The results of the study indicate that male firefighters who organisational system were 8 times more likely to become depressed than those who did not. This result was consistently observed even after adjusting for age and job position. This is similar to other previous studies reporting a relationship between organisational justice and depression. A cohort study conducted with civil servants in England showed that adverse change system increased psychiatric risks [8], as well as the risk of poor self-rated health [9]. In addition, previous studies were conducted to examine the connection between job stress and depression in firefighters. As a result, a subscale of job stress related to role conflict (OR = 1.8) and lower self-esteem (OR = 5.8) was presented as a risk factor. In particular, firefighters working for 24 hours showed work stress factors similar to procedural injustice such as inter-group conflict (OR = 1.7) and role ambiguity (OR = 1.6) as risk factors for depression.

This study, however, found that the difficult physical environment of firefighters had a negative association on depression (OR = 0.20, 95% CI [0.04–0.92]). The public workers such as rescue workers [10] and police officers [11] are reported that they are relatively resilient to mental health problems as they have been trained to cope with critical incidents since the beginning of their employment. Notably, this can also be seen as a result of the healthy worker effect, which can introduce a bias. Firefighters are selectively placed because they are physically and mentally healthier than the general population. Moreover, it is generally known that firefighters experience extremely strong social bonding, referred to as a ‘brotherhood’; this shows that the basic organisational culture of firefighters is likely to attenuate the influence of job stress on depression [12].

On the other hands, it has been reported that subjectively perceived incidents are classically more influential than objective incidents or situations as the pathologic mechanism of depression [13]; a lack of coping skills for responding to this is known to be an important mechanism [14]. The results of this study, that organisational system felt by individuals acts as a stronger risk factor for depression than frequently experienced job stress caused by a difficult physical environment, contrast with the results of existing studies on the pathogenesis of depression.

However, while the mental prescription for physical incidents can be filled by superficial contemplation, an invisible injustice at organisational system is a problem that is difficult to actively address and resolve.

A qualitative study in Korea indicated that in organisational systems, firefighters’ complaints include low allowances, bottlenecked promotions, and a rigid corporate culture [2]. Using the same job stress questionnaire used in this study, research conducted by KOSS found that organisational system, as a job stress factor of firefighters, was significantly higher than that found in the general population [15].

KOSS was validated in Korea and then applied to various types of occupations, presenting reference values for Korean workers [7] and a discussion to understand items through correlation analysis among the subscales. And organisational system in KOSS is also referred to as organisational culture, showing a lack of reward and strong correlation (r = 0.67) with other KOSS subscales [16]. A previous study [17] noted effect-reward imbalance as a predictor of poor mental health. Moreover, the ‘lack of reward’ could have detrimental effects on self-esteem. A Japanese study similar to this one investigated risk factors of depression in firefighters and reported that low self-esteem was five times more relevant to depression among them. Meanwhile, relational injustice is a component of organisational injustice that most previous studies have dealt with [18,19]. Studies on the absence of sickness [20] or self-rated health [9] have focused mostly on relational injustice. This study differs from the existing research in that it focused on procedural injustice. In the Korean municipal worker environment, organisational system is realistically close to procedural injustice. In fact, the internal congestion of promotions for municipal workers is attracting attention from society. A survey [21] showing that promotion was influenced by personal connections internally supports the assumption that unfairness is widespread in the decision-making process of the municipal employment hierarchy. This study found that such a situation might act as a severe stressor for staff in the organisation, and, if chronic, could cause depression after a year. In contrast, research that examined the relevance to depression using the same KOSS survey of all workers in Korea unexpectedly showed that organisational system did not have a great influence on depression [16,18,22]. However, this study discovered the adverse effects of organisational system on mental health, similar to previous research findings on firefighters, instead of the above study targeting all workers in Korea. In particular, unlike the aforementioned studies, the strength of this study lies in its longitudinal design and clear temporal relationships, as well as in the maintenance of internal consistency by restricting subjects to a certain panel of firefighters.

Injustice at organisational system also differs according to type of health effects. Stansfeld reported that organisational injustice is a major factor in short-term absence due to illness in women and long-term absence in men [23].

In our study, female firefighters were excluded from the final study group. The duties of female firefighters were mostly administrative jobs, different from those of male firefighters. Furthermore, subjects who presented as depressed in the initial survey were excluded to avoid confounding results caused by medication, and because this could be an interruption of the natural course of the study.

There are several known risk factors for depression including young age, being unmarried, and low socioeconomic status [24]. In our results, a univariate analysis of the above known risk factors according to depression revealed that they did not show any statistically significant association with depression. The point prevalence of depression in our study was 13.4% at the initial survey and 9.1% at the follow up survey. In the general population of Korea, the point prevalence of depression is 5%, 2.5% for lifetime prevalence according to an interview-based survey [25]. Using CES-D, the same self-rating scale used in the present study, the point prevalence for males in the general population was found to be 6.5% [26]. A similar study of depression in firefighters in Taiwan found a point prevalence for depression of 5.4%, and 10.5% for PTSD. In the same study, current PTSD status was shown to be a significant predictor for current major depression (OR = 1.157) [27]. Among firefighters present at the World Trade Center incident (also called 9/11), the point prevalence of combined PTSD and depression was 16.1%, which is higher than the prevalence of depression alone (5.9%). Because of limitations in the self-rated scale, the prevalence of depression in our results could not be adequately discriminated from depression co-morbid with other mental disorders. However, the healthy worker effect likely attenuated symptoms, so the actual prevalence could be understated due to the effect of bias toward the null. Besides, the size of depression group is slightly small and might be limitation of our study design. Because of panel study, the size and spectrum of participant was small and narrow and the depression group as well.

Notabley, this study was designed as a longitudinal panel study design which have a strength that it could be catch-up the level and trend of change of variance in single panel at the dynamic view point among other longitudinal studies. The dynamic changes of variance and status should be plausible evidences of the policy and management in a homogeneous panel. In case of this study, there is additional study about the depression-recovered group who would be also important group in practical view point of job-fitness. The risk factor and natural course of mild self-rated depression of firefighter should be noticed for management of public health. On the other hands, panel study design has a limitation to show exact causal relationship than the other longitudinal study design. A longitudinal study with incident case at the follow-up survey would be most appropriate to evaluated the causal relationships rather than cross-sectional study.

In terms of keep validity of panel study, the attrition rate should be worried for overestimation of variance of retention group. In our study, the follow-up loss due to job circulation was 38 subjects (13%) out of the final panel. Subgroup analysis of the 38 lost follow-up participants and high group in physical environment was conducted to assess the internal consistency of the panel. Because the distribution of socioeconomic status was not different statistically from that of the final subjects, the original characteristics of the panel were maintained.

Furthermore, panel studies are vulnerable to the weaknesses of instruments, which could be memorized by the subjects in the panel and therefore present biased results in the final analysis. Therefore, the questionnaires were varied to prevent subjects from learning the pattern of answers. For instance, the 2 types of self-rated scales for depression (the BDI and CES-D) were administered but only the BDI was used at the time of the follow-up survey. However, the validity of the self-rated scales of the BDI was used to measure symptoms of depression, not clinical diagnosis. Comparing BDI scales was suggested to be valid, and there should be little discordance between the questionnaire and actual clinical diagnosis [28]. A survey of 120 psychiatrists found that 70.89% of them suggest the BDI, with their suggestion confirmed by clinical practices [29].