Overview about study designs and sampling

Defined as the distribution and the determinants of health related states and events in a particular population and it's application in controlling health problems.

Major study designs

1) Observational
2) Experimental 

a) Observational

i) Descriptive studies
Features :

No comparison is being made. (hence, no control group is required)
Application of statistical test of significance is therefore not required.
The prevalence and incidence can be multi-centric, when >2 centres are used. 
The main purpose of observational studies is to generate a hypothesis.


Eg.

i) Case reports - Reporting of an interesting case/finding from a laboratory experiment
ii) Case series analysis - Reporting of large amount of similar cases in a particular time period
*Note that both the above studies are retrospective.

iii) Ecological - A representative from a community is interviewed about the psycho-socio-cultural background of the community

iv) Cross-sectional studies - Field of survey within a short period of time for identification of both old and new cases.

v) Prospective studies - Follow-up done in a community in order to identify new cases

ii) Analytical studies
Features :

Since comparison is being made, a control group is match for at least their age and sex
Hence, application of statistical test of significance is mandatory.
The purpose : to confirm a hypothesis created.

Eg,

i) Case-control studies (Retrospective)
ii) Cohort studies (Retrospective/Prospective/Retro-prospective)
iii) Nested case control studies (Case control embedded in Prospective studies)

b) Experimental studies

Comparison is required. Hence, control group is obtained after informed consent, whenever feasible.
The study design can be multi-centric, when >2 centres are used. 
Application of statistical test of significance is mandatory.
To access the relationship in between cause and effect.

4 phases :

Phase I trial - done on animals in laboratory settings (minimally on terminally-ill patients)
Phase II trial - either therapeutic (drugs)/preventive (vaccine), on diseased patients
Phase III trial (field trial) - Done on individuals who are apparently healthy.
Phase IV trial - Done on community as a unit.

Sampling

a) Probability sampling
b) Non-probability sampling

The major difference in between the above two is : the former uses randomized method for sampling, the latter uses non-randomized method.

a) Probability sampling

Simple random sampling (SRS)
By using random number table or lottery method, samples are obtained from a homogenous population.
Each individual has equal chance of being selected, with or without replacement.

SRS using probability proportionate sample size method (PPS)
Here, emphasis is placed on proportionate representation of homogenous population according to their locality (district/state), with or without replacement.

Stratified random sampling
Proportionate strata of a heterogenous population, by age, sex, occupation, social status, etc.

b) Non-probability sampling

Purposive sampling
The respondents are chosen based on the desire of the selector.

Snowball technique of purposive sampling
This is used if there's difficulties in gaining assess into a community.
The participants of a study is asked to recommend their friends to join the study.

Sequential sampling
A sequential set of samples are being pooled until a desirable and suitable sample is present.

Quota sampling
Here, sample units are chosen with non-randomized method under a fixed quota.
There are 2 types :

i) Proportional QS : A major characteristic of a population needs to be represented, by sampling them proportionately to each other.

Eg, In a population, is known to consists of 40% of females and 60% of males.
Sampling will continue even if the quota of 40% of females is achieved, if the 60% quota is yet to be achieved.

ii) Non-proportional QS : Here, it's less restrictive. After clearly stating the minimum number of participants in each category of a study, we are not concern of whether the number of participants actually proportionately represents the population. As long as there's sufficient amount of people in the study to the extent of even the small groups are adequately represented.

Procedures of sampling

Source of samples : Obtainable from the house-hold list and updated family folders.
Then, sample units are selected by lottery method, with or without replacement.
Application of the PPS technique.
Sample fraction = Total sample size required / Total number of population study
By arranging the family folders according to their locality, random number table is used to select the respondents.

Category: Community Medicine