Biomedical Informatics 4th Edition by Edward Shortliffe, James Cimino

Exercise 6

To decide whether a patient has a signifi cant urinary tract infection, physicians commonly use a calculation of the number of bacterial organisms in a mil liliter of the patient's urine. Physicians generally assume that a patient has a uri nary tract infection if there are at least 10,000 bacteria per milliliter. Although laboratories can provide such quantifi -cation with reasonable accuracy, it is obviously unrealistic for the physician explicitly to count large numbers of bac teria by examining a milliliter of urine under the microscope. As a result, one article offers the following guideline to physicians: "When interpreting … microscopy of … stained centrifuged urine, a threshold of one organism per field yields a 95 % sensitivity and fi ve organisms per field a 95 % specifi city for bacteriuria [bacteria in the urine] at a level of at least 10,000 organisms per ml." (Senior Medical Review 1987, p. 4)
(a) Describe an experiment that would have allowed the researchers to determine the sensitivity and specificity of the microscopy.
(b) How would you expect specifi city to change as the number of bacteria per microscopic fi eld increases from one to five?
(c) How would you expect sensitivity to change as the number of bacteria per microscopic fi eld increases from one to five?
(d) Why does it take more organisms per microscopic fi eld to obtain a specifi city of 95 % than it does to achieve a sensitivity of 95 %?