This paper focuses on reviewing an article that uses both two-group design and factorial design. The article was published in 2011 and the authors of the article are Wolbers et al.
Tuberculosis meningitis (TBM) is considered the most dangerous infection of the M. Tuberculosis form. The infection either kills or disables approximately half of the infected patients necessitating the intervention of new effective strategies. The test hypothesis in this study is that the anti-mycobacterial regimes used currently are impotent and that an increase in the anti-mycobacterial levels drugs in the cerebrospinal fluid might improve the clinical outcomes (Wolbers et al., 2011). To test this hypothesis the researchers simultaneously increased the rifampicin dose and added levofloxacin to the standard treatment. A two-group comparison was then done between standard treatment and intensified treatment. The two-group comparison provided a good basis for testing the primary hypothesis of the study but had one major disadvantage since it does not allow quantification of the treatment effects of each drug or even facilitate exploration of the interactions between them. A 2 x 2 factorial design was, therefore, necessary in order to explore the interaction by randomly grouping the patients into two-factor levels (Wolbers et al., 2011). Factor 1(standard treatment versus standard +treatment A) and factor 2 (Standard versus standard + treatment B), such that 25% of the patients receive each combination of treatment. An analysis of factorial designs generally estimates the treatment effects of treatment A by pooling across factor 2 levels.
Delegate your assignment to our experts and they will do the rest.
Overall survival was the primary focus of the TBM study during a 9 months follow-up. It was expected that there will be a mortality rate of about 40% in the control (standardized) group and a significant mortality risk reduction of approximately 10% due to intensified treatment. A sample size of 706 patients was required in order to achieve a power of 80% and a 5% significance level (Wolbers et al., 2011). The dependent variable, in this case, is the mortality rate which depends on two independent variables, treatment A (intensified rifampicin) and treatment B (levofloxacin).
The researchers investigated the treatment effects of each drug with the assumption that each drug alone leads to an equal reduction of the hazard. Sample sizes were varied for these investigations. From the results obtained, if the overall treatment is contributed by only one drug, the power of the 2x2 factorial design to detect the effect of individual treatment is almost equal to the power of detecting a combined treatment effect of the two drugs (Wolbers et al., 2011). If the two drugs contribute to the combined effect, the power is far much reduced. In order to detect the individual effects of the drug, the sample size should be increased by nearly 8 folds. The results also show that the power of the combined effect is further reduced in case of a negative interaction. In order to detect -100% interaction (effect of one drug equals the combined effect of the two drugs), with a power of 80%, the sample size would have to be increased 16 times (Wolbers et al., 2011). The reasons for this kind of results is the fact that the two drugs have equal contribution to the overall treatment effect and second is the fact that interaction tests often lack power. An increase in the sample size would detect any differences (if any) in the treatment effects of the two drugs.
The researchers conclude by indicating that the use of a two-group design is more appropriate than factorial design when several assumptions are met. First, a feasible powered design is not possible due to the large sample size. Second, both interactions test the same hypothesis (Wolbers et al., 2011). Third, when both drugs are considered ineffective. Fourth, none of the drugs is more toxic or costly than the other and lastly is that such combination is would not be accepted by the regulatory authority. The sample in this study is randomly selected.
It is evident that in cases where multiple interventions are to be tested, group designs are the most appropriate. It is generally difficult to use a factorial design where the treatments have the same effects. One major disadvantage of this study is that it does not present the actual effects of the treatments, A and B.
Reference
Wolbers, M., Heemskerk, D., Chau, T. T. H., Yen, N. T. B., Caws, M., Farrar, J., & Day, J. (2011). Sample size requirements for separating out the effects of combination treatments: randomised controlled trials of combination therapy vs. standard treatment compared to factorial designs for patients with tuberculous meningitis. Trials , 12 (1), 26.
