The following are summaries and excerpts of what I take to be the
most important stuff concerning the benefits of I&M programs.  I
have not attempted to pull out THE critical information, which I
think we will be able to summarize in a paragraph.  Perhaps
someone working on the project can take a stab at that and then
send it to everyone for criticism (including me). 

It seems that the main effects of pollution are those of ozone
(O3) and particulates on temporary symptoms, including hospital
admissions, and the effect of particulates on death rates.

We need to find the effects of I&M on ozone and on particulates. 
Beaton et al., below, provide an estimate.

Probably the best single paper, and a must-read if you are
working on this, is the following:

Small, K. A., & Kazimi, C. (1995).  On the costs of air pollution
from motor vehicles.  Journal of Transport Economics and Policy,
29, 7-32.

"With respect to helth effect, there are several reasons to
accept linearity between concentrations and aggregate health
costs as a good approximation for aggregate analysis."

For death rates as a function of PM10 (particulate matter less
than 10 microns in diameter), "The relevant coefficients are
measured as the increased annual metropolitan-area deaths per
100,000 population for a unit increase in particulate
concentration in micrograms per cubic meter (m3)."  The article
estimates that the coefficient for PM10 is about .52 (p. 18). 
This is derived from a Schwartz study which says that the
relative risk is .06% per unit increase (micgrograms/m3), and
extrapolated from a death rate in LA of 870 per 100,000 pea year.

For effects on temporary symptoms, this articles notes that the
following article by Hall et al. was criticized for being on the
high side.  It favors the estimates of Krupnick and Portney
(who, however, pretty much ignore mortality effects).  The
present article concludes that the mortality effects are the main
costs.

Hall, J. V., Winer, A. M., Kleinman, M. T., Lurmann, W. W.,
Brajer, V., & Colome, S. D. (1992).  Valuing the health benefits
of clean air.  Science, 255, 812-817.

This paper looked at overall health effects of ozone (O3) and
particulate matter (PM10), the main culprits, in Southern
California, where pollution is high.  (How high relative to
Philly?)  It considered the effects on the 12 million inhabitants
of the South Coast Air Basin.  It assumed that O3 affects mostly
temporary symptoms and PM10 affects mortality.  It looked at the
value of getting from current levels to national attainment
standards (NAAQS)

It used previously estimated monetary values for cough,
headache,chest discomfort, sort throat, and eye irritation.
Values were inferred from expenditures and contingent-valuation
stueies.

"Several studies have reported an association between O3 levels
and minor restricted activity days (MRAD), days on which activity
is reduced, but not severely restricted.  The relaionship we used
is Mj = .077 Oij Pj, where Mj is the change in the number of
MRADs for population j on day i, Oij is the change in the daily
high O3 level for a one-hour period in parts per million on day i
for population j, and Pj is the population in demographic group
j."

"A restriced activity day (RAD) includes days missed from work,
spnt in bed, or otherwise  measurably constrained.  ... the
relation simplifies to PM10-related RADs per person equals .0556
times the annual average PM10 concentration."

"The PM10 estimates indicated that approximately 10 million daily
exposures to concentrations in excess of the 24-hour NAAQS of 150
micrograms per cubic meter occurred per year, slightly less than
one per capita."

"Basin-wide O3-related occurrences of cogh were estimated to
decrease by about 120 million days annually after attainment, or
11 fewer days per capita.  Analogous results for the other
symptoms are 190 million fewer days of eye irritation (16 per
capita); 180 million days of sort throat (17 per capita); 120
million fewer days of headache; and 65 million fewer days of
chest discomfort (5 per capita).  Also O3-related MRADs decrease
by nearly 18 million days annually, or 1.5 per capita.

"Although effects related to PM10 exposure occure less frequently,
they are more serious.  Attainment of the federal PM10 standard is
estimated to result in almost 15 million fewer RADs each year, or
1.2 per capita.  Premature death is the most serious, and least
frequent, of all effects related to any air pollutant. ... We
estimate that attaining the standard will prevent approximately
1600 annual deaths attributable to PM10 levels.  ... Thus, the
average annual risk of death is 1 in 10,000 greater than it would
be if the ... standard were attainted.  To place this result in
perspective, in California in 1987, the risk of death in a motor
vehicle accident was 2 in 10,000; nationwide, the risk of
job-related death was .5 in 10,000."

"We assumed that a day spent suffering from a severe symptom
would be highly correlated aith at least some restriction in
activity and that the value of avoiding a severe symptom would
therefore closely approximate the value of reducing an MRAD.
Human clinical studies suggest that more than one thpe of symptom
- upper respiratory difficulty and eye irritation, for example -
may occur during a single day.  We account for this possibility
by terming these multiple minor symptoms days (MMSDs) to avoid
overvaluing their joint occurrence."  The article becomes a
little unclear about how precisely this was done.

The following is from the first major reported cost-benefit
analysis.  Unfortunately, it is limited in several ways.

Krupnick, A. J., & Portney, P. R. (1991).  Controlling urban air
pollution: A benefit-cost assessment.  Science, 252, 522-528.

"We estimated the reduced incidence of the quantifiable adverse
health effects in the year 2004 accompanying a 35% reduction in
emissions of VOCs for an estimated 129 million people living in
the 94 metropolitan areas (322 counties) predicted to be in
nonattainment in 2004 19, 20 . For each metropolitan area, we
made separate calculations on the basis of the predicted change
in air quality there and then aggregated these estimates to
obtain the national estimates.
    From the epidemiologic studies, we found that the average
asthmatic will experience about 0.2 fewer days per year on which
he or she has an asthma attack and that the average nonasthmatic
will experience about 0.1 fewer minor restricted activity days
per year because of reduced VOC emissins and subsequently
improved air quality 21 . In addition, nonasthmatics will
experience other minor health benefits as well in the form of
reduced number of symptom days.
    To convert these predicted changes in physical health into
economic benefits, it is necessary to ascertain individuals'
willingness to pay for a reduced incidence of illness and adverse
symptoms. To do so, we drew on a number of studies designed to
uncover these values, primarily through questioning of both
healthy and infirm respondents with supplemental data on the
out-of-pocket  medical costs and lost income that may be
associated with illness or symptomatic effects 22 . These studies
have found an average value of $ 25 for each asthma attack
prevented, $ 20 for a reduction of one restricted activity day
(on which  an individual is neither bedridden nor forced to miss
work but must alter his or her usual pattern in some way), and $
5 for one fewer day of occasional coughing. When reduced
incidence is combined with these values, the predicted aggregate
dollar benefits across the United States from these improvements
in individuals' acute health status amount ot $ 250 million per
year.
    By using clinical rather than epidemiologic studies to
estimate health benefits, we arrive at a somewhat larger value
for acute health benefits. For example, we predict that the
number of coughing spells of 2 hours' duration would be reduced
by as much as 2.5 episodes per person per year. Also, fewer
episodes of shortness of breath and pain on deep inspiration are
predicted to occur. Both are important consequences of air
pollution control. We estimate that the annual monetary benefits
associated with these improvements in health would be on the
order of $ 800 million annually 23 .
    Comparison. To summarize, according to OTA, the costs
associated with a 35%  reduction in nationwide emissions of VOCs
in nonattainment areas will be at least $ 8.8 billion annually by
the year 2004 and could be as much as $ 12 billion. Yet the acute
health improvements that we predict to result from these  changes
are valued at no more than $ 1 billion annually and could be as
little as $ 250 million. The high estimate relies on the most
generous of the four clinical studies that the EPA sanctioned in
its staff paper on the health effects associated with O.sub.3 and
other photochemical oxidants 24 . We also assumed that exercise
rates would be high in the exposed population (which increases
health benefits), and we included benefits even for those engaged
in light or moderate exercise. Subject to the caveats discussed
below, total health benefits are still relatively small.
    In contrast to, say, the removal of lead from gasoline, for
which estimated  benefits are well in excess of costs 25 , the
benefit-cost comparison for national O.sub.3 control is
unfavorable. The reasons for this are, in part, the  relatively
small improvements in ambient O.sub.3 levels that the controls
effect (which in turn imply fairly small benefits) as well as the
high costs of control."

I am not going to bother to try to make these figures relevant
because it seems we have better ones.

In a comment on the article to the New York Times (4/30/91),
Michael Oppenheimer of the Environmental Defense Fund said, "The
study's Achilles heel is that it cannot capture the nonmontizable
aspects of cleaner air" such as aesthetic benefits.  I'm not
sure.  It ignored the effects on death.

The following quotes are useful background, but they contain no
useful numbers.  They are what gave me the idea of limiting our
analysis to I&M programs.  The quotes from the next article, by
Beaton et al., start where this one leaves off.  And Beaton et
al. provide an estimate of the benefits of an I&M program at
reducing pollution (at the end of the part quoted here).

Calvert, J.G., Heywood, J.B., Sawyer, R.F., & Seinfeld, J.H.
(1993).  Achieving acceptable air quality: some reflections on
controlling vehicle emissions.  Science, 261, 37-45.

"High emitters. It has become increasingly apparent that most
of the mobile source emissions are caused by a small percentage
of the vehicles. A compelling body of remote-sensing and
roadside data shows, more or less regardless of locale, that
about 50% of the CO and HC emissions come from 10% of the
vehicles 7, 8 . Figure 2 shows the CO and HC emissions from the
1989 random roadside survey conducted by CARB. The data set
comprises 4400 vehicles sampled at 60 locations throughout the
state. These plots show just how significant are the CO and HC
emissions from the dirtiest one-fifth of the fleet in each model
year. A relatively small amount of the overall emissions comes
from the very old cars because they contribute such a small
amount to the total vehicle miles traveled. The high emitter
problem appears to span all model years and results from dirty 
vehicles that are driven significant distances. One would expect
older cars to have higher emissions, and some have very high
emissions (super emitters). What was not expected was the high
emission rates detected in the worst 20% of more recent model
cars. This evidence of super emitters or of modes of
vehicleoperation with high emission rates (such as heavy
accelerations) that occur more commonly than expected represents
the main problem of excessive motor vehicle 
emissions.
  Old cars have high emissions because their standards were
higher and they have had a long time either to deteriorate or to
be altered somehow. However, why are there so many recent model
super emitters? The limited evidence from roadside inspections
indicates that several factors are important. A sizable fraction
(15 to 30%) of cars have had their emission controls tampered
with, by either the owner or a mechanic. The implementation of
emission. controls 15 to 20 years ago did worsen driveability and
fuel economy. However, today's engines have been carefully
optimized to combine control of emissions with good fuel economy
and driveability. However, when these engines do fail they tend
to become fuel-rich, leading to excessive HC and CO emissions.
Some proportion of vehicles have been misfueled - that is, used
with the often cheaper leaded gas instead of the required
unleaded gas. One tankful of leaded gas causes major poisoning of
the catalyst. Because leaded gas has essentially disappeared from
the U.S. market, this problem is no longer important.
  In both older and newer cars, components of emission control
systems do malfunction and required maintenance is not always
done or done properly. If key engine components (such as the
airflow meter or exhaust gas oxygen sensor)malfunction, emissions
can increase markedly. Such component malfunctions and failures
do occur, with owners who apparently do not notice or ignore
service or check-engine warning lights."

"Inspection and maintenance programs. Inspection and
maintenance programs are intended to reduce in-use vehicle
emissions through the identification and repair of vehicles that
do not meet exhaust emission standards. Today, IM is typically a
measurement of tail pipe emissions at two different engine speeds
with no load on the engine. The EPA established guidelines for IM
programs, and states were allowed to take a credit for a 25%
reduction in vehicle emissions if they had an IM program.
Simply having a program in place was sufficient to earn credit
for the 25% reduction; no requirement existed to link IM credits
to enforcement or actual in-use vehicle emissions. Of the
options available for the reduction of motor vehicle emissions, 
a well-designed IM program is among the most cost-effective.
Unfortunately, the performance of IM programs has generally not
been subject to the same degree of scrutiny as have other
elements in the matrix of motor vehicle emissions controls.
  These programs have generally not met their goals for several
reasons: (i) evaporative emissions have not been tested, (ii) the
repair of too many high-emitting vehicles has been waived because
of cost limits, (iii) the tail pipe test on the operation of
unloaded engines is not representative of on-road emissions,
(iv) testing has often been performed incompetently, and (v)
cheating has taken place both through collusion between the
tester and vehicle owner and by vehicle owners modifying their
vehicles before and after scheduled testing. The EPA has proposed
an enhanced IM program that would require a dynanometer test
under varying engine loads, would test evaporative emissions, and
would raise the repair cost limit substantially. Although these
are clearly steps in the right direction, the enhanced IM program
still has some serious defects. Most notably, it lacks the
necessary component of significant on-road, in situ remote
sensing to validate the emissions reductions and to catch cars
that have been tampered with between inspections.  An
important aid to the maintenance of good emissions control over
the useful life of each vehicle is the incorporation of on-board
engine and emission control system diagnostics (OBD). These
devices are combinations of sensors, computer diagnostics, and
warning lights that alert the driver and maintenance personnel to
problems that affect the emission control system. Both the 1990
Clean Air Act Amendments and CARB require that during the next
few years, extensive OBD capability be built into new vehicles.
To be effective, these diagnostics must be robust and the OBD
regulations must be stated largely in terms of the requirements
for the performance of emission control systems rather than the
specification of individual technologies. Unfortunately, the OBD
requirements as currently stated by EPA and CARB seem to be
excessively detailed and headed in conflicting directions. This
potential clash of requirements should be rectified.
  Remote sensing. As discussed above, there is considerable
evidence from tunnel tests and measurements of off-cycle mobile
fleet emissions that suggests real world emissions from the
mobile fleet are probably much higher than those produced from
the stationary FTP test on individual cars. Thus, an IM program
should include the measurement of emissions of the mobile fleet
as they occur in actual operation. The remote monitoring of tail
pipe emissions of passing cars is a cost-effective tool that can
allow the identification of high emitters, which can then be
targeted for repair. Monitoring sites, relocated regularlyto
appropriate positions that are invisible to the drivers, will
allow sampling of actual emissions. Notification of the car
owner (identified from a license plate photograph) can be given
so that a more detailed examination can be made by inspectors,
high emissions verified, problems identified, and correction of
problems ensured. The combination of remote-sensing programs with
IM programs to focus inspection resources on the higher emitting
vehicles is an especially attractive strategy."

Beaton, S. P., Bishop, G. A., Zhang, Y., Ashbaugh, L. L., Lawson,
D. R., & Stedman, D. H. (1995).  On-road  vehicle emissions: 
regulations, costs, and benefits.  Science, 268, 991-993.

"For any region violating air quality standards, a State
Implementation Plan  (SIP) must be submitted to and approved by
the U.S. Environmental Protection Agency (EPA). In approving the
SIP, EPA grants credit for each portion of the plan, including
new  vehicle emission  standards, new fuels, vehicle scrappage, 
and inspection and maintenance (IM) programs, on the basis of
predictions from a spreadsheet computer model (3). This EPA model
treats all cars of a given model  year as having the same
odometer reading, the same annual mileage accumulation,  and an
equal likelihood of emission control system problems, The model
has had little success in predicting urban on-road  vehicle
emissions  (4), and the use  of unverified computer models as the
sole guide for public policy decisions is controversial (5).
Calvert et al. (6) criticized the model and recommended in-use
surveillance programs to identify high-emission vehicles and to
monitor progress in emissions reduction. We present such a study
and discuss its policy implications.
    The California Study   During the summer of 1991, we placed
an on-road remote sensor of exhaust CO and HC emissions (7) at
various urban locations throughout California. We identified
66,053 different vehicles for which we collected 91,679 records
with  valid HC and CO measurements (8). The emission distribution
is highly skewed; the half of the fleet with the lowest emissions
contributed less than 10% of the CO and HC, while a few
high-emission vehicles dominated the mean values. In this
instance, 7% of the vehicles accounted for 50% of the on-road CO
emissions, and  10% of the vehicles accounted for 50% of the
on-road HC emissions. These vehicles we call gross polluters.
About 5% of the vehicles were gross polluters  for both HC and CO
(9).
    It is often assumed that gross polluters are simply old
vehicles. In fact, all model years of vehicles we have measured
on the road include some proportion of gross polluters, as shown
in Fig. 1. We found that the highest emitting 20% of the newest
cars were worse polluters than the lowest emitting 40% of
vehicles from any model year, even those from model years before
the advent of catalytic  converters (1970 and earlier). These
data are typical of CO and HC results across the United States
and at many other locations worldwide (10): Differences in
emissions within a model year are greater than differences
between the averages of the various model years. Correlation of
average on-road emissions and vehicle age shows that these
results cannot be dismissed as random samples of normal vehicle
behavior (11).    For 2 weeks during the California study, we
operated two remote sensors on Rosemead Boulevard in South El
Monte near Los Angeles (12). Vehicles identified  by these
sensors as gross polluters were immediately pulled over, and a
voluntary California Smog Check--an emission control system
inspection and tailpipe test--was administered. The remote
sensors measured 58,063 unique vehicles, and we obtained Smog
Check data on 307 of these vehicles. Of these, 126 (41%) showed
deliberate tampering, and another 77 (25%) had defective or
missing equipment that may not have been the result of tampering
(for example, missing air pump belts). Overall, 282 (92%) failed
the inspection even though all had valid registrations. Thus,
less than 8% of the vehicles identified as gross polluters by
remote sensing passed an immediate roadside test. When random
pullover studies were carried out by the California Bureau of
Automotive Repair  and the California Air Resources Board,
approximately 60% of the vehicles passed the roadside test,
whether or not they were registered in a region of California
with a scheduled IM program (13). Of the vehicles inspected in
the present study, 76 were recruited for an immediate IM240 test
(a loaded-mode dynamometer  test) and their emissions were
compared with EPA-recommended pass-fail values. All but three
vehicles failed the IM240 test, and these three had already
failed the roadside Smog Check (14). These data show that
vehicles identified as gross  polluters by the remote sensors
were poorly maintained or had been tampered with (the majority
apparently illegally); they were not a subset of normally
maintained vehicles that were temporarily emitting more
pollutants becausetheir engines were cold or they were
accelerating hard (11). If we could have inspected all 3271 gross
polluters identified by the remote sensors, we estimate that only
266 of the 58,063 vehicles (0.5%) would have passed the roadside
inspection.
    This study indicates that the large difference between the
majority of cars  and the few gross polluters (the front-to-back
difference in Fig. 1) is caused by poor maintenance or tampering.
The smaller dependence of average emissions on vehicle age may be
the combined result of deterioration, older technology, and
poorer maintenance of older vehicles.
    Implications for Emission Control Policy and Cost
Effectiveness
   On-road emissions must be controlled to reduce ambient
pollutants. The skewed distributions shown in Fig. 1 imply that
policies that treat all vehicles equally, or that target new
vehicles, are likely to be less cost-effective than  those that
recognize the overriding importance of maintenance and that
target poorly maintained vehicles regardless of their age. We
used the California data  to estimate the cost effectiveness of
several proposed or hypothetical programs."

[Several omitted paragraphs expand the last statement.]

"In a targeted repair program, the worst 20% of all vehicles
from each model  year (the back row in Fig. 1) would be repaired
to achieve the average emissions of the remaining 80% of the same
model year. This action would result in a 50% reduction in HC
emissions and a 61% reduction in CO emissions. Scrappageprograms
pay $ 700 to $ 1000 per vehicle, whereas the repairs necessary to
move  a gross polluter to the lower emission categories average
around $ 200 (21). The result is a 57% reduction in HC emissions
and a 69% reduction in CO emissions per billion dollars spent.
The full cost of this identification and repair program could be
raised over a 4-year period by means of an annual $ 11 fee per 
vehicle (22). Because the pullover study showed that about half
of the gross polluters had been illegally tampered with, the cost
of the program would be cut in half if the owners of these
vehicles were required to pay for their own repairs. Even if
repair costs were as high as $ 400 per vehicle, the targeted,
subsidized repair program is still estimated to be the most
cost-effective option."

Schwartz, J., Wypij, D., Dockery, D., Ware, J., Zeger, S.,
Spengler, J., Ferris, B. Jr. (1991).  Daily diaries of
respiratory symptoms and air pollution: Methodological issues and
results.  Environmental Health Perspectives, 90, 181-187.

"Any effect of pollution exposure is not nececessarily
contemporaneous.  ... For instance, Dockery et al. () reported a
lag of 1 to 2 weeks between exposure to high levels of
particulates and reductions in lung function.  In contrast,
Spector et al. () and Kinney et al. () reported that high ozone
exposure causes almost immediate reductions in lung function."

The following paper was added after I wrote the rest, so it may
not agree with other things here, but it is quite recent.

Pope, C. A., Bates, D. V., & Raizenne, M. E. (1995).  Health
effects of particulate air pollution: Time for reassessment?
Environmental Health Perspectives, 103, 472-480.

This is a review of recent studies of health effects of
particulate air pollution.  It draws the following conclusions:

Acute morbidity: A 10 microgram/m3 increase in PM10 [the standard
increase hence forth] was typically associated with a 1-10%
increase in symptoms such as cough, combined lower respiratory
symptoms, and asthma attacks.  These effects were also observed
at comparable PM10 lebels near or even below 150 micrograms/m3.

A [] increase on the day of the visit or 1-2 days before the
visit was typically associated with a 1-4% increase in hospital
visits.

Chronic morbidity: .. a 10-25% increase in bronchitis or chronic
cough.

Acute mortality: .. an increase in daily mortality equal to
0.5-1.5%.

Chronic mortality:  .. in increase in daily mortality equal to 3%
or more.  The strongest associations were observed with
cardiopulmonary disease and lung cancer deaths...

Pope, C. A., Thun, M. J., Namboodiri, M. M., Dockery, D. W.,
Evans, J. S., Speizer, F. E., & Heath, C. W. Jr. (1995).
Particulate air pollution as a predictor of mortality in a
prospective study of U.S. adults.  American Journal of
Respiratory and Critical Care Medicine, 151, 669-674.

Abstract: "... The present
study evaluates effects of particulate air pollution on mortality
using data from a large cohort drawn from many study areas. We
linked ambient air pollution data from 151 U.S. metropolitan
areas in 1980 with individual risk factor on 552,138 adults who
resided in these areas when enrolled in a prospective study in
1982. Deaths were ascertained through December, 1989. Exposure to
sulfate and fine particulate airpollution, which is primarily
from fossil fuel combustion, was estimated from national data
bases. The relationships of air pollution to all-cause, lung
cancer, and cardiopulmonary mortality was examined using
multivariate analysis which controlled for smoking, education,
and other risk factors.  Although small compared with cigarette
smoking, an association between mortality and particulate air
pollutionwas observed.  Adjusted relative risk ratios (and 95(
confidence intervals) of all-cause mortality for the most
polluted areas compared with the least polluted equaled 1.15
(1.09 to 1.22) and 1.17 (1.09 to 1.26) when using sulfate and
fine particulate measures respectively. Particulate air was
associated with cardiopulmonary and lung cancer mortality but not
with mortality due to other causes. ..."

The paper uses a data set from Philadelphia for extensive
analysis, finding a relative risk of 1.07.  That is, death rates
were 7% higher when particulates increased by 100 micrograms per
cubic meter.  Figures in the paper suggest that the normal daily
death rate in Philadelphia is 50, so this means an increase of
3.5 deaths out of a population of 1,586,000 (1990 census).  (A
check of the cited paper revealed that deaths were in Philadelpha
proper.)  The paper also gives a mean level of 77 TSP (total
suspended particulates) for Philadelphia.  The increased
probability of death is therefore .0000023 per day for a 100
microgram/m3 increase in TSP.  This is not age adjusted.

The following abstract is interesting, and supports the
conclusion just reached.  I would guess that .0000025 is a better
estimate, but this is the least of our worries.

Schwartz J.
Environmental Epidemiology Program, Harvard School of Public Health,
Boston, Massachusetts 02115.
What are people dying of on high air pollution days?.
Environmental Research.  64(1):26-35, 1994 Jan.

The air pollution disasters in London in 1952, the Meuse valley in 1930,
and in Donoroa, Pennsylvania, in 1948 made it clear that extremely high
levels of particulate-based smog could produce large increases in the
daily mortality rate. Recent studies of fluctuations in daily air
pollution and daily mortality have reported associations at much lower
concentrations in London during the 1960s and in Philadelphia,
Steubenville, Santa Clara, St. Louis, Utah valley, Detroit, and eastern
Tennessee in the 1970s and 1980s. Whether these associations are causal or
not is a matter of considerable public health concern. If the detailed
pattern of the deaths at these lower concentrations appeared similar to
the pattern in London, this would strengthen the argument for causality.
To examine this issue, the death certificates from Philadelphia were
examined on the 5% of the days with the highest particulate air pollution
and the 5% of the days with the lowest particulate air pollution during
the years 1973-1980. There was little difference in weather between the
high and low pollution days, but total suspended particulate matter
concentrations averaged 141 micrograms/m3 on the high pollution days
versus 47 micrograms/m3 on the low pollution days. The relative risk of
dying on the high pollution days was 1.08 P < 0.0001. The relative
increase was higher for COPD (1.25) and pneumonia (1.13). Deaths were also
elevated for heart disease and stroke; however, there was a substantial
increase in the reports of respiratory factors as contributing causes for
those underlying causes of death. Dead-on-arrival deaths and deaths
outside of hospitals and clinics were also disproportionately increased.
This paralleled the pattern seen in London in 1952. The age pattern of the
relative risk of death was also similar. This adds to the evidence that
the association is causal.

The following abstract is the first report of the original study. 
It breaks down the death-rate increase by age.

Schwartz J.  Dockery DW.
Increased mortality in Philadelphia associated with daily air pollution
concentrations.
American Review of Respiratory Disease.  145(3):600-4, 1992 Mar.
Cause-specific deaths by day for the years 1973 to 1980 in Philadelphia,
Pennsylvania, were extracted from National Center for Health Statistics
mortality tapes. Death from accidents (International Classification of
Disease, Revision 9 greater than or equal to 800) and deaths outside of
the city were excluded. Daily counts of deaths were regressed using
Poisson regression on total suspended particulate (TSP) and/or SO2 on the
same day and on the preceding day, controlling for year, season,
temperature, and humidity. A significant positive association was found
between total mortality (mean of 48 deaths/day) and both TSP (second
highest daily mean, 222 micrograms/m3) and SO2 (second highest daily mean,
299 micrograms/m3). The strongest associations were found with the mean
pollution of the current and the preceding days. Total mortality was
estimated to increase by 7% (95% CI, 4 to 10%) with each 100-micrograms/m3
increase in TSP, and 5% (95% CI, 3 to 7%) with each 100-micrograms/m3
increase in SO2. When both pollutants were considered simultaneously, the
SO2 association was no longer significant. Mortality increased
monotonically with TSP. The effect of 100 micrograms/m3 TSP was stronger
in subjects older than 65 yr of age (10% increase) compared with those
younger than 65 yr of age (3% increase). Cause-specific mortality was also
associated with a 100-micrograms/m3 increase in TSP: chronic obstructive
pulmonary disease (ICD9 490-496), +19% (95% CI, 0 to 42%), pneumonia (ICD9
480-486 & 507), +11% (95% CI, -3 to +27%), and cardiovascular disease
(ICD9 390-448), +10% (95% CI, 6 to 14%). These results are somewhat higher
than previously reported associations, and they add to the body of
evidence showing that particulate pollution is associated with increased
daily mortality at current levels in the United States.

BURNETT RT; DALES RE; RAIZENNE ME; KREWSKI D; SUMMERS PW;
ROBERTS GR; RAADYOUNG M; DANN T; BROOK J
EFFECTS OF LOW AMBIENT LEVELS OF OZONE AND SULFATES ON THE
FREQUENCY OF RESPIRATORY ADMISSIONS TO ONTARIO HOSPITALS
ENVIRONMENTAL RESEARCH V0065 N2 MAY 1994 pp. 172-194.
To investigate the acute respiratory health effects of ambient
air pollution,the number of emergency or urgent daily respiratory
admissions to 168 acute care hospitals in Ontario were related to
estimates of exposure to ozone and sulfates in the vicinity of
each hospital. Ozone levels were obtained from 22 monitoring
stations maintained by the Ontario Ministry of the Environment
for the period January 1, 1983 to December 31, 1988. Daily levels
of sulfates were recorded at nine monitoring stations
representing three different networks operated by the Ontario
Ministry of the Environment and Environment Canada, Positive and
statistically significant associations were found between
hospital admissions and both ozone and sulfates recorded on the
day of admission and up to 3 days prior to the date of admission.
Five percent of daily respiratory admissions in the months of May
to August were associated with ozone, with sulfates accounting
for an additional 1( of these admissions. Ozone was a stronger
predictor of admissions than sulfates. Positive and statistically
significant associations were observed between the
ozone-sulfate pollution mix and admissions for asthma, chronic
obstructive pulmonary disease, and infections. Positive
associations were also found in all age groups, with the largest
impact on infants (15% of admissions associated with the
ozone-sulfate pollution mix) and the least effects on the elderly
(4%). Temperature had no effect on the air pollution-admission
relationship. ...

"Model III predicted a 6.0% reduction in admissions corresponding
to a reduction in the daily 1-hr maximum ozong level, lagged 0 to
3 days, from its mean value of 50 to 0 ppb."  (Note although this
estimate is slightly reduced when sulfates are included, it seems
reasonable to assume that reduction in O3 will correlate with
reduction in sulfates, and the overall effect of O3 plus sulfates
is also 6.0%.)

"Asthma accounted for 39% of respiratory admissions, COPD
(chronic obstructive pulmonary disease) for 25%, and infections
for 36%.  Approximately 12% of admissions occurred in infants,
30% in the 2-34 age group, 22% in the 35-64 age group, and 36% in
the elderly."

"These results pertain to a population of approximately 8.7
million."  Average number of daily admissions for the whole
sample was 108.  We can thus estimate that a reduction of 50 ppb
in O3 will result in a reduced risk of (.06)(108)/8,700,000, or
.0000007 per day.  That's about 1 in a million.

Ostro, B. D., Lipsett, M.J., Mann, J.K., Krupnick, A., &
Harrington, W. (1993). Air pollution and respiratory morbidity
among adults in Southern California. American Journal of
Epidemiology, 137, 691-700.

Figure 1, p. 696, shows the relation between lower-respiratory
symptoms (cough, wheezing, chest pain, etc.) and ozone
concentration measured in 1-hour maximum pphm (parts per hundred
million) for each day. Symptoms were reported daily. Reading
from the graph, the level is .011 for no ozone and .021 for 30
pphm. Effects on upper-respiratory symptoms (nose, etc.) were
statistically significant only among people without air
conditioners.

National Research Council. (1991). Rethinking the ozone problem
in urban and regional air pollution. Washington, DC: National
Academy Press.

p. 283-4 - "motor vehicle emissions typically account for 40% of
the total VOC and NOx emissions in any region."

284 - in a CA study, "for the no-load 1000 RPM idel test ... 10%
of the vehicles were responsible for about 60% of the exhaust
idle CO, and ... another (not necessarily the same) 10% were
responsible for about 60% of the exhaust VOCs. The results
showed only a weak relationship between vehicles that emit large
amounts of CO and those that emit large amounts of VOCs, based on
the idle test.

Lawson, D. R. (1995).  The Costs of 'M' in I/M -- Reflections on
Inspection/Maintenance Programs."  Journal of the Air & Waste
Management Association.  45, 465-476.

The study they report on shows that high-emitters are the only cars worth 
going after.  Over 1100 vehicles which failed California's Smog Check 
program were studied.

"The repair costs averaged $89 per vehicle."

"In the case of HC [hydrocarbons], more than half of the emission
reductions come from the dirtiest decade [tenth]."  For these
cars, average coss were $103 per vehicle.  (I can't tell whether
this is per vehicle in the sample or per vehicle repaired.)