PG GUIDE: DR S R SURYAWANSHI STUDENT: DR SAIPRASAD BHAVSAR

Designing and methodology of

a study

Research is a topic that many find intimidating

This issue of Research methodology will look at the concept of research, as well as the related concepts of evaluation and quality improvement, study design and analysis activities which many health workers are familiar with through their current work.

Basic classification

Various issues in any of the research methodology can be conveniently divided into six steps for the better understanding of the process.

1. Defining what is research2. Defining what is evaluation3. How to formulate a research

question.

4. Reviewing the literature5. Study design6. Quantitative data

analysis.

What is researchA broad definition of research is any

activity which is undertaken to increase the knowledge.

In the health field research can be defined as the systematic investigation of a problem, issue or question, which increases the knowledge and understanding of health and the provision of care.

Evaluation:- People who work In the health area are involved in the implementation of projects, which aim to improve the health or health service delivery.

Evaluation is the integral part of this implementation.

It can be defined as the process of value of an intervention by systematically gathering information to make more informed decisions.

Quality improvement(QI):- QI provides a framework for monitoring and improving performance, by systematically reviewing care provided, or outcomes achieved, against explicit criteria.

Thus it can be stated in short that, Research questions aim to

discover new knowledge; evaluation questions aim to judge

the worth of an intervention QI questions aim to examine how

well something is done and improve performance

What is evaluation?

Evaluation is the process of judging the worth of something.

It can determine if an intervention worked, help decide if it should continue, and provide evidence of effectiveness to obtain additional funding.

Some basic principles of evaluation

Why to EVALUATE?:- A good program evaluation will tell you

and others: what your program has done; how well it has contributed to the goal,

met the objectives and undertaken the strategies;

what worked well and what didn’t, and why;

whether there were any unintended outcomes; and

what can be learnt from the program to improve practice and inform other programs.

When should evaluation occur?

Evaluation should run in parallel to the planning and implementation of your program.

It should become part of a continual development process by providing feedback about progress, encouraging reflection about outcomes, and providing a basis for considering future strategies.

What steps are involved in evaluating a program?

There are three broad tasks involved in evaluating a program

developing an evaluation plan;assessing the results;communicating the results and

recommendations.

What should be in an evaluation plan?

Developing an evaluation plan involves:

clarifying the purpose of your evaluation;

selecting the scale and scope;

determining the methodology;

organising how the evaluation

will be conducted.

What types of evaluation are there?

Formative and summative1) Formative evaluation:--conducted early in the

implementation of a program;-aims to identify problems that arise

during development and allows modification.

2) Summative evaluation:-conducted at the end;-looks at effects or impact;-helps to decide what to do next.

Process, impact and outcome evaluation

1) Process evaluation:-focuses on how the program has been

implemented;-assesses whether activities were

conducted as planned.2) Impact evaluation:-focuses on the immediate effects of the

program;-judges how well the objectives were met.3) Outcome evaluation:-focuses on the longer term effects of the

program;-judges how well the goal has been

achieved

How to formulate a research question

Research originates with an idea about some general problem or question.

This problem or question is narrowed down to a more specific research question, which then represents the central issue being addressed.

First, it is important to distinguish between descriptive and analytical studies.

Descriptive studies ask simpler questions about what is going on.

For example, .How many or what proportion of patients admitted to hospital are with a fractured neck of femur?.

Analytical studies compare one or more interventions or exposures.

For example, Is it more effective to educate GPs about depression guidelines with group education sessions or practice visits?. or .Is lung cancer associated with cigarette smoking?.

First steps in formulating your question

Focusing analytical questions:-Well-built clinical questions usually

contain four elements-Patient or problem. Starting with your

patient, ask yourself, How would I describe a group of patients similar to mine?

-Intervention or exposure: Ask, Which main intervention am I considering?

-Comparison intervention: Ask, What is the main alternative to compare with the intervention?

- Outcomes: Ask, What can I hope to accomplish? or What could this intervention really affect?

Criteria for a good questionA good research question is

described by the acronym FINERFeasible (adequate subjects,

technical expertise, time and money, and scope)

Interesting to the investigatorNovel (confirms or refutes

previous findings, provides new findings)

EthicalRelevant (to scientific knowledge,

clinical and health policy, future research directions)

Reviewing the literatureWhy do a literature review? Determines to what extent the issue or

research question has been previously researched

Identifies past relevant studies as well as the methods used

Assists in refining your research question

Puts the project and methodology into a relevant context

Adds valuable background to the study or formal report

Suggests areas requiring further investigation

Required for funding applications

The two main components are conducting the search and critically appraising the results of your search, the published papers.

There are a number of courses on how to do literature searches.

These are available through Area Health Services (e.g. through the Clinical Information Access Program - CIAP) and university libraries.

Free databases are available at Pub Med, Cochrane Library and PEDro.

Critical appraisal skillsCritical appraisal skills are essential

for helping to decide if published research is of sufficiently high quality. Critical appraisal checklists have been produced by many authors.

CASP (2000) identified three broad issues that need to be considered when appraising research. They are:

Are the results of the study valid?What are the results?Will the results help locally?

STUDY DESIGNVarious designs are available

depending on the kind of question/s being asked.

For example, descriptive questions do not require a randomised controlled trial (RCT), but studies to evaluate a new treatment would benefit from a RCT.

Following are the various types of the study designs used for the study depending upon the conditions appropriate for the study.

Descriptive and analytical studies

One way of classifying study design is to divide them into descriptive studies that describe a situation or analytical studies that try to explain a situation by formulating and testing hypotheses

Non-experimental and experimental studies

Studies can also be categorised as non-experimental (or observational) with no intervention or experimental where the researcher intervenes, e.g. by introducing a new treatment

Main study designs

Randomised controlled trials (RCT)

Intervention trialsLongitudinal studiesCross-sectional studiesCohort studiesCase-control studiesCase study or series

RANDOMIZED CONTROLL TRIAL

In true sense this is an experimental epidemiological study.

Heart of this control trial is Randomization

Randomization is the statistical procedure by which the participants are allocated in the groups usually ‘study’ and ‘control’ groups. To receive or not to receive the intervention.

Thus it ensures that the investigator has no control over the allocation of participants thus eliminating what is known “Selection Bias”

Thus every individual gets an EQUAL opportunity to get selected into any of the trial group.

This can be done by BlindingThere are three typesa) Single Blindingb) Double Blindingc) Triple Blinding

Types of Randomized trials are as follows

a)Clinical trials:- E.g., The evaluation of beta-blockers in reducing the cardiovascular mortality in patients surviving the acute phase of myocardial infarction.

b)Preventive trials:- E.g., Vaccination trial

C) Risk factor trial:- E.g., four main intervention to reduce the CHD are reduction in the blood cholesterol, cessation of smoking, control of hypertension, and promotion of regular physical exercise.

d) Cessation experiments:- E.g., decreased incidence of lung cancer in the group who have given up the smoking.

e) Trial of etiological agent:- E.g., Retrolental fibroplasia as a cause of the blindness in premature babies exposed of high concentration continuous supply of oxygen than the control group who have been given oxygen only on emergency basis.

f) Evaluation of health service:- E.g., Excellent example is that of chemotherapy of tuberculosis in India which showed ‘Domiciliary treatment’ of TB was as effective as much costlier hospital or sanatorium treatment.

COHORT STUDY Is the another type of analytical or

observational study Known as prospective study,

longitudinal study, incidence study, forward looking study.

Distinguishing features are-a) Cohorts are identified prior to the

appearance of the diseaseb) Study group so defined are observed

over a period of time to determine the frequency of the disease

c) Study proceeds forwards from cause to effect.

Advantages are-a) Incidence can be calculatedb) Several possible outcomesc) Direct estimation of relative risk

and Attributable riskd) Dose response ratio can be

calculatede) Bias like misclassification of

groups into diseased and non diseased is minimized

CASE CONTROL STUDY

Often called “retrospective Study” is a common first approach to test the causal hypothesis.

Case Control study has three distinct featuresa) Both exposure and outcome have occurred

prior to the start of the studyb) The study proceeds backwards from effect

to causec) It uses control or comparison group to

support or refute the study.d) This is basically a comparison study

e) Estimation of the relative risk and odds ratio is the main distinguishing feature of case control study

Quantitative data analysis

Type of variablesAn item of data that can be observed or

measured is called a variable. There are two main types of variables.

1) Numerical variables can be: -Discrete variables - values that are

separate and distinct e.g. number of GP visits; or

-Continuous variables - when all values are possible e.g. blood pressure, weight.

2 )Categorical variables represent membership of a particular category. They can be:

-Ordinal variables – several categories where order is relevant e.g. physical activity measured as minimal, moderate or vigorous;

- Nominal variables - no natural order e.g. area of residence; or

-Dichotomous variables – only two responses e.g. yes/no.

In summarising numerical data, some common measures are:

Mean - sum of all individual counts or measures divided by the number of individuals;

Mode - most frequently occurring count or measure across a group of individuals;

Median - middle observation in a sample of individuals;

Range - difference between the maximum and minimum observations; and

Standard deviation - a measure of how much the individual data tend to deviate from the mean.

To describe the relationship between numerical variables, a common test used is the correlation coefficient, r. Correlations range between –1 and +1 and describe the nature and degree of association between two variables.

It is important to remember that correlations are not concerned with causality. An additional factor may underlie both variables.

In summarising categorical data, counts are used. They can also be expressed as proportions or percentages by dividing the count by the total number of individuals.

While categorical variables may be coded using numbers, it is important not to summarise them as numerical data. e.g.average coded numbers in Likert scales as below:

Very dissatisfied very satisfied

The relationship between categorical variables is usually presented in a contingency table and tested using chi-square test .

Statistical tests:- There are many statistical tests available. Two of the most common tests are the t-test for numerical data and the chi-square test for categorical data.

Inferential statistics In inferential statistics you are trying to reach conclusions about a population, based on a sample of individuals from the population.

In hypothesis testing, statistical methods are used to determine the probability of obtaining the observed effect by chance. The p value of your chosen statistical test is compared to the level of significance (usually set at 0.05 or 0.01).

For example, a t-test with a p-value of 0.03 (and level of significance of 0.05) indicates that the results are not due to chance and are statistically significant.

Confidence intervals can also be used. Based on a sample, they are estimations of a range of values (confidence interval) within which the population parameter is likely to lie.

With a level of significance set at 95%, you can say that based on your sample, you are 95% confident that the population value lies within your confidence interval.

REMEMBER

The many steps as explained earlier in planning, implementing and completing research

Among these, it is essential to choose the appropriate study design

Plan carefully before implementing your study

Choice of a Good Study Design is very important because

A badly designed study can NEVER be RETRIEVED…….

THAT’S ALL FOR THE DAY….

THANK YOU VERY MUCH…..