Overall summary estimates of effect for the associations between telomere length (TL) at times 1 and 2 and physical performance (PP) at times 1 and 2 from random effects meta-analyses.

n = Sample size in age and sex adjusted analyses;*Number of studies in analyses;**T1 and T2 are telomere lengths at times 1 and 2 respectively. PP1 and PP2 are physical performance measures at times 1 and 2 respectively.††Walking speed includes TUG measures in CaPS and NSHD. T2T1 is telomere length at time 2 conditional on telomere length at time 1 and is a measure of change in telomere length given the initial level. ΔPP is conditional change in physical performance between time 1 and time 2;†Mean difference in standardised walking speed; Mean difference in standardised chair rise speed; Odds Ratio of poor balance (lowest 20%). For change in balance in NSHD an ordinal group was created for balance from time 1 to time 2 where good balance was balancing ≥5 s and poor balance for ‡P-value is obtained from the heterogeneity χ2;‡‡Fully adjusted model is for age, sex, BMI, smoking status, SEP and health status; Telomere length measures at times 1 and 2 have been z-scored. Random effects meta-analyses were undertaken.