Encyclopedia of Environmental Science and Engineering, Volume I and II

382 EPIDEMIOLOGY

to fi nd higher rates of colon cancer among those consum-

ing more than the average. But this information we do not

posses, though it is precisely what is required to demonstrate

the relationship conclusively.

LATENCY PERIOD

Another consideration, which is of particular relevance in the

fi eld of cancer and other areas of epidemiology, is that there

is a latency period, often of at least 20 years, between expo-

sure to a substance (like a carcinogen) and the development

of a clinically observable disease (again, like cancer). Thus

the per-capita fat consumption should refer to fi gures of 20 or

more years ago. If the cancer incidence is not showing rapid

secular change, this consideration may not be important. It has

also been shown that there is a similar relationship between the

consumption of protein and colon cancer. It may well be that

both fat consumption and a diet high in protein are among the

causative factors of colon cancer, but it may also be that some

other factor or group of factors, correlated with these two in

particular, is more directly relevant. Lifestyle factors or social

status may each be included, and both subsume at the same

time a wide variety of other measures, some of which may

be more legitimately described as directly etiological. In sum,

therefore, a correlation diagram can lead to the generation of

hypotheses of causation but cannot of itself prove the rela-

tionship. The gross correlation between national averages and

disease incidence needs to be investigated on individual cases

of the disease, for each of which measures of consumption of

putatively carcinogenic items of diet can be obtained, prefer-

ably over a large period of time in retrospect. Data of this kind,

if suffi cient in quantity and reliability in substance, can form

the basis of an informative etiological study, probably using

multivariate analytical techniques.

REGRESSION AND CORRELATION ANALYSIS

The relationship between the incidence of lung cancer and

the number of cigarettes smoked is now well known, and

has been verifi ed many times in a variety of situations. For

most of these studies it is possible to obtain a graph of the

mortality (or morbidity) rate against the number of cigarettes

smoked per day, yielding a straight-line relation of the form

y  a 
 bx between them, where y is the morbidity rate and

x the cigarettes smoked per day, a and b being appropriate

constants. The values of a and b can of course be readily

obtained from the data, being the parameter of the simple

linear regression between the two quantities. Many books on

statistics prescribe the technique of fi tting regression lines.

INTERACTION

A relationship of a similar kind has been shown between the

incidence of esophageal cancer and the number of cigarettes

smoked. The same disease is also related in the same way to

quantity of alcohol consumed. We could combine these two

fi ndings to obtain a regression equation using two quantities

as infl uencing a third (the morbidity rate), of the form

z  a 
 bx 
 cy

where z is the morbidity rate,

x is the number of cigarettes smoked,

y is the quantity of alcohol consumed,

and a, b, and c are appropriate constants.

The form of this equation assumes independence between

the actions of each quantity on the morbidity rate. It has been

found in actual study that the combined effect of both quan-

tities in the same individuals results in an enhanced rate of

morbidity, above the additive effect of the two separately. This

enhancement, amounting to a multiplicative rather than an

additive effect, is known as “synergism” (Figure 8). A similar

example of a synergistic effect is found in the occupational

fi eld by the combination of the effects of exposure to asbes-

tos and cigarette smoking on the development of lung cancer.

Figure 9 shows the separate effects of each agent in terms of

a rate set at 1 for those exposed to neither and also the rate for

both together, which corresponds to multiplying rather than

adding the separate rates. The establishment of a genuinely

synergistic effect requires both extensive and reliable data.

ANALYSIS OF OCCUPATIONAL DISEASE

In studies of occupational disease, the basic question to be

answered is whether there is an excess of cases of the disease

0–40 41–80 81+

0–9

10–19

20+

TOBACCO

(g/day)

10 7.3 18.0

3.4 8.4 19.9

5.1 12.3 44.4

FIGURE 8 Cancer of the esophagus in relation to alcohol and

smoking.

C005_011_r03.indd 382C005_011_r03.indd 382 11/18/2005 10:25:43 AM11/18/2005 10:25:43 AM