Short-term effects of gaseous pollutants on cause-specific 1 cerebrovascular disease among patients with type 2 diabetes: case- 2 crossover evidence from Beijing, China

65 Background : The association between gaseous pollutants and cerebrovascular morbidity has been widely studied; however, the association in patients with type 2 67 diabetes has remained unknown in developing countries. Methods : A time-stratified case-crossover design combined with a distributed lag 69 nonlinear model was adopted to estimate the short-term effects of gaseous pollutants 70 (nitrogen dioxide [NO 2 ], sulfur dioxide [SO 2 ], ozone [O 3 ], and carbon monoxide [CO]) 71 on cerebrovascular cases among patients with type 2 diabetes in Beijing. In addition, 72 our study explored the variability across sex and age groups. 73 Results : A total of 223,216 (male 57.6% and elderly 61.7%) cerebrovascular cases 74 were reported from 2014 to 2018. The cumulative exposure-response curves were U- 75 shaped for NO 2 , J-shaped for SO 2 , and V-shaped for O 3 and CO. Extreme low-O 3 , low- 76 CO and high-CO increased the risk of cerebrovascular morbidity, with the maximum 77 relative risk (RR) of 1.14 (95% CI: 1.01-1.28) (1st vs median), 1.02 (95% CI: 1.01- 78 1.03) (1st vs median) and 1.25 (95% CI: 1.09-1.45) (99th vs median), respectively, 79 appearing at lag0-13. Elderly individuals aged over 65 years were susceptible to 80 extremely low-O 3 (maximum RR: 1.27, 95% CI: 1.08-1.48). Men and elderly 81 individuals were more susceptible to extremely high-CO, and the maximum RR was 82 10% [1.31 (1.11-1.54) vs. 1.19 (1.00, 1.41)] and 26% higher [1.41(1.17-1.69) vs. 1.12 83 (0.97, 1.30)] than women and adults, respectively. Conclusions : NO 2 , SO 2 , O 3 and CO presented nonlinear and lagged effects which may vary by sex 86 and age. Our study added to the limited evidence for the risk of cerebrovascular disease 87 due to gaseous pollutants among patients with comorbid type 2 diabetes. 88


91
Cerebrovascular disease remains a global challenge to public health and is the leading 92 cause of mortality (Feigin et al., 2017). The cerebrovascular disease burden has 93 increased in China (Wang et al., 2017), accounting for approximately one-quarter of the 94 overall mortality (Zhou et al., 2016). One-third of cerebrovascular cases are comorbid 95 with type 2 diabetes (T2D), leading to poor prognosis (Buddeke et al., 2017). To reduce 96 the disease burden of comorbidity of cerebrovascular disease and T2D, it is crucial to 97 determine modifiable risk factors.  (Chiusolo et al., 2011) found that patients with preexisting diabetes appeared to 106 confer a stronger susceptibility to NO2 exposure. A recent study suggested that 107 individuals with comorbidities, especially diabetes, tend to be more susceptible to air 108 pollution (Li et al., 2019). 109 The disease burden of comorbidities has become a serious issue for public health 110 in China. However, the short-term effects of ambient gaseous pollutant exposure on 111 cerebrovascular disease onset among diabetic participants in China remain largely 112 unknown. Additionally, these associations at the city-specific level have not been 113 thoroughly investigated, particularly in Beijing. 114 In this study, we aimed to examine the associations between ambient gaseous Information Center (http://www.phic.org.cn/). The data recording system in the study 124 area has been proven to be of high validity and good representativeness, covering more 125 than 95% of medical services for permanent residents in Beijing (Aklilu et al., 2020). 126 The geographic locations of 258 hospitals included in the analysis have been described 127 in our previous study (Li et al., 2018). 128 Data capture from the medical record system, including patient age, sex, date of

149
In this study, data were collected as part of health surveillance mandated by government 150 and all data were desensitized. Since we only analyzed the aggregated data, which were 151 typically used for administrative purposes, informed consent was waived. However, the

155
A time-stratified case-crossover design was adopted to remove control confounders and 156 long-term trends (Liu et al., 2021). A distributed lag nonlinear model (DLNM) was 157 applied to estimate the nonlinear and lagged influence of gaseous pollutants on 158 cerebrovascular disease onset. Based on bidimensional functions called "cross-basis", 159 DLNM can reflect both lag structure of the association and the nonlinear exposure 160 responses (Gasparrini, 2014). Therefore, the combination of the case-crossover design 161 with DLNM was applied to simultaneously estimate the short-term, nonlinear, and 162 lagged effects of gaseous pollutants on cerebrovascular disease onset.

163
In our study, a DLNM model combined with a time-stratified case-crossover 164 design was built as follows:    The results of the multipollutant models indicated that SO2, NO2, O3 and CO were 223 significantly associated with both overall and cause-specific cerebrovascular morbidity.

238
Positive associations were observed between NO2 concentrations and risk for overall 239 and cause specific cerebrovascular cases except hemorrhagic stroke (Fig. 2b).

240
High O3 concentrations were associated with an increased risk of overall 241 cerebrovascular cases    Table 2 Cumulative RRs and 95% CIs of extreme gaseous pollutants concentrations on cerebrovascular cases using multi-pollutant models.

313
To the best of our knowledge, this is the first study to examine the associations between 314 gaseous pollutants and cerebrovascular case onset among patients with comorbid T2D 315 in China. In this study, we observed significant nonlinear and lagged effects of NO2, 316 SO2, O3 and CO on the morbidity of cerebrovascular disease among patients with T2D.

317
The extreme effect of high and low concentrations was not consonant, depending on 318 sex, age, and type of gaseous pollutants. Our study added to the limited evidence for 319 the risk of comorbidities of T2D and cerebrovascular disease due to gaseous pollutants 320 in developing countries.

321
The nonlinear and lagged effects of gaseous pollutants were robust when adjusting 322 for co-pollutants using principal component analysis in multipollutant models. T2D is 323 the leading risk factor for cerebrovascular cases and related death in Chinese adults.  Subgroup analysis of sex supported that male participants were more susceptible 376 to extremely high-CO exposure than females. A study from Shenzhen showed that 377 males were vulnerable to extremely high-SO2, high-NO2 and low-O3, and females 378 were vulnerable to extremely high-O3 (Wang et al., 2020). Another study found that 379 among stroke patients male were more vulnerable than female to changes in 380 concentrations of SO2 and NO2 (Wang et al., 2016). In short, the potential sex 381 differences in the associations between gaseous pollutants and cerebrovascular cases 382 remain to be determined.

383
Subgroup analysis of age showed that elder individuals were more susceptible to 384 extremely high-CO and low-O3 exposure and were at higher risk of cerebrovascular-385 related hospitalizations. Unsurprisingly, compared with younger adults, elderly 386 participants are more vulnerable to gaseous pollutants and have a higher risk of 387 24 cerebrovascular diseases. Wang and colleagues found that younger adults were more 388 immune to extremely low-NO2 than elderly adults (Wang et al., 2020). Another study 389 of Beijing also supported that increased concentrations of NO2 and O3 were associated

392
There are some strengths of our study. First, the data on cerebrovascular cases has 393 been proven to be of high validity and good representativeness, covering more than 95% 394 of permanent residents in Beijing, and the sample size was relatively large. Therefore, 395 the present results may be representative of the authentic associations between gaseous 396 pollution and cerebrovascular cases in the study area. Second, to the best of our 397 knowledge, this was the first time that a nonlinear and lagged association were observed 398 between NO2, SO2, and O3 and cerebrovascular cases in a population with T2D, which