Background: A validated frailty index (eFI) based on cumulative deficits in the electronic primary care record has been implemented in the UK NHS primary care. eFI scores categorise people into fit, mild, moderate and severely frail, whilst ignoring transitions between stages.
Objectives: To identify clusters of frailty trajectories over five years; and to examine their association with subsequent mortality.
Methods: A cohort of 270,908 older people, aged 65+ in 2009 and alive in 2013, were identified from UK primary care records (CPRD GOLD) linked to hospital inpatient and mortality data. Each calendar year, eFI index was applied to calculate the cumulative number of deficits. These deficit counts were modelled with latent class growth analyses to identify clusters of people with differential frailty trajectories. Bayesian information criterion (BIC) was used to select the optimal number of clusters and distribution. To illustrate transition between stages over five years, the emerging clusters were retro fitted to predefined frailty stages at each year. Follow up after classification in frailty trajectories went from 1/1/2014 until end of 2018, death, or loss to follow-up, and Cox models used to estimate relative risk of death according to frailty trajectory clusters, adjusted for age and gender.
Results: Overall, mean (SD) eFI changed from 0.11(0.08) to 1.48 (0.09), indicating slow worsening frailty. Five-cluster Poisson models of frailty trajectories had the smallest BIC. Two clusters (C5, 6%, C4, 20%) included fit people remaining fit with zero deficit throughout the 5-year period, and with slowly onset frailty (increasing from 0 to up to 2 deficits over the 5-year period), respectively. One cluster (C3, 37%) could be classified as slow frailty progression from pre to mild frail. Patients in C2 (29%) and C1 (8%) experienced slow accelerating rates of mild and moderate frailty respectively. Patients with slowly onset frailty cluster, C4, and slow frailty progression cluster, C3, were 24% (95% CI: 1.17, 1.31) and 53% (1.46, 1.61), respectively, more likely to die in 5 years compared to C5. The death risk increased to twice for patients in the two clusters of slow acceleration of mild (C2) and moderate (C1) frailty.
Conclusions: Five distinct frailty trajectories are identified in older patients including one cluster experiencing transitions from fit to mild frailty and over 50% higher risk of mortality. Further research is needed, including the validation of the identified clusters in external datasets.