Main findings were that gait and cognition did not decline markedly over 5 years, a slower gait but not cognitive impairment significantly increased falls risk for non-Māori only, being male increaesd falls risk for Māori, and MCR prevalence was low especially for non-Māori. We rejected our hypothesis that gait and cognition would be independently associated with falls risk for both cohorts.
Of note are the high attrition rates for both cohorts, especially in the later years. Around 65% of the total sample was lost to follow-up, or death which although not surprising given the advanced age, will have undoubtedly biased results towards participants who were cognitively and motorically resilient, and who fell less. Kerse and colleagues report retention rates and reasons for drop out for the first 2 years of LiLACS, and note the higher drop-out for Māori for reasons other than death . They also report significant differences in baseline characteristics for those retained in the study compared with those who died or were lost to follow-up.
Exceptions to the mostly stable scores for both cohorts over the 5 years were the 2 lowest cognitive scoring groups for Māori whose trajectory markedly declined, and who together comprised just over a quarter of the total Māori cohort. A further exception was the increase in the number of non-Māori fallers at Year 5. The modest decline in gait and cognition for those who remained in the study is an important finding and perhaps surprising given the multiple morbidities associated with advanced age, also evident for 93% of LiLAC NZ participants who reported at least 2 diagnosed health conditions at baseline .
Around a third of Māori and approximately 40% of non-Māori fell at any one time over the course of the study, which is broadly comparable with published data for adults over 65 years . Non-Māori were slightly older which may have impacted on falls prevalence, especially by Year 5 where almost half non-Māori fell compared to under a third of Māori. Based on PASE scores at baseline, Māori were significantly more active than non-Māori, suggesting that lower falls prevalence was not related to lower amounts of activity. Falls prevalence for Māori did not increase over time, although the reasons for this are not clear.
Gait speed for non-Māori participants was comparable to norm-referenced data. Oh-Park  reports 0.95 m/s for men 80 to 84 years lowering to 0.88 m/s over 85 years, and for women 80 to 84 years 0.87 m/s lowering to 0.78 m/s over 85 years. Average gait speed for Māori was significantly slower, which may reflect anthropometric differences related to BMI which was significantly higher, or culturally different gait dynamics. Importantly, trajectories for gait speed for around 90% of Māori did not vary over the course of the study. Cognitive outcomes were considerably higher than published data for community dwelling older adults, for both cohorts. Bassuk and Murphy (2003) report an average 3MS score of 75.4 for people 75–84 years, lowering to 63.3 for those over 85 years. Only the 2 lowest Māori cognitive clusters had scores comparable to these, with all other groups scoring higher.
MCR prevalence at baseline was low particularly for non-Māori, although estimates are comparable to younger cohorts from Australia and UK . Estimates for Māori were higher, although still lower than overall multi-country prevalence of 9.7% . It is unwise to draw any conclusions from these results other than to note the low estimates probably reflect the high functioning nature of both cohorts, despite their advanced age. Kerse and colleagues documented the prevalence of dementia in LiLACS NZ as defined by 3MS scores (≤ 80 for Māori; ≤ 84 non-Māori), noting that the proportion of participants with no dementia stayed steady at approximately 87% . Overall these findings point to a stable cohort with mostly preserved gait and cognition, both of which are critical to independent living and overall functioning.
Results from the multivariate modelling distingished between cohorts. For Māori, being male was the only significant predictor of falls when gait and cognition were entered separately into the model but not when MCR was included as an independent predictor. This may reflect a modifying effect of gender on gait speed. Māori men walked at a significantly slower pace than non-Māori men, but this was not the case for women. Earier reports suggest that being female confers a higher risk of falling , although data are not specific to adults of advanced years. For non-Māori, gait was a significant predictor of falls, and cognition trended towards significance. The findings were comparable when MCR criteria were included in the model. These differerences are interesting, although our interpretation is cautious in line with the small proportion with MCR. Results for non-Māori conform to a substantial body of research supporting the association between gait, cognition and falls in older adults [8, 27]. Results for Māori suggest the need for a broader repetoire of predictors to idenfity associations between falls and socio-cultural features such as whānau (family), living environment and support networks. These outcomes show that risk factors for falls are inconsistent across ethnic groups. Intervention strategies may also need rethinking.
The key strength of this study is the robust sampling which enabled reporting of novel data for a large cohort of Maori and non-Māori of advanced years with acceptable rates of attrition. Key limitations include sample bias as noted above, and the approach to collecting falls data which was retrospective via self-report and does not conform to best practice . Also, gait speed was collected over a 3 m distance which is short, although congruent with population-based, longitudinal studies . Lastly, because Māori were not involved in developing the outcomes for this study they are unlikely to reflect key aspects of Māoritanga (Māori culture, beliefs, traditions), and may therefore be culturally-biased towards non-Māori participants.
In conclusion this study contributes to our understanding of falls evolution and the role of gait and cognition to falls in community-dwelling adults of advanced age. For both Māori and non-Māori who live on into advanced years, fall frequency is relatively stable and there is minimal decline in gait and cognition For non-Māori, slow gait speed is a significant predictor of falls. These findings resonate widely, given current and projected demographic trends. Further work is needed to identify cultural appropriate strategies to improve falls risk for those living to advanced age.