The Effects of Antihypertensive Step-Therapy Protocols on Pharmaceutical and Medical Utilization and Expenditures

The samples were relatively comparable in terms of the comorbidity measures. For all enrollees there was no statistically significant difference in the count of 3-digit ICD-9-CM diagnostic categories (P = .113). The Deyo-Charlson Comorbidity Index and the Chronic Disease Score were slightly higher in the step-therapy group (0.18 vs 0.16, P = .001, and 1.75 vs 1.57, P <.001, respectively). For antihypertensive users, there was no statistically significant difference between the step-therapy and comparison groups in the Deyo-Charlson Comorbidity Index (P = .053). The Chronic Disease Score was slightly lower in the step-therapy group (5.67 vs 5.78, P <.001), and the ICD-9-CM counts were slightly lower in the step-therapy group (8.31 vs 9.15, P <.001).

Table 2 describes prescription drug utilization and spending in 2006—the year that all plans implementing step therapy had step therapy in place. It should be noted that the effect of step therapy cannot be easily discerned from this table because the effects were measured about a year after implementation for 1 health plan, while effects were measured about 3 years after implementation for the other plan. For antihypertensive users, the mean number of days supplied per year of antihypertensives (ACE/ARB) was lower in the step-therapy group than in the comparison group (224.5 vs 252.4 days, respectively), while the antihypertensive discontinuation rate per user for the step-therapy group was higher than that for the comparison group (.13 vs .10, respectively) (all comparisons P <.001). The relationships between the step-therapy group and the comparison group were consistent across all service utilization and spending measures. Specifically, prescription drug utilization and spending measures were lower in the step-therapy group. However, the step-therapy group had more emergency room utilization, outpatient visits, and inpatient admissions. For all enrollees, the step-therapy group had more antihypertensive (ACE/ARB) days supplied per year than the comparison group.

Multivariate Model Results (Utilization)

In the multivariate models of antihypertensive users (Table 3), step therapy was associated with a higher rate of discontinuation of antihypertensive (ACE/ARB) medications, as the immediate effect on discontinuation was significant (P <.001), and the time-varying effect on discontinuation was positive and grew with time (an expanded version of Table 3 is available at www.ajmc.com as eAppendix Table 3). The effects of step therapy on the number of antihypertensive (ACE/ARB) days supplied per antihypertensive user was more complex, with an immediate 7.9% drop in the number of antihypertensive days supplied after implementation (coefficient -0.079, P <.001). However, the number of antihypertensive days supplied increased with time (coefficient 0.016 per quarter, P = .002), and 5 quarters after implementation of step therapy, the number of antihypertensive days supplied in step-therapy plans began to exceed the days supplied in comparison plans (ie, in the fifth quarter after implementation the combined effect was -0.079 + 5*0.016 = .001). Similarly, the number of antihypertensive days supplied for all enrollees dropped after step therapy was initiated (coefficient –0.071, P <.001), but then grew to equal that of non–step-therapy plans. Mirroring these effects, the total number of prescriptions per antihypertensive user that were filled in all medication classes dropped after implementation of step therapy and, after an initial decline, began to increase.

For antihypertensive users, step therapy was associated with an increase in outpatient office visits and inpatient admissions (Table 4). Step therapy also was positively associated with the number of emergency room visits, and the increase in emergency room visits grew with the amount of time elapsed since step therapy was implemented (an expanded version of Table 4 is available at www.ajmc.com as eAppendix Table 4).

Table 5 displays the coefficients of the spending models for antihypertensive users. For inpatient spending, the immediate effects of step therapy were not significantly different than zero (P = . 977), but the effects increased with time and step therapy was not associated with a significant effect on outpatient spending. In contrast, emergency room spending increased immediately after implementation of step therapy and the effects increased with time. Finally, while prescription drug spending declined 3.1% (P <.001) after implementation of step therapy, spending on prescription drugs in step-therapy plans grew over time to be closer to that in non–step-therapy plans (an expanded version of Table 5 is available at www.ajmc.com as eAppendix Table 5).

Table 6 shows the predicted results of the inpatient, emergency room, and prescription drug spending models—the 4 expenditure categories significantly affected by implementation of a step-therapy program. In the first quarter after step therapy was implemented, inpatient costs were lower in the step-therapy plan ($18.84). However, by quarter 4 the reverse was true; step-therapy inpatient costs grew relatively more expensive, reaching a $52.57 difference by quarter 8. Similarly, step-therapy prescription drug expenditures were lower through quarter 4 but exceeded the comparison group expenditures by $15.84 by quarter 8. Also, emergency room spending for step-therapy plans consistently exceeded the spending levels of the comparison group.

DISCUSSION AND CONCLUSION

Step therapy has become a common aspect of private health insurance plans; nevertheless, the research examining its effects has been limited. In particular, little attention has been paid to outcomes beyond reduced medication use and spending. In theory, if step 1 (preferred) medications are perfect substitutes for step 2 medications, and selecting the preferred drug was administratively seamless, then one would anticipate that step therapy would lower medication costs with no negative effects on drug use patterns, outcomes, or expenditures. One might hypothesize that step-therapy plans also could improve effectiveness by substituting more efficacious medications.