Do not get burnt by damn statistics

Mary Schmich’s (1997) Chicago Tribune column provides advice about how to live a happier life and avoid common frustrations. These words were later spoken by Lee Perry in Baz Luhrmann’s 1999 song ‘Everybody's Free (To Wear Sunscreen)’ also known as ‘The Sunscreen Song’ (Schmich et al., 1997). Much of Schmich’s advice is sound. Especially about wearing sunscreen!! However, when it comes to a journal’s authorship guidelines ‘Read the instructions even if you don’t follow them’ is NOT advice you should adhere to.

A fundamental element of research is advancing knowledge. This includes developing and testing hypotheses. It is expected that Journal of Child Health Care (JCHC) published papers will report in detail their methods and assumptions of analysis. Papers should also discuss their findings in the context of available literature with some author critique. Then JCHC’s audience can use their expertise to critically interpret implications of results and any associated uncertainties. To aid with this there is a requirement for consistent reporting to an audience who is primarily comprised of our expert peers.

Reporting of statistical methodology, including design, analysis, and presentation of results obtained varies considerably. This can make reading of manuscripts and interpretation of results incredibly difficult for JCHC’s readers. Particularly for those who are not statistically minded. Hence, there is a need to present all statistics in an understandable and readable format.

Statistical methods employ extensively accepted tools for analysing data and drawing appropriate conclusions (Charan and Saxena, 2012). However, inappropriate statistical methods may lead to unsuitable or false conclusions (Shann, 1996). An example is when analysing data measured on a continuous scale, applying a parametric test, such as the ubiquitous Students t-test, might produce a false result if the data is not normally distributed (Charan and Saxena, 2012). Similarly, over reporting of probability values, rather than 95% Confidence Intervals, can lead to misleading results (Harrington et al., 2019).

Lack of relevant statistical information can lead to review process delays. If information is unclear, or missing, there is a need to request clarifications. This may require additional information being sought from authors. This increases a manuscripts review time. These delays can often include a requirement to identify new reviewers. This can lead to interruptions in a manuscript’s assessment with subsequent frustrations for all concerned.

To assist authors, and speed-up the review process, JCHC has recently updated its statistical guidelines (JCHC, 2020). These guidelines are both a requirement and a recommendation for reporting of a study’s quantitative component, trial or other investigation. Though they should also be used for qualitative or mixed methods studies, and reviews, that are reporting numbers.

JCHC’s updated statistical guidelines aim to harmonise and standardise reporting of quantitative components of a study for the benefit of JCHC, reviewers, editors, authors and most importantly its readers. They are designed to improve the quality, openness, and transparency of studies JCHC reports. This is particularly relevant given the amount of sometimes spurious information that statistical software packages produce following each analysis.

‘Good statistical practice is fundamentally based on transparent assumptions, reproducible results, and valid interpretations’(Committee on Professional Ethics of the American Statistical Association, 2018: 1). JCHC’s statistical guidelines provide a concise and practical overview of general and specific principles relevant to reporting of statistical work. It is imperative that authors reporting is consistent with these guidelines. This can begin with the design of the study or trial, estimation of the sample size to identify a desired clinical effect, participant randomisation, data collection and finally data analysis (Charan and Saxena, 2012). Then in turn these results form a basis on which conclusions can be drawn.

JCHC’s updated statistical guidelines provide clear and detailed information and recommendations regarding information required. It is important that the authors provide all relevant information. Clear and concise reporting, including all relevant information, can help to shorten the review process and lead to quicker publishing.

A key change to JCHC’s revised statistical guidelines is that if requested, as a condition of acceptance, authors will make data available. In return JCHC’s duty is to ensure that data is treated confidentially and returned to the author once examined.

JCHC’s statistical guidelines have been clarified and expanded from our original guidelines. For authors they cover both presentation and reporting of terminology, methods, and results. These updated guidelines highlight many aspects of study reporting and include:

Manuscripts should transparently describe all study designs, statistical methods, any assumptions, and detail software used.

Preference for replacing probability (p) values with 95% confidence intervals.

Consistency of reporting results and data. This includes notations for figures and tables.

Report missing values and any assumptions made and methods used when simulating missing data.

Use of an appropriate measure of central tendency.

Consideration to number of decimal/significant digits that are being reported and ease of reading these numbers.

JCHC’s mission is to focus on issues related to the health and health care of neonates, children, young people, and their families. This includes areas such as illness, disability, complex needs, well-being, quality of life, and mental health in a diverse range of settings. JCHC’s updated statistical guidelines will assist authors, reviewers, and readers to publish, review and interpret submitted manuscripts. Above all, we encourage future authors to read and follow our new statistical guidelines.

… and of course, to wear sunscreen!