- Predicting sales of new medicines is highly inaccurate and subject to significant and often costly errors.
- While investment analysts can draw on research tools and experience, history suggests new drug forecasting will remain more art than science.
- Despite the high level of uncertainty and variability in new drug forecasts, the innovative medicine industry is alive and well.
Harlan Sonderling, CFA, Senior Healthcare Analyst | April 14, 2014
Predicting sales of new medicines is highly inaccurate and subject to error on both the low- and high-end. A recent McKinsey study found that this inaccuracy holds regardless of whether the drug is first in its class or a follow-on/improvement to an established drug class. For example, Lipitor, developed by Warner-Lambert and launched in 1996 with partner Pfizer (which soon acquired Warner) was the fifth and, at the time, the most potent statin drug approved to reduce LDL cholesterol. Consensus forecasts in 1995 were for sales five years out of $500 million. By 2000, however, sales had reached $5 billion on their way to the 2006 peak of nearly $13 billion. Conversely, at its launch in 1999, sales of GSK’s Agenerase for HIV were forecast to reach $1.5 billion by 2002; in fact, 2002 Agenerase sales were $70 million, an expensive disappointment.
A study by Citi Research concluded that over two-thirds of recent novel drug launches have failed to meet analysts’ first-year launch estimates and that there is a strong correlation between first-year and subsequent launch performance. Citi found that of launches exceeding first-year sales forecasts, 65% and 53% exceeded them in years 2 and 3, respectively. Of launches that missed first-year forecasts, 78% and 70% missed them in years 2 and 3, respectively. Work by the brokerage firm Cowen & Co. over many years confirms the regularity of drug companies’ and analysts’ significant and often costly forecast errors. For example, from years ago:
Source: Cowen and Company
Much is at stake in predicting market size and product sales. Drug companies create forecasts in order to guide huge R&D investments both inside their companies and with external collaborators and to build large marketing and sales infrastructure to support the product as it nears approval. Potential competitors make similar forecasts to guide their responses. Investment analysts forecast sales of drugs in development and newly launched products for their own obvious reasons. Forecasters are often highly educated and have broad backgrounds across the pharmaceutical and investment industries. Furthermore, they don’t operate in a vacuum. A drug’s therapeutic and product profile takes shape over a number of years as data emerge from clinical trials and are published and discussed at a wide variety of medical meetings. So, leaving aside the many predictive biases inherent in human behavior, why are these important forecasts rarely correct and can their quality improve?
One answer is found in therapeutic categories in which existing drugs are not very effective, the market is small, and a new drug’s potential is underestimated. Prior to the introduction of the “blockbuster” histamine H-2 antagonists, first Tagamet and then Zantac, for gastrointestinal (GI) reflux and ulcers, the GI market was seen as limited. Patients were treated with antacids, which had only modest efficacy, and anti-cholinergics, which were poorly tolerated. The H-2s, in turn, were later dwarfed by the more powerful acid anti-secretory drugs, called proton pump inhibitors, led by Prilosec and then Nexium. These drugs worked better, had fewer side effects, and won expanded indications like esophageal repair. Nexium sales peaked at $5 billion.
Conversely, Pfizer’s Viagra was the first PDE-5 inhibitor for erectile dysfunction (ED), a breakthrough in efficacy with few safety concerns. In the early 1990s, Pfizer forecast Viagra’s peak annual sales to be $500 million. By 1997, however, one year before the drug’s launch, the average brokerage forecast was for sales of $5 billion by 2002. In fact, Viagra’s sales peaked in 2012 at just over $2 billion, above Pfizer’s modest initial estimate, but well below Wall Street’s exuberance. Viagra was limited by high initial pricing, impediments to insurance coverage, extensive counterfeiting and the launch of two competing drugs. Drugs that treat orphan (rare) diseases were initially thought of as limited commercial opportunities because of small patient population. However, with competition limited by the FDA to encourage innovation, some orphan drugs have been priced in multiple hundreds of thousands of dollars, leading to bonanzas for their developers.
Sometimes drug companies simply fail to support product launches, fail to seek expanded indications or shy away from the risk of comparative clinical trials that could differentiate their drugs. Increasingly, large payers – governments, managed care insurers and pharmacy benefit managers (PBMs) – are negotiating directly with, or attempting to dictate prices to drug manufacturers with the return promise of higher volume through better formulary positioning. Representative Henry Waxman recently called on a hepatitis C innovator to justify its price, which directly caused industry analysts to question their revenue assumptions. Even doctors have become more sensitive to drug prices, having heard from their patients and their insurers; they are increasingly pushing back on higher prices and seeking less expensive alternative therapies. Patent challenges have become increasingly unpredictable, and patent settlements against generic challengers are subject to more regulatory oversight. Emerging market sales have become more important to the pharmaceutical industry. Developing economies fluctuate more often and with higher variability than developed economies, and their level of regulatory oversight is lower. Several major manufacturers recently learned the hard consequences of rapid growth in China. One of the key reasons behind global, specialty and generic pharmaceutical industry consolidation may be the companies’ strategic desire to diversify product portfolios in order to hedge commercialization risk in a difficult and less predictable new drug launch environment.
Can the quality of commercial sales predictions improve? Despite analysts’ research tools and experience, history suggests new drug forecasting will remain more art than science. At Columbia Management, our healthcare team’s analysis involves using probability-adjusted therapeutic category and drug class forecasts and responding more quickly than in the past to initial launch data. Our “art” lies in our mix of science and business, with a clear eye both on the past and future. We dig deeply into therapeutic categories to attempt to identify products under development or at or near launch that are marked improvements over existing therapies or that have potential to fill unmet patient needs, what are commonly called “first-in-class” or “best-in class.” How far behind are the second and third drugs in the new class, and are they material improvements in efficacy or safety, or both, for which the market is willing to wait? Is the drug in clinical trials across a range of indications beyond its initial one, and can the product win regulatory approval across geographies?
We interact regularly with both doctors/scientists and payers; the former on drug mechanisms, side effects, patient need and alternatives, and the ease of administration relative to other therapies; the latter on their interpretation of novelty (whether the drug fills an unmet need or represents a significant improvement, and at what perceived cost-benefit to the patient and payer), existing or imminent alternatives, pricing and the product’s likely position within increasingly restrictive prescription drug formularies. For drugs that are promoted similarly to other consumer products, that is, more