Table 2.

Negative binomial growth model of the number of depressive symptoms (2002/2003–2014/2015)

Growth parameters	Initial modela			Fully adjusted modelb
	b	SE	p	b	SE	p
Intercept	− 0.409	0.031	< 0.001	− 0.735	0.043	< 0.001
Intercept regressed on
Female	0.388	0.022	< 0.001	0.351	0.021	< 0.001
Baseline age (years)c	0.023	0.002	< 0.001	0.010	0.002	< 0.001
Medium PSC	0.056	0.033	0.092	0.051	0.030	0.088
Low PSC	0.347	0.033	< 0.001	0.235	0.029	< 0.001
Baseline age × medium PSC	− 0.004	0.003	0.193	− 0.005	0.003	0.097
Baseline age × low PSC	− 0.011	0.003	< 0.001	− 0.011	0.003	< 0.001
Sloped	− 0.091	0.048	0.060	− 0.068	0.070	0.335
Slope regressed on
Female	0.017	0.034	0.621	− 0.007	0.034	0.835
Baseline age (years)	0.015	0.004	< 0.001	0.017	0.004	< 0.001
Medium PSC	0.069	0.048	0.154	0.077	0.047	0.103
Low PSC	0.102	0.046	0.027	0.119	0.045	0.009
Baseline age × medium PSC	0.011	0.005	0.039	0.010	0.005	0.041
Baseline age × low PSC	0.014	0.005	0.003	0.013	0.005	0.007
Intercept variance	0.937	0.023	< 0.001	0.666	0.019	< 0.001
Slope variance	0.557	0.039	< 0.001	0.558	0.039	< 0.001
Intercept–slope covariance	− 0.147	0.026	< 0.001	− 0.127	0.023	< 0.001

^aAdjusted for the covariates shown in the table, plus white/non-white group and self-reported ever doctor-diagnosis of depression

^bAdjusted for the covariates shown in the table, plus white/non-white group, self-reported ever doctor-diagnosis of depression, educational qualification, work status, total non-pension wealth, self-rated health, and self-reported limiting long-term illness

^cCentred at 65 years

^dAs the time scale (years) was divided by 12, the average slope and effect of covariates on the slope describe the total rate of change over the 12-year study period