A scoping review of ICHOM sets identifies a small number of commonly recommended baseline characteristics

The capture of patient baseline characteristics in randomized controlled trials (RCTs) is essential for assessing external validity and for exploring treatment effect heterogeneity. Yet, the capture of such information is highly inconsistent across trials with limited harmonization efforts to date. The International Consortium for Health Outcomes Measurement (ICHOM) has issued expert recommendations for baseline characteristics to capture in clinical practice, as part of Sets of Patient-Centered Outcome Measures ("Sets"). We reviewed which characteristics were recommended across different conditions and how they were to be measured.

We systematically retrieved and included all ICHOM Sets that were published in peer-reviewed journals and issued measurement recommendations for adult populations. For each Set, we extracted measurement recommendations pertaining to patients' individual baseline characteristics ("case-mix factors") from the Set's data collection reference guide. We used inductive thematic coding to harmonize variable names and group them into thematic factors and domains. Two reviewers independently coded factors and resolved disagreements. We also extracted the recommended approach to measuring each factor.

We identified 32 eligible ICHOM Sets covering a range of conditions that make important contributions to the global disease burden (eg, coronary artery disease, diabetes, depression). These Sets identified 616 original baseline variables for measurement, which we mapped onto 89 thematic factors. We identified factors that were repeatedly recommended across conditions, with a focus on sixteen factors appearing in at least 15% of Sets. These included demographics such as age, sex, gender, race/ethnicity; clinical factors such as comorbidities and disease onset; and psychosocial factors such as education, tobacco and alcohol consumption, employment, relationship status, and physical activity. For most of these factors, ICHOM recommended brief assessments via one or two questions. Only for comorbidities and physical activity did the recommended approaches include multi-item instruments, namely the self-administered comorbidity questionnaire and the International Physical Activity Questionnaire Short Form, respectively. Only six factors were recommended in more than 50% of Sets.

There is scope to harmonize baseline measurement in RCTs both within conditions, drawing on clinical and physiological patient characteristics with condition-specific relevance, and across conditions, based on recurrent demographic and psychosocial factors.

Clinical trials need to collect clear and consistent information about the people who take part. This information is essential for understanding who the trial findings apply to and whether different groups of people may respond differently to a treatment. However, this type of information is currently collected in very inconsistent and incomplete ways across trials. In this study, we examined recommendations for patient characteristics to capture that have been developed by expert groups convened by the International Consortium for Health Outcomes Measurement (ICHOM). Although these expert recommendations were designed for measurement in everyday clinical settings rather than research studies, they can potentially inform future efforts to identify similar recommendations specifically for research studies. We reviewed 32 sets of ICHOM measurement recommendations that covered a range of different health conditions. Across these Sets, we found 616 individual baseline variables, which we grouped into 89 thematic factors. Many of these factors were specific to particular diseases, for example clinical measurements or the clinical history of a patient's condition. However, we also identified some demographic and psychosocial factors that were recommended repeatedly for measurement in the context of different conditions. Sixteen such factors were found in at least 15% of Sets, including, for example, age, sex and gender, race and ethnicity, co-occurring health problems, when the disease began, education level, smoking and alcohol use, employment and relationship status, living situation, and physical activity. Most of this information can be obtained from patients using only one or two questions. These findings suggest that it may be possible to develop guidance on what patient information to collect in clinical trials and how, both in terms of information that is relevant for specific conditions and information that has general relevance across diseases. If differences in measurement approaches could be reduced, this would make it easier to interpret, compare, and combine research findings, and could improve our understanding of which treatments work best for which groups of patients.