Falls, sarcopenia and growth in early life

METHODS

RESULTS

The characteristics of the 2148 study participants are shown in Table 1 according to falls history. The prevalence of having had a fall in the past year was 14.3% in men and 22.5% in women (odds ratio for falls history in women compared with men: 1.74 [95%CI 1.38, 2.19], p<0.001). Men who had fallen in the past year had lower weight at one year and conditional infant growth, lower grip strength, shorter stature and were more likely to report slower walking speeds. Women who had fallen in the past year had lower grip strength, shorter stature, greater body mass index, and were more likely to report slower walking speeds. Birth weight, adult weight, fat-mass and non-fat mass, social class, smoking habit and alcohol consumption were not related to falls history in men or women and were not considered further in analyses. Chronological age was not associated with falls history in men or women but this variable was included in mutually adjusted analyses. Conditional infant growth was more strongly related to falls history in men than weight at one year and was therefore included in mutually adjusted analyses. Conditional infant growth was also included in mutually adjusted models for women owing to a priori interest in its relationship with falls risk in adulthood.

Multivariate analyses were used to investigate the relationship between falls history and conditional infant growth after adjustment for adult grip strength, height, age and walking speed in men, and after adjustment for adult grip strength, height, age, walking speed and body mass index in women. Among men, grip strength and walking speed remained strongly associated with a history of falls, but the relationships between conditional infant growth or adult height and falls history were attenuated (Table 2). Walking speed was the only characteristic associated with falls history in the mutually adjusted model for women (Table 2). Figure 1 displays the univariate and mutually adjusted odds ratios for falls history in men. Consideration of effect size in a sequence of three separate multivariate models for men (for conditional infant growth and grip strength, conditional infant growth and walking speed, and finally conditional infant growth and adult height as predictors of falls history) suggested that attenuation of the association between conditional infant growth and falls history was slightly greater after adjusting for grip strength than after adjusting for walking speed or adult height although all factors reduced the effect size (odds ratios for falls history per SD decrease in conditional infant growth were: 1.21 (95%CI 0.98, 1.48) p=0.08 adjusted for grip strength; 1.24 (95%CI 1.00, 1.52) p=0.05 adjusted for walking speed; and 1.23 (95%CI 0.99, 1.52) p=0.06 adjusted for adult height). There were no quadratic relationships between early growth and falls.

DISCUSSION

To our knowledge, this is the first study to demonstrate an association between poor growth in early life and falls in older people. We have shown that lower conditional infant growth is significantly associated with a history of falls in older men. There was a similar but non-significant relationship in the women. However there was no relationship between birth weight and falls in either men or women. The other significant predictors of falls in this population were gender, height, grip strength, and walking speed. BMI was only associated with falling in the women and neither fat mass nor non-fat mass were related to falls in either gender. Age within the narrow range studied was not significantly associated with falls history in men or women.

Multivariate analyses allowed further investigation of the relationship between conditional infant growth and history of falls in the men. In particular, low social class was a potential alternative explanation for the relationship between poor early growth and increased falls. However social class, ascertained either at birth or in adulthood, was not related to history of falls and furthermore social class was not significantly associated with infant growth although there was a relationship with adult walking speed. The association was also not explained by height but was attenuated after adjustment for grip strength and walking speed. The contribution of low grip strength to the relationship between poor conditional infant growth and falls is consistent with findings from previous studies demonstrating a relationship between poor early growth and reduced adult muscle strength(21;23;24). It suggests that early environmental influences may not only result in long term impaired muscle function but also impact on the important clinical sequelae of falls. Supportive evidence for the link with falls comes from a study reporting an association between poor childhood growth and risk of hip fracture in later life(34). Sarcopenia and increased falls risk, as well as reduced bone mass, may lie on the causal pathway.

Associations between early growth and adult health have been explained by the phenomenon of programming which is the persisting influence of exposures occurring at critical periods of early development on long-term organ structure, function and regulation(35). Programming is an example of developmental plasticity which is the ability of a single phenotype to produce more than one alternative form of structure, physiological state or behaviour in response to environmental conditions(36). This enables the production of phenotypes that are better suited to their environment than would be possible if the same phenotype was produced regardless of environmental conditions(37). The consequences of developmental plasticity for human health were the subject of a recent paper(38).

Programming of muscle has been documented in animal models, for example in the field of animal husbandry where prenatal nutritional manipulation of muscle growth and quality is of particular interest to the meat industry. Prenatal undernutrition has been associated with reduced neonatal muscle weight but not fibre number in the sheep(39) and a reduction in postnatal muscle fibre number in the pig(40), guinea pig(41) and rat(42). There is evidence that these effects persist (43). There is evidence that the muscle phenotype can also be influenced by postnatal nutrition(44) and the observations from this study linking falls history with conditional infant growth rather than birth weight suggests that postnatal rather than prenatal influences on muscle growth and development may be more important for risk of falls in later life.

The relationship between conditional infant growth and falls appeared to be stronger in the men than in the women although there was no statistical evidence for different associations between the genders. Health care utilization is known to differ between men and women but the use of self-reported fall data in this study avoided information bias due to differential recording of falls in medical records. Gender differences in programming have been described previously, associated with differing growth trajectories and susceptibility to environmental influences between male and female fetuses (45). However there are also differences in adult body composition that may be relevant. Women have a lower muscle to fat ratio than men and the relatively lower proportion of muscle may attenuate associations between growth in early life and adult muscle function in women. The weaker although still significant relationship between grip strength and falls in women supports the concept that gender differences in body composition are relevant in understanding the differing patterns of associations.

There are a number of potential caveats to the interpretation of our findings. We did not have prospective falls data and had to use history of falls in the last year to characterise fallers. This approach can lead to under-reporting because of problems with recall but is unlikely to have affected differentially those with different patterns of early growth, so no observation bias should have been introduced. In contrast, response bias may have occurred because of losses to follow up which occurred during tracing and in gaining consent to participate. However we were able to address this by characterising those who did not take part in the study in a number of ways. There were no substantial differences in birth weight, weight at one year or conditional infant growth between subjects who were traced and eligible to participate in the study but did not, and those who had a home interview. Furthermore, there were no major differences in age, social class, alcohol consumption or activity level, between interviewed subjects who did or did not attend clinic.

The proportion of current smokers was lower among interview subjects who did come to clinic than those who declined suggesting that there may have been a ‘healthy subject’ effect in this study. However our comparisons were internal, therefore unless the relationship between early size and falls history differed between those who did and did not come to clinic, no bias should have been introduced(27). We used muscle strength rather than muscle mass to define sarcopenia. The latter approach has been reported more commonly in the literature but there is growing support for the use of strength because of its greater functional relevance to physical performance in older people(31).

The potential relevance of these findings to clinical practice needs to be considered. At the population level, poor early growth appears to be a risk factor for falls. Knowledge of early growth and muscle strength may contribute to risk stratification tools for the primary prevention of falls in the future. Greater understanding of the mechanisms underlying the association between early growth, sarcopenia and increased falls risk may also facilitate the development of beneficial interventions across the life course to preserve muscle function and prevent falls. Our findings support an association between poor growth in early life and falls in older men. This relationship appears to be mediated partly through sarcopenia. The weaker, non-significant association in women may reflect gender differences in body composition. The absence of a relationship with birth weight suggests that postnatal rather than prenatal influences on muscle growth and development may be important for risk of falls in later life.