Similar topics
community sheet # 4 of Dr farouq  Hadeel Almasri
JU.De :: 3rd year :: Sheets and slides :: Biostatistics
Page 1 of 1 • Share
community sheet # 4 of Dr farouq  Hadeel Almasri
بسم الله الرحمن الرحيم
__________________________________
mediafire.com ?hqi19nch7qhqvpi
__________________________________
__________________________________
mediafire.com ?hqi19nch7qhqvpi
__________________________________
بسم الله الرحمن الرحيم
Epidemiology
As mentioned by the Prof. this lecture will have a portion of 50% of Epidemiology exam questions… but don’t worry, it’s an easy one!!
In the last lecture, we started talking about Epidemiology, it means: the study of the distribution and the determinants of a disease in a human population.
We also talked about what is called “Risk Factor”... the factor that is significantly associated with the occurrence of the disease, in which the “PValue” is less or equal to 5%.
The Criteria Of Causation…at least there are five criteria:
1The type of the study.
2The strength of association, which can be measured by the:
Relative Risk (RR), or
Odds Ratio (OR), or
RValue, which is the regression and correlation coefficient.
3Temporality, means that the risk factor comes before the onset of the disease. Such as in the case of a smoker (the risk factor is smoking) who is prone to have lung cancer.
4DoseResponse, and this means that if the Risk Factor increases, the incidence rate or the probability of the disease increases as well.
5Consistancy, which is the similarity of the results of a study to the results of other studies that are carried out in different parts of the world.
Also we’ve explained in more details the first criterion which is the “type of the study”, and we’ve classified the Epidemiological Studies according to this criterion into:
1Observational studies.
2Experemintal studies.
Then we’ve gone in more details about each, as the following:
Observational studies, classified into:
1Descriptive studies.
2Analytic studies.
The Descriptive studies, the most common type of them is the CrossSectional studies. The CrossSectional studies have 2 objectives:
1Finding out the prevalence rate of the disease.
2Finding out the Risk Factor.
The analytic studies, are 2 types:
1Case Control studies, they have only one objective, which is to find out the Causation or Causality or CauseEffect relationship.
2Cohort studies, which have 2 objectives:
To find out the incidence rate.
To find out Causation or Causality or CauseEffect relationship.
Experimental studies, are 2 types:
1Clinical trials
2Community trials.
And they have 3 objectives:
1To find out the incidence rate.
2to investigate Causation.
3which is the most important; to find out the efficacy or efficiency rate of intervention.
Please note that:
**the incidence rate was mentioned in the “Experimental” and “Cohort” studies only.
**Caseeffect relationship was mentioned in “CaseControl”, “Cohort” and “Experimental” studies only.
** Efficacy or efficiency rate was mentioned in “Experimental” studies only.
**Prevalence rate and Risk Factor were mentioned in “CrossSectional” studies only.
You might be asked in the exam about these notes, for example, the prevalence rate can be calculated in which of the following studies??
1……, 2……,3…….
So memorize them well….
NOW…the second criterion of causation; the strength of association:
It means in Arabic: مقدار العلاقة بين متغيرين
Actually, it’s an excellent field for exam questions..
As aforementioned, it has 3 measures, but here we’ll talk about only two of them:
1Relative Risk (RR), it’s used in cohort studies only.
RR=IREXP/IRNONEXP
IREXP : Incidence Rate of the disease in the exposed group. Exposed group means those who have the risk factor like smoking, so in a community there are smokers (exposed to the risk factor) and nonsmokers (nonexposed to the risk factor).
So in this example, RR=IRsmokers/IRnonsmokers
RR might be:
**Equal to 1 : and this means that there is no association between the two variables.
**More than 1 : this means there is +ve association
**Less than 1 : this means that the factor is protective, the presence of this factor protects against the occurrence of the disease –lung cancer in our example.
So obviously, the determining the IR is important to calculate the RR.
IR of the disease=
(new cases of the disease(over a specific period of time)/sample size)X K(constant)
“K” can be 100 or 1000 or 10000 or 100000
2Odds ratio (OR): is the degree of association between 2 variables (the disease and the factor), which is used in CaseControl studies only, and it’s a cross product ratio ضرب تبادلي…
Let’s take this table as an example:
Disease
+ve ve
+ve a b
Risk Factor (RF)
ve c d
RF +ve : exposed to the factor
RF –ve : nonexposed to the factor
Disease +ve : affected by the disease
Disease –ve : unaffected
Odds Ratio (OR) = (a X d) ÷ (b X c)
BUT…what about if you have a table with altered places of the positive and the negative values in relation to one another?? Something like this:
Disease
ve +ve
+ve a b
Risk Factor (RF)
ve c d
In the previous table, we’ve exchanged the places of the +ve and the –ve values of the Disease (the negative is first unlike the first example)… so here, will you divide the product of a and d by the product of b and c to find the Odds Ratio as in the example before?? …the answer is NO.
Actually, the solution is very simple, you know that Odds Ration = (# X #) ÷ (# X #) …. as a cross product. Now what is important to determine in order to find out the Odds Ratio in any table is the FIRST # (underlined in the equation), if you know it, then you’ll be able to determine all the others..
You can determine the FIRST # by knowing that it is the # of those who have +ve disease and +ve risk factor … you only look for it in the table whatever it was, and then multiply it with the opposing number, and finally divide the result by the product of the remaining two opposing numbers in the table. In our previous table, the FIRST # will be “b” …+ve with +ve … so Odds Ratio = (b X c) ÷ (a X d).
Remember that this is a CaseControl study.
Like in the relative risk, the Odds Ratio might be:
**equal to 1 : No association between the 2 variables
**greater than 1 : +ve association
**less than 1 : the factor is protective
But please remember that Odds ratio is used only in CaseControl studies while the relative risk is used in Cohort studies.
Now we’ll talk about the Attributable Risk % (AR %) :
AR% =( (IREX – IRNONEX) ÷ IREX ) X 100%
IREX : incidence Rate of the disease in the exposed group (who have the risk factor like smoking).
IRNONEX : incidence Rate of the disease in the nonexposed group (do not have the risk factor).
So obviously, to calculate AR% you need to know the incidence rate (IR)
Remember that you can obtain the IR from 2 types of studies: Cohort and Experimental studies.
For example, you might be asked this question: if you have a Crosssectional study carried out somewhere... Can you find out the AR%?? Absolutely NO, because Crosssectional studies cannot provide us with the (IR) that’s needed in the calculation of the AR%.. So we cannot calculate the AR% in crosssectional studies, while we can in Cohort and Experimental studies that provide us with the (IR).
Example:
A Cohort study or a clinical trial one of the 2 types of Exp. studies was carried out in a certain area, in this study, the number of the exposed group (let’s say smokers) = 200, while the number of the nonexposed (nonsmokers) = 500, the number of bronchitis cases(disease) in smokers=30, the number of bronchitis cases in the nonsmokers=40 …here we can create this table:
Bronchitis(disease) Total#
+ve ve
+ve 30 170 200 (all smokers)
Smoking(RF)
ve 40 460 5oo (all non smokers)
In the previous table, Smoking +ve : all smokers, smoking –ve : all nonsmokers, 30 of the smokers(+ve) were affected with bronchitis(disease +ve) so 170 of the smokers(+ve) were not affected (disease –ve), the same issue for the nonsmokers(smoking –ve), 40 were affected(disease +ve), so 460 remain from the the total 500…so 460 of the nonsmokers were not affected (disease –ve).if we assume that this is a Cohort study, then we can calculate the relative risk (RR), AR% and populational Attributable Risk (POP AR).
To calculate the AR%, you have to know the IR of the disease in the exposed (smokers) and the nonexposed(nonsmokers) groups.
IREX = (new cases of the disease ÷ total sample) X 100% = ( 30÷ 200) X100%
IRnonEx = (new cases of the disease ÷ total sample) X 100%=(40÷ 500) X100%
Note that the total sample in each equation means the total number of smokers and the total number of nonsmokers ..
Now you can calculate the AR%= ( (IREX – IRNONEX) ÷ IREX ) X 100%
Relative Risk (RR)= IREX ÷ IRnonEx
POP AR …to be able to calculate it, you should calculate the IR of the disease in the population in this study:
IRpop=TOTAL # of new cases ÷ TOTAL sample size=((30 +40)÷ (200+500))X K
SO..now we can easily calculate the POP AR% :
POP AR% = ((IRpop – IRnonEX(nonsmokers)) ÷ IRpop ) X 100%
By this… the lecture is finished, but there was a question asked by the end of this lecture and the prof. answered BUT not for the sake of the exam “as he mentioned”, SO, I’ll write it down here just for benefit:
The question was: what is the difference between Cohort and CaseControl studies??
The answer:
Cohort studies, are those studies carried out on a group of a population classified into Risk Factor +ve and Risk Factor –ve ..then these two groups are followed up in the future for a certain period of time..months, years,… to obtain the incidence rate (IR) in the exposed and the nonexposed groups.
For example, we divide the students in a class into smokers and nonsmokers, then we follow them up for 2 months for example, after that, we find out the number of bronchitis cases in the first group and in the second group, so we can find the incidence rate of bronchitis in the exposed and the nonexposed groups, relative risk, AR% and POP AR%...So in cohort studies we have FOLLOW UP.
In CaseControl studies, “Cases” means those who have the disease(disease +ve), “Controls” means the comparative group(disease –ve)
Disease
+ve ve
+ve a b
Risk Factor (RF)
ve c d
So… the column (a & c) is the “Cases”, and the column (b & d) is the “Controls”. We have a total # of bronchitis (cases) for example=200 (diseas +ve), and usually the controls (disease –ve) are equal in # to the Cases, so controls=200 as well.
So what we want is to create a table:
Disease(bronchitis)
+ve ve
+ve a b
Risk Factor(smoking)
ve c d
200 200
To fill the data, we take Case#1, asking: “were you smoking before the onset of the disease??”, then we ask the same Q to all of the cases(200 case)..we find out how many answered “yes” y3ny RF +ve for example =120, and how many answered “No” y3ny RF –ve =80
Then we do the same with the “controls” … But we don’t ask them the same previous question “were you smoking before the onset of the disease??”, simply because they do not have disease!!! So the Q will be: “Do you smoke??” and again we find out how many answered “yes” y3ny RF +ve for example =60, and how many answered “No” y3ny RF –ve =140
Disease(bronchitis)
+ve ve
+ve 120 60
Risk Factor(smoking)
ve 80 140
200 200
Then you can calculate the Odds Ratio as aforementioned… and that’s it.
A final note: about 99% of the studies carried out in Jordan and the region is crosssectional, actually they don’t do casecontrol because they don’t know how to do it, despite its simplicity!!!
GOOD LUCK
Hadeel Almasri
Monday
Apr 25th 2011
Dr.Farouq Shakhatreh
JU.De :: 3rd year :: Sheets and slides :: Biostatistics
Page 1 of 1
Permissions in this forum:
You cannot reply to topics in this forum

