A randomization tool for the daily routine
The case study in this Newsletter illustrates very well the importance of study subject randomization and blinding for clinical studies to prevent selection bias. The same is true for preclinical animal studies where it is also absolutely necessary to ensure that, for example, animals for an analysis are well distributed over the measurement period. This is especially true for experiments which take several hours. The easiest way to equally distribute all animals is via simple randomization, which will be discussed here in the context of in vivo but also in vitro experiments. In the following newsletters we will then introduce block randomization and stratified randomization.
Randomization based on a single sequence of random assignments is known as simple randomization. It is commonly used for in vivo experiments to ensure that scientists do not use all animals in an orderly way, but randomly choose animals from different groups during the experiment. The most common and basic method of simple randomization is flipping a coin or rolling a dice to assign a study subject to a treatment group. However, for animal studies in a laboratory environment, a free and easy-to-use online randomization tool is available: http://www.randomizer.org. For example, having two treatment groups (control versus treatment) and 20 animals in total, all animals will be numbered from 1 – 20 with the animals from 1 – 10 belonging to the control group and 11 – 20 to the treatment group. To randomize these animals using the online “randomizer”, the number of sets in the matrix should be chosen as “1” and the numbers per set as “20” (number of animals). The number range should be from 1 to 20 and each number should be unique. The numbers should not be sorted but it might be helpful to choose “Place Markers Within” as the last parameter. Pressing “Randomize now” will provide a random sequence of numbers that can be used to treat, analyze or measure the animals in an unbiased sequential arrangement.
How is this done for in vitro experiments? From our experience, samples or treatments are usually not randomized here. However, for the following scenarios, applying simple randomization techniques might be useful or even necessary to avoid a false experimental outcome:
Scenario 1 ‘Immunoprecipitation experiments comparing unstimulated vs stimulated samples’. Here, the unstimulated sample is always the first one that is treated, washed and analyzed. Wouldn’t it make sense to process this important control (to which all other treatments are compared to) chronologically in between all other stimulated samples – to make sure that the first sample is not always subjected to the same kind of bias during the experimental procedure?
Scenario 2 ‘Spatial arrangement of samples’. Many in vitro molecular biology studies use multi-well microtiter plates. The arrangement of samples on these plates is important, as technical artifacts can enhance or attenuate any treatment effects, which makes the interpretation of results difficult. One type of artifact is the ‘edge-effect’, where wells near the edge tend to be different than wells in the middle of the plate. This is quite often due to the fact that wells near the edge tend to lose any aqueous solution faster, which can change the concentration of substances remaining in the well. Irregular coating of plates can result in ‘gradient-effects’ across a plate with higher values obtained at one side compared to the other. A third artifact, the so called ‘plate-effect’, occurs when systematic differences between plates prevent the comparison of samples on different plates. In general, the layout of samples within and across plates should be arranged so that valid inferences can be obtained even if these artifacts are present. Distributing samples or rows (to avoid pipetting errors) randomly over the microtiter plate ensures that the experimental interpretation is not influenced by these effects when comparing experiments.
Scenario 3 ‘Experiments comparing cell death-inducing reagents’. Randomizing different treatments will ensure that the unstimulated sample is not always the first one and the highest test reagent concentration not always the last one to be processed. Cells in petri dishes might just be left dry for a few seconds longer during a medium change which could already have a significant impact on cell viability and, thus, could change the outcome of the experiment dramatically!
There are certainly many other ways to prevent and control for such events during the experimental procedure, however, randomizing samples also in in-vitro experiments could contribute to eliminate experimental bias and ensure higher confidence in experimental data.