. 2020 Mar 6;36(5):1045–1053. doi: 10.1007/s13187-020-01732-2

Table 1.

Logistic regression analyses on determinants associated with infrequent or none use of sunscreen

		Chi² test		Logistic regression model
		Infrequent or none use of sunscreen		Model I		Model II		Model III		Model IV
		%	p value	OR	p value	OR	p value	OR	p value	OR	p value
Sociodemographic variables
Sex			< 0.001
	Male	24.7		1.000						1.000
	Female	16.4		0.595	< 0.001					0.610	< 0.001
Age groups			< 0.001
	14–17	22.0		1.000						1.000
	18–25	18.6		0.804	0.227					0.713	0.057
	26–35	14.6		0.612	0.008					0.537	< 0.001
	36–45	28.0		1.471	0.034					1.101	0.567
Immigrant background			< 0.001
	No	19.1		1.000						1.000
	Yes	27.0		1.523	< 0.001					1.625	< 0.001
Education			0.010
	Low	20.2		1.000
	Medium	23.6		1.154	0.288
	High	18.2		0.872	0.329
Employment			0.591
	None	21.5
	Full-time/part-time	20.5
Partnership			0.752
	No	21.0
	Yes	20.5
Skin cancer risk
Skin type			< 0.001
	I/II	12.7				1.000				1.000
	III–VI	24.9				2.192	< 0.001			2.127	< 0.001
Sunburn before the age of 15			0.327
	Rarely/ Do not know	20.9
	Often	18.1
More than 40 birthmarks			< 0.001
	No/do not know	22.9				1.000				1.000
	Yes	15.7				0.690	< 0.001			0.644	< 0.001
Family history of malignant melanoma			0.002
	No/do not know	21.4				1.000				1.000
	Yes	13.5				0.609	0.006			0.587	0.004
History of malignant melanoma			0.371
	No/do not know	20.8
	Yes	16.9
Tanning behavior
Sunbathing in summer			< 0.001
	Rarely/ never	28.6						1.000		1.000
	Sometimes	12.9						0.373	< 0.001	0.358	< 0.001
	Very often/ often	19.3						0.594	< 0.001	0.550	< 0.001
Sunbed use			< 0.001
	Never	18.4						1.000		1.000
	Past	28.3						1.707	< 0.001	1.574	< 0.001
	Current	23.5						1.434	0.023	1.458	0.023
n				2726		2977		2977		2956
r²_Nagelkerkes				0.060		0.046		0.052		0.139

Dependent variable: 1 = using sunscreen rarely or never, 0 = using sunscreen sometimes/often/very often

Models I–III: logistic regression models only included variables that were significant in bivariate analyses

Model IV: logistic regression model only included variables that were previously significant in models I–III

Data from the fourth wave of the National Cancer Aid Monitoring (NCAM)

n = 3000 individuals aged 14–45 years old living in Germany

Data weighted by age, sex, education, and federal state

OR odds ratio