Delineating the third age: joint models of older people's quality of life and attrition in Britain 2002–2010

Figures & data

Table 1. Description of baseline sample ELSA wave 1.

Covariate	Mean	SD	Covariate	Mean	SD
Age	62.1	8.1	Limiting illness
Sex			None (68%)	45.3	7.1
Male (46%)	42.5	8.6	Yes (31%)	37.5	9.1
Female (53%)	43.1	8.6	CESD 8 items	1.43	1.9
Wealth quintile			ADL, IADL 0-11	0.45	1.2
1 (12%)	41.0	9.4	CVD
2 (19%)	41.8	8.9	0 (52%)	44.0	8.0
3 (21%)	42.0	8.8	1 (47%)	41.6	9.0
4 (22%)	43.2	8.2	COPD
5 (24%)	45.1	7.4	0 (93%)	43.2	8.4
Education			1 (6%)	37.9	9.7
<degree (87%)	42.5	8.7	Cancer
Degree (12%)	45.3	7.3	0 (94%)	42.9	8.6
Cantril-like ladder	58.1	17.2	1 (5%)	41.6	8.9
Occupational class			Smoking
Non-routine (59%)	44.1	8.0	Non-smoker (81%)	43.4	8.3
Routine (40%)	41.0	9.1	Smoker (18%)	40.5	9.6
Marital status			Drinking
Single (5%)	41.8	8.7	No/light drinker (70%)	42.3	8.8
Married/cohabiting (73%)	43.4	8.3	Drink daily (29%)	44.3	8.0
Separate (21%)	41.2	9.3	Exercise: rigorous
Soc. Support			No (79%)	42.2	8.8
0 (2%)	34.8	10.5	Yes (20%)	45.5	7.3
1 (98%)	43.0	8.5	Exercise: moderate
Social connection (0-18)	9.8	3.9	No (38%)	40.3	9.4
			Yes (61%)	44.5	7.6

Figure 1. CASP-19, social class and dropping out.

Table 2. Trajectories of CASP-19 in England 2002–2010.

	Initial	Social	Health	Behaviour	Joint
	Coef/SE	Coef/SE	Coef/SE	Coef/SE	Coef/SE
Sex, F	0.457†	1.155‡	1.283‡	1.349‡	1.438
	0.169	0.17	0.135	0.135	0.128
Depression era	−0.261	−0.313	−0.321	−0.31	−0.497
	0.803	0.721	0.58	0.575	0.581
Interwar cohort	−0.763	−1.057*	−0.711	−0.667	−0.744*
	0.507	0.458	0.372	0.369	0.364
War baby cohort	−0.337	−0.661*	−0.424	−0.397	−0.505*
	0.302	0.271	0.218	0.217	0.202
Age	1.594‡	1.637‡	1.180‡	1.114‡	1.112‡
	0.091	0.091	0.085	0.085	0.083
Age^2	−0.013‡	−0.012‡	−0.009‡	−0.008‡	−0.008‡
	0.001	0.001	0.001	0.001	0.001
Married/coh.		0.941‡	0.625*	0.557*	0.540*
		0.321	0.264	0.262	0.246
Separated		−0.1	0.366	0.347	0.264
		0.33	0.274	0.272	0.255
Widowed		0.578	0.982‡	0.941	0.908
		0.348	0.292	0.29	0.277
Lower-middle Q		0.753	0.439	0.408	0.337
		0.286	0.226	0.224	0.219
Middle Q		1.191‡	0.578*	0.565*	0.517
		0.286	0.226	0.224	0.216
Upper-middle Q		1.911‡	1.093‡	1.013‡	0.911‡
		0.292	0.231	0.229	0.220
Top Q		3.096‡	1.769‡	1.614‡	1.452‡
		0.311	0.246	0.244	0.233
Degree		0.038	−0.316	−0.489*	−0.442*
		0.243	0.192	0.191	0.181
Routine workers		−1.191‡	−0.561‡	−0.458‡	−0.432‡
		0.167	0.133	0.132	0.128
Cantril-like ladder		0.102‡	0.093‡	0.092‡	0.094‡
		0.003	0.002	0.002	0.008
Social support		2.093‡	1.874	1.871	1.946
		0.29	0.275	0.274	0.271
Social connections		0.187	0.176	0.170	0.175
		0.011	0.01	0.01	0.010
Household size		0.033	−0.068	−0.074	−0.124
		0.089	0.071	0.071	0.064
Limiting illness			−2.249	−2.150	−2.171
			0.083	0.083	0.083
ADL IADL			−1.062	−1.016	−0.922
			0.034	0.034	0.033
CES-D			−1.141	−1.127	−1.150
			0.02	0.02	0.021
CVD			−0.183	−0.183	−0.158*
			0.069	0.069	0.069
COPD			−0.440	−0.418	−0.347
			0.133	0.133	0.134
Cancer			−0.415	−0.400*	−0.439
			0.161	0.161	0.160
Smoker				−0.666	−0.672
				0.130	0.126
Alcohol daily				0.378	0.358
				0.089	0.089
Vigorous exercise				0.627	0.615
				0.086	0.086
Mod exercise				0.623	0.540
				0.069	0.070
Constant	−6.501*	−21.869	−4.034	−2.245	−1.200
	3.259	3.26	2.959	2.956	2.883
Random int.					3.618
					9.598
Age					6.377
					0.064
Age					−0.648
					0.010
Female					0.053
					0.029
Constant					−15.201
					0.093
	6.519	5.350	4.060	4.071	0.001
	0.137	0.227	0.204	0.201	0.001
	058	0.058	0.041	0.040	0.073
	0.004	0.005	0.004	0.005	0.001
	19.961	19.750	18.824	18.783	19.046
	0.194	0.194	0.188	0.187	0.186
	0.20	0.20	0.44	0.45	0.61
BIC	199437	197103	192044	191877	305209
N	30654	30651	30651	30651	30651

Significance: * 5%, 1%,

Table 3. Top panel: the peaks of the third age when attrition is assumed random in growth model versus when attrition in jointly modelled; bottom panel: the level-difference across cohorts in CASP-19, reference is the baby boomers cohort, based on the joint model.

Model	Peak age	Standard deviation	Upper peak
Growth curve	66.4	5.8	77.7
Joint model	67.0	5.8	78.3
	Difference in	Standard	Max
Cohort	CASP-19	deviation	difference
War baby	−0.51	0.20	−0.90
Interwar	−0.74	0.36	−1.46

Figure 2. Comparing trajectories of cohorts. Top panel: joint model (higher overall) accounts for attrition while growth curve model assumes attrition at random; bottom panel: both models account for attrition and show early baby boomers have higher CASP-19 scores than those born between the wars.