Steven Stellman - Reliability of smoking history in medical charts in relation to interview data

Version 1

      Publication Details (including relevant citation   information):

      Pollack, D., Jubas, J., Zhang, Z., Stellman, S., Harlap, S.   7 (6) 1379-82-

      Abstract: We previously reported on a series of   studies exploring relationships between cigarette smoking and   mutations of tumor suppressor genes; the studies relying on   medical charts for data on smoking. To assess the accuracy of   these data we compared them with data obtained using a   conventional epidemiologic interview. As part of an ongoing   case-control study of tobacco-related cancers, a trained   interviewer questioned 144 patients about their smoking habits   using a structured interview instrument; the medical charts were   then reviewed. Using the Spearman test to compare the data   obtained from the two sources, the correlation coefficient was   0.85 for cigarettes per day; 0.88 for years of smoking and 0.95   for pack-years, suggesting that the data in the charts were   reliable. The kappa coefficient was 0.96 for current smoking   status, indicating a very strong agreement between the two   sources of data. The sensitivity and specificity for chart review   were 96.3% and 100% respectively, compared with the 'gold   standard' of a structured interview. These results show that when   patients cannot be interviewed, for example when performing   retrospective studies using existing collections of archived   tumors or other biologic materials, then data on smoking   abstracted from medical charts may be a reliable substitute.

      Address (URL): http://www.ncbi.nlm.nih.gov/pubmed/21552976