Health behavior of working-aged Finns predicts self-reported life satisfaction in a population-based 9-years follow-up

Health behavior is one of the presumed pathways in the well-established association between subjective well-being and in various ways determined good health [1, 2]. However, further understanding of this link is needed in longitudinal settings [3]. Subjective well-being refers to the extent to which a person feels that his or her life is going well [3] and life satisfaction represents the cognitive component of the personal evaluation [4].

The four major domains of modifiable health behavior with substantial impact on the risk of non-communicable diseases comprise of dietary habits, smoking, alcohol consumption, and physical activity [5, 6]. Thus far, the association of health promoting behavior with life satisfaction is well established in cross-sectional settings [7, 8], but few studies have focused on corresponding longitudinal associations. Healthy dietary patterns have been positively associated with good life satisfaction on a day-to-day basis [9, 10], have improved life satisfaction if changed in a favorable direction even when adjusted for other health behaviors [11, 12], and have predicted better life satisfaction in a 2-year follow-up [11]. Furthermore, being a smoker predicted poorer life satisfaction 4 years later [13], and high alcohol consumption poorer life satisfaction during a 15-year follow-up [14]. In addition, physical activity has shown a positive association with life satisfaction on a day-to-day basis [15] and increased physical activity has been shown to improve life satisfaction [16]. The effect has varied according to intensity of exercise, i.e. moderately intensive exercise improves life satisfaction, but vigorous does not [17]. Further, an overall additive effect has been observed, i.e. the more health behavior domains are on a favorable level, the stronger their association with life satisfaction [18].

Studies combining various principal domains of health behavior are encouraged [19] and for that purpose a sum score can be constructed (e.g. [20, 21]). A sum score has major advantages as health behaviors tend to cluster [19] and depicts the general situation of health behavior when co-occurrence is analyzed [22].

Previous studies have indicated several factors that can influence the results obtained. Age, sex, education, socioeconomic status, and health are the main covariates that have been repeatedly explored. Life satisfaction generally follows a U-shaped curve by age, the lowest life point being in the middle age [23, 24] and men generally show better life satisfaction than women [25]. Further, higher education generally results in better socioeconomic status that associates with better life satisfaction [26] but the effect of education does not disappear when adjusted for socioeconomic status [27]. Better health status (both objective and subjective) also associates with higher life satisfaction [2].

Due to the continuous increase of chronic diseases, innovative strategies are needed to decrease their burden in the individual and healthcare levels [5]. A better understanding of the link between health behavior and life satisfaction could enhance and support new and positive strategies for health behavior change. Thus far, however, research has mainly limited to single health behaviors and short follow-ups. As better health behavior leads to better health and prevents diseases, we hypothesized that it also predicts better life satisfaction [2]. Thus, we expect that a better health behavior score at baseline will result in a better subsequent life satisfaction score at follow-up. In a long follow-up, the effect could, however, be confounded by multiple changes happening in life. Adjusting for covariates, especially for baseline life satisfaction, assumably reduces the effect as they can correlate with baseline health behavior and are also determinants of subsequent life satisfaction.

The aim of the study was to explore how the four principal health behavior domains predict life satisfaction as a composite score in a previously unexplored longitudinal setting. Thus, we analyzed the predictive power of dietary habits, smoking, alcohol consumption, and physical activity as a composite measure in a 9-year prospective follow-up (in 2003–2012) in a large population-based and randomly selected sample of working-aged Finns.

In the study participants, there was a higher share of women (n = 6930, 63.8%) than men (n = 3925, 36.2%). In addition, the age group for 55–59 years in 2003 was slightly overrepresented (32.2%) compared to the younger age groups. The largest educational group (39.0%) among the participants had college-level education. Over half of the study population (58.8%) reported at least two major diseases.

The highest percentage of participants (42.4%) reported three favorable health behaviors, 26.1% reported all four favorable health behaviors, 24.5% reported two, 6.2% reported one, and 0.9% none. The mean LSscore after the follow up was 8.37 (SD 3.18). Women more often reported all health behavior domains as protective (HBSS₂₀₀₃ = 4) than men (30.9% vs. 17.7%). For details of the distribution of the study characteristics, see Table 1. The estimates for Cronbach’s alpha were as following: HBSS₂₀₀₃, 0.28; LSscore₂₀₀₃, 0.76; and LSscore₂₀₁₂ 0.76.

The protective behavior of the four studied behavior domains predicted significantly better life satisfaction after follow-up (p < .001) in all the linear models (Table 2). In the crude model 1, the β estimate for subsequent LSscore was − 0.51. Thus, each protective health behavior changed the LSscore towards better life satisfaction by 0.51 points, resulting in a total of 2.04 points for those with all favorable health behaviors. In model 2 with sex, age, education, and major diseases as covariates, the β decreased to − 0.47. In model 3, also adding the baseline life satisfaction in 2003 in model 2, reduced the β to − 0.24 (p < .001), resulting in a total of 0.96 points better life satisfaction after follow-up for those with all protective health behaviors compared to those with none.

The details of the estimates of model 3 are presented in Table 3. Baseline life satisfaction increased life satisfaction after follow-up by a β estimate of 0.44 (p < .001) i.e. one point lower life satisfaction score in 2003 indicated a decrease of 0.44 in the life satisfaction score after follow-up. Age (p < .001), education (p = .02), and diseases (p < .001) were statistically significant covariates. Belonging to the age group 2, having lower education, and having more than two diseases was associated with poorer life satisfaction and a smaller probability of having all protective health behaviors (see also Table 1). Sex was not a significant covariate.

When studying linear regression models for dichotomized individual health behaviors all showed significant (p < .001) predictive power and similar Akaike Information Criterion (AIC) where alcohol consumption had the highest estimate for prediction as follows: alcohol consumption 0.62, smoking status 0.38, physical activity 0.26, and dietary habits 0.19. Details of the models on individual health behaviors are provided in Additional file 2, Tables 2-5.

The present 9-year follow-up study of a population of 11,000 working-aged Finns found that health behavior predicted life satisfaction even after controlling for sex, age, baseline education, major diseases, and baseline life satisfaction. It is, to our knowledge, the first study to assess the predictive power of four major domains of modifiable health behavior as a sum score on subsequent life satisfaction. However, the strongest predictor was baseline life satisfaction among the covariates 9-years previously, which halved, but did not eradicate the significant impact of health behavior on subsequent life satisfaction. Alcohol consumption and smoking showed stronger effect on life satisfaction compared to physical activity and dietary habits.

Our study contributes significantly to research on multiple health behaviors and their impact on supporting or creating good life satisfaction. Life satisfaction can here be regarded as also representing more generally good mental health. However, it is obvious that for most people over the course of time, life circumstances, health behavior, and factors influencing life satisfaction vary substantially. Therefore, it is remarkable that a statistically significant longitudinal association between health behavior and life satisfaction persisted in a 9-year follow-up. Previously, heavy alcohol consumption predicted poorer life satisfaction during a 15-year follow-up with the same, but categorized, life satisfaction scale as used here [14]. Another study [13] concluded by structural equation modeling that smoking predicted significantly poorer life satisfaction 4 years later. Further, the study by Mujcic and Oswald [11] explored the association between initial fruit and vegetable consumption and later life satisfaction in a 2-year follow-up adjusting for exercise, smoking, alcohol consumption and eating patterns; consumption of fruit and vegetables predicted subsequent life satisfaction and an average increase of 0.24 in life satisfaction (range: 0–10) was observed when the consumption of fruits and vegetables was increased by eight portions a day. The study by Ocean et al. [12] concluded that a change in fruit and vegetable consumption resulted in better well-being 5 years later when adjusted for smoking status, walking 10 min per day and longstanding health condition. In line with earlier research [23, 24], a U-shaped curve was observed in life satisfaction by age where the oldest age group showed best life satisfaction. Furthermore, lower life satisfaction was observed in those having two or more diseases, but not having one, when compared to no diseases, which is in line with earlier research showing better health resulting in better life satisfaction [2]. Higher education was also associated with better life satisfaction as in earlier research [26], but not consistently when comparing no education or college level education to university level education. In contrast to earlier research [25], no difference between sexes was observed which might be due to the sex equality in Finland. None of the covariates showed significant interaction in how health behavior predicts life satisfaction suggesting that all covariate groups benefit similarly of good health behavior in respect to life satisfaction.

In a non-response analysis of the Time 1 study population [39], externally caused mortality of the non-respondents was found to be significantly higher for men whereas disease mortality was significantly higher for women as compared to respondents in the 7-year follow-up, although the absolute differences were small. In another earlier non-response analysis of the present data, the respondents in Time 1 showed satisfactory representativeness to national morbidity statistics of the corresponding Finnish population, although women, younger people, those in employment, and individuals with a higher education were somewhat over-represented. On the other hand, smoking and the use of anti-depressants were less prevalent than among the corresponding Finnish population [23]. Compared to Time 1, the present study population showed a higher proportion of women (63.8% vs. 59.2% in 1998) and the proportion of those with high education was lower (19.9% vs. 33.8% in 1998). In 2003, the oldest age group represented the highest proportion of respondents (32.2%) whereas in 1998 the four age groups were fairly equally represented. Therefore, even though the Time 1 study population was mostly fairly representative of the general population, attrition could also be responsible for some changes. However, the covariate groups showed a satisfactory distribution and thus the results can be generalized to the working-aged Finnish population.

To the best of our knowledge, the present study is the first to identify a long-term predictive effect of a sum score comprising multiple health behavior domains to life satisfaction in the future. The observed difference in subsequent life satisfaction between individuals reporting four protective health behaviors in comparison to those with none was almost a third of the magnitude of the standard deviation of the outcome. Furthermore, the difference in life satisfaction related to baseline health behavior reached a twofold level as compared to the corresponding difference between those having a baseline university education and those lacking any professional education. Since good health behavior improves health, better health is presumably one of the routes by which health behavior can influence life satisfaction [3].

The results of the present study may have many implications for clinical medicine, health policy, and health promotion. Innovative solutions for health promotion are needed as the increasing prevalence of chronic non-communicable diseases in combination with advanced but increasingly expensive medical technology challenges healthcare around the world [5]. As focusing on the long-term risk perspective alone is generally perceived by most people as arduous, our results could provide a more beneficial approach by emphasis on future mental well-being (i.e. life satisfaction and happiness) [4, 40] that can be supported by good health behavior. Moreover, improved health behavior reduces the costs of healthcare by prevention of diseases. It can also increase individual productivity via enhanced life satisfaction [41]. From an individual’s perspective, pursuing happiness and life satisfaction is of central importance. Recognizing the role of health behavior in improving one’s life satisfaction could strengthen motivation for making changes for the better. In addition, the results of the present study emphasize that improvement in one health behavior domain might already exert a protective impact. Thus, improving health behavior should not necessarily focus concomitantly on all domains to enable a favorable development in life satisfaction.

Good health behavior is protective to subsequent life satisfaction. This knowledge could strengthen the motivation for improvement of health behavior particularly on an individual level but also on a policy level.