Table 3.
Odds ratios [OR] for the univariable associations of ICD10 depression at 18 years with previous depressive symptoms and indicators of socioeconomic position
| SEP indicator | N | OR 95% CI | OR 95% CI multiple imputations | OR 95% CI multiple imputations log[IMOR] = 1a | OR 95% CI multiple imputations mutually adjustedb |
|---|---|---|---|---|---|
| Previous depressive symptoms | 4563 | 7.18 [5.58 9.24] | 6.85 [5.20 9.05] | 7.20 [5.60 9.26] | 5.66 [4.19 7.64] |
| Manual social class | 3984 | 1.20 [0.94 1.52] | 1.16 [0.91 1.47] | 1.47 [1.35 1.61] | 0.99 [0.74 1.33] |
| Rented accommodation | 4176 | 1.58 [1.20 2.09] | 1.47 [1.11 1.95] | 1.63 [1.45 1.82] | 1.32 [0.94 1.83] |
| Major financial problems | 3873 | 1.45 [1.04 2.01] | 1.44 [1.04 1.99] | 1.08 [0.95 1.24] | 1.02 [0.70 1.49] |
| No car access | 4173 | 1.54 [0.99 2.42] | 1.36 [0.88 2.11] | 1.60 [1.33 1.93] | 0.97 [0.58 1.62] |
| Material hardship | 4045 | 1.74 [1.35 2.24] | 1.73 [1.34 2.24] | 1.47 [1.32 1.63] | 1.57 [1.16 2.12] |
| Maternal education | 4152 | ||||
| ≥ A-level | 1 | 1 | 1 | 1 | |
| O-level | 1.27 [1.00 1.63] | 1.24 [0.96 1.60] | 1.12 [1.02 1.22] | 1.20 [0.89 1.61] | |
| CSE/vocational | 1.11 [0.82 1.52] | 1.10 [0.80 1.53] | 1.65 [1.49 1.84] | 0.99 [0.68 1.47] | |
aThe NMAR assumption is quantified through a missing not at random OR (IMOR). IMOR (informative missing OR) corresponds to the difference: delta = logit(Prob(case diagnosed if missing) − logit(Prob(case diagnosed if observed). IMOR of 0.1 would indicate a very strong association between being a ‘case’ (i.e. having depression) and missing data. A very small value of log(IMOR) close to 0 would represent an assumption of missing at random. The sensitivity analyses also assumed that missing data dependencies are distributed at random among those with low and high SEP
bAdjustment also included gender