mild dysfunction’ as MMSE = 24-27; ‘severe dysfunction’ as MMSE <24. Additional use of MMSE as continuous measure.||Age, sex, treatment arm, education, duration of diabetes, blood pressure, hypertension, HbA1c, cholesterol, body mass index, macrovascular disease, microvascular disease, smoking, and alcohol||Prospective analyses (unadjusted): increased risk of SH (but not any hypoglycemia) in groups with ‘mild dysfunction’ and ‘severe dysfunction’ (versus ‘normal’ group). For ‘severe dysfunction’, but not ‘mild dysfunction’, finding survived full adjustment. Each unit-lower baseline MMSE score associated with 10 % increased risk of SH (adjusted for age, sex, education, and treatment group). Increased risk of hypoglycemia in treatment group with intensified glycemic control, but finding similar across cognitive groups. No difference in cognitive decline between treatment groups.|
|Feinkohl et al.  (2014)||Patients with type 2 diabetes participating in the Edinburgh Type 2 Diabetes Study; Scotland||4-year prospective, observational||1,066||Mean 68 ± 4 years||Self-reported history of episode requiring assistance at baseline (prevalent SH) and during follow-up (incident SH)||MMSE, composite score from seven cognitive tests, and estimate of pre-morbid ability||Age, sex, cholesterol, blood pressure, smoking, HbA1c, TIA, stroke, MI, and angina||History of SH and incident SH both associated with lower cognitive function at year 4 and with increased rate of 4-year cognitive decline. Incident SH associated with steeper estimated lifetime decline. Baseline lower cognitive function predicted increased risk of incident SH.|
|Launer et al.  (2011)||Patients with type 2 diabetes participating in ACCORD and ACCORD-MIND, with HbA1c targets of less than 6.0 % versus 7.0 % to 7.9 %; North America||40-month trial on blood pressure, lipids, and glycemic control||2,977||Mean 62 ± 6 years (ACCORD-MIND)||Increased risk of episode of hypoglycemia requiring medical assistance and of episode of hypoglycemia requiring any assistance in treatment arm with intensified glycemic control in ACCORD (n = 10,251)||Total brain volume at baseline and 40 months. Digit Symbol Coding (primary outcome), MMSE, Rey Auditory Verbal Learning, and Stroop (secondary outcomes) at baseline and 20 and 40 months.||Second trial assignment (lipid or blood pressure trials), group allocation within second trial assignment, clinical center, and history of cardiovascular disease||No difference in 20- or 40-month cognitive decline between treatment groups in ACCORD-MIND substudy of ACCORD (fully adjusted analyses). Total brain volume declined at slower rate in intensively treated compared with standard treatment group (independent of adjustment variables, and of age, sex, duration of diabetes, and Digit Symbol Coding score). Greater abnormal white matter in intensively treated compared with standard treatment group at 40 months.|
|Lin and Sheu  (2013)||Patients with diabetes (>45 years); Taiwan||
3-year retrospective ascertainment of hypoglycemia.|
Dementia ascertained in subsequent 4 years
|15,000||Mean 64 ± 10 years||ICD codes for any hypoglycemia from inpatient and outpatient medical records||ICD codes for dementia from inpatient and outpatient medical records||Age, sex, insulin use, cardiovascular disease, hypertension, ischemic heart disease, chronic kidney disease, and cholesterol||Hypoglycemia associated with increased risk of subsequent dementia diagnosis. Linear relationship of number of episodes with dementia risk.|
|Punthakee et al.  (2012)||Patients with type 2 diabetes participating in ACCORD-MIND, with HbA1c targets of <6.0 % versus 7.0 % to 7.9 %; North America||40-month trial on blood pressure, lipids and glycemic control||2,956||Mean 62 ± 6 years||
1. SH defined as self-reported <2.8 mmol/L or symptoms that resolved with use of glucose or similar.|
2. HMA defined as episode requiring medical assistance (hospitalization; care in emergency department/ by emergency personnel). 3. HAA defined as episode requiring any assistance.
|Digit Symbol Coding (primary outcome), MMSE, Rey Auditory Verbal Learning, and Stroop (secondary outcomes) at baseline and 20 and 40 months||Age, education, language of test administration, depression, second trial assignment (lipid or blood pressure trials), group allocation within second trial assignment, duration of diabetes, stroke, HbA1c, ethnicity, body mass index, peripheral neuropathy, urine albumin-to-creatinine ratio, and baseline use of insulin||Cross-sectional analyses (unadjusted): association of history of HMA with lower cognitive function at baseline. Lower baseline cognitive function predicted increased risk of first-ever HMA and HAA (but not recurrent HMA or HAA) during follow-up across intervention groups (fully adjusted analyses). Association between 20-month cognitive decline and risk of first-ever HMA in subsequent 22 months (finding restricted to group scoring in lowest tertile of Digit Symbol Coding at baseline; analysis adjusted only for second trial assignment, group allocation within second trial assignment).|
|Whitmer et al.  (2009)||Patients with type 2 diabetes; USA||18-year retrospective, observational||17,000||Mean 65 ± 7 years||ICD codes for emergency treatment or hospitalization for hypoglycemia, 1980 to 2002; additional analysis of 1980 to 1985.||ICD codes for any dementia in inpatient and outpatient medical records 2003 to 2007; additional analysis of 2005 to 2007.||Age, sex, education, ethnicity, duration of diabetes, anti-diabetes treatment, duration of insulin use, HbA1c, body mass index, hyperlipidemia, count scores of co-morbidity based on ICD codes for hypertension, cardiovascular disease, stroke, and end-stage renal disease||In 5- and 18-year analyses: hypoglycemia (versus none) 1980 to 2002 or 1980 to 1985 associated with increased risk of subsequent dementia (analyses of dementia 2003 to 2007). Association of ≥2 episodes of hypoglycemia (but not of single episode of hypoglycemia) versus none with increased risk of dementia in analysis of dementia 2005 to 2007. Findings similar for codes of any hypoglycemia and for episodes resulting in hospitalization.|
|Yaffe et al.  (2013)||Patients with diabetes participating in Health ABC; USA||12-year prospective, observational||783||Mean 74 ± 3 years||Hospital records identifying hypoglycemia as primary or secondary diagnosis related to hospitalization during follow-up.||Participants cognitively unimpaired at baseline. Identification of dementia cases on the basis of hospital records showing ICD codes for dementia as primary or secondary diagnosis related to hospitalization, or dementia medication on medication inventory during annual visit; MMSE administered at 2-year intervals.||Age, sex, education, ethnicity, diabetes at baseline, insulin use, HbA1c, APOE e4 status, baseline MMSE, MI, stroke, and hypertension||Cross-sectional analysis (unadjusted): association of history of hypoglycemia with lower cognitive function. Prospective analyses: hypoglycemia associated with increased risk of dementia (fully adjusted analysis; survived additional adjustment for slope of MMSE over time). Dementia associated with increased risk of hypoglycemia (analysis adjusted for adjustment variables, minus HbA1c and APOE e4).|