< Back to Newsletter


Being Happier May Lead to Better Health: Positive Psychological Well-Being and Lifestyle Over 20 Years of Follow-Up.

Claudia Trudel-Fitzgerald, PhD1 (pictured left);  Julia K. Boehm, PhD2; Shelley S. Tworoger, PhD1,3; Laura D. Kubzansky, PhD, MPH1

1Harvard T.H. Chan School of Public Health, Boston, United States
2Chapman University, Orange, United States
3Moffitt Cancer Center, Tampa, United States

Corresponding author: Dr. Claudia Trudel-Fitzgerald, E-mail: ctrudel@hsph.harvard.edu

Facets of positive psychological well-being -including positive emotions, optimism, and life satisfaction- are related to future lower risk of cardiometabolic diseases like heart disease and hypertension (Boehm & Kubzansky, 2012; Trudel-Fitzgerald, Boehm, Kivimaki, & Kubzansky, 2014). Healthy behaviors may be a key mechanism underlying these associations. Although there is emerging evidence to suggest that higher well-being levels are associated with individual health behaviors (e.g., physical activity, favorable diet), its role in overall lifestyle remains understudied. Investigating the role of psychological well-being on the likelihood of adopting a set of healthy behaviors is important, because behaviors tend to cluster and have a multiplicative impact on cardiometabolic disease morbidity and overall mortality (Loef & Walach, 2012; Noble, Paul, Turon, & Oldmeadow, 2015). We hypothesized that individuals with higher well-being levels would be more likely to engage in a healthy lifestyle over time.


             Data are from the Nurses’ Health Study cohort which has been following 121,700 women for >40 years with biennial questionnaires. Women were included in the analytic samples if they were free of chronic conditions when they reported on their well-being. In 1992 women self-reported happiness on a single item (i.e., “How much of the time during the past 4 weeks have you been a happy person?”); in 2004 women reported optimism assessed by the Life Orientation Test–revised(e.g., “In uncertain times I usually expect the best.” ; Scheier, Carver, & Bridges, 1994). Health-related behaviors (i.e., physical activity, body mass index [BMI], diet, alcohol and tobacco consumption) were measured in 1992, and repeatedly every 4 years until the last assessment available (2014).

A composite measure of healthy lifestyle was derived from these five health-related behaviors, which have been validated in this cohort. Based on prior work (Chiuve et al., 2008; Trudel-Fitzgerald, Tworoger, Poole, Williams, & Kubzansky, 2016)and prevention guidelines for age-related chronic diseases (World Health Organization, 2014),each behavior was dichotomized based on whether women met recommended levels or not (1/0). A score of 1 is given for ≥150 min/week of moderate-to-vigorous physical activity queried with a validated scale (Chasan-Taber et al., 1996), BMI is derived using women’s self-reported height and updated weight. Optimal weight is defined as BMI≤25 kg/m2. Dietary data was obtained from the 131-item Food Frequency Questionnaire; our diet score is derived from the validated Alternative Healthy Eating Index(AHEI; McCullough et al., 2002)which defines healthy diet as total scores in the highest 40% of the cohort distribution (Chiuve et al., 2008). Healthy alcohol intake is defined as drinking ≤1 drink/day (Mosca et al., 2011); healthy smoking status is defined as never/past smoking. Dichotomized scores for each behavior were summed for a lifestyle score ranging from 0 “least healthy” to 5 “most healthy.” In previous research, having ≥4 healthy behaviors was related to 50% lower stroke risk (Chiuve et al., 2008)and 66% lower all-cause mortality risk (Loef & Walach, 2012).

Final samples included 52,259 women for the happiness analyses (2 to 5 lifestyle follow-up assessments) and 32,384 women for the optimism analyses (2 lifestyle follow-up assessments). Women were categorized according to their initial level of happiness (i.e., low versus moderate versus high). Logistic regressions were used to estimate the odds ratio (OR) and their 95% confidence intervals (CI) of reporting a healthy lifestyle (i.e., ≥4 healthy behaviors) at least twice over the study duration. Models were adjusted for potential confounders measured on biennial questionnaires (i.e., age, education level, marital status, physical exam in the last 2 years). Because unhealthy participants might be more likely to drop out of the study, inverse probability weights were developed and included in the models. Secondary analyses stratified models by baseline lifestyle level (healthy versus unhealthy) to account of initial lifestyle levels at the study outset. Sensitivity analyses further controlled for anxiety and depression symptoms.


             At the 1992 baseline, women were 57.62 years on average, 83.10% were married/in a relationship, and 87.56% had a recent physical exam. The distribution of covariates was comparable for the 2004 baseline, but the mean age was 68.19 years. As shown in Table 1,

Table 1. Healthy versus unhealthy lifestyles levels and relative amounts of relatively low, moderate, and high levels of happiness and optimism.

OR 95%CI p-value
Low 1.00 (reference group)
Moderate 1.25 1.18-1.31 ≤.0001
High 1.48 1.42-1.54 ≤.0001


1.00 (reference group)
Moderate 1.19 1.13-1.26 ≤.0001
High 1.40 1.33-1.48 ≤.0001
OR=odds ratio; 95%CI=confidence intervals

compared to women with low levels of happiness, those with moderate and high levels had a 25% and 48% greater likelihood, respectively, of reporting a healthy lifestyle at least twice over the next 22 years. Similarly, women with moderate and high (versus low) levels of optimism were 19% and 40% more likely, respectively, to report a healthy lifestyle at least twice during the next 10 years. No interaction between baseline lifestyle and psychological well-being was evident (p-values ˃.05), suggesting that the relationship of happiness and optimism with future lifestyle was similar regardless of women’s baseline health behavior pattern. Moreover, associations remained significant, although somewhat attenuated, after controlling for anxiety and depression symptoms in the models. Figure 1 displays the levels of lifestyle scores over time according to baseline level of happiness. While the differences in lifestyle scores between groups reduced over time, the dose-response effect was sustained throughout two decades. Furthermore, women who reported low happiness levels barely engaged in 3 healthy behaviors, on average. Patterns were similar across optimism levels.