An international study of the performance of sample collection from patients


  • W. H. Dzik,

    Corresponding authorSearch for more papers by this author
  • M. F. Murphy,

  • G. Andreu,

  • N. Heddle,

  • C. Hogman,

  • R. Kekomaki,

  • S. Murphy,

  • M. Shimizu,

  • C. T. Smit-Sibinga,

  • the Biomedical Excellence for Safer Transfusion (BEST) Working Party of the International Society for Blood Transfusion

Walter H. Dzik, MD, Blood Transfusion Service, J-224, Massachusetts General Hospital, 55 Fruit Street, Boston, Massachusetts 02114, USA E-mail:


Background and Objectives Collection of a blood sample from the correct patient is the first step in the process of safe transfusion. The aim of this international collaborative study was to assess the frequency of mislabelled and miscollected samples drawn for blood grouping.

Materials and Methods Hospitals in 10 countries provided data on sample error rates during a period of at least 3 months, including the last quarter of 2001. Mislabelled samples were defined as those not meeting local criteria for acceptance by the laboratory. Miscollected samples [wrong-blood-in-tube (WBIT)] were defined as samples in which the blood group result differed from the result on file from prior testing. WBIT rates were corrected for the proportion of repeat samples and for undetectable errors occurring as a result of chance collection of blood from the wrong patient with the same ABO group. Participants also completed a questionnaire on current policies regarding sample collection.

Results A total of 71 hospitals completed surveys describing policies related to sample collection. Sixty-two hospitals provided usable data on the frequency of mislabelled and miscollected samples. Mislabelled and miscollected samples were common. Based on results from over 690 000 samples, the median hospital performance resulted in a rate for mislabelling of 1 in every 165 samples (6·1 per 1000; interquartile range 1·2–17 per 1000). The presence of national patient identification systems in Sweden and Finland was associated with rates of miscollected samples that were too low to estimate. Outside these nations, miscollected samples demonstrating WBIT occurred at a median rate of 1 in every 1986 samples (0·5 per 1000; interquartile range < 0·3–0·9 per 1000). There was great variation worldwide in the reported frequency of mislabelled samples, probably resulting from variation in policies for sample acceptance. Miscollected samples occurred at a more constant rate.

Conclusions The rate of mislabelled samples and miscollected samples is 1000–10 000-fold more frequent than the risk of viral infection. Rates of mislabelled samples and WBIT can be tracked as key indicators of performance of an important step in the clinical transfusion process. WBIT episodes represent important ‘near-miss’ errors. By providing baseline performance data for the collection of patient blood samples, this study may be useful in formulating future national standards of performance for sample collection from patients.