There were three main results of this study. First, experts listed a large number of baseline characteristics that described patients with osteoarthritis of the knee included in trials that evaluated treatment effects. Second, experts agreed on the relevance of only one baseline characteristic. All other baseline characteristics received ratings scattered over a broad range, which indicated disagreement among experts. Third, the relevance of baseline characteristics varied according to the outcome measure in a trial.
Researchers have published a number of relevant articles that emphasized the definitions and measurements of outcomes in clinical trials that evaluated treatment effects in patients with knee osteoarthritis [6–9]. Despite a thorough search in various databases, we could not find any publications that focused on how to select baseline characteristics of patients that participated in trials on osteoarthritis of the knee. However, we identified a few publications that summarized the evidence for prognostic factors that characterized patients with knee osteoarthritis. Cheung et al.  stated that strong or moderate evidence indicated that progression was associated with age, generalized osteoarthritis, knee malalignment, and serum hyaluronic acid concentration; limited evidence indicated associations with knee pain, synovitis, the adduction moment of the knee, vitamin D and C concentrations, and MRI bone marrow lesions in the knee; and conflicting evidence indicated associations with body mass index, initial severity of x-ray changes, cartilage oligomeric protein (Comp), and urinary CTX-II. In a recent systematic review, Chapple et al.  reported some of the same results. They found that age, generalized osteoarthritis, varus knee alignment, and radiographic features, particularly joint space narrowing were strongly associated with prognosis. The latter review  provided no specific statements about the prognostic relevance of serum hyaluronic acid concentration.
In part, our results were in agreement with the previous studies [1, 10]; but in part, our findings disagreed with those studies. For example, the panel members of our survey considered psychosocial factors important, e.g., anxiety and fear; however, the supporting evidence for these factors appeared to be scarce. The most striking discrepancy was the difference between the number of prognostic factors gathered from the synthesis of original studies and the number collected from the clinical experts of the present study. The clinical experts listed a much higher number of relevant factors than the numbers listed in the current literature.
The results of our survey might be helpful for clinicians and researchers. This study aimed to provide guidance to clinicians for assessing the applicability of trial results to a different clinical application. After reading the results of a clinical trial, the main task of the clinician is to assess which patients might benefit from the treatment. Apart from the inclusion/exclusion criteria, the most significant information for this assessment are the baseline characteristics of study participants. The present study provides a list of relevant factors based on clinical expert opinions. Clinicians can consult this list to evaluate the comprehensiveness of the baseline characteristics in the reports they are considering.
Researchers may also find this list of baseline characteristics important for two reasons. First, our results may inform the design of future trials in patients with knee osteoarthritis. Researchers can consult the present list of baseline characteristics for each outcome of interest to decide which patient characteristics should be reported. The careful selection and reporting of baseline characteristics can facilitate the translation of research results into patient care, and this increases the usefulness of trial results. Second, researchers may find the list relevant when synthesizing the results of original studies. Guidelines for preparing systematic reviews by meta-analyses recommend checking the comparability of patient populations between original studies before pooling the results to derive a single value [2, 4, 11]. A prerequisite for this type of assessment is the availability of detailed information about the distribution of baseline characteristics among the patients included in the original studies.
Our study had both strengths and limitations. The primary limitation, inherent in most surveys, was that a different panel of experts may provide different results. A strength was that the members of the panel were experts in the field and had authored two or more clinical trials that evaluated the effects of treatments for patients with osteoarthritis of the knee. Furthermore, we included a large number of panel members, and they were from different countries. A panel with about 15 members is recommended for surveys to reach a consensus or to assess the degree of disconsensus . With 23 panelists, we exceeded that recommended number. The panel member internationality assured a broad spectrum of opinions and eliminated the domination of an opinion based on a single clinic or region-specific beliefs. A further limitation of our study was that, in the first questionnaire, we only included pain-related and functional outcomes, but no structural outcomes. In the second questionnaire, we included the structural outcomes. We assume that the addition of an outcome parameter did not impact the results different results.