Validation of the Croatian Version of the Sense of Coherence 29-Item Scale in Croatian Nurses

The Sense of Coherence (SOC) is a key concept of salutogenic model proposed in 1979 by the American-Israeli medical sociologist Aaron Antonovsky (1). It was seen as a coping resource that helped individuals to identify and use their external and internal resources for solving problems and managing life events (1). In order to measure the SOC, Antonovsky developed a special instrument: the Orientation to Life questionnaire, also named the SOC scale. The original version of the instrument consisted of 29 questions (SOC-29), while a shorter version contained 13 questions (SOC-13) (2). Current research by Mittelmark et al. indicates that the SOC-29 and SOC-13 have been used in at least 49 different languages in at least 48 different countries around the world (2). Both instruments were also assessed for their reliability and validity in different population groups, from the general population to various groups of patients (2, 3, 4, 5, 6, 7, 8 , 9). The results of validation studies proved both instruments to be reliable and cross-culturally appropriate tools (2, 3, 4). In these studies, a one-factor structure was often tested against a three-factor structure. Some authors have concluded that the SOC scale is more appropriate as a one-dimensional scale (3, 7), while others have suggested a multidimensional concept (2).

Research has proved that nurses constitute a vulnerable population group that faces significantly more stress compared to other professions (10), which may have a negative impact on their health (3, 4, 5, 6, 7, 8, 9, 11, 12, 13, 14). Research findings on SOC in the nursing profession have so far shown that SOC is a protective factor against stressors in the work environment (15). A strong SOC is also a good predictor of nurses’ ability to function healthily in the workplace, protecting them against burnout (16). Burnout is also a feature of the nursing profession in Croatia (11, 17). According to Mijakoski et al. (17), Croatian nurses reported high levels of depersonalisation and of organisational and emotional work demands. The study highlighted the need for specific organisational interventions to be implemented in hospital settings. An exploration of SOC among Croatian nurses is therefore urgently required.

In 2005 the SOC-29 was translated into Croatian (SOC-29-CRO) and psychometrically tested on the sample of 822 individuals from the general population (18). It has not, however, so far been used to assess the Croatian nursing population. Moreover, to our knowledge the literature currently contains only a handful of studies that attempt to validate the SOC-29 among nurses (19, 20). The study aimed to validate the SOC-29-CRO for use as part of health promotion research among Croatian nurses. Its objective study was to assess selected psychometric characteristics of the instrument within this particular occupational group.

The results of our study show that the SOC-29-CRO successfully fulfilled the required psychometric criteria for being used within the Croatian nursing population.

The results of the reliability analysis in our study are consistent with the results reported in other similar studies. First, the reliability obtained in our study is in the upper range of values for this measure obtained in other similar studies in general (range 0.70–0.95) (2, 3, 4, 5).

Finally, the results of our study are similar to the results of the study conducted among Portuguese nurses with a reported Cronbach’s alpha value of 0.87 (19). The reliability obtained in our study is also completely consistent with the reliability obtained in the study by Kardum et al. (18). The value of Cronbach’s alpha coefficient in this study was almost identical (α=0.88).

The analysis of factor structure in our study first identified five factors (a similar result was obtained in the studies conducted by Frenz et al. (7) and Paika et al. (5). In both studies in continuation one-factor structure was revealed. In the study by Frenz et al., subsequent analysis of the factor scores suggested the scale had one core factor. The authors stated that the results suggested that the SOC was best understood as having a single core factor (7). Paika et al. repeated the factor analysis using the fixed number of factors as the extracted model, which resulted in one factor being extracted. Their study revealed that an SOC-29 one-factor structure showed a coherent structure with remarkable stability (5). The study conducted among Portuguese nurses also confirmed the one-factor structure (19). As in our study all the items showed saturation with the first factor, we decided to use the one-factor structure advocated by other authors (5, 7). This was despite the fact that the percentage of the explained variance could not be considered satisfactory. However, we can justify this by the fact that we investigated a specific sample, i.e. a population of nurses employed in a precisely designated hospital. The coefficient of relative variation for SOC can also be considered low. Unlike our study, the study by Kardum et al. (18) showed a three-factor structure. However, in their conclusions they drew on the use of the overall result rather than the three components. This was because they discussed the inadequacy of their functional differentiation. In order to make a comparison with this study, we also tested a three-factor model. However, the three-factor structure deviated significantly from the structure advanced by other authors and, in our case, also proved uninterpretable, despite the rotations. For this reason, we decided to adopt the one-factor structure. These leads us to conclude that authors generally prefer the one-factor structure of the instrument. The one-factor model was also advocated by Antonovsky himself, since the instrument was not intended to measure dimensions individually (1).

Since the SOC-29-CRO has been already validated among the Croatian general population (18), it is important to explain why the authors considered it necessary to validate the instrument again on a specific nursing population. The first reason has been already presented: in the first SOC-29-CRO validation study, the factor structure of the instrument was not clear, even though the authors had proposed that a one-factor structure be used. This ambiguity could be the result of the internal heterogeneity of the sample of the general population. Consequently, we decided that it was important to reexamine the factor structure on a specific population (like the nursing population addressed in this study) for the purposes of further research. The nursing population predominantly consists of women who are of reproductive age, have a higher level of education and are also exposed to extremely high levels of workplace stress (12). A study on the perception of stress conducted in Slovenia, a country similar to Croatia, found that employed women with at least college-level education were at very high risk of suffering from frequent stress perception (25). Various studies show that differences exist between the genders as regards experience of stress, with women estimating stressors as more severe than men and tending to report higher rates of psychological distress (26). The reason for this may lie in the fact that employed women have to cope with various roles in their professional and private lives and that their experience of the environments in which they live and work is probably different from that of men. It is therefore reasonable to conclude that this group deviates very significantly from the general population, which could also be reflected in the results of the instrument validation.

Our study has some potential limitations. First, one can argue that this study only includes nurses from one hospital in Croatia. A counter-argument would be that this hospital is a typical large healthcare institution that employs a large number of nurses of different profiles. This allows in-depth research into health problems related to the nurses’ workplace. Second, the response rate/number of participants was towards the lower recommended limit, but nevertheless still within the values that allow robust correct conclusions to be drawn. Third, one could argue this study used no method of measurement of stability of the instrument over time, e.g. the test-retest method. We counter this by the fact that the reliability of any self-reported outcome measure can be evaluated using measurement stability methods and/or measurement equivalence methods. The latter were developed from social science research for situations in which it was not possible to perform repeated measurements because the measured phenomenon had changed or could change over time (27). Finally, one can argue that this study tested only one- and three-factor models. Our argument here is that, in accordance with the existing theoretical assumptions (2), the results of other studies (2, 3, 4, 5, 7, 8) that have the confirmed one- and three-factor structure of the SOC questionnaire, and with the results of PCA in our study, it was rational to test only the one- and the three-factor models.

The study also has several important strengths, the most prominent being that it provided additional and important information and knowledge about the psychometric properties of the SOC-29-CRO when evaluated in a Croatian nursing population. The study showed the one-factor structure more directly than the study on the general population. As a result, the one-factor structure of the SOC-29-CRO can be more reliably used in subsequent research on nurses. Finally, the fact that the study was conducted among the nurses of a single hospital only can be an advantage as well as a limitation, since this institution is similar to many other institutions in countries that are experiencing similar transition processes to Croatia. The results of this study could apply to these hospitals as well, especially those in the countries of former Yugoslavia.

The study results are significant for occupational medicine, health organisations and nursing management in Croatia and beyond. Nurses are often exposed to occupational hazards in their work environment which can lead to stress (11). The outcome of these stressful events depends on how nurses cope with stress. Weak SOC prevents nurses from actively managing stressful events, which can lead to work-related health problems. The SOC-29-CRO instrument could be used by nursing managers to identify nurses with weak SOC and implement interventions to create healthy working environments and protect and promote nurses’ health. Actively managing nurses’ health can help to increase their work ability. This instrument could also be helpful when planning the introduction of coping strategies among nursing students as well as among students of other healthcare professions, since it has already been proven that high levels of perceived stress predispose health students to anxiety and a lower quality of life (28). A number of challenges remain for those studying the use of SOC-29-CRO in nursing populations. The dynamics/stability of the SOC-29-CRO must first be checked over a longer period of time in time, especially in relation to those nurses with more demanding jobs (e.g. in intensive care departments). With a focus on studying the properties of the SOC-29-CRO in the nursing population, our work could usefully be continued by analysing additional aspects of validity, e.g. criterion validity.

The adequate psychometric properties of the instrument indicate that SOC-29-CRO is a reliable tool for use in further research. In the near future, it may also be important to the process of designing measures for enhancing nurses’ internal resources for the management of workloads.