Analysis of Texas Cervical Cancer Vaccine Decision

Texas Governor Rick Perry has ordered that all Texas 6th-grade girls receive the Gardasil vaccine to prevent HPV (and, by extension, reduce the chance of cervical cancer.)  This is controversial for a number of reasons.  But since this is Applied Genius and not Fulminating Bloviator, I thought I'd provide a few facts and a little bit of basic math: 

(1) According to Wikipedia (citing the WHO), "Worldwide, cervical cancer is the fifth most frequent cancer in women, when ordered by number of deaths. Cervical cancer affects about 16 per 100,000 women per year and causes death in about 9 per 100,000 per year. "  Last year, 1,169 new cases were reported and almost 400 Texas women died from it. (Dallas Morning News).  If 1 in 6,250 women get cervical cancer EVERY YEAR, it's obviously not mathematically correct that only 1 in 20,923 (the 2002 SEER estimate) will get it EVER.  SEER estimates for the Hispanic population in the USA range from 14.6-17.1 per 100K women, consistent with the 16 WHO estimate.   If 9 per 100K (about 1 in 11K) die per year and 400 died last year in Texas, that would imply that Texas has roughly 4.4 million women; the 2000 Census shows about 10.5 million women, which would seem to indicate that either (a) there are unreported cervical cancer deaths in Texas, (b) that Texas women are unusually resistant to it, or (c) that the definition of "women" that the WHO is using is more restrictive than the Census' version, probably limited by age.

(2) The vaccine costs $360 retail; hopefully the state will get a volume discount.

(3) The vaccine apparently prevents 70% of the cervical cancer cases.

(4) There are approximately 170,000 6th-grade girls in Texas; the vaccine for them would cost approximately $60 million (New York Times.)

Let's look at the math on a year-by-year basis.  Texas spends $60 million to save 280 lives (70% of 400) and save money for cancer treatment for 538 patients (70% of the 769 new cases who didn't die.)  I don't know how much cervical cancer treatment costs (and hope never to find out) but let's assume it's $50,000 on a present-value basis.   If the $50K figure is right, that's $40.9 million in medical savings alone.  That means that the lives were saved at a cost of only $68,214 each.  Given that wrongful death settlements are approximately $2 million each, the benefit/cost ratio seems to be about 30:1.   If the cancer treatment costs $75K rather than $50K, the lives saved are FREE.

How about this deal: rather than doing the vaccine now, we can simply wait until 280 girls (who are in sixth grade this year but will grow up to be women, of course) eventually die of cervical cancer.  Then the state of Texas can pay their families $68,000 each, for having failed to prevent the cancer when we could.  (Everyone who opposes the vaccine can chip in.)  Do you think the families would be happier to have lost their child but gained $68K because they avoided the vaccination?

This vaccine is a no-brainer.  Sure, Texas should get a big discount for buying a lot of it, and $60 million sounds like a lot, and it looks shady that Merck will get a lot of money after donating to the Governor's campaign -- but the savings are 30x the cost, even at retail prices.

P.S.  It's tempting to say "Since 280 women die each year, and 1,169 get cervical cancer, each day that we delay vaccinations costs us about 4 cases, including nearly 1 death."  Although I admire the attempt at quantification, this is fuzzy thinking at its worst.   Even if the deaths and incidence of cancer are indeed randomly distributed thoughout the year, earlier vaccination doesn't prevent the early deaths -- it only prevents the INITIAL INFECTION, and I haven't seen any info on when that happens.    In particular, I would guess that the number of 6th-grade girls in Texas who contract HPV between now and when school starts, who THEN go on to have cervical cancer, is either none or 1., with a high likelihood of "none."  After all, only 1 in 21,000 women gets cervical cancer over the course of her lifetime, which contains about 80x the time delay we're talking about (6 months). 

Comments

[tomjedrz]

I think your arithmetic is incorrect, as is your conclusion.

- 170K Texas 6th graders
- 1 in 21K women get the disease
- so ~8 women who are now 6th graders will get the disease
- the vaccine prevents 70% of the cases, the vaccinations will prevent ~6 cases.
- $60 million to prevent 6 cases -- $10/million per case prevented.

AND, this argument ignores the costs of any side effects.

This is a stupid decision by the Texas governor. The disease does not occur often enough for vaccination absent other risk factors. It is political payback plain and simple, and a waste of Texas taxpeyers money. The $60 million could be much better spent.

(Mycroft: I clarified the numbers, above. My original analysis is based on the reported incidences and deaths in TEXAS, not the per-year hazard rate -- which, in any event, seems to be higher than 1 in 21K by any calculation.)

[geoff]

This analysis isn't quite complete. At least two very important questions have not been answered. First, in order to assess the desirability of the program, we'd need to know what else the money could be spent on. Do you know that there aren't more effective (and perhaps cheaper) vaccines that could be given but aren't? What about programs entirely unrelated to vaccination--are any other welfare programs a better cost-benefit deal? Even more importantly, what would be the benefit of not taking this money for government use at all? Why does this form of redistribution (on average, from parents of healthy daughters to those of girls who would otherwise have gotten cancer) improve upon a world where the money to fund the program is left in the hands of the taxpayers to invest, to use for other forms of preventive medicine, or even to spend on vacations or food or iPods? I'd think one would need to know or estimate the answers to these questions before knowing whether to support the plan.

Second, is there no way to improve upon the proposed plan? Most obviously, is there no way to determine which girls are more likely at risk? Does the vaccine need to be given to every girl, indiscriminately, or could we improve the cost-benefit ratio by targeting girls with a family history of cervical cancer (if that matters), girls who are sexually active, etc? Relatedly, what are the costs of giving every girl the vaccine? Your analysis doesn't address problems arising from giving the vaccine (side effects, e.g.). There are surely some costs other than the price of the vaccine itself--a proper analysis would seek to maximize effectiveness while minimizing these costs (meaning the optimal number of girls to vaccinate is almost certainly not "all"). I believe that a proper analysis of the optimal program need not take the proposed regime at face value, nor should it.

There are other complications, as well, but these will suffice for now.