New guidelines aiming to improve tools used by doctors in making diagnoses and treatment decisions are published today in the Annals of Internal Medicine and 10 other journals.
The TRIPOD guidelines were developed by a consortium of international investigators, led by researchers from the University of Oxford and the University Medical Center (UMC) Utrecht in the Netherlands, alongside health care professionals and journal editors. TRIPOD stands for Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis.
Clinical prediction models guide medical practitioners and patients through diagnostic and treatment decision-making. Such models are in fact tools to convert multiple pieces of information from a patient or individual into a probability that a specific condition is present or may occur in the future. The research that has led to these prediction models is generally reported in the medical literature. For medical practitioners to make accurate predictions and informed decisions, it is highly important that such research is transparently reported in medical journals.
The TRIPOD guideline aims to enhance this transparent reporting. It consists of a checklist of 22 items for reporting research on so-called prediction models which aim to help clinical decision-making. The checklist is designed to allow researchers, peer reviewers and journal editors make sure a set of minimum criteria are included a published article.
Gary Collins, Associate Professor at the Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences at Oxford University and lead researcher in TRIPOD, says: "We know clinical prediction models are being developed and used in all areas of medicine, with an ever increasing number being recommended in clinical guidelines. These models are being used to support clinical judgment. For example, NICE guidance recommends that people with a 10-year risk of developing cardiovascular disease in excess of 10% should be prioritised to be prescribed statins and/or blood pressure lowering drugs - this estimation of risk is done using a prediction model. It is therefore crucial that before using one of these models on individuals or patients, all aspects of the studies describing these models are accurate, complete and transparent so that methodological rigour, accuracy and applicability can be examined. We hope that journals will rapidly adopt and endorse the TRIPOD Statement."
Karel Moons, Professor of Clinical Epidemiology of the Julius Center for Health Sciences and Primary Care, UMC Utrecht, The Netherlands, adds: "The past decades have seen a wild growth in clinical prediction models, across the entire medical field. But the scientific evidence whether these models are accurate enough to use them in our patients is in the vast majority lacking and certainly unclear reported. This should change quickly: besides that these models are often recommended in clinical guidelines as Gary says, it worries me even more that they show up almost daily on websites and in medical apps. Everyone can thus use these prediction models to ‘quickly’ calculate their risk of having or developing, for example, a osteoporosis, diabetes, colon cancer, heart failure, or depression. We have no clue whether the underlying models were correctly developed let alone properly tested. When researchers, editors and guideline developers take notice and ideally follow our guideline, it will become transparent for health care professionals, patients and other users, to judge whether published prediction models are indeed accurate, useful and trustworthy. It will separate the chaff (bad models) from the wheat (good models). Moreover, comprehensive and honest reporting facilitates others to check what was done, confirm or falsify the study results, and to detect misconduct."
Currently, the quality of reporting of prediction model studies in medicine is generally poor, across different disease areas and different journals, as shown by a number of scientific reviews. Poor reporting ultimately leads to weak prediction models that are not or should not be widely implemented or used in clinical practice.
The TRIPOD researchers intend their Statement to change the landscape of clinical research reporting - and therefore research design - in the coming years. This will ultimately lead to better health care by improving the clinical prediction models used and their uptake by medical practitioners, say the researchers. TRIPOD will also help other scientists understand what has been done, provide sufficient information for others to replicate the study, identify scientific misconduct, and improve accountability.
To encourage dissemination of the TRIPOD Statement, the article is also published in BJOG: An International Journal of Obstetrics & Gynaecology, British Journal of Cancer, British Journal of Surgery, BMC Medicine, British Medical Journal, Circulation, Diabetic Medicine, European Journal of Clinical Investigation, European Urology, and Journal of Clinical Epidemiology.
The Oxford University and UMC Utrecht researchers have previously been involved in developing similar reporting guidelines for clinical trials, observational studies, systematic reviews, animal research and diagnostic test research, called CONSORT, STROBE, PRISMA, ARRIVE and STARD, respectively. These guidelines are changing the quality of research reporting found in the literature.
The TRIPOD guidelines are also part of the EQUATOR Network, an international initiative that seeks to improve the reliability and value of published health research literature by promoting transparent and accurate reporting and wider use of robust reporting guidelines.
“The EQUATOR Network’s online platform provides easy access to all guidelines and other resources supporting scientists in responsible publication of their research. The TRIPOD guidelines are a valuable addition to our collection”, says Iveta Simera, Head of Programme Development for the EQUATOR Network. “Following reporting guidelines when writing research manuscripts is a simple, cost effective solution for improving completeness, accuracy and usability of medical research papers. In the current financial climate it is unjustifiable not to implement something so simple that can substantially improve the quality of research output and increase its benefits to patients.”
Pictured: Associate Professor Gary Collins