Dr Prasanth V
RESEARCH DESIGNS – A Quick Guide
EXPERIMENTAL DESIGN CONCEPT
Two or more groups are involved in the study. One group is the intervention group (new treatment/drug) and the other forms the Control group (standard/ placebo). Investigator decides in the choice of intervention.
This includes randomized and nonrandomized control trials.
OBSERVATIONAL DESIGN CONCEPT
Investigator stands apart from the events taking place from the study, simply observe and record.
DESCRIPTIVE DESIGN CONCEPT
First scientific toe in water in new areas of inquiry. Good descriptive study answers five ‘W’ questions, who, why, what, when and where and the 6th one. ‘so what’. They describe or estimate a phenomenon in terms of ‘person’, ‘place’ and ‘time’. This design generate ‘hypothesis’ (no relationship can be proved) which can be tested using ‘analytical’ type of designs.
ANALYTIC DESIGN CONCEPT
Used to test hypothesis concerning relationship between a suspected risk factor and an outcome and to measure the magnitude of the association and its statistical significance. They are called natural experiments.
Observational and descriptive. It’s a careful detailed report of the profile of a single patient or case. Eg- Case of 40 year old woman developing pulmonary embolism within 5 weeks of oral contraceptive usage in the year 1961.
Observational and descriptive. Series of homogenous cases are studied and described. Eg – A review of all patients diagnosed as endometrial carcinoma and look for history of hormone replacement therapy.
CROSS SECTIONAL STUDY (Snapshot studies, Prevalence Studies) Observational and descriptive (can be
analytical also if there is comparison of groups based on
exposure or outcome). These studies provide information concerning a situation (exposure and outcome/disease) at
a given time and hence provides information concerning
a situation at a given point time only. Not feasible to study
rare conditions and not ideal for studying rare exposures.
Eg – Prevalence of perceived stress among students of
various Medical Colleges under KUHS.
Observational and descriptive (can be analytical also). Here the unit of study is not individuals, but populations. Exposure and outcome obtained from different groups or populations are compared. Since ecologic studies refer to ‘populations’ and NOT ‘individuals’, it is not possible link the exposure and outcome in an individual level (ecologicfallacy).
COHOHT STUDY (Longitudinal / Incidence Studies)
Group of persons (Cohort) are identified based on their exposure. They should be free of disease/outcome of interest, but differ in exposure selected. They are followed identically until they develop outcome/disease. Incidence of disease and risk calculated. Eg–Risk of passive smoking among pregnant women on low birth weight of babies.
Easy Learning - In one vineyard pesticides are routinely applied and in other vineyard it’s not applied normally. There was a hypothesis that, the spoiled grapes were less in the first vineyard. To verify that, compare the amount of good and spoiled grapes at the time of cultivation obtained from both farms.
It’s a retrospective study design because group of subjects with disease/outcome of interest (cases) and group of subjects without disease/outcome of interest (control) are identified first. Cases and controls are investigated for presence or absence of hypothesized risk/exposure factors. Analysis in this type of studies are done by calculating Odds Ratio (OR). Eg – Is post-menopausal hormone replacement therapy associated with increased risk of endometrial cancer?
Easy Learning - Just the reverse of what we saw in cohort study. Here we start with identification of spoiled grapes (case) and good grapes (control). History of pesticide exposure in both groups are investigated and revealed.
RANDOMIZED CONTROL TRIALS (RCT)
It offers highest form of evidence regarding association and causation. The word ‘control’ means that observations are conducted under controlled circumstances in which the researcher manipulates the conditions. This ‘controlled’ experiment can be randomized or non-randomized. ‘Randomization’ ensures that the assignment of study participants into ‘study’ and ‘control’ groups are done by ‘chance’ alone. This is almost similar to Cohort study, but the intervention is decided by the investigator.
Easy Learning - We want to study the effect of rain on the health of children If observe children standing in the rain with and without umbrellas whom we divide into two groups (one with umbrella and the other without umbrella) and study the effects, then that will be a cohort study, a natural experiment. If the investigator divide children standing in rain randomly into two groups, give umbrellas to one group (study group) and no umbrellas to the other group (control group) and both groups are then compared, it will be a RCT.
Parallel Design, Cross over design, Factorial Design, Multi-centric trials, Mega trials and Split half designs (split mouth, split eye etc) are various types of clinical trials. According to purpose of research, trials can be therapeutic trials, prophylactic trials, safety trials and risk factor trials.
HIERARCHY OF RESEARCH DESIGNS
OBJECTIVES AND STUDY DESIGN