---start----
smith
epidemiology
2/4/98

and now for something completely 
different...
there's really something very attractive about a good 
insult.

first person said: do you mind if i smoke?
reply: i don't care 
if you burn
i wish i'd said that
you will

winston, if i were married to 
you, i'd put poison in your coffee
nancy, if i were married to you i'd 
drink it

Clare Booth: age before beauty
Dorothy Parker: pearls before 
swine

sandwich to wilks: really, i don't know whether you'll die on the 
gallows or of the pox. 
wilks: that depends on whether i embrace your 
principles or your mistress.

Now changing step completely and thinking 
about infxs dz and their dynamics:

we're going to discuss the dynamics 
of transmission of infxs diseases. 
prefacing this more 
mathematically...qualitative aspects of disease transmission. you know 
that morbidity and mortality vary from disease to disease.

why is this? 
why do m&m vary from infection to infection? why is cholera more severe 
than rhinitis? an answer to this is being suggested - most of this 
lecture is taken from some dude's paper about the evolution of virulence, 
and you might want to read Ewald's papers about this. there are 
biochemical and genetic mechanisms of virulence.

virulence here is 
describing the degree of sickness, case fatality rate. we know a lot 
about biochemistry nd genetics, but til recently had no idea how to 
explain why differences in virulence developed. no one cared. accepted 
dogma used to be that parasite:host associations (including things like 
viruses, bacteria, and protozoa) always eventually evolve to a state 
where a parasite does the host very little harm. if a parasite kills its 
host, it reduces its own chance of surviving, so an adapted parasite 
doesn't kill its host. this would mean that very virulent infections are 
newer than nonvirulent infections. however, over the last few years this 
has been challenged. a continual decrease in virulence isn't the only 
evolutionary pathway that makes sense or that can be followed according 
to the new thinking. two lines of reasoning: first, the kinds of 
mathematical models we're going to deal with were used to ask what if 
questions...you ask it questions and it delivers back to you sometimes 
challenging ideas. you find that there are circumstances where infections 
with organisms that have very high virulence can maintain themselves 
well. second, in detailed examination of relationship b/w mode of 
transmission and degree of virulence, and the balance that is 
established, we found - assuming virulence is proportional to the number 
of viruses or bacteria that a host contains, the more shedding and more 
transmission occurs - so lots of virus equals lots of transmission. a 
host with a lot of bugs is likely to be very sick, and very infectious. 
tradeoff b/w making large numbers of bugs and increasing chance of 
transmission, and having the host survive long enough to contact another 
susceptible host to transmit it to. so there is this trade off. as a 
consequence of this idea - selection pressures continue to push pathogen 
to make more of itself, unless some more important imperative comes into 
play. pathogens are trying to ensure their own existence, to mke a lot of 
themselves. that will happen as long as nothing more important gets in 
the way. that would be the host dying. if the host dies, the whole thing 
fails. 

slide: selection for resistance to myxoma virus has occurred, 
but strains of intermediate virulence were preferentially selected rather 
than strains of very low virulence.

myxoma was introduced into australia 
in 1951, and had a 99% mortality rate in lagomorphs. it was awful. the 
rabbits were covered in sores and transmission occurred when bugs 
mechanically transmitted the virus. at first, it caused a huge decline in 
rabbit population density. there were fears that it would kill all the 
rabbits in the country but that didn't happen. the first strains 
introduced had very very high virulence. under the old dogma, you'd 
expect it to shift toward very low virulence. but, what happened is that 
it shifted to an intermediate level of virulence - most of the virus 
isn't the very low virulence type, it's intermediate level. now, if 
animal dies really quickly, it's unlikely to be able to spread the 
disease. also, for disease to last overwinter, you need some rabbits to 
harbor the virus over the winter. you can't have rabbit die too soon, you 
can't kill all the rabbits, but you do need the sores to occur, so you 
have intermediate virulence. this has some clinical consequences esp for 
human dz.

consider modes of transmission and how this affects our 
thinking...
this hypothesis suggests that parasites, viruses, bacteria, 
whatever, that are transmitted by biting arthropods, should evolve 
towards higher levels of virulence than if it's directly transmitted say 
by droplet inhalation. transmission of pathogens by arthropod vector 
isn't very adversely affected if host gets sick and stops moving around. 
transmission from host doesn't require host to move around. sick hosts 
are bitten by more bugs than well hosts, because their defenses fail, 
they don't move away from bugs or try to brush them off or whatever. so 
it doesn't matter to pathogen that it makes so much of itself that it 
makes the host very sick (high morbidity). now, it depends on the vector 
to move from host to host, so it can't make the vector sick. it wants to 
make the host kinda sick, but the vector not sick. this holds up. malaria 
and other arthropod borne infections are often more virulent than 
airborne infections. sick,immobile hosts make a lot of pathogen which is 
sucked up by arthropod and easily transmitted.

waterborne transmission. 
consider this in the same light. suppose for example that if there is a 
severe case of diarrhea, many pathogens are released and they enter water 
supply. if this mixes with unprotected drinking water, large numbers of 
people can be infected - so high morbidity doesn't impair transmission. 
if high morbidity results in release of many pathogens, it may enhance 
transmission. so likelihood is waterborne infections will be more 
virulent than directly transmitted infections, too.

many diseases have 
more than one mode of transmission. the proportion of outbreaks of a 
particular disease that are waterborne rises directly with fatality rate. 
waterborne diseases have greater virulence than directly transmitted 
diseases.

hospital attendants, nurses, doctors, veterinarians - all are 
disease vectors. vectors of neonatal diarrhea, puerperal sepsis, etc. if 
you aren't careful it is easy to transfer diseases between wards, etc. 
this isn't to say that you will get sick - you could just mechanically 
carry the infection from one room to another or one animal to another. 
sick, immobilized people or animals tend to get a lot of attention and 
handling - so they are more vulnerable to getting a disease vectored by 
an attendant or handler. many hospital based infections show greater 
virulence than the same infections outside the hospital. there's no 
penalty for immobilizing the host in the hospital, b/c people vector 
infections for you. in fact, immobilizing host creates need for more 
handling, which is an advantage to the pathogen. hospitals are dangerous 
places wrt infectious diseases. 

consequences: in a nutshell. 
high 
levels of shedding are associated with high levels of morbidity and 
mortality. infections relying on animal to animal contact for 
transmission, eg infected animal must be mobile, then selection pressures 
are such that there are fewer pathogens/host, less morbidity 
andmortality, and infected animals are more mobile. this is directly 
transmitted infections, like a cold. 

in vector borne diseases, water 
borne disease, and hospital attendant borne dz, high levels of pathogen 
production and high morbidity are beneficial to pathogen, transmission 
doesn't depend on host moving around. 

1. virulence responds to 
selection pressures quickly
2. virulence is often sassociated with large 
numbers of pathogens and host responses that enhance transmission 
(coughing, diarrhea)
3. when the likelihood of transmisison is high, 
selection often favors increased virulence, and vice versa. 

if you want 
proof of evolution, look no further than evolution of antibiotic 
resistance, which takes only months. 

there was a cholera outbreak in 
tanzania; the people infected were given tetracycline. in the first 
month, there was no measurable resistance to the drug. within 5 mos, 75% 
of people given tetracycline weren't cured. the pathogen got resistant. 
even nematodes can develop resistance quickly within a couple of years. 
if you want a rule of thumb for any pathogen - be it fungus, bacterium, 
nematode, insect - you generally get resistance w/in 50 generations, on 
average in 5-10 generations. the lesson here is that virulence can 
respond very rapidly to changed environemental circumstances. this is bad 
wrt abx resistance, but sometimes is good. if you try to disrupt normal 
transmission, you have to be careful, b/c you're changing the 
environment, and you might increase virulence or decrease virulence. you 
want to decrease it. you should be able to influence it quickly.


consider influenza. 10 million soldiers were killed in combat in WWI; 20 
million were killed by influenza in that pandemic of 1918, which 
persisted for 2 years after that. now, the flu kills about 1 person/100. 
before, it killed about 10/100. why? the soldiers fought in extensive, 
crowded trenches, with little freedom of movement, for a long time. you 
were going to spread disease very easily. selection was in direction of 
high number of pathogens unless something stops it. if you got the flu 
and early immobilization and death didn't stop you from spreading it, the 
pathogen would be very very virulent, because there was no reason not to. 
then the war ended. within about 3 years, that same strain reduced its 
virulence - because the soldiers were't smushed together anymore, and 
that highly virulent strain became very difficult to transmit because 
people density decreased. 

you can make changes in environment in which 
disease occurs that change the nature of the disease. 

HIV/AIDS


initially there were just a few cases in big cities. 8 yrs later, it was 
all over both coasts, and now it's all over the country. it's a terrible 
disease but - there are two things about it that are interesting. one, 
why has it evolved to its current level of virulence, and two, how will 
that virulence change? well, an idea to explain the current virulence is 
the transfer to humans from monkeys - perhaps was benign in the monkeys, 
due to long association b/w monkey and virus - and then was virulent in 
humans since it had just entered humans. evolution hasn't had time to 
push it toward benign association. but, that was the old thinking. ewald 
suggests that as a sexually transmitted disease, it evolved to ensure 
transmission in very low population density areas. infection in man, 
therefore, probably was sputtering around in rural areas for a very long 
time. he then suggests that in fact something happened and virulence 
started to increase rather than decrease. why? if there are frequent 
enough contacts b/w susceptible individuals, the tradeoff is pushed 
toward producing more virus. since more virus == more shedding, this 
enhances transmission. but when contacts are infrequent, there are 
penalties for high virulence - infecteds die before spreading it. so, 
long, benign infections provide long opportunity to infect someone else. 
so aids sputtered along in low density, stable, african villages for a 
long time. virulence was low. we know HIV was present in africa at least 
20 yrs prior to the current pandemic, but the clinical signs/disease 
associated with HIV were not present. from about 1960 onwards, partly as 
result of decolonization of africa by european countries and partly for 
other reasons, there was massive migration from rural to urban africa, 
which disrupted cultures, and more virulent strains appeared. patterns 
repeated in other cultures - many very promiscuous men, etc etc. the 
number of new sexual partners per unit time increased, and virulence 
increased, because there was no worry about the infected person not 
contacting a new person to infect. as rate of new partners increases, 
probability of infection increases. in the 70s, many gay men had very 
large numbers of sexual partners - so there was no reason for the 
pathogen not to be very virulent, because chance of spreading was so 
high. in the heterosexual population, there were many fewer partners per 
year, and therefore increased virulence was dangerous to the pathogen, 
because it could kill infected people before they could spread it.

when 
it was realized that HIV was a sexually transmitted virus, all we could 
do was try to educate people. so they explained that the rate of 
aquisition of new sexual partners was a major factor, and they suggested 
that for personal safety and public health reasons, people should reduce 
their numbers of sexual partners. for a time, the average number of new 
sexual partners for homosexual men decreased significantly. now, if this 
continued to the present day, virulence of HIV would decline. why? 
because opportunity for transmission is lessened. now, virus has to 
maintain itself in individual host for longer time. so, educational 
programs will protect individuals, will slow the rate of spread of 
disease, but also will decrease the virulence of the disease. 

there are 
many veterinary diseases where the same kind of thing is true. there may 
be things that mere education can do that affect the actual character of 
the disease.
---break---

ecological epidemiology is that branch of the 
discipline which deals with the dynamics of infections in populations of 
hosts...we shall begin with an essentially qualitative discussion of 
epidemic and endemic behavior. 

this is the most fascinating part of 
epidemiology.

an epidemic is said to occur when the occurence of a 
disease increases above expected levels. this is the best definition we 
have. classical swine fever doesn't occur in the US, but it does in 
Germany. it's a bad thing. It's endemic in much of europe and there are 
occasional epidemics. normally, 0-5 outbreaks/month occur. but sometimes 
they have epidemics where up to 25 outbreaks/month occur. they have to 
control it by depopulation (slaughter), quarantine, extensive monitoring.


some guy in the Netherlands, some epidemiologist, was supposed to predict 
the size of the outbreak of swine fever - then he found out it was being 
transmitted in semen from an AI unit - so he threw up his hands in 
despair.

bubonic plague - there are about 10 cases a year here, but we 
worry about it.
classical swine fever - doesn't occur here at all, but it 
could, nd we need to consider it.

we almost never have time series 
records of the prevalence of infection.
a change in the proportion of 
animals at one time that is infected isn't something people record. the 
best you can do is measure prevalence once. so how are epidemics 
measured? one way is to look at consequences of dz. trichomoniasis, a 
protozoal std of cattle - it causes dairy cows to fail to conceive. so 
the # days open is an index of length of time it takes for cow to 
conceive. a sudden increase in #days open indicates an epidemic of 
trichomoniasis. so we never have measures of prevalence; we make do with 
surrogate measures. often, we don't even have records for individual 
animals - we just have a number of outbreaks, or number of affected herds 
- herd prevalence. 

another example of using consequences of 
infection...infectious laryngotracheitis - respiratory dz of chickens. 
you can record mortality - the number of birds killed rises and falls, 
and when it is high, there is an epidemic. frequently, the epidemic is 
recorded in terms of the consequences of infection, like mortality. 


rabies in raccoons in mid-atlantic states - is this an epidemic? well, 
frequency rose and plateaued. so yes, we're in the midst of an epidemic. 
it doesn't have to drop off again to be an epidemic. 

now for the ILT, 
if 50 birds die in one day the whole flock is killed and sold (in the 
USA), because otherwise, you lose out big time. so we don't get epidemic 
curves for this disease in the US, because as soon as the epidemic 
starts, the whole flock is depopulated. 

often, the epidemic is recorded 
in terms of the number of new cases/week.  in veterinary medicine, the 
number of new cases/week is often applied to valuable animals. in Hong 
Kong in 1992, horses were imported and monitored in preparation for a 
race - number of cases per day were monitored (not per week).

in 
veterinary literature, the number of new cases per time is called 
"incidence" very often, but this is a colloquial use of the term, and 
doesn't really refer to incidence as we've defined it. the WHO actually 
confers a seal of approval on this use of the word incidence, when you're 
really referring to a # of new cases per day or week or year, not the 
instantaneous per capita rate of infection. 

salmonellosis in horses - 
new cases per month were reported, chart shows recurrent epidemics.


infectious bovine keratoconjunctivitis:
chart - plots number infected, 
new infections, and new cases
# of eyes vs weeks
this is pinkeye 
infection in cattle (note to self - diagram drawn inside poultry medicine 
folder!) pinkeye tends to be worse in the summer, when other irritants 
and UV radiation abound. pinkeye causes a drop in production. when we 
consider an epidemic, we think intuitively of a proportion of animals 
that are affected. on this graph, the best index we have of that is the 
"number infected" line. PI went to the same herd every week, tested every 
animal in the herd. that meant that once an animal was infected, they 
retested it every week,so it could be positive more than once of course, 
if it remained infected. so the line that reaches the largest number of 
eyes is the nearest we have to the prevalence of infection curve which we 
intuitively consider as epidemic curve.
the next line is the number of 
new infections. it drops off way before the number of infections does - 
because it doesn't include the old infections. the number infected keeps 
rising though, until there are no more new infections, then it drops. 

worse still is the number of new cases - those animals actually showing 
signs of disease - not all the new infections, just the ones we happen to 
notice! this is much smaller than the new infection line. doesn't really 
represent what's going on with the herd. but this is the most frequently 
reported number!

so realize, not every infected animal turns into a 
case. non case animals can still transmit disease.

ok. endemic 
infectious diseases: an endemic infection is one that has persisted for a 
long time in a given population of hosts. hard to define but this is the 
best we can do. chart: coccidial infections in dogs at VHUP

prevalence 
of coccidia in dogs vs year. on the whole, the prevalence is between 0.1 
and 0.5, so coccidiosis is endemic among dogs in Philadelphia area. but - 
these are generally referrals. a, they may not be from philadelphia, and 
b, they are generally preselected as sicker than normal.

what do we mean 
by "a given population of hosts"?
among ALL pigs in germany, there is 
always at least one pig infected with classical swine fever. so CSF is 
endemic in german pigs. now, there may be some herds which have had swine 
fever for many years...but there are many farms which have never had 
classical swine fever or have been free of it for decades. CSF isn't 
endemic in those populations, is it? no. 

how long does an infection 
have to persist before being "endemic"?
rabies in woodchucks (marmota 
monax) in midatlantic states - this is the slide, but he's talking about 
raccoons. rabies in raccoons has been in US since at least 1940, when 
epicenter was in Florida. when they talk about rabies in raccoons in FL 
and GA they call it an endemic. when they refer to it in midatlantic 
states, they call it epidemic. it came here in 1978 when hunters moved 
infected raccoons up to virginia. so it's been here for 20 yrs, but it's 
still not called endemic. 

does it have to persist at any particular 
level to be called endemic? yes and no. the woodchucks with rabies have 
recurrent epidemics - but it's called recurrent epidemic behavior of an 
endemic. every summer, there are sharp rises in recognized cases. but the 
pattern repeats every year. plus, hey - recognized cases isn't the same 
thing as true cases - maybe in the summer, people see more woodchucks. in 
the winter, people stay inside and don't see the woodchucks. 

endemic 
infections often seem to undergo regular oscillations in prevalence - we 
call this recurrent epidemic behavior. they may be simple seasonal 
variations, they may be real, or artifacts of observation. very often, 
viral dzs will oscillate in prevalence on a 2-4 yr time span. rabies in 
foxes does this. in raccoons it does it to a somewhat lesser extent. 


simple endemic models tend to show oscillations, but even in a closed 
population it is often difficult to decide whether the oscillations are 
explicable in terms of the infection dynamics, or merely the result of 
continual reintroduction of the virus. 

dogs w/parvo - closed colony - 
are recurrent oscillations an inherent characteristic of the virus, or 
are they discrete, separate epidemics, caused by the attendants 
introducing the virus from the outside? who knows? it's hard to figure it 
out. 

simple epidemic model consists of three compartments. notice there 
are no births in the simple epidemic model.

      BY          d

|X|-------->|Y|-------->|Z|

three compartments - X, Y, and Z. The 
letters name the compartments, and stand for the density of animals in 
the compartment. Compartment X comprises the susceptible, naive animals, 
those that are at risk of infection. compartment Y is those that are 
infected and infectious. compartment Z is those that are recovered and 
immune. some books call the compartments S, I, and R (susceptible, 
infectious, recovered/resistant) - often this is called the SIR model. 


d (delta) is the recovery rate. ** we estimate the numerical value of d 
by discovering how long the animals remain in compartment Y, and then 
taking the reciprocal of it. If an animal is infectious for 10 days, d = 
1/10. ** this is important.

BY = rate of infection, aka incidence. 
B=beta, btw. B = transmission constant. Y is the density of animals in 
compartment Y. so infection rate changes based on number of infectious 
animals. how do we measure this? the reciprocal of the average time spent 
in the preceding box. BY is the reciprocal of average time to disease 
onset. if it takes 3 weeks to get infected, then BY = 1/3, or 1/21, 
depending if you're using weeks or days. Now, reciprocal of time to 
disease onset we already know is incidence. so BY = i, true incidence. 
this constantly changes as Y changes. initially it will rise and then it 
will fall. (in an epidemic). in simple epidemic model we assume there are 
no deaths. we assume it happens so quickly that we can ignore births and 
deaths.

this model differs from endemic model because there are no 
births. the density of susceptibles can't change. we aren't adding any 
new animals into X. X either stays the same (no epidemic) or gets 
smaller, as it flows into Y and then Z. 

simple endemic model - the 
same, except you have mu (u) entering and exiting X, and exiting Y and Z. 
u = constant death and birth rates. birth rate equals death rate. 
therefore X+Y+Z doesn't change. 

the simple endemic model also has three 
compartments but now we add births to ensure that there are always new 
susceptibles, and deaths to ensure that population density stays 
constant. in this model we're assuming that animals are being born 
susceptible, not diseased or resistant. it's the new pool of susceptible 
animals that allows disease to persist.

----end----