The results of these analyses highlight firstly, the relationship between actual behaviour and perception of behaviour with large proportions of the population having an incorrect perception of their risk. Secondly, the analysis highlighted the proportions of people who worry about their health as a result of not undertaking the correct behaviour with substantial proportions of this population reporting high levels of worry. The results of the multivariable analyses highlight the similarities and dissimilarities between a wide range of demographic, socioeconomic and other related variables for the six key behavioural indicators. The multivariate analysis concentrated on those whose perception is that they undertake unhealthy behaviours, based on the premise that this perception is required before any behaviour change can be undertaken. If a stage of change model was employed these people would be in the contemplation and preparation stages . What this analysis has shown is that there are clear demographic and health-related variables that are different between the groups who are, and are not, worried about the health effects of their actions.
One of the most striking features of the multivariate analyses was the markedly different profiles for different risk factors. Noticeable in these results is the ‘different’ profile for smokers (less likely to fit with the other risk factors) and the range of positive associations with BMI. This highlights the fact that campaigns need to be targeted differently depending upon the profile of the population who are most likely to act upon the message. As argued by others [26, 27], the tailoring of specific messages to specific groups is an important endeavour to counteract the broad, population-wide, non-specific messages commonly used. There is a need to look past the demographic areas of research so that additional detail on the broader life and health context details are provided.
The most striking commonalities across the behaviours was age with all risk factors associated with at least one age group. The 45 to 54 year olds were most likely to have increased odds for each risk factor. This highlights the middle age groups as key targets for interventions, with those who are in the risk categories and are worried about the effect the risk factor is having on their health, being perfect targets for interventions. Other studies have found that midlife is an important time of life to make positive behavioural changes [28–30]. Interestingly, a trend was apparent for smoking and psychological distress with each younger age group more likely to have higher odds indicating that the young smokers and the younger persons with high levels of psychological distress are prime targets for interventions.
While research has highlighted the socio-economic differences apparent in risk behaviours with lower income groups more likely to be smokers , undertake less exercise , and have higher rates of obesity , this analysis showed that the relationship is not necessarily as straight forward as it seems. While our only measure of socio-economic status was annual household income, it was the middle household income level ($40,000 to $80,000 per year) who were more likely to be in the final models for BMI and high psychological distress indicting that campaigns targeting middle income levels for this risk factor should be considered. The lower income level (<$40,000) was also statistically significantly more likely to be included in the BMI model indicting that for BMI both lower income groups are also targets for intervention. In contrast, the middle income level was statistically significantly lower for physical inactivity indicting that this income group were less likely to be worried about their inactivity. No such clear message was apparent in our analysis for fruit and vegetable, alcohol and smoking with household income not included in the final models. Again the need for more detailed, topic-specific interventions are warranted.
A visit to a doctor was a variable included in the final model for BMI highlighting the important opportunity the general practitioner has in influencing these adults. Not surprisingly, smokers were significantly less likely to visit a doctor 10 or more times in the past year. This pattern was repeated for visits to other health professionals with smokers statistically significantly less likely to visit other health professionals while those at risk for low fruit and vegetable consumption and alcohol were statistically significantly more likely to visit other health professionals at least one to four times per year. Previous research has highlighted the important role that general practitioners and other medical specialists have in encouraging and influencing positive behavioural change of their patients [34, 35], although concerns have been expressed on how successful the uptake of guidelines in this area have been .
Interestingly the overall health status variable was included in only three of the models (physical inactivity, smoking and high psychological distress) with higher odds for those respondents reporting fair/poor health. The variable that assessed anger with current health status was also included in these three models in addition to the fruit and vegetable model. While it is acknowledged that anger is associated with many chronic diseases including heart disease , depression and other mental health problems , diabetes , and arthritis  the relationship with risk factors has not been explored and highlights an area for further research.
One of the major strengths of this study is the use of a large randomly selected sample of the Australian population. The large sample size allows for greater generalisation of results. The weaknesses of this study include the cross-sectional nature of the data collection with the consequent inability to determine direction of effect. The reliance on self-report for some of the assessed variables is vulnerable to social desirability or other biased responses and is also a weakness of this study. In addition, sampling by telephone directory is likely to under sample some groups in the community. The study only involves community living adults and as such people living in supported accommodation such as aged care facilities would be missed from the sample. The response rate of nearly 44% is acceptable for this type of survey but the potential for survey non-response bias is acknowledged. Response rates are declining in surveys based on all forms of interviewing [41, 42] as people have become more active in protecting their privacy. The growth of telemarketing has disillusioned the community and diminished the success of legitimate social science research by means of telephone-based surveys. In addition, the increased use of mobile telephones and decreased use of land-lines could result in an under-representation of younger respondents (with younger persons more likely to have mobile telephones only and hence be excluded from sampling frames based on listed telephone numbers). Up to 5% of telephone calls made were on mobile telephones (those that are listed in the EWP or those that are obtained when contact is made with the household).
Other weaknesses of the study are the lack of validation of some of the variables and the fact that these data elements were collected with a range of other variables that were not included in the analysis. This exclusion of these other variables did not allow for consideration of potential confounders. Only using questions pertaining to fruit and vegetable consumption to represent a balanced diet could also be seen as a weakness of the study. Notwithstanding these weaknesses, the overall prevalence estimates obtained from this survey are in line with state and national estimates indicating a non-biased sample.