[extropy-chat] Are vaccinations useless?

[Hal Finney]

I was able to get a copy of one of the references Robin cited,
<http://www.milbank.org/720203.html>: "Improving Health: Measuring
Effects of Medical Care", by Bunker, Frazier and Mosteller; Milbank
Quarterly, 72(2), 1994.  Unfortunately I could not find the text online;
I had to xerox it at the library, so I can't easily make it available.
But I will type in here their discussion of immunization.  Their goal
in this section is to evaluate the contribution of medicine to life
span since 1900.  During this time average life span in industrialized
countries went from 45 to 75 years, a gain of 30 years.

The authors divide their analysis into preventive and curative services.
Immunizations fall into the preventive category.  Their analysis
attributes approximately 18 months of life expectation increase to
immunization.  Here is the text of the analysis (typos are mine!):

> Infectious disease, the most common cause of death in childhood at
> the beginning of the century, has become a rare cause of death today.
> In 1900 the annual death rates of diphtheria, measles, and pertussis
> were 40, 13 and 12 per 100,000 respectively, whereas by 1960 there were
> no deaths reported for diphtheria, and only 2 and 1 per 100,000 for
> measles and pertussis.  Most of the fall in death rates from measles and
> pertussis occurred before the introduction of their respective vaccines
> or the availability of antibiotics (McKeown 1979), and we can only credit
> the introduction of diphtheria antitoxin and subsequent immunization
> for the observed fall in mortality, equivalent to an increase in life
> expectancy of approximately 10 months.
>
> Death rates from poliomyelitis and tetanus before the introduction of
> immmunization against each were less than for measles, diphtheria, and
> pertussis; the rate for poliomyelitis varied between 0.4 and 1.8 per
> 100,000 in the 1920s, 1930s, and 1940s; and that for tetanus, between
> 0.7 and 1.7 in the 1920s and 1930s.  Their virtual elimination following
> the achievement of nearly universal immmunization represents an increase
> in life expectancy of about three weeks for both combined.
>
> What conclusions can we draw regarding the contribution of immunization
> practices to health today?  Measles offers a case in point.  Although
> death attributed to measles is rare in the United States, the recently
> reported marked upsurge in cases of measles in unvaccinated, preschool
> children in the inner city raises the possibility of the return of
> measles-related death, deafness, and mental retardation (Hersh et
> al. 1992), as we discuss elsewhere (Bunker, Frazier, and Mosteller 1994).
> A waning appreciation for the importance of, and declining confidence in,
> immunization gives added cause for concern.  Koplan et al. (1979) have
> explored the potential effect that curtailment of pertussis vaccination
> might produce.  Using decision analysis, he and his colleagues predicted
> that there would be a 71-fold increase in numbers of cases and an almost
> 4-fold increase in deaths were pertussis immunization to be discontinued.

This last paragraph kind of undercuts the overall conclusion that
immunization has not made a very large contribution towards the 30
years of lifespan extension.  They do have nice things to say about
the practice, although it is not clear that their comments are fully
grounded in the data.  This may be the kind of attitude that caused
Robin to consider them somewhat overly generous in their assessment.

I can also show an excerpt from their Taqble 1, which estimates the gain
in life expectancy from various interventions, including immunization.
The columns are: "Clinical preventive service (immunization)";
"Individuals affected by condition in the absence of preventive service";
and "Gain per individual receiving preventive service".  All of the
2nd-column incidence rates are per year:


Diphtheria		40 deaths		10 months
			per 100,000

Poliomyelitis		2,500 deaths
						3 weeks (polio+tetanus)
Tetanus			2,500 deaths

Smallpox		NA			3-6 months

Influenza		10,000-40,000		3 weeks
			deaths

Pneumococcus		400,000 cases		6 weeks

Hepatitis-B		21,000 cases		1.5 - 2 weeks


I added these myself to get the figure I reported above of approximately
18 months out of 30 years of improvement.

They have a couple of footnotes on the smallpox line.  The incidence
is NA, footnoted as "not applicable following worldwide eradication".
And the life extension level is footnoted "Limited to this century only".
Since the smallpox vaccine was introduced in the 19th century it lies
outside the "window" of life extension that the authors are analyzing.
I'm not sure what exactly the 3-6 months means; maybe this is due to
either improvements in the smallpox vaccine or an increase in its use.

It's worth noting that diphtheria immunization provided the biggest
improvement, an average of 10 months of life per person.  This is actually
the single largest value of all of the preventive or curative measures
they analyze.  The next most effective preventive measure is screening for
hypertension, which they estimate adds 1.5-2 months per person.  And on
the curative side, the most effective are diabetes treatment, worth
6 months per person; appendicitis, 4 months; hypertension treatment,
3.5-4 months; and infant respiratory failure (in premature babies),
3-4 months.  Note that these values are not the benefit to the people
being treated, but are in effect discounted by the incidence rates within
the population.  Modern medicine is of great value if you get diabetes,
with an estimated gain of 25 years of life; but only one person in 50
has diabetes so medicine only gets credit for 1/2 year of benefit.

The bottom line is that they estimate that preventive and curative
medicine together account for five years of life expectancy gain, out
of 30 which have been observed.

I haven't read the whole paper yet; the next section is about improvements
in quality of life rather than lifespan.  But I think this gives the
flavor of their analysis.  Probably the most important factor is that
the incidence of measles and pertussis fell a great deal even before
their vaccines were introduced, or antibiotics.  So medical treatment
gets zero credit for the increase in lifespan due to these therapies.

In a way, you can see how they might be too generous in some measures.
Diphtheria's vaccine gets most of the credit; but if that vaccine had
not been available until the 1960s, chances are the rate of diphtheria
would have fallen just like measles.  In that case we would not credit
diphteria vaccination with adding 10 months of life.  So to some extent
it looks like a matter of timing and luck as to whether a treatment
came along early enough to get credit or was so late that other factors
appeared to be responsible.

And the puzzle remains; how could medicine only be good for 5 years out
of 30?  Or maybe it's not a puzzle.  Obviously there are a host of reasons
why people live longer today: improved nutrition, sanitation, safety, and
so on.  A priori there is no particular reason why medicine should get the
lion's share of the credit, over all these other factors.  Maybe there are
six things that are each good for 5 years' worth of benefit.  In that case
the limited value of medicine should not come as so much of a surprise.

Hal