Table 3.
Results of regression analyses comparing outcomes of CANTAB tests for SAM survivors v. controls, seven years after surviving an episode of SAM, in a cohort of Malawian children
| CANTAB outcome† | SAM survivors (n 171) | Controls (n 155) | SAM survivors v. controls, adjusted only for age | SAM survivors v. controls, adjusted for age, sex, HIV, SES | SAM survivors v. controls, adjusted for HAZ, age, sex, HIV, SES | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Mean | sd | Mean | sd | Difference | 95 % CI | P | Difference | 95 % CI | P | Difference | 95 % CI | P | |
| BLC % correct | 94·0 | 13·1 | 97·8 | 5·4 | −4·02 | −6·5, −1·6 | 0·001* | −3·35 | −6·1, −0·6 | 0·02* | −1·78 | −4·1, 0·5 | 0·13 |
| IED total errors (adjusted) | 93·7 | 81·5 | 75·3 | 77·3 | 18·34 | −1·1, 37·7 | 0·06 | 6·79 | −13·5, 27·1 | 0·51 | 3·75 | −16·8, 24·3 | 0·72 |
| MOT mean error | 10·0 | 2·8 | 9·8 | 2·9 | 0·15 | −0·5, 0·8 | 0·64 | 0·31 | −0·4, 1·0 | 0·37 | 0·29 | −0·4, 1·0 | 0·41 |
| MOT mean latency (ms) | 1347 | 502 | 1257 | 423 | 86·4 | −16·0, 188·8 | 0·09 | 52·9 | −60·0, 166·0 | 0·36 | 22·3 | −90·2, 134·8 | 0·70 |
| PAL total errors (adjusted) | 111·1 | 70·4 | 96·3 | 71·7 | 13·9 | −1·5, 29·4 | 0·08 | 8·04 | −8·8, 24·9 | 0·35 | 5·57 | −11·4, 22·5 | 0·52 |
| PAL total errors (six shapes, adjusted) | 31·2 | 19·7 | 28·4 | 20·8 | 2·61 | −1·8, 7·1 | 0·25 | 1·23 | −3·6, 6·1 | 0·62 | 0·59 | −4·3, 5·5 | 0·81 |
| PRM % correct | 63·6 | 16·0 | 69·5 | 16·3 | −4·36 | −8·0, −0·71 | 0·02* | −3·73 | −7·6, 0·1 | 0·06 | −3·6 | −7·5, 0·3 | 0·07 |
| IED total stages completed (ordered logistic) | 5·77 | 3·5 | 6·5 | 3·3 | −0·46 | −0·9, −0·02 | 0·04* | −0·32 | −0·8, 0·2 | 0·21 | −0·24 | −0·8, 0·3 | 0·35 |
CANTAB, Cambridge Neuropsychological Testing Automated Battery; SAM, severe acute malnutrition; SES, socio-economic status; HAZ, height-for-age Z-score; BLC, Big/Little Circle; MOT, Motor Screening Test; PAL, Paired Associated Learning; PRM, Pattern Recognition Memory; IED, Intra/Extradimensional Set Shift.
Indicates significant difference (P < 0·05). Test outcomes quantifying the number of total errors are adjusted for incomplete tests, as participants who fail at earlier stages of the test have fewer opportunities to make errors.
Linear regression used unless otherwise stated.