The document provides an overview of basic statistics and research methodology, focusing on study designs. It discusses observational studies like cross-sectional, case-control and cohort studies as well as experimental studies like clinical trials. For each study design, it describes the key elements including temporal sequence, intervention, sampling methods, and how they differ from one another. It emphasizes the importance of selecting the appropriate study design based on the research question and highlights factors to consider like ability to determine causation, study of rare diseases, costs and time involved.

1. Basic Statistics (FK6163)
Principles of Research
Methodology
Study Designs
A presentation by
Assc. Prof. Dr Azmi Mohd Tamil,
Department of Community Health, Faculty of Medicine,
Universiti Kebangsaan Malaysia

2. Research
1. A systematic & organised scientific
process to find answers to the
questions.
2. Getting specific answers to specific
questions.
3. Involves data collection, analysis &
interpretation.

3. Research Process
Problem selection
Literature review
Formulation of research objectives
Selection of variables
Selection of the appropriate study
design
Sampling of population
Data collection
Analysis of data
Presentation of findings

4. Identifying the appropriate
study design
Research Methodology

5. STUDY DESIGNS
OBSERVATIONAL EXPERIMENTAL
STUDY STUDY
CLINICAL COMMUNITY
DESCRIPTIVE
STUDY TRIAL TRIAL
ECOLOGICAL
CORRELATION CROSS-SECTIONAL
CASE CONTROL
ANALYTIC COHORT
STUDY

6. Comparison between
different designs

7. The study design differs from
one another by;
Intervention – present/absent
Temporal sequence – when is the risk
factor and outcome measured;
– At the same time
– Risk factor before, outcome later
– Outcome first, then risk factor
(retrospectively)
Sampling methods

8. Difference Between
Study Designs
Study Design Intervention Temporal Sequence Sampling
Risk Factor and
Cross Sectional Absent Yes
Outcome at same time.
Outcome first, then risk
Case Control Absent Matching
factor (retrospectively).
Risk Factor first, then
Cohort Absent Maybe
Outcome.
Clinical/ Intervene, then measure
Present Randomisation
Community Trial Outcome.

9. SAMPLING
The process of selecting study subjects from
a larger population
extent to which research findings can be
generalised to a larger population
to save time, money, efficiency and safety.
PROBABLITY SAMPLING - equal chance to be
chosen ex. simple random, systematic, stratified,
multistage, cluster
NON-PROBABILITY SAMPLING - convenience,
quota, purposive.

10. POPULATION TO CHOOSE FOR
THE SURVEY
SELECT REPRESENTATIVE POPULATION
? sampling methods
- simple random sampling (may not be
practical in national study)
- stratified random sampling (in hetero/
stratum)
- multistage sampling (national-state-
district-sub district-village)
- cluster sampling

11. Which design is appropriate
for my study?

12. How to study the problems
or prove the hypothesis?
Select appropriate study design
- Descriptive - case report, case series,
ecological correlation
- cross-sectional / survey
- case-control
- cohort
- intervention - clinical trial/
community trial

13. Questions that need to be answered
?distribution of blindness - descriptive
?correlation between import of fruits
and visual acuity among the
population - ecological correlation
?prevalence of blindness, cataract or
poor vision - cross-sectional/survey
?association between vit A def. and
blindness - case-control

14. Questions that need to be answered
?incidence and relative risk of radiation
cataract among radiographers- cohort
?therapy/preventive methods useful or
effective , daily vit A supplementation
to prevent xeropthalmia - intervention

15. CROSS-SECTIONAL
STUDY
Also known as Prevalence Study
or Survey.

16. Cross-Sectional Study
Measures the relationship of variables
in a defined population at one
particular time
Both risk factors (exposure) and
disease outcome are observed at the
same (point in) time in a sample (or
the entire population) of subjects.

17. Exposure & Outcome
Exposure Outcome
Time
Confounders

18. Cross-Sectional Study
Risk Factor Outcome
Time
Confounders
Both Risk Factor & Outcome measured at the same time.

19. Cross-Sectional Study
Risk Factor -Race Outcome-Diabetes Mellitus
Disease + (14%)
Indians
(15%)
Disease - (86%)
Sample
ratio
Disease + (8%)
Others
(85%)
Disease - (92%)
Time
Both Risk Factor & Outcome measured at the same time.

20. RESEARCH QUESTIONS
What is the nature / magnitude of the
problem?
Who is affected?
How do the affected people behave?
What do they know, believe, think
about the problem?
We know very little about the
problems and its possible causes.

21. CROSS-SECTIONAL
SURVEY
MAY BE REPEATED - to
measure changes over time
LARGE SURVEY - limited
variables
SMALL SURVEY - unlimited

22. COMPARATIVE CROSS
SECTIONAL STUDY

23. COMPARATIVE CROSS-
SECTIONAL STUDY
AN ANALYTICAL STUDY
attempts to establish causes or risk
factors for certain problems e.g.
– obesity and IGT
– level of cholesterol and CHD
– betel leaves and NIDDM
– milk consumption and IDDM

24. Comparative Cross-
Sectional Study
Both risk factor(s) and outcome were
measured at the same point in time in
the selected sample or population.
The sample may have been selected to
represent the population being studied.
The selection or sampling method may
be random or not-random (refer to
sampling method notes).

25. COMPARATIVE CROSS-
SECTIONAL DESIGN
disease present
factor present
disease present
POPULATION factor absent
disease absent
factor present
disease absent
factor absent

26. HOW TO COMPARE?
PREVALENCE OF DISEASE IN
DIFFERENT SUBGROUP
– Rate of DM among obese vs rate of DM
among normal BMI
PRESENCE OF VARIABLES (OR
ABSENCE) IN DISEASE VS NON-
DISEASE
– Rate of contact with pigs among those
afflicted with Nipah virus against those
free from the disease.

27. COMPARE RATES
Disease Rate of disease
TOTAL among the
+ -
Exposure
exposed
+ A B A+B = A/(A+B)
- C D C+D Rate of disease
among the non-
TOTAL A+C B+D N exposed
= C/(C+D)
Disease Prevalence
=(A+C)/N

28. ADVANTAGES
cheaper, easier and faster
can study association
able to generalise findings to the
larger population
prevalence – for planning, measure burden
of disease, identify high risk population.
As a baseline for future cohort study

29. DISADVANTAGES
Can show association only but NOT
CAUSATION – no temporal sequence
survivors problems – only those who
still survive are studied, may miss the
contribution of those that already died
from the disease.
not suitable for rare diseases, or in
remission etc.
possible biases - selection,
misclassification

30. CASE-CONTROL STUDY

31. CASE CONTROL STUDY
- CONCEPT
comparison of group of diseased
person (cases) with another group of
non-diseased person (control) for past
exposure to a suspected cause of the
disease.
arises because of hypothesis that the
risk factor (exposure) causes the
disease

32. Case-Control Study
Exposure Outcome
Look back
in time
Time
Confounders
Starts with Outcome, then trace the retrospective exposure

33. Case-Control Study
Outcome-Cataract Risk Factor-Diabetes Mellitus
DM + (50%)
Cataract
Sample DM - (50%)
ratio
(1:1) DM + (8%)
Normal
vision
DM - (92%)
Time Past
Starts with Outcome, then trace the retrospective exposure

34. CASE CONTROL STUDY
FREQUENCY CASE WITH
OF THE DISEASE
PAST
EXPOSURE Assessment
TO THE of exposure
SUSPECTED CONTROL WITHOUT
CAUSE IN DISEASE
EACH
GROUP
Past Time

35. CASE SELECTION
Determine clear and reproducible
definitions of the health problems to be
studied (avoid misclassification bias)
source of cases
All persons with the disease seen in
particular facility(ies) in a specified period
of time.
All persons with the disease found in
general population.
Incidence cases (newly diagnosed cases)
preferred

36. CHOICE OF CONTROLS
Controls should ideally be selected from the
same population gave rise to cases
Similar to cases in regard to past potential
exposure
Free from study disease
If controls are patient with other diseases
then select only diseases that are not known
to have relationship with factors under
study.

37. MATCHING
to take account for potentially
confounding variables
type of matching :
• Frequency matching : selection of controls
with the same proportionate distribution of
the particular variable as the cases. (eg. age
and sex)
•Individual matching : pairing the controls
with the cases on some common variables
such as age, sex and ethnic group.
• Matched variables cannot be evaluated

38. ASSESSMENT OF EXPOSURE
Exposure should be assessed by the
same method for both cases and
controls
blinding
methods of assessment :
• available records - hospital, vital, employment
•interview
•self-administered questionnaire
•direct measurement
• Comparability of the accuracy and completeness
of data

39. UNMATCHED CASE-
CONTROL STUDY
OUTCOME
CASE CONTROL
R
I Exposed a b
S
K
F
A
C Not
T
O
exposed c d
R
Analysed using Chi-Square &
Odds Ratio = ad/bc

40. MATCHED PAIR CASE-
CONTROL STUDY
CASES
Exposed Not exposed
C
O
Exposed a b
N (both pairs exposed) ( pairs of controls exposed)
T
R Not
O exposed c d
L (pairs of cases exposed) (both pairs not exposed)
Analysed using McNemar and Odds Ratio = c / b

41. ADVANTAGES
able to study rare diseases
can explore multiple exposures
relatively inexpensive
can calculate Odds Ratio
can support causation but not prove it
easy to get cases

42. DISADVANTAGES
Affected by biases such as selection
bias, information bias, recall bias.
Temporal relationship may not be
clear.
Inefficient for rare exposure
Can study only one outcome at a time
Cannot measure prevalence or
incidence.

43. COHORT STUDY

44. COHORT STUDY
BASIC CONCEPT
Group or groups of individuals are
studied over time as to onset of new
cases of disease and factors
associated with the onset of disease.
Synonyms : incidence study,
longitudinal study, prospective
study.

45. Cohort Study
Exposure Outcome
Look
forward
in time
Time
Confounders
Starts with Risk Factor, then measure the future Outcome

46. COHORT STUDY
FOLLOW-UP
DISEASE
EXPOSED GROUP
NONDISEASE
Free from
disease
DISEASE
UNEXPOSED GROUP
NONDISEASE

47. Cohort Study
Risk Factor-Weight Outcome-Diabetes Mellitus
DM + (32%)
Overweight
Free Sample DM - (68%)
from ratio
DM (1:1) DM + (7%)
Normal
DM - (93%)
Time Future
Starts with Risk Factor, then measure the future Outcome

48. THE PURPOSE OF
COHORT STUDY
To identify risk factors for disease
to identify protective factors against
disease
to identify prognostic factors for
outcome of disease
to describe the natural history of disease
to determine the number of new cases in
a population over time for :
planning acute care services
assessing effectiveness of preventive
measures

49. TYPE OF COHORT STUDY
PROSPECTIVE COHORT STUDY:
the cohort assembled at the start of the
study, followed over time (into the
future) to determine the incidence of
disease
HISTORICAL COHORT STUDY:
the cohort assembled in terms of a
particular exposure in the past and are
followed through existing records into
the future.

50. CLASSICAL VS DYNAMIC
COHORT
CLASSICAL OR CASE-NONCASE
- cohorts are counted by person
contribution overtime, the
denominator for incidence rate is
population at risk.
DYNAMIC OR POPULATION TIME
cohorts are counted by person-time
contribution overtime, the
denominator for incidence density is
person-time at risk

51. Recruitment
Those recruited must be free from the
disease of interest at the beginning of
the study.
Those with sub-clinical presentations of
the disease may miss from being
excluded. This is one of the challenges.

52. SELECTION OF COHORTS
SOURCES :
- geographically defined groups
- special resource groups -doctors,
nurses, factory workers etc.
- special exposure group - expose to
radioactive, asbestos, benzene, haze
COMPARISON GROUPS :
- similar in all respects except exposure

53. Cohort Definition
Both groups (E+ & E-) must have equal
chance of being followed up.
Types of cohort;
– Representative – low exposed subjects
– Enriched – high exposed subjects
– Specific group – occupational, institutional
etc.

54. ASSESSMENT OF
EXPOSURE AND OUTCOME
INFORMATION ON EXPOSURE:
records, cohort members, medical
examination and measures of the
environment
INFORMATION ON OUTCOME
periodic reexamination
surveillance of deaths, hospitalization,
clinic visits

55. Follow-up
Keep participation at > 90%
Must have equal ability to detect disease in all
subjects and all groups, with standard
measurement
Active vs Passive follow-up
Verbal Autopsy
Blinding of the assessor
Assess both primary and secondary outcomes

56. COHORT STUDY
Relative risk (risk ratio)
A measure of how many times more likely
are exposed persons to get the disease
relative to non-exposed
Useful for causative associations
Size of RR does not necessarily indicate
magnitude of incidence rates in exposed
and non-exposed groups
Compares the relative contributions of risk
factors no matter how much of the disease
exists

57. Cohort Study: Basic Steps
Total population/sample
No disease (a) Disease (b)
No exposure (c) Exposure (d) Exclude
Disease (e) No disease (f) Disease (g) No disease (h)
Relative risk = Ie/Ie' = (g/d)/(e/c)
ranges from zero (strong negative
association) to infinity (strong positive
association)

58. Relative Risk
Cohort Study
OUTCOME
Disease + Disease -
R
I Exposed a b
S
K
F
A
C Not
T
O
exposed c d
R
Relative Risk = (a/(a+b))/(c/(c+d))

59. Advantages
Able to show causation
Able to measure risk
May represent population
If the exposure occurs rarely, then
cohort is a good way to study the
outcome.
Can directly measure incidence

60. Disadvantages
High cost, longer time & relatively
more difficult to execute.
Many subjects lost to follow-up.
Control subjects may end up being
exposed i.e. start smoking
Not suitable for rare diseases
Biases – more scrutiny of the exposed
group.

61. Clinical Trials

62. CLINICAL TRIAL
Any prospective controlled assessment of :
1. A treatment method
2. Diagnostic technique
3. Preventive measures
Characteristics
PROSPECTIVE – cohort or group of patients,
followed over a period of time, evaluation of
outcome.
CONTROLLED – 2 or more groups of patients. All
groups are comparable to one another with respect to
all factors relevant to the outcome

63. Clinical Trial
Intervention Outcome
Look
forward
in time
Time
Confounders
Starts with Intervention, then measure the future Outcome

65. Clinical Trial
Intervention Outcome-Improved?
Improved (75%)
Fluoxetine
Sample No improvement (25%)
Depressed
patients ratio
(1:1) Improved (70%)
Sertraline
No improvement (30%)
Time Future
Starts with Intervention, then measure the future Outcome

66. EXPERIMENTAL STUDY
STUDY POPULATION
SELECTION
ELIGIBLE
PARTICIPANTS NOT ELIGIBLE
PARTICIPANTS NON-PARTICIPANTS
RANDOMISATION
TREATMENT CONTROL

67. EXPERIMENTAL STUDY
C+
T+
C-
Study Eligible
Population Participants
C+
Selection T-
Randomisation C-
Future

68. SELECTION OF SUBJECTS
Number, sources
Inclusion criteria : age range, sex, weight,
diagnostic criteria, informed consent,
cooperation
Exclusion: lack of inclusion criteria,
pregnancy or lactation, drug allergy, disease
severity, other disorders, requirement of
other drugs, unresponsive cases, mental
status.

69. RANDOMISATION
Method for inducing comparability between
groups
Ensures that characteristics known or
unknown did not influence the treatment
assigned
Treatment given to the patient/next patient
to enter the trial is determined purely by
chance, not by any characteristics of the
patient

70. RANDOMISATION
Allows the computation of the probability
that the groups differ for both known and
unknown characteristics.
Does not guarantee perfect comparability
Better than any known procedure
Each patient enter the trial has an equal
chance of receiving which ever therapy

71. COMPARISON GROUP
To allow the evaluation of the outcomes
of interest in a comparable group of
patients who received a standard or
best available relevant alternative
treatment
The effect of treatment are compared to
the effects of a control treatment

72. COMPARABILITY
The patients should not differ in any
characteristics, known or unknown,
relevant to the outcomes of interest
other than the treatment employed.

73. Parallel vs Crossover
Start
B
Start
A B A
A B
B A
End
A B End
B A

74. Crossover Design
Disease under study must be chronic
and treatable but not curable
Disease must be stable over course of
study
Upon analysis, must consider;
Period effect
Carryover/residual effect

75. ADMINISTRATION OF STUDY
Institutional review – ethical committee
Informed consent
Receipt, distribution and storage of
investigational supplies
Instructions to recorders
Instructions to subjects
Adverse reaction reporting
Monitoring
Reporting results

76. PROCEDURE
Assignment of subjects to treatments
Interview and examinations
Methods of assessment
Lab. Studies
Treatment schedules: number of units per
visit, rules fo changing dosage, compliance
checks
Adverse reactions: definition and grading,
inquiry, management.
Drops out: definition, handling and
recording, terminating and extending study

77. BLINDING
Single blind : either the patient or the
physician knows the treatment which
has been assigned
Double blind: neither the patients nor
the physician are aware of the treatment
assigned
Triple blind: even the statistician not
aware of the assignment.

78. FOLLOW-UP SCHEDULE
The schedule of visits
The duration of follow-up
Measurement and procedures to be
conducted at each visits
In multi-centre trial, the methodology
of all measurements and procedures
should be specified thoroughly.

80. EXPERIMENTAL STUDY
ADVANTAGES DISADVANTAGES
the best design to Exposed to biases
determine causal – selection,
association and – attrition,
evaluate program – compliance and
performance – contamination
biases.
ethical implications

81. COMMUNITY TRIALS
1. All prophylactic vaccines, such as those against
measles and rubella (german measles), diphtheria, and
polio, were tested on populations of children to prove their
efficacy in preventing the diseases.
2. Prophylaxis with drugs such as penicillin to prevent
episodes of rheumatic fever or isoniazide hydrochloride
(INH) in the prevention of active tuberculosis.
3. Antihypertensive drugs for the reduction of blood
pressure and prevention of complications such as stroke.
They were proven effective in clinical trial experiments.
4. Testing various forms of health service delivery, such
as comparing family practitioner services with physician
specialty services.
5. Health effects of radiation following the atom bomb
explosions of 1945, the famine in Africa of 1974, of loss of
jobs in the recession of 1980.

82. TERIMA
KASIH