In this multicentre survey of students in Germany, the associations between health beliefs and health behaviour were assessed. Among the three studied HLOC dimensions, ILOC had the highest scores, whereas the scores for PLOC and CLOC were nearly equal and much lower. Students reported a frequent consumption of alcohol, relatively frequent physical activity, but a low importance of healthy nutrition. Multivariable analyses showed multiple associations between the HLOC dimensions and selected health behaviour forms: However, these associations differed across the dimensions indicating an importance of a holistic view on health control beliefs.
In general, HLOC scales only accounted for a small fraction of variation in health behaviour, which is in agreement with previous studies in this field
]. However, this finding led to different interpretations in previous research. On the one side, Calnan
] raised doubts about the importance of the HLOC construct and suggested that health behaviour may not be associated with beliefs regarding control of health but rather with concerns over risky health behaviours like smoking or alcohol consumption. On the other side, Steptoe and Wardle
] took a different position and argued that the effects of health control beliefs were rather strong in population terms in contrast to other psychological factors like social support. They also pointed out that correlation measures (and therefore their derivate “explained variance” in linear model or its equivalent in logistic regression) do not convey the public health importance of an association. Explained variation is driven by the effects observed in the vast majority of a population, but still there can be specific characteristics strongly related to specific risk behaviours (for example people with very high scores on PLOC have substantially higher likelihood of being current smokers). Identifying such subgroups can help in the development of targeted interventions
In contrast to some previous studies among working age
] or adolescent samples
], there were fewer associations between the examined forms of health behaviour and ILOC. ILOC was neither associated with a higher likelihood of being a former or current smoker, nor for more frequent drinking, nor for using illicit drugs. However, high ratings on the ILOC scale were associated with an increased chance of more physical activity during a normal week and with paying more attention to healthy nutrition. This is consistent with the theory postulated by Steptoe and Wardle
] that ILOC has stronger effects on health maintenance behaviour (e.g. healthy nutrition)
] than on multiply determined risk behaviours (e.g. smoking, alcohol consumption and drug use).
In our study, ILOC was not related to a higher likelihood of being under- or overweight. Adolfsson et al.
] reported significant associations between the two but they focussed on participants in a weight losing programme. In this subgroup, the association with ILOC may be due to the fact that losing weight is to some extent associated with goal attainment. In general, the observed lack of association can be linked to the fact that weight is not a direct outcome of a single aspect of behaviour and therefore the association between health beliefs and weight can be more complex.
For PLOC, our findings with regards to alcohol consumption were in line with data from previous studies among students as well as adults
]. Additionally, our findings indicated an association between PLOC and drug intake. To our knowledge, no other study has evaluated this association. Whereas a positive correlation between PLOC and smoking was found in a population-based study by Bennett et al.
], no association was observed in our study. It could be argued that behaviour with short-term consequences is more affected by PLOC than behaviour with long-term consequences. Alcohol and drug use can be associated with immediate health complaints like injuries, memory loss or sexual harassment that are directly visible for students
]. In contrast, students might not feel to be at risk of health complaints due to smoking because smoking does not affect their immediate state of health and related diseases appear in a distant future
]. The observation that students with higher CLOC beliefs have higher odds to be current smokers is in agreement with a population-based study in Wales
]. Frequent alcohol consumption was associated with higher CLOC scores in our study. These results agree with the previous study addressing HLOC in university students
]. Finally, our results regarding the importance of healthy nutrition were in line with another population-based study in Wales
]. In summary, CLOC demonstrated reversed associations to ILOC as expected in the HLOC theory. Interestingly, CLOC was the only HLOC dimension associated with smoking. Given the omnipresent smoking prevention advertisement and increased awareness of consequences of smoking, being a smoker has to be associated with beliefs in chance and vice versa smoking prevention based on fear appeal might not be effective for those with strong beliefs that chance determines their health.
Strengths and limitations
The strength of the present study was the large homogeneous sample of students from multiple institutions, but there were also several limitations. The MHLC scale was applied in a reduced form with only three items for each dimension, which was less than in other studies (e.g.
]); this may have affected the precision of the measurement. Despite the self-administered questionnaire, there could have been some reporting bias for health behaviour. For healthy nutrition only the importance, but not the actual behaviour was measured and these two concepts may not be equivalent. Additionally, the measurement of smoking habits did not allow a distinction between occasional and regular smokers. Additionally, we included BMI as a measure of health with respect to weight which does not take a person's body fat content into account.
Another limitation was the lack of measurement of the value of health in our survey. Several previous studies concluded that the value of health is a moderator variable with respect to HLOC and health behaviour
]. However, there are studies which do not confirm this moderator effect
Overall, the sample of the current survey (as a regional sample) is not representative for all students in Germany. Furthermore, the sample differed from the total population of students in NRW in terms of study subjects and gender distribution. This may have had an impact on reported distributions of forms of health behaviour and potentially also on the strength of observed associations. Finally, the cross-sectional study design does not allow for causal inferences.