Stability of Balance Performance From Childhood to Midlife

The 1970 British Cohort Study comprises over 16 000 individuals born in England, Scotland, or Wales during the same week in 1970.³⁵  Data collection has occurred at 10 waves, from birth to the most recent follow-up at age 46, which consisted of home visit interviews and a biomedical assessment. At age 46, 8581 study members participated; the remaining individuals had either died, emigrated, withdrawn from the study, could not be contacted or did not consent to participate. To be included in the analytical sample, individuals needed to have balance data at ages 10 and 46. Data at age 10 were available for those who completed a medical examination and were able to understand and perform the motor coordination tests (87.3%; 12 984 of 14 870 who participated in age 10 wave). Balance data were available for 85.8% (n = 7363 of 8581) of those participating at age 46. Reasons for nonparticipation in balance assessments at age 46 included: recent injury on preferred standing leg, lower back problems, hip problems, did not feel that it was safe to attempt the test, or refusal (n = 228). A total of 5990 and 6024 had data at both ages for static and dynamic balance scores, respectively.

Participants were asked to balance for 30 seconds on each leg, with their suspended foot against the knee of their standing leg and their hands on their hips. As balance times were heavily left skewed (>60% of the cohort could balance for 30 seconds), a categorical variable was derived from additional observations made by the assessors. One point was awarded if the foot did not move and 1 point if the hand did not move, for a maximum of 4 points across both legs. Those with 3 or 4 points were categorized as high balance, those with 2 points as medium balance, and those with 0 or 1 as low balance.²⁸ 

Participants were asked to place their hands on their hips and walk backward along a straight line by placing their feet toe to heel for 20 steps. Assessors recorded the number of steps before an error occurred (ie, deviation from line, movement of hands from the hips, or failure to maintain the toe-to-heel movement). If an error occurred in the first 5 steps, nurses counted the number of steps before the next error instead. As the number of steps were also heavily left skewed (>45% of the cohort completed 20 steps), a categorical variable of high (20 steps), medium (10–19 steps) and low (<10 steps) balance was also derived.²⁸ 

Participants were instructed to balance on their preferred leg for 30 seconds with eyes open, while keeping their other leg off the ground. They could use their arms, bend their knee, or move their body during the trial. Timing stopped if the raised foot touched the ground, or the standing foot moved. If successful in achieving 30 seconds, the participant was asked to repeat the test with eyes closed. As balance times for both eyes open and closed were heavily skewed (>85% balanced for 30 seconds with eyes open; >65% for <10 secons eyes closed), balance performance was considered in 5 ascending categories, following an approach from previously published analyses.³⁶  The 5 categories were: low (<15 seconds with eyes open), low-medium (15 to <30 seconds with eyes open), medium (30 seconds with eyes open, <15 seconds with eyes closed), medium-high (30 seconds with eyes open, 15 to <30 seconds with eyes closed) and high (30 seconds with eyes open, 30 seconds with eyes closed).

Covariates were identified a priori based on known associations with balance performance across the life course.^11,15,36,37  At age 10, childhood cognition (Edinburgh Reading Test standardized z-scores),³⁸  childhood SEP (categorized using the Registrar General’s Social Classification of the father’s occupational class: I professional and II intermediate, III skilled nonmanual and manual, or IV partly skilled and V unskilled),³⁹  and childhood illness (number of overnight hospitalization admissions between ages 5 and 10) were ascertained. Highest academic qualification, as an indicator for adulthood SEP,³⁹  was derived using records from all waves (categorized as none; General Certificate of Education (GCE) Ordinary Level, generally attained at age 16 years; and GCE Advanced Level, generally attained at age 18 years; diploma or degree level; or higher degree). The remaining covariates were assessed at age 46. Height (in cm) and BMI (as kg/m²; calculated with height and weight) were measured by research nurses; self-reported values were used where nurse measures were unavailable. Individuals reported how many days per week they exercised for 30+ minutes, working hard enough to raise their heart rate and break into a sweat. Verbal memory was assessed with a word recall test^11,13,40  in which participants were played 10 words at 2 second intervals by a computer program and then asked to recall as many as they could within 2 minutes. Grip strength was assessed as the maximum score of 6 trials (3 in each hand) using a Smedley spring-gauge hand-held dynamometer.

Given sex differences in balance performance in childhood and adulthood, sample characteristics were described by sex across each category of static and dynamic balance at age 10. χ² tests assessed the cross-tabbed proportions of balance at ages 10 and 46. The correlation between balance measures at age 10 was assessed using Spearman’s rank correlation. Multinomial logistic regressions were used to assess associations between each measure of balance at age 10 with balance at age 46. The following steps were conducted for static balance and repeated for dynamic balance. First, an interaction between sex and childhood balance was assessed (P <.05); if present, models would be stratified by sex. Second, preliminary sex-adjusted (or stratified) models were assessed. Finally, each covariate was added in turn, and the final-adjusted model of all covariates is presented using relative risk ratios. Following a missing at random assumption, covariate data were imputed using multiple imputation chained equations⁴¹  and estimates from 20 imputed datasets were combined using Rubin’s rules.⁴²  Supplementary analyses compared characteristics between: (1) those with low static childhood balance and either high or low balance at age 46; (2) those with high static childhood balance and either high or low balance at age 46; and (3) the main analytical sample and those excluded because of missing balance or covariate data. Sensitivity analyses repeated regression models using complete cases data. Analyses were performed in Stata MP 16 and RStudio 4.0.3.

Characteristics of the analytical sample are provided in Table 1. Compared with those with low or medium static balance at age 10, those with high static balance had higher SEP and cognitive performance in childhood and adulthood, lower BMI, better grip strength, and exercised more frequently. There were no differences in childhood hospital admissions or adult height. Associations between dynamic balance groups and sample characteristics were similar, however there were no associations with BMI or exercise frequency.

At age 10, females had better static (high: 51.2% vs 38.8%) and dynamic balance (high: 51.2% vs 47.2%) than males. Conversely at age 46, males had slightly better balance than females (high: 14.1% vs 11.6%; Table 2). Spearman rank correlations between static and dynamic balance at age 10 were weak (0.25 in males and 0.20 in females); 27.8% had high performance on both tests, whereas 5.0% had low performance on both.

χ² tests suggested that static and dynamic balance at age 10 tracked strongly to balance at age 46 (Fig 1, Supplemental Table 4). Of those with high static balance, 31.1% had high or medium-high balance at age 46, compared with 18.2% of those with poor static balance. Similarly, 22.1% of those with poor static balance continued to have low or low-medium balance at age 46, compared with 8.8% of those with high balance. Notably, tracking was weaker between childhood dynamic balance and balance at age 46; 28.2% and 21.4% of those with high and poor childhood dynamic balance, respectively, had high or medium-high balance at age 46.

There was no interaction between sex and either measure of childhood balance, thus males and females were included in the same model. Poor static balance at age 10 was consistently associated with poorer balance performance at age 46 (Table 3). Relative to the highest performing group (30 seconds with eyes open and closed), children with poor static balance had a greater risk of being in any of the lower 4 balance categories at age 46 and children with medium static balance had a greater risk of being in any of the lowest 3 categories. For example, in the sex-adjusted model, those with poor static balance had a 7.07 (95% confidence interval [95% CI]: 4.92–10.16) greater risk of being in the worst performing balance group (<15 seconds with eyes open). There was only minimal attenuation of estimates in the adjusted model (risk ratio: 5.04 [3.46–7.37]), with no single covariate driving the attenuation.

Poor or medium dynamic balance was also associated with greater relative risks of poor midlife balance (Table 3). For example, relative to the highest performing group and in sex-adjusted models, those with poor dynamic balance at age 10 had 2.33 (1.66–3.26), 1.82 (1.31–2.52), and 1.27 (1.09–1.72) greater risk of having low, low-middle, or middle balance at age 46, respectively. Adjustment for covariates somewhat attenuated the results, with poor or medium dynamic balance remaining associated with a lower relative risk of being in the poorest balance group only.

Compared to those with consistently low balance (n = 237), individuals who improved from low static balance at age 10 to high balance at age 46 (n = 195) were more likely to be male, have higher childhood cognition and SEP, lower childhood illness, higher education, lower BMI, taller height, higher word recall, greater strength, and exercise more frequently (Supplementary Table 5). Conversely, compared with those with consistently high balance (n = 843), those who declined from high static balance at age 10 to low balance at age 46 (n = 238) were more likely to be female, have lower childhood cognition and SEP, lower education, higher BMI, shorter height, lower word recall, lower strength, and exercise less frequently.

Individuals who were missing balance data at age 46 because of attrition or not having a valid measurement were more likely to have poor static (55.8% vs 44.2%) and dynamic balance (55.2% vs 44.8%) at age 10 than the analytical sample. Compared with complete cases, those who were missing covariates performed worse on the balance test at age 46 (low group: 12.1% vs 5.9%), however there was no difference in age 10 balance. Results did not change when complete cases data were used (Supplementary Table 6).

In a large prospective birth cohort study, poor childhood balance performance was strongly associated with poor balance performance in midlife. Specifically, children with poor static balance and, to a slightly lesser extent, poor dynamic balance were at greater risk of having poor 1-legged balance ability at age 46. Associations were not explained by childhood SEP, cognition, and illness nor adult indicators of exercise, strength, SEP, or verbal memory. This suggests that the ability to balance in midlife largely depends on ability in childhood, reflecting an important period of development of postural strategies.

Key strengths of this study include the prospective ascertainment of balance ability in childhood and adulthood, the large population representative sample, and the novel ascertainment of balance ability in midlife before onset of major age-related disease. The study does have some limitations. Notably, balance performance could not be modeled continuously; there was a strong ceiling effect at age 10, whereas at age 46, only individuals who could balance with their eyes open for the full 30 seconds performed the eyes closed trial. Although models are adjusted for potential adulthood factors, causal interpretation of the associations should be done with caution because of the observational nature of the data and the potential for residual confounding. Dynamic balance was not ascertained in midlife, limiting our ability to examine how dynamic balance tracks across life. Finally, as in any birth cohort, there was loss-to-follow; despite this attrition, the sample remains large and representative of the mainland UK population in many respects.⁴³  As those with missing data at age 46 had poorer childhood balance than the analytical sample, estimates may be underestimated. Finally, the lack of ethnic diversity (95.3% White British) is a limitation.⁴⁴ 

This is 1 of the first studies to examine associations between childhood and adulthood balance ability, and as such, comparisons with previous studies are limited. However, other studies have identified a similar positive relationship between childhood cognition and standing balance in mid and later life.^16,17,45  This supports the hypothesis that establishment of early neural pathways may play a crucial role in balance across the life course via a sensorimotor integration mechanism.^16,46  Advantageous early motor development, including better balance and coordination, is also associated with positive health outcomes in adulthood.^15,47  For example, a recent study using 1958 British birth cohort data demonstrated that better childhood psychomotor coordination (eg, balance, ball catching, dexterity, and jumping) was associated with lower risk of death by age 60.⁴⁷  Possible mechanisms include direct tracking of coordination skills across life (eg, minimizing risk of accidental death) and indirect life course pathways leading to high disease prevalence. As balance appears to track strongly from childhood to adulthood, future research must explore if poor adulthood balance ability mediates the association between poor psychomotor coordination and premature mortality.

We tested 2 sets of factors that may explain the association between childhood and adulthood balance. First, it was hypothesized that better balance in childhood may reflect current and future cognitive and socioeconomic advantages, with evidence showing that better cognition and SEP across the life course contribute to better adult balance.^{11–13,16,34}  Second, children with better motor development tend be more physically active throughout life^26–28,48  and, as a result, may demonstrate greater muscular strength in adulthood.^29,49  As adults age, maintenance of muscle strength and power may facilitate the motor components of balance.^50,51  However, neither set of factors explained the association between childhood and adulthood balance.

Instead, there was robust evidence of an association between childhood and adulthood balance ability, independent of the hypothesized adulthood factors, suggesting that the ability to balance is largely established in early life. Developmental balance curves demonstrate steep nonmonotonic improvements in postural control between ages 8 and 13, at which age balance ability is nearly equivalent to adult-like performance.^23,52–55  Stark improvements in balance performance during these ages are a result of both refined motor skills and changes in strategies used to maintain balance.⁵⁶  Early postural control strategies (eg, <10 years) focus on relatively large and reactive adjustments to body position using an open-loop system of control,⁵⁴  which does not incorporate somatosensory feedback. Around age 10, children learn to refine these strategies with smaller, more frequent postural adjustments^54,57–59  using a closed-loop system of control, resulting from nonlinear development of the 3 afferent sensory systems involved in balance: visual, proprioceptive, and vestibular input.^54–59  As postural strategies are largely refined by midchildhood, this provides further support that maturation of balance control occurs in early adolescence.

The mechanisms involved in static and dynamic balance differ, as demonstrated by the low correlation and overlap in performance at age 10, and may explain the stronger association between static balance measures at ages 10 and 46. Dynamic balance ability requires higher level cognitive processes to successfully respond to goal-oriented movement (eg, stepping off a curb or avoiding an obstacle).⁶⁰  Static and dynamic balance performance are both closely linked to fall risk in mid and later life.^61,62  Therefore, when data becomes available, further research must examine if similar associations extend from dynamic balance in childhood to dynamic balance in adulthood and if childhood balance directly contributes to balance-related outcomes, such as fall risk.⁶³ 

The maturation of balance control in childhood highlights the need and opportunity for early intervention in children with lower balance abilities. A systematic review of 17 studies has demonstrated a moderate to large effect of balance training on both static and dynamic balance performance in children and adolescents.⁶⁴  Associations were strong, regardless of age, sex, training history, and intervention frequency or setting, highlighting the feasibility and potential impact of early intervention. This could manifest as clinical screening and subsequent referral to balance training during regular physician check-ups or as national public health policies that incorporate balance development into educational curricula and physical activity guidelines. Additionally, we identified characteristics of individuals with high childhood and low adulthood balance (eg, declining) and those with low childhood and high adulthood balance (eg, improving) that could inform evidence-based interventions. These included sex, cognition, SEP, BMI, strength, and exercise frequency.

With a rapidly ageing population, public health efforts increasingly aim to minimize fall risks and improve physical function in older adults. Physical activity guidelines recommending balance exercises target only adults over the age of 65.^8–10  In addition to existing interventions targeting older adults, the strong continuity of balance from childhood to adulthood provides support for earlier integration of these health promotion strategies and guidelines. This could have an important impact in improving adulthood balance ability and reducing balance-related outcomes.