Volume 20, Number 5—May 2014
Dispatch
Factors Associated with Antimicrobial Drug Use in Medicaid Programs
, Joshua P. Metlay, Steven C. Marcus, and Jalpa A. Doshi
Table 2
Factors associated with antimicrobial drug use among 194,874 adult Medicaid patients, 40 US states, 2007*
| Variable | % Visits at which drugs were prescribed | Odds ratio (95% CI) |
|---|---|---|
| Age, y | ||
| 21–29 | 49.8 | Referent |
| 30–39 | 53.7 | 1.10 (1.05–1.16) |
| 40–49 | 53.8 | 1.09 (0.99–1.20) |
| 50–59 | 53.3 | 1.09 (0.96–1.23) |
| 60–64 | 51.3 | 1.02 (0.89–1.16) |
| ≥65 |
43.2 |
0.77 (0.63–0.94) |
| Sex | ||
| F | 52.4 | Referent |
| M |
51.1 |
0.94 (0.90–0.99) |
| Race | ||
| White | 56.9 | Referent |
| Black | 48.9 | 0.82 (0.74–0.91) |
| Hispanic | 45.6 | 0.73 (0.64–0.82) |
| Other |
45.0 |
0.75 (0.61–0.92) |
| Diagnosis at index visit | ||
| Cold or acute URI, ICD-9 codes 460 and 465 | 40.0 | Referent |
| Acute bronchitis, ICD-9 code 466 |
69.0 |
3.32 (2.78–3.95) |
| RxHCC score† |
0.94 (0.88–1.00) |
|
| Quarter of index visit date | ||
| Jul–Sep | 52.8 | Referent |
| Jan–Mar | 52.3 | 1.04 (1.01–1.08) |
| Apr–Jun | 52.1 | 1 (0.97–1.04) |
| Oct–Dec |
51.1 |
0.98 (0.95–1.01) |
| Residence in low-education county‡ | ||
| No | 52.1 | Referent |
| Yes |
52.1 |
0.96 (0.85–1.09) |
| County–level annual per capita income (in $1,000)§ | 1.00 (1.00–1.00) | |
| Residence in urban area | ||
| No | 55.8 | Referent |
| Yes |
50.2 |
0.91 (0.82–1.00) |
| Residence in state participating in CDC Get Smart campaign¶ | ||
| No | 57.7 | Referent |
| Yes |
50.7 |
0.74 (0.62–0.88) |
| No. primary care physicians/10,000 persons in county# | ||
| <2.2 | 56.8 | Referent |
| 2.2–3.4 | 56.2 | 0.96 (0.87–1.07) |
| 3.5–4.7 | 55.4 | 0.91 (0.80–1.04) |
| 4.8–6.5 | 51.4 | 0.84 (0.73–0.96) |
| >6.5 | 48.2 | 0.76 (0.66–0.88) |
*Data are from the 2007 Medicaid Analytic Extract files linked with the Area Resource File. URI, upper respiratory tract infection; ICD-9, International Classification of Diseases, Ninth Revision; RxHCC, prescription drug Hierarchical Coexisting Condition; CDC, Centers for Disease Control and Prevention.
†Modified RxHCC score used here, wherein coefficients for age and sex are zeroed out in the score calculation because regression models separately control for these variables. Range in sample described here 0–5.3. A higher score indicates a higher medical comorbidity burden.
Odds ratio indicates the increased odds of antimicrobial drug use associated with per unit increase in the score.
‡The categories were based on quintile of county-level number of PCP physicians per 10,000 persons. Each category includes 644 counties.
§Defined as a county with ≥25% adults without a high school diploma.
¶In separate analysis, the county-level annual per capita income was coded as categorical variables according to the quintile of the measure across all counties in the Area Resource File. Similar to the continuous variables, the categorical variables were not significant.
#In our sample, 33 of 40 states participated in the CDC Get Smart campaign during 2002–2006.