Table 1 provides descriptive statistics of all variables used in this study per cohort. Descriptive statistics of these variables stratified by sex are presented as a supplementary material in Table S1.
![](https://myfiles.space/user_files/58653_1b1c6aeb34a62c68/58653_custom_files/img1602523137.jpg)
High depressive symptoms (PHQ-2 (3-6)) were reported by 23.0% of French students (15.4% for males and 25.0% for females), and by 3.5% of Japanese students (3.5% for males and 3.3% for females). Poor self-rated health was reported by 2.8% of French students (2.3% for males and 2.9% for females), and by 4.5% of Japanese students (5.0% for males and 2.9% for females). Other major differences between countries concerned regular physical activity (87.5% of French students versus 66.7% of Japanese students), smoking (24.9% of French students versus 2.7% of Japanese students), and alcohol consumption (92.3% of French students versus 57.5% of Japanese students).
Table 2 presents the univariate associations between PHQ-2 score and all other variables including self-rated health. Results stratified by sex are presented as a supplementary material in Table S2.
![](https://myfiles.space/user_files/58653_1b1c6aeb34a62c68/58653_custom_files/img1602523178.jpg)
![](https://myfiles.space/user_files/58653_1b1c6aeb34a62c68/58653_custom_files/img1602523210.jpg)
There was a similar pattern of association between PHQ-2 and self-rated health in both cohorts. Poor self-rated health was significantly and positively associated with high depressive symptoms in both French and Japanese students (p<0.001). 6.2% of French students with high depressive symptoms (PHQ-2 score (3-6)) and 1.7% of students with low depressive symptoms (PHQ-2 score (0-2)) reported poor self-rated health. On the other hand, 31.7% of Japanese students with high depressive symptoms (PHQ-2 score (3-6)) and 3.5% of students with low depressive symptoms (PHQ-2 score (0-2)) reported poor self-rated health. Concerning the detection of depressive symptoms by self-rated health, sensitivity was 6.2% and 31.7% in French and Japanese students, and specificity was 98.3% and 96.5% in French and Japanese students, respectively. Similarly, bad sleep quality, BMI<18.5, BMI≥30 and smoking habit were associated with high depressive symptoms in both cohorts with p values comprised between 0.049 and <0.001. Never drinking alcohol was associated with high depressive symptoms only among French students (p<0.001).
On the other hand, the two cohorts reported different patterns of association between PHQ-2 and sex, age, year of study, and physical activity. Female sex was associated with high depressive symptoms in the French cohort (p<0.001), whereas there was no sex difference in the Japanese cohort (p=0.466). French students with high depressive symptoms were younger than those with low depressive symptoms (p<0.001), whereas Japanese students with high depressive symptoms were older although the association was not significant (p=0.064). Moreover, the 1st undergraduate year was associated with high depressive symptoms in French students (p<0.001), whereas the 1st undergraduate year was associated with low depressive symptoms in Japanese students (p<0.001). Less physical activity was associated with high depressive symptoms in Japanese students (p<0.001), whereas this association was not significant in French students (p=0.448).
Finally, Table 3 illustrates the results of the multivariate logistic regression analyses. Results stratified by sex are presented in Table S3.
![](https://myfiles.space/user_files/58653_1b1c6aeb34a62c68/58653_custom_files/img1602523357.jpg)
![](https://myfiles.space/user_files/58653_1b1c6aeb34a62c68/58653_custom_files/img1602523368.jpg)
Poor self-rated health was significantly associated with high depressive symptoms in both cohorts although the level of the point estimate was higher in Japanese students (i-Share: OR 2.82, 95%CI 1.99-4.01, p<0.001, Kyoto: OR 7.10, 95%CI 5.76-8.74, p<0.001). In addition, bad sleep quality (i-Share: OR 2.30, 95%CI 1.98-2.67, p<0.001, Kyoto: OR 2.94, 95%CI 2.44-3.53, p<0.001), BMI≥30 (i-Share: OR 1.48, 95%CI 0.99-2.21, p=0.075, Kyoto: OR 2.21, 95%CI 1.34-3.64, p=0.015) and never drinking alcohol (i-Share: OR 1.42, 95% CI 1.12-1.80, p=0.003, Kyoto: OR 1.44, 95%CI 1.20-1.73, p<0.001) were also associated with high depressive symptoms in both cohorts. Smoking was not significantly associated with PHQ-2 in both cohorts.
Female sex was associated with high depressive symptoms in French students (OR 0.61, 95%CI 0.51-0.73, p<0.001), but not in Japanese students (OR 0.96, 95%CI 0.78-1.18, p=0.681). French students with high depressive symptoms were younger (OR 0.95, 95%CI 0.92-0.98, p=0.001), whereas there was no association between age and PHQ-2 in Japanese students (OR 1.00, 95%CI 0.97-1.02, p=0.645). In addition, years other than the 1st undergraduate year were associated with high depressive symptoms in Japanese students (OR 2.40, 95%CI 1.65-3.48, p<0.001), whereas 1st undergraduate year was associated with high depressive symptoms in French students (OR 0.85, 95%CI 0.73-1.00, p=0.056) although the association was not significant. BMI<18.5 was associated with high depressive symptoms in French students (OR 1.20, 95%CI 1.00-1.45, p=0.075), whereas not in Japanese students (OR 1.03, 95%CI 0.80-1.32, p=0.015). Less physical activity was associated with high depressive symptoms in Japanese students (OR 1.35, 95%CI 1.13-1.61, p=0.001), but not in French students (OR 1.06, 95%CI 0.87-1.29, p=0.592).
As shown in Table S3, there were some different results depending on sex. Among French males, no significant association between PHQ-2 and self-reported health was identified (OR 2.17, 95%CI 0.86-5.47, p=0.100). The correlation between self-rated health and depressive symptoms, as well as the score of PHQ-2 and the numbers and percentages of students with elevated depressive symptoms in each five categories of self-rated health are reported as Supplemental Material.
Finally, since age and year of study were correlated, we performed a sensitive analysis in both cohorts by running models with either age or year of study, and we observed that results were the same across all models. Moreover, since we found different interactions in each cohort, we performed another sensitive analysis to take interactions into models respectively and confirmed that the final results did not change.