By Keith Passwater & Dave Nelson
Extending life and improving quality of life for literally millions, drug treatments such as aspirin, penicillin, insulin, AZT, cancer drug Herceptin, Harvoni, and Lipitor are nothing short of miracle drugs. Recently developed therapies are also impressive, such as single-dose Kymriah, which has an unprecedented 83 percent remission rate for children with acute lymphoblastic leukemia. Another example is the newly available gene therapy Luxturna, which can cure blindness in carefully selected patients.
Such therapies can achieve outcomes that are truly miraculous for those cured patients; they can also be staggeringly expensive.
While anyone with a heart would agree that these miracle drug benefits are wonderful for patients, we need to find ways to harness these invaluable benefits while also finding ways to curb the growth in drug costs in the U.S. for all parties—consumers, payers, and manufacturers alike.
The cost of prescription drugs in the U.S. (not including drugs dispensed in a doctor’s office or hospital) is more than 14% of total health care costs and will equal a whopping $500 billion in 2018. Further, prescription drug spending is growing faster than other health care spending and, if not curbed, will continue to consume a greater share of total health care dollar every year.
A key contributor to increased pharmaceutical cost is the development of new, expensive pharmaceutical treatments, which require a great deal of time and money to develop. At least part of the cost of therapies that reach the market goes to fund the >99.99% of the initially promising new chemical entities (NCEs) fail to meet that promise; medicines that survive the development gantlet in a way pay for those that fall away so manufacturers can continue searching for new therapies that, in some case, prove to be miracle drugs that cure previously incurable medical conditions.
The American Academy of Actuaries in March 2018 released an issue brief, Prescription Drug Spending in the U.S. Health Care System.The issue brief sets the stage for any discussion of drug prices in the U.S.; in a nonpartisan, objective way, it highlights prescription drug issues present in our health care system, including cost drivers, the effect on various stakeholders, and possible approaches that might help address the health care cost growth associated with prescription drugs.
Government, insurers, plan sponsors, pharmaceutical companies, and patients are seeking to find ways to reduce the cost of drug treatments. This article will describe the U.S. drug development process, cost drivers, cost control options, and our perspective with respect to these issues.
U.S. Drug Development Process
The process for a new medicine to be offered in the U.S. is regulated by the Food & Drug Administration (FDA) and is often described in five steps (the verbiage that follows is specific to small-molecule drug development, although there is an analogous process for biologics):
Drug discovery usually begins with a spark of inspiration that a compound might be effective in treating a given health condition. This inspired concept is then tested in the laboratory and, if the results look promising, the concept moves to the next step. It is estimated that of 10,000 to 30,000 NCEs in discovery, a mere 10-20 will graduate to the next step.
- Preclinical Research
Once necessary funding is secured, the manufacturer will launch additional testing to determine whether the NCE has the potential to cause serious harm in humans, also called toxicity. This is also known as preclinical safety assessment. This is where potential human adverse events are first identified and safe clinical starting dose projections are made. The two types of preclinical research are in vitro (basically research in a test tube) and in vivo (which involves testing in a live animal). The beginning of in vitro testing (used to study the drug mechanism) is commonly understood to mark the transition from drug discovery to drug development. In vivo testing is more expensive and generally takes longer but is far more useful in predicting human toxicity.
- Clinical Research
Once a manufacturer is convinced that an NCE is worthy of further pursuit, the manufacturer submits an Investigational New Drug (IND) application to the FDA, which is a request for authorization to administer an investigational drug or biological product to humans to initiate testing in human volunteers. Completing the IND will require the manufacturer to determine the clinical endpoints. These predetermined critical success factors are outlined as the primary and secondary endpoints. Those monitoring drugs in the development pipeline will often hear that an experimental drug “missed” its primary endpoint (e.g., progression-free survival, for an oncology drug) but “hit” its secondary endpoint (e.g., overall response rate, for an oncology drug). In such circumstances, it would mean that patients showed response to the drug (e.g., the tumor stopped growing) but their lives were not extended beyond what is seen with the current standard of care. In that case, the program to develop this drug would usually be terminated before any sunk costs could be recouped.
Clinical trials are expensive in part because they present such a significant challenge to the drug concept under review. If the FDA finds a problem, it can order a clinical hold to delay a trial, or interrupt a clinical trial if problems occur during the study. By a wide margin, this is the most expensive stage of drug development.
A step not formally included in the FDA timeline, but critical to the commercialization of a new drug treatment, is the development of the drug manufacturing process. This step may not be obvious because one assumes researchers must have a process for developing the drug that has been produced for study to this point. However, chemical engineers understand that a process appropriate for developing small quantities in the lab will generally require considerable modification to perform at the vast scale required for mass production. Furthermore, the FDA requires that a very detailed description of the development, manufacturing, and testing specifications and control points and parameters be submitted to the agency as part of the drug approval process. The FDA will also perform a preapproval inspection prior to granting marking authorization. The inspection consists of an onsite review of the manufacturing sites by agency officials.
Failure to receive approval due to whatever reason is extremely costly for the manufacturer. A recent example is the migraine drug Ajovy produced by Teva.
Teva was eager to begin selling the promising miraine drug Ajovy after having passed many of the development hurdles. But, its South Korean manufacturer partner, Celltrion, was issued eight regulatory violations by the FDA, including a lack of employee training and inadequate protocols for assuring product quality and purity. As a result, Teva lost the ability to market Ajovy for three months, during which time other manufacturers caught up in the development of their own pipeline migraine drugs.
- FDA Review
Once a manufacturer obtains clear evidence that a compound meets safety and efficacy requirements, a New Drug Application (NDA) must be filed with the FDA to start the review and approval process. A proper NDA must include information such as clinical results; dosage form and strength; the route of administration; the identification numbers of all INDs; all testing results; and the drug’s proposed indications for use. Additionally, an NDA requires payment of a $2.4 million user fee to the FDA. After approval, the FDA works with the manufacturer to develop and refine prescribing information for best use of the drug, also known as labeling which takes additional time.
- FDA Post-Market Safety Review
Even though a drug has been reviewed and approved for sale and use, there is still a risk that new safety issues will arise once the drug is used by a wider patient pool. This is because clinical trials are often performed on very select patient panels in order to minimize the influence of confounding variables (e.g., co-morbidities). For example, Vioxx was approved by the FDA in 1999 as a pain reliever. Unfortunately, an estimated 40,000 people died from heart attacks due to an undetected cardiovascular risk. Merck, the manufacturer of Vioxx, withdrew the product from the market in 2004. In response to that and other incidents, in May 2008 the FDA launched the Sentinel Initiative, which requires ongoing monitoring and reporting of adverse reactions to approved compounds in the market. As a result, the public is safer but drug patent holders do not have certainty about a compound’s permanent availability even after FDA approval.
In total, the required time and attrition of each step puts the odds of a drug compound making it to market at approximately 1 in 20,000, with a timeline that would average between eight and 12 years. The total cost of advancing a drug therapy through approval is estimated at nearly $3 billion.
Efforts to Accelerate the Process
In the last few years, the Food and Drug Administration has adopted new approaches that are intended to speed up drug development. Here are four specific new approaches which may be applied singularly or in combination for a particular new drug:
- Priority Review commits the FDA to taking action on an application within six months.
- Breakthrough Therapy expedites the development and review of drugs that may demonstrate substantial improvement over available therapies.
- Accelerated Approval allows drugs to be approved based on a surrogate endpoint if they target an unmet medical need for a serious condition.
- Fast Track is a process designed to facilitate the development and expedite the review of drugs that target an unmet medical need for a serious condition.
In addition, the recently passed sixth authorization of the Prescription Drug User Fee Act (PDUFA VI in 2017) requires the FDA to respond more quickly to requests for approval of new drugs. Time will tell if improvements like these make the drug approval process more efficient, faster, and cheaper.
Long-Term Changes in Drug Development
Scientific advance has changed drug development significantly over the past 50 years. Historically, scientists looked for small molecules in the natural world to cure disease. That served as a significant limiter on discovery. Over time, synthetic methods were developed to generate small molecule candidates, and high-throughput screening robots dramatically accelerated the search for new candidate molecules.
Additionally, biotech companies have developed new approaches beyond small molecules, which present a great breakthrough in allowing scientists to address the many antibody targets that are not druggable with small molecules. These new approaches include determining the precise genetic cause of an illness and then searching for proteins that may control the disease. Proteins allow for more targeted and often more powerful disease inhibition as compared to small molecule drugs. These therapies are often administered via intravenous infusion or subcutaneous injection as they typically cannot survive the harsh environment of the stomach. Major companies like Amgen and Genentech started to use biotech methods to compete with the traditional methods of small molecule discovery with significant success. And, now most large pharmaceutical companies include both small molecule and protein therapeutics in their portfolio to access a broader spectrum of biologically relevant targets for clinical evaluation.
In another major development, researchers on April 14, 2013, published the final sequence mapping of the human genome. Knowledge of the human genome has made it possible to identify the cause of many illnesses. At this point, most new drugs—whether small molecule, protein therapy, or another type—are influenced by modern genomics.
Just in the last 10 years, advances in identifying the specific genes that cause various cancers has led to artificial stimulation of the immune system to improve a patient’s natural ability to fight cancer (so-called immunotherapy agents). For instance, former President Jimmy Carter was recently treated successfully with Keytruda, a PD-1 inhibitor, which targets and blocks the Programmed Death receptor protein allowing the body’s immune system to attack the cancer. The 2018 Nobel Prize in Medicine was awarded to James Allison and Tasuku Honjo for their discovery of advanced immuno-oncology (which is the basis for Keytruda and related therapies).
As scientific advance has changed drug development over the last 50 years, health insurers, pharmacy benefit managers (PBMs), retail pharmacies, independent specialty pharmacies, and medical providers have all organized to give special treatment to patients receiving new costly drugs and biological products. It is predictable that continued scientific advance will result in a growing specialty pharmacy market, which helps a small but rapidly increasing number of people obtain significant (often life-saving) health improvements. It is also likely that the costs of specialty pharmacy benefits will continue to increase much faster than other health care costs.
Drug Development and AI
Discovering a new drug is a long, expensive process as thousands of compounds go through a series of tests to find each viable drug. Another category of technologies in which companies are investing heavily is artificial intelligence (AI) in hopes that it will make the process faster, cheaper, and more effective. While there is some skepticism in the clinical research community, one large European AI company (Benevolent AI) claims to have achieved a two-year reduction and 60 percent cost decrease for the early stages of drug development. AI proponents aim to improve on that by eliminating three to five years from the development process, which would be a huge financial benefit to all stakeholders, including patients.
As a result, pharmaceutical companies are investing significantly in AI and hope to apply the technology in at least five ways: 1) find new drugs that could be potential therapies, 2) predict how well drugs will perform in testing, 3) discover previously untried drug combinations that work as effective treatments, 4) find new uses for previously tested drugs, and 5) create personalized medicines based on genetic markers.
In this debate between those who claim AI can solve vexing problems and those skeptical of AI’s capabilities, actuaries can add tremendous value by substitut[ing] facts for appearances and demonstrations for impressions.
Drugs in the Pipeline
All of this information leads one to ask, So, what sorts of drugs are in the development pipeline now that might impact cost and quality of life in the future? In addition to the cancer-fighting drugs mentioned above, experts predict the emergence of new drugs to treat hemophilia; immunotherapies for peanut allergies; treatments for nonalcoholic steatohepatitis (NASH); and new therapies for cystic fibrosis, Alzheimer’s and rheumatoid arthritis. Given the rising prevalence of these conditions, the commercial market for such drugs is growing, which explains why pharmaceutical companies are investing so much in the pipeline for these drugs. In addition, it is likely that the increased knowledge of the human genome will allow drug manufacturers to target drugs based on a person’s individual genetic makeup.
A major driver of high drug prices is high launch prices along with high annual increases for patented brand-name drugs. While some argue that the high price of drugs is commensurate with the high risk and long duration of drug development, the degree to which that linkage is causal is less than clear and continues to be studied. One thing all can agree on, though, is that the costs of new drugs are very high.
In addition to drug development cost, the manufacturer of a new drug with potentially dramatic benefits also incurs the cost of marketing, such as presenting to health insurers and pharmacy benefit management companies for addition to their formularies. These costs outpace the drug development costs over the life of the product and provide little benefit to the patient but manufacturers deem these necessary to get the drug to the patient.
Compounding the development cost for pharmaceutical companies is the rapidly accelerating business environment. Historically, while a new drug might stand alone in the market for a decade, today a new drug may face competitors within one year, further diluting the company’s likely recovery of return on investment.
Also, a significant increase in drugs recognized under the Orphan Drug Act contributes to the high cost of drugs. Passed in 1983, the Orphan Drug Act incentivizes rare disease drug treatments by offering incentives such as tax credits, fee waivers, and seven years of post-approval market exclusivity for the approved indication. As a result, rare disease treatments have been developed—which is, of course, a good thing. However, since the early 2000s there has been a dramatic spike in the number of orphan-designated drugs with many targeted at biological subsets of larger disease categories. Clearly, the law worked to spur investment and treatments are now available to patients that would normally not have treatment options. There is also some debate as to whether the definition of orphan has been stretched beyond the original intent of the act.
As mentioned earlier, new drugs are usually protected under patent during development. The patent protects the manufacturer’s investment in the drug’s development by giving them the sole right to sell the drug while the patent is in effect. When the patents or other periods of exclusivity on brand-name drugs expire, manufacturers can apply to the FDA to sell generic versions at lower prices than the patented brand. And this price competition from generics tends to drive down the cost of drug treatments; however, pharmaceutical patentholders increasingly protect adjacent elements of their intellectual property, such as manufacturing methods for biologics. Further, some brand owners have been known to pay generic manufacturers to delay production so that the brand owner can realize high prices longer.
Adding to the cost of drug treatments is the reality that patients, providers, and payers lack information about the comparative effectiveness of drugs at the point in time when critical health care decisions are made. Further compounding the problem is that the pharmaceutical distribution system does not make essential pricing information available to patients, providers, and payers at the point of care—information that patients and their providers need when deciding on the best course of treatment. Without this critical information at the point of decision, providers and patients might revert to a higher-priced, better-known brand drug, rather than a comparable generic.
Recently pharmaceutical companies have considered what they view as value-based pricing to ensure that the prices charged for drugs reflect the benefits they provide, either in terms of longer life or better quality of life. For instance, Spark Therapeutics announced that its gene therapy treatment for a very specific inherited eye disease (Luxturna) will cost $850,000 per patient but be subject to a rebate that returns the money if the patient is not cured. In another instance of value-based pricing, Gilead priced hepatitis C drugs based on the avoidance of a transplant.
In addition to traditional drug pricing practices, there are outlier practices that need to be understood. For example, Mylan drew the ire of customers and Congress when it raised the price of an epinephrine pen to more than $600 for an EpiPen two-pack, having patented the delivery device for the generic (expired patent drug). Mylan later announced plans to sell a generic version for $30.
Government Cost Control Options
While it’s not anticipated that federal regulation will soon make radical changes to the pharmaceutical market, some options with many pros and cons have been debated. For example, the FDA could further accelerate the drug approval process, although that might expose patients to great safety risks. Another federal option would be to prevent brand manufacturers from paying generic companies to delay manufacturing of generics. More possible federal actions could include taking effective steps to speed up the drug development process, implementing pricing controls, or allowing Medicare and Medicaid drug prices to be directly negotiated with manufacturers, which federal law currently limits.
Health Insurance Company Cost Control Options
Health insurers have some history of seeking drug cost control. One of the methods insurers use to control drug costs is to try to design a cost-effective benefit plan. Health insurance companies typically organize prescription drug benefits into at least four tiers: generic, preferred brand, non-preferred brand, and specialty. Each tier offers different deductibles, co-insurance, and copays that can be configured to steer members to cost-effective usage. In some cases insurers require that drugs be delivered from a mail order pharmacy. With these practices, health insurers aim to ensure what they consider appropriate prescription treatment for the patient together with optimal cost for the plan sponsor. (For a variety of reasons, drug manufacturers also may offer copayment coupons and discount card programs designed to lower the patient’s share of the cost. These coupon programs tend to increase the insurers’ cost because in some cases the cost of the drug net of the coupon exceeds the cost of the alternative.)
Besides benefit plans, insurance company utilize other methods to manage prescription drug costs. A partial list of observed additional cost control methods follows:
- Partial Fill Programs that require less-than-30-day fill for very-high-cost specialty drugs. These programs reduce waste if some patients cannot tolerate the drug and would quickly discontinue use.
- Drug Utilization Review programs typically apply for chronically ill patients that are taking many drugs. These programs are viewed by insurers as likely to reduce adverse drug interactions, improve health, and lower costs.
- Formularies are typically designed to increase use of cost-effective drugs and to allow the insurer to negotiate rebates.
- Step Therapy is used to make patients first try a preferred drug and switch to a non-preferred drug only if the preferred drug is ineffective.
- Preferred Pharmaceutical Networks allow the payer to negotiate larger discounts. Patients are incented to use the preferred pharmacies through lower copays.
- Prior Authorization requires review for medical necessity before a drug is dispensed.
- Reference Pricing ties reimbursement to the national market price of equivalent drugs.
Our Actuarial Perspective
In our view, the first thing an actuary can do when working in the pharmaceutical area is develop a relationship with pharmacists and physicians, who know more about drugs than you will ever know. If an actuary wants to successfully identify the cost impact of new drugs and predict future drug costs, they must learn the language of the drug industry and cultivate relationships with knowledgeable experts, particularly those who are involved in drug discovery and development.
While actuaries must rely on others for assistance, there is a broad and important role for actuaries in the understanding and management of drug costs. Actuaries have the set of analytical skills that are critical to drug cost management, trend identification, product pricing and general issue analysis.
Job two, after developing a relationship with physicians and pharmacists, is setting up data stores that enable thorough analysis of drug costs. Only with advance planning are actuaries able to measure drug trends by type of drug (branded, generic, specialty, etc.), by drug, by drug category, by plan of benefits, compared to predicted trends, for larger customers, by site of service (hospital, doctors office, home, etc.), in comparison to industry benchmarks, and over time.
Actuaries who work with prescription drugs are often asked to cost different drug benefits and work with marketing to identify benefit options that are attractive to customers. To do this well, one will want to keep abreast of drug plan design features that are offered by Insurance company competitors. Drug cost management savings are critical to pricing and product development efforts. When pricing drug plans, actuaries are often asked to predict (and later monitor) the value of drug cost management and network development efforts.
Special expertise is required for Medicare drug pricing. Actuaries need to understand and comply with regulations from the Centers for Medicare and Medicaid Service (CMS) while meeting business goals. Without specialized Medicare expertise, the company is not going to be able to offer sellable products.
Learning is a key part the job. It is helpful to keep abreast of earnings call information from drug companies and other insurance companies; earnings calls are a publicly available resource for information that may be germane to the actuary’s work. It is also helpful to learn by working with consultants who have pharmaceutical pricing expertise and by building a network of colleague actuaries at other companies who do similar drug pricing work.
Finally, given the knowledge actuaries can develop in the pharmaceutical area, actuaries could be asked to help with other jobs. For example, actuaries might:
- Make clear how much money the insurance company’s PBM is making (for use in PBM contract negotiations);
- Explain drug trends to management in advance of questions from the board, customers, at-risk providers, and investors; and
- Cost legislative change options, as needed for lobbying efforts.
A couple of new areas have been emerging that we believe are ripe for actuarial engagement. The first emerging area relates to the AI technologies discussed earlier in this article. Many actuaries have trained on AI techniques, affording them the abilities to contribute directly to the use of AI in drug development. A second emerging area for actuaries, although some limited actuarial involvement occurs today, is in capturing and analyzing medical outcomes by patients of different types to determine cost benefit at the patient characteristic level. The result of this work could be not only greater efficiency of medical cost spend but also better treatments in light of what’s learned about who responds to which drugs.
KEITH PASSWATER, MAAA, FCA, FSA, is managing director of KTPassCo, providing leadership, strategy, finance, and actuarial consulting; he can be reached at KPasswater@ktpassco.com.
DAVE NELSON, MAAA, FSA, is a retired actuary, having served as vice president and chief actuarial officer at Blue Cross Blue Shield of Michigan.
The authors wish to thank Carol McCall, MAAA, FSA; Dr. Peter Canning; Dr. Michael Watkins; Jeff Klein, MAAA, ASA; Karen Sheridan, RPh; Jennifer Wang, PharmD, MS; and Cameron Bradshaw for their invaluable contributions to this article.
 www.fda.gov/downloads/BiologicsBloodVaccines/CellularGeneTherapyProducts/ApprovedProducts/UCM573941.pdf, accessed Jan 8, 2019
 C Roehrig, Projections of the Prescription Drug Share of National Health Expenditures Including Non-Retail, Altarum, May 2018
 F Pammolli, et. al, The Productivity Crisis in Pharmaceutical R&D, Nature Reviews Drug Discovery 10, 2011
 Available at http://www.actuary.org/content/prescription-drug-spending-us-health-care-system.
 US Food & Drug Administration, The Drug Development Process, www.fda.gov/ForPatients/Approvals/Drugs, accessed Jan 8, 2019
 Charles D. Hebert, Ph.D., D.A.B.T., Basic Overview of Preclinical Toxicology Animal Models, Dec 5, 2013
 J Bell, Teva confident migraine drug will launch this year despite new manufacturing woes, Biopharmadive.com, August 7, 2018
 US Code of Federal Regulations 21 CFR 314.50
 US Food & Drug Administration, The Drug Development Process, www.fda.gov/ForPatients/Approvals/Drugs/ucm405570.htm, accessed Jan 8, 2019
 S Findley, The FDA’s Sentinel Initiative, Health Affairs/Health Policy Brief, June 4, 2015
 J DiMasi, et. al, Innovation in the pharmaceutical industry: New estimates of R&D costs, Journal of Health Economics, May 2016
 U.S. Department of Health and Human Services, www.fda.gov/ForPatients/Approvals/Fast, accessed Jan 8, 2019
 B. Nelson, Why Big Pharma and biotech are betting big on AI, NBC News, March 1, 2018
 J. Walker, Machine Learning Drug Discovery Applications—Pfizer, Roche, GSK, and More, Emerj, December 12, 2018
 J Ruskin, a long time ago
 A Kesselheim, et. al, Biomarker-Defined Subsets of Common Diseases: Policy and Economic Implications of Orphan Drug Act Coverage, PLoS Medicine, Jan 3, 2017
 A Halvorson et. al (American Academy of Actuaries Prescription Drug Work Group), Prescription Drug Spending in the US Healthcare System: An Actuarial Perspective, American Academy of Actuaries Issue Brief, March 2018