Resources
RESOURCES
Given its commitment to enhance the efficiency in conducting quality research within the department, the ARC has created several technical tools and packages to better facilitate your research projects. We have also included helpful links that can help navigate the research process at MGH.
Please click here to access the presentation library, including previous ARC applied research didactics and research presentations.
Tools & Resource Library
SAP Template
Provided is a Statistical Analysis Plan (SAP) template that describes in detail the planned analyses of a research study, which should be used in conjunction with the protocol.
Codebook Template
A codebook describes the contents, structure, and layout of a data collection. An example of a codebook is attached.
Datasheet Template
Researchers are encouraged to use REDCap for PI initiated studies. If not using REDCap the attached excel template can be customized for your data collection needs.
SAP Guidelines
Recommendations are provided for a minimum set of items that should be addressed and included in SAPs for clinical trials.
Click here to access the article.
Publishr R Packages
In this package is a collection of functions for medical publication.
Click here to access this package.
Writing in the Sciences
This free course provided by Stanford teaches scientists to become more effective writers, using practical examples and exercises.
Click here to access access the course.
Literature Reviews & Meta-Analyses
The HMS Countway and MGH Treadwell Libraries have resources available to help you get started on a narrative review or meta-analysis.
Find out more about these resources here.
IRB Template for Medical Record Reviews
For studies that include only secondary use of health / medical record data the IRB has developed an expedited review process. ARC has put together template language that may assist in the submission process, if applicable.
The Importance of Statistical Power Calculations
In this article members of ARC describe the importance of power calculations, as well as important components to consider.
Click here to access the article.
Observational One Page Template
Members of the ARC team have put together a one page template for outlining your observational research project. This document often serves as the SAP for short analytic efforts, or informs the discussion for the initial study consult. Click below to download and complete this form.
Suggested References
Missing Data
To find out more about missing data, the following open source articles are suggested:
——-
Fixed and Random Effects
To better understand linear mixed effect models (and how to code them in R) please see the following curated references from ARC:
- Introduction to Linear Mixed Models
- Linear Mixed Models
- Fitting Linear Mixed-Effects Models Using lme4
Confounder Control
Often times when we start to develop models for a cohort study it is important to consider the role of ‘control’ or ‘confounding’. More information to spot a ‘bad’ control can be found here.
Time Series Analyses
This branch of statistics is wonderful but requires a considerable number of unique considerations. More information on items to consider when evaluating and constructing time-series models can be found here.
Propensity Scores
A propensity score is the probability of treatment assignment conditional on observed baseline characteristics. To find out more, please see these suggested references:
- An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies
Bias-Variance Tradeoff
Often when building a model we have to consider the bias-variance tradeoff, in which we simultaneously attempt to minimize the these two sources of error. As the complexity of the model increases the bias decreases, however the variance of the parameter estimates increases, thus this must be considered when selecting an ‘optimal model’.
- Bias-Variance Tradeoff
- Understanding the Bias-Variance Tradeoff
- Understanding the Bias-Variance Tradeoff
Sparse Data
Small samples sizes, and particularly sparse data, are often problems in clinical research and can introduce bias into your analysis and interpretations. Several statistical approaches have been identified to address these problems and are discussed briefly here.
Attribution
Whenever projects are completed using ARC personnel, services or equipment, please be sure to include attribution, as appropriate. Recommended language is provided in the paragraph below. It is suggested that additional considerations for authorship of team member(s) be made in accordance with the International Committee of Medical Journal Editors (ICMJE) Criteria for Authorship. More information about this can be found here.
Acknowledgements: [List the Service(s) Provided – e.g. “Statistical and regulatory”] support was provided by the Anesthesia Research Center within the Department of Anesthesia, Critical Care and Pain Medicine at Massachusetts General Hospital.
This text is recommended but can be customized to comply with journal requirements. Please contact ARC with any questions related to attribution.
Helpful Links
- MGH Research Intranet
- MGH Research Navigator
- Insight
- CITI Online Training
- Harvard Catalyst
- https://clinicaltrials.gov/
- Clinical Trials.gov Login
- Grants.gov
- REDCap
- MGH Funding Opportunities
- MGH IRB Forms
- MGH IRB Instructions for New Applications (see ‘Preparing IRB Submissions for Approval of Human Subjects Research’)
- Reporting Guidelines for Main Study Types (ex. STROBE, CONSORT)
- Partners Logos
- Guidelines on Publication Ethics