A PROPOSAL TO USE CONFIDENCE-INTERVALS FOR VISUAL ANALOG SCALE DATA FOR PAIN MEASUREMENT TO DETERMINE CLINICAL-SIGNIFICANCE

Visual analog scales (VAS) ranging from 0 cm (no pain) to 10 cm (worst imaginable pain) are used widely for pain measurement, but various in vestigators have not treated these data consistently. Conventional sta tistical tests of such data, although evaluating the 'statistical sign ificance' may obscure the clinical value of a treatment. On the other hand, confidence intervals (CIs) can illuminate both statistical and c linical importance. CIs give a range of values based on the observed d ata which contain, with a specified probability, a true but unknown va riable typifying a population. We reviewed 112 articles published rece ntly in anesthesia journals for statistical reporting of VAS data. Of the 112 articles, only two used CIs to report mean pain scores and one used CIs to report differences in median pain scores between the stud y groups. Only two articles presented 95% CI for the mean pain scores graphically Analgesic techniques that produce VAS values in the range of 0-3 have been reported to represent adequate analgesia. A graphical method using CIs is proposed that allows ready interpretation of VAS data. With this approach, one evaluates whether the 95% CI for the mea n pain score in a group during a particular period lies entirely withi n the zone defined as ''analgesic success'' (0-3). Such an analysis al lows a visual assessment of whether a particular technique would produ ce clinically important effects in the population at large. This appro ach seems to provide more information than the use of conventional hyp othesis testing in the interpretation of VAS data for pain measurement .