We studied the association of blood glucose and lung function cross-sectionally and longitudinally among 11,107 participants who attend physical examination from 2009 to 2019.
We found that both pre-diabetic and diabetic adults had lower lung function compared to normal population at baseline, indicating that lung function decrement may start from prediabetic status, giving a hint to the inflexion point. Our study not only qualitatively but also quantitively analyzed this question, suggesting that with 1 mmol/L increase of FPG, FVC, FEV1, FVC% and FEV1% lowered 25ml, 13ml, 0.71%-1.03% and 0.46%-0.72% respectively (Table 2). Previously, most of the researches on this topic used baseline FPG to represent long-term situation, however, our results doubted the reliability of this method, pointing out that it might underestimate the decrease of lung function. We used simultaneous FPG and baseline FPG respectively to conduct the analysis. Although both methods supported that increased FPG correlated to more severe lung function decline, the decrease was not so definitive using baseline FPG. High blood glucose may have accumulative function, making it inaccurate to use baseline data to predict the function for a while. Our results suggest that even for predicting long-term lung function change, FPG following each lung function test is necessary for a more exact study.
Our longitudinal analysis tried to answer the intriguing question if the decreasing rate of lung function was the same in diabetic patients and normal adults. Our study supported that no statistical significance was found for the decline rate in these two populations, based on a large number of participants. This result is surprising as we observed clearly lower lung function parameters in population with higher blood glucose at baseline. Three hypotheses help to explain this interesting result. Firstly, the speed of lung function decline may be faster when blood glucose first becomes abnormal, and with time goes on, it slows down to a “normal” speed, which cannot be differentiate from the decreasing speed of the non-diabetic population. This hypothesis is supported by two studies [8, 17], suggesting that lung function declined faster when adults were first diagnosed as hyperglycemia. Therefore, further investigations are required to figure out the inflexion point of lung function decline in diabetic patients. Secondly, our observation time is not long enough for a mild factor, hyperglycemia, to create a difference of lung function decline rate between diabetic and non-diabetic populations. As we studies lung function changes for 3 to 6 years, to get a more affirmative result, future research can prolong the study time. Thirdly, blood glucose may be adjusted to normal during our study. As we measured the diabetic status based on baseline FPG, which only reflected the blood glucose at one time point, it is not perfect for representing the long-term blood glucose level. Abnormal FPG from health examination rings a bell for the participants to adjust their lifestyle toward a healthier one. A parameter reflecting the long-term situation of blood glucose is necessary to avoid short-term changes that may disturb the result.
Although most of the researches show that the magnitude of lung function impairment secondary to diabetes will cause only subclinical abnormalities, a loss of pulmonary reserve capacity may become clinically important in the context of superimposed conditions, such as acute or chronic lung disease. Researchers start following pulmonary function impairment in a much more detailed degree. For example, epidemiology study of PRISm (Preserved Ratio Impaired Spirometry), which represents slight impairments of both FEV1 and FVC, shows that the mortality of it is comparable to that of diabetes [18]. However, for now, PRISm is still not considered as a disease state and the mechanism of it is unclear, and our study may provide clues for it.
There are some limitations in our study. Firstly, our study is a single center study in Beijing, not enough to represent the whole nation. However, since pulmonary function test is not well promoted in China until recent years, it’s tough to find both large number of participants and reliable lung function data. Secondly, we use FPG to assess blood glucose, which may lead to random error. The other parameter, HbA1c, can represent blood glucose level for a long-term. However, HbA1c is not an index that is regularly included in biochemical check-up in most examination centers. Besides, most of the participants who tested HbA1c had been diagnosed as diabetes, which may lead to selection bias. Thirdly, we did not analyze the lung function change of participants whose blood glucose was high at first and then went back to normal. We are interested in this group because they are on behalf of people who care about their health and this population becomes larger these days. Analyzing this population helps to give advice for better lifestyle. Meanwhile, we did not record the exact time when diabetic patients took anti-diabetic medication. As a result, we couldn’t analyze the lung function of people who took the medicine and recovered from hyperglycemia. To did a basic analysis of medication use, we adjusted diabetes medication history at baseline in model 3 (Table 3, Fig. 1 and Table S2) and the result was the same as it was in the other two models.