By Mary Gabb (

Writing a manuscript can be daunting, even for a seasoned researcher. How does one synthesise a coherent publication out of the mountain of accumulated data?

The most traditional approach is to start with the IMRaD principle: Introduction, Methods, Results, and Discussion. But how to summarise all of those results?

This is a common challenge, but it can be overcome with a straightforward approach and by being mindful of two key principles:

  • Always keep in mind the original research question. All of the data you choose should be able to answer the original research question. Data that do not answer the primary question should not be included (although if secondary endpoints or hypotheses were tested, relevant data for those questions can be included, depending on your word count limits).
  • Be as clear and transparent as possible. After reading your manuscript, the reader should be able to duplicate your study design and results. Regarding the data, be sure to include:

• All data related to the study sample (eg, race/ethnicity, education, socioeconomic status, etc)
• Statistical analyses
• Characteristics of diagnostic tests
• Study limitations (eg, sample size, type of data, low response rate).

It is also important to present your data clearly:
• Write in short, direct sentences
• Try to present as much data as possible, and in short tables and figures
• Report data comparisons clearly (for example, actual “raw” data as well as absolute or relative risk reduction, confidence intervals as well as P values).

Finally, consult checklists used by journal editors, such as the Consolidated Standards of Reporting Trials (CONSORT), the Quality of Reporting of Meta-analyses (QUOROM), the Meta-analysis of Observational Studies in Epidemiology (MOOSE), and the Standards for Reporting of Diagnostic Accuracy (STARD).