Barry Blackwell: Corporate Corruption in the Psychopharmaceutical Industry

Charles M. Beasley’s response 3 to Barry Blackwell’s response 2

 

       I thank Barry for his additional comments on April 18, 2019.  I have enjoyed this written dialogue, and as he said of me, Barry has also been calm and fair-minded.  I wish, and believe both of us wish, that those who govern in this country could also be calm, avoid vitriolic rhetoric and work collaboratively to address the matters that we have discussed.  These are the persons who need to find ways that allow for an adequate profit to encourage high-risk investment in our capitalist economy for the development of new treatments.  New treatments should be superior treatments for disorders where treatments of limited effectiveness exist, safer and more tolerable treatments where treatments exist but carry a substantial burden of adverse drug reactions, and, most importantly, novel treatments for disorders where no treatments exist.  My belief, with which Barry might disagree, is that this capitalist economic system does more to foster innovation, creativity and productivity than any alternative.  Perhaps it could do better with some fine-tuning in some areas.  Those who govern also need to find ways of ensuring that some minimal level of adequate health care is available to all while not discouraging individuals from accepting some personal responsibility for their health maintenance.

       Barry has focused on the cost of prescription medications as an unreasonably priced component of health care costs in the United States.  Out-of-pocket payments for prescription medications by patients are a small percentage of total payments for health care.  It should be kept in mind that retail drug costs make up a small fraction of U.S. health care costs. 

       An online report /Research-St(https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports /National Health Expend Data /downloads /highlights.pdf) by Centers for Medicare & Medicaid Services (CMS) titled “National Health Expenditures 2017 Highlights” provides the latest figures (2017, percentage of total cost and increase over 2016) for U.S. health care costs and who pays these costs as follows:

·         Costs

o   Hospital care:  33% - increased 4.6%

o   Physician and Clinical Services:  20% - increased 4.2%

o   Retail Prescription Drugs:  10% - increased 0.4% ($350 billion cost in 2017)

o   Other Health, Residential, and Personal Care Services:  5% - increased 5.6%

o   Nursing Care Facilities and Continuing Carew Retirement Communities:  5% - increased 2%

o   Dental Services:  4% - increased 3.2%

o   Home Health Care:  3% - increased 4.3%

o   Other Professional Services:  3% - increased 4.6%

o   Other Non-Durable Medical Products:  2% - increased 2.2%

o   Durable Medical Equipment: 2% - increased 6.8%

·         Payors

o   Private Health Insurance:  34%

o   Medicare:  20%

o   Medicaid:  17%

o   Out-of-Pocket:  10%

 

       The total cost of the healthcare services in the U.S. in 2017 was $3.5 trillion / $10,739 per person.  On average, then, out-of-pocket medical expenses were approximately $1,074 per person and prescription medication was only some proportion of this $1,074.  Of course, the actual out-of-pocket costs for some persons with serious illnesses is much, much higher, especially for those not covered by private health insurance, Medicare or Medicaid.  Healthy persons, such as children not suffering from congenital anomalies or disorders highly influenced by Mendelian or non-Mendelian complex genetic patterns raised in affluent and caring environments, have little in the way of health care costs other than for preventive services.  However, health care costs toward the end-of-life are not as universally great for all persons at the end-of-life as some might believe.  French, McCauley, Aragon, et al. (2017) reported the cost of medical care toward the end-of-life for nine developed countries based on spending in these countries between 2009-2011.  During the last year of life, spending ranged from a low of 8.5% to a high of 11.2% of total costs.  However, care during the last three years of life reached a high of 24.5% in one country and was generally higher than the range during the last year of life.  These percentages include long-term care.

       As Barry notes, the problem is complex.  Multiple factors likely drive the various domains of health care costs noted above.  For prescription medications, these factors include the pharmaceutical company’s pricing before discounts to large volume payors (e.g., insurance companies, retail pharmacy corporations).  Then, multiple additional factors impact the actual price paid by the multiple parties paying for the medication before it is placed in the individual patient’s hands.  Wholesalers/distributors and retail pharmacy corporations (few independent pharmacies continue to exist) sit between the pharmaceutical company and the retail pharmacy. 

       For patients for whom a substantial portion of the cost of their prescription medication is paid, not by them but by some insurance entity, the actual out-of-pocket cost is impacted by “clawbacks.”   Clawbacks are reviewed in a 2018 online report ("https://healthpolicy.usc.edu/wp-content/uploads/2018/03/2018.03_Overpaying20for20Prescription20Drugs_White20Paper_v.1-4.pdf">https://healthpolicy.usc.edu/wp-content/uploads/2018/03/2018.03_Overpaying20for20Prescription20Drugs_White20Paper_v.1-4.pdf) by Van Nuys, Joyce, Ribero and Goldman titled “Overpaying for Prescription Drugs: The Copay Clawback Phenomenon,” a report sponsored by the USC Schafer Center for Health Policy and Economics.  A “clawback” is an amount of money paid by the patient out-of-pocket that exceeds the actual cost of the medication paid by the insurance entity.  This “clawback” constitutes an additional source of revenue for the insurance entity paying for the medication.  This revenue might not be entirely profit as some portion might be funding another entity (e.g., a pharmacy benefits manager) working on behalf of the insurance entity to negotiate the price paid by that insurance entity.  Based on available data,  Van Nuys and colleagues estimated that in 2013 “clawbacks” were paid on 28% of generic drug prescriptions and 6% of branded drug prescriptions and that average amounts of these “clawbacks” were $7.32 for a generic prescription and $13.46 for a branded prescription.  These authors estimated the total cost of “clawbacks” in 2013 was $135 million.  This $135 million would be out-of-pocket expenses for those with insurance covering the bulk of the drug costs.  Therefore, this is an added out-of-pocket cost for only some proportion of the total expenditure for retail prescription drugs in 2013.  Hartman, Martin, Lassman et al. (2015), Centers for Medicare & Medicaid Services staff, estimated the cost of retail prescription drugs in the U.S. in 2013 at $271.1 billion.  I lack the data to compute the “clawback” amount as a proportion of the retail prescription drug costs that it affected.  The $135 million would represent only 0.04% of the total retail prescription drug costs.  Again, however, the $135 million impacted only a proportion of those $271.1 billion costs (proportion paid for by an insurance entity) and therefore the percentage of the retail prescription costs affected would likely be higher than 0.04%.  Finally, in absolute terms, $135 million is a substantial amount of money.

       As I have previously noted, I can generally pay less for my several generic medications if I pay completely out-of-pocket using one of several generic discount programs (e.g., GoodRx) than if I use my Medicare Part D coverage (managed by Express Scripts) and search for the pharmacy with the lowest overall price with the generic discount program.  This greater cost when using insurance strikes me as quite strange and something of which the average Medicare Part D covered patient would not be aware.

       Barry cites as an example of egregious price increases, the cost of an epinephrine pen.  He and I are in robust agreement about the matter of the increases in the prices of some generic drugs, and I have concerns about the initial prices of some generic drugs.  These are matters that I have previously addressed.  The epinephrine pen introduces an important matter that persons not familiar with pharmaceutical regulation or the pharmaceutical industry are likely not familiar.  An epinephrine pen is not a generic drug.  It is a combination of a generic drug and a medical device (the pen).  All medical devices, be they a delivery pen or a substantially more complex device, require extensive testing as well as regulatory review and approval before they can be marketed and used.  I do not know if the pens with astronomical price increases were newly designed and developed pens (although I doubt that they were because new pens were not discussed at the time of the price increase announcements). 

       Furthermore, I do not know, assuming that they were new pens (I doubt but do not know), that the new pen design was required by some deficiency in an older pen design or some perceived insufficiency in the documentation of testing on the part of regulators forcing new pen development.  If the pens were not new or if new but not mandated by regulators, then these price increases are another example of profiteering and not a reward for a risky investment (I would suspect to be the case but do not know).  Any drug delivered by a device is not simply a drug and requires two distinct development programs, one for the drug and one for the device.  Device development for simple delivery devices is a small portion of total development costs, but it is a cost. 

       I found one source ( href="https://www.pharmacircle.com/rc/files/docs/1442408626_Chpt_8_DD2011%202015-08-03.pdf">https://www.pharmacircle.com/rc/files/docs/1442408626_Chpt_8_DD2011%202015-08-03.pdf) that estimated the cost of several drug delivery systems at a mean of $85.5 million with a range of $24.9 to $167.2 million in 2009 dollars.  The authors based these estimates on regulatory documents filed with the United States Securities and Exchange Commission by Drug Delivery and Specialty Pharma companies.  While these development costs are a small fraction of the cost of the development of the drug being delivered, they are not insubstantial.

       Barry suggests two additional studies (or type of studies) be required for drug approval and maintenance of approval.  First, he suggests that all development programs should include a low-cost generic comparator, if available, for the disorder.  For this to be a requirement for U.S. regulatory approval by the Food and Drug Administration (FDA), the legislative remit of the FDA would need to be expanded.  Briefly, that remit is to require adequate demonstration of efficacy and sufficient assessment of safety to provide adequately informed use concerning safety.  The FDA’s mission statement includes: “protecting the public health by assuring the safety, effectiveness, quality, and security of human and veterinary drugs.”  As we have discussed, it is not scientifically possible to “assure safety” as all drugs, if effective, will have some off-target effects that are likely to be adverse.  Adequate testing and assessment before approval aim to characterize these off-target effects such that use of the drug has a positive benefit-risk balance across the population that might be treated with the drug.  For the FDA to be able to require what would be labeled a cost-effectiveness study, its legal duty and authority would need to be expanded by legislation to include requiring assessment (or demonstration) of cost-effectiveness. 

       For many therapeutic areas, such a legal requirement is unnecessary because such studies are already being conducted.  They are conducted for multiple reasons.  First, a positive control allows for the distinction between a so-called “failed trial” and a “negative trial.”  With a positive control and placebo included in a trial, if both test drug and the active control do not discriminate from placebo, the trial failed.  The general interpretation is that some aspect of the trial (e.g., placebo response) caused both non-placebo treatments to fail and the trial does not cast a cloud of doubt over the potential efficacy of the test drug. Without a positive control, if the test drug does not discriminate from placebo, the trial is negative and often interpreted (incorrectly) as demonstrating that the test drug does not have efficacy.  Even if not misinterpreted, the result does raise some degree of uncertainty about efficacy, even with subsequent successes.  Of course, if the positive control discriminates from placebo, but the test drug does not, that outcome would increase concern about the efficacy of the test drug but would not conclusively demonstrate a lack of efficacy.  An argument for lack of efficacy, although still not definitive proof, would be strengthened by the positive control being superior to both placebo and the test drug.  These adverse outcomes for the test drug are rare, and sponsors generally recognize that the potential benefits outweigh the risks of the use of active controls.  Active controls do increase study cost and time to completion of a development program.  Some sponsors (those with constrained resources) balk at their use because of both time and cost considerations.

       Such protection against misinterpretation of failure to discriminate from placebo is especially important in disorders with high variability in responsiveness to drug and placebo (e.g., psychiatric disorders).  For example, psychiatric drugs that were developed by Eli Lilly and Company while I was an employee all included active controls in one or more large Phase 3 studies.  With fluoxetine, the largest Phase 3 study included imipramine.  With duloxetine, the active control was fluoxetine.  With olanzapine, haloperidol was included in two large Phase 3 studies, one with placebo and the other with pseudo-placebo that was a 1 mg per day dose of olanzapine.  With atomoxetine for ADHD, the active comparator was methylphenidate.  When atomoxetine was being pursued as an antidepressant, with the generic name of tomoxetine, I designed two large Phase 3 trials with placebo and desipramine as an active control (both drugs failed to separate from placebo in both studies with high placebo response rates). I was responsible for the design of a large Phase 3 study of a novel, non-benzodiazepine interacting molecule for the potential treatment of GAD with lorazepam as the active control.

       In my post-Lilly consulting career, one client, a company larger than Lilly, employed my services to assist in planning and implementing a moderate sized Phase 2 study to attempt to minimize placebo response.  The study included an active control and the VP for clinical development for the therapeutic area told me explicitly that she/he was less interested in the efficacy of the test drug compared to placebo than in the performance of the active control.  The assay validity of the study was much more important than the performance of the test drug within the study.  The test drug had a novel mechanism of action for the disorder for which it was being evaluated.  It was important to have a clear answer as to whether this mechanism had any efficacy.  Without such a clear answer, internal factions favoring and opposing the further pursuit of the mechanism would likely waste time debating additional work with the drug and potentially waste material resources in possible additional studies for a drug with potentially no efficacy.  Most companies with adequate resources understand the importance of active controls in both Phase 2 and Phase 3 studies and use them.

       The use of active controls is also seen with disorders in other medical domains.  For example, many of the studies of what might be broadly termed second-generation anti-inflammatory biologics (e.g., IL-17 antagonists) have included first-generation anti-inflammatory biologics (e.g., TNF inhibitors). 

       Some therapeutic areas do not lend themselves well to use of active comparators as, although the drug under development is intended ultimately for solo use, studies in those disorders must be conducted adding the test drug to another ongoing treatment.  Adverse event (AE) interpretation and judgment as to whether any given AE is an adverse drug reaction (ADR), is challenging in an add-on study design and this difficulty grows more complex when attempting to understand the profiles of two different add-on drugs.  Some desirable active comparators might be inappropriate for add-on due to pharmacokinetic and/or pharmacodynamic factors and potential drug-drug interactions.

       A second reason that active-comparator studies are frequently seen is that a reason for developing the new test drug might be the intent to provide a more tolerable, and in some cases, a demonstrably more effective treatment.  If either of those goals can be demonstrated, the new drug can command a pricing premium over the active comparator.  Several of the second-generation biologic anti-inflammatory drugs have been able to demonstrate this superiority in the treatment of psoriasis, as one example.

       Finally, major third-party payors and their representatives (benefits managers and formulary boards) increasingly determine what medication a patient receives, not the prescribing physician.  Formulary boards and benefits plan managers are responsible for determining which medications will be covered by a given third-party payor.  These boards and managers, while extremely interested in the absolute relative costs of different medications for a given disorder, are also interested in evidence of relative ultimate cost-effectiveness.  Cost-effectiveness considers such things as the impact of the drugs on hospitalizations, cost of dealing with ADRs, the complexity of managing the use of the medication (e.g., frequency of physician visits required for most effective use), etc.  Positive cost-effectiveness data makes for wider access to more patients (greater sales) and the potential for premium pricing.  If the sponsor is interested in marketing the drug in countries other than the U.S.,  in a number of those countries such data have been required for placement on a national formulary.

       Barry has also suggested that a large effectiveness trial conducted within three years of initial approval should be a condition of continued approval to market a drug.  Such a trial would be what might be termed a study of large-simple design.  Ideally, the study would involve random assignment to the newly approved drug versus a single alternative drug or perhaps any alternative approved drug.  In some cases, a large-simple trial can be conducted without random assignment and statistical techniques might or might not be used to attempt to adjust for the potential differences between patients assigned to the new drug and alternative treatments.  Without random assignment, the study is a prospective, epidemiological, cohort study.  Occult differences that lead to differences between treatment group outcomes will not be considered.   Also, some differences that are known to investigators but thought irrelevant, while they are actually influential, will not be considered in statistical adjustments.  A clinician unblinded to choose between a very new treatment and a more established treatment is likely influenced by a wide range of substantial differences between the patients receiving new versus old treatment.  Even with random assignment occult or known differences between treatment groups might impact differences in the outcome but the random assignment to the independent variable alternatives is one of the foundations of well designed, comparative experiments.

       In the abstract, I agree with Barry about the desirability of the conduct of such a study.  I have worked with a regulatory venue where such a study is a requirement.  Lilly conducted such a study with olanzapine, the SOHO (Schizophrenia Outpatient Health Outcomes) study.  The SOHO study was a 36-month non-random assignment study enrolling more than 10,000 patients conducted in 10 countries.  Various 36-month outcomes comparisons have been published (e.g., Novick, Haro, Perrin et al. 2009a; Novick, Haro, Suarez et al. 2009b).  The ZODIAC (Strom, Eng, Faich et al. 2011) trial of ziprasidone would be another example of such a study with a psychiatric medication.  While the ZODIAC trial was focused primarily on the assessment of a potential AE/ADR with ziprasidone, SOHO was conducted to capture information on a wide variety of effectiveness, safety, quality of life and health economic outcomes.

       Such studies can be quite useful from several perspectives but how useful can they be in observing infrequent or rare AEs which is a different matter from determining that an observed AE is an ADR by either soft criteria or robust criteria as we have discussed elsewhere (Beasley and Tamura 2019)?  A 10,000-subject study with random or non-random assignment to only two treatments results in roughly 5,000 patients assigned to the new drug.  The statistical “Rule-of-3” approximates the appropriate interpretation of failure to observe a specific AE in this population of 5,000 patients.  This interpretation is that in a large population of comparable patients, there is a 95% probability that the actual incidence of the AE is ≤1/1,667 (1/(5000/3)).  By inference, a study with 5,000 subjects has a 95% probability of observing an AE that occurs with an incidence of 1/1,667 (and a higher probability if the incidence is greater).  For an important AE with an incidence in the total population of <1/1,667, the 5,000-patient study lacks the desired probability (i.e., 95%) of observing even a single occurrence of that AE.  Observation of a single case or very few cases of an AE only alerts us to the possibility of an ADR which can be very important information so long as the distinction between the possibility of an ADR and the “proof” of an ADR is kept clearly in mind for all infrequent/rare AEs.  I do favor the conduct of such studies as long as all who use the data produced have a clear understanding of what those data do and do not tell us as opposed to suggesting possibilities to us.

       These studies are expensive.  I do not believe they should be mandatory.  For some drugs, they would be practical impossibilities for a variety of reasons.  When conducted, they do substantially increase total research investment in drug development.  Such studies have a potential impact on drug pricing, especially for drugs used in smaller populations, and that might be potentially priced with lesser profit margins than other drugs in a company’s portfolio.

       In conclusion, perhaps Barry and I are in closer agreement on many matters than might initially appear to the readers of our interchange.  We likely do differ somewhat on where on some continuum between an extreme, government enforced socialist economy and an unconstrained capitalist economy (the latter not something that I favor) is the point that optimizes a population’s welfare and fosters maximum creativity and productivity to, hopefully, the ultimate benefit of the population as a whole.  My beliefs about the optimal point along this continuum is not a hypothesis supported by consistent, multiple empirical studies (impossible to conduct).  I frequently refer to such ideas of mine with the pejorative characterizations as “prejudices” or “biases” to remind myself of their lack of good empirical support and that these beliefs could be incorrect.  There are, again, many points of agreement.  Barry is a scholar and commands a great depth of expertise in many domains.  He has been highly cordial throughout our discussions, and for this, I thank him deeply.

 

References:

Beasley CM Jr, Tamura R.  What We Know and Do Not Know by Conventional Statistical Standards About Whether a Drug Does or Does Not Cause a Specific Side Effect (Adverse Drug Reaction) 3.  Potential sampling error in an RCT and what we learn from the lack of occurrence of an AE in an RCT (Rule-of-3) and impact on sample size calculations.  inh.org.ebooks.  January 10, 2019.

Blackwell B. Barry Blackwell’s response 2 to Charles M. Beasley’s response 2.  inhn.org.controversies.  April 18, 2019.

French EB, McCauley J, Aragon M, Bakx P, Chalkley M, Chen SH, Christensen BJ, Chuang H, Côté-Sergent A, De Nardi M, Fan E, Échevin D, Geoffard PY, Gastaldi-Ménager C, Gørtz M, Ibuka Y, Jones JB, Kallestrup-Lamb M, Karlsson M, Klein TJ, de Lagasnerie G, Michaud PC, O'Donnell O, Rice N, Skinner JS, van Doorslaer E, Ziebarth NR, Kelly E. End-of-life medical spending in last twelve months of life is lower than previously reported. Health Affairs 2017;36:1211-17.

Hartman M, Martin AB, Lassman D, Catlin A, National Health Expenditure Accounts Team.  National health spending in 2013: growth slows, remains in step with the overall economy.  Health Affairs 2015;34:150-60.

Novick D, Haro JM, Perrin E, Suarez D, Texeira JM. Tolerability of outpatient antipsychotic treatment: 36-month results from the European schizophrenia outpatient health outcomes (SOHO) study.  Eur Neuropsychopharmacol 2009a;19:542-50.

Novick D, Haro JM, Suarez D, Vieta E, Naber D. Recovery in the outpatient setting: 36-month results from the schizophrenia outpatient health outcomes (SOHO) study.  Schizophr Res. 2009b;108:223-30.

Strom BL, Eng SM, Faich G, Reynolds RF, D'Agostino RB, Ruskin J, Kane JM. Comparative mortality associated with ziprasidone and olanzapine in real-world use among 18,154 patients with schizophrenia: the ziprasidone observational study of cardiac outcomes (ZODIAC).  Am J Psychiatry 2011;168:193-201.

 

October 31, 2019