This study evaluated the explanatory power of an operationalized typological definition for burnout syndrome using the “Burnout Clinical Subtype Questionnaire”, in its long (BCSQ-36) and short (BCSQ-12) versions [19, 22], regarding the standard offered by MBI-GS [6, 47]. Multiple regression analysis enabled us to see that the dimensions of the MBI-GS were captured by the BCSQ-36 and BCSQ-12 subscales, with an adequate fit. Moreover, the distribution of residuals was approximately normal and no autocorrelation problems were detected.
As limitations, we should not overlook the fact that participant assessments were self-reported, and therefore may be weakened by socially desirable responses. Equally, the response rate obtained may seem low, although these values were similar to those found in other studies using similar on-line data collection procedures [45, 46], and they enabled a sample size to be obtained that was not far off that initially estimated to be necessary, contributing evidence in relation to the aims originally set out. It should be pointed out that the distribution of the response rate was uneven for occupational strata, which could lessen the generalizability of our results. Finally, test-retest measurements were not taken for the variables under study, and therefore this aspect of their reliability could not be quantified. Nevertheless, we consider that the strength of this study lies in the work carried out with a broad and multi-occupational sample of employees in at-risk occupations with face-to-face personal contacts, in jobs with very different characteristics, which allows our conclusions to be generalized. Additionally, data quality was controlled by eliminating possible errors in the questionnaire transcription process through the use of purpose-designed software.
As we have explained previously, BCSQ-36 and BCSQ-12 were able to explain a high proportion of the variability contained in the criterion dimensions of the standard MBI-GS, although they were significantly higher in BCSQ-36, as we had established initially as a working hypothesis. All the dimensions of both typological models showed adequate internal consistency, and were significantly associated with some of the criterion dimensions of the standard on an individual basis. On the whole, the dimensions of the long and short typological models contributed to the explanation of each of the classic dimensions according to the proposed hypothesis, given that the “frenetic” profile presented the dimension that contributed most to the explanation of 'exhaution', as did the “underchallenged” profile with ‘cynicism’ and the “worn-out” profile with ‘efficacy’. However, as can be seen, the pattern of contributions obtained was somewhat more complex than initially expected.
First, ‘overload’ and ‘lack of control’ were the dimensions that basically explained ‘exhaustion’, something that is coherent with the Karasek’s demand–control model , according to which psychological strain is caused by the combination of high demands and low control. This result is also in line with the areas of worklife model , according to which workload and lack of control are important correlates of the syndrome, and with the more recent demands-resources model , in which personal resources are more important when coping with work-related demands. All of this is congruent with the process of stress caused by lack of control over results and over decision-making, with the association established between excess work and the appearance of fatigue and low levels of empathy, and with the development of emotional disorders caused by chronic stress [53–58]. We see that ‘lack of control’ contributed to the explanation of all the criterion dimensions, and that it can therefore be accepted as a key dimension when it comes to explaining the development of burnout symptoms in general, although it was in fact more strongly correlated with ‘exhaustion’.
On the other hand, ‘lack of development’ and ‘indifference’ were the dimensions that most contributed to explaining the criterion dimension of ‘cynicism’. Using Karasek’s framework with non-linear effects as proposed in a previous study , a manner of interpreting these results is that just as high demands may be overwhelming, or “toxic” to use Warr’s word , low demands may also be so unchallenging as to create feelings of frustration and monotony. This perspective is also included in the model by Schwab, Jackson and Schuler , which considers monotony to be an antecedent for the syndrome. Moreover, the ‘indifference’ variable contributed significantly to the explanation of all criterion dimensions, and therefore may be another key dimension for explaining the development of symndrome symptoms in general, although this variable was strongly correlated with ‘cynicism’ in particular, and both could eventually reduce satisfaction, interest and productivity in this subtype of workers [30, 61–63].
Finally, ‘neglect’ and lack of ‘ambition’ were the dimensions that best explained the factor of lack of ‘efficacy’. These variables have also traditionally been associated with low performance levels in Bandura's theory of perceived self-efficacy and lack of it may also cause difficulties when it comes to alleviating perceived stress [43, 64–66]. In general, it is understood that a progressive decrease in levels of engagement seems to be the kind of response adopted by burnout workers to cope with frustration, as described in the demand-resources model , and could be an important factor in explaining the differences between the subtypes from a longitudinal perspective [11–22]. These differences, explained by the BSCQ-36 and BCSQ-12 models by means of the degree of dedication to tasks as a criterion of typological classification, are not explained by previous models of burnout.
We have seen how that ‘overload’, ‘lack of development’ and ‘neglect’ variables of the BCSQ-12 contributed significantly to the explanation of ‘exhaustion’ and ‘cynicism’; however, of these three variables in BCSQ-36, only ‘overload’ contributed to that of ‘exhaustion’ and only ‘lack of development’ contributed to ‘cynicism’. This apparent inconsistency is the result of the control exerted by a number of variables over others when included together in the regression model. This effect can be understood if we observe that, while on a bivariate level significant correlations were obtained between the referred to independent and dependent variables (and generally between most of the variables under study), the ‘lack of development’ and ‘neglect' variables in the BCSQ-36 regression model did not provide new information on ‘exhaustion’ than that provided by the other variables. Likewise, no new information was provided by the ‘overload’ and ‘neglect’ variables on ‘cynicism’ in the BCSQ-36 regression model. This effect is clear if we observe the values provided by the partial and semi-partial correlation coefficients (Ry3.12 and Ry(3.12)). As previously mentioned, this is due to the information that could have been added in both cases being contained in the ‘indifference’, ‘lack of acknowledgement’ and ‘lack of control’ variables. We have already mentioned that ‘indifference’ and ‘lack of control’ could be dimensions with great explanatory power over all the classic symptoms, so they should perhaps be taken into account generally in the design of any intervention on the syndrome. However, ‘lack of acknowledgement’ was more important for explaining ‘exhaustion’ and ‘cynicism’ and not so much for lack of ‘efficacy’. These apparent inconsistencies did not occur in the models in relation to the ‘efficacy’ dimension, given that in both BCSQ-12 and BCSQ-36 the ‘overload’ and ‘lack of development’ variables did not contribute significantly to expaining it. In this case, 'neglect' was seen to be the dimension with the greatest explanatory power over lack of ‘efficacy’.
As we have seen, the explanatory power of BCSQ-36 was high and significantly greater than that of BCSQ-12. Given its length, complexity and the information it contributes, this questionnaire could be a very suitable instrument for use in mental health services, facilitating the design of interventions adapted to the characteristics of each particular case. For example, the “frenetic” subtype may benefit more from an intervention focusing on decreasing levels of activation, distress and fatigue. On the other hand, the “underchallenged” subtype may need to recover interest and enthusiasm to regain satisfaction and meaning with regard to the tasks assigned. Finally, the “worn-out” subtype needs to address feelings of hopelessness, lack of perceived efficacy and sense of abandonment at work. The source of the discomfort experienced in each subtype of burnout seems to come from very different coping strategies and dysfunctional attitudes based on the level of dedication at work . In general, this approach is more in tune with how clinicians group symptoms and define disorders, something which may facilitate the use of specific forms of therapy. As Kokkinos  points out, the fact that each dimension of the syndrome is predicted by different variables should not remain unnoticed especially when designing and implementing intervention programmes to reduce burnout.
BCSQ-12 was also seen to have high explanatory power, very close to that of the long version. Given its brief and functional nature, and by making use of the already proposed cut-off points , it could be a very useful screening instrument in primary care consultations. In other words, this questionnaire could provide detection and recognition of burnout syndrome in cases where a commorbid association with anxiety, depressive or psychosomatic symptoms could lead to latent work-related psychosocial problems being overlooked . We have seen that the subscales of ‘overload’, ‘lack of development’, and ‘neglect’ that comprise the BCSQ-12 were highly associated in a bivariate way with the criterion dimensions of ‘exhaustion’, ‘cynicism’, and ‘efficacy’ respectively, and contributed significantly to its explanation in multivariate models, while being relatively unrelated with each other , meaning that besides significant convergence, they present great discriminative power for differentiating the clinical subtypes. So, these subscales approach both burnout perspectives, that of typology and the traditional perspective. Taken separately, as they are presented in BCSQ-12, they could provide a brief description of the history of syndrome development in an operative way and with high convergent validity.
When these findings are seen within the context of accumulated clinical experience on burnout syndrome, it can be observed that as with other disorders (such as anxiety and depression), burnout appears to show itself in different ways, which require specific evaluation and possibly different intervention approaches [17–23]. Vercambre, Brosselin, Gilbert, Nerrière and Kovess-Masféty  take this perspective when they propose the use of different interventions depending on the characteristics presented by affected individuals. These authors recognize the multi-dimensional nature of burnout, but they set out their differential proposal over the classic dimensions of the MBI. These dimensions could include the core definition of burnout, but they do not facilitate a differentiation of the syndrome that would allow the history of the development of the disorder to be understood as is manifested in each particular case, something that can be done by means of the identification of the “frenetic”, “underchallenged” and “worn-out” subtypes of burnout. The properties making up the identified burnout subtypes may have different types of associations with the mediator variable of guilt, as suggested in other studies [26, 68], thus contributing to explain the evolution of the different forms in which burnout is manifested , and perhaps enabling their influence on health to be differentiated . Another interesting line of research that could lead to the establishment of specific biological markers for the syndrome may arise from the study of possible associations between the burnout subtypes and physiological correlates for the syndrome in current use, such as prolactin, cortisol, Immunoglobulin A, natural killer cell activity (NKCA) or mononuclear antibiodies CD16 and CD57 [70–73], which are associated with the functioning of the hypothalamo-pituitary-adrenal axis and the immune system.