Incorporating frailty to address the key challenges to geriatric economic evaluation

BMC Geriatrics

Table 6 Logistic regression for engaging in regular unpaid work

Dependent variable: Unpaid work¹in Wave 5 (N = 6,205)
Explanatory variables	Coefficient (SE)²	P-value
Constant	-12.856 (2.951)	< 0.001
Age W4	0.331 (0.083)	< 0.001
Age^2 W4	-0.002 (0.0006)	< 0.001
Female	0.313 (0.065)	< 0.001
SES (ref: Most privileged quartile)
2nd quartile	-0.266 (0.095)	0.005
3rd quartile	-0.233 (0.080)	0.004
Most deprived quartile	-0.236 (0.098)	0.016
Frailty W4 (0-100)	-0.013 (0.005)	0.010
Change in frailty³	-0.012 (0.006)	0.039
Cognitive impairment W4	-0.379 (0.091)	< 0.001
Abnormal gait/balance W4	-0.299 (0.097)	0.002
Unpaid work¹ W4	1.944 (0.065)	< 0.001

¹ ELSA W4-5 contained information on the frequency of ‘formal’ volunteering activities (i.e., as part of a volunteering organisation) in the past 12 months: at least once a week; less than once a week; and one-off. Similar frequency data was reported for provision of unpaid help (i.e., volunteering on a less formal basis), including informal caregiving for sick persons, childcare, and helping people with daily activities such as cooking, cleaning, and transporting. Together, they constituted unpaid work performed by older persons. A binary variable was created to indicate weekly or more regular unpaid work.
² Coefficient greater than zero implies the explanatory variable increased the odds of the dependent variable relative to its reference level, and vice versa.
³ Two-year change in frailty between ELSA W4 and W5.
Abbreviation: ELSA: English Longitudinal Study of Ageing; MA fall: fall requiring medical attention; Ref: reference; SE: standard error; SES: socioeconomic status; W4: ELSA Wave 4; W5: ELSA Wave 5

ISSN: 1471-2318