Multivariable logistic regression model of propensity score for participation in the self-administered OPAT versus healthcare-administered OPAT program.

The model’s area under the receiver operating characteristic curve was 0.91. Patients’ predicted probability of S- OPAT participation from the model is the propensity score used to control for selection bias in later outcome modeling. *The p-values for the main category terms (e.g., health funding source) are from the type 3 analysis of the main effects of the nine categorical variables, and the p-values for the individual category terms test the difference between each category (e.g., Medicaid) and its referent category (indicated by aOR = 1.00; e.g., Medicare), all based on a sample size of 1,168 patients. ‡Fiscal years run from 1 October to 30 September. For H-OPAT, fiscal year 2010 also includes the 9 mo before the fiscal year (1 January 2009 to 30 September 2009). aOR, adjusted odds ratio; ENT, ear/nose/throat; ref, referent category. Multivariable logistic regression model of propensity score for participation in the self-administered OPAT versus healthcare-administered OPAT program.