Spatiotemporal Patterns of Small for Gestational Age and Low Birth Weight Births and Associations With Land Use and Socioeconomic Status

In addition to small for gestational age (SGA) and low birth weight at term (LBWT), critically ill cases of SGA/LBWT are significant events from outcomes and economic perspectives that require further understanding of risk factors. We aimed to assess the spatiotemporal distribution of locations where there were consistently higher numbers of critically ill SGA/LBWT (hot spots) in comparison with all SGA/LBWT and all births. We focused on Edmonton (2008-2010) and Calgary (2006-2010), Alberta, and used a geographical information system to apply emerging hot spot analysis, as a new approach for understanding SGA, LBWT, and the critically ill counterparts (ciSGA or ciLBWT). We also compared the resulting aggregated categorical patterns with proportions of land use and socioeconomic status (SES) using Spearman correlation and logistic regression. There was an overall increasing trend in all space-time clusters. Whole period emerging hot spot patterns among births and SGA generally coincided, but SGA with ciSGA and LBWT with ciLBWT did not. Regression coefficients were highest for low SES with SGA and LBWT, but not with ciSGA and ciLBWT. Open areas and industrial land use were most associated with ciLBWT but not with ciSGA, SGA, or LBWT. Differences in the space-time hot spot patterns and the associations with ciSGA and ciLBWT indicate further need to research the interplay of maternal and environmental influences. We demonstrated the novel application of emerging hot spot analysis for small newborns and spatially related them to the surrounding environment.

. In Canada, the average rate of SGA was reported to be 9.1% and LBW (all gestational ages <2,500 g) was 6.4%, during 2015-2017 5; whereas in Alberta, the rate of SGA was 10.1% and LBW was 7.1%. These values have been increasing since before the beginning of our 2006-2010 study involving the Census Metropolitan Areas (CMAs) of Calgary and Edmonton.   In Calgary there were six distinct areas of hot spot patterns for all births (indicated by red toned symbols in Figure S4). The largest patch was in the northeast, and smaller ones in the northwest, northcentral, central, southcentral, and southeast. The five distinct areas of SGA occurred in the northeastern (largest), northcentral, central, southcentral, and southeast ( Figure S5). Much smaller areas were observed for ciSGA: central and scattered in the northwest ( Figure S6). Figure S7 shows two separate hot spot patterns for LBWT in the northeast, one in the east, one central, and an outlying community. The distinct areas for ciLBWT were northeast, central (but expanded beyond LBWT), and in the southeast ( Figure S8).
Note that Figures S9 through S13 are enlargements of Figures 4 through 6 of the main publication. Table S1. Statistically significant hot spot categories for the 5-and 3-year study periods are defined in terms of the total months aggregated by 3-month time steps. Table continues.

Pattern Category Emerging Hot Spot Definition New Hot Spot
A hot spot location for the last 3 months of the time series (the final time-step interval) and has never been a hot spot before. Consecutive Hot Spot A never-been-hot-before location with a single uninterrupted run of hot spot bins in the final time-step intervals, and for <90% of time-step intervals (Calgary: <54 months; Edmonton <32.4 months). Intensifying Hot Spot A hot spot location for ≥90% of the time-step intervals (Calgary: 54 of the 60 months; Edmonton: 32.4 of the 36 months), including the last 3 months (final time step), and there is an increase in the intensity of clustering of high counts in each 3-month time step. Persistent Hot Spot A hot spot location for ≥90% of the time-step intervals (Calgary: 54 of the 60 months; Edmonton: 32.4 of the 36 months) and has no increasing/decreasing trend in the intensity of clustering over time. Diminishing Hot Spot A hot spot location for ≥90% of the time-step intervals (Calgary: 54 of the 60 months; Edmonton: 32.4 of the 36 months), including the last 3 months (final time step), and there is a decrease in the intensity of clustering of high counts in each 3-month time step. Sporadic Hot Spot A hot spot location that is on-again then off-again for <90% of time-step intervals (Calgary: <54 months; Edmonton <32.4 months), and none of the time-step intervals have been cold spots. Oscillating Hot Spot A hot spot location for the last 3 months (the final time-step interval) that has previously been a cold spot, and <90% of time-step intervals (Calgary: <54 months; Edmonton <32.4 months) have been hot spots.

Historical Hot Spot
A location that is not a hot spot for the last 3 months (the final time-step interval), but ≥90% of the timestep intervals (Calgary: 54 of the 60 months; Edmonton: 32.4 of the 36 months) have been hot spots.

No Pattern Detected
Does not fall into any of the hot or cold spot patterns defined above or below. New Cold Spot A cold spot location for the last 3 months of the time series (the final time-step interval) and has never been a cold spot before. Consecutive Cold Spot A never-been-cold-before location with a single uninterrupted run of cold spot bins in the final time-step intervals, and <90% of time-step intervals (Calgary: <54 months; Edmonton <32.4 months).
Pattern Category Emerging Hot Spot Definition Intensifying Cold Spot A cold spot location for ≥90% of the time-step intervals (Calgary: 54 of the 60 months; Edmonton: 32.4 of the 36 months), including the last 3 months (final time step), and there is an increase in the intensity of clustering of low counts in each 3-month time step. Persistent Cold Spot A cold spot location for ≥90% of the time-step intervals (Calgary: 54 of the 60 months; Edmonton: 32.4 of the 36 months) and has no increasing/decreasing trend in the intensity of clustering over time.

Diminishing Cold Spot
A cold spot location for ≥90% of the time-step intervals (Calgary: 54 of the 60 months; Edmonton: 32.4 of the 36 months), including the last 3 months (final time step), and there is a decrease in the intensity of clustering of low counts in each 3-month time step.

Sporadic Cold Spot
A cold spot location that is on-again then off-again for <90% of time-step intervals (Calgary: <54 months; Edmonton <32.4 months), and none of the time-step intervals have been hot spots. Oscillating Cold Spot A cold spot location for the last 3 months (the final time-step interval) that has previously been a hot spot, and <90% of time-step intervals (Calgary: <54 months; Edmonton <32.4 months) have been cold spots.

Historical Cold Spot
A location that is not a cold spot for the last 3 months (the final time-step interval), but ≥90% of the time-step intervals (Calgary: 54 of the 60 months; Edmonton: 32.4 of the 36 months) have been cold spots.  Table S5. Logistic regression β coefficients (and 95% CI) for all and critically ill SGA/LBWT, modelled with proportions of only socioeconomic status (SES) and sum of all births for each location. Calgary also included industrial land use because it was not correlated with SES (see Table S3). Significant coefficients (p<0.05) marked by an asterisk (*); number of locations are indicated in Table 3.