Recently, Broers and colleagues1 reported that injecting drug users were as compliant with zidovudine as patients with human immunodeficiency virus infection in other risk groups.1 The investigators chose two compliance measures for their study: estimation of compliance by the physician treating the patient and the rise in mean corpuscular volume (MCV) of erythrocytes.
Numerous studies over the last 30 years have compared physician assessments of compliance with objective measures such as pill counts or electronic medication monitors and have demonstrated that physicians are inaccurate judges of their patients' compliance.2-6 In these studies, the sensitivity of physician judgments of reduced compliance has ranged from 30% to 65% and specificity from 43% to 68% (likelihood ratios, 0.9 to 1.1).3 Product-moment correlations (Pearson or Spearman r) between physician estimates and compliance as measured by pill counts or electronic medication monitors have ranged from 0.01 to 0.48.24"6 Physicians have