Table 2. Association of Pharmacy Brand With Tobacco Sales to Minors in FDA Inspection of Chain Pharmacies, 2012-2017.
| Factors Associated With Sale to Minor in Inspection (1 = Yes, 0 = No) | OR (95% CI) for Model 1 | aOR (95% CI) for Model 2 |
|---|---|---|
| Pharmacy brand | ||
| Walgreens | 1 [Reference] | 1 [Reference] |
| CVS | 0.63 (0.53-0.75) | 0.70 (0.58-0.85) |
| Rite Aid | 0.40 (0.34-0.46) | 0.41 (0.35-0.47) |
| Other chains | 0.61 (0.45-0.84) | 0.65 (0.47-0.89) |
| Year | NA | 1.00 (0.96-1.04) |
| Lung Association grade (coded: 0 = F, 4 = A) | NA | 0.74 (0.52-1.04) |
| % of Non-Hispanic white individuals | NA | 0.95 (0.92-0.97) |
| Median household income, z-scored within county | NA | 1.00 (1.00-1.01) |
| Intercepta | 0.10 | 0.12 |
Abbreviations: aOR, adjusted odds ratio; FDA, US Food and Drug Administration; NA, not applicable; OR, odds ratio.
Random intercept for state to address within-state dependence (intraclass correlation, 0.11). Intercept is exponentiated. In model 1, n = 23 841 (after exclusion of 22 CVS inspections decided in 2015-2017 after sales had stopped). Final model is n = 23802 owing to sporadic missingness in neighborhood-level variables, geocoding, and 2015-2017 CVS inspections. Neighborhood characteristics are scaled for ease of interpretation and model convergence; z scores were multiplied by 10 (eg, z = 0.13 is coded as 1.3), and the percentage of white individuals was divided by 10 (eg, 13% white is coded as 1.3). Nevada did not inspect any of the pharmacy chains we examined during this time period. We tested a quadratic term of year squared; it improved the fit of the model and was included in model 2. The model is not sensitive to inclusion or exclusion of the 22 CVS inspections conducted after 2014.