. 2021 Nov 23;4(11):e2136113. doi: 10.1001/jamanetworkopen.2021.36113

Table 3. Logistic Regression Models for Association Between Social Media Use at Initial Survey and Clinically Significant Increase in PHQ-9.

Platform	Full cohort, OR (95% CI)^a	P value for interaction with age	OR (95% CI)
Platform	Full cohort, OR (95% CI)^a	P value for interaction with age	≥35 y	<35 y
Facebook	1.42 (1.10-1.81)	.04	1.12 (0.85-1.48)	2.60 (1.46-4.62)
Instagram	1.18 (0.95-1.46)	.17	NA	NA
LinkedIn	1.03 (0.80-1.33)	.08	NA	NA
Pinterest	0.94 (0.76-1.16)	.20	NA	NA
TikTok	1.39 (1.03-1.87)	.01	1.67 (1.14-2.45)	1.36 (0.86-2.16)
Twitter	1.05 (0.84-1.31)	.08	NA	NA
Snapchat	1.53 (1.19-1.96)	<.001	1.96 (1.44-2.65)	1.17 (0.78-1.77)
YouTube	1.16 (0.93-1.44)	<.001	1.31 (1.03-1.67)	0.68 (0.41-1.11)

Abbreviations: NA, not applicable; OR, odds ratio; PHQ-9, 9-item Patient Health Questionnaire.

^{^a}

Cohort includes all participants with PHQ-9 scores less than 5 at first survey who responded to a subsequent survey. Age-stratified analysis was completed only if age-by-platform interaction was nominally significant.