Study Development and Management
Once an observational study proposal is approved by a Working Committee, it is assigned a CIBMTR study number and a provisional timeline is established. The Principal Investigator (PI) is asked to sign a letter of commitment and agrees to place reasonable efforts into completing each stage of study development as efficiently and quickly as possible.
Stages of study development
Protocol development
Supplemental forms/data collection
Data file preparation
Analyses
Manuscript preparation
Submission
Considerations for studies using NMDP samples
Stages of study development
The following describes the stages of study development as they are conducted by CIBMTR Working Committees. The process pertains to studies for which CIBMTR provides both data and statistical expertise. Datasets may be made available to investigators who have their own statistical resources. Manuscripts resulting from these analyses are reviewed and approved by CIBMTR prior to submission to a journal.
Protocol development
The first responsibility of the PI is to develop the study proposal into a comprehensive study protocol, with the help of invited Working Committee members. Study protocols are used by CIBMTR biostatisticians to prepare data files for analyses and define the detailed study design. This crucial document drives everything that happens once the study is accepted by CIBMTR.
The proposal must include:
- Hypothesis: scientific assumption that is the basis for the study;
- Objectives: specific aims that will be achieved by the proposed analysis;
- Scientific justification: summary of the study rationale, conveys the importance of the study;
- Study population: definition of selection criteria;
- Outcomes: clear definition of study outcomes, including time-points if relevant;
- Variables to be analyzed: listing of explanatory variables, based on biological principles, available in the observational database and proposed format/categories for analysis;
- Data collection: specifics of supplemental data required and a plan for data collection;
- Study design: approach to achieving each objective (this will be refined with support of CIBMTR Statistical Center biostatisticians);
- Bibliography
Once received by the CIBMTR Statistical Center, the draft is further refined in collaboration with the Working Committee MS and PhD biostatisticians, Scientific Director and Chairs. A table with a preliminary description of the proposed population is added and the draft protocol is presented at a weekly Statistical Center meeting.
Supplemental forms/data collection
Most studies will use data collected routinely, but in special cases where an important study cannot be done without supplemental data, collection of a minimal amount of additional data may be undertaken. If the study requires supplemental data collection (e.g. information not collected on CIBMTR Comprehensive Report Forms), development of a supplemental form will be required prior to approaching centers for that additional data. These forms are prepared by CIBMTR Statistical Center staff in collaboration with the study PI and Committee Co-Chairs.
Data file preparation
The objective of study file preparation is to create a data file of eligible subjects who are consecutively treated at participating centers with adequate follow-up, with minimal missing data fields, and in large enough numbers to give the analysis sufficient statistical power to meet the stated study objectives. This process involves a series of steps by the MS biostatistician, sometimes working together with a Clinical Research Coordinator, to ensure data quality. It often involves consultation with the PI and the physicians and biostatisticians at the CIBMTR Statistical Center.
These steps include finalizing:
- Selection criteria;
- The adequacy of follow-up, and taking steps to obtain additional follow-up information if necessary;
- The extent and nature of missing values and their potential effect on the study, and taking steps to obtain missing data if necessary and feasible;
- Data discrepancies/outliers, and reconciling these by examination of data collection forms, communication with centers and the PI. This culminates with the MS biostatistician providing a detailed table describing the patient population for the PI, the Chairs, the Scientific Director and the PhD biostatistician to review before analysis;
- Appropriate groups for continuous and categorical variables, if not already specified in the protocol;
- Included and excluded patients so that the investigators can determine whether the final study population is representative of the target population (unbiased sample).
Analyses
Analyses proceed in several phases. The first analysis generally includes a detailed description of the patient population and univariate analysis of study endpoints. A multivariate analysis is then performed by a PhD biostatistian. These data are distributed to Writing Committee members, requesting their suggestions and comments. An iterative process then ensues. The PI works with CIBMTR Statistical Center staff to summarize and discuss comments from the Writing Committee. Revised analyses, with a description of steps taken to address comments, are distributed to the Writing Committee. It is the responsibility of the PI to draft this memo with input from Statistical Center staff. If additional substantive comments are made by the Writing Committee, the process is repeated until a final analysis is available. This final analysis serves as the basis for the manuscript.
Manuscript preparation
The PI has primary responsibility for manuscript preparation, and is expected to prepare a draft manuscript within 30 days of receipt of analysis results. This draft is then reviewed, revised as necessary, and approved by Working Committee leadership. They ensure that description and interpretation of the statistical analyses are accurate, as well as contributing to the fundamental message of the manuscript. The first draft is then distributed to the Writing Committee by the MS biostatistician for their comments.
The Writing Committee ideally includes persons who make a significant contribution to study development. The PI may wish to incorporate relevant comments into a subsequent draft. As with the analysis, this is an iterative process until all involved agree that the manuscript is ready for submission. Writing Committee members have two weeks to respond to each circulated draft. The approved final draft is the version submitted by the CIBMTR Statistical Center to the appropriate peer-reviewed journal.
The final author list is determined at this stage since it depends on the value and extent of contribution of each individual throughout the study process. To assure co-authorship, members of the Writing Committee must make timely and substantive contributions to study design, data analysis, interpretation of results, and preparation of the manuscript for publication.
Submission
CIBMTR Statistical Center staff are responsible for submission of the manuscript as well as any correspondence with the chosen journal. In most cases, the corresponding author is the Working Committee Scientific Director. The study MS biostatistician forwards editor and reviewer comments to the PI, the Scientific Director and the PhD biostatistician. The PI is expected to prepare a response, if necessary, working with Statistical Center staff, who will provide additional analyses of the data as needed. Communication with the journal, including re-submissions, is continued as necessary by Statistical Center staff.
Considerations for studies using NMDP samples
The following guidelines apply to studies that require samples from the National Marror Donor Program® (NMDP) Research Sample Repository (samples collected by the NMDP from related and unrelated transplant donors, cord blood units and recipients):
- The Working Committee Chairs may require inclusion of CIBMTR and/or NMDP personnel who contribute to specimen procurement and data analysis.
- The PI is responsible for abiding by the NMDP Research Sample Reporting Requirements for submission of laboratory data.
- In return for access to the samples, the PI will be required to submit the interpreted results of all assays performed on the samples. Data should be submitted to the NMDP as directed prior to submission of the study for publication.
- The data submission requirement ensures that all sample testing yields information that is readily available to the research community for subsequent analysis and eliminates or reduces duplicative testing to preserve resources and sample inventory.
- Data will only be made available for subsequent study following publication of the original analysis.
