The objective of this study is to determine the reliability of the Ame
s automated method as a screening procedure for the detection of red b
lood cells and leukocytes. 5486 urines were analyzed by an automated d
ipstick technique. 3127 urines were completely negative on the stick a
nd 2359 showed a positive finding. All 2359 dipstick positive urines a
nd 456 of the 3127 dipstick negative urines (total of 2815) were read
blindly by one of two nephrologists working on different days, 1743 by
M. K. and 1072 by P. S. Red cells, leukocytes and casts were enumerat
ed per high power field and compared to the dipstick findings. Urines
that were negative by dipstick for protein, blood leukocytes, nitrites
, glucose and ketones were generally negative on microscopic examinati
on, with only 5.3% having any abnormality. Urines positive for one or
more of these findings correlated poorly with the microscopic findings
due to the number of false positive and false negative dipsticks for
red cells and leukocytes. Sensitivities were 75.3% and 81.0% and speci
ficities were 88.6% and 64.3%, for red cells and leukocytes, respectiv
ely. Increasing the dipstick cutoff point improved sensitivity but low
ered specificity. Urines positive for glucose and ketones had lower se
nsitivities for the detection of leukocytes, but not urines positive f
or protein. Microscopic examination should be done on dipstick abnorma
l urines. This may not be necessary for negative urines unless specifi
cally indicated for clinical reasons. Urines positive for ketones or g
lucose should be screened systematically since they give a larger prop
ortion of false negatives for leukocytes.