Skip to main content
NCBI home page
PLOS One logoLink to PLOS One
. 2021 Oct 29;16(10):e0259280. doi: 10.1371/journal.pone.0259280

Longitudinal stability and interrelations between health behavior and subjective well-being in a follow-up of nine years

Säde Stenlund 1,2,*, Niina Junttila 3, Heli Koivumaa-Honkanen 4,5, Lauri Sillanmäki 1,2,6, David Stenlund 7, Sakari Suominen 1,2,8, Hanna Lagström 1,2,9, Päivi Rautava 1,2
Editor: Marcel Pikhart10
PMCID: PMC8555827  PMID: 34714864

Abstract

Background

The bidirectional relationship between health behavior and subjective well-being has previously been studied sparsely, and mainly for individual health behaviors and regression models. In the present study, we deepen this knowledge focusing on the four principal health behaviors and using structural equation modeling with selected covariates.

Methods

The follow-up data (n = 11,804) was derived from a population-based random sample of working-age Finns from two waves (2003 and 2012) of the Health and Social Support (HeSSup) postal survey. Structural equation modeling was used to study the cross-sectional, cross-lagged, and longitudinal relationships between the four principal health behaviors and subjective well-being at baseline and after the nine-year follow-up adjusted for age, gender, education, and self-reported diseases. The included health behaviors were physical activity, dietary habits, alcohol consumption, and smoking status. Subjective well-being was measured through four items comprising happiness, interest, and ease in life, and perceived loneliness.

Results

Bidirectionally, only health behavior in 2003 predicted subjective well-being in 2012, whereas subjective well-being in 2003 did not predict health behavior in 2012. In addition, the cross-sectional interactions in 2003 and in 2012 between health behavior and subjective well-being were statistically significant. The baseline levels predicted their respective follow-up levels, the effect being stronger in health behavior than in subjective well-being.

Conclusion

The four principal health behaviors together predict subsequent subjective well-being after an extensive follow-up. Although not particularly strong, the results could still be used for motivation for health behavior change, because of the beneficial effects of health behavior on subjective well-being.

Introduction

The cross-sectional association between health behavior and various measures of subjective well-being is well-established [1–4], but the longitudinal associations need further research [5]. Only a few studies have focused on them [6]. In a follow-up of 15 years, a bidirectional relationship was evident between adverse alcohol consumption and life dissatisfaction, but the former was a somewhat stronger predictor of the latter than vice versa in a large sample of adults [7]. Fruit and vegetable consumption predicted better life satisfaction in a two-year follow-up when adjusted for other health behaviors, but not vice versa [8]. Positive changes in dietary patterns [8,9] and physical activity [10] have resulted in better life satisfaction. In addition, measures of subjective well-being have predicted greater reduction in smoking [11] and less relapses [12]. Earlier reviews have stated that the effect of subjective well-being on subsequent health behavior is largely explained by baseline health behavior [13] but that the bidirectional nature of the relationship requires further study [14]. Subjective well-being refers to a personal evaluation and appraisal of one’s life including both a cognitive judgement (such as life satisfaction) and an emotional response on life (such as happiness) [15].

Structural equation modeling is suitable for health behavior research due to the possibility of including multiple causes and outcomes, lower risk of type I error compared to univariate or bivariate testing, the possibility to specify relationships between variables, reduced effect of measurement error, and advanced treatment of missing data [16]. It enables a more detailed analysis of individual components of health behavior and subjective well-being as their distinctive impacts on the latent variable are investigated. In a study using structural equation modeling [17], the effects of smoking on subsequent lower life satisfaction, lower optimism, and less positive affect (path coefficients = 0.10–0.025) were stronger than the effects in the opposite direction (path coefficients = 0.04–0.08) in a four-year follow-up on older adults (mean age = 64 years). The cross-sectional associations between smoking and subjective well-being at baseline were also statistically significant (path coefficients = 0.04–0.05). Nevertheless, the strongest path coefficients were observed between baseline and follow-up for either smoking (1.77) or subjective well-being (0.43–0.64). To the best of our knowledge, however, the associations between subjective well-being and multiple health behaviors or single health-promoting behaviors, such as physical activity and dietary habits, have not previously been studied using structural equation modeling.

The aim of the present study was to explore the cross-sectional, longitudinal, and cross-lagged relationships of health behavior and subjective well-being by structural equation modelling, as shown in Fig 1. Our hypothesis was that health behavior predicts subsequent subjective well-being and vice versa, because bidirectionality of the relationship has been suggested earlier [14]. Based on earlier research, we also anticipated that the cross-sectional relationships between health behavior and subjective well-being would be statistically significant. Furthermore, we presumed that health behavior at baseline would be a significant predictor of health behavior at follow-up, as well as subjective well-being predicting subsequent subjective well-being.

Fig 1. Structural model showing the connections between latent variables that were tested in the analysis.

Fig 1

Open in a new tab

Methods

The data used in this study comes from a random sample (n = 13,050) of the Finnish working-age population that was collected in the Health and Social Support (HeSSup) study. We used postal survey data from the second (2003) and the third wave (2012) of the study. The survey questions were identically phrased and were presented in an identical or almost identical order. Individuals with missing information on any of the covariates (n = 1,244) were excluded from the analysis. The selection process of the study population is outlined in Fig 2. The final study population (n = 11,806) did not differ considerably from the respondents of the two waves (n = 13,050) with regard to age and gender. For details of the comparison, see S1 File.

Fig 2. Outline of population data selection.

Fig 2

Open in a new tab

Inclusion and exclusion criteria for the present study population from the Health and Social Support (HeSSup) prospective population-based follow-up study.

Four age groups were represented: 25–29 years; 35–39 years; 45–49 years; 55–59. When the HeSSup-study commenced in 1998, the age groups were chosen to be non-continuous to capture certain life transition periods by creating clearly distinct groups that would be fairly homogeneous, i.e., ages not spreading over an entire decade. Four educational groups were represented: no professional education; vocational course/school/apprenticeship contract; college; university degree/university of applied sciences. Major diseases from a pre-defined list of 32 conditions were grouped as follows: none; one; two or more.

The concurrent joint Ethical Committee of the University of Turku and the Turku University Central Hospital approved the Health and Social Support study. The present study was carried out according to the Declaration of Helsinki. Participants signed a written consent agreeing to a prospective follow-up including the registry data.

Measures

Health behavior

Data on four principal health behaviors were dichotomized with the two categories beneficial and risky behavior. Physical activity was measured by metabolic equivalent task (MET) where a score of 2, corresponding to 30 minutes of walking per day, was the cut-off point between active and inactive [18]. Dietary habits were dichotomized at the median of a dietary index (range 0–10) measuring adherence to Nordic nutritional recommendations [19] for ten food items in a non-validated food questionnaire. Each choice that was in line with the following recommendations contributed to one point: dark bread (≥ 2/day); pastries and sweets (≤ 1–2/week); fat free milk (≥ 1/day); sausages (≤ 1–2/week); red meat (≤ 1–2/week); chicken or turkey (≤1–2/week); fish (≥ 1–2/week); fresh fruits and berries (≥ 2/day); vegetables (≥ 2/day); alcohol use (< 10g/day women, 20g/day men) [20]. Alcohol consumption was dichotomized according to the at-risk use level in Finland, excessive consumption being ≥ 140g/week for women and ≥ 280g/week for men [21]. Smoking status divided participants into current smokers and others.

Subjective well-being

Items from a four-item life satisfaction scale [22] were used to reflect subjective well-being with a reversed scale (i.e. higher scores indicating better life satisfaction). The items assessed interest (1–5), happiness (1–5), and ease in life (1–5), and perceived loneliness (1–4) as follows: very boring/unhappy/hard/lonely = 1; fairly boring/unhappy/hard/lonely = 2; cannot say = 3; fairly interesting/happy/easy and not at all lonely = 4; very interesting/happy/easy = 5 [7,22]. On the original scale, perceived loneliness ranged from 1 to 5 but had no response alternative 2. The scale was compressed into a scale ranging from 1 to 4.

In general, subjective well-being comprises both cognitive and affective components [23]. Thus, the four items in the life satisfaction scale can be regarded to reflect subjective well-being–as we have assumed here–rather than solely that of life satisfaction, which has been defined as the cognitive component of subjective well-being [5]. It is also more appropriate to let its four components represent subjective well-being in the structural equation model, in which the components represent distinct observed variables and, thus, different aspects of subjective well-being.

Statistical analysis

The hypothesized models were tested using structural equation modeling with the MPlus software, version 7.4 [24]. The analyses, based on the covariance matrices of the data, were performed using the mean- and variance-adjusted weighted least squares estimator (WLSMV). This estimator was used because the observed health behavior variables were categorical. The fit of the models was evaluated using the following indexes (levels of acceptable fit): the comparative fit index (CFI > 0.90), the Tucker-Lewis Index (TLI > 0.90), and the root mean square error of approximation (RMSEA < 0.08) [25]. The χ2 values were not primarily used for model fit estimation, because a good fit using χ2 can be hard to achieve in large samples [26], but the χ2 values were still used for comparison between different models.

The latent variables used in our model were health behavior and subjective well-being, each at two time points (2003 and 2012). The structural validity of these latent variables in 2003 was confirmed through confirmatory factor analysis (CFA), which tests the adequacy of the specified relations between the constructed latent variables and their corresponding indicators [27]. Single items and errors were assumed to be uncorrelated. The fit of the CFA model was good, with CFI = 0.990, TLI = 0.985 and RMSEA = 0.019 in 2003, and only slight differences in these values were observed in 2012. When performing CFA separately on the four individual latent variables, the fit was acceptable in all cases, except that TLI was somewhat under the cut-off value (TLI = 0.887) for health behavior in 2003. Despite this slight discordance, no further modifications on the health behavior latent variable were made, as the CFA with both latent variables included showed good fit in 2003 and also in 2012, and the goal of the present study was to explore the relationship of the four principal health behaviors with subjective well-being.

To study the bidirectional relationships between health behavior and subjective well-being in 2003 and 2012, a latent variable structural model was constructed as shown in Fig 1. For the model to be estimated, autocorrelations of the individual components of health behavior and subjective well-being were included. This crude model was tested for the respondents of the two waves (n = 13,050) and for the population used in the final model (n = 11,806), which confirmed that the omission of participants having missing data on covariates did not substantially alter the results. For details, see S1 File. Thereafter, the covariates gender, age, education, and diseases were included to serve as potential predictors of health behavior and subjective well-being. The model adjusted for the covariates showed good fit: CFI = 0.958, TLI = 0.944, and RMSEA = 0.035. The MPlus software suggested the following connections as potential modifications to the model: physical activity with alcohol consumption, dietary habits with smoking, and interest in life with ease of living. These connections were excluded after testing, however, since the fit of the model did not improve substantially.

A discussion on reliability measures is included in S2 File. Since the assumption of tau-equivalence does not hold, McDonald’s ω is preferable over the commonly used Cronbach’s α as a measure of reliability. Values of both α and ω are given in Table A in S2 File.

Results

We explored the bidirectional longitudinal relationships between health behavior and subjective well-being, when concomitant influence of several confounders was taken into consideration, due to the applied statistical method of analysis. The baseline characteristics of the participants are presented in Table 1. The final model is shown in Fig 3, where all indicated connections are statistically significant (p < 0.001). Health behavior in 2003 predicted subjective well-being in 2012 by a standardized path coefficient of 0.156, but subjective well-being in 2003 did not show a statistically significant prediction for health behavior in 2012. However, both predicted (in 2003) their own subsequent levels (in 2012), with the standardized path coefficients being 0.896 for health behavior and 0.468 for subjective well-being. Their cross-sectional associations were also significant, the standardized path coefficients being 0.302 in 2003 and 0.159 in 2012. The path coefficients and R-values are presented in Table 2, the descriptive means of the observed variables in Table 3, and the changes in fit indexes adjusted for covariates and suggested modifications in Table 4. Table 5 summarizes the path coefficients and autocorrelations of the observed variables.

Table 1. Baseline characteristics of the participants in 2003.

Variable Category Share of the study population % (n)
Study population 100 (11,806)
Age 25–29 20.7 (2,449)
35–39 20.5 (2,422)
45–49 26.7 (3,155)
55–59 32.0 (3,780)
Gender Male 37.1 (4,382)
Female 62.9 (7,424)
Education No professional education 12.2 (1,436)
Vocational school 28.7 (3,391)
College 39.0 (4,599)
University level education 20.2 (2,380)
Diseases 0 18.0 (2,129)
1 23.4 (2,759)
2 or more 58.6 (6,918)
Open in a new tab

Fig 3. Path diagram of final structural equation model.

Fig 3

Open in a new tab

The structural equation model showing standardized path coefficients for cross-lagged, cross-sectional, and longitudinal connections for health behavior and subjective well-being in a nine-year follow-up adjusted for covariates.

Table 2. Standardized path coefficients and R-values in the structural equation model.

HB2003 SWB2003 HB2012 SWB2012 R
HB 2003 0.302 0.896 0.156 0.183
SWB 2003 ns 0.468 0.032
HB 2012 0.159 0.791
SWB 2012 0.294
Age 0.144 0.037 ns 0.041
Gender a –0.598 ns ns 0.089
Education 0.292 0.087 ns ns
Diseases –0.092 –0.156 ns –0.039
Open in a new tab

Estimates for health behavior, subjective well-being and covariates based on data from the Health and Social Support (HeSSup) study.

HB2003 = Health behavior in 2003.

HB2012 = Health behavior in 2012.

SWB2003 = Subjective well-being in 2003.

SWB2012 = Subjective well-being in 2012.

ns = non-significant, for all other values p < 0.001.

CFI = 0.958, TLI = 0.944, and RMSEA = 0.035.

a Gender is a binary covariate (female = 0, male = 1), and the values are therefore STDY standardization, while all other values are STDYX.

Table 3. Means (standard deviations) of observed variables in the Health and Social Support (HeSSup) study.

Latent variable Observed variable 2003 2012
Health behavior
Risky = 1
Beneficial = 2		 Physical activity (1 or 2) 1.72 1.71
Dietary habits (1 or 2) 1.39 1.47
Alcohol consumption (1 or 2) 1.95 1.95
Smoking status (1 or 2) 1.81 1.86
Subjective well-being Interest in life (1–5) 3.90 (0.93) 3.88 (0.92)
Happiness in life (1–5) 3.97 (0.83) 3.98 (0.82)
Ease of living (1–5) 3.47 (1.02) 3.63 (0.99)
Not feeling lonely (1–4) 3.43 (0.91) 3.46 (0.89)
Open in a new tab

Table 4. Model fit statistics on the bidirectional longitudinal association between health behavior and subjective well-being.

Model χ2 (df) CFI/TLI RMSEA
Model 1: Crude model 432 (90) .992/.990 .017
Model 2: Age, gender, education, major diseases includeda 2079 (138) .958/.944 .035
Model 3: Suggested modificationsb added to Model 2 1966 (132) .960/.944 .034
Open in a new tab

a Final model, see Fig 1.

b Suggested modifications were connections between physical activity and alcohol consumption, dietary habits and smoking, as well as interest in life and ease of living.

Table 5. Standardized path coefficients, correlations, and autocorrelations of observed variables.

Observed variable 2003 2012 Autocorrelation Correlation 2003–2012
Physical activity 0.288 0.362 0.415 0.492
Dietary habits 0.502 0.488 0.291 0.477
Alcohol consumption 0.564 0.569 0.510 0.757
Smoking status 0.557 0.579 0.669 0.917
Interest in life 0.799 0.796 0.076 0.412
Happiness in life 0.843 0.847 0.001 (ns) 0.363
Ease of living 0.497 0.531 0.169 0.306
Not feeling lonely 0.644 0.647 0.153 0.399
Open in a new tab

ns = non-significant.

All path coefficients and autocorrelations are statistically significant (p < 0.001) unless indicated otherwise.

Discussion

Our study on 11,800 working-age Finns was focused on the association between four principal health behaviors and subjective well-being as latent variables in a structural equation model adjusted for age, gender, education, and major diseases at baseline. The results suggest that health behavior predicts subjective well-being after a nine-year follow-up at a weak but still significant level, but not vice versa. Baseline health behavior is a strong determinant of subsequent health behavior and baseline subjective well-being is a moderate determinant of subsequent subjective well-being. The cross-sectional associations of health behavior and subjective well-being are evident.

Our study contributes to the understanding of the relationship between health behavior and subjective well-being. Health behavior is highly correlated with its subsequent level, which indicates that it is a stable characteristic. In addition, 79% of its subsequent level can be accounted for by the factors in our model. Physical activity seems to be the least strongly reflected behavioral mode among the observed indicators of health behavior. Alcohol consumption and smoking seem to be the most stable characteristics. For subjective well-being, the factor loadings are higher than for health behavior. This suggests that the components of subjective well-being are more consistently co-varying than those of health behavior, which is also reflected in Cronbach’s alphas being higher for subjective well-being than for health behavior. Note, however, the caveats of using Cronbach’s alpha discussed in S2 File. However, subjective well-being was a less stable characteristic during follow-up than the latent variable of health behavior when using continuous measures for subjective well-being and dichotomized for health behavior. Happiness and interest in life have the strongest impact on the latent variable. About 30% of the subsequent level of subjective well-being was determined by factors in our model, even when the model included baseline subjective well-being. Thus, subjective well-being is not primarily determined by factors in our model, but largely by factors outside it. The results also suggest that ease of living has less impact on subjective well-being than the other measures. Significant cross-sectional correlations between health behavior and subjective well-being are observed, which could partly be caused by external factors and partly by their effect on each other.

Being older, being a woman, having higher education, and having less diseases were associated with somewhat better health behavior at baseline. However, the effect was not present at the follow-up. This could be explained by the fact that the covariates account for interindividual differences and do not substantially change during follow-up. Therefore, the effect of covariates is included in the effect that baseline health behavior has on subsequent health behavior. Diseases at baseline are associated with both worse health behavior and subjective well-being at baseline and worse subjective well-being at the follow-up. Men showed slightly better subjective well-being at the follow-up. Better education has positive impact both on the baseline subjective well-being and on health behavior. Age has an impact on subjective well-being at both time points and on health behavior at baseline, which is in line with earlier research [15].

In another study using structural equation modelling, non-smoking predicted measures of subjective well-being and vice versa in older adults (mean age = 64 years) with a follow-up of four years [17]. The highest path coefficient was observed for smoking status predicting subsequent smoking (1.77), which is in line with our study showing high longitudinal correlations of health behaviors and smoking having the highest value of these. Similar to our model, the magnitudes of the path coefficients from baseline subjective well-being to subsequent subjective well-being at follow-up were higher (0.43–0.64) than for the cross-sectional (0.04–0.05 or non-significant) or cross-lagged associations (0.04–0.25 or non-significant). The magnitudes of the path coefficients from smoking to subsequent subjective well-being were higher (0.10–0.25) than vice versa (0.04–0.08 or non-significant), which also is in line with the results in our study. However, our study observed stronger cross-sectional associations than cross-lagged associations. There may be multiple reasons for this disparity. The follow-up was shorter in the study by Lappan, which might result in stronger longitudinal associations. Inclusion of multiple health behaviors could also strengthen the concurrent bidirectional effect to subjective well-being; exercising and eating a healthy diet could promote subjective well-being, and higher levels of subjective well-being could promote self-efficacy to maintain healthy behavior. The age difference in the studies can also have an effect; health behavior of working-aged persons is perhaps guided more by lack of time compared to that of older adults, for whom other factors such as emerging health conditions might be more dominant.

Implications

The results demonstrate the stability of health behavior, which underlines the importance of targeting health behavior early in life when health behavior patterns are formed. Health behavior shows a longitudinal association with subjective well-being, which could serve as a motivator for health behavior change on an individual level. The results could also emphasize the relevance of political actions targeting health behavior, not only to reduce the increasing costs of non-communicable diseases, but also to maintain and improve people’s subjective well-being. Good mental health and subjective well-being of citizens are valuable goals in itself, but they can also lead to greater productivity [28]. The results could be generalized to the whole working-age population in Finland and presumably to corresponding populations of other Western countries.

Evaluation

The use of structural equation modeling enables deeper and more detailed understanding of the bidirectional association between health behavior and subjective well-being than previous research has provided. The large population-based sample yields solid results. Many of the challenges in structural equation modeling [27] were considered by the following strategies: large population-based random sampling, longitudinal study set-up, use of multiple fit indexes, inclusion of autocorrelations, theory-based and limited use of modifications and covariates. However, only a limited number of models were tested. This was partly done because of the interest to deepen the understanding from earlier research and partly because the observed variables were based on earlier literature. Including autocorrelation accounted for measurement error, which is of importance especially for the continuous measure of subjective well-being, resulting in more reliable estimates of variance.

The association between health behavior and subsequent subjective well-being was statistically significant but fairly weak. However, its strength was about a third of the path coefficient of how subjective well-being predicted its subsequent level and of similar magnitude to the results in the study by Lappan et al. [17] exploring the relationship between smoking and measures of subjective well-being. The four principal health behaviors associated to the risk of chronic diseases of public health concern [29,30] were the focus of this study. However, their factor loadings varied considerably. Different components of health behavior or individual health behaviors could yield more reliable factors and results. The measures for subjective well-being were items from the four-item life satisfaction scale [22], which as an indicator of subjective well-being reflects its cognitive and affective components. It could be argued that loneliness is not part of the standard subjective well-being, but it was included because of a good fit in the factor analysis and a significant path coefficient. Health behavior was a more stable characteristic than subjective well-being, but it was measured by dichotomized variables, where small changes remain unrecorded compared to the scales for subjective well-being. The study was conducted in Finland, where the social security system is strong and for example gender or economic inequality low. Therefore, the results are most reliably generalized to Western nations with low inequality. In general, research on subjective well-being has been conducted mostly in Western nations, and it is still unsure how the results can be generalized to other cultures [15]. Furthermore, studies on ethnic differences are still rare, and firm conclusions are still impossible for the impact of race [6]. Therefore, caution should be applied when comparing the results with other cultures.

The use of more than two time points could have strengthened the results, but subjective well-being was not measured in the 1998 survey. Impairment caused by a particular disease might differ considerably between individuals, but the severity was not reported in the survey. However, multimorbidity has been shown to linearly associate with life satisfaction [31] and, more generally, be an important health indicator. Grouping according to the number of reported diseases was considered a suitable way to differentiate between participants. This grouping was used in the study due to a statistically significant effect.

The dietary habits were measured by self-report, where information about special diets was not included. However, as the observed variable for dietary habits was dichotomized, individual dietary restrictions are likely to have a minor impact. Lastly, non-response and attrition has resulted in some underrepresentation of men and individuals having a lower level of education, fewer healthy behaviors, or lower subjective well-being. The length and the sensitive nature of the survey were reasons for non-response in 1998 [32] and have presumably also resulted in increasing attrition of participants during the follow-up.

Further study

In our study, we examined the relationship of the four principal health behaviors that have substantial impact on non-communicable diseases and subjective well-being. However, the health behaviors do not covary consistently and it is also unclear whether the included health behaviors are the principal ones associated with subjective well-being. Additional health behaviors–such as meditation, avoiding sedentary behavior, and sleep–could also have a role and would therefore be worth studying. Health behavior as a latent variable could also be studied in more detail. Additional observed variables, a larger variety of covariates, and a more detailed scale for all health behaviors could be tested, e.g., including former smokers as a separate group of smoking status. The study of the relationship between health behavior and subjective well-being would benefit from experimental interventions where habits improving subjective well-being would be encouraged to support a change in multiple health-behaviors.

Conclusion

Our results suggest that health behavior partly predicts subjective well-being in a longitudinal follow-up. The study also underlines the stability of health behavior. These results could serve as motivators for health behavior change in health promotion.

Supporting information

S1 File. Comparison of the study population and respondents to the survey.

Data showing that omitting participants due to missing information on covariates does not cause major changes in the characteristics of the study population or distort the SEM model.

(DOCX)

Click here for additional data file. (18.5KB, docx)
S2 File. Reliability measures.

(DOCX)

Click here for additional data file. (20.6KB, docx)

Data Availability

We kindly notify that study data contains personal and sensitive information and due to the present legislation in Finland cannot be made publicly available inside or outside Finland. Data inquiries may be sent to Prof. Markku Sumanen (markku.sumanen@tuni.fi), leader of the HeSSup research group. Additionally, Lauri Sillanmäki (lauri.sillanmaki@utu.fi) is responsible for the data storage and can be contacted concerning data inquiries. Due to the national data protection legislation, the data used in this study cannot be shared without applying for permission to use the data with a specific study protocol and scientifically justified study questions.

Funding Statement

This work was supported by personal grants by the Signe and Ane Gyllenberg Foundation (SSt:5175) and (HKH:5525) [2020], https://gyllenbergs.fi/fi/apurahat-ja-symposiumit; the Waldemar von Frenckell Foundation (SSt) [2020], http://www.foundationweb.net/frenckell/; the Päivikki and Sakari Sohlberg Foundation (HKH:2020), https://pss-saatio.fi; and the Magnus Ehrnrooth Foundation (DS) [2020], https://www.magnusehrnroothinsaatio.fi/en/. The funding sources did not participate in designing or conducting the study; collection, management, analysis or interpretation of the data; or preparation, review, or approval of the manuscript.

References

  • 1.Kushlev K, Drummond DM, Diener E. Subjective Well-Being and Health Behaviors in 2.5 Million Americans. Appl Psychol Heal Well-Being. 2019;12: 166–187. doi: 10.1111/aphw.12178 [DOI] [PubMed] [Google Scholar]
  • 2.Grant N, Wardle J, Steptoe A. The Relationship Between Life Satisfaction and Health Behavior: A Cross-cultural Analysis of Young Adults. Int J Behav Med. 2009;16: 259–268. doi: 10.1007/s12529-009-9032-x [DOI] [PubMed] [Google Scholar]
  • 3.Strine TW, Chapman DP, Balluz LS, Moriarty DG, Mokdad AH. The associations between life satisfaction and health-related quality of life, chronic illness, and health behaviors among U.S. community-dwelling adults. J Community Health. 2008;33: 40–50. doi: 10.1007/s10900-007-9066-4 [DOI] [PubMed] [Google Scholar]
  • 4.Liu B, Floud S, Pirie K, Green J, Peto R, Beral V. Does happiness itself directly affect mortality? the prospective UK Million Women Study. Lancet. 2016;387: 874–881. doi: 10.1016/S0140-6736(15)01087-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Diener E, Lucas RE, Oishi S. Advances and open questions in the science of subjective well-being. Collabra: Psychology. University of California Press; 2018. doi: 10.1525/collabra.115 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Diener E, Pressman SD, Hunter J, Delgadillo-Chase D. If, Why, and When Subjective Well-Being Influences Health, and Future Needed Research. Appl Psychol Heal Well-Being. 2017;9: 133–167. doi: 10.1111/aphw.12090 [DOI] [PubMed] [Google Scholar]
  • 7.Koivumaa-honkanen H, Kaprio J, Korhonen T, Honkanen RJ, Heikkilä K, Koskenvuo M. Self-reported life satisfaction and alcohol use: A 15-year follow-up of healthy adult twins. Alcohol Alcohol. 2012;47: 160–168. doi: 10.1093/alcalc/agr151 [DOI] [PubMed] [Google Scholar]
  • 8.Mujcic R, Oswald AJ. Evolution of Well-Being and Happiness After Increases in Consumption of Fruit and Vegetables. Public Health. 2016;106: 1504–1510. doi: 10.2105/AJPH.2016.303260 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Ocean N, Howley P, Ensor J. Lettuce be happy: A longitudinal UK study on the relationship between fruit and vegetable consumption and well-being. Soc Sci Med. 2019;222: 335–345. doi: 10.1016/j.socscimed.2018.12.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Headey B, Muffels R, Wagner GG. Choices Which Change Life Satisfaction: Similar Results for Australia, Britain and Germany. Soc Indic Res. 2012;112: 725–748. doi: 10.1007/s11205-012-0079-8 [DOI] [PubMed] [Google Scholar]
  • 11.Haller CS. Trajectories of smoking behavior as a function of mood and satisfaction with life: What matters most? J Affect Disord. 2016;190: 407–413. doi: 10.1016/j.jad.2015.10.038 [DOI] [PubMed] [Google Scholar]
  • 12.Leventhal AM, Ramsey SE, Brown RA, LaChance HR, Kahler C. Dimensions of depressive symptoms and smoking cessation. Nicotine Tob Res. 2008;10: 507–517. doi: 10.1080/14622200801901971 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Boehm JK, Vie LL, Kubzansky LD. The Promise of Well-Being Interventions for Improving Health Risk Behaviors. Current Cardiovascular Risk Reports. 2012. pp. 511–519. doi: 10.1007/s12170-012-0273-x [DOI] [Google Scholar]
  • 14.Boehm JK, Kubzansky LD. The heart’s content: The association between positive psychological well-being and cardiovascular health. Psychol Bull. 2012;138: 655–691. doi: 10.1037/a0027448 [DOI] [PubMed] [Google Scholar]
  • 15.Diener E, Oishi S, Tay L. Advances in subjective well-being research. Nat Hum Behav. 2018;2: 253–260. doi: 10.1038/s41562-018-0307-6 [DOI] [PubMed] [Google Scholar]
  • 16.Buhi ER, Goodson P, Neilands TB. Structural equation modeling: A primer for health behavior researchers. Am J Health Behav. 2007;31: 74–85. doi: 10.5555/ajhb.2007.31.1.74 [DOI] [PubMed] [Google Scholar]
  • 17.Lappan S, Thorne CB, Long D, Hendricks PS. Longitudinal and Reciprocal Relationships Between Psychological Well-Being and Smoking. Nicotine Tob Res. 2018;22: 18–23. doi: 10.1093/ntr/nty185 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Kujala UM, Kaprio J, Koskenvuo M. Modifiable risk factors as predictors of all-cause mortality: The roles of genetics and childhood environment. Am J Epidemiol. 2002;156: 985–993. doi: 10.1093/aje/kwf151 [DOI] [PubMed] [Google Scholar]
  • 19.Becker W, Lyhne N, Pedersen AN, Aro A, Fogelholm M, Phórsdottir I, et al. Nordic nutrition recommendations 2004—Integrating nutrition and physical activity. Scand J Nutr. 2004;48: 178–187. doi: 10.1080/1102680410003794 [DOI] [Google Scholar]
  • 20.Lagström H, Halonen JI, Kawachi I, Stenholm S, Pentti J, Suominen S, et al. Neighborhood socioeconomic status and adherence to dietary recommendations among Finnish adults: A retrospective follow-up study. Heal Place. 2019;55: 43–50. doi: 10.1016/j.healthplace.2018.10.007 [DOI] [PubMed] [Google Scholar]
  • 21.Alho H, Eskola K, Aalto M, Autti-Rämö I, Holopainen A, Kaarne T, et al. Alkoholiongelmaisen hoito. Duodecim. 2010;126: 1353–1354. 20681358 [Google Scholar]
  • 22.Koivumaa-Honkanen H, Honkanen R, Viinamäki H, Heikkilä K, Kaprio J, Koskenvuo M. Self-reported Life Satisfaction and 20-Year Mortality in Healthy Finnish Adults. Am J Epidemiol. 2000. doi: 10.1093/aje/152.10.983 [DOI] [PubMed] [Google Scholar]
  • 23.Busseri MA, Sadava SW. A review of the tripartite structure of subjective well-being: Implications for conceptualization, operationalization, analysis, and synthesis. Personal Soc Psychol Rev. 2011;15: 290–314. doi: 10.1177/1088868310391271 [DOI] [PubMed] [Google Scholar]
  • 24.Muthén L, Muthén B. Mplus Version 7 user’s guide. Los Angeles, CA Muthén Muthén. 2012. doi: 10.1111/j.1532-5415.2004.52225.x [DOI] [PubMed] [Google Scholar]
  • 25.Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equ Model. 1999;6: 1–55. doi: 10.1080/10705519909540118 [DOI] [Google Scholar]
  • 26.Bentler PM, Bonett DG. Significance Tests an Goodness of Fit in the Analysis of Covariance Structures. Psychol Bull. 1980;88: 588–606. [Google Scholar]
  • 27.MacCallum RC, Austin JT. Applications of structural equation modeling in psychological research. 2000; 201–226. [DOI] [PubMed] [Google Scholar]
  • 28.Diener E. The Remarkable Changes in the Science of Subjective Well-Being. Perspect Psychol Sci. 2013;8: 663–666. doi: 10.1177/1745691613507583 [DOI] [PubMed] [Google Scholar]
  • 29.OECD. Health at a Glance 2019. OECD; 2019. doi: 10.1787/4dd50c09-en [DOI] [Google Scholar]
  • 30.WHO. World health statistics 2020. World health statistics 2020: monitoring health for the SDGs, sustainable development goals. Geneva: World Health Organization; 2020. Licence: CC BY-NC-SA 3.0 IGO. Cataloguing-in-Publication. 2020. [Google Scholar]
  • 31.Lukkala PS, Honkanen RJ, Rauma PH, Williams LJ, Quirk SE, Kröger H, et al. Life satisfaction and morbidity among postmenopausal women. PLoS One. 2016;11: 1–10. doi: 10.1371/journal.pone.0147521 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Korkeila K, Suominen S, Ahvenainen J, Ojanlatva A, Rautava P, Helenius H, et al. Non-response and related factors in a nation-wide health survey. Eur J Epidemiol. 2001;17: 991–999. doi: 10.1023/a:1020016922473 [DOI] [PubMed] [Google Scholar]

Decision Letter 0

Marcel Pikhart

21 Jun 2021

PONE-D-21-14933

Longitudinal stability and interrelations between health behavior and subjective well-being in a follow-up of nine years

PLOS ONE

Dear Authors,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by 17 July 2021. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Marcel Pikhart

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2.  We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

3. We noted in your submission details that a portion of your manuscript may have been presented or published elsewhere. 

"The data has been used in two other manuscripts exploring the relationship between health behavior and life satisfaction previously by linear regression models and sum scores of health behavior and subjective well-being. In the present study a more intricate method is applied to gain deeper understanding of the relationship."

Please clarify whether this publication was peer-reviewed and formally published. If this work was previously peer-reviewed and published, in the cover letter please provide the reason that this work does not constitute dual publication and should be included in the current manuscript.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This longitudinal follow-up research is promising. There are some critical issues whose resolution will increase the strength of the manuscript and contribute to its publication, as mentioned below:

-You have cited a lot of reference based on participants from different nationalities such as US, UK, and so on. Do different races have an impact on the research results?

-The grouping of participants in this study is confusing to some extent, for example, the age grouping in not continuous, and the major diseases grouping is based on the number rather than the type of disease. Is there any reason for this grouping?

-When studying smoking status, you have only divided it into current smokers and others, while the time after quitting smoking may make a difference in health behavior. Could you divide this group into several subgroups?

-The response rate of Survey (2003) sent to persons who responded in 1998 was 80.2% while the response rate of Survey (2012) sent to persons who responded in 1998 was 57.4%. It is interesting to find this big difference, can you explain why were there so much censored data?

-This research has a large number of participants which adds credibility to results, could you provide a table describing the baseline information of participants?

PAGE 9, LINE 206: The R value is overwritten by other numbers.

Reviewer #2: Background

There is no clear what concept of subjective well-being was considered in the study. This should have significant implications for its operationalization.

The aim of study is presented too general. It is recommended to add more specific problems with hypotheses and graphic presentation of the tested model (here rather not later).

Methods

Were the populations of a random sample (n=13,050) and this finally included in the study (n=11,806) equal according to study variables?

Health behaviours measure: as for adherence to Nordic nutritional recommendations - how responses of people on gluten-free or lactose-free diets were classified? (was the diet controlled for?)

What was the reliability of health behaviour and subjective well-being indicators? Goodness of fit of the model (CFA) does not imply reliability of the scales. In the Discussion section, there are some considerations on stability of subjective well-being but it is not sufficient to understand the character and the role of this indicator. Kuder-Richardson’s and Cronbach’s alfa coefficients (respectively) should be presented.

Additionally, the following issue demands explanation: one of the items included in subjective well-being indicator differs in the scale of responses (criterion for creating psychometrically correct indicators).

It would be also valuable if the Authors explain what was the aim and the effect of inclusion of autocorrelations in the model.

Results

Table 2 – please check the title and content (should it variance or SD?)

I couldn’t find chi2 tests for models’ comparisons.

Discussion

In the Discussion section, there is excessive and not supported by data concentration on the effect of health behavior in 2003 on subjective well-being in 2012, ignoring the fact the this relationship is very week (especially taking into account the sample size). It is rather difficult to treat it as a factor encouraging to health behaviour change. It seems also that health behaviour indicator has low consistency and maybe it would be better to check the effect of individual health behaviors on subjective well-being (assuming that reliability of this scale is satisfactory).

There are single editing and typing errors (e.g. line 130).

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2021 Oct 29;16(10):e0259280. doi: 10.1371/journal.pone.0259280.r002

Author response to Decision Letter 0

20 Aug 2021

For better layout, see the attached Response to reviewers.

Dear Marcel Pikhart,

Thank you for the feedback on our manuscript. We found the reviewer comments helpful for improving the manuscript. Below we have addressed all the comments (numbered) point by point (in bold) and included the new text (in blue) also after the response. The pages and lines mentioned refers to the final version without any visible changes.

JOURNAL REQUIREMENTS:

Comment 1: Please ensure that your manuscript meets PLOS ONE’s style requirements, including those for file naming.

Response 1: The style instructions resulted in the following changes:

a) Background section was renamed as Introduction on page 3, line 59.

b) Supporting information captions were included as follows on page 22, line 506.

Supporting information

S1 File. Comparison of the study population and respondents to the survey. Data showing that omitting participants due to missing information on covariates does not cause major changes in the characteristics of the study population or distort the SEM model.

S2 File. Reliability measures.

Comment 2: We note that you have indicated that data from this study are available upon request. In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

Response 2: The following information was updated to the cover letter:

… study data contains personal and sensitive information and due to the present legislation in Finland cannot be made publicly available inside or outside Finland. Lauri Sillanmäki (lauri.sillanmaki@utu.fi) is responsible for the data storage and can be contacted concerning data inquiries.

Comment 3. We noted in your submission details that a portion of your manuscript may have been presented or published elsewhere. "The data has been used in two other manuscripts exploring the relationship between health behavior and life satisfaction previously by linear regression models and sum scores of health behavior and subjective well-being. In the present study a more intricate method is applied to gain deeper understanding of the relationship."

Please clarify whether this publication was peer-reviewed and formally published. If this work was previously peer-reviewed and published, in the cover letter please provide the reason that this work does not constitute dual publication and should be included in the current manuscript.

Response 3: The following information was updated to the cover letter:

Two manuscripts have been submitted using the current data. The first has been peer-reviewed and resubmitted waiting for final decision. The second manuscript was suggested to be transferred to another journal according to the peer-reviewers’ comments. Both of them use general linear regression models whereas the current manuscript uses a more complex method i.e. structural equation modeling (SEM). In SEM, each component of health behavior and subjective well-being have their distinct impact in contrast to the sum scores used in the previous two manuscripts with only linear regression analyses. SEM enables the inclusion of the cross-sectional, cross-lagged, and longitudinal relationships between health behavior and subjective well-being at both baseline and at follow-up. Therefore, SEM is a more advanced method for detailed analysis. It provides a different perspective and gives deeper knowledge compared to our earlier manuscripts. Thus, the present manuscript gives further knowledge on the studied relationship not given by the previous manuscript.

REVIEWER REPORTS:

REVIEWER 1:

Comment 4: This longitudinal follow-up research is promising. There are some critical issues whose resolution will increase the strength of the manuscript and contribute to its publication, as mentioned below:

-You have cited a lot of reference based on participants from different nationalities such as US, UK, and so on. Do different races have an impact on the research results?

Response 4: Most studies on subjective well-being are conducted in Western nations and generalizability to other nations is still unsure. As a Western nation, Finland has a strong social security system and low inequality, which might affect the results. However, generalizability between Western nations is usually considered acceptable, although differing characteristics of the nations where the results are obtained is important to take into account. An elaboration was added in the evaluation section as follows on page 17 line 371:

“The study was conducted in Finland, where the social security system is strong and for example gender or economic inequality low. Therefore, the results are most reliably generalized to Western nations with low inequality. In general, research on subjective well-being has been conducted mostly in Western nations, and it is still unsure how the results can be generalized to other cultures [27]. Furthermore, studies on ethnic differences are still rare, and firm conclusions are still impossible for the impact of race [6]. Therefore, caution should be applied when comparing the results with other cultures.”

Comment 5: The grouping of participants in this study is confusing to some extent, for example, the age grouping in not continuous, and the major diseases grouping is based on the number rather than the type of disease. Is there any reason for this grouping?

Response 5: The following clarification on age groups was added on page 5 line 119:

“When the HeSSup-study commenced in 1998, the age groups were chosen to be non-continuous to capture certain life transition periods by creating clearly distinct groups that would be fairly homogeneous, i.e., ages not spreading over an entire decade.”

An evaluation of the covariate of diseases was included on page 17 line 380:

“Impairment caused by a particular disease might differ considerably between individuals, but the severity was not reported in the survey. However, multimorbidity has been shown to linearly associate with life satisfaction [31] and, more generally, be an important health indicator. Grouping according to the number of reported diseases was considered a suitable way to differentiate between participants. This grouping was used in the study due to a statistically significant effect.”

Comment 6: When studying smoking status, you have only divided it into current smokers and others, while the time after quitting smoking may make a difference in health behavior. Could you divide this group into several subgroups?

Response 6: When preparing the first version of the manuscript, smoking was a three-category variable, but it was converted into a two-category variable to conform with the other health behavior indicators, all of which were dichotomized. Mixing dichotomized and other categorical observed variables connected to the same latent variable is possible in SEM, but not preferable. Furthermore, previous research analyzing multiple health behaviors frequently uses dichotomized health behaviors. Exploring more detailed health behavior scales could be a possible direction for future research. A short addition was included in the text on page X line X.

“Additional observed variables, a larger variety of covariates and a more detailed scale for all health behaviors could be tested, e.g., including former smokers as a separate group of smoking status.”

Comment 7: The response rate of Survey (2003) sent to persons who responded in 1998 was 80.2% while the response rate of Survey (2012) sent to persons who responded in 1998 was 57.4%. It is interesting to find this big difference, can you explain why were there so much censored data?

Response 7: The survey was long: 29 pages containing 112 detailed and personal items. Complying to filling out such a survey requires effort and the probability of not responding increases as the number of waves increase. The following text was added to the manuscript page 17 line 389:

“Lastly, non-response and attrition has resulted in some underrepresentation of men and individuals having a lower level of education, fewer healthy behaviors, or lower subjective well-being. The length and the sensitive nature of the survey were reasons for non-response in 1998 [32] and have presumably also resulted in increasing attrition of participants during the follow-up.”

Comment 8: This research has a large number of participants which adds credibility to results, could you provide a table describing the baseline information of participants?

Response 8: A table is provided with baseline information about the participants. The corresponding percentages were removed from the text in the methods section. A reference to the table is provided on page 9 lines 211:

“The baseline characteristics of the participants are presented in Table 1.”

Table 1. Baseline characteristics of the participants in 2003.

Variable Category Share of the study population

% (n)

Study population 100 (11,806)

Age 25–29 20.7 (2,449)

35–39 20.5 (2,422)

45–49 26.7 (3,155)

55–59 32.0 (3,780)

Gender Male 37.1 (4,382)

Female 62.9 (7,424)

Education No professional education 12.2 (1,436)

Vocational school 28.7 (3,391)

College 39.0 (4,599)

University level education 20.2 (2,380)

Diseases 0 18.0 (2,129)

1 23.4 (2,759)

2 or more 58.6 (6,918)

Comment 9: PAGE 9, LINE 206: The R value is overwritten by other numbers.

Response 9: The line numbers had overwritten the R-values and therefore the table has now been adjusted on page 10 and line 234.

REVIEWER 2:

Background

Comment 10: There is no clear what concept of subjective well-being was considered in the study. This should have significant implications for its operationalization.

Response 10: Thank you for reviewing the manuscript and for the comments. The concept of subjective well-being was added on page 3 line 70:

“Subjective well-being refers to a personal evaluation and appraisal of one’s life including both a cognitive judgement (such as life satisfaction) and an emotional response on life (such as happiness) [15].”

Comment 11: The aim of study is presented too general. It is recommended to add more specific problems with hypotheses and graphic presentation of the tested model (here rather not later).

Response 11: The aim was specified and the hypothesis stated on page 4, line 91.

“The aim of the present study was to explore the cross-sectional, longitudinal, and cross-lagged relationships of health behavior and subjective well-being by structural equation modelling, as shown in Fig 1. Our hypothesis was that health behavior predicts subsequent subjective well-being and vice versa, because bidirectionality of the relationship has been suggested earlier [14]. Based on earlier research, we also anticipated that the cross-sectional relationships between health behavior and subjective well-being would be statistically significant. Furthermore, we presumed that health behavior at baseline would be a significant predictor of health behavior at follow-up, as well as subjective well-being predicting subsequent subjective well-being.”

Additionally, the graphic representation of the tested model (Figure 1) was moved below the aims as suggested.

Methods

Comment 12: Were the populations of a random sample (n=13,050) and this finally included in the study (n=11,806) equal according to study variables?

Response 12: The close similarity between the population used in the final model and the total respondents of the two waves is shown in two ways on page 5 lines 111 and page 8, line 193:

“The selection process of the study population is outlined in Fig 2. The final study population (n = 11,806) did not differ considerably from the respondents of the two waves (n = 13,050) with regard to age and gender. For details of the comparison, see S1 File.”

“This crude model was tested for the respondents of the two waves (n = 13,050) and for the population used in the final model (n = 11,806), which confirmed that the omission of participants having missing data on covariates did not substantially alter the results. For details, see S1 File.“

S1 File displays data on these comparisons.

Comment 13: Health behaviours measure: as for adherence to Nordic nutritional recommendations - how responses of people on gluten-free or lactose-free diets were classified? (was the diet controlled for?)

Response 13: The dietary measure did not take into account special diets. However, as the dietary index contains information on 10 food items and a special diet usually has major impact on only one, the impact of special diets is not seen as a major concern. Furthermore, in Finland lactose- and gluten-free products are readily available and therefore does not necessarily have a large impact. An additional comment on special diets was included on page 17 line 387:

“The dietary habits were measured by self-report, where information about special diets was not included. However, as the observed variable for dietary habits was dichotomized, individual dietary restrictions are likely to have a minor impact.”

Comment 14: What was the reliability of health behaviour and subjective well-being indicators? Goodness of fit of the model (CFA) does not imply reliability of the scales. In the Discussion section, there are some considerations on stability of subjective well-being but it is not sufficient to understand the character and the role of this indicator. Kuder-Richardson’s and Cronbach’s alfa coefficients (respectively) should be presented.

Response 14: Kuder-Richardson (KR-20) and Cronbach’s alpha estimates were computed using SPSS. However, since the factors are not tau-equivalent, the validity of these coefficients is questionable, and the values are therefore included in S2 File together with some comments. Instead, the omega coefficient described in S2 File is preferable. The omega values were calculated using Mplus and are given in S2 File. This was clarified shortly in the a) Statistical analysis section on page 9 line 204 and b) Discussion section in on page 13 line 284:

a) “A discussion on reliability measures is included in S2 File. Since the assumption of tau-equivalence does not hold, McDonald’s ω is preferable over the commonly used Cronbach’s α as a measure of reliability. Values of both α and ω are given in Table A in S2 File.”

b) “This suggests that the components of subjective well-being are more consistently co-varying than those of health behavior, which is also reflected in Cronbach’s alphas being higher for subjective well-being than for health behavior. Note, however, the caveats of using Cronbach’s alpha discussed in S2 File. “

In addition, the stability of health behavior and subjective well-being respectively was clarified in two places i.e. a) on page 13 line 287 and b) on page 17 line 368:

a) “However, subjective well-being was a less stable characteristic during follow-up than the latent variable of health behavior when using continuous measures for subjective well-being and dichotomized for health behavior.”

b) Health behavior was a more stable characteristic than subjective well-being, but it was measured by dichotomized variables, where small changes remain unrecorded compared to the scales for subjective well-being.

Lastly, the focus in the current study was to explore the effect of the four principal health behaviors on subjective well-being. Therefore, these four were included in the model, even though the fit or reliability could probably be improved by a different selection of health behaviors. To underline this focus, clarifying additions were provided a) and b) in the abstract (page 2, lines 34 and 38), c) in the beginning of the discussion (on page 13, line 269). The issue was further addressed d) in the evaluation section (on page 16, line 361) and e) in future research (page 18, line 397):

a) “In the present study, we deepen this knowledge focusing on the four principal health behaviors and using structural equation modeling with selected covariates.”

b) “Structural equation modeling was used to study the cross-sectional, cross-lagged, and longitudinal relationships between the four principal health behaviors and subjective well-being at baseline and after the nine-year follow-up adjusted for age, gender, education, and self-reported diseases.”

c) “Our study on 11,800 working-age Finns was focused on the association between four principal health behaviors and subjective well-being as latent variables in a structural equation model adjusted for age, gender, education and major diseases at baseline.”

d) “The four principal health behaviors associated to the risk of chronic diseases of public health concern [29,30] were the focus of this study. However, their factor loadings varied considerably. Different components of health behavior or individual health behaviors could yield more reliable factors and results.”

e) “However, the health behaviors do not covary consistently and it is also unclear whether the included health behaviors are the principal ones associated with subjective well-being.”

Comment 15: Additionally, the following issue demands explanation: one of the items included in subjective well-being indicator differs in the scale of responses (criterion for creating psychometrically correct indicators).

Response 15: A clarification on the item of loneliness was included on page 7, line 154. In SEM, observed variables of differing scales can be attributed to the same latent factor.

“On the original scale, feelings of loneliness ranged from 1 to 5 but had no response alternative 2. The scale was compressed into a scale ranging from 1 to 4.”

Comment 16: It would be also valuable if the Authors explain what was the aim and the effect of inclusion of autocorrelations in the model.

Response 16: This is now clarified in the evaluation section on page 16, line 353.

“Including autocorrelation accounted for measurement error, which is of importance especially for the continuous measure of subjective well-being, resulting in more reliable estimates of variance.”

Results

Comment 17: Table 2 – please check the title and content (should it variance or SD?)

Response 17: In Table 2 (now Table 3) the variance was reported, as stated in the title. We changed this and now the standard deviation is reported instead on page 11, line 246.

Table 3. Means (standard deviations) of observed variables in the Health and Social Support (HeSSup) study.

2003 2012

Health behavior

Risky = 1

Beneficial = 2 Physical activity (1 or 2) 1.72 1.71

Dietary habits (1 or 2) 1.39 1.47

Alcohol consumption (1 or 2) 1.95 1.95

Smoking status (1 or 2) 1.81 1.86

Subjective well-being Interest in life (1–5) 3.90 (0.93) 3.88 (0.92)

Happiness in life (1–5) 3.97 (0.83) 3.98 (0.82)

Ease of living (1–5) 3.47 (1.02) 3.63 (0.99)

Not feeling lonely (1–4) 3.43 (0.91) 3.46 (0.89)

Comment 18: I couldn’t find chi2 tests for models’ comparisons.

Response 18: �2 tests comparing the different models were not performed, but the �2 values for the individual models are given in Table 4 (earlier Table 3). A reformulation on �2 was provided on page 7, line 172:

“The �2 values were not primarily used for model fit estimation, because a good fit using �2 can be hard to achieve in large samples [26], but the �2 values were still used for comparison between different models.”

Discussion

Comment 19:

In the Discussion section, there is excessive and not supported by data concentration on the effect of health behavior in 2003 on subjective well-being in 2012, ignoring the fact the this relationship is very week (especially taking into account the sample size). It is rather difficult to treat it as a factor encouraging to health behaviour change. It seems also that health behaviour indicator has low consistency and maybe it would be better to check the effect of individual health behaviors on subjective well-being (assuming that reliability of this scale is satisfactory).

Response 19: The a) abstract, b) discussion , c) implications, d) evaluation and e) conclusion, have been adjusted according to the reviewer’s views on the association between health behavior in 2003 and subjective well-being in 2012. Some elaboration on future research is also included.

a) Abstract, page 2, line 50: “The four principal health behaviors together predict subsequent subjective well-being after an extensive follow-up. Although not particularly strong, the results could still be used for motivation for health behavior change, because of the beneficial effects of health behavior on subjective well-being.”

b) Discussion, page 13, line 271: “The results suggest that health behavior predicts subjective well-being after a nine-year follow-up at a weak but still significant level, but not vice versa.”

c) Implications, page 15, line 336: “The results could also emphasize the relevance of political actions targeting health behavior, not only to reduce the increasing costs of non-communicable diseases, but also to maintain and improve people’s subjective well-being.”

d) Evaluation, page 16, line 357: “The association between health behavior and subsequent subjective well-being was statistically significant but fairly weak. However, its strength was about a third of the path coefficient of how subjective well-being predicted its subsequent level and of similar magnitude to the results in the study by Lappan et al. [17] exploring the relationship between smoking and measures of subjective well-being.”

e) Conclusion, page 18, line 409: Our results suggest that health behavior partly predicts subjective well-being in a longitudinal follow-up. The study also underlines the stability of health behavior. These results could serve as motivators for health behavior change in health promotion. “

Comment 20: There are single editing and typing errors (e.g. line 130).

Response 20: The typing error mentioned and a few more have been corrected.

i.e. “on loneliness”and “influence of several confounders was taken into consideration”

Additional journal requirements:

Comment 21: Thank you for including your ethics statement on the online submission form: "The concurrent joint Ethical Committee of the University of Turku and the Turku University Central Hospital approved the Health and Social Support study. The present study was carried out according to the Declaration of Helsinki. Participants signed a written consent agreeing to a prospective follow-up including the registry data." To help ensure that the wording of your manuscript is suitable for publication, would you please also add this statement at the beginning of the Methods section of your manuscript file.

Response 21: The paragraph was added in the first part of methods section. On page 5, lines 127.

Comment 22: In line with our goal of ensuring long-term data availability to all interested researchers, PLOS’ Data Policy states that authors cannot be the sole named individuals responsible for ensuring data access (http://journals.plos.org/plosone/s/data-availability#loc-acceptable-data-sharing-methods).

Therefore, please provide a non-author, point of contact where others can request access to your minimal data set.

Response 22: Prof. Markku Sumanen (markku.sumanen@tuni.fi), leader of the HeSSup research group, agreed to be a non-author contact person for the data, which was also communicated to the journal through the submission system.

Attachment

Submitted filename: Response to reviewers.docx

Click here for additional data file. (55.7KB, docx)

Decision Letter 1

Marcel Pikhart

18 Oct 2021

Longitudinal stability and interrelations between health behavior and subjective well-being in a follow-up of nine years

PONE-D-21-14933R1

Dear Authors,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Marcel Pikhart

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #2: The introduced changes correctly completed the missing information and contributed to the improvement of the quality of the manuscript. I have no further questions or comments.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Acceptance letter

Marcel Pikhart

22 Oct 2021

PONE-D-21-14933R1

Longitudinal stability and interrelations between health behavior and subjective well-being in a follow-up of nine years

Dear Dr. Stenlund:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Marcel Pikhart

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 File. Comparison of the study population and respondents to the survey.

    Data showing that omitting participants due to missing information on covariates does not cause major changes in the characteristics of the study population or distort the SEM model.

    (DOCX)

    Click here for additional data file. (18.5KB, docx)
    S2 File. Reliability measures.

    (DOCX)

    Click here for additional data file. (20.6KB, docx)
    Attachment

    Submitted filename: Response to reviewers.docx

    Click here for additional data file. (55.7KB, docx)

    Data Availability Statement

    We kindly notify that study data contains personal and sensitive information and due to the present legislation in Finland cannot be made publicly available inside or outside Finland. Data inquiries may be sent to Prof. Markku Sumanen (markku.sumanen@tuni.fi), leader of the HeSSup research group. Additionally, Lauri Sillanmäki (lauri.sillanmaki@utu.fi) is responsible for the data storage and can be contacted concerning data inquiries. Due to the national data protection legislation, the data used in this study cannot be shared without applying for permission to use the data with a specific study protocol and scientifically justified study questions.

    Articles from PLoS ONE are provided here courtesy of PLOS

    RESOURCES