I’m going to begin on a completely unrelated note by mentioning that a book I requested at the library recently became ready for pickup. Listed as being there when I ordered it, it has since been listed as “In Transit” for about two months. Now I feel slightly guilty that I’m not going to pick it up, because some time after I originally ordered it I picked up a rather lengthy book from a university library from when I temporarily had privileges, and hence won’t be able to review. I’ve been holding off on that in part because I’ve procrastinated so much when it comes to writing this review.

I forget where exactly I heard of Brad Spellberg’s “Rising Plague”. Megan McArdle seems most likely, since she occasionally references increasing antibiotic resistance. It seems like an important problem about which little is done, so good thing there’s a book about it. Unfortunately I can’t give a hearty recommendation while at the same time that’s not because of some objective flaws. I didn’t care much for his writing, which makes some sense since he’s a doctor rather pop-science writer (although both Oliver Sachs and Michael Crichton have a background in practicing medicine). Sometimes I was irked at how much it was aimed at a broad audience with its emotional anecdotes illustrating the need for action (he’s quite explicit about soliciting and collecting them for that purpose) rather than mere statistics, but clearly pleasing me was not his primary goal.

I’ll begin with some statistics that illustrate how large a problem it is. There are a lot of microbes out there, they make up 60% of the earth’s biomass (90% excluding plant cellulose). Infections are the third leading cause of death in the United States, and since many cancer deaths are proximately caused by the infection of weakened immune systems, adding them could bump it up to number two (even without doing so, it is number two worldwide). In 1996 (he might have chosen that year because it was particularly high) 170000 Americans died of infections. The top three infections (sepsis, influenza and pneumonia) are estimated to kill 250000 to 300000 Americans per year. Those death rates alone don’t suffice to indicate the importance of antibiotics, because they have become common supplements to medical practices that either expose the body to infections (whether through full-blown surgery or ventilators or something milder like catheters or I.Vs) or weaken the immune system (as with chemotherapy, one of the first modern medical treatments to directly attack infectious agents). The end of antibiotics would make those standard procedures liable to cause death. A point that Spellberg emphasizes is that this is not a matter of some substandard hospitals following bad practices or sloppily overprescribing antibiotics, the standard procedure deemed optimal at hospitals will inevitably result in selection for antibiotic resistance. And even attempting to isolate patients with such strains away from everyone else in the hospital frequently fails to prevent spread.

Antibiotics have been around long enough that most folks today have little memory of how things were before. Back then relatively mild injuries that could be easily be treated today led to death, and doctors just had to accept there was nothing they could do. There were treatments going back to the ancient Greeks wherein an injury would be treated with a caustic substance, which Spellberg speculates caused inflammation and an immune system response in that local area, but it would not directly affect any infection and I don’t understand why the injury itself wouldn’t suffice for that sort of reaction. The Chinese and Egyptians both made use of mold, which of course was discovered to produce penicillin. As late as 1937 most cases of pulmonary tuberculosis in the U.S were treated with “artificial pneumothorax”, which is medical-latin for jabbing a large needle into one lung in order to collapse it, starving the bacteria of the oxygen it needs while the patient limps by on the other lung. There had already been a chemical treatment for syphilis published by 1910, but the chemical was arsenic, which is of course poisonous to people as well as bacteria (chemotherapy thus precedes the more patient-friendly treatment of antibiotics).

The story of how in 1928 Alexander Fleming accidentally allowed the mold to grow and repel bacteria is well known, but what I didn’t know was that nothing came of the discovery because Fleming failed to isolate a pure enough version of penicillin to be effective and gave up. Instead the first antibiotic to be used came was discovered in 1931 when a group of scientists at I. G. Farben found that injections of a certain red dye were able to cure streptococcus in mice. It failed to kill bacteria in test tubes because the dye itself wasn’t an antibiotic. What they found was that the liver metabolized the sulfurous dye into the first antibiotic, sulfanilamide. Spellberg calls it “one of the great ironies” that a few years later the director of the research laboratory saved his daughters life with that drug, but it’s not ironic at all that someone so closely connected to a potentially lifesaving new drug would be treated with it. Pure penicillin was still in short supply in 1940, so that a British police officer treated it with it had to have his urine recycled for retreatment when the hospital’s meager experimental supply from their lab ran out (the infection returned and he died).

What I didn’t know before reading this book is that “Virtually all of the antibiotics we now use are either harvested directly from microbes or are made synthetically based on the design of naturally occurring antibiotics” resulting from the competition between various microbes with each other, and that these microbes simultaneously evolved resistance to their own antibiotics in order to avoid committing suicide. Combine that with the ability of bacteria to swap genes even across extremely separate family trees, and it means that all manufactured antibiotics are running down a clock that started before they came into existence. And of course the more an antibiotic is used, the more selection there is for resistance. Spellberg notes the same resistance problem with bacteria also exists for fungi and parasites (viruses aren’t alive enough to have analogous drugs), but doesn’t explain why he wrote a book just about the first. Judging by “Parasite Rex” and this underwhelming warning about the rising fungal threat, I’d guess it’s because people in the first world just aren’t affected by the latter much. 3% of a bacteria’s offspring are reported to have mutations, which sounds lower than I thought the rate to be for humans (maybe our larger genomes means errors are more likely), but the shorter reproductive cycle compounds that.

The problem wouldn’t be quite so bad if antibiotic development always outpaced resistance. That doesn’t seem to be the case. Spellberg and some colleagues wrote a report for the Antimicrobial Availability Task Force. This is my attempt to convert their bar-graph into a table:

Year Range Number of New Antibiotics
1983-1987 16
1988-1992 14
1993-1997 10
1998-2002 7
2003-2007 5

It takes on average over ten years from the discovery of an antibiotic to approval for use on patients, and in 2004 product pipelines in publicly available listings of the fifteen largest pharmaceutical companies contained five new antibacterial drugs in development. Contrast that to eight for bladder hyperactivity and seven each for acid reflux and irritable bowel syndrome. At least it’s one more than the four for erectile dysfunction. They found one more antibacterial in development among biotechnology companies. Because they’re popular to mock, it might seem that “lifestyle” drugs are sucking up all the oxygen, but they actually make up only a small portion (which is why they are comparable in number with antibiotics). Most new drugs are for conditions like cancer, blood pressure, cholesterol, diabetes, arthritis, inflammation, heart attacks, stroke or dementia. Judging from that list, treating chronic conditions seems to be more profitable than curing diseases, and Spellberg actually seems to embrace Chris Rock’s logic on polio.

Rock cites AIDS there, and ironically enough Spellberg notes that there has been much more success treating that than bacteria. This is in accordance with Mike Darwin’s account of the early “AIDS underground”, which almost seems to have been written to illustrate Sailer’s Mancur Olson-esque point about group mobilization in illness. Spellberg notes that over the last fifteen years virtually the same number of numbers have been brought to market for HIV as for all bacterial infections combined. In the last five years, there have actually been more of the former than the latter. In 2004, there were actually twice as many HIV drugs in development as antibiotics. This despite the fact that MRSA*, by itself, kills more Americans every year than HIV. Approximately 10% of the roughly $500 million that NIAD**/NIH spends on drug-resistant infections goes to bacterial infections, with most of the rest going toward HIV. As a result, Spellberg says “in many ways it is now easier to treat HIV than diabetes”.
*That’s gotten enough press that you’ve probably heard of it. But Spellberg claims that both it and extreme drug resistant tuberculosis are less of a threat than multi-drug resistant nosocomial gram negative rods. I’d never heard of any of the spectacularly resistant non-fermenting species, but E. coli is a gram negative rod we should be familiar with.
**National Institute of Allergy and Infections Diseases. Why are those together?

Those numbers are still a small fraction of the total amount of money spent on drug R&D. The members of the Pharmaceutical Research and Manufacturer’s Association was budgeted $44.5 billion for research and development in 2007. Comparing budgets in previous years, that is up 70% from 2000, 430% from 1990 and 2125% from 1980, meaning there was an average annual increase of 79% from 1980 to 2007. Increasing expenditures on R&D is one of the datapoints cited to show diminishing marginal returns in Tainter’s “Collapse of Complex Societies” and fits with Tyler Cowen’s story of consuming the low-hanging fruit. Greg Cochran, in contrast, thinks ignorance is pervasive enough to leave plenty low-hanging fruit unpicked (though the book claims drug companies do actually harvest microbes from soil and mass-screen to see if they kill other microbes). I would have to count Spellberg as relatively optimistic, since he thinks the problem is that companies can’t make as much money on antibiotics as other drugs, and (though the proposals are more complicated and will be discussed later) can be ameliorated by throwing more money at it. As mentioned, development has shifted from antibiotics to treating more chronic conditions, and there is a rather convincing graph in the book showing that fluctuations in the gross margins of pharmaceutical companies are closely tracked by R&D outlays. The information he presents about how pervasive failure is makes me more pessimistic. In 1995 GlaxoSmithKline started with over 300 targets from the published genome of Hemophilus influenzae, initiated 67 distinct campaigns (each costing approximately $1 million), of which 16 identified potential gene targets in initial laboratory screens, 5 turned out to be real, and by 2001 they gave up developing drugs for any of them. That’s just one example of an unusually targeted approach, the success rate for the more common route of starting with molecules is reported to be 1 in 10000. Even among drugs that are found safe and effective in animals as well as being drug-able (effective in concentrations feasible to dose people with), 90% fail in the three phases of human clinical trials (the last of which costs an average of $105 million).

The argument must be that there is some particular problem with antibiotics. If the author spoke economese, he would say “market failure” or “public good”. As mentioned, he takes the Chris Rock line that the very effectiveness of antibiotics makes them a less profitable investment. That didn’t really make sense for me. The more effective and useful a drug is, the more should be willing to pay for it. And there are plenty of goods like cars (though in that clip Rock uses them as a counter-example), houses or higher-education that have high costs and infrequent purchases on a per-capita basis. Even within medicine there are plenty of surgeries only expected to be performed once on a person. It’s a common truism that the first dose of a new drug costs extravagant amounts, but every subsequent one costs peanuts to produce, hence generics can easily undercut the original in the absence of patent protection*. But the same problem should apply to all drugs, not just antibiotics. Spellberg claims there is a unique problem in that the use of antibiotics (which of course would increase profitability) has to be restricted in order to prevent resistance, but the same is true for H.I.V. I have a conjecture that precisely because drugs for chronic conditions are not expected to “cure” them, there is a lower bar to be deemed effective. Prescription drugs are ascribed 10% of all healthcare costs, which perhaps shouldn’t be surprising since they can benefit from the technological progress of the manufacturing field, while Baumol’s cost disease (together with licensing) continues to inflate the service side**. Medicare not covering prescription drugs until 2006 may have helped keep the price down, since the moral hazard of third-party payment is well illustrated by $5 a day (over a two-week course) amphotericin B deoxycholate being driven off the market by a newer $300 a day derivative which was no more effective but much less toxic.
*David Friedman argued that patents should internalize the externality of antibiotic resistance, but I’m not sure how they would work in his preferred system of anarcho-capitalism.
**Despite Baumol, pharamaceuticals still seem to be slightly less cost-effective than surgery per one of Spellberg’s tables.

Intervention Type Median cost (1998 dollars) per
quality adjusted life-year saved
Vaccinations 2000
Delivery of Care in Appropriate Setting
(e.g., ICU bed versus regular ward bed)
Surgery 10000
Pharmaceutical Treatment 11000
Screening Tests 12000
Other Public Health
(e.g., airbags, folic acid supplements)
Health Education/Counseling 20000
Diagnostic Studies 20000
Medical Devices
(e.g., pacemakers)
Other Medical Procedures
(e.g., transfusions, catheterization)

Throughout the book Spellberg tends to be sympathetic to the pharmaceutical industry, which he thinks is unjustly derided in the popular discourse. He does note that in 2004 Fortune ranked industries by their profit margins, and at 14.3% pharmaceuticals had the third highest (purportedly to compensate for the risk of failing to develop any drug, and to make up for costs before the patent expires and generics step in). An industry probably more derided than pharmaceuticals is the defense industry, at 30th place with 3.2% margins. That’s relevant because it illustrates the difficulty of the government taking a larger role in drug development. During the Bush years there was a move toward drug development for defense purposes, and while one might expect (as with the popular conception of the defense industry) the result to be loads of sweetheart-contracts, the 2004 “Bioshield” bill resulted in almost no takers despite the $6 billion in potential federal contracts on offer. There was one biotech firm which received $429 million (recall that’s ten times as much as the federal government spends on antibacterial research) for a second-generation smallpox vaccine, despite the fact that smallpox doesn’t exist in the wild because the first vaccine eliminated it. Other actions taken by the administration made them highly unreliable partners for the pharmaceutical industry. Ciprofloxacin (together with a close relative) is the only drug capable of treating many drug-resistant gram-negative bacteria. The hysteria over the anthrax attack (which the administration inflamed to help justify a response) resulted in prescriptions increasing by 160000 from 2000 to 2001 (and recall the attacks took place near the end of the year). The administration threatened to revoke Bayer’s patent on the drug (which wound up not being necessary, since there weren’t follow-up attacks) if they didn’t supply an enormous quantity at half the usual cost. Hollis-Eden spent $100 million under Bioshield for a drug to treat radiation poisoning in case of a dirty bomb (which again was fortunately not required), but the former Amtrak lawyer appointed to the Bioshield program then changed the order from ten million to one hundred thousand doses, sending the stock price plummeting. None were ever delivered because the order was ultimately cancelled.

I’m willing to believe the Bush administration was particularly incompetent, but that doesn’t necessarily generalize. The American healthcare system is often contrasted unfavorably with Europe, so it’s a natural point of comparison. Part of the reason for that is price controls which result in Americans paying more for the same drugs. Fifteen years ago, before such controls became the norm, European pharmaceutical R&D spending far outstripped that of the U.S. Now U.S expenditures are estimated to be 60% higher. U.S biotech companies account for over 75% of global biotech revenue and R&D. The density of biotech patents per capita is 80% higher in the U.S. There are twice as many new drugs in development in the U.S as Europe. Approximately 400000 biomedical scientists educated in Europe now live and work in the US, just 13% of those surveyed plan to eventually return. I suppose that’s all very nice for the world, but analogous to the realists advocating the U.S leave more military spending to its allies, I don’t think we should encourage so much free-riding.

As an academic physician, Spellberg notes that his promotions and salary are mostly derived from securing NIH grants. He also notes that they reject about 90% of grant applications (although if it’s cheap to generate an application, we should expect them to dwarf the number of available grants). He also discloses other possible conflicts of interest resulting from grants or consulting work with pharmaceutical companies, but on average over 80% of his pay comes from the NIH (this may change if his biotech startup to develop a vaccine he researched is successful). Based on the principle that “Where you stand depends on how you sit” you might think that he’d see the NIH as extremely important and advocate expanding its capacity, but when he mentions their budget failing to keep pace with inflation (for the first time since the 80s) or that combined corporate pharmaceutical R&D expenditure was more than twice the NIH budget, he treats this as set in stone. When arguing that such companies are the only option we have to go beyond basic science and actually deliver drugs, he points out that the anti-malarial drug artemisinin is the only anti-infective in the world exclusively developed by direct government sponsorship (more specifically, the Chinese army, and herbalists had been using the source plant millennia before the PLA developed new techniques to extract it). He doesn’t seem to consider the “Lucal critique” that if governments decide to leave drug development to private companies, of course they won’t develop many themselves, without providing any evidence that a change in policy would be futile. On the other hand, it is suggestive if that policy stance is so universal that there are no other exceptions to his rule.

Spellberg’s proposals for dealing with the problem set out basically amount to sweetening financial incentives. He divides those into “push incentives” that can be small amounts for the initial phases of development (which may be attractive to small companies) and “pull incentives” that are much larger and come in at a later end in such forms as “extended patents, periods of market exclusivity, and possibly major tax credits”. He gives surprisingly short shrift to direct subsidies as a “pull strategy”, including such methods as prizes (advocated by Joe Stiglitz, among others). The “true subsidy” nature of a prize seems a political non-starter to Spellberg, but from an economic perspective his recommendations are as well. The only circumstance I can think of in which a patent (or patent extension) is less costly than a prize is a case in which the drug is never sold (or for some reason sold at generic price), or in other words is a failure. Another case where his perspective seems very different from that of an economist is the Orphan Drug Act, which he regards as a great success. That a condition which afflicts fewer people has fewer resources devoted to its amelioration than those that afflict more is just how things ought to be. Resources are limited after all. A more interesting program is the Vaccine Injury Compensation Program, a trust fund that receives $0.75 in excise tax from each dose of vaccine sold and is paid out to patients who experience the (very rare) side effects from vaccines (or at least claim to). Pharmaceutical companies could similarly pool funds in case one is sued for toxicities not found in clinical trials of an antibiotic. As with many of his other incentives, Spellberg would like to limit these to “priority pathogens”, using similar logic that led to the VICP that the public health mission was too important to be blocked by fear of lawsuits. Limiting coverage by the frequency of a side effect would have the beneficial result of incentivising companies to closely track the frequency of a side effect both before and after FDA approval in order to prove that it was low enough to waive their liability (whereas today they may not care to track any evidence post-approval until enough people have been affected to mount a lawsuit). Finally, he raises the “lightning rod” possibility of transferable patent extensions, where receiving approval for a “priority pathogen” could result in a patent extension for an entirely different (and presumably more profitable) drug. In the case of normal patents, the benefit is limited by the usefulness of the product (hence why the complete failure hypothesized earlier could have its patent extended costlessly), this proposal could permit monopoly profits entirely out of proportion to the usefulness of the “priority” antibiotic (putting a lot of weight on the judgment of the federal commission that decides what’s a priority). In order to make it more acceptable, or maybe just because it should receive more funding, Spellberg proposes that 10-20 percent of the extended patent drugs’ profit be funneled through the NIAID to fund grants for studying antibiotic resistance. He discusses a cost-benefit analysis he did showing that (using some conservative assumptions to understate his case) the benefits could match the extra costs of these “wild card” extended patents over a period of ten years and exceed over twenty years, but strangely he doesn’t seem to factor in increased resistance for the marginal antibiotic over those years. Yes, in a book devoted to the topic of antibiotic resistance and discussing new drugs for resistant bacteria, he doesn’t mention anything about resistance toward the new drug (although it’s possible he factored that in and failed to mention it)!

Weirdly, he devotes just half a page to vaccines as a complementary method of antibiotics, even though he acknowledges (and see the chart above) that they’re pretty much the most cost-effective way to reduce the global burden of infectious disease. His argument for why antibiotics are important is that no strategy is 100% effective. Well whoopity doo, but that doesn’t explain why you wrote a whole book about antibiotics but devoted such little space to vaccines! I presume there must be some scientific reason (aside from the fact that vaccines are a preventative measure, and we often need a “pound of cure” after the fact) why they can’t shoulder too much of the burden, hence why the discovery of antibiotics was such a huge deal.

I’d write a conclusion, but I think I’ve said enough about my attitude toward the book. I’ve also procrastinated to the extent that I’m nearly out of extensions and would like to move on to the more interesting next book before that runs out as well.