Background
The effect of lifestyle on morbidity and mortality is increasingly being recognized [
1‐
3]. The disease burden attributed to lifestyle choices primarily consists of non-communicable diseases (NCDs). May et al., [
4] among others, have shown that making healthy choices regarding smoking, nutrition, alcohol consumption and physical activity (SNAP) (here used to define lifestyle), has a strong impact on the prevention of NCDs. However, contrary to what would be desirable from a public health perspective, studies have shown that adherence to a healthy lifestyle (making healthy choices) has decreased over the past decade [
5,
6].
Adherence to a healthy lifestyle may have decreased in general, however in the last decade a strong reduction in the prevalence of smoking has been observed. Over 20% of the worldwide population smokes, which leads to high numbers of premature deaths [
7]. Nutrition also plays a major role in premature deaths and disability. It has been estimated that in 2017 a poor diet was a risk factor in one in five of all deaths globally [
8]. Excessive alcohol intake has been linked to 3 million deaths in 2016 [
9]. Furthermore, almost a quarter of the adult population is physically inactive. Sedentary lifestyles are increasing in varying rates across countries, but seem to currently be most persistent and alarming in developed countries [
10].
Healthy lifestyle promotion requires a comprehensive understanding of the way people behave. Mostly, unhealthy lifestyle choices do not occur in isolation, but in different combinations [
11]. Engaging in a combination of unhealthy behaviours has been shown to have an additional negative influence on health [
12,
13]. A holistic approach to lifestyle interventions may therefore result in more health gains.
Frequent combinations of unhealthy behaviours can be referred to as clusters. Noble et al., [
14] conducted a systematic review of the clustering of SNAP health risk factors (referred to from now on as unhealthy SNAP behaviours). They found that the most frequently reported cluster of unhealthy SNAP behaviours was the absence of any of the behaviours, followed by a cluster of excessive alcohol consumption and smoking, a cluster including all behaviours and a cluster with an unhealthy diet and physical inactivity. To understand behavioural choices, it is relevant to have insight into the way unhealthy SNAP behaviours cluster. However, not much research has been conducted on the potential drivers of these clusters.
Our understanding of attitudinal characteristics that influence people’s lifestyle choices remains limited, both in terms of underlying causes and in the way resulting consequences are perceived. Such information can be useful in the context of promoting healthy lifestyles and changing health behaviours. Here, we focus on two attitudinal concepts that may be associated with (the onset of) unhealthy behaviour: time orientation and risk attitude. Various studies show that smokers are less concerned with future consequences of their health behaviour than non-smokers [
15‐
17]. Furthermore, research shows that risk attitude is associated with risky behavioural choices, like smoking [
18]. However, associations between these concepts and the engagement in multiple unhealthy SNAP behaviours have not yet been studied. People engaged in multiple unhealthy SNAP behaviours, or in certain combinations of these behaviours, might differ in their attitudinal characteristics.
Engagement in unhealthy SNAP behaviours may also result in (or result from) differences in subjective health experiences and expectations. Subjective health has been shown to be an independent predictor of morbidity and mortality [
19,
20] and as such can be considered to carry relevant information in relation to health behaviours. Several studies have shown the association between self-rated health (SRH) and single lifestyle factors [
21‐
23], however few studies have investigated the association between a number (or certain combinations) of healthy lifestyles and SRH [
24]. Subjective life expectancy (SLE) is also an indicator for subjective health; it captures how old people expect to become. SLE was found to be associated with smoking behaviour, which may reflect people’s expectations of the increased risk of dying due to smoking, either directly or indirectly through poorer experienced health due to smoking [
25]. Associations between SLE and unhealthy dietary choices have also been found [
26]. Note that the causal direction between subjective health and unhealthy behaviour can go in both directions. People with an
ex ante low SLE may for instance be more prone to smoke, as they may expect to have less to loose from smoking. Studying these associations between subjective health and lifestyle factors, while also including behavioural characteristics, and acknowledging that unhealthy behaviours do not occur in isolation has, to our knowledge, not been done before.
Here, we present the results from a study that measured attitudinal factors, subjective health and unhealthy SNAP behaviours simultaneously in the same population. Such information can help to understand potential drivers of unhealthy lifestyle choices, both in terms of causes and consequences of unhealthy behaviours. The objectives of this study were therefore (i) to identify how unhealthy behaviours cluster in a sample representative of the adult population of the Netherlands in terms of sex, age, and education, and (ii) to associate combinations of unhealthy behaviours with attitudinal factors (time orientation, risk attitude) and subjective health (SRH, SLE).
Discussion
In the current study, unhealthy SNAP behaviours were studied independently and in combination with each other. The prevalence of smoking, unhealthy diet and physical inactivity was comparable to figures for the general Dutch population [
42]. However, the prevalence of excessive alcohol consumption (29%) was considerably higher than reported in official Dutch population statistics (9%) [
42]. Half of our study population was engaged in two or more unhealthy SNAP behaviours. The most prevalent combination was an unhealthy diet combined with physical inactivity (17%). Smoking, drinking excessively and the lifestyle index were significantly associated with an increased focus on the immediate consequences of behaviour (i.e., the CFC-I). On the other hand, we also found that being physical inactive was significantly associated with an increased focus on the future consequences of behaviour (i.e., the CFC-F). This latter finding is contradictory to what one may expect. These findings may have implications for public health policy, but need to be confirmed longitudinally.
Applying the two-factor structure of the CFC in our regression analysis revealed that smokers were significantly more oriented on immediate consequences compared to non-smokers. However, we did not find a future-oriented attitude among non-smokers. This finding underlines the added value of a two factor structure for the CFC. Previous studies also found that smokers are more present oriented, both when using the CFC as one scale [
43] or two sub-scales [
17]. The CFC also has been used in relation to healthy eating, physical activity and BMI [
17,
34,
43,
44]. Our results confirm previous findings, indicating that people engaged in unhealthy behaviour(s) are especially oriented towards the immediate consequences of their behaviour. This finding does not apply to physical activity, however. We even found a more future oriented attitude for physically inactive people. This finding is counter intuitive since physical activity typically provides gains on the long term. Doing sports is also found to bring positivity and reward just after the exercise and therefore a more present-oriented attitude may also suit athletic people [
45]. In this study the question concerning physical activity not only involved “physical exercise” or “sports” but also walking or climbing stairs. Therefore, an appropriate interpretation of this finding is complicated. We found that the people who had both an unhealthy diet and were physical inactive were significantly less oriented on immediate consequences. This implies that time orientation for unhealthy SNAP behaviours can differ between a single behaviour and a particular combination of behaviours. Findings regarding risk attitude were in line with the general risk attitude hypothesis. People engaged in an unhealthy SNAP behaviour, except for excessive drinking, were more risk seeking than those people not engaged in this unhealthy SNAP behaviour. This association was persistent when considering multiple unhealthy SNAP behaviours. The HRAS-6 (the instrument we applied for risk attitude assessment) has recently been introduced and was shown to be a valid and reliable measure of health-risks attitudes [
38]. The different results found for alcohol consumption could be related to the high percentage of excessive alcohol drinkers in our population, which might be less representative of problematic drinking populations.
The presence of unhealthy SNAP behaviours was associated with significantly lower SRH, although for smoking this was not confirmed in the regression analyses. Two potential explanations can be put forward. First, in the regression analyses we controlled for the potential differences in SRH attributed to co-variates. It is conceivable that the co-variates (sociodemographic-characteristics) explain differences in SRH more than smoking does. Second, it is suggested in the literature that smokers tend to understimate short-term risks of smoking [
46]. This phenomenon might be reflected in the current SRH status of smokers. The clustering of unhealthy SNAP behaviours and the association of these clusters to SRH has been studied before [
47]. Conry et al., (2011) found that respondents with multiple unhealthy SNAP behaviours reported less good SRH scores than respondents with less unhealthy SNAP behaviours.
In the logistic regression analyses, the association between unhealthy SNAP behaviours and SLE only remained significant for smoking. This may imply that people are aware of negative long term health consequences of smoking, which has also been found in previous studies [
48‐
51]. Hence, reiterating long-term consequences in preventive messages may have little effect on behaviour, which is emphasized by the finding that people engaged in an unhealthy SNAP behaviour are more focussed on immediate rather than future consequences. Two other remarks can be made about the association between SLE and unhealthy behaviour. First, a person engaged in risky behaviour might already experience decreased health due to the chosen lifestyle, which in turn negatively affects SLE. Second, one might expect a lower life expectancy when family members on average died relatively young [
27,
52]. Unhealthy habits may then be expected to not or only marginally affect the already low life expectancy.
Study limitations and strengths
A number of limitations of this study need to be mentioned. First, the unhealthy SNAP behaviours were operationalised through dichotomisation, with people either having the risk factor or not. Cut-off points from national guidelines were used to do so. It is important to note that these cut-off points remain somewhat arbitrary and our findings may be sensitive to the cut-off point chosen. For instance, using the national guidelines we observed a considerably higher prevalence of excessive drinking in our sample as compared to national statistics. However, this prevalence would have been even higher if we had adopted alternative, often stricter, international guidelines for excessive drinking, for example the classification defined by the National Institute on Alcohol Abuse and Alcoholism (NIAAA). Second, unhealthy SNAP behaviours were self-reported, which might result in an under- or over estimation of certain habits. Third, alcohol consumption in this study population substantially differed from that of the Dutch population (29% versus 9%). This may highlight that the panel of the sampling agency reached a particular selection of Dutch individuals. This limits the generalisability of our findings. Fourth, we do not know how many people declined to participate in the survey, or dropped-out, as this information was not made available by the survey company for commercial reasons. This information is important to examine potential selection bias in the sample, and given its unavailability, we cannot rule out potential selection bias. Fifth, due to the cross-sectional design of our study, we could not investigate causal relationships. For instance, our data does not allow us to investigate whether people become more present-oriented because they smoke, or whether people become smokers more easily because they are more present-oriented. While examining this further is important, knowing the associations may already be useful for designing interventions and future research.
Several strengths of this study also deserve to be highlighted. Except for the over-representation of excessive alcohol drinkers, our sample appears to be fairly representative of the adult population of the Netherlands in terms of sex, age, educational level and unhealthy SNAP behaviours. The study sample was large with almost 1000 respondents. Moreover, our dataset was relatively rich in terms of the wide variety of included variables. Finally, we tested different clustering techniques to identify combinations of unhealthy SNAP behaviours in the sample, but in the end opted for the simpler and more straightforward approach presented here. The results for this approach were essentially the same and had a somewhat clearer interpretation. In addition, this approach for clustering the unhealthy SNAP behaviours is easier to communicate to a general audience with a less advanced background in statistics.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.