Mortality in the Hertfordshire Ageing Study: association with level and loss of hand grip strength in later life

Abstract

Background: weak hand grip strength in later life is a risk factor for disability, morbidity and mortality and is central to definitions of sarcopenia and frailty. It is unclear whether rate of change in grip strength adds to level of grip strength as a risk factor for poor ageing outcomes.

Methods: study participants were 292 community-dwelling men and women whose grip strength was measured during the 1994/5 (average age 67) and 2003/5 (average age 76) phases of the Hertfordshire Ageing Study, UK. Individual rate of change in grip strength was estimated using a residual change method. Mortality was followed-up to 2011 (42 men and 21 women died).

Results: average grip strengths in 2003/5 were 38.4 kg (standard deviation [SD] = 8.1) and 23.7 kg (SD = 6.6) for men and women respectively. Average annualised rates of change in grip strength (2003/5 minus 1994/5) were modest owing to a healthy-participant effect (men: -0.12 kg/y, SD = 0.71; women: 0.08 kg/y, SD = 0.54) but varied widely. Mortality risk varied according to level and rate of change in grip strength (P = 0.03); death rates per 100 person years of follow-up were 6.7 (95% CI: 4.6, 9.6) among participants who lost grip over time and had low grip in 2003/5, in contrast with 0.8 (95% CI: 0.1, 5.8) among participants whose grip changed little over time and remained high in 2003/5.

Conclusions: levels of grip strength in later life should be considered in conjunction with estimates of change in grip strength identified by repeat measurement over time. Normative data for longitudinal change in grip strength are required.

Keywords: change; epidemiology; longevity; older people; physical function; sarcopenia.

Figure 1. Mortality rates in HAS men and women combined subsequent to 2003/5 clinic according to level of grip strength at the HAS 2003/5 clinic and change in grip strength between 1994/5 and 2003/5 clinics

Footnotes: P-value for interaction between level and residual change in grip strength, adjusted for gender: p=0.03. P=0.02 also adjusted for age, height and weight residual; p=0.01 also adjusted for social class, walking speed, smoking habit and alcohol intake; p=0.06 also adjusted for number of systems medicated Error bars represent 95% confidence limits. Thirds of the grip strength distribution at 2003/5 clinic were identified by: “lowest” third 15-34kg for men and 2-22kg for women; “middle” third 35-42kg for men and 23-27kg for women; “highest” third 43-58kg for men and 28-39kg for women. Median crude annualised changes in grip strength (2003/5 minus 1994/5, divided by follow-up duration) within each third of the residual change in grip strength distribution were: “loss” -0.72 kg/year among men and -0.44 kg/year among women; “no change” -0.05kg/year for men and 0.08kg/year for women; “gain” 0.61kg/year among men and 0.59kg/year among women. Only 2 men and 1 woman experienced loss of grip strength whilst maintaining a grip strength in the highest third of the sex-specific distribution at 2003/5 clinic; none of these men and women died. This death rate estimate has been regarded as missing and omitted from the figure.

Figure 2. Contour plots for combinations of grip strength measurements at HAS 1994/5 and 2003/5 clinics which were estimated to have the same hazard ratio for mortality in comparison with a reference group of individuals with average grip strength at each phase of follow-up.

Footnotes: These contour plots were derived by fitting a second order model for mortality on linear terms for grip strength at 1994/5 and 2003/5 clinics, in addition to an interaction term, and quadratic terms for each of these grip strength values. A second order model provided a better description of mortality risk than a model that only included linear terms for grip strength at each follow-up (p<0.05 with or without adjustment for potential confounders). Average grip strengths among men were 39.5kg and 38.4kg at HAS 1994/5 and 2003/5 clinics respectively; corresponding figures for women were 23.0kg and 23.7kg. The diagonal lines reflect identical values of grip strength at 1994/5 and 2003/5 follow-ups; combinations of grip strength measurements located to the bottom right of the lines reflect loss of grip strength (lower in 2003/5 than 1994/5) and combinations to the top left of the lines reflect gain in grip strength.