Diabetes and Mortality From Respiratory Diseases: The Japan Collaborative Cohort Study

Background Little evidence is available about the association between diabetes and respiratory disease mortality among Japanese populations. We aimed to explore the association between diabetes and the risk of respiratory diseases mortality through a nationwide prospective study in Japan. Methods We followed 95,056 participants (39,925 men and 55,131 women) for a median 17.1 years. The information about diabetes status, sociodemographic characteristics, and lifestyles was collected at baseline. Cox proportional hazards regression models were used to estimate hazard ratios (HRs) of mortality from respiratory diseases associated with baseline diabetes status. Results We identified 2,838 deaths from total respiratory diseases (1,759 respiratory infection, 432 chronic obstructive pulmonary disease, and 647 other respiratory diseases). The association between diabetes and total respiratory disease mortality was statistically significant among women (HR 1.81; 95% CI, 1.39–2.37) but of borderline statistical significance in men (P for interaction <0.01). Besides, there were significant associations between diabetes and mortality from respiratory infection among both men and women (HR 1.39; 95% CI, 1.10–1.76 and HR 2.30; 95% CI, 1.71–3.11, respectively; P for interaction <0.001). However, we failed to detect any statistically significant association between diabetes and COPD mortality. Moreover, the subgroup analysis revealed that the association between diabetes and total respiratory disease mortality was stronger in never smokers when compared with ever smokers (P for interaction = 0.02). Conclusions Significant association was observed between diabetes and the risk of total respiratory disease mortality, in particular from respiratory infection. Prevention and control of respiratory diseases, especially respiratory infection, should be paid more attention among people with diabetes in clinical and public health practice.


Introduction
As an important risk factor for cardiovascular diseases, diabetes has significant impact 32 on global health, and contributed to 1.5 million deaths in 2012 worldwide. 1 In addition, 33 diabetes is responsible for the morbidity and mortality of a number of diseases including 34 stroke, ischaemic heart diseases, and cancer. 2,3,4 In Japan, the prevalence of diabetes has 35 increased rapidly over the last decades. 5 It is estimated that approximately 7.2 million 36 Japanese people had diabetes in 2013. 6

38
Although several studies have showed that clinical outcomes in patients with diabetes 39 have improved over time, diabetes-related complications have risen due to the increased 40 prevalence of the disease. 7,8 Thus, the influence of diabetes on the disease-related 41 outcomes still needs further study. Previous studies have demonstrated associations 42 between diabetes and risk of mortality from respiratory diseases such as chronic 43 obstructive pulmonary disease (COPD), pneumonia, and acute respiratory distress 44 syndrome (ARDS), 9,10,11 yet the results were inconclusive among different studies. 45 Bragg et al 9 found that diabetes was significantly associated with increased mortality 46 from chronic respiratory diseases, mainly COPD through a nationwide prospective 47 study of 512,869 adults in China (rate ratio=1.29, 95% CI, 1. 10-1.51). An Australian 48 prospective study which enrolled 1,108,982 individuals with diabetes indicated that 49 standardized pneumonia mortality was 1.22-fold (95% CI, 1.17-1.27) higher in those 50 with type 2 diabetes compared with general populations. 10 Moreover, Soubani et al 11 51 conducted a retrospective cohort study of ARDS patients, which showed that diabetes Study design and participants 66 The Japan Collaborative Cohort Study is a nationwide population-based cohort study 67 with the study design and methods described in detail previously. 13 Briefly, a number  in the study by choosing "agree to participate in this study" for the first question in 74 baseline questionnaire, we considered that informed consent was valid. In several 75 communities, informed consent was obtained from leaders of local government level. 14

76
The study was approved by the Ethics Committees of the Nagoya University School of 77 Medicine and Osaka University.   Information on tobacco smoking was obtained through asking the participants to 94 describe their smoking status: never, former, or current, where former smokers and 95 current smokers were defined as ever smokers. In addition, drinking status was asked 96 to classify subjects into never-, ex-, and current drinkers. Moreover, information of     The Cox proportional hazards regression model was adopted to calculate hazard ratios 117 (HRs) of mortality from respiratory diseases. In the first model, all HRs were adjusted 118 for age at baseline. We also adopted multivariable model to adjust for the potential hour/day, 0.5-1 hour/day, ≥1 hour/day), smoking status (never smokers, ex-smokers, 123 current smokers), alcohol use (never drinkers, ex-drinkers, current drinkers), and family 124 history of diabetes (yes/no). In addition, tests for interaction were carried out with Cox 125 proportional hazards regression analyses by setting variable cross-product terms of 126 diabetes with sex, age (<65 and ≥65 years), smoking (never smoking and ever smoking), 127 and BMI (<25 and ≥25 kg/m 2 ) in the model.

134
Of the 95,056 participants (aged 57.1±10.1 years), the prevalence of self-reported 135 diabetes was 4.5%, of which women was 3.6% and men was 5.8%, respectively.

136
Diabetic patients were older than non-diabetic participants at baseline. The percentage 137 of obesity was higher among diabetic patients when compared to non-diabetic 138 participants among both men and women. In addition, diabetic patients had less alcohol 139 consumption and walking time but more sports activity than non-diabetic participants 140 (Table 1). for educational level, BMI, sports activity, walking, smoking, alcohol consumption, and 146 family history of diabetes, we failed to detect statistically significant association 147 between diabetes and the risk of mortality from total respiratory diseases among men 148 (P>0.05). However, in the age-adjusted model, diabetic women had a higher risk of 149 mortality from total respiratory diseases when compared to those without diabetes (Table 2). Thus, sex may significantly modify the association between diabetes and risk 153 of mortality from total respiratory diseases (P for interaction<0.01). Moreover, there 154 was significant association between diabetes and the risk of mortality due to respiratory 155 infection among both men and women in the age-adjusted model ( In the present prospective study of 95,056 Japanese populations, diabetes was 189 significantly associated with higher mortality from total respiratory diseases among 190 women rather than men. However, we found a significant association between diabetes 191 and increased mortality due to respiratory infection among both men and women. The 192 study indicated that diabetes was associated with risk of mortality from respiratory 193 diseases mainly due to respiratory infection.  19 However, our study failed to find significant association between 216 diabetes and the risk of mortality from COPD and other respiratory diseases, which was 217 different from the findings in the cohort study of Chinese populations. 9 The number of 218 deaths from COPD (1,941) in the Chinese cohort was much larger than that in the 219 current study, which could be an explanation for the inconsistent results between the 220 two studies. Since COPD has a long disease process with diabetes often coexisting, as 221 well as the complicated underlying mechanism about the relationship between the two 222 diseases, 20 the causal relationship between diabetes and COPD mortality remains to be 223 further explored.

225
In the current analysis, we found that baseline history of diabetes was significantly 226 associated with risk of mortality from total respiratory diseases among women but not 227 for men, where sex may significantly modify the association. Gordon-Dseagu et al 16 228 also reported that an increased odds of mortality from respiratory diseases was only 229 observed among women rather than men after adjusting for age. The underlying 230 mechanism about the sex-specific association between diabetes and the risk of mortality 231 from total respiratory diseases needs further study.

233
Moreover, we conducted a subgroup analysis to explore the association between 234 diabetes and respiratory diseases mortality according to different age groups, and found 235 that the associations between diabetes and the risk of mortality from total respiratory 236 diseases and respiratory infection were stronger among participants aged 40-64 years 237 than those aged 65-79 years. A previous study also showed the significant interaction 238 between age and type 2 diabetes for respiratory disorders, where the association 239 between diabetes and respiratory disorders was stronger for younger age classes. 21 In the subgroup analysis based on baseline smoking status, we observed a significant 242 association between diabetes and the risk of mortality due to total respiratory diseases 243 among never smokers rather than ever smokers, and the interaction between smoking 244 and diabetes was significant. The relationship between diabetes, smoking, and the risk We also conducted stratified analysis according to baseline BMI levels, and found that 251 the interaction between diabetes and BMI was not statistically significant although There are several potential mechanisms for the association between diabetes and the 260 risk of respiratory diseases mortality such as infection and inflammation, 261 hyperglycemia, oxidative stress, and decreased lung function. 25,26,27,28,29 Several studies 262 showed that common indicators of inflammation could predict the development of both 263 diabetes and COPD. 25,26,27 In addition, hyperglycemia may harm immune function, 264 antioxidant systems, and complement activation, thus increasing the risk of deaths due 265 to infection. 28 Moreover, as a target organ for diabetes, the function of lung may be 266 impaired by diabetic microangiopathy. 29 The potential relationship between diabetes 267 and respiratory disease mortality still needs more research in the future.  In conclusion, the prospective cohort study revealed that diabetes was significantly 302 associated with increased mortality from total respiratory diseases among women but 303 the association was of borderline statistical significance in men. However, diabetes was 304 associated with mortality due to respiratory infection among both men and women. The