Table 2: Effect of Different Assumptions About Missing Outcome Data on the Effectiveness of an Early Childhood Caries Prevention Programme

Assumptions About Missing Outcomes Number with S-ECC* Outcome Parameters (95% Confidence Interval)
Intervention Control Relative Risk Risk Difference NNT **
All lost to follow-up have S-ECC 84 124 0.70 (0.56 - 0.88) 0.12 (0.04 - 0.19) 8 (5 - 25)
None lost to follow-up have S-ECC 4 20 0.21 (0.07 - 0.60) 0.05 (0.02 - 0.08) 20 (13 - 50)
All have S-ECC frequency as intervention group 5 22 0.24 (0.09 - 0.61) 0.05 (0.02 - 0.08) 20 (13 - 50)
All have S-ECC frequency as control group 12 30 0.41 (0.22 - 0.79) 0.06 (0.02 - 0.09) 17 (11 - 50)
All have S-ECC frequency as entire population 8 26 0.32 (0.15 - 0.69) 0.06 (0.02 - 0.09) 17 (11 - 50)
Multiple imputation of missing outcomes 7 29 0.25 (0.11 - 0.56) 0.07 (0.03 - 0.10) 14 (10 - 33)

* Cases of severe early childhood caries (S-ECC) that would have been observed in the intervention (n = 307) and control group (n = 318) with mother-infant pair as randomised, if the assumption about missing outcomes would be correct.
** Number Needed to Treat to prevent one case of S-ECC.