. 2020 Jun 10;156(8):1–9. doi: 10.1001/jamadermatol.2020.1602

Table 4. Multivariate Analyses of Dietary Patterns From a Principal Component Analysis of the Study Population Using a Multinomial Logistic Regression.

Dietary pattern^a	Multivariate analyses vs never acne^b
	Past acne		Current acne
	aOR (95% CI)	P value^c	aOR (95% CI)	P value^c
Healthy	0.98 (0.94-1.01)	.17	0.88 (0.83-0.94)	.001
Fatty and sugary	1.05 (1.01-1.08)	.006	1.13 (1.05-1.18)	<.001
Animal products and refined cereals	0.98 (0.95-1.01)	.24	0.99 (0.92-1.07)	.87

Abbreviation: aOR, adjusted odds ratio.

^{^a}

Healthy indicates frequent consumption of fruit, vegetables, and fish; fatty and sugary, frequent consumption of fatty and sugary products, including chocolate; and animal products and refined cereals, frequent consumption of animal products (eg, meat and milk) and refined cereals.

^{^b}

Analysis included 11 068 never acne participants (54%), 7983 past acne participants (39%), and 1361 current acne participants (7%). Data are missing for some variables used in adjustment; therefore, the sample number (20 412) differed from the overall population.

^{^c}

P values were determined in a multivariate logistic regression after adjustment for total calorie intake, the number of dietary records completed, sex, age, smoking status, physical activity, educational level, body mass index, and history of cancer, diabetes, cardiovascular disease, or depressive symptoms.