RESEARCH

Prof. Dr. Sidhartani Zain, MSc, SpAK

Definition & purpose of research
 Research is a scientific investigation to discover new knowledge and
to test this knowledge.
 Purpose : to develop new knowledge and technology which can be
translated to skills and tools to improve the practice of the
profession.

New knowledge Skills

Research improve practice & Better health

delivery of services

New technology Tools

Type of research

 Basic research
 Epidemiological research
 Clinical research
 Research methods
 Pure basic research
 Applied research

Basic strategies are :

1. Description of the nature of events by observation of all members of the
defined population
2. Generation of hypothesis of relationship/association from analysis of
prevalence rates
3. Testing of hypothesis by comparison of frequencies of factors
4. Formulation of solutions/intervention based on previous studies
5. Testing of solutions by comparison of frequencies of outcome in
experimental and control groups
 Basic steps of research
I. Technical
 Identification and definition of the
problem  Implementation of plan
- Analysis of needs - Data collection
- Review of literature - Data processing
- Determination of the significance of - Data analysis
the problem
- Formulation of hypotheses and
categorization of variables
 Interpretation and conclusion
 Planning the research
the plan of investigation includes :
- Statement of objectives
- Study population and selection of  Reporting of study
subjects
- Research design
- Method of data collection
- Plan of data processing and analysis
 Basic steps of research
II. Administrative
 General preparations, includes :
 Feasibility study
- Procurement of funds
‘dry run’ to pretest the
- Purchase of materials, measuring devices and the
equipment, etc practicability of the
- Hiring and training of methodology.
personnel
- Scheduling of activities
 Termination of study
- Preparation of study manual
- Preparation/orientation of Relocating the personnel,
study area disposition of equipment,
supplies and records.
- Sampling of study group
 Research design

 It is a plan or course of action to solve the research problem.

 Basic research designs classified as descriptive or analytic

Descriptive

Basic research

Observational (cross sectional, case control, cohort)

Analytic
Experimental (true & quasi experimental)
Research design

Descriptive :
inquiry into the nature of unknown phenomenon or
occurrence of events.
It does not explain relationship, but seek knowledge
for basis of some future actions.
Research design

Observational : to test hypothesis of relationship

1. Cross sectional : relate prevalence of the independent
variable (cause) to the prevalence of the dependent var
(effect)
2. Case – control : compare the frequencies of the
independent var (cause) in those who have (affected) and
who do not have (unaffected) the dependent var.
3. Cohort : compare the occurrence of the dependent var
(effect) between those with and those without the
independent var (cause) under investigation
How is a research design planned & implemented ?

 Choice of study population

 Selection and classification or assignment of subjects
 Assesment or observation of variables
 Processing and analysis of data
 Descriptive study
1. To choose the population to be studied
2. To draw an adequate number of subjects
3. To collect data from each subjects

Diagram 1. Scheme of Descriptive study

Choice of population

Study population
Sampling

Subjects
Assesment of variables

Examined subjects
Categorization of subjects

Analysis Data
 Cross-sectional study
 Also called a survey or prevalence study

Diagram 2. Scheme of Cross-sectional study

Choice of population

Study population
Classification of subjects and
Assesment of variables

Without Without With factor With factor

factor & factor but but without and
outcome with outcome outcome outcome

Analysis of data
 Case-control study
 Requires two population (with the outcome & without the outcome)
 The Cases : individuals with the outcome or effect
 The controls : individuals without the outcome or effect
 Analysis : odds ratio/relative risk

Diagram 3. Schema of case-control study

Choice of population
Study population
Classification of subjects

Cases (with outcome) Control (free of outcome)

Assesment of factor

Without With Without

With factor
factor factor factor
Analysis data
Cohort study
 Starts from the suspected cause and looks forward at the outcome
 Relationship by comparing the frequency of the outcome in those exposed & unexposed
 Concurrent cohort study/prospective cohort study : two groups are followed for a given
period of time to identify members who will develop the outcome.
 Non-concurrent/retrospective cohort study : previously collected data may be used after
the population have been classified as having or not the factor.

Choice of population
Study population
Exclude those with
Classification of subjects outcome already

With factor Without factor

Detection of outcome

With Without With Without

outcome outcome outcome outcome
Analysis of data
 Experimental study
 The strongest of all study design
 Investigator’s artificial manipulation or introduction of the study factor
 Randomization of the study factors adds to the strength of the experiment.
Randomization equalizes the distribution of extraneus factors

Choice of population

Study population
Allocation of intervention

intervention No intervention

Detection of outcome

Without With Without

With outcome
outcome outcome outcome
Analysis of data
 Appropriate design selection

 If the objective is to know more about the phenomenon or event 

descriptive
 If the objective is to determine the prevalence of the outcome or
examine relationship between a factor and an outcome based on
prevalence data  cross-sectional study design
 If the objective is to test a hypothesis of relationship between
variables and not enough knowledge on the subject or the
outcome/effect is rare  case-control approach
 If the objective is to test a hypothesis and there are already enough
evidence that the hypothesis is likely to be true  cohort study
 If the objective is to evaluate the efficacy and/or effectiveness of an
intervention  experimental study
 Requirement to get a valid result

 Adequate number of subjects or observations

 Produce accurate and precise data

 Objectivity of observation
 Validity of measurement device
Characteristic of study design

 Descriptive study : the population needed is a target population from which a

representative sample must be drawn, otherwise the findings cannot be
generalized to the target population
 Cross-sectional study : the requirement are the same as those of a descriptive
study, hypothesis of relationship is generated from the result, the finding also
apply to the target population.
 Case-control study : estimates and compares the prevalence of the independent
variable (exposure) in those with the dependent variable (outcome) and in those
without the dependent variable. Mind the case & the control group. Confounding
variables should be controlled in the design or analysis .
 Cohort study : the frequencies of occurrence of the dependent variable (outcome)
in those with and in those without the independent variable (exposure). Requires
comparability of the exposed & unexposed in all respects except in exposure are
estimated.
 Experimental study : the principal requirement of an experiment is the same as
the cohort. Comparability of the experimental & control groups must be
achieved by scientific allocation.
 Sampling

Sampling techniques
 Sampling is a procedure of drawing a fraction or segment of a population for
the purpose of determining certain characteristic of the population or the
occurrence of some events in that population
 Various methods of sampling :

1. Random sampling : selecting a number of subjects (n units) from a population

or universe (N) such that every member has equal chance of being drawn into
the sample.
2. Systematic sampling : this is a method where every Kth unit (such every
fourth or fifth member) is selected from a population N which is arranged in
some definite way to obtain a sample of units.
3. Stratified sampling : in this method the population is divided into subpopulation or
strata. Samples from each stratum are selected either by random or systematic
sampling. The whole procedures are described as :
a). Stratified random sampling
b). Stratified systematic sampling
4. Cluster sampling : the sampling unit consist of group or cluster of similar unit. Instead
of choosing 1000 houses at random, 20 blocks of 50 houses are chosen.
5. Multi-stage sampling : for this purpose the population is divided into a set of first
stage sampling unit (primary sampling unit) and a sample of these units is selected.
Then each of these selected first-stage units is further divided into a second stage
units.
Sample size
Determination of sample size
 An adequate sample of the study population is essential in research. If the size is
inadequate the accuracy of the result is compromised.
 The researcher must also set the level of confidence for the study and the
maximum amount of deviation from the true value or the minimum difference
between parameters.
 Basic methods of determining sample size :

1. Problem dealing with counts

a). Descriptive study or one population problem
n = k2pq
d2
where : n = sample size
k = reliability coefficient based on the level of confidence
p = estimate of frequency of event
q = 1-p
d = maximum amount of deviation from the frequency
b). Analytic study comparing two frequency from different population
n = k22pq
d2
where : n and k = same as above
p = average of the estimates of the frequencies of the event in
the two population
q=1–p
d = minimum difference between the frequencies to be
detected
2. Problems dealing with measurements
a). Descriptive study or one population problem
n = k2SD2
d2
where : n and k = same as above
SD = estimate of standart deviation
d2 = maximum amount of deviation from true mean
b). Analytic study comparing two means from different population
n = k22SD2
d2
where : n and k = same as above
SD = estimate of standart error of the difference
d = minimum difference between means to be detected
RANDOMIZED CLINICAL TRIAL

Prof. Dr. Sidhartani Zain, MSc, SpAK

Clinical decisions on treatments

Balancing of the magnitude of the :

 Expected therapeutic benefit ( intended effects,

main effects)
 Adverse effects (unintended effects, side effects)

 Costs
 Outcome

 An (untoward) clinical event, over a fixed period of time

 Occurrence of the event can be verified for each patient
 Proportion of patients with event characterizes the occurrence
for the whole group
Primary outcome parameter

 One single parameter related to primary objective

 Identified in advance
 Specified in objective of the trial
 Related to study size calculation
 Determines positive/negative trial
A trial is a measurement

 Object of measurement
 Treatment effect

 Measurement errors

 Systematic errors (fatal)

 Random errors
Lack of systematic errors

 When index treatment is as effective as reference

treatment
 RR = 1
(except for random variation)
 Principles of trial design follow from the need to
avoid systematic errors

Lack of systematic errors = validity

 Comparability

 Extraneous factors
- Placebo effects
* Attention of the physician
* Psychology of patient
- Adjustment life style
- Co-medication
 Prognostic factors/natural history
- Risk profile of the patients
 Information on outcome
Techniques

 For comparability of extraneous factors ?

 For comparability of prognosis ?
 For comparability of information ?
Comparability of prognosis

Ordinary clinical circumstances :

 The physician assigns the more seriously afflicted

patients to the favored or most intensive treatment

 Perceived high risk or poor prognosis constitute an
indication for intervention
Principles of randomization

Intervention Outcome

X ?!
reason to initiate
of refrain from intervention
Extraneous factors

 What aspect of the index treatment is to be studied?

 Chemical substance
- Placebo treatment
 Procedure as a whole (strategy)
- Conventional treatment
Comparability of information

 Information bias (or detection bias) must be avoided

- Biased patient reporting
- Biased ascertainment of information by physician
- Biased assesment of information by physician, data
manager, statistician
 Solution : BLINDING
Importance of comparability of extraneous effects is
depending on the objective of a study :

 Interest in one solitary aspect of intervention

chemical compound (drug)
 explanatory
 Interest in procedure, eg. Surgery vs medical
treatment
 pragmatic
 Per protocol analysis

 Principle
- Outcomes are available in patients as long as they
are on treatment
- Outcomes are counted among patients on (study)
treatment
 Consequence
- Accords (intuitively) with scientific principles
- Leads to over-optimistic treatment effects
THANK YOU