Evidence before this studyWe have searched PubMed and Scopus from 1990-2017, to identify relevant studies that contain terms for older people; all-cause mortality; glycaemic control; glycaemic variability (with synonyms). We also identified current international guidelines for older people. Overall, the data on optimal glycaemic targets for older people are scant, particularly from prospective studies. In terms of the association between glycaemic control and mortality in older populations the finding have suggested a 'J' shaped distribution in that relationship, although the point at which a significant mortality hazard is observed at the lower end of the glycaemic range has varied between studies. In terms of glycaemic variability, it has been identified that longer term variations in glycaemic control are associated with mortality risk. However, these analyses have not been graduated for magnitude or direction of variability. In addition, previous analyses have not considered the impact of low HbA1c levels, which are associated with mortality risk independent of diabetes intervention.
Added value of this studyIn this large population study we are the first group to consider both glycaemic control and glycaemic variability together. We have also employed a new metric for variability which considers exposure to clinically significant changes in glycaemic control. This metric enabled us to assess the direction of change as well as the overall variability. Integrating glycaemic control and variability in our modelling enabled us to consider the importance of stability as a potential factor in understanding the mortality hazard in this population. Additional nuances to our analysis include: consideration of low HbA1c values; higher levels of granularity compared to previous studies in terms of glycaemic thresholds, with 0.5%(5.5mmol/l) HbA1c increments; consideration of gender differences; and the distinction between those who develop diabetes in midlife and those who develop it in older age.
Implications of all the available evidenceOur data suggest that we may need to rethink how we consider glycaemic targets in the older diabetes population, in a number of ways: firstly, that variability expresses significant hazard in older people; secondly, that variability may be independent of diabetes therapies and may be related to other factors related to aging; thirdly, stability seems to attenuate hazard in medium to higher ranges of glycaemic control; and finally, there may be some important gender differences in relation to glycaemic control and hazard which are not considered in current guidelines. Therefore, while we recognise that observational data can often raise more questions than answers; we would advocate that we reconsider glycaemic control not simply as a target to direct therapeutic management, but as an important piece of information in relation to assessing individual risk. Perhaps in the past we have been too polarised in our view of glycaemia as purely indicative of optimal control, rather than as a p...