RESEARCH METHODS

FOR RESIDENT PHYSICIANS

MODULE ONE
RESEARCH PROTOCOL
DEVELOPMENT

NOEL L. ESPALLARDO, MD
CLINICAL ASSISTANT PROFESSOR
DEPARTMENT OF FAMILY AND COMMUNITY MEDICINE
UNIVERSITY OF THE PHILIPPINES COLLEGE OF MEDICINE
 LECTURE I
RESEARCH IDEAS, RESEARCH QUESTION AND
RESEARCH OBJECTIVES
Preparing for your research project

RESEARCH IDEAS
One conducts research because of pure interest, or it is a career or it is a requirement for
promotion, training or even recognition. Whatever is the purpose of conducting a research one
has to start with a good research idea.

Sources of Research Ideas

Experience
An expert researcher will always have his experience as the best source of new research
idea. Unfortunately for those who are still in residency or are just trying to start a research project
experience may not be enough. Other source of ideas should also be explored.
Another source of idea is the medical literature. They may be in an electronic data base
like the MEDLINE, INTERNET or publications like medical journals in the library. These sources
should be scanned very well in order to get a good and relevant research idea. It is therefore
necessary for a serious researcher to have an adequate knowledge and skill in browsing through
this medical literature.
Morbidity and mortality statistics in your area of practice can also be a good source of
research idea. This will surely be relevant in your setting.

Be alert to new ideas

New ideas can also be taken from scientific gatherings, discussion with friends or
colleagues or experts in your area of practice. Make sure that every time you attend a scientific
convention or meeting, try to pick up new ideas or problem areas.
Every now and then you will also be assigned to report or discuss a clinical case in your
group discussions. In most cases they are part of your training program. This exercise can also be
a source of research ideas.
Oftentimes a research idea can pop out of the blue anywhere or anytime and then
disappear. It is therefore advisable that when it occur write it down in any piece of paper or
notebook that can be retrieved when needed.

Keep your imagination roaming.

A skeptical attitude to current practice will stimulate your mind to be creative and
imaginative. Considering the increasing cost of treatment, there is always a need for alternative
and equally effective but less expensive modalities.

Choose a good mentor.

If after going through the sources listed above you still don’t have a good research idea,
you still have one last alternative. Ask your consultant!

2
Focusing Research Ideas
Many research ideas do not materialize because they are vague or too broad. Although a
research idea may have one primary problem and a couple of secondary problems, it is advisable
that for a beginning researcher focusing on just one research problem is enough. There is no
formula on how to focus the research idea into a research question. It all depends on the interest
of the researcher. The examples below can serve as a guide on how to focus the research idea.

A young researcher wanted to study better ways to improve the success of TB treatment.
There are so many ways that TB treatment can be improved. So the researcher focused in
improving compliance because it is a grave problem among TB patients. The researcher further
considered that decreasing the frequency of medication with slight dose modification or directly
observing the intake of medication will improve compliance. In this scenario however, two
factors are being considered i.e. decreasing frequency and observing therapy that may have a
confounding effect on compliance. To avoid complications the researcher eventually chose to
study the effect of directly observed therapy on treatment compliance.

The example above started from a general idea of improving success of TB treatment to a
more focused research idea of investigating the effect of directly observed treatment to
compliance of TB treatment.

Another researcher wanted to investigate alternative ways of improving the control of

hypertension other than or in addition to pharmacological treatment. Recent guidelines suggest
that lifestyle modification and risk factor modification should always be considered as an
additional intervention. However, encouraging the patients to adhere to lifestyle modification is
often difficult. The researcher thought that if someone else probably a family member can
encourage and remind the patient constantly, adherence will be improved. The researcher later
decided to investigate the effectiveness of educating the patient and a family member in
controlling the blood pressure and modifying the risk factors. The research idea is now more
focused on the effect of educating the patient and the relative.

RESEARCH QUESTION

The research question is a long statement intended to focus the research project. It is
usually written in the latter part of the introduction. In a protocol it should be stated in such a
way that it can be answered by a yes or no. It usually embodies the research objectives.

Components of a Good Research Question

A good research question should contain at least all of the following:
1. Biologic or theoretical rationale.
2. Specific population to be studied.
3. Specific test intervention.
4. Specific primary outcome of interest.
5. Specific comparative intervention if a comparative study is planned.
6. Suggestion of a study design.

3
 Below is an example of a research question on the effectiveness of a surrogate observer in
the treatment of tuberculosis.

“Among patients consulting at the local health center and diagnosed to have PTB
randomized to a clinical trial, will compliance to TB treatment be improved by encouraging a
family member to be a surrogate observer to drug intake compared with usual clinic advise?”

Biologic rationale yes, family members can encourage drug intake thereby
improving compliance
Specific population patients diagnosed to have PTB in a community health center
Specific test intervention use of surrogate observer for drug intake
Specific primary outcome of improving compliance to TB chemotherapy
interest
Specific comparative usual clinic advise
intervention
Suggestion of a study design randomized clinical trial

Another example is given below.

“Among the newly diagnosed hypertensive patients consulting in an outpatient clinic,

will compliance to lifestyle modification and control of hypertnesion be improved by
educating the patient and family member about lifestyle modification than just educating the
patient alone in a randomized clinical trial design?”
Criteria for a Good Research Question

The criteria for a god research question is best remembered by the mnemonics F-I-N-E-S-
T (which stands for F-feasible, I-interesting, N-novel, E-ethical, S-significance and T-time bound).
Thus in undertaking a research project, the research question must at least fulfill these criteria.

Feasibility
Before you waste your time and effort in a particular research plan, first make sure that
the undertaking is feasible. Feasibility can be affected by several factors, but the most prominent
may be the cost of the project and the availability of study population. It is extremely difficult to
conduct a study that needs P 1 million in funding if the available amount is only P 1 hundred
thousand. Likewise it is difficult to study a disease that is very rare in a population. It will take a
whole lifetime before you can adequately accumulate your cohort.

Interesting
This may depend on whose perspective is the research question being evaluated. It is
helpful to view the interest first from the investigator, second from the patient and third from the
health care provider.

Novel
A good research always contribute new information.

Ethical
A research project will definitely not pass the institutional review board if the study is
not ethical. A study therefore must not place the participants in unnecessary risk or deny them of
the necessary benefits. It should also not infringe on their privacy.

4
 Significance
Just like interest, significance may be viewed from a different perspective. Significance
may also be evaluated in the same manner as interest.

Time Bound
It is very important that when you undertake a study, it should be finished at least within
your lifetime (or within your training, or within your deadline).

To make sure that your research question is a good one, apply the checklist as shown in
Table 1.

Table 1 Checklist for a Good Research Question

Yes No Not Sure
Is it feasible?
Do I have enough subjects?
Do I have enough funding?
Is it interesting?
For the investigator?
For the patient?
For the health care provider?
Is it new?
Had the medical literature been searched?
Had the experts been consulted?
Are there any controversy that need answers?
Is it ethical?
Is there any risk for the subjects?
Will the study violate their privacy?
Is the study significant?
For the investigator?
For the patient?
For the health care provider?
Can the study be completed within a given period?

Deciding to Undertake the Research Project

After subjecting your research question to the checklist for a good research question it
may be helpful to ask the following questions below before embarking on your study.

Elaborate significance

1. Is there a scientific rationale?

The answer to this question may be implied or you can design your theoretical
framework based on your previous readings.

5
2. Is enough known about the problem?

To answer this question a thorough search of the medical literature should first be done.
An electronic database search is usually more efficient than a hand search of printed journals in
the medical library.

3. Is the objective directed toward the improvement in health care?

4. Is it a priority problem?

If the answers to the questions above are all yes, then your study is significant. The next
question to answer is “Is it feasible?” To answer this first make a general plan of your study.
Follow the outline suggested in Table 2. Then ask the three questions below.

1. Is the proposed methodology feasible?

2. Can I finish it in time?

3. Can I afford the cost of the study or is funding available?

If the answers to the questions above are all yes, congratulations you’re now ready to
prepare your research protocol.

After formulating the research question and study plan a few problems might arise and
this might lead you to discard the research project. However there are solutions I would like to
suggest before discarding your research idea. First is consult Table 3 and adopt the suggested
solutions for the problems. Then present your study plan to you friends, colleagues and mentors
for comments and suggestions. This is the iterative process of forming a research proposal. Lastly
rely on your creativity, judgment and tenacity.

No amount of problem will be left unsolved to a determined investigator.

6
Table 2 The Research Plan
Title

Investigator/s

Research Question

Objectives

Study Design

Study Population
Inclusion Criteria
Exclusion Criteria
Sample Size

Intervention
Experimental
Control

Outcomes

Procedures for
Measurement of
Outcomes

Statistical Analysis

Total
Budget/Schedule/Perso
nnel

7
Table 3 Problems and Solution for a Difficult Research Question or
Research Plan
Problems Solutions
Vague or inappropriate Write the research question at an early stage.
research plan Get specific in the study plan about:
how the subjects will be sampled
how the outcomes will be measured
Think of ways on how to make:
the subjects more representative of the population
the measurements more representative of the phenomena of
interest
Not Feasible
Too broad Specify a smaller set of outcomes or variables
Narrow the question

Not enough subjects Expand the inclusion criteria

available Eliminate exclusion criteria
Add other sources of subjects, multi-center
Lengthen the time frame for entry into the study
Use more efficient variables or design

Methods inadequate or Consult experts or review literature for alternative methods

beyond the skills of the Learn the skills
investigator Collaborate with colleagues who have the skills

Too expensive Consider less costly study design and measurement methods
Seek additional financial support
Decrease sample size?
Not interesting, novel or Modify the research question
significant
Uncertain ethical Consult institutional review board
considerations Modify research question

RESEARCH GOALS AND OBJECTIVES

After formulating your research question, the next step is to formulate your research
objectives from the research question. The aims, goals, or objectives are the building blocks of a
research proposal. They provide a picture of what you plan to accomplish in your research
project.

The characteristics of a good research objective is also best remembered by the

mnemonics S-M-A-R-T (which stands for S-specific, M-measurable, A-attainable, R-realistic and
T-time bound). This criteria should be applied to both the general and specific objectives.

8
General Objective

Usually expressed in a broad health care issue or statement. The statement should
include the disease it wants to study, the aspect of disease that the project intends to change and
the setting where the change will occur.

For example, a researcher wanted to investigate the use of quality of life as an outcome
for the treatment of hypertensive patients in family practice, the general objective may be stated
as:
“The general objective of this study is to improve the care and treatment of hypertensive patients
in family practice.”

Disease to be studied Hypertension

Aspect of disease to be changed Care and treatment
Setting Family practice

This general objective can also apply to a study that test a new anti-hypertensive
medication that may offer better BP control , improved survival and lower side effects.

Avoid incomplete statements like these:

“Improve the treatment of hypertensive patients.”
“Investigate the quality of life of hypertensive patients.”

Another researcher who is interested in trying to validate the modified Prime MD, a
screening instrument to detect psychological problems in family practice may state the general
objective as:
“The purpose of this study is to improve the screening program for psychological problems in
family practice.”

Disease to be studied Hypertension

Aspect of disease to be changed Care and treatment
Setting Family practice

Avoid these:
“To validate the modified Prime MD.”
“To use the Prime MD as a screening tool in primary practice.”

In the example of the study on directly observed therapy the general objective can be
stated as:
“This study aims to improve the treatment of tuberculosis in the community setting.”

In the example of the effect of education on the patient and family members, the general
objective can be stated as follows:
“The general objective of this study is to improve the treatment of hypertensive patients.”

Specific Objectives

The specific objective involve a specific clinical question that is usually embodied in the
research question. Some author write their specific research objectives based on the methodologic
strategy like describing the population, measuring the outcome and analyzing the outcome as
shown below.

9
 Specific objectives for the study on TB compliance:
1. To diagnose patients with PTB based on chest x-ray or sputum examination.
2. To educate the patient or the patients family to act as surrogate observer for drug intake.
3. To measure compliance of drug intake in the two groups.
4. To compare drug compliance between the two groups.

Although the above specific objectives can be summarized into like;

1. To educate TB patients and their relatives to act as surrogate obsrver for drug intake.
2. To compare compliance to drug intake between those with surrogate observer and no
surrogate observer.

It can also be stated as a single specific objective like;

1. To determine the effect of surrogate observer on compliance to treatment of TB patients.

There is no definite rule on how to formulate the specific objectives but it should follow
the SMART criteria. Special attention however should be given to the action verb used in the
statement of objective. The action verb must be clear and specific. The choice of the verb may
actually be the basis of the statement fulfilling the SMART criteria. For example the action verb
“to know” may be more appropriately stated as “to determine”.

10
 LECTURE II
METHODOLOGY I: OVERVIEW OF STUDY DESIGNS

PRIMARY STUDY DESIGNS

Non-comparative or Descriptive studies

The word comparative here is used as a method of distinguishing between two
populations at the start of the study. These studies are designed simply to describe certain
characteristics of a problem. Cause and effect relationship is not being answered by this design.
The importance of this design is that it can be a source to generate hypothesis that can serve as a
topic for future research using more complicated designs. A typical example of this design is the
cross sectional study design.

Cross sectional studies

The essential feature of this type of design is that the cause and the outcome are
measured at the same point time. There is no attempt to establish a temporal relationship
between the cause and effect. They are basically descriptive and try to establish prevalence.
Example of cross-sectional studies would be a study to describe the prevalence of
different diseases in population, or sex distribution among hemophiliacs or prevalence of use of
tobacco among cancer patients.
Data is gathered using a structured questionnaire or data extraction form.

General steps in doing a cross-sectional study:

1. Select a sample from the population
2. Collect data using a standardized data collection method
3. Analyze data

Comparative Studies
These study designs compare the presence or absence of a cause or an outcome
characteristics between two groups. It is a better design to establish cause and effect relationship.

Observational Studies
These studies are designed to attempt to define the relationship between the outcome
and its causes (cause and effect relationship). The outcome can be a development of a disease or
cure of a disease and the cause can be a risk factor such as genetic predisposition, an
environmental exposure or unhealthy behavior like smoking, sedentary lifestyle etc. In this
designs the researcher do not manipulate the exposure of the subject to the cause but only
observe for their presence or absence. For example the researcher do not decide on the
environmental exposure, or who should have sedentary lifestyle or smoke. Manipulating them
may be difficult or even unethical especially when the outcome being observed is potentially
harmful.

There are two types i.e. case-control studies and cohort studies. They are described
below.

11
Case-control Studies
In a case-control study, inclusion of subjects starts with defining or selecting those who
have the outcome or effect. These are considered as cases. Then this group is compared with
subjects who don’t have the outcome or effect. These are considered as the controls. Both the
cases and the control should be taken from within the same population. Then the two groups are
investigated as to the presence or absence of hypothesized causes or risk factors for the outcome.

Case-control study can be prospective or retrospective depending on the manner of

patient recruitment. If recruitment is being done as cases develop forward in time it is
prospective, but if the cases have already developed in the past and patient recruitment is being
done by reviewing existing clinical records then it is retrospective.

General steps in doing a case-control study:

1. Identify cases in a certain population
2. Identify controls from the same population matched to the cases based on certain
characteristics
3. Collect data from both cases and controls
4. Analyze data

Example in the literature (Shapiro et al. Oral contraceptive use in relation to myocardial
infarction. Lancet 1979; 1:743-747.)
A group of researcher examined the relationship between oral contraceptive and
myocardial infarction. They selected 234 women who develop myocardial infarction (cases) and
1,742 women who did not have myocardial infarction (control). Then they were all questioned
about history of oral contraceptive use. The results are shown below.

MI No MI
OC use 29 135
Non OC user 205 1,607
Total 234 1,742

A history of OC use was reported in 12.4% of cases and 7.7% of controls. We might say
that there is an association between OC use and MI because if there is none the proportion of MI
cases should be the same for both groups. The validity of the association however is dependenet
on the manner of population sampling or selection and determination of exposure.

Cohort Studies
A cohort is any group of individuals who share the same characteristics. In a cohort
study, selection of subjects start with identifying individuals who have the same characteristics or
presence or absence of a particular cause or exposure. They are then divided into two groups,
those with the characteristics or causes and those without the characteristics. They are then
observed forward in time and determine who among them develop the outcome or effect.

Cohort studies can also be prospective or retrospective depending on the manner of

patient recruitment. If recruitment is being done forward in time it is prospective, but if the
cohort already existed in the past and data gathering is being done by reviewing existing clinical
records then it is retrospective.

12
 General steps in doing a cohort study:
1. Identify a group or cohort who have the exposure or characteristics of interest in a
certain population.
2. Identify a group or cohort who don’t have the exposure or characteristics of interest.
3. Observe these two groups forward in time for the outcome of interest (data collection).
4. Analyze the data

Example in the literature

The Framingham study is the longest cohort study to date. In 1948 the investigators took
a random sample of 5,209 men and women from the general population in Framingham,
Massachusetts. Baseline characteristics were determined and re-examined every two years. The
result of one of the sub-study is shown below.

Alive at 20 years Dead at 20 years % Dead

Non smokers 819 132 16.1
Smokers 1,489 333 22
Total 2,317 465 20.1

Thg results show that 22.2% of cigarette smokers died compared to only 16.1% of non-
smokers within 20 years of follow-up. If there is no association between smoking and early
mortality the observed proportion should be the same in both groups. Thus smoking increases
mortality in the general population. The validity of this conclusion also depend on the selection
of the population and determination of exposure and outcome.

Experimental Studies
In this studies, there is manipulation in the exposure to the cause to establish its relation
with the outcome. Manipulation may be ethical when the outcome being studied is potentially
beneficial, for example testing a new drug that has advantage over the old one. It is therefore not
ethical to do experiments in humans if the experimenter wants to see the effect of different
radiation exposure to radiology technicians.
True experiments are randomized controlled trials. Non-randomized or uncontrolled
trials are considered quasi-experimental design.

Quasi-experimental Studies
Quasi-experimental designs are non-randomized, or non-comparative studies that
involve observation of the effect of a particular intervention. In a non-randomized comparative
study, the investigator choose two groups and assign each group to the two type of intervention
by convenience. The outcome is then compared between the two groups. In a non-comparative
study the investigator administer an intervention to a group of patients and watch the effect
before and after the study. The outcome is then described before and after the intervention. Some
authors consider non-comparative studies as case series, others as before-and-after design, but for
the purpose of the classification we have adopted, we will label them as quasi-experimental
study.

Example (Villamangca, D. Effects of Patient Education on the Quality of Life of Patients with
Bronchial Asthma, Dept Internal Medicine, Rizal Medical Center)
An investigator assembled a group fo asthmatic patients and enrolled them to an asthma
education program. The program consists of a series of group lectures and interactive discussions
patterned after the comprehensive asthma education program of the Philippine General Hospital.

13
At the same time he also assembled a group of asthmatic patients and educate them in the
outaptient clinic. He measured the quality of life of both groups before and after the educational
intervention. The results of AQLQ after the educational intervention are shown below.

Education Program Usual Education

AQLQ overall 4.68 3.85
Symptoms 4.44 3.58
Activity 4.94 4.11
Emotional 4.77 3.58
Environment 4.6 4.29

The results showed higher scores for the intervention group but the difference were not
statistically significant. Further analysis showed that baseline AQLQ in the usual education
group was really low from the start compared with the intervention group. This is the usual
problem with a non-randomized design.

Randomized Controlled Trial

This is the strongest design of all study designs. If done properly the result will surely be
of highest validity and reliability. In this type of design, individuals are randomly assigned
(randomization) to two or more groups, one with the exposure, intervention or cause the other
without the exposure, intervention or the cause. Randomization try to make the two groups
similar for both known and unknown factors that may affect the outcome other than the
exposure, intervention or cause being tested. Then they are observed forward in time and their
outcome compared. The outcome can be the cure of a disease, relief of symptoms or
improvement in quality of life.

In most but not all cases blinding is done. Blinding is the process in which the subjects,
investigator and other personnel in the study is not made aware of the type of intervention the
subject is recieving. This will be discussed in more detail later.

General steps in doing a clinical trial:

1. Select a sample from a population
2. Measure baseline variables
3. Randomize into intervention groups
4. Apply intervention
5. Measure outcome in both groups
6. Analyze data

Example in the literature

The TRACE Study was designed to determine improvement of survival among patients
with LV dysfunction after MI. Patients were randomized to receive either trandolapril 1-4 mg for
two years or to placebo aside from their usual medications. Deaths and their causes were noted in
both groups during the follow-up period of two years. At the end of the study the results are
presented below.

14
 Trandolapril Placebo
Overall deaths 304 (34.6%) 369 (42.3%)
Cardiovascular deaths 226 (25.8%) 288 (33.0%)
Non-cardiovascular deaths
78 (8.9%) 81 (9.3%)

The results of the study showed that the all-cause mortality, cardiovascular and non-
cardiovascular mortality were all lower in the trandolapril group. Trandolarpil therefore improve
survival among patients who have LV dysfunction and post-MI. The validity of this conclusion
depends on the selection and randomization of patients, manner of intervention and
determination of outcome.

SECONDARY STUDY DESIGNS

Meta-analysis
A meta-analysis is a procedure that integrates and combine the results of two or more
primary studies that are similar in the population enrolled the intervention used and the outcome
measured. The pooled result is then subjected to a statistical analysis. A well conducted meta-
analysis allow a more objective appraisal of the existing evidence about a problem than a
traditional review or overview. It may also be biased owing to the inclusion or exclusion of some
irrelevant or relevant studies respectively.

General steps in doing meta-analysis

1.Formulate the objective of the meta-analysis
2. Formulate the collection of data, data to be included and excluded
3. Collect and pool the data
4. Analyze the data

Economic Analysis
Economic analysis can be defined as an analysis that uses analytic techniques of from
primary studies to define the choices in resource allocation.
In this design the cost of a particular intervention is estimated. Estimation include direct and
indirect costs. There are three types of economic analysis depending on the type of outcome. If
the outcome being considered is effectiveness of treatment, it is called cost-effectiveness analysis.
If the outcome is savings in terms of monitary units it is called cost-benefit analysis. If the
outcomes are equal and the cost is the only one being compared it is called cost minimization.

CHOOSING A STUDY DESIGN

The choice of a study design entirely depend on your research question. If you want to
describe characteristics of an interesting population, you might want to choose a cross-sectional
survey. If you want to establish a cause and effect relationship you might do an observational
study or if you want to compare effectiveness of two interventions then you can do an

15
experimental study. The algorithm in Figure 1 might be very helpful in deciding what research
design to use.

Are you going to compare

two or more groups?

Yes No
Are you studying causation, Are you describing a
prognosis, treatment or diagnostic single case or groups of cases?
tests or just compare
clinical characteristics?

Single case Groups of cases

Case Report Case Series
Cross-sectional
Studies

Causation, prognosis, Diagnosis, compare

Treatment clinical characteristics
Are you going to recruit Cross-sectional Studies
subjects forward in time?

Yes No
Are you going to assign Are you going to recruit
treatment or exosure? those with exposure or
those with outcome?

Exposure Outcome
Retrospective Cohort Retrospective Case-
control

Yes No
Will you randomize? Are you going to recruit
those with exposure or
those with outcome?

Yes No Exposure Outcome

Randomized Quasi-experimental Prospective Cohort Prospective Case-
Controlled Trial control

Figure 1 Algorithm for Choosing a Research Methodology

16
 WORKSHOP I

Objectives of the workshop

At the end of the workshop the participant should be able to:
1. Identify a problem in his/her area of practice.
2. Formulate a research question and research plan for his/her idea.
3. Decide which of his/her research idea to undertake as his/her research project.

1. Enumerate at least 3 research ideas you are interested in.

1.

2.

3.

2. Focus your research ideas.

1.

2.

3.

3. Prioritize your research ideas.

1.

2.

3.

4. Translate the top research idea into research question.

1.

5. Formulate the general and specific objectives of your research question.

1. General

Specific

17
 LECTURE III
METHODOLOGY II: STUDY POPULATION,
RANDOMIZATION, AND SAMPLE SIZE ESTIMATION

STUDY POPULATION
The larger group to which the study results are to be generalized is called the target
population. Thus if one is studying a new drug for the treatment of hypertension, the target
population is all the hypertensive patients in the world. It is however impossible to study all of
them thus we only choose a representative group. This is called the study or sample population.

The study population must be defined in the early stage of the study. It should be
appropriate enough to attain the objective of the study. For the study population to be clearly
defined, we should be guided by the following questions:
1. What or who should be the study population?
2. When and where should the study population be recruited?
3. How should the study population be selected?

What and who should be the study population

The exact characteristics of the study population must be defined. As much as possible it
should correspond to the characteristics of the target population. Defining the population is
usually stated as an inclusion and exclusion criteria. Table 4 illustrates how to formulate the
inclusion and exclusion criteria.

When and where the study population be recruited

A statement indicating when and where the study population will be recruited is
oftentimes necessary. A study on the treatment of hypertension may be recruited as stated:
“Patients consulting for hypertension at the Family Medicine Clinic of the Philippine General
Hospital between the period January 1 to December 31, 1998 will be recruited for the study”.

How should the population be selected

The method of selecting the study population is called sampling. It can be categorized as
probability sampling or non-probability sampling. Probability sampling is made through a
process called random sampling where every individual in the population has the same chance of
being included in the study. Random sampling should be distinguished from randomization
which will be discussed later. Non-probability sampling is made by non-random methods. This is
often used in clinical research because of the difficulty of identifying the target population.

Probability Sampling
Random sampling is a process whereby each unit in the population has the same
probability of being chosen – chance alone will decide which of the unit will be included. The
steps in doing random sampling is illustrated in Table 5.

18
Table 4 Formulating Inclusion and Exclusion Criteria

Inclusion Criteria Specifying the characteristics A trial of calcium supplementation

(should be specific) that define populations that are for preventing osteoporosis might
relevant to the research specify that the subjects should be:
question.

Target population Demographic characteristics Females, age 45-50

Clinical characteristics In good health with no known life

threatening illness, not previously
diagnosed to have osteoporosis and
no history of neurologic deficit or
taking corticosteroid

Accessible population Geographic Patients attending the medical clinic

at the investigator hospital

Temporal characteristics Between January 1 and December

31, 1998
Exclusion criteria Specify subsets of the The calcium supplementation trial
(be parsimonious) population that will not be might exclude subjects who are:
studied because of:

A high likelihood of being lost Alcoholic or plan to move out of the

to follow-up city or country

Inability to provide good data Disoriented or have a language

barrier

Ethical barriers Kidney stone formers

(contraindicates oral calcium)

The subject’s refusal to Unwilling to accept random

participate allocation to placebo

Table 5 Steps in Doing Random Sampling

Step 1 Prepare the sampling frame, usually a list of the target or accessible population
Step 2 Decide on the size of the sample
Step 3 Get a table of random numbers, arbitrarily select a starting point then get a series of
random numbers equal to the sample size. The numbers in the series that correspond
to the list are the subjects of the study. This process is called simple random sampling.
Another way is by systematic random sampling where the total number in the list is
divided by the sample size to get the sampling interval. Randomly select a starting
point in the list and include the subject in the list at every sampling interval. This is
equal to simple random sampling as long as the list is not arranged in a particular
order.

19
Special Types of Random Sampling
Stratified random sampling – the accessible population is first divided into non-
overlapping strata such as range of age, sex, economic status etc. In each strata subjects are
randomly selected either by simple or systematic random sampling.

Disproportional sampling – if different strata have different size, sampling can be done
whose size is proportional to the size of the strata that may lead to different sizes between
different strata.

Cluster sampling – sampling is done from similar units of the population like 5 clusters
of provinces in the Philippines. This can also be done at several stages or multi-stage cluster
sampling. For example we want to study the prevalence of TB in the Philippines and we need
1,500 barangays, we can start with a cluster of 30 provinces, a cluster of 5 towns in each province
then a cluster of 10 barangays in each town selected.

Non-probability Sampling
The most common non-probability sampling is convenience sampling. In this method
subjects are chosen based on their availability i. e. during the time of consultation, those who
respond to the announcement or those who are near the study center etc. Purposive sampling is
another common method wherein a researcher selects a certain subject because they fulfill some
specific criteria. In snowball sampling, a subject who was already included are asked to identify
others who also have the same requisite characteristics. Stratified non-probability sampling can
also be done by quota sampling where volunteers are called to join the study and stop
recruitment in each strata once the proper size is achieved.

RANDOMIZATION
As previously mentioned the randomized controlled trial is the design considered as the
“gold standard” in clinical research. Randomization is an essential feature of such design. It is
defined as the process in which each subject is given the same chance of being assigned to the
different study groups. The purpose is to make the groups comparable with respect to known
and unknown variables that might affect the outcome of the study.

Several methods of randomization is available, but in this module we will only discuss
randomization with fixed allocation. This means that the randomization will not be alter as the
study progress.

Simple Randomization
The simplest method of randomization is by a toss of a coin. If the coin turns up heads
the subject is assigned to A, if the coin turns up tails the subject is assigned to B.

Another simple randomization procedure can be done using the table of random
numbers. The steps in doing randomization using the table of random numbers are described in
Table 6.

20
Table 6 Steps for Randomization Using the Table of Random Numbers

Step 1 Decide what group to assign odd numbers and even numbers and how to generate a
series of random numbers in series or in columns
Step 2 Get a table of random numbers and point with a pencil where to start while being
blindfolded (another way is to use a series of seed numbers)
Step 3 From the starting point get the sequence of random numbers equivalent to the
expected sample size in the study
Step 4 Assign the groupings whether the number is even or odd

The use of computer generated randomization can also be done using the software called
RALLOC.

Simple randomization is easy to implement but may result to unequal sample sizes in the
study groups. This problem can be solved by blocked randomization.

Block Randomization
Blocked randomization is used to avoid imbalance in the number of subjects assigned to
each group. If 20 subjects will be randomized to two groups using simple randomization might
result to 12 subjects being assign to one group and 8 to the other. In blocked randomization each
block, say 4, is designed to have equal number of A and B by enumerating the possible
combinations of 2 A’s and 2 B’s (AABB, ABAB, ABBA, BAAB, BABA, BBAA). The combinations
are then selected at random until all 20 subjects are randomized (see Table 7).

Table 7 Block Randomization of 20 Subjects to Two Groups in Blocks of 4

Step 1 Number the following combinations of A and B in blocks of four as shown below.
1 2 3 4 5 6
A A A B B B
A B B A A B
B A B A B A
B B A B A A
Step 2 Randomly select five sequence of numbers from 1 to 6, say 3, 5, 1, 6, 1.
Step 3 The randomization therefore is as follows
Subj Assign Subj Assign Subj Assign Subj Assign Subj Assign
1 A 5 B 9 A 13 B 17 A
2 B 6 A 10 A 14 B 18 A
3 B 7 B 11 B 15 A 19 B
4 A 8 A 12 B 16 A 20 B

Stratified Randomization
In stratified randomization, prognostic factors that may affect the outcome such as age,
severity of illness etc. are identified. They are then divided into different strata such as less than
20 as the first strata, 20-50 as the second strata, and greater than 50 as the last strata. The subjects
in each strata are then randomized (simple or blocked) to their group assignments.

21
 SAMPLE SIZE ESTIMATION
The formula to be used for estimating the sample size may depend on the type of the
study and the outcome to be measured. For the purpose of simplification, we will discuss
computation primarily based on the type of outcome being measured and whether we are just
describing a single population or comparing two populations. Only the formulas and the needed
data to compute for the sample size will be presented.

Tables 8 and 9 are values of Za and Zb.

Table 8 Values for (Za + Zb)2

Alpha = 0.05
Beta 1-tailed 2-tailed
0.5 2.71 3.84
0.2 6.18 7.85
0.1 8.56 10.51

Table 9 Values of Za at Different Confidence Levels

Confidence Level
0.80 0.90 0.95 0.99
Za 1.28 1.64 1.96 2.58
Note Value of Zb at 0.10 is 1.28

Sample size estimation to describe a single population, rate or proportion as the outcome
measure:

N = Za2 PQ
d2

Where:
N – sample size needed
Za – value of Z in normal distribution at desired alpha level
P – estimated rate or proportion of the population with the outcome
Q – estimated rate or proportion of the population without the outcome
or (1-P)
d – maximum tolerable error

Example
A community physician wanted to estimate the prevalence of TB among schoolchildren
in a certain community. Previous study showed that the prevalence was 20%. He wanted to
estimate the present prevalence with a desired precision of 5% and confidence level of 95%.

N = (1.96)2 (.2)(.8)
(.05)2

N = 246

22
Sample size estimation to describe a single population with mean as the outcome measure:

N = Za2 S2
d2

Where:
N – sample size needed
Za – value of Z in normal distribution at desired alpha level
S – standard deviation of the mean of variable being studied
d – maximum tolerable error

Example:
An obstetrician wanted to determine the mean hemoglobin of pregnant patients
consulting at the local health center during their first pre-natal visit. She wants an estimate of at
least within 2 units of the true value of hemoglobin with a confidence of 90%. Previous survey
showed the standard deviation of hemoglobin among pregnant patients during their first pre-
natal visit was 10 grams %.

N = (1.645)2(10)2
(2)2

N = 68

Sample size estimation to compare two populations with rate or proportion as the outcome
measure:

N = Za2 2PQ + Zb2 (P1Q1+P2Q2)

(d)2

Where:
N – sample size needed per group
Za – value of Z in normal distribution at desired alpha level
P – estimated mean rate or proportion in the two population with the
outcome or (P1/2+P2/2)
Q – estimated mean rate or proportion of the population without the
outcome or (1-P)
P1 – estimated rate or proportion in the first population with the outcome
Q1 – estimated rate or proportion in the first population without the
outcome or (1-P1)
P2 – estimated rate or proportion in the second population with the
outcome
Q2 – estimated rate or proportion in the second population without the
outcome or (1-P2)
d – maximum tolerable error

Example:
A researcher wants to determine the response rates of patients with CHF taking diuretics
alone compared with trandolapril plus diuretic. The investigator want to detect a 20%
improvement in response from the 60% who respond favorably to diuretics alone based on
previous studies. He wants to detect this at the level of significance of 0.05 (1-tailed) and a power
of 90%.

23
Computation (exercise)

Sample size estimation to compare two populations with mean as the outcome measure:

N = 2(Za+Zb)2 S2
d2

Where:
N – sample size needed per group
Za – value of Z in normal distribution at desired alpha level
Zb – Valuae of Z in normal distribution at the desired beta level
S – standard deviation of the mean of variable being studied
d – maximum tolerable error

Example:
An investigator wants to determine the effectiveness of a new anti-asthma agent
(Asmalin inhaler) using improvement in PEFR as the outcome. Previous study showed that the
variation in PEFR measurement in the ER was 12 L/min. He wants to detect a difference of at
least 5 L/min between the new drug and the old standard at alpha of 0.05 and a power of 90%.

Computation (exercise)

24
 WORKSHOP II

Objectives of the workshop

At the end of this workshop the participant should have:
1. Written the contents of the methodology section (Subjects, Subject selection,
Randomization, Intervention) of his protocol by completing the answers to all the questions
below.

1. Based on your research question and objectives, describe in general your planned
methodology.

Describe the population you plan to include in your study.

Inclusion criteria

A. Exclusion criteria

2. Describe how you plan to recruit your subjects.

25
3. Compute for your sample size.

4. If randomization will be done, describe how you will randomize your subjects.

5. Describe the exposure, treatment or diagnostic tests you plan to observe or compare.

Experimental

Control

26
 LECTURE IV
METHODOLOGY III: DATA COLLECTION AND
ANALYSIS

DATA COLLECTION
This section will discuss the process of collecting data. The lecture is divided into
subsections that are important in collecting accurate and reliable data.

Data Collection Forms

Before data collection the researcher must review the objectives of the study and the
design. Consultation with the statistician to design data collection forms that is adapted for the
planned statistical analysis will always be helpful. Before designing data collection forms it might
be wiser to use existing data collection forms that have withstood pre-testing and modifications.

Investigators should take special care in developing these forms so that data are complete
and accurate. Before designing the form, the investigator should make an outline of all the data
that are to be collected. Each page in the data collection form should contain an identifying mark
for each subject. The data must be organized (modular) for entry into the computer software that
is planned to be used. Modular structure means separating the data collection form into sections
like demography, history, physical examination, laboratory testing, test drugs, other outcomes to
be monitored (quality of life), adverse events etc. Having such ready made data collection forms
improves efficiency and cost of reproduction.

Some rules in designing data collection forms:

1. collect data pertinent to the study objectives
2. be simple, easily understood, short as possible and yet comprehensive
3. use forms that can be completed with a check mark rather than requiring fill-in
4. implement easy to use automation system for data entry, editing and analysis

I hope you’ll find helpful the sample data collection forms in Appendix I.

Blinding
Bias can occur at different stages of the research project. It can occur at the time of
assignment to treatment groups. Bias in this case is reduced by randomization. It can also occur
during data collection or observation. Bias in the second situation is reduced by blinding.
Blinding is the process of concealing the treatment or intervention to the patient or investigator.
As much as possible clinical trials should be double blind, meaning both the investigator and the
patient is not aware of the treatment. But if it is impossible as in surgical trials, a single blind
approach or other methods to reduce bias like concealing the intervention to the evaluator or the
statistician can be done. Table 10 enumerates the types of blinding you can use.

27
Table 10 Types of Blinding

Open label Both the subject, the investigator and other members of the research team
know the treatment assignment
Single blind Only the subject is not aware of the treatment assignment
Double blind Both the subject and the investigator is not aware of the treatment
assignment
Triple blind This is just an extension or modification of the double blind design, where
aside from the investigator and subject, other persons involved in the
study like the monitoring team, data encoders and statisticians are also
not aware of the treatment assignment

Making a Questionnaire
A questionnaire is a measurement instrument for assessing individual’s attitudes, beliefs,
behavior or attributes. It is important that the researcher should establish first hand what he
wants to measure. Measuring attitude is often difficult because they are very sensitive to
variations in words. Attributes on the other hand is less sensitive to wording variation. Before
constructing your questionnaire be clear about its purpose and the information you want to get.
Make sure that a questionnaire is the best possible method to get that information.

A questionnaire can be open-ended or close-ended. Close-ended questions limit the

responses and the information to be collected to the available choices. The advantage is that
coding and analyses are often easy. To avoid losing information it is best for close-ended
questions to include all possible answers. If this is not possible include “others and specify” as
one of the choices. Open-ended questions allow the researcher to obtain greater information
especially on attitudes and opinions. But coding and analyses are often difficult. The most
appropriate questionnaire will probably be a combination of both open and close-ended
questions.

When constructing your questionnaire make sure that your grammar is correct and the
wordings are simple. It should be targeted to the lo est educational level of potential
respondents. Sequence questions into logical order and group them into topics to make the
respondents’ task pleasurable. Keep the questionnaire as short as possible. After constructing the
questionnaire give it to your peers for comments. Pre-test it to a sample of possible respondents.
When changes are made pre-test the revision until you have decided on the trial questionnaire.

If you don’t intend to do these don’t gather your data by means of a questionnaire! Table
11 presents the listing of problems associated with use of questionnaires.

28
Table 11 Common Problems in Questionnaire Construction and Use

Lack of attention by the The construction of a good questionnaire requires a high level
researcher of experience and skill and is not to be undertaken lightly
without extensive planning and consultation.
Ethics The respondents’ privacy and dignity should be respected.
Respondent characteristics Factors like age, sex, literacy and educational level may
influence questionnaire results and their reliability.
Response error Response error may arise from failure of memory, motivation,
communication and knowledge. In other words respondents
may not answer correctly and accurately.
Response bias Social desirability bias is present when patients try not to
offend other people or try to respond to behavior questions that
are socially desirable when it may not be the real behavior of
the respondent.
Return rate Low return rates often affect the reliability of the study.
Asking threatening or personal Threatening questions may cause the respondents to be
questions embarrassed and feel uncomfortable. This may also affect the
accuracy of the responses.

Measurement of Study Variables

Measurement is defined as the process of assigning numerals to objects to represent
quantities of their characteristics according to certain rules. We assign numerals to responses to a
questionnaire like “0” for a no answer and “1” for a yes answer. We can also assign numerals to
severity of symptom or illness like “0” for none, “1” for mild, “2” for moderate and “3” for
severe. In some cases numeral assignment is very evident like for height and weight, age, blood
pressure, temperature etc. This assignment must follow certain rules that must be set by the
researcher. Thus a pediatric researcher might want to measure age in months but an adult
medicine researcher might want it measured in years. Some might measure weight in pounds
while others in kilograms. These assignment of numbers and rules must be explicitly set by the
investigator.

Measurement is used in research to describe the quality or quantity of an existing

variable i. e. the characteristics of what we want to observe. For example, if we want to analyze
demographic characteristics in terms of age we measure it in number of years, if in terms of
education we measure it in educational level (0 for none, 1 for elementary, 2 for high school, 3 for
college etc.), if in terms of economic status, we measure it in terms of annual income.
Measurement can also help us make clinical decisions when we measure and compare
effectiveness of two alternative drugs or diagnostic tests. Finally we use measurement in research
to help us draw some conclusion about the relation between two variables. This relation may be
an association or difference.

29
Levels of Measurement
At the start of the study the investigator decides on what outcome to observe and how it
will be measured or recorded.
There four different levels of measurement namely nominal, ordinal, interval and ratio.
Each level needs a different kind of analysis.
1. Nominal data – these are qualitative data which are mutually exclusive and
exhaustive and do not mean hierarchy. Numerals represent category or classification
labels only. For example name, sex (male does not mean being better than female),
address etc.
2. Ordinal data – reflects a rank order among the categories used to measure a variable,
example is economic class (low, middle, upper), scale of severity (none, mild,
moderate, severe, very severe).
3. Interval measures – have more meaning than ordinal measures because the
differences between two categories are known and definite. The intervals between
numbers are equal but not related to true zero. They do not represent true quantity.
For example temperature, calendar years etc.
4. Ratio scale – Numbers represent units with equal intervals measured from true zero.
For example distance, weight, blood pressure etc.

Coding
Coding is the process of assigning numbers to answers or data collected for data entry
and analysis. It must be differentiated from measurement where the purpose is to quantify
characteristics. In pre-coding the data or response is coded before collection. This is applicable if
the responses are known in advance and the questions are close-ended. It is more efficient
because it makes data entry and analysis easier. However it may not be applicable to open-ended
questions. In open-ended and complex questions, post-coding is usually done. In post-coding
numbers are assign to answer categories after they have been collected.

Analysis
Statistical analysis should be planned before data collection. Since the type of data to be
collected or measurement to be done is already planned prior to collection, statistical analysis can
be planned in advance as well. The analysis must be simple as much as possible. Complex
analysis might be impressive but interpretation becomes difficult. Consult a statistician (familiar
with clinical research) for advice on the appropriate statistical analysis for your data. Make use of
existing computer packages to analyze data.

General classification of statistical techniques commonly used in clinical research.

1. Correlation coefficients – a measure of the strength of relationship between two
variables
2. Linear regression – given that a strong relationship exists between two variables, this
technique allows the prediction of one variable from a given values of another
variable
3. T-tests – compares the variance (spread of scores) of values obtained for two
variables – indicates whether the two samples or variables are significantly different
4. Analysis of variance – a more complex form of the t-test, where the variation in one
variable is broken down and attributed to other variables selected for the study

30
 5. Factor analysis – a method of statistically grouping items which are answered by the
respondents in similar ways

We will not go farther in discussing how to compute for these statistical techniques. Try
to find the algorithm in Figures 2 to 9 helpful in deciding what statistical techniques you will use
in analyzing your data.

Sample Population

Difference Association

Measure the size Testing for Measure of the Testing for Extent
of the difference statistical degree of the statistical association
See Figure 3 significance of association significance of explains
the difference See Figure 4 the association variations
See Figure 5 between groups
See Figure 6

Nominal Data Ordinal Data Continuous Data

See Figure 7 See Figure 8 See Figure 9

Figure 2 Starting Point When Deciding What Statistical Techniques to Use

Measuring the size of the difference

Nominal or Ordinal Data Continuous Data

Descriptive Statistics Descriptive Statistics

Figure 3 Measuring the Size of the Difference

Measure degree of association

Nominal Data Ordinal Data or when no Continuous data when a linear

Odds Ratio or linear relationship is relationship is suspected
Relative Risk suspected Pearson’s Correlation
Spearman’s Rho or Kendall’s Coefficient
Tau

Figure 4 Measurement of Degree of Association

31
 Testing for statistical significance of association

Nominal Data Ordinal data or when no Continuous data or when a

Statistical Significance of linear relationship is linear relationship is
Odds Ratio or Relative Risk suspected suspected
Statistical Significance of Statistical Significance of
Spearman’s Rho or Kendall’s Pearson’s Correlation
Tau Coefficient

Figure 5 Testing for Significance of Association

Extent of association explains variation between groups

Nominal data Ordinal data or when no Continuous data or when

Attributable Risk linear relationship is linear is suspected
suspected Pearson’s Coefficient of
Spearman’s Rho2 or Kendall’s Determination (R2)
Tau2

Figure 6 Extent of Association Explains Variation Between Groups

Nominal Data

Small unmatched Small matched Large unmatched Large matched

sample sample sample sample
Fisher’s Exact Test Sign Test Chi-square with Mac Nemar’s Test
Yates Correction

Figure 7 Testing for Significance of Difference Between Two Nominal Data

Ordinal Data

1 comparison >1 comparison

(2 groups) (>2 groups)

Unmatched sample Matched sample Unmatched sample Matched sample

Mann-Whitney U or Wilcoxon Matched Kruskal-Wallis one- Friedman two-way
Median Test Pairs or Signed way Analysis of Analysis of Variance
Ranks Tests Variance

Figure 8 Testing for Significance of Difference Between Two Ordinal Data

32
 Continuous data

1 comparison >1 comparison

(2 groups) (>2 groups)

Unmatched sample Matched sample Unmatched sample Matched sample

Student’s Unpaired Student’s Paired T- F-test for Analysis of F-test for Analysis of
T-test test Variance followed by Variance with
Pair-wise comparison Blocking or Analysis
of co-variance

Figure 9 Testing for Significance of Difference Between

Two Continuous Data

The Consent Form

Informed consent is a very important factor for justifying your research. Thus it is very
important that the consent form be designed according to ethical standards (amended Helsinki
declaration). The form should be written in understandable language and contain the following:
1. A statement of the purpose of the study
2. Description of procedures both experimental and routine
3. Duration of the subject’s involvement in the study
4. Whom to contact in terms of adverse events or additional questions
5. Risks and discomfort associated with participation in the study
6. Alternative appropriate treatment available in place of experimental treatment
7. Benefits the subject may expect from participation in the study
8. A statement that participation is voluntary
9. A statement guaranteeing confidentiality
10. A statement regarding compensation due to adverse events

33
 LECTURE V
WRITING THE RESEARCH PROPOSAL
The initial stages and often the most difficult part of the research process is the
development of the research proposal. The proposal describes the purpose of the study, the
importance of the research question, the methodology and justifies the feasibility of the project.

The proposal serves several purposes:

1. represents the synthesis of the researcher’s critical thinking and the scientific
literature to ensure that the research question is refined enough to be studied
2. serves as an application for review by peers, administrative committees or funding
agencies
3. enhance communication among colleagues who may be co-investigators
4. serves as guide throughout the study to ensure that the researchers follow the
outlined rules of conducting the study

General Rules in Writing the Proposal

These are the basic grammar guidelines in writing a research proposal.

Word Choice
The words chosen should be simple, precise, necessary and familiar. Highly technical
and scientific terms should be used less often and only when necessary i.e. no other familiar
word with the same meaning. Avoid jargon or inventing new words by adding suffixes or
prefixes to familiar words. Use few abbreviations as least as possible.

Wrong word choice Suggestions

Renal blood flow is drastically compromised if greatly reduced or reduced by more than 50%
the aorta is obstructed.
The change in current produced by M major increase
protein was greater than 85%of the maximal
response to isoproterenol.
Infusion of serotonin was associated with an resulted to or led to
increase in microvascular pressure.
After 4 hours of hemodialysis, we abruptly of hemodialysis and abruptly can be removed
ended the hemodialysis procedure. i.e. After 4 hours, we ended the hemodialysis
proceedure.
Heat stable materials will be utilized in the used
isolation and processing of samples.
We endorphinized the dogs. injected endorphins to

Sentence Structure
Use simple and direct sentences. This can be done if the core or the message is conveyed
in a simple sentence structure i.e. subject, verb and predicate or completer. The topic should be in
the subject and the action in the verb.
Another common mistake is the piling of nouns into noun cluster or putting too many
ideas in one sentence. A sentence should only talk about one thing at a time. Aim for a mean
sentence length of no more than 20 words per sentence.

34
 Wrong sentence Suggestions
The children with arteriovenous shunts had The shunts of children arteriovenous shunts were
their shunts opened and heparin injected. opened and heparin was injected
The new drug caused a decrease in heart rate. The new drug decreased heart rate.

Noun clusters Suggestions

filament length variability variability of filament length
peripheral chemoreceptor stimulation stimulation of peripheral stimulation

Paragraph Structure
The paragraph should convey an organized idea and the continuity of these ideas must
be clear. To do this a paragraph should have a definite structure. It should be started with a topic
sentence followed by a series of logically arranged supporting sentences.

Sample Discussion
(A) There are three different theories put A – topic sentence
forward for the very slow relaxation of catch
muscles of molluscs. (B) One theory holds that B – States first theory
catch is due to some unusual property of
myosin in these muscles that produce a slow
rate of detachment. (C) In this theory C – Explain first theory
paramyosin would have no special role beyond
that of providing the long scaffolding on which
the myosin is positioned as well the mechanical
strength for the large tension developed. (D)
The second theory holds that tension is D – States second theory
developed by actinomysin interaction but is
maintained by paramyosin interactions. (E)
Because the thick filaments are of limited E – Explain second theory
length, interactions have to occur through
fusion of thick filaments. (F) A third theory, to F – States third theory
which I subscribe, pictures a structural change
in paramyosin core affecting the rate of
breaking of myosin-actin links at the filament
surface.

35
Outline of the Clinical Research Protocol
This should serve as a guide to the researcher in writing his/her research protocol. The
protocol should contain the following parts:

1. Title of the protocol – contains key words found in the study objectives or the research
question. The title describes the main idea of your project. It serves to orient the reviewer as
to what he or she is about to read. The title should be not so brief that it says nothing or
leaves some things hanging, but it should not be so long that a reviewer has to think hard to
figure out what it means.
2. List of investigators – name of investigators (main investigator first and not the most senior),
their institution and addresses. Limit authors who will actively participate in the study from
planning, conduct of the study, analysis and writing the final paper.
3. Protocol abstract – should not exceed 300 words, single spaced and following standard
structure as shown below:
Objectives – state the main question of the study
Study design – describe design as to the appropriate use of randomization, blinding and
temporal direction
Setting – indicate the study setting (hospital, clinic, community)
Patients/participants – state selection procedures, entry criteria and numbers of
participants entering and completing the study
Interventions – describe the essential features of any intervention including their method
and duration of administration
Main outcome measures – primary outcome measures should be indicated as planned
before data collection begins
Data analysis – state sample size and considerations for its calculation and exact level of
significance. State specific statistical methods to test difference or association
Conclusion – state expected primary conclusions according to study objective along with
the clinical application
4. Introduction
Scientific significance
The introduction should be able to supply background information, rationale
and purpose of the study. It should include a critically appraised and up-to-date review
of literature. Justify that despite the literature cited, there is still a need to answer the
research question and conduct the study.
Research Question
Goals and Objectives
A goal is a general statement that reflects what will be accomplished at the end of
the research. The objectives are similar to the goals of the study but they are more
specific, measurable, attainable, realistic and time bound.
5. Methodology section
Research design or strategy – state the design (randomized controlled trial, cohort or
case-control, etc.), justify choice of the design
Sample population – define target population, demographic area, how they will be
selected, describe randomization if it will be done
Experimental intervention – describe how intervention will be given, describe the process
of blinding if it will be done, how to encourage compliance and other co-intervention
Outcome measurement – specify outcome attributes being measured, how they will be
measured, and ways of minimizing bias in its measurement
Analysis – specify the statistical analysis to be undertaken
Pilot study – describe how and to whom it will be done

36
 Ethics – emphasize informed consent, confidentiality, insurance policy if any
6. References
7. Time schedule and duties of personnel
8. Budget – indicate source of budget, specify budget for the following:
Personnel
Data handling costs (computer hours)
Materials and supplies
Equipment
Maintenance and operating expenses
Travel
Analysis
Write-up
9. Appendices – include letters of agreement, data collection forms etc.

Suggestions on How to Write Other Parts of the Protocol

Introduction
The introduction should awaken the interest of the reader and prepare him or her to
understand the paper. To accomplish this the introduction should be direct to the point, specific
rather than vague and general.
The most important point in the introduction is the statement of the research question. In
general the structure of the introduction is like a “funnel”. The broader mouth of the funnel
represents topics known about the subject. This is followed by topics not yet known. What is not
yet known usually leads to the research question. The research question can also be written as the
research objectives.

Methodology
Since you are writing a research protocol the methodology should be written in future
tense. It should be specific and detailed enough that the reader or reviewer will be able to
conduct or replicate the research methodology.

Budget
In preparing a budget you need to know the following, a) policies and requirements of
the agency from which grant funds are being sought, b) the policy and requirements of your
institution, and c) the resources and costs associated with each task or activity of the research
project.
The budget should cover for the following items:
A. Direct costs
1. Personnel (salary, allowances or honoraria)
2. Equipment
3. Supplies
4. Training expenses
5. Travel
6. Alterations and renovations
7. Consortium/Contractual costs
8. Other expenses
B. Indirect costs

37
 1. Some activities of the project will be shouldered by the host institution such as
building or office maintenance, utilities and other administrative expenses.
2. Rate is usually 40% to 60% of the total direct costs.
C. Institutional commitments
1. Ethical review boards fee
2. Donations

Ethical Review Boards

Before submitting a research proposal inquire from the ethical review board of your
institution about their review criteria and availability of funding. It may also be helpful to ask for
the names of the reviewers and ask for their comments and suggestions prior to submission.
Make sure that the protocol you will submit answers the review criteria.

38