Psychosocial work stressors and depressive symptoms in bank employees

Dear Sir,

Valente et al. investigated the association between exposure to psychosocial work stressors and depressive symptoms in bank employees [1]. The authors used the Patient Health Questionnaire-9 (PHQ-9), and two levels of depressive symptoms: major depressive symptoms (MDS) and other forms of depressive symptoms (ODS) were specified. A logistic regression model was used to estimate associations between the level of depressive symptoms and scores from the two job stress models (Demand-Control-Support and Effort-Reward Imbalance). I have a query on their study.

The authors analysed 1046 participants from 2806 eligible subjects (37%) and concluded that psychosocial conditions in bank employees with high strain, low social support at work, high effort with low reward and over-commitment represented possible risk factors for MDS and/or ODS. They adopted PHQ-9 as a screening tool for depression according to Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I). Allgaier et al. reported the screening ability of PHQ-9 by two procedures—the ‘categorical scoring procedure’ and the ‘dimensional scoring procedure’ by using cut-off values—and concluded that there was an advantage to using the ‘dimensional scoring procedure’ for PHQ-9 screening for clinical depression in children [2]. Recent reports also specified the advantage of ‘dimensional scoring procedure’ by setting the cut-off value of 10 [3,4]. I previously reported the relationship between overtime work and depression by using the PHQ-9 [5], and I found that half the number of subjects with two to four of nine depressive symptom criteria were not categorized as subjects with ODS, because they lacked depressed mood and/or anhedonia. Valente et al. adopted a ‘categorical scoring procedure’, and the screening procedure should be taken into account for their study outcome.

Relating to the first query, the discrepancy of odds ratios of effort-reward imbalance for ODS and MDS should be considered precisely. Although the authors speculated the interaction between psychosocial factors and other work-related variables, there is space for making a final conclusion. The authors described in the footnote of Tables 2 and 3 that they adopted multivariate multinomial logistic regression model and this model handles two or more dependent variables. I suppose that the statistical model would clarify the U-shaped relationship between three levels of depressive symptom and psychosocial factors. Anyway, I recommend the authors specify the non-linear relationship between depressive symptoms by PHQ-9 and psychosocial factors.