Article Text
Abstract
Introduction/Background Nausea and vomiting are one of the most common and challenging side effects related to chemotherapy. The aim of the study was to develop a predictive score for chemotherapy-induced nausea and vomiting (CINV) in patients with gynaecological cancers planned for chemotherapy by identifying non-pharmacological, patient-related risk factors.
Methodology A research-based questionnaire of 27 risk factors was designed and handed out to chemotherapy-naïve patients with gynaecological malignancies. Data on nausea and vomiting from at least 3 cycles of therapy was collected. Variable selection via stepwise and LASSO regression combined with patients’ history was used to determine few questions with high predictive power. Bayesian logistic regression (risk prediction model) was implemented with a cut-off chosen to reach a sensitivity of 80%. Area under the curve analysis (AUC) was performed and the accuracy of prediction calculated.
Results 191 patients were enrolled, of which 174 (91.1%) received at least one dose of chemotherapy (intention-to-treat population). Most patients suffered from ovarian cancer (68.0%) and received the carboplatinum/paclitaxel chemotherapy combination (57.5%). Leading predictive factors for CINV were educational status, nausea and vomiting due to other medication, motion sickness, anxiety from therapy in general, anxiety from nausea due to therapy, emetogenic potential of the therapy and distress level. 142 (81.6%) patients answered all questions concerning these factors. Among those, 107 (66.0%) were affected by nausea or vomiting. The AUC of the predictive score based on the above mentioned factors was 0.727 (95% CI [0.636, 0.818]), with a sensitivity of 80.4% [72.9%, 87.9%], a specificity of 48.6% [31.4%, 65.7%] and an overall accuracy of 72.5% [65.5%, 79.6%].
Conclusion To this day, a patient-related predictive model for the occurrence of CINV is missing, making the choice of the right antiemetic prophylaxis difficult. The score featured in our study showed very promising predictive power and is currently being validated.