Event dependent overall survival in the population-based LIFE-Adult-Study

Backround Information about the direct comparability of big data of epidemiological cohort studies and the general population still is lacking, especially regarding all-cause mortality rates. The aim of this study was to investigate the overall survival and the influence of several diagnoses in the medical history on survival time, adjusted to common risk factors in a populations-based cohort. Methods From 10,000 subjects of the population-based cohort LIFE-Adult-Study (Leipzig Research Centre for Civilization Diseases), the medical history and typical risk factors such as age, smoking status and body-mass-index (BMI) were assessed. The survival status was identified from the saxonian population register. Univariate and multivariate analyses were used to determine the influence of the medical history and risk factors on overall survival. To develope an optimal model, the method by Collet [1] was used. Results The mortality rate of the participants is approximately half the mortality rate expected for the german population. The selection bias in epidemiological studies needs to be considered whenever interpreting results of epidemiological cohort studies. Nevertheless we have shown that several diagnoses proved to have a negative influence on overall survival time even in this relatively healthy cohort. This study showed the significantly increased mortality risk if the following diseases are reported in medical history of the participants in a large population-based cohort study including adults aged 18 and over: diabetes mellitus (HR 1.533, p = 0.002), hypertension (HR 1.447, p = 0.005), liver cirrhosis (HR 4.251, p < 0.001), osteoporosis (HR 2.165, p = 0.011), chronic bronchitis (HR 2.179, p < 0.001), peptic ulcer disease (HR 1.531, p = 0.024) and cancer (HR 1.797, p < 0.001). Surprisingly, asthma has the opposite effect on survival time (HR 0.574, p = 0.024), but we believe this may be due to an overrepresentation of mild to moderate asthma and its management, which includes educating patients about a healthy lifestyle. Conclusion In the LIFE-Adult-Study, common risk factors and several diseases had relevant effect on overall survival. However, selection bias in epidemiological studies needs to be considered whenever interpreting results of epidemiological cohort studies. Nevertheless it was shown that the general cause-and-effect principles also apply in this relatively healthy cohort.