Author | Year | Country and dataset | n | Study design | Mean age (sd) | Male (%) | Brain function | Muscle structure | Associations* |
---|---|---|---|---|---|---|---|---|---|
1. Berryman et al. [65] | 2013 | Canada, Training Intervention Study | 48 | Baseline characteristics from a large physical training intervention study | 70.8 (5.4) | 41.67 | MMSE & modified Stroop test | LBM (DEXA) | Study: none |
Calculated: A GLM showed no association between LBM and MMSE, Stroop naming, reading or inhibition tasks, adjusted for sex and age. However there was an association between the Stroop flexibility task and LBM (t 2.126, p = 0.039, partial eta squared 9.3%), however after adjusting for education and height the effect was attenuated (p > 0.05). | |||||||||
2. Bites et al. [66] | 2013 | Chile | 306 | Retrospective study | M 74.9 (61–91), F 75.5 (69–90) | 24.5 | MMSE | TLM, Arm LM and Legs LM (DEXA) | Study: none |
Calculated: Authors sent one data sheet for this study and Bunout et al., as there is a large amount of overlap between the studies. N = 401, mean age 75.3 (sd 4.8), males 28.7%. GLM performed adjusting for sex and gender. Total LM (t 2.38, p = 0.018, partial eta squared 1.4%) and Leg LM (t 3.53, p < 0.001, partial eta squared 3.1%) were both associated with MMSE score but Arm LM is not. After adjusting for height the relationship between total LM and MMSE is non-significant and between leg LM and MMSE is attenuated (t 2.09, p = 0.038, partial eta squared 1.1%). | |||||||||
3. Bunout et al. [67] | 2005 | Chile | 298 | RCT | M 75.4 (4.8) F 75.8 (4.7) | 29.2 | MMSE | TLM, Arm LM and Legs LM (DEXA) | Study: none |
Calculated: See Bites et al. 2013 for analysis using this dataset | |||||||||
4. Auyeung et al. [68] | 2013 | Chinese University of Hong Kong - 4y f/u | 3153 | Prospective observational study | M 71.76 (4.67) F 72.03 (5.07) | 49.7 | CSI-D and MMSE | ASM, LLMM, FFM (DEXA) | Study: none |
Given: CS-CSID did not predict TLM or ASM at baseline or at 4 years (all p > 0.05). However baseline MMSE was associated with baseline TLM (rho = 0.058, p = 0.002) and ASM (rho = 0.061, p = 0.001) and at follow-up (TLM rho = 0.058, p = 0.002, ASM rho = 0.054, p = 0.005). MMSE at follow up was not associated with TLM or ASM at baseline or follow-up (p > 0.05). | |||||||||
5. Auyeung et al. [69] | 2011 | Chinese University of Hong Kong - 4y f/u | 2737 | Prospective observational study | M 71.6 (4.58) F 71.5 (4.85) | 55.3 | CSI-D and MMSE | ASM (DEXA) | Study: In men, low baseline ASM predicted lower MMSE score after 4 years (B = 0.246, p < 0.01) however after adjustment for age, years of education and baseline MMSE it no longer did (p > 0.05). In women, ASM did not significantly predict MMSE after 4 years, either before adjustment or after (p > 0.05). |
Given: see Auyeung et al. (2013) for analysis using this dataset | |||||||||
6. Lee et al. [70] | 2011 | Chinese University of Hong Kong | 4000 | Prospective observational study | M 72.3 (5.0) F 72.5 (5.3) | 50 | CSI-D and MMSE | ASM, LLMM, FFM (DEXA) | Study: none |
Given: see Auyeung et al. (2013) for analysis using this dataset | |||||||||
7. Auyeung et al. [71] | 2008 | Chinese University of Hong Kong - baseline | 4000 | Prospective observational study | M 72.3 (5.0) F 72.5 (5.3) | 50 | CSI-D and MMSE | ASM (DEXA) | Study: none |
Given: see Auyeung et al. (2013) for analysis using this dataset | |||||||||
8. Pedersen et al. [72] | 2012 | Denmark | 72 controls | Cross-sectional study | Median 53 (48–60 inter quartile range) | 46 | DART, WAIS-III information subtest, TMT-A&B, Rey Auditory Verbal Learning Test (RAVLT), Symbol Digit Modalities Test (SDMT), and fluency tests | FFM (DEXA) | Study: None |
Calculated: FFM did not predict the cognitive z score with or without adjusting for BMI and childhood intelligence (Danish Adult Reading Test, DART). The six individual cognitive tests were then analysed: FFM did not predict RAVLT, SDMT, category fluency (using animals) or TMT-b test, with or without adjusting for BMI and childhood intelligence (DART). Unadjusted, there was no significant association between the letter fluency test (using “s”) and FFM (P > 0.05), however after adjustment for BMI and DART, letter fluency was significantly associated with FFM (t 2.34, p = 0.02, partial eta squared 7.7%). TMT-a test did significantly predict FFM (t 3.08, p = 0.003, partial eta squared 12.3%). After adjusting for BMI and DART the relationship became non-significant. | |||||||||
9. Magri et al. [73] | 2006 | Italy | 27 controls | Cross-sectional case–control study | Controls 33.3 (7.15) | 0 | MMSE | FFM (BIA) | Study: none |
Calculated: FFM did not significantly predict MMSE (p > 0.05), adjusting for age. Adjustments for BMI and educational level did not significantly affect the results. | |||||||||
10. Lasaite et al. [74] | 2009 | Lithuania | 29 healthy controls | Observational case–control study | 66.2(6.3) | 0 | TMT-A and B and digit span | FFM (BIA) | Study: none |
Calculated: FFM does not significantly predict TMT-A or B adjusting for age +/− height (p > 0.05). Trend with FFM predicting digit span (t 1.96, p = 0.06, partial eta squared 13%) but attenuated when adjusted for height in addition to age (p = 0.37). | |||||||||
11. Liu et al. [75] | 2014 | Taiwan, I-Lan Longitudinal Aging Study | 983 | Population based ageing cohort study | 65.2 (9.3) | 50.6 | MMSE | LBM and Relative ASM (=ASM/ height2) (DEXA) | Study: A t test comparing mean MMSE in those with normal RASM and those within the lowest 20 % of RASM found a significant difference in men and women of all ages (p < 0.05). |
Given: In a MLR, RASM did not predict MMSE after adjusting for age and sex (beta −0.003, p = 0.940). Adjusting for education in addition did not affect the results. | |||||||||
12. Moore et al. [76] | 2012 | USA, Baltimore Longitudinal Study of Aging | 786 | Longitudinal cohort study | 66.3 (range 26–96) | 51.9 | California Verbal Learning Test (CVLT), digit-span test, TMT A & B | Mid-femur thigh CSA (CT) | Study: none |
Given: In a linear regression, none of the cognitive tests predicted thigh CSA, adjusting for age and gender. After adjusting for age, gender and height, the digit-span backward test became significantly associated with thigh CSA (beta −1.55, p = 0.024). | |||||||||
13. Kamijo et al. [77] | 2014 | USA, FITKids Study | 37 (healthy weight) | Cross-sectional study (case–control substudy comparing obese and healthy weight children) | 8.8 (0.6) | 46 | Kaufman Brief Intelligence Test (K-BIT) | TLM (DEXA) | Study: none |
Calculated: Authors sent one data sheet for the FITKids study as there is considerable overlap in subjects between the two Kamijo et al. papers [77, 78], (n = 139, mean age 8.8 (sd 0.6), male 51.1%). A GLM found that TLM did not predict K-BIT after adjustment for age and gender (p > 0.05). Adjusting for BMI in addition did not alter the results. | |||||||||
14. Kamijo et al. [78] | 2012 | USA, FITKids Study | 126 | Cross-sectional study | 8.9 (0.5) | 50 | Kaufman Brief Intelligence Test (K-BIT) | TLM (DEXA) | Study: none |
Calculated: as per Kamijo et al. [77] |