Random number tables are a statistical method for developing random population samples from which to conduct studies on. The tester selects the inclusion criteria for his study and then proceeds to use a random number table to determine who or what is in the population sample. Any subject that does not find the criteria is thrown out and those that do are introduced into the sample. In a study by Chapman, et al. (2013), a random number table was used to select patients for a trial design to determine whether immunosuppression can preserve renal function in patients with membranous nephropathy. The inclusion criteria required patients with biopsy proven disease, plasma creatinine concentration <300mol/L, and a 20% decline in renal function over 2 years.
Using a random number table the patients were assigned to 1 of 3 groups: supportive treatment only, supportive treatment + 6 months of alternating cycles of prednisolone and chorambucil, or supportive treatment +12 motnhs ciclosporin. Point estimation is a statistical method in which a value serves as a general value for the population at large. In a study by McInnis, et al. (2013), it was thought that the HOXB13 missense mutation causing a 20-fold increase in the rates of prostate cancer was too high. By screening for the mutations in a large Australian sample and then using questionnaires to obtain a detailed cancer history and subsequent pedigree analysis, the authors concluded that the age-specific incidence of disease for the carriers of the mutation was 16.4x that of the general population. Point estimation is contrasted to confidence interval, insofar as confidence interval expresses a range in which the population may be at risk, whereas the point estimate expresses a relatively certain integer of the population at risk. A phase III clinical trial is a large, randomized, multicenter evaluation of the therapeutic efficacy of a drug compared to current treatment methods. Because of the size and potential time they take to complete they tend to be expensive and difficult to conduct.
Once the phase III trial is completed they make up the bulk of the regulatory report that will be submitted to a governing agency charged with approving a new drug. The randomization is done in order to get a representative sample from which satisfactory conclusions can be drawn for a large population of patients that may or may not benefit from the drug. In these studies both physicians and biostaticians must carefully plan sample sizes in order to determine statistical significance of a treatment (Rohrig, 2010) In 2008, Scagliotti, et al. conducted a phase III trial comparing cisplatin plus gemcitabine with cisplatin plus pemetrexed in patients with advanced stage non-small cell lung cancer who had never been previously exposed to chemotherapeutic agents.
The randomization method involved in this study was the Pocock and Simon method, which is a method that is designed to limit imbalances that might be obtained in a research study because of imbalances favoring a subgroup (for instance, in this case, the vast majority of patients were probably smokers – and thus, the general statistical value of the study would’ve been compromised by this overrepresented subgroup). Pocock and Simon stratify patients based on a variety of factors and ultimately add imbalances together. A group can then be chosen where the imbalances are kept low in order that general conclusions can be more easily drawn. Using good CI’s, it was shown in the Scagliotti study that the new treatment regimen had significantly better survival times for patients with large-cell and adeno - carcinomas than the original cisplatin/gemcitabine treatment protocol. For other subgroups there was no statistically significant survival.
Howman, A., et al (2013). Immunosupression for progressive membranous nephropathy:
a UK randomized controlled trial. Lancet, 381, 744-751. http://dx.doi.org/10.1016/S0140-6736(12)62033-9
MacInnis, R.J., et al (2013). Population-Based Estimate of Prostate Cancer Risk for
Carriers of HOXB13 Missense Mutation G84E. PLOS ONE, 8(2), e54727. doi:10.1371/journal.pone.0054727
Röhrig, B., et al (2010). Sample Size Calculation in Clinical Trials. Deutsces Arzteblatt
International, 107(31-32, 552-556. doi: 10.3238/arztebl.2010.0552
Scagliotti, G.V., et al (2008). Phase III Study Comparing Cisplatin Plus Gemcitabine
With Cisplatin Plus Pemetrexed in Chemotherapy-Naïve Patients with Advanced-State Non-Small-Cell Lung Cancer. Journal of Clinical Oncology, 26(21), 3542-3551. DOI: 10.1200/JCO.2007.15.0375