No abstract

Clinically significant bacteraemia in adults is characterized by low numbers of circulating bacteria. Assuming a Poisson or a binomial distribution we have calculated the probability of detecting bacteria as a function of the concentration in blood, estimated the concentration of bacteria in blood from a given test result, and calculated the blood volume required to detect bacterial growth with a probability of 95% at a given mean number of colony‐forming units (cfu) per sample unit. These theoretical assumptions have been used on an empirical population of patients with proven bacteraemia. Results: 18% of Staphylococcus aureus bacteraemias and 29% of Escherichia coli bacteraemias have circulating bacteria with an estimated spread of less than 0.04 cfu/ml. With a 95% probability of detection of a bacteraemia, a concentration in blood corresponding to 3 cfu/sampling unit is necessary. In our empirical material, where 30 ml was cultured, the probability of detection of E. coli bacteraemias would have decreased by 11% if 20 ml had been cultured, and 27% if only 10 ml had been cultured. The corresponding figures for S. aureus were 6% and 15%, respectively. For low grade E. coli bacteraemias (<0.04 cfu/ml) the decrease would have been 33% and 67%, respectively.

Appropriate quality assurance should focus on patient value rather than technical details to provide an appropriate clinical service.

Blood volume is the most important variable for the detection of microorganisms in blood cultures (BCs). Most standards recommend 40 to 60 ml blood, collected in several BC bottles filled up to 10 ml. We measured blood volume in individual BC bottles and analyzed the associations of hospital, bottle type, day of the week, daily sampling time, and age and sex of the patient with sampling volume and BC result. The variation in blood volume per BC bottle was analyzed in a mixed linear model using hospital, bottle type, weekday, sampling time, age, and sex as fixed factors and patient identification (ID) and episode as random factors to control for repetitive sampling of individual patients. Only 18% of all bottles were filled with the recommended 8 to 10 ml, and 47% were filled with less than 8 ml. The mean (± standard error) volume was larger in positive bottles (9.09 ± 0.15) than in negative bottles (8.47 ± 0.07) (P < 0.001). Blood volume was larger in BacT/Alert-FA Plus bottles than in -FN Plus BC bottles (P < 0.001). There were significantly lower volumes collected during the night (P < 0.001). The volume of blood collected decreased significantly with increasing patient age (P < 0.001). Larger volumes were collected from male patients than from female patients: 8.78 (± 0.06) versus 8.36 (± 0.06) ml (mean ± standard error [SE]), respectively (P < 0.001). The odds of detecting a positive patient increases by 13% for each additional milliliter of blood drawn. Our results show that we need to work actively with the development of blood sampling routines to overcome age and sex effects and to optimize blood sampling volumes.