Statistics is an important part of the research, because regardless of the study design, the researchers need to summarize the information collected so that it can be interpreted and presented to others. Therefore, we must pay attention to it when creating a data analysis plan. In fact, even before data collection begins, we need a clear analysis plan, SAP, which will guide us from the initial stages to the end to summarize and describe the data to testing our assumptions.
The statistical analysis plan (SAP) describes the planned analysis of clinical trials. Compared to the protocol that outlines the analysis, the SAP is a technical document detailing the statistical techniques of research analysis. The SAP defines all statistical outputs that will be included in the clinical study report. The SAP and the annotated CRFs are the most commonly used documents for statistical programmers to create their deliverables. In general, the followings should be included in an SAP (Figure 1).
Figure 1. Contents of a SAP
Our Services
CD BioSciences' statistical experts will help you develop a clear and comprehensible plan to build tables, data and lists for dataset analysis and plan preparation. To achieve that, our work is constituted with the design of SAP for the various clinical trials as follows:
Generally, there are three types of missing data: missing completely at random, in which there is no pattern in the missing data on any variables; missing at random, in which there is a pattern in the missing data but not in your primary dependent variables; missing not at random, in which there is a pattern in the missing data that affect your primary dependent variables.
Here we can use listwise deletion, and recover the values, educated guessing, average imputation, common-point imputation, regression substitution or multiple imputation to help you to handle with the missing data.
In SAP, you should provide the types of statistical tests to be used, the layering method, the type of squared sum (if applicable) and other information needed for the analysis. For example, when a formal meta-analysis is to be performed, the model should indicate which terms are considered fixed effects and which terms are considered random effects.
In addition, with parameters/nonparametric/binary endpoints, or continuous/category/binary data, our experts will help you add substructures to break through the statistical methods.
The multiplicity may appear in many different situations in clinical trials and the following picture (Figure 2) shows some of them. If the multiplicity is not properly handled, the effectiveness of a drug may be made as a consequence of an inflated rate of false positive conclusions. Using usual Bonferroni and sequentially rejective Bonferroni methods or Hochberg method, our experts could evaluate the various approaches and determine which adjustment approach is best suited to the situation in your study.
Figure 2. Multiplicity arises in many different situations.
A confounder is a variable whose existence affects the variables being studied so that the results will not reflect the real relationship. When study designs are premature, impractical, or non-executable, investigators must rely on statistical methods to adjust them for potentially confounding effects. We usually use logistic regression, linear regression and ANOVA to help our clients to adjust confounders.
Heterogeneity is an unwanted variable when analyzing aggregated datasets from multiple sources (e.g. clinical trial datasets from multiple study centers). Our statisticians will develop an independent correction method for heterogeneity adjustment and make it the best fit for your research.
Sensitivity analysis usually is to reanalyze the same results using different methods or different definitions of results - the main goal is to assess how these changes affect the conclusions. If the results are different or result in a different conclusion from the original results, all results of the sensitivity analysis should be reported. However, if the results are still robust (i.e., remain the same), a brief statement may suffice.
Every non-clinical or clinical trial has its own algorithms for generating tables and results. Based on your research question, study design, the level of measurement, and the level of significance, our experienced statisticians will provide you with an appropriate statistical method and lead you to draw a correct conclusion.
We guarantee the confidentiality and sensitivity of our customers' data. We are committed to providing you with timely and high-quality deliverables. At the same time, we guarantee cost-effective, complete and concise reports.
If you are unable to find the specific service you are looking for, please feel free to contact us.
References:
1. Simpson, S. H. (2015) ‘Creating a data analysis plan: what to consider when choosing statistics for a study’, Canadian Journal of Hospital Pharmacy, 68(4), 311.
2. Thabane, L., Mbuagbaw, L., Zhang, S., Samaan, Z., Marcucci, M., & Ye, C., et al. (2013) ‘A tutorial on sensitivity analyses in clinical trials: the what, why, when and how’, Bmc Medical Research Methodology, 13(1), 92.
3. Yu, J., Zhou, H., Liu, X., Zou, F., & Willke, R. J. (2014) ‘Adjusting complex heterogeneity in treatment assignment in observational studies’, Powder Technology, 169(2), 71-76.
4. Pourhoseingholi, M. A., Baghestani, A. R., & Vahedi, M. (2012) ‘How to control confounding effects by statistical analysis’ Gastroenterology & Hepatology from Bed to Bench, 5(2), 79.
5. Proschan, M. A., & Waclawiw, M. A. (2000) ‘Practical guidelines for multiplicity adjustment in clinical trials’, Controlled Clinical Trials, 21(6), 527-39.