Page 17 - Delaware Medical Journal - July 2016
P. 17
SCIENTIFIC ARTICLE
FIGURE 4
Analysis of accidents by sex distribution within Delaware. The top panel shows a map indicating the percentage of females in the population by census tract in Delaware. The bottom panel shows ANOVA comparison of normalized accident frequency across quintiles of population. Stars are placed atop bars if that quintile is significantly different from the others indicated. Where only a single difference is noted, the star may be placed between the connector. Double stars (**) indicate a significant difference as per the BH correction, whereas single stars (*) indicate a difference at the P
< 0.05 level. The census tracts with the highest quintile in female population also have the highest normalized frequency of accidents (P < 0.05).
cities are shown in Figure 1b with points indicating crash locations and shading indicating cluster centers as determined by kernel density estimation. Figure 1b also overlays road map data to help further localize the injury events by street and neighborhood. In Figure 1c, injury counts per 10,000 census tract inhabitants are presented to offer a population normalized view of prevalence. Wilmington and Dover are again highlighted as the areas with the highest injury frequencies across the state.
ANOVA
Figures 2–6 indicate census tract demographic data by
quintile in the top panels and comparisons of injury frequency by quintile in the bottom panels. The color scheme has been preserved from top to bottom panels to better illustrate the trends in the census tracts. Analysis of variance in these quintiles indicated a number of trends that tracked with demographics. Normalized injury frequency increased with increasing black population such that it was 5.2 times higher (P < 0.001) from the highest to lowest quintiles (Fig. 2). Similarly, census tracts with the lowest median incomes had 6.8 times
the normalized frequency of injuries as those with the highest median incomes (P < 0.001), with injury frequency decreasing
as income increased (Fig. 3). On the whole, injury frequency increased with increasing female population in a census
tract; although, detailed analysis indicated that, in more rural areas, female population percentage did not correlate to injury (Fig. 4)
< 0.05 level. A similar trend existed between injury frequency and lack of high school degree. On average, injury frequencies were lower among more educated individuals (Fig. 5). However, while very rural areas had larger proportions of inhabitants without a high school degree, there were proportionally fewer accident frequencies in the lowest two educated quintiles (P
< 0.01). Finally, when considering population of children in a given census tract, the injury rates in the highest quintile were at least 3.7 times higher (P < 0.05) than any other quintile (Fig. 6). The only census tracts with such a high proportion of children are in Wilmington.
DISCUSSION
are meant to help understand the patterns related to pedestrian motor vehicle injury in this state. Data extracted from the 2012 Delaware Census indicate 6.1 percent of the population is under
Del Med J |
July 2016
| Vol. 88
| No. 7
209
