Improving reporting of ICU outcome data

Clinical studies reporting outcomes after intensive care admissions are often published, including in this journal 1 . This is a field with much research activity, and with widespread clinical interest. For observational cohort studies, it is strongly recommended that authors form their reports using guidelines presented by the STROBE (Strengthening the reporting of observational studies in epidemiology) group 2 , and these recommendations are designed to strengthen the report based on their observations.


Improving reporting of ICU outcome data
Clinical studies reporting outcomes after intensive care admissions are often published, including in this journal. 1 This is a field with much research activity, and with widespread clinical interest. For observational cohort studies, it is strongly recommended that authors form their reports using guidelines presented by the STROBE (Strengthening the reporting of observational studies in epidemiology) group, 2 and these recommendations are designed to strengthen the report based on their observations. This guideline reminds authors what to include in the different parts of the manuscript, focusing on key elements. However, STROBE does not describe in detail what to report from the clinical settings and what variables that is necessary in order to interpret the results in a correct manner. Such more detailed guidelines for clinical studies are emerging now for some specific categories of patients. 3 Such detailed guidelines will obviously also be important for journals, editors, and reviewers in their assessment of manuscripts. An obvious question follows: how rigorously do authors and peer-reviewers use these detailed guidelines? While it is relatively simple to state that one or another reporting recommendations or study design recommendations were followed, how rigorously are the recommendations actually followed is another matter.
The STROBE recommendations do not specify (for example) which variables-outcomes, exposures, predictors, confounders or effect modifiers-to include for a specific study question. Nor is it specified exactly what authors should do to address potential sources of bias.
But these critical elements need to be present if the report is going to be able to answer the study question as best it can. This is true for observational studies in intensive care medicine, as well.
Why is it important to raise awareness of the content, or critical elements, of outcome studies from intensive care? Broadly, such outcome studies can be divided into 2 types: one with reporting of objective data as survival, length of stay, duration of mechanical ventilation for example, and the second with studies analysing patient-reported outcomes like quality of Life. This editorial note will focus on the first type.
The quality and hence impact of many studies on ICU outcomes are often modest, usually because the findings are not including important clinical data either on outcome, or more commonly they lack factors needed to understand why outcomes become as they are reported. As a part of the quality process for reporting such analyses, it is required that a minimal data set is studied and described properly. This can be difficult in retrospective studies, since the investigators often have only limited access to the data which often also are missing. Such data can be a description of the severity of illness, which is extremely important if the study reports outcomes for mortality or survival. Luckily, in most Nordic ICUs collection of such data is mandatory for reporting to national ICU process registries in Finland, Sweden, Denmark, and Norway. However, not all ICUs contribute to such registries, and certainly not elsewhere in Europe. Therefore, a retrospective study on objective outcomes without any severity data will be difficult to compare to similar studies using the same cohorts. Also, case-mix is important. Far too often, all kinds of ICU admissions are added in one large "melting pot" of a cohort. This lack of refinement of cohorts can be misleading. A good example is planned vs unplanned ICU admission.