WP08 Statistics


 

Work Package Lead: Prof. Marcus Lind

08. VASTRA GOTALANDS LANS LANDSTING – NU-SJUKVARDEN (SE)

 

Objectives:

  • Write all statistical parts of the protocol including statistical methods and sample size for primary efficacy variable.
  • Review the study CRFs.
  • Provide randomisation system ( Web-based) using minimisation algorithm to the technology platform (EVIDEM).
  • Write detailed Statistical Analysis Plan (SAP).
  • Produce a Tables Manual, a document that gives details for all Tables, Listings and Figures in the Statistical Report and in the Clinical Study Report, as part of the Operations Manual.
  • Check data for irregularities with statistical methods during the study in coordination with CRO.
  • Create Statistical Analysis SAS-datasets that are the basis for all Analyses, Tables, Figures and Listings and validate them.
  • Make all the statistical analyses, produce all Tables, Figures and Listings for the Statistical Report and validate all results.
  • Write or review the statistical parts in the Clinical Study Report.
  • Characterise the trial population and compare it with the data from the DECODE study, matching randomised patients by age and sex, etc with the DECODE participants.
  • Develop various layperson-oriented prediction tools for diabetes complications based on the data of the DECODE and externally validate the tools based on the trial data.
  • Economic burden of diabetes complications and cost-effectiveness analysis of the intervention project.
  • Design optimized OGTT or meal tests suitable for calculation of model-based insulin sensitivity and beta-cell function indices (CNR).
  • Contribute to produce an operation manual for the experimental tests.

 

Main tasks:

1. Statistical parts of the Protocol to the e-PREDICE study:

  • Suggest statistical methodology for analysis of primary variable and the most important secondary variables. Perform a sample size calculation for the primary variable after statistical methodology defined.
  • Check that all variables needed in the statistical analyses are measured appropriately.
  • Provide randomisation system (Web-based) using minimisation algorithm.

2. Produce a final Statistical Analysis Plan (SAP) to the e-PREDICE study:

  • The SAP should describe the general statistical methodology and in detail how each variable should be analysed and described in the study. The SAP should also contain a listing of all Tables, Figures and Listings that are part of the Statistical Report and the Clinical Study Report. The draft SAP will be distributed to be signed by the executive committee and the Study Statistician.
  • The team will also develop Statistical Analysis Plans for each planned article from the study not included in the main SAP, together with co-writers, executive committee.
  • All SAPs should be finalised and signed before locking the e-PREDICE database. Any changes in the SAP prior to database lock will be carefully documented in the Memo-to-files.

3. Produce Tables Manual, a document that gives details for all Tables, Listings and Figures in the Statistical Report and in the Clinical Study Report:

  • After the detailed main Statistical Analysis Plan (SAP) has been signed the first draft version of the Tables Manual (TM) will be produced. The Tables Manual contains a detailed description (mock table) of all tables and figures and a list of variables that are to be listed in the listings that should be produced for the Statistical Report and the Clinical Study Report. The draft TM will be distributed to all members in the executive committee and to other selected. The team will take into account all suggested improvements into the TM. In case of disagreement a meeting will be held to discuss the TM. The final TM will be signed by the executive committee.

4. Check data for irregularities with statistical methods during the study:

  • We plan to check blinded study data during the study for irregularities with statistical SAS-programs.

5. Create Statistical Analysis SAS-datasets that are the basis for all Analyses, Tables, Figures and Listings:

  • From the locked database we will create Statistical Analysis SAS-datasets to be used in all SAS-programs for statistical analyses, Tables, Figures and Listings. For each Statistical Analysis SAS-dataset a requirement specification will be written together with a Test Plan. All derived datasets SAS-programs will be validated (double programming).

6. Perform all the statistical analyses, produce all Tables, Figures and Listings and validate all results:

  • All results will be generated using SAS-programs. All results given in Tables and Figures will be validated (double programming). It will be checked that all Listings include all planned variables. The programming of all datasets, analyses, tables, figures and listings might be done during the study in a blinded manner, but before the database lock.

7. Write or review the statistical parts in the Clinical Study Report:

  • After the Statistical Report has been delivered, the team will assist the study Medical Writer by reviewing or writing the statistical parts of the Clinical Study Report, as found most appropriate.
  • Conventional descriptive statistics will be applied to compare the trial data with that of the DECODE. Cox regression analysis will be used to develop the risk prediction models, and the performance of the models will be evaluated with regard to the discrimination, calibration and net reclassification improvement (NRI). 15D (http://www.15d-instrument.net) developed in the University of Helsinki will be used to measure health- related quality of life (HRQoL). The 15D has been translated into 25 different languages including English, French, German, Italy and Spanish. Both cost-effectiveness analysis (CEA) and cost-utility analysis (CUA) are planned to be made to measure the amount spent on preventing one case of diabetes complications and the number of quality-adjusted life years (QALYs) gained within the intervention program.

8. Metabolic modelling:

  • Insulin sensitivity and beta-cell function will be determined by modelling analysis of OGTT or meal tests, according to validated methodologies. Upon completion of the tests, the data will be collected in electronic form and analyzed by modelling to obtain an insulin sensitivity index and several beta-cell function parameters representing various aspects of the beta-cell response. Quality controls will be performed during the analysis. The calculated indices will be used for assessing the treatment effects and for the study of the underlying mechanisms.

 

Scroll to Top