Background Currently, needs assessments for older people are based on self-reported measures of physical capability, whilst performance-based measures are commonly used in the academic literature. By comparing these measures in the same population, two distinct groups of discordant individuals may be identified. One group reports poor capability yet performs to higher levels than expected (underestimators), whilst the other group, report higher levels of capability than expected given their poor performance (overestimators). The main objective of this study was to explore factors across life associated with these two types of discordance. By characterising these groups it may be possible to identify those individuals whose needs may go undetected when only one type of assessment is used.

Methods Analyses were performed using self-reported and performance-based measures of physical capability collected at age 60–64 in the MRC National Survey of Health and Development. Summary scores of the self-reported and performance-based measures were produced and plotted against each other to identify two discordant groups, and a concordant group (the reference outcome category). Multinomial logistic regression was used to test the associations of discordance with: sex, childhood and adult socioeconomic position (SEP), depression and obesity from midlife onwards, and marital status at age 60–64. Associations between each factor and discordance were first adjusted for sex and then were mutually adjusted for the other factors investigated (N = 1733).

Results In sex-adjusted models all factors investigated, with the exception of obesity, were associated with at least one type of discordance. In fully-adjusted models women, those of higher SEP and those with symptoms of anxiety and depression at age 43 were at higher risk of being categorised as underestimators vs. concordant; for example the RRR for women vs. men = 9.97 (95% CI: 5.33–18.66). Conversely, women and those with symptoms of anxiety and depression at age 43 were at lower risk of being categorised as overestimators vs. concordant; for example the RRR for symptoms of anxiety and depression vs. no depression = 0.34 (0.12–0.94). Participants who were single and never married, were at increased risk of being overestimators vs. concordant when compared with those currently married (RRR = 2.49 (1.18–5.25)).

Conclusion These findings suggest that a range of factors across life are associated with discordance. These factors require further investigation and should be taken into consideration when either self-reported or performance measures are used in isolation to assess physical capability, to ensure all those with needs related to their physical capability are identified.