Influence of geocoding quality on environmental exposure assessment of children living near high traffic roads

BMC Public Health

Table 2 Bias and error in determining children at-risk based on proximity to major roads in Orange County, Florida.

Buffer Radius (m)	Number of children within buffer zone						Comparison measures
	Parcels	Street	Parcels yes/Street yes	Parcels yes/Street no	Parcel no/Street yes	Parcel no/Street no	Sensitivity (%)¹	Specificity (%)²	Odds Ratio (95% CI)³
50 meters	391	1413	151	240	1262	103212	39	99	0.28 (0.25 – 0.31)
100 meters	2090	2851	1392	698	1459	101316	67	99	0.73 (0.69 – 0.77)
150 meters	4717	5276	3899	818	1377	98771	83	99	0.89 (0.85 – 0.93)
250 meters	10729	10945	9704	1025	1241	92895	90	99	0.98 (0.95 – 1.01)
500 meters	26347	26597	25201	1146	1396	77122	96	98	0.99 (0.97 – 1.01)
1000 meters	54500	54614	53838	662	776	49589	99	98	01.00 (0.980 – 1.01)

¹ Sensitivity is the percentage of parcel geocoded children residing within the buffer zone that were correctly classified using street geocoding.
² Specificity is the percentage of parcel geocoded children residing outside the buffer zone that were correctly classified using street geocoding.
³ The odds-ratio is determined by comparing the parcel and street geocoded populations. The odds-ratio represents the odds that parcel geocoded addreses are within a given buffer radius, given that street-geocoded locations for the same address fall inside vs. outside the buffer radius. A value less than 1 indicates that the odds for the parcel geocoded population residing within the buffer radius is lower than for the street geocoded population

ISSN: 1471-2458