Pooled Gingival Samples are Less Statistically Sound than Individuals

Objective: Periodontal research frequently requires sampling the gingival tissue from small size animals such as rat.  Due to limited quantity of available gingival tissue in such animals, it has been reported that instead of processing each sample individually, samples in the same group are pooled, processed and analyzed, e.g. measurements of proteinase activities.  Means of pooled samples are then compared between groups.  Since samples are pooled, the results may be different from those when each sample is processed individually.  We investigated this issue with statistical tools to determine if “pooled sample” protocol can produce the same results as those with the individual sample protocol. Method: A series of Monte Carlo simulation studies were conducted to investigate the power of t -test when samples were pooled or individually processed.  Data of different sets of statistical parameters such as variance, sample size and proteinase activities were generated by computer software. Result: The simulation results indicated that the power of t-test decreased with pooled samples because final sample size was reduced. It was calculated that four animals in each pool could triple the power loss compared with only two animals in each pool. If tissues from six animals were pooled, then the power loss was more than fivefold. It was also shown that the larger the sample size and the smaller the variance can reduce the difference in power values of t-test between the two protocols, as demonstrated in our simulation tests with various sample sizes and variances. Conclusion: We concluded that “Pooled sample” protocol was less sound statistically compared to individual sample protocol because the former resulted in smaller sample size and reduced power value of t-test.