Clinically important changes in acute pain outcome measures: a validation study

J Pain Symptom Manage. 2003 May;25(5):406-11. doi: 10.1016/s0885-3924(03)00162-3.


The purpose of this study was to validate the changes in acute pain measurement scales that are most strongly associated with a patient-determined indicator of clinical importance. Measures of pain intensity and pain relief are commonly used outcomes in therapeutic clinical trials. Recent studies of the properties of acute pain measures have provided data defining the cut-off points that are best associated with clinically important differences. Validation of these findings in another clinical trial data set is important. Data were obtained from the titration phase of a recently conducted randomized controlled clinical trial of oral transmucosal fentanyl citrate (OTFC), which compared OTFC to immediate release morphine sulfate (MSIR) for the treatment of cancer-related acute breakthrough pain. Changes in pain intensity and pain relief were recorded every 15 minutes for 60 minutes and global medication performance recorded at the end of each study pain episode. At any titration step, if the patient felt that the first dose of the study medication did not provide adequate relief within 30 minutes, an additional rescue medication could be taken. To find the level of each pain scale best associated with this measure of the adequacy of pain relief, the calculated sensitivity, specificity, and accuracy for different cut-off points of the measured pain scales were compared to whether or not the patient needed rescue medication. The overall ability of the pain measures to discriminate episodes for which a rescue was not needed was calculated using area under the receiver operating characteristics (ROC) curves. Data were analyzed from 134 OTFC-naive patients who collected data on 1307 episodes of breakthrough pain. Using the criteria of a balanced sensitivity and specificity, the best cut-off points were determined to be: 33% for the percent pain intensity difference; > or =2 for the raw pain intensity difference on a 0-10 numeric rating scale; > or =2 (i.e., moderate or better) for pain relief; > or =33% for the percent maximum total pain relief; and > or =2 (good or better) for global medication performance. ROC area under the curve ranged from 0.839 to 0.862 for each of the pain measures listed above, calculated at 60 minutes. These data indicate that the pain scale cut-off points that are best associated with a patient-derived measure of a clinically important difference closely approximate those found in an earlier study. ROC analysis provided evidence that the overall pain measures were strongly associated with not requiring an "additional dose of rescue medication." Thus, the cut-off points determined for these pain scales provide a good surrogate measure of a patient-determined clinically important response. This provides support for the usefulness of these values in future clinical trials of pain therapy.

Publication types

  • Research Support, U.S. Gov't, P.H.S.
  • Validation Study

MeSH terms

  • Acute Disease
  • Administration, Buccal
  • Analgesics, Opioid / administration & dosage
  • Analgesics, Opioid / therapeutic use*
  • Fentanyl / administration & dosage
  • Fentanyl / therapeutic use*
  • Humans
  • Neoplasms / complications*
  • Outcome Assessment, Health Care*
  • Pain / drug therapy*
  • Pain / etiology*
  • Pain Measurement*
  • Reproducibility of Results
  • Sensitivity and Specificity


  • Analgesics, Opioid
  • Fentanyl