Table 2.

Comparison of Preemie Tots CDI scores in treatment and placebo groups across the study

	Baseline		Final visit		Change from baseline to final visit
	Treatment (n = 14)	Placebo (n = 13)	Treatment (n = 12)	Placebo (n = 12)	Treatment	Placebo
CDI words and gestures	12.5 (2.4)	12.1 (1.4)	12.9 (1.5)*^	11.0 (2.3)^	0.4τ	−1.1τ
CDI gestures	9.6 (2.3)	9.1 (1.4)	10.9 (1.4)*	9.0 (2.3)	1.3	0.1
CDI words	40.7 (37.7)	31.0 (28.2)	45.8 (33.4)^	39.7 (30.4)^	5.1τ	8.7τ
Deictic gestures	4.8 (0.4)	4.5 (0.7)	5.0 (0)	4.8 (0.5)	0.2	0.3
Representational gestures	5.3 (1.4)	4.7 (1.8)	6.2 (1.1)*	4.3 (2.3)	0.9	−0.4
Probability of having all deictic gesturesZ	0.8 (0.2)	0.8 (0.2)	0.8 (0.2)	0.8 (0.2)	0	0

All values represent the count of words or gestures or both except the probability of having all deictic gestures

CDI MacArthur-Bates communicative development inventory

^*Significantly different from the placebo group at the same time point (p < 0.05).

^{^}Significantly different from baseline in the same group (p < 0.05)

^τSignificantly different change from baseline to final visit

^ZAt baseline, one toddler in the placebo group had only three deictic gestures, and three toddlers in the placebo group and two toddlers in the treatment group had only four deictic gestures. The rest had all five. At the final visit, all children in the treatment group had all five deictic gestures, and two children in the placebo group only had four deictic gestures. Therefore, the model was run after first creating a binary variable where 0 = did not have all deictic gestures and 1 = had all deictic gestures. The likelihood of having all deictic gestures was then predicted by weeks of completed gestation, BITSEA ASD scores, and PDDST scores. The likelihoods were output as a separate variable and predicted by time and treatment group with the original mixed model. The values in the table represent the likelihood that a child in each group at each data collection time had all deictic gestures