Concerns about rising drug costs continue to grab headlines, but a new analysis suggests that health care organizations looking for ways to manage these costs may be relying on seriously flawed numbers that routinely overestimate cost growth.
The Partnership for Health Analytics Research (PHAR, LLC), a leading health services research consultancy, announced that predictions of health care costs made prior to the introduction of new drugs are often dramatically overestimated. These overestimates may, in turn, cause health insurers to make access to drugs more difficult for patients than may be clinically appropriate.
In one example, the Institute for Clinical and Economic Review (ICER) predicted the one-year cost of the two PCSK9 inhibitors, a new class of injectable treatments for patients with high cholesterol who are at high risk of CV events, would be $7.2 billion. Based on the initial quarters of reported sales, the actual number will be closer to $83 million, or 1.2 percent of the prediction, an overestimate of $7.1 billion. As Dr. Michael S. Broder, MD, President of PHAR notes: “Overestimating drug costs by so much cannot lead to good decision making. In fact, it is likely that patients feel the negative effects of such predictions in the form of early access restrictions and higher copayments.”
The overall study examined publicly released predictions for the cost of 14 drugs launched in a variety of indications since 2012. In addition to the two cholesterol lowering drugs, PHAR studied three drugs to treat various forms of cancer; three for hepatitis C; two for obesity management; and one each for cystic fibrosis, heart failure, psoriasis, and diabetes.
The researchers found 24 published predictions for the 14 drugs and compared them to actual sales. On average, predictions were 11 times higher than actual sales. That is, for every $11 of predicted cost there was only $1 of actual cost to the healthcare system.
In other examples, PricewaterhouseCoopers and Bloomberg both predicted the drug cost to the healthcare system for Viekira Pak, for hepatitis C, and Opdivo, for metastatic melanoma skin cancer. Their predictions for Viekira Pak were over $2.9 billion; actual first year sales were 28 percent of the prediction. Similarly, the predicted cost of Opdivo was $1.7 billion; actual sales were 48 percent of predicted. Underestimates were less common—Cosentyx, originally approved for psoriasis, was predicted to cost $120 million in the first year after launch; actual sales were more than twice that. Taken together, these findings raise concern about the usefulness of pre-launch predictions for payers, patients, and prescribers.
The study findings are consistent with other recent research. A prominent physician predicted in the New England Journal of Medicine that the PCSK9 inhibitor class alone would raise annual health care premiums by $124 per person. Yet, a report by the advisory company Avalere Health showed that all prescription drugs accounted for only a $3.29 average increase in per member per month premiums.
The results of the current study were submitted for presentation at the Association of Managed Care Pharmacy’s (AMCP) upcoming NEXUS conference.
Dr. Broder summarized the significance of this research, saying, “Drug usage predictions are used to inform pharmacy policy. Our goal in pointing out these discrepancies is to help improve predictions and decision making. Ideally, payers and policymakers would be able to use them as planning tools. In their current form, I’m afraid these predictions are so far off that they may just be scaring people into making bad decisions about limiting access to new drugs.”