The Predictive Acquisition Cost® (PAC) is an estimate of the acquisition cost of a drug created from a multi-dimensional predictive analytics model.
Motivating Idea – Predictive Analytics
Predictive analytics has been employed by the financial services industry for several decades to drive decision-making through the customer life-cycle, from solicitation to originations to account management to collections. Predictive models use proven statistical methods to analyze from thousands to millions of historical outcomes in order to identify combinations of factors that help to predict that outcome.
The use of predictive analytics to improve decision-making is a trend that is now revolutionizing many other industries, improving decision-making and producing compelling impact across a variety of domains, ranging from credit cards and insurance to direct marketing and major league sports recruiting. Predictive analytics is beginning to be applied to key problems in Life Sciences, such as detecting healthcare fraud and abuse. Drawing on our team’s expertise in developing commercially successful predictive analytics enabled solutions for many industries, including healthcare, we have designed what we believe can serve as a new drug-pricing standard for the pharmacy industry.
The PAC Multi-Factor Model
Using various available factors associated with the cost of a drug, via a multi-dimensional predictive analytics model we can estimate the acquisition cost to within sufficient accuracy to support pricing activity. Our statistical model is trained to synthesize various known attributes into an overall estimation of acquisition cost.
For illustrative purposes, consider a drug group’s various data elements that can be used to generate input factors that can in turn be synthesized by the model into our estimate of acquisition cost PAC.
The PAC multi-factor model provides multiple outputs on a daily basis to meet the various needs of the pharmacy industry from analytics to drug pricing benchmarks:
- PAC – estimate of the acquisition cost for a drug group
- PAClow and PAChigh – “low” and “high” range for the estimate of the acquisition cost for a drug group; the size of the range indicates the level of accuracy associated with the PAC model’s estimate of acquisition cost
- PACretail – proxy for use in existing contract vehicles (in place of AWP)
The graph below helps us quantify how superior PAC is to the AWP for tracking actual acquisition cost. For instance, the PAC value pinpoints acquisition cost within 15% for over 50% of drug groups, while AWP can only do so for fewer than 10% of the drug groups.
Another way to assess effectiveness of a benchmark is to see how closely it sort orders drugs compared to acquisition cost. Assign each drug group a Rank of 1 through 10 based on the actual acquisition cost (lower rank means lower acquisition cost). As the graph shows, the pool of drugs in each decile bin resulting from a PAC-based sorting has an average Rank very similar to the optimal acquisition-cost sort order; the same cannot be said for AWP.