Exposure assessment
The National Emergency Response Plan for Public Emergencies by the China State Council defined 4 levels of emergency response: Level I (extremely serious), Level II (serious), Level III (relatively serious), and Level IV (common) (22). After the outbreak of COVID-19, the Guangdong Provincial Government announced a Level I response on 1/23/2020 and later degraded the response level to Level II and Level III on 2/24/2020 and 5/9/2020, respectively. The Level III response was maintained after 5/9/2020. During the Level I response, offices, shops, colleges, schools, childcare facilities, and all other non-essential institutions were shut down. Residents’ social activities and gathering were rigorously restricted. Most of the workforce adapted to a new work-from-home mode due to traffic and mobility restrictions. Fewer restriction measures were implemented during the Level II and Level III responses. During the Level II response, crowded areas were temporarily closed and disinfected before reopening. During the Level III response, people’s lives gradually returned to normal. All shopping malls, supermarkets, hotels, restaurants, and other living areas were reopened with routine precautionary measures such as wearing masks and practicing social distancing (Table S1).
We defined the time period with a Level I response (1/23-2/24/2020) as Level I lockdown. Women who were pregnant during the Level I lockdown period were defined as the exposed group (N=101900). Women who were pregnant during the same calendar months in 2015-2019 were defined as the unexposed group (N=493496). This served to control for the seasonal effect, as our data indicated a significant variation in PTB rate across calendar months of conception (Figure S2).
To further explore the potential susceptible exposure window, we divided the exposed group into 11 subgroups according to their GW on 1/23/2020. We determined the day of conception based on the gestational length and date of birth. For example, women who were conceived during the Level I lockdown period were defined as the first subgroup, and women whose GWs were less than four weeks on 1/23/2020 were defined as the second subgroup (Figure S3). The gestational age of women with over 41 GWs was grouped into 41 weeks. Similarly, the unexposed group was divided into correspondingly matching subgroups. With each pair of subgroups (exposed vs unexposed), we estimated the associations of lockdown exposure with gestational length and PTB.
Restriction measures during the Level II and Level III responses may also have adverse effects on PTB risk. Therefore, we quantitatively estimated individual cumulative exposure dose to lockdown by assigning different weightings to days with different levels of emergency responses: 1/22/2020 or earlier (no response, weighting=0), 1/23-2/24/2020 (Level I, weighting=3), 2/25-5/9/2020 (Level II, weighting=2), and 5/10-12/31/2020 (Level III, weighting=1). Moreover, to account for the potential effect modification by timing of exposure, we only estimated the cumulative exposure dose in their first 22 GWs, a conventional cut-off value of the shortest GW for a newborn to survive with current medical technology (Figure 2) (23). The distribution of the lockdown exposure dose in the exposed group is shown in Figure S4.