Introduction
Rationale
Objectives
Methods
Search Strategy and Selection Criteria
Study Selection
Study characteristics | Effect sizes (MD = BL age minus ML age, 95%: CI) | ||||||||
---|---|---|---|---|---|---|---|---|---|
Study | N (% of females) | Mean age of diagnosis & symptom onset | Education level | Language measure | MCI diagnosis and type | MMSE scores | Age of diagnosis | Age of onset | Cognitive impairment |
Bialystok et al., 2014 | ML: 28 (50%); BL: 26 (56%) | ML: 66.5 (12.3) BL: 70.0 (10.7) | ML: 15.5 (3.8) BL: 14.3 | LSBQ | MCI | ML: 29 (1.4) BL: 28.4 (1.9) | 3.5 (−1.77–8.77) | 4.7 (0.97–10.37) | 0.6 (−0.17–1.37) |
Ossher et al. 2012 | ML (SDaMCI): 49 (55%) BL (SDaMCI): 19 (32%) ML (MDaMCI): 22 (45%) BL (MDaMCI): 21 (43%) | ML (SDaMCI): 74.9 (6.9) BL (SDaMCI): 79.4 (6.3) ML (MDaMCI): 75.2 (8.5) BL (MDaMCI): 72.6 (7.2) | ML (SDaMCI): 14.7 (2.5) BL (SDaMCI): 14.5 (3.9) ML (MDaMCI): 14.9 (3.3) BL (MDaMCI): 15.0 (3.3) | Questionnaire | Clinical interview including neuropsychological tests; SDaMCI, MDaMCI | ML (SDaMCI): 27.7 (1.6) BL (SDaMCI): 27.6 (1.9) ML (MDaMCI): 27.9 (1.4) BL (MDaMCI): 27.7 (1.8) | SDaMCI: 4.50 (0.93–8.07) MDaMCI: −2.60 (−7.32–2.12) | NA | SDaMCI: −0.1 MDaMCI: −0.2 |
Ramakrishnan et al., 2017 | ML: 22 (18.2%) BL: 93 (20.4%) | ML: 58.1 (11.4); 55.8 (12.2) BL: 65.2 (9.9); 63.2 (10.1) | ML: 10.4 (3.7) BL: 15.5 (3.3) | NA | Clinicians used Petersen criteria for final diagnosis; MCI: amnestic MCI & non-amnestic MCI | NA | 7.1 (2.36–11.84) | 7.4 (2.46–12.34) | NA |
Study characteristics | Effect sizes (MD = BL age minus ML age, 95%: CI) | ||||||||
---|---|---|---|---|---|---|---|---|---|
Study | N (% of females) | Mean age of diagnosis & symptom onset | Education level | Language measure | Dementia diagnosis and type | MMSE/3MSE scores | Age of diagnosis | Age of onset | Cognitive impairment |
Alladi et al., 2017 | ML: 72 (52.8%) BL: 121 (36.4%) | Dementia ML: 61.0 (9.5), 58.4 (9.3) BL: 64.2 (9.4), 61.7 (9.1) | ML: 6.9 (5.3) BL: 13.9 (4.3) | Case records | MMSE, ACE-R, FrSBe; bvFTD, PNFA, SD, FTD-MND, CBD, PSP | ML: 15.9 (10.3) BL: 18.1 (10.2) | FTD 3.2 (0.43–5.97) | FTD 3.3 (0.61–5.99) | FTD 2.2 (−0.81–5.21) |
Alladi et al., 2013 | ML: 257 (49.0%) BL: 391(25.1%) | Dementia ML: 63.4 (11.4), 61.1 (11.4) BL: 68.1 (10.0), 65.6 (10.0) AD (onset) ML: 65.4 (10.0) BL: 68.6 (9.6) | ML: 5.9 (5.1) BL: 12.9 (4.9) | Family member interview | DSM-IV; AD, VaD, mixed AD with CVD, FTD, DLB | ML: 16.7 (7.5) BL: 18.9 (8.0) | Dementia 4.7 (3.03–6.37) AD 3.2 (0.67–5.73) | Dementia 4.5 (2.83–6.17) AD NA | Dementia NA AD 2.2 (0.97–3.43) |
Bialystok et al., 2014 | ML: 35 (54%) BL: 40 (55%) | ML: 74.2 (11.2), 70.9 (11.0) BL: 81.4 (8.4), 78.2 (8.9) | ML: 12.5 (3.7) BL: 12.2 (4.9) | LSBQ | NA; Probable AD | ML: 23.4 (3.8) BL: 22.3 (4.5) | 7.2 (2.68–11.72) | 7.3 (2.72–11.88) | 1.1 (−0.83–3.03) |
Bialystok 2007 | ML: 91 (53%) BL: 93 (59%) | Dementia ML: 75.4 (9.3); 71.4 (9.6) BL: 78.6 (8.4); 75.5 (8.5) AD ML: 75.8 (9.8) BL: 79.2 (8.7) | ML: 12.4 (3.8) BL: 10.8 (4.2) | Medical records | NINCDS–ADRDA; AD, possible AD, dementia due to other neurodegenerative disorders, and CVD | ML: 21.3 (6.4) BL: 20.1 (7.1) | Dementia 3.2 (0.62–5.78) AD 3.3 (0.241–6.505) | 4.1 (1.4–6.74) | AD 1.2 (−3.24–5.64) |
Chertkow et al., 2010 | ML: 379 (63%) BL: 253 (51%) | ML: 76.7 (7.8) BL: 77.6 (7.2) | ML: 10.9 (3.5) BL: 10.7 (3.8) | Patient & caregiver interviews | NINCDS-ADRDA; probable AD | ML: 23.1 (3.9) BL: 22.9 (4.3) | 0.9 (− 0.31–2.11) | NA | 0.2 (−0.45–0.85) |
Clare et al., 2016 | ML: 49 (45%) BL: 37 (57%) | ML: 76.2 (8.8), 73.7 (9.9) BL: 79.3 (6.8), 76.9 (7.1) | ML: 12.31 (3.04) BL: 11.84 (2.46) | LQ-SV | ICD-10; AD | ML: 23.90 (3.19) BL: 22.68 (3.16) | 1.22 (−0.16–2.60) | 3.21 (−0.65–7.07) | 1.94 (−1.33–5.21) |
Craik, Bialystok, & Freedman 2010 | ML: 109 (55%) BL: 102 (59%) | ML: 76.5 (10), 72.6 (10.0) BL: 80.8 (7.7), 77.7 (7.9) | ML: 12.6 (4.1) BL: 10.6 (5.1) | NA | NINCDS-ADRDA; probable AD | ML: 21.5 (5.7) BL: 20.4 (5.6) | 4.3 (1.87–6.63) | 5.1 (2.64–7.56) | 1.1 (−0.43–2.63) |
Lawton et al., 2015 | ML: 54 (65%) BL: 27 (63%) | ML: 81.10 (NA) BL: 79.31 (NA) | ML: 4.99 (4.17) BL: 7.70 (4.88) | ARSMA-II | ADDTC NINCDS–ADRDA; VaD Possible and probable AD | ML: 78.87 (9.90) BL: 79.56 (15.57) | 1.79 (−4.55–0.97) | NA | 0.69 (−4.97–6.35) |
Perani er al., 2017 | ML: 40 (52.5%) BL: 45 (71%) | ML: 71.4 (4.9) BL: 77.1 (4.5) | ML: 10.5 (4.07) BL: 8.26 (4.55) | Questionnaire | NIAAA; probable AD | ML: 21.10 (4.84) BL: 22.40 (4.19) | 5.70 (3.71–7.71) | NA | 1.3 (−0.65–3.25) |
Schweizer et al., 2012 | ML: 19 (70%) BL: 20 (70%) | ML: 77.3 (6.8) BL: 78.9 (7.7) | ML: 13.6 (3.5) BL: 11.6 (4.5) | Interview with patient and significant-other | CDR; probable AD | ML: 23.2 (3) BL: 22.1 (5.1) | 1.60 (−2.95–6.15) | NA | −1.10 (−3.87–1.67) |
Woumans et al., 2015 | ML: 69 (69%) BL: 65 (69%) | ML: 72.5 (9.4) BL: 77.3 (10.5) | ML: 13.5 (2.8) BL: 14.7 (3.1) | Patient and caregiver interviews using Likert scale | Neurologist in consultation with a neuropsychologist; AD | ML: 24.2 (3.1) BL: 23.8 (3.4) | 4.80 (1.43–8.17) | 4.6 (1.17–8.03) | 0.40 (−0.70–1.50) |
Zheng et al., 2018 | ML (Cantonese): 48 (85%) ML (Mandarin): 20 (45%) BL: 61 (57%) | Diagnosis ML (Cantonese): 67.7 (9.9) ML (Mandarin): 67.0 (9.1) BL: 74.4 (9.4) Onset ML (Cantonese): 63.9 (9.7) ML (Mandarin): 63.4 (8.9) BL: 70.9 (9.4) | ML (Cantonese): 4.92 (3.85) ML (Mandarin): 10.95 (3.28) BL: 10.79 (4.32) | BAT | Two neurologists delivered the diagnosis using the NINCDS–ADRDA; probable AD | ML (Cantonese): 12.25 (5.39) ML (Mandarin): 15.75 (6.75) BL: 16.43 (6.46) | |||
N (% of females) | Mean age | Education level | Language measure | Dementia diagnosis and type | MMSE/3MSE scores | Prevalence | Rate ratio | ||
Estanga et al., 2017 | ML: 100 (58%) Early BL: 81 (54.3%) Late BL: 97 (60.8%) | ML: 57.82 (6.42) Early BL: 56.82 (6.48) Late BL: 57.56 (6.57) | ML: 12.33 (3.37) Early BL: 14.35 (3.76) Late BL: 14.98 (3.77) | BLPQ | Cognitively intact | ML: 28.44 (1.34) Early BL: 28.81 (1.09) Late BL: 28.81 (1.09) | ML: (stage 1: 11.9%; stage 2: 6.8%; and SNAP: 6.8%); Early BL: (stage 1: 3.6%; stage 2: 1.8%; and SNAP: 1.8%) | The prevalence of subjects in preclinical AD stage 1 (abnormal amyloid), stage 2 (abnormal amyloid and tau), and SNAP (abnormal tau) was significantly different (p = 0.02) between early bilinguals (stage 1: 3.6%; stage 2: 1.8%; and SNAP: 1.8%) and monolinguals (stage 1: 11.9%; stage 2: 6.8%; and SNAP: 6.8%) | |
Yeung et al., 2014 | ML: 913 (60.4%) BL: 81 (61.7%) ESL: 622 (57.4%) | ML: 77.4 (6.7) BL: 77.0 (6.5) ESL: 77.1 (7.1) | ML: 10.4 (2.9) BL: 11.9 (4.2) ESL: 8.1 (3.7) | Self-repot | DSM-IIIR; Dementia | ML: 89.0 (8.1) BL: 89.3 (6.7) ESL: 83.3 (11.0) | ML: 197 (31%), 440 (69%); BL: 86 (20%), 344 (80%) | Bilingualism was not associated with risk of developing dementia |
Study characteristics | Effect size | |||
---|---|---|---|---|
Study | N (% female); mean age (SD), Ed, MMSE/3MS | Language measure, dementia diagnosis and type | % with dementia ML/BL % no dementia ML/BL | RR, OR, HR |
Wilson et al., 2015 | 964 (76.8%), 78.7 (7.4), 14.6 (3.2) | Self-report NINCDS–ADRDA MCI | During a mean of 5.8 years (SD = 3.5) of annual follow-up evaluations, 396 individuals (41.1%) developed MCI | Higher levels (>4 years) of foreign language instruction: HR = 0.687, 95% CI: 0.482, 0.961 |
Study characteristics | Effect size | |||
---|---|---|---|---|
Study | N (% female); mean age (SD), Education level, MSSE/3MSE | Language measure, dementia diagnosis, severity, and type | % with dementia ML/BL % no dementia ML/BL | RR, OR, HR |
Hack et al., 2019 | 325 (100%); 75+ years (NA), Grade school (n = 15), High school (n = 14), Bachelor’s degree (n = 123), Master’s degree or higher (n = 173), NA | Questionnaire developed by the School Sisters of Notre Dame DSM-IV, ADLs, CERAD, MMSE, Delayed Word Recall, Verbal Fluency, Boston Naming, Constructional Praxis | ML: 27/109; BL: 82/109 ML: 60/216; BL: 156/216 | Bilingualism was not associated with a reduced risk of dementia (OR: 1.17, 95% CI: 0.69, 1.98) |
Lawton et al., 2015 | 81 (64%) Baseline ML: 4.99 (4.17), 3MSE 78.87 (9.90) BL:7.70 (4.88), 3MSE 79.56 (15.57) Follow-up ML: age 81.10 BL: age 79.31 | ARSMA-II | ML: 54/1154 BL: 27/624 | BL did not decrease the risk of dementia p = .72, AD p = .59, or VaD p = .53 |
Ljungberg et al., 2016 | 818 (51%) 73.6 (8.9) Baseline ML: 73.8 (9.0), 6.9 (1.5), 26.6 (2.3) BL: 65.7 (6.6), 14.2 (4.3), 28.7 (1.7) Follow-up ML: 78.1 (6.1), 6.5 (1.6), 25.3 (2.3) BL: 76.0 (7.7), 12.0 (2.4), 26.8 (1.6) | Language History Questionnaire DSM-IV, NINCDS–ADRDA AD, VaD, LBD, FLD, PD, and UD | ML: 102 (13.86%), 634 (86.14%) BL: 10 (12.2%), 72 (87.8) | BL did not decrease risk of dementia (p = .50) or AD (p = .36), even after adjusting for age and sex (p = .29) |
Yeung et al., 2014 | ML: 576 (61.6%), 76.1 (6.2), 10.7 (2.8), 3MS 91.2 (5.7) BL: 54 (70.4%), 75.5 (5.6), 12.4 (4), 91.1 (5.6) ESL: 360 (60.6%), 75.7 (6.4), 8.7 (3.5), 87.4 (6.9) | Self-report ML: Dementia 9.4%. 3MS 91.2 (5.7) BL: Dementia 11.1, 3MS 91.1 (5.6) ESL: Dementia 9.7%, 3MS 87.4 (6.9) | ML 54 (9.4%), 492 (85.4%) BL 6 (11.1%), 46 (85.2%) ESL 35 (9.7%), 285 (79.2%) | Model 1: 1.06 (0.69, 1.63) Model 2: .13 (0.73, 1.79) Model 3: 7 (0.67, 1.72) Model 4: (0.61, 1.59) Time 1 3MS, Time 2 3MS, and Change in the 3MS: Unadjusted model, English bilingual: Time 1, 0.6 (−1.8, 2.9), Time 2, 2.5 (−0.7, 5.7), Changed in 3MS, −1.7 (−4.2, 0.8) |
Zahodne et al., 2014 | ML: 637 (72%), 75.66 (5.79), 5.05 (3.61) BL: 430 (64%), 74.78 (5.66), 8.30 (4.22) | Self-report (four-point Likert-type) DSM-III Probable and possible AD, VaD, LBD, and other dementias | ML: 198/637 BL: 86/344 | Better self-rated bilingualism was associated with lower odds of dementia conversion. Each point on the self-report scale was associated with 0.291 lower log odds of conversion to dementia |
Description of Mild Cognitive Impairment and Dementia
Data Extraction and Risk of Bias
Data Analysis
Results
Data Collection Process
Study Characteristics
Risk of Bias for Cross-Sectional Studies
Study | Selection | Comparability | Outcome | Total/10 | |||||
---|---|---|---|---|---|---|---|---|---|
Representativeness of the sample | Sample size calculation | Non-respondents | Ascertainment of exposure | Controls for most important factor | Controls for any additional factor | Assessment of the outcome | Statistical test | ||
Alladi 2017 | – | – | ★ | – | ★ | ★ | ★ | ★ | 5 |
Alladi 2013 | – | – | ★ | – | – | ★ | ★ | ★ | 4 |
Bialystok 2014 | – | – | – | ★ | ★ | ★ | ★ | ★ | 5 |
Bialystok 2007 | – | – | ★ | ★ | ★ | ★ | ★ | ★ | 6 |
Chertkow 2010 | – | – | ★ | ★ | ★ | ★ | ★ | ★ | 6 |
Craik 2010 | – | – | ★ | – | – | ★ | ★ | ★ | 4 |
Clare et al., 2016 | – | ★ | ★ | ★ | ★ | ★ | ★ | ★ | 7 |
Estanga 2016 | – | – | ★ | ★ | ★ | ★ | ★ | ★ | 6 |
Lawton 2015 | – | – | ★ | ★ | ★ | ★ | ★ | – | 5 |
Ossher 2013 | – | – | – | – | ★ | ★ | ★ | ★ | 4 |
Perani 2017 | ★ | – | – | ★ | ★ | – | ★ | ★ | 5 |
Ramakrishnan 2017 | – | – | ★ | – | ★ | ★ | ★ | ★ | 5 |
Schweizer 2012 | – | – | – | – | ★ | ★ | ★ | ★ | 4 |
Woumans 2015 | – | – | ★ | – | ★ | ★ | ★ | ★ | 5 |
Yeung 2014 | – | ★ | ★ | – | ★ | ★ | ★ | ★ | 6 |
Zheng 2018 | – | – | ★ | ★ | ★ | ★ | ★ | ★ | 6 |
Risk of Bias for Longitudinal Studies
Study | Selection | Comparability | Outcome | Total/9 | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Representativeness of the exposed cohort | Selection of the non-exposed cohort | Ascertainment of exposure | Demonstration that outcome of interest was not present at start of study | Controls for most important factor | Controls for any additional factor | Assessment of outcome | Was follow-up long enough for outcomes to occur | Adequacy of follow up of cohorts | ||
Hack 2019 | – | – | – | ★ | ★ | ★ | ★ | ★ | ★ | 6 |
Ljungberg 2016 | – | – | – | ★ | ★ | ★ | ★ | ★ | ★ | 6 |
Wilson 2015 | – | – | – | ★ | ★ | ★ | ★ | ★ | ★ | 6 |
Yeung 2014 | – | – | – | ★ | – | – | ★ | ★ | – | 3 |
Zahodne 2014 | – | – | ★ | – | ★ | ★ | ★ | ★ | – | 5 |