Change between study enrolment and month 12 | Correlation coefficient† | Lost at least 0.25 HAZ* | Gained at least 0.25 HAZ* | ||||

Adjusted RRR‡ | 95% CI | P values | Adjusted RRR† | 95% CI | P values | ||

Change in weight per kg | 0.55 | 0.46 | 0.39 to 0.56 | <0.001 | 2.15 | 1.82 to 2.54 | <0.001 |

Change in WAZ per z-score | 0.61 | 0.44 | 0.35 to 0.55 | <0.001 | 2.19 | 1.77 to 2.71 | <0.001 |

Change in WHZ per z-score | 0.15 | 0.80 | 0.69 to 0.92 | 0.002 | 1.10 | 0.95 to 1.27 | 0.19 |

Change in MUAC per cm | 0.46 | 0.69 | 0.60 to 0.80 | <0.001 | 1.46 | 1.28 to 1.68 | <0.001 |

*Compared with children with minimal change in HAZ (±0.25Z).

†Spearman’s rank correlation coefficient.

‡Adjusted for co-trimoxazole randomisation arm, child carer, born premature (<37 weeks), height-for-age at study enrolment, outpatient treatment for diarrhoea and readmission during follow-up for diarrhoea. Relative risk ratios are computed using multinomial logistic regression with minimal HAZ change (−0.25 to 0.25Z) as the reference. P values from multivariable multinomial logistic regression.

HAZ, height/length-for-age z-score; MUAC, mid-upper arm circumference; RRR, relative risk ratio; WAZ, weight-for-age z-score; WHZ, weight-for-height/length z-score.