Skip to main content
. Author manuscript; available in PMC: 2026 Mar 7.
Published in final edited form as: Prev Med. 2025 Mar 29;194:108274. doi: 10.1016/j.ypmed.2025.108274

Table 3.

Odds of in-range A1C1 follow-up tests associated with physical activity facility density and park distance among United States veterans with a new type-2 diabetes diagnosis in 2008–2018 by urbanicity, examined using two statistical modeling approaches.

 
Odds of all A1C1 follow-up tests in-range
Odds of repeated A1C1 follow-up tests in-range
Adjusted Odds Ratios2 95 % confidence intervals Adjusted Odds Ratios3 95 % confidence intervals
Density of physical activity facilities
 High density urban 1.00 1.00, 1.02 1.01* 1.00, 1.01
 Low density urban 1.04* 1.01, 1.07 1.03* 1.01, 1.04
 Suburban/small town 1.06 0.99, 1.14 1.02 0.97, 1.08
 Rural 1.19 0.93, 1.54 1.08 0.90, 1.29
Distance to seven closest parks
 High density urban 1.02 0.98, 1.07 1.00 0.97, 1.03
 Low density urban 1.02* 1.01, 1.03 1.01* 1.00, 1.02
 Suburban/small town 1.00 1.00, 1.01 1.00 1.00, 1.01
 Rural 1.00 0.98, 1.00 1.00 1.00, 1.00
1

A1C =glycated hemoglobin, an indicator of glycemic control.

2

Logistic regression used to model odds of all A1C follow-up tests in-range vs. less than all tests in-range with an asterisk indicating significance. All logistic regression models controlled for baseline A1C, age at diabetes diagnosis, sex, race/ethnicity, income/disability, marital status, comorbidities, and neighborhood-level socioeconomic status, land use mix, % black, and % Hispanic.

3

Generalized estimating equations (GEE) used to model odds of repeated A1C follow-up tests in-range vs. not-in-range with an asterisk indicating significance. All GEE models also controlled for time of A1C test in years since new type-2 diabetes diagnosis.