Cancer pain assessment in clinical trials. A review of the literature (1999-2002)

The aim of this review was to evaluate the methods of pain measurement in controlled clinical trials in oncology published between 1999 and 2002. An electronic literature search strategy was used according to established criteria applied to the Medline database and PubMed search engine. Articles were selected to include only studies that had chronic cancer pain as the primary or secondary objective of a controlled clinical trial. A specific evaluation scheme was used to examine how pain measurement methods were chosen and implemented in the study procedures. The search strategy identified 613 articles, and 68 were selected for evaluation. Most articles (69%) chose unidimensional pain measurement tools, such as visual analogue scales, numerical rating scales and verbal rating scales, whereas others used questionnaires. The implementation of the pain assessment method was problematic in many studies, especially as far as time frame of pain assessment (70%), administration modalities (46%), and use of non-validated measurement methods (10%). Design of study and data analysis were often unclear about the definition of pain outcome measure (40%), patient compliance with pain assessment (98%), and impact of missing data (56%). Statistical techniques were seldom appropriate to the type of data collected and often inadequate to describe the pain variable under study. It is clear from this review that most authors were aware of the need of valid pain measurement tools to be used in clinical trials. However, too often these tools were not appropriately used in the trial, or at least their use was not described with sufficient accuracy in the trial methods.