Phenome-wide association of physical activity with morbidity and mortality risk in China: A prospective cohort study

Abstract

Research in high-income countries has established the health benefits of physical activity (PA), but evidence from low- and middle-income countries, including China, where PA patterns vary from those in high-income countries, remains limited. Moreover, previous research, mainly focused on specific diseases, failing to fully capture the health impacts of PA. We investigated the associations of PA with 425 distinct diseases and 53 causes of death using data from 511,088 participants aged 30-79 years in the China Kadoorie Biobank. Baseline PA was assessed using a questionnaire between 2004 and 2008, and usual PA levels were estimated using the resurvey data in 2013-2014. Cox regression was employed to estimate the associations between PA and outcomes, adjusting for potential confounders. During a median follow-up time of 12 years, 722,183 incident events and 39,320 deaths were recorded across 18 chapters of the International Classification of Diseases, 10th Revision (ICD-10). Total PA was significantly and inversely associated with incidence risks of 14 ICD-10 chapters, specifically 65 diseases and 19 causes of death, with the highest quintile group of PA showing a 14% lower disease incidence and 40% lower all-cause mortality compared with the lowest group. Of these diseases, 54 were not highlighted in World Health Organization PA guidelines. Dose-response analyses revealed L-shaped associations for most PA types, except moderate-to-vigorous intensity PA, which showed a U-shaped relationship. In this population, physical inactivity accounted for 12.8% of PA-related deaths. The findings underscore the broad health benefits of PA across a variety of body systems and the significant disease burden due to inactivity in China, highlighting the urgent need for PA promotion.

Keywords: China; morbidity; mortality; phenome-wide; physical activity.

Graphical abstract

Figure 1

Wide landscapes of diseases associated with the highest quintile group of physical activity after FDR adjustment by ICD-10 chapters The number of diseases with FDR significant associations with physical activity is 69. The y axis represents the negative log₁₀ of the phenome-wide p value after FDR adjustment. The horizontal gray dashed line indicates the cutoff for 0.05. The size of the point is proportional to the number of cases. The models were stratified by age at risk (5-year groups), sex, and 10 study areas, and were adjusted for education, drinking status, and smoking status. ICD-10, International Classification of Diseases, 10th Revision; FDR, false discovery rate.

Figure 2

Adjusted HRs for specific diseases showing significant associations after FDR adjustment with physical activity by ICD-10 chapters The x axis is on a log scale. The black squares correspond to the HR values, and the size of the squares is inversely proportional to the standard error; the black horizontal lines represent the 95% confidence interval. HRs were stratified by age at risk (5-year groups), sex, and 10 study areas, and were adjusted for education, drinking status, and smoking status. The individual diseases listed exhibited statistically significant associations between the highest quintile group and the lowest after FDR adjustment in overall analyses. #, an outcome that is critical to decision-making defined by WHO for those aged ≥18 years; ∗, an outcome that is important, but not critical to decision-making defined by WHO for those aged ≥18 years; †, an outcome that is critical to decision-making defined by WHO for those aged ≥65 years. HR, hazard ratio; CI, confidence interval; WHO, World Health Organization; ICD-10, International Classification of Diseases, 10th Revision; FDR, false discovery rate.

Figure 3

Wide landscapes of diseases associated with the highest quintile group of physical activity after FDR adjustment by ICD-10 chapters in sex subgroups The number of diseases with a significant FDR association with physical activity is 37 and 36 for men and women, respectively. The y axis represents the negative log₁₀ of the phenome-wide p value after FDR adjustment. The horizontal gray dashed line indicates the cutoff for 0.05. The size of the point is proportional to the number of cases. The models were stratified by age at risk (5-year groups), sex, and 10 study areas, where appropriate, and were adjusted for education, drinking status, and smoking status. ICD-10, International Classification of Diseases, 10th Revision; FDR, false discovery rate.

Figure 4

Wide landscapes of diseases associated with the highest quintile group of physical activity after FDR adjustment by ICD-10 chapters in age subgroups The number of diseases with a significant FDR association with physical activity is 55 and 9 for the groups aged <65 and ≥65 years, respectively. The y axis represents the negative log₁₀ of the phenome-wide p value after FDR adjustment. The horizontal gray dashed line indicates the cutoff for 0.05. The size of the point is proportional to the number of cases. The models were stratified by age at risk (5-year groups), sex and, 10 study areas, where appropriate, and were adjusted for education, drinking status, and smoking status. ICD-10, International Classification of Diseases, 10th Revision; FDR, false discovery rate.

Figure 5

Associations of selected PA-related diseases with total and domain-specific PA levels Restricted cubic splines with three knots were used to graphically estimate the associations of PA with aggregated diseases. Solid lines represent HRs, and the shaded areas represent 95% CIs. All p values for nonlinearity ≤0.001. HRs were stratified by age at risk (5-year groups), sex, and 10 study areas, and were adjusted for education, drinking status, and smoking status. In the occupational and non-occupational, moderate-to-vigorous intensity and low-intensity PA analyses, additional mutual adjustments were made. PA, physical activity; WHO, World Health Organization; CKB, China Kadoorie Biobank; HR, hazard ratio; CI, confidence interval; MET-h/day, metabolic equivalent of task per hour per day.

Figure 6

Incidence and mortality rates from CKB PA-related diseases and diseases listed in WHO PA guidelines The bar diagrams showed absolute incidence and mortality rates per 100,000 person-years for physically active and inactive participants. The numbers on the bar graph are the absolute excess incidence or mortality rates in the physically inactive group compared with the physically active group, as well as PAR% (95% confidence interval) for CKB PA-related diseases and diseases listed in WHO PA guidelines, both overall and separately for those <65 and ≥65 years. PAR% was calculated from hazard ratio stratified by age at risk (5-year groups), sex, and 10 study areas and adjusted for education, drinking status, and smoking status. PA, physical activity; PAR%, population attributable risk percent; WHO, World Health Organization; CKB, China Kadoorie Biobank.