Case study of the effects of office-based generic drug sampling on antibiotic drug costs and first-line antibiotic prescribing ratios.

Background: Health plans and members benefit from the substitution of lower-cost drug therapies that achieve the same clinical outcomes as higher-cost drugs. Previous research suggests that generic sampling programs produce drug cost savings overall, but the effects attributable to acute therapies are unknown. Encouraging physicians to prescribe less expensive, first-line antibiotics may help reduce direct drug costs associated with prescribing potentially unnecessary, and more expensive, second-line agents.

Objectives: To determine the effects of an automated, office-based generic drug sampling kiosk on (a) prescribing of first-line oral antibiotic agents as a ratio of total antibiotic prescribing and (b) average antibiotic drug cost per claim.

Methods: This managed care organization of 2.3 million members with pharmacy benefits collaborated with a vendor that developed an automated generic drug kiosk that allows for the dispensing of samples of generic medications within the prescriber's office. Among the samples contained in the kiosk were 6 generic, first-line oral antibiotics, representing 8 unique drug-strength options. Drug costs were defined as the ingredient cost of the drug claim, which includes plan cost, member cost share, and any dispensing fees or administrative program costs associated with the sampling program. In a difference-in-difference analysis, changes in outcome measures (antibiotic drug cost per claim and dispensing rates of first-line antibiotics) from 2003 (baseline year) to 2005 (post-implementation year) were compared among kiosk prescribers (n=179) and nonkiosk prescribers who were part of the same provider network (n=7,236). A cross-sectional analysis of the same outcome measures compared kiosk (n=396) and nonkiosk prescribers (n=10,267) in 2006. All statistical analyses were performed using t-tests of logtransformed data.

Results: The mean cost per claim dropped by $4.14 (12.3%) from $33.56 in 2003 to $29.42 in 2005 for the kiosk prescribers and by $3.35 (8.8%) from $38.26 in 2003 to $34.91 in 2005 for nonkiosk prescribers, but the mean change from 2003 to 2005 in the difference-in-difference analysis was not statistically significant (P=0.123). The first-line antibiotic prescribing ratio declined by 2.1 absolute points from 49.1% in 2003 to 47.0% in 2005 for the kiosk prescribers and by 3.4 points from 46.0% in 2003 to 42.6% in 2005 for the nonkiosk prescribers, but the difference- in-difference analysis showed that the change was not statistically significant (P=0.901). A cross-sectional analysis of 2006 data revealed significant differences between the kiosk prescribers versus their nonkiosk network counterparts for both first-line antibiotic prescribing rates (42.0% and 41.4%, respectively; P=0.028) and antibiotic cost per claim ($28.44 and $32.40, respectively; P<0.001). While the results of the cross-sectional analysis are statistically significant, the practical significance of the results is less evident.

Conclusions: The benefits of including short-term medications such as first-line antibiotics in a generic drug sampling program are difficult to quantify, and the cost outcomes are unlikely on their own to justify generic drug sampling. However, acute (short-term) medications may be an effective part of an overall program to promote generic prescribing for all conditions.