The Research on a Predictive Model of Depression among College Students
|Keywords||college students depression PHQ-9 DSM-IV Logistic regression cognitive factors|
This research mainly studied the risk factors of depression among college students, and a Logistic regression model was built to predict the probability of how much is the chance that a student would get obvious depression after 6 month. The difference with previous studies is that we tried to build a multi-factors model to predict depression in 6 months, and also to see how these factors affect depression. Firstly, we choosed 13 risk factors as below:population data(sex, major, grade, economical condition, race/living place and religion belief, family conditions and psychological heritable factors), personality(EPQCRS), physical and mental health(SF-12), life events(ASLEC), abnormal experiences and discrimination in childhood(added to ASLEC), attributive styles(Internal-External Locus of Control Scale), negative automatic thoughts(ATQ), coping methods(Middle School Students’Coping Scale), and perceived social support(PSSS). At last, there are 7 factors entered the logistic regression equation. The results can be demonstrated as:(1) 1427 college students in Hangzhou participated in this research. With the depression detection tool—Patients Heath Questionnaire-9(PHQ-9),we found 153(12.60%) students had obvious depressive mood, and the re-test reliability of this questionnaire is 0.609(significant at 0.01 level). Also the correlation of PHQ-9 with SF-12 mental dimension is 0.450(significant at 0.01 level), which proved the compatibility validity of PHQ-9 is good. And with the《Diagnostic and Statistic Manual of mental disorders》(DSM-Ⅳ)formulate by APA, we defined the Minor Depression(Mind) diagnostic method as criteria to study the sensitivity and speciality, which respectively was 94.44% and 90.91%. And by making Receptive Operation Curve(ROC), we found the best cut off score of PHQ-9 to test depression is 10. So PHQ-9 is a brief, reliable, convenient, and time-saving tool in college students.(2) To do the correlation test of the 13 factors, results showed that except part of the population data was insignificantly related with PHQ-9, and SF-12-physical division was significantly related at 0.05 level, other factors as negative automatic thoughts, personality, attributive factor, social support, coping methods, life events and childhood experiences, mental health condition, mental disorders history, family heritability, and only-child or not, these factors were all significantly related with PHQ-9 at 0.01 level. And there was also interrelation existing between factors.(3)Based on the screen score of PHQ-9 after 6 months, we builded the Logistic model through 11 correlative factors and the regression model showed that 7 factors entered our equation:negative automatic thoughts(B=0.025), attribution(B=0.069), perceived social support(B=0.029), life events and childhood experiences, personality-Psychoticism division(B=0.022), mental disorders history(B==1.209) mental and physical health condition(B-mental=-0.099,B-physical=-0.035). And there were 3 inconstant factors and 2 constant factors. To estimate the observed values by this logistic model, the correct percentage was as high as 90.9%.So after this study, we find the negative automatic thoughts, attribution methods, perceived social support, personality-Psychoticism division, mental and physical health, and metal disorders history are all predictive factors to depression of college students. And the first 3 factors are all cognitive factor, so this results also give psychotherapy some reference by acquiring a predictive probability of how many chance a student will depressed in a certain time.