Randomization in clinical research

Sampling is the process of selecting a small group from the entire population for research. Selection of sampling method based on the research question should align with the specific objectives and goals of the research; (1) Understand the characteristics of the population, including subgroups or clusters. (2) Assess the available resources, including time and budget. (3) Consider the level of precision needed for your study. Some methods may yield more precise estimates than others. (4) Evaluate the practicality of implementing the chosen sampling method in the context of your study. (5) To consult with statisticians or experts in research methodology when making decisions about sampling methods.

To attain an unbiased outcome, it is important to use a systematic selection process to ensure the samples include all characteristics that represent the actual population for which the research is conducted. The use of a larger number of dental graduates as a sample to find the incidence of dental caries may deviate the outcome, as they may be well aware of oral hygiene practice. This disproportionate selection of sample causes selection bias and hence sampling is an essential for any research.

Sampling methods are either probability (randomization) or nonprobability sampling. Probability is a random selection of participants so that any individual of the whole population has an equal chance to be included in the study, whereas in nonprobability sampling, the researcher deliberately selects the participants for his/her research goals.

Randomization/probability sampling is a crucial aspect of research design, as it helps minimize bias and ensures that the groups being compared are similar in their basic characteristics at the baseline. Here are some key considerations for randomizing samples in dental research so that the outcome of the research is more reliable:

• Select samples based on the research question before randomization: For example, to evaluate the effect of implants in improving brain activity, we need a population with the same mental characteristics in both control and study groups. A subset/subgroup from the sample can be selected based on the objective of the research, such as gender, age, or any other such criteria

• Allocation of participants with the use of appropriate randomization method:[1,2]

  • Random sampling is done with the use of computers (computer-generated random numbers) and mathematical tables (random number tables) to generate sequences of participants that need to be included in the test and control groups. For example, to evaluate the effect of the implant on brain activity, use random number tables to select the required number of samples from the selected population

  • A systematic random sampling is performed when we have a defined sequence of populations. For example, the patient's unique identity number at the hospital can be used to collect data on oral manifestations caused by COVID-19. The researcher can recall every n^th number (like 5, 10, 15…) from the hospital data of such patients. This method is efficient and often more practical than simple random sampling when a defined list is available

  • Stratified random sampling ensures a precise balance by blocking or stratifying the systematically selected sample. In this method, a subgroup/subset based on age, sex, or educational level is stratified from the selected population. After stratification of the population, for example, based on sex, a simple random sampling is conducted under each stratum of male and female. This method ensures proportional representation for each stratum. For example, if age is a critical factor, use stratified randomization by dividing the participants based on age group (10–20; 20–30; 30–40, etc.), and later random sequence generation is done for each stratified group

  • Cluster random sampling is more practical to sample groups or clusters of individuals in a geographical region. This method is useful when the population is naturally organized into clusters. For example, a questionnaire study especially requires clusters of samples based on geographic area. The population in India is divided into clusters such as Delhi, Chennai, and Mumbai. If one of the cities is selected, then all individuals within the selected clusters are included as the sample in this method.

• Allocation concealment is blinding the researcher or participant to minimize bias in the assignment of a group and in the assessment of its outcomes. This helps to prevent conscious or subconscious selection bias by researchers or participants. The researcher should employ specialized software or statistical packages that include randomization algorithms to automate the randomization process and ensure its integrity in allocating participants

• Tailor the randomization approach to fit the specific design of your dental research: Specific randomization strategies are required for different designs/research objectives (e.g., parallel groups, crossover, and factorial)[3,4]

  • In a parallel group randomization, each group receives a specific treatment that is comparable. This has a test group and a control group or more than two test groups undergoing parallel treatment

  • In crossover randomization, the groups are interchanged with the rendered treatment after a specific period. This confirms that the obtained outcome is not based on patient characteristics

  • Factorial randomization is when two different interventions that do not interfere with each other are assessed on the same participants.

• Ensure that the randomization process is ethical and transparent. Participants should be informed about the random assignment process and the possibility of being assigned to any group. Clearly document the randomization process, including the method used, any blocking or stratification criteria, and the allocation sequence

• After randomization, perform statistical analysis to confirm that the randomization was successful and that the groups are comparable at baseline. This can involve comparing demographic and clinical characteristics between groups to ensure there is no statistical difference between the groups.

A detailed sampling technique is less commonly mentioned in an article, except for a statement that a randomized sampling is done. The research is considered successful only when it is applicable to the entire population, and hence, researchers should be familiar and take an effort in use of appropriate sampling methods during the conduction of the research.