Symptom prevalence rates based on patient-reported outcomes

Introduction Within the last two decades the use of patient-reported outcomes (PROs) in oncological trials has increased considerably. Available PRO assessment instruments such as the EORTC (European Organisation for Research and Treatment of Cancer) or the FACIT (Functional Assessment of Chronic Illness Therapy) measurement system describe physical and psychosocial symptom burden on continuous scales. Whereas this is advantageous for many research purposes, it does not allow calculating symptom prevalences based on these measures. Availability of PRO-based symptom prevalence rates, would relate this type of data to a large amount of prevalence data derived from proxy-ratings from physicians. In addition, such figures are easier to understand for researchers and clinicians not being familiar with PRO research and certain PRO instruments. A further use of cut-off scores on which such prevalence rates are based, would be within symptom screening in daily clinical practice. Objectives Based on such considerations the proposed study aims at developing a methodology for establishing prevalence rates for key cancer symptoms (physical functioning, fatigue, pain, and nausea/vomiting) based on cut-off scores for the EORTC measurement system. The development of such cut-off scores to determine prevalence rates will be done separately for patients with solid tumours and patients with haematological malignancies. A further focus will be the analysis of cross-cultural differences in cut-off scores between Dutch and Austrian patients. Finally, change of cut-off scores over the disease trajectory (e.g. due to response shift) will be investigated within this study. Patients and Methods For developing cut-off scores for the EORTC measures for physical functioning, fatigue, pain, and nausea/vomiting, patient ratings on symptom relevance will be collected as external criteria. EORTC measures will comprise scales from the QLQ-C30 and the corresponding computer-adaptive EORTC measures. Developed cut- off scores will be compared to percentiles of score distributions from cancer patient samples with similar characteristics. The study will include a cross-sectional Dutch sample of 300 patients with solid tumours (colorectal, bladder, and lung cancer), a longitudinal Dutch sample of 100 breast cancer patients, and a cross-sectional sample of 600 Austrian cancer patients (300 patients with solid tumours and 300 patients with haematological malignancies). Statistical analysis will comprise mainly Receiver Operating Characteristic analyses to assess diagnostic accuracy of the employed PRO measures, and binary logistic regression to investigate differences in optimal cut-off scores between patient subgroups and changes of optimal cut-off scores over time.

The research project conducted within the Erwin-Schrödinger-Fellowship focused on improving the interpretability of quality of life (QOL) data that is frequently collected in the medical field. QOL data is not only used in studies to evaluate medical treatments, but also in daily clinical practice to obtain important information on the health status and situation of individual patients. The project was conducted within a two-year research stay at the Netherlands Cancer Institute in Amsterdam and a one-year return phase at Innsbruck Medical University. QOL data comprises physical, psychological and social aspects of diseases and treatments and is collected directly from the patient via self-report (so-called patient-reported outcomes). Standardised questionnaires, which are the usual method to assess QOL, allow to calculate score points from individual questions to describe the degree or level of the various symptoms and problems assessed. Such score points allow to discriminate between patient groups with different diagnoses or treatments and provide detailed information on changes over time. However, whereas score points are helpful for comparisons of groups and time points, they are difficult to interpret without the availability of thresholds for clinical importance. Such thresholds are essential to distinguish those problems and symptoms that require a health professionals attention from others. Therefore, this research project aimed at developing such thresholds for two applications, one in oncology, and one in orthopaedics. Thresholds for clinical importance were defined using additional aspects of a symptom or problem (e.g. limitation in everyday life, patient satisfaction, need for help or care). In the study in the field of oncology 548 cancer patients from the Netherlands, Austria, Poland and the UK participated and provided detailed information on the impact of cancer symptoms that allowed to define thresholds for clinical importance for one of the internationally most widely used QOL questionnaires, the Quality of Life Questionnaire Core 30 developed by the European Organisation for Research and Treatment of Cancer (EORTC). The study in the field of orthopaedics used data from 1055 patients with an artificial knee joint, available from the arthroplasty registry in St. Gallen (Switzerland), to define thresholds for treatment success with regard to pain, function and QOL.The development of such thresholds increases the interpretability of patient-reported outcomes and therefore facilitates the use of questionnaires for the evaluation of diseases and medical treatments. This strengthens the integration of the patients individual perspective into medical research and daily clinical practice. Taking the patients perspective better into account provides more comprehensive and valid information on the consequences of medical treatments.