2. Types of Epidemiological Research
• Descriptive.
• Analytical
• Experimental and clinical trials
• Meta-analysis.
3. Types of Analytical Epidemiological
Studies
• Retrospective studies.
• Prospective studies.
• Historical prospective studies.
• Cross sectional, prevalence or a survey
study.
4. Retrospective studies and /or Case
Control Studies
• In this kind of studies, the subject under the study have
the disease, and their past experiences are compared
with other persons who do not have this or related
disease.
• Selection of cases should consist of all newly diagnosed
cases with specified parameters under study during a
specified period of time.
• Controls are representative of the general population in
terms of probability of exposure to the risk factor under
the study.
5. Prospective studies and /or Cohort
• The philosophy of this approach is that exposed
subject in the investigation are representatives
of all exposed persons in regard with risk under
consideration.
• Healthy individuals , cohorts are allocated and
followed forward in time for development of
specific disease.
• Types of cohort :
Birth cohort, marriage , specific graduation.
6. Historical prospective studies.
• Include the follow up of healthy exposed
and unexposed subjects, cohort, for the
development of disease.
• However this cohort are allocated
retrospectively through medical records.
7. Cross sectional, prevalence or a
survey study.
• Both the risk factor and the disease are
examined at the same time .
• Temporality of the risk is not evident .
8. PAST PRESENT FUTURE
retrosp
control prosp
hist.
prosp
types of analytic studies
Select
cases
Look for
exposure
to risk factor
Select cohort
according
to exposure
Follow up
To record the
Disease
level
According to
Existing
Records, determine
Exposure in
the past
Identify
cohort
in the past
Development
Of disease
9. Research /service project model
• Type of study.
• Budget.
• Site, community.
• Date, start, close.
• Criteria ( inclusion and exclusion).
• Procedure plan.
11. Introduction for Biostatistics
The main goal to improve patient care
through more understanding of research
and to be critical thinkers, do study design
and do statistics.
Of course , you may need statistician to be
with you in advanced issues.
12. Population vs. samples
As a researcher , we are interested in
finding results that apply to entire
population of people or things. ( we cannot
collect data from every human being).
Therefore , we collect data from a small
subset of population ( known as sample).
13. Source of Data in Population
(Epidemiological data )
• Census.
• Vital statistics.
• Morbidity data.
• National health network.
14. Sampling
How to collect data that represent
population ??????>>>> reduce
the population to a statistical
model>>>>>so this statistical
model make predictions about the
real – world phenomenon.
16. Hypotheses
• A hypothesis is a proposed explanation for
the occurrence of a phenomena that a
researcher formulates prior to conducting
an experiment.
• Types of hypotheses.??????
• How to test your hypotheses.????
18. Null hypothesis, Vs. Alternative
hypothesis
If non directional ,
H0:µ or P = K ( i.e. mean has no diff. to a value).
HΑ:P # K ( i.e. mean is not equal to a value).
if directional
H0:P ≤ K
HΑ:P <K
H0:P ≥ K
HΑ:P >K
20. Testing Null Hypothesis
• N0:= hypothesis that there is no relation or
difference .
• If significant , P >0.05, reject N0, , i.e. false
hypothesis , there is a relation or there is a
difference.
• If non significant ,accept N0, i.e. true hypothesis,
there is no relation , there is no difference. ( it is
not no relation , but it is only statistically non
significant ).
21. Available data and hypothesis ,
what we will do ???
Statistical tools ,results to
discuss
22. Statistics
• A branch of applied mathematics
concerned with the collection and
interpretation of quantitative data , and the
use of probability theory to estimate
population parameters.
• Concerned with treatment of quantitative
information from groups of individuals.
23. What can statistics do?
• Provide objective criteria for evaluating
hypothesis.
• Synthesis of information.
• Help to detect the pattern of data
( descriptive statistics).
• Help to evaluate argument ( research
questions and hypothesis ).
24. Statistics Cannot?
• Tell the truth ( it can only give probability
only ).
• Compensate poor design.
• Indicate clinical significance.
25. Statistics don not Prove any thing
• Statistics suggest a relationship.
• In order to make conclusion you need :
- Multiple converging indicators.
- Multiple confirmatory studies.
- Temporal relationship.
- Dose response.
- Biological response.
- Biological plausibility ( reasoning).
26. Think in Research? How to
• Hypotheses and introduction.
• Is the research quantitative or qualitative .
• Collecting data . Sampling , prepare tools and survey
method.
• Preparing Data. Types of variables, Dependent,
Independent, Categorical ,Continuous.
• Data entry.
• Exploring Data. ( parametric, nonparametric ).
• Descriptive statistics.
• Inferential / analytical Statistics.
• Results, discussion , conclusion .
27. Types of Variables
string and numerical
• Qualitative:
- categorical,
- Nominal
- Usually independent
- Analyzed by
frequency table.
- Example?
• Quantitative
• Continuous
• Scale, ordinal,
• Usually dependent on
predictor
• Analyzed by
examining central
tendency (mean..etc.)
• examples
28. In Qualitative Research N.B:
• Prepare information as variables.
then
• Descriptive and analytical statistics.
29. Statistics & Population
• Descriptive : frequencies.
• Inferential .
• Periodic report : SWAT Analysis
Strengths, weakness, opportunities, threats
• Ratio and percentages.
30. Statistics & Population
• Incidence rate =
no. of new cases at point of time * 100 or 1000
population at risk
no. of new cases during a period of time * 100 or 1000
population at risk
31. Statistics & Population
• Incidence rate of rare disease =
no. of new cases during a period of time
population at mid of the year during this period of time
• Incidence rate in outbreak situation = attack rate
• Inception rate = new attacks of illness in
a population / year .( attacks may exceed the
number of population).
32. Statistics & Population
• Prevalence rate =
no. of existing cases at a point of time * 100 or 1000
total number of population
No of existing cases during a period of time * 100 or 1000
total number of population
Annual prevalence : total no. of disease at any time during a year.
Life time prevalence : total no. of individuals known to have the
disease at least part of their life time.
33. Statistics & Population
• Segmentation: Divide populations into
segments.
• Profiling: Develop profiles of hotspot segments.
• Drill-down: drill-down dimensions and numerical
value ranges.
• Variable selection: select variables used in
profiling and segmentation.
• Ranking: Order segments based ranking
criteria.
• Visualization: Visualize result statistics.
35. Statistics & Sample
• The sample can be summarized statistically by what is called
“mean”. The center of distribution of the scores.
• It is hypothetical value of typical score X . ????
• Sum of Deviances from the mean = total error = of course 0
• To be considered mathematically
Sum of squared error (SS) are done.
• To avoid the effect number of sample on the error, to estimate the
error in the population ----variance = SS
n-1 ?? df?
• SD+ = Square root of variance.
• SD, a measure of how well the mean represent the data.
• Small SD, indicates that the data points are close to the mean.
• Large SD, indicates that the data points are distant from the mean.
• Larger SD , i.e. that the mean is not accurate representation of the
data .
• Smaller SD, i.e. that the mean is of small fluctuation.
36.
37.
38. Statistics & Sample
• of course , in different sample the mean and SD , shows
that the sample is not in normal distribution .
• By z score , ( when any sample can be reformed to a
normal distribution , by making mean =0 and SD = 1 ) we
can calculate the probability, cumulative percentage of
any values in the data, and how the distribution.
• e.g when 95% z score lies between + 1.96
• Standard error.SE is SD of sample means. Small SE
indicates that most sample means are similar to the
population mean, and so our sample is likely to be an
accurate reflection of the population .
39. Statistics
Zscore: writing score
N Valid 200
Missing 0
Mean .0000000
Std. Error of Mean .07071068
Median .1292387
Std. Deviation 1.00000000
Minimum -2.29728E0
Maximum 1.50075
Percentiles
25 -7.9389478E-1
50 1.2923869E-1
75 7.6224449E-1
40.
41. Statistics & Sample
• Another way to think in the sample and
represent the data than mean is “linear
model “. It s the basic of ANOVA &
regression .
• Linear model is based on central tendency
and means .
42. Descriptive Statistics
• Method of organizing and summarizing
data in table , graph or numbers.
• Frequencies , %, cumulative %
• Mean , median ,mode
• SD , SE of mean. Level of confidence
• Skewness ,kurtosis , SE of skew , SE of
kurtosis . Parametric / non parametric
43. Inferential analysis
• It s a decision to choose the right way to do your analysis according
to :
1- parametric vs. non parametric.
2- level of confidence.
3- hypothesis
4- type of independent variable/s.
5- type of dependent variables/s
6- number of group / means .
7- related participants or not .(one or more)
8- repeated means .
8- difference , correlation, or regression.
9- reliability and validity for scales .
47. SPSS, training
• View : data/variable
• Creating data file
• Name of variable
• ID
• Abbreviation list
• Variable type, width, decimal, label,value
• Missing value
• Measurement.
• Entering variable.
48. SPSS, training
• Option
• Help : topic , tutorial , statistical coach
• Transform ( recode – compute variables)
• Analyze :
Frequency, descriptive, cross table,
compare means t test , GLM, correlation ,
regression, log linear , scale
nonparametric