A systematic review and meta-analysis finds that psychological well-being predicts all-cause mortality independently of psychological ill-being

BACKGROUND: Does psychological well-being (e.g. life satisfaction, happiness, meaning in life) predict the length of a person’s life, or are measures of well-being simply assessing the absence of ill-being (depression, loneliness, sadness) and are therefore redundant with ill-being measures? Individual studies have examined this question by measuring and controlling for both well-being and ill-being, but inconsistent findings have been reported. METHODS: A systematic review and random-effects meta-analysis was conducted on the literature of prospective longitudinal studies linking all-cause mortality to well-being. Studies that controlled for ill-being were compared to studies that did not control for ill-being to identify the unique predictive ability of well-being. Separate effects were estimated for different well-being constructs (e.g., optimism) and study characteristics (e.g., patient population). Publication bias was assessed via sensitivity analyses such as removing outliers, trim-and-fill, and RoBMA. RESULTS: A total of 99 prospective longitudinal studies predicted future all-cause mortality from current measures of well-being and controlled for concurrent measures of ill-being. Overall, well-being predicted significantly lower mortality rates when controlling for ill-being (Hazard Ratio [HR] = 0.86 [0.84-0.88]). This finding was apparent when well-being was operationalized with very different constructs and assessments, ranging from optimism to life satisfaction to perceived social support. The magnitude of this association was highly heterogeneous across studies (I2~0.75) with the 95% Prediction Interval for the Hazard Ratio ranging from 0.72 to 1.04. Findings were robust to methods that adjusted effect sizes for the influence of outlying observations, publication bias, or substantial heterogeneity. Surprisingly, the average effect of well-being on mortality was not any stronger when studies failed to control for ill-being (HR = 0.90 [0.88-0.92]), which suggested that well-being predicted mortality independently of the association with an absence of ill-being. CONCLUSION: These results justify research seeking to evaluate the causal role of a broad range of well-being constructs on longevity because they imply that well-being is not simply a lack of ill-being, but more research is needed to identify the causal impacts and mechanisms of these associations.