A frailty, depression and osteoporosis cluster marked higher mortality in older adults

Design

Secondary analysis of a cohort using latent-class analysis and Cox models

Population

16,095 community-dwelling adults aged ≥70 without prior dementia or cardiovascular disease

Primary outcome

All-cause mortality by multimorbidity cluster

Effect

Frailty/depression/osteoporosis cluster mortality hazard ratio 1.56 (men) and 1.68 (women)

Multimorbidity is the rule rather than the exception in older adults, but it is usually counted, not characterised. This analysis of 16,095 community-dwelling Australians aged 70 and over used latent-class analysis to find which conditions cluster together and which clusters carry the worst survival, over a median 10.85 years.

Almost 88% were multimorbid at baseline. Five clusters emerged, and the pattern that mattered most for death was a frailty, depressive symptoms and osteoporosis cluster, associated with higher mortality in both men and women. Kidney-disease-containing clusters — metabolic and gout in men, gout and anaemia in women — also carried excess mortality.

The value is in pattern recognition. These groupings are hypothesis-generating, not a validated tool, but they reinforce a practical instinct: the older adult whose problems cluster around frailty, low mood and bone disease is at particular risk and warrants proactive, goal-focused review rather than disease-by-disease management.

• Almost 88% of adults aged ≥70 were multimorbid at baseline; five condition clusters were identified.
• A frailty, depressive-symptoms and osteoporosis cluster carried higher mortality in both sexes (adjusted hazard ratio 1.56 men, 1.68 women).
• Kidney-disease clusters also raised mortality — metabolic/gout in men (1.63), gout/anaemia in women (1.96).
• Flag the frailty–low-mood–bone-disease pattern for proactive, goal-focused review rather than siloed management.

Why it matters

It turns an undifferentiated list of diagnoses into recognisable high-risk patterns that should change how intensively a patient is followed.

Don't overread it

Hypothesis-generating latent-class analysis from a single cohort — it identifies risk patterns, not a tool ready to guide an individual's care.

The statistics, in plain English

A hazard ratio around 1.5–2.0 for a cluster means a meaningfully higher death rate than peers without it. These are associations from one cohort, so the clusters are a lens for risk rather than a validated score to apply to an individual.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.