Reference drug pricing

In my recent commentary,1 I was remiss in not noting that the “10% decline in the use of antihypertensives” was not statistically significant (p = 0.15). Nonetheless, regardless of statistical significance, the decline was real. This specific statistic reported the difference in utilization of antihypertensives between the Jan. 1997 predicted utilization (as extrapolated from the November 1996 observed utilization, Figure 2) and observed utilization.2

Instead, if one were to make the comparison using the Oct. 1996 utilization relative to the March 1997 utilization, the decline in antihypertensive utilization appears to be approximately 15%, a decline that is perhaps now also statistically significant. Moreover, the estimated trend in overall antihypertensive utilization does not catch up with the predicted utilization until the very last data point presented: April 1998 (Figure 2). In fact, during the entire 16-month post-reference-pricing period studied, the estimated trend is below the predicted line. This represents a real decline in antihypertensive utilization, in which health consequences remain unmeasured.

In the same paper, the authors estimated the cost savings to the BC drug plan from the application of reference pricing of ACE inhibitors to be $6 700 000 in the first year alone. This estimate is substantially larger than those found by Grootendorst and colleagues, who found savings of an average of $1 200 000 per year over the first 2 years after the policy was introduced.3 Given that both studies were based on the same administrative source data, could it be that one or both of the studies are methodologically incorrect? An indepth exploration of the causes of the divergence in cost savings is beyond the scope of this rebuttal.

Finally, recognizing that the source data for both of the above studies were collected to process prescription claims and not to perform outcomes research, and that underlying these claims are a complex array of business and personal incentives, exact estimates and an assessment of statistical versus clinical significance should be of secondary importance. What is important is deciding whether, on balance, reference pricing was good or bad for the average person with hypertension in BC.

Aslam H. Anis Department of Health Care and Epidemiology University of British Columbia Vancouver, BC