How to write a research paper

Abstract

It is not an easy task to perform and report a good study or review, and therefore quite a number of papers have been published on presenting and explaining guidelines on how to optimally do this. In line with this topic, it is also useful to pay attention to the craft of writing a scientific paper in general. Indeed, even if a study has been appropriately conducted and technically well presented, it may have been written in such a way that its message will not be recognized [1,2]. In this issue, after an introductory paper by Kotz et al, Kotz and Cals publish the first of a series of monthly compact one-page papers, each highlighting an essential step in preparing and writing a research paper. This series, containing a total of 12 one-pagers, originates from a PhD student course organized at Maastricht University, and is especially recommended to young investigators who would appreciate efficient guidance based on extensive practical teaching experience. But senior authors may also find useful writing tips in this series. Diagnostic and prognostic research is a major topic in this issue. In a Commentary, Weiss analyzes the opportunities and challenges in studying the relationship between test results and the effectiveness of treatment. The author presents a broad methodologic overview, including both clinical epidemiological and ethical issues. Austin and his group compared the performance of conventional classification and regression trees, including logistic regression, with modern flexible tree-based methods from the data-mining and machine-learning literature, in predicting and classifying heart failure (HF) patients according to subtypes. It turned out that each of these two approaches had specific strong points in different classification tasks. Dataanalytic work on diagnostic performance has also been conducted by Spruijt et al, who studied how vital signs such as heart and respiratory rates should be included in clinical prediction models for serious bacterial infections in febrile children. Using data from a large prospective observational study of febrile children, they compared various ways to handle these rates as predictors, and concluded that maintaining them as continuous variables results in a better predictive ability than dichotomization. Simel and co-authors present a simple method to calculate sensitivity, specificity, and likelihood ratios when, in studies of diagnostic tests, the odds ratios and marginal values in a 2 2 tables are given. This can help to retain studies in meta-analyses of characteristics of diagnostic tests when only the odds ratio is reported.