Urine analysis and decision-making on cystitis in general practice.

Decision-making on cystitis in general practice was studied by addressing 1171 general practitioners of whom 909 responded. A case history was supplied with a test-strip result where leucocyte esterase was 1 + and the nitrite field either positive or negative. The respondents' estimated probability of urinary tract infection from the history had a median at 50% (10 and 90 percentiles, 30% and 90%). The test-strip results changed the estimate to 90% (50% and 100%) in the nitrite-positive group and 50% (10% and 90%) in the nitrite-negative group. Likelihood ratio was calculated from each respondent's estimated pre- and post-test probability to express the test's assumed diagnostic power. The median likelihood ratios were 4.75 (10 and 90 percentiles, 1.00 and 35.70) in the nitrite-positive group and 1.00 (0.11 and 4.00) in the nitrite-negative group. The resulting actions taken by the respondents were: no action (14 respondents, corresponding median estimated post-test probability 27.5%), await further examination (482, 50%) and prescription of antibiotics (366, 90%). Further examination included urine microscopy (13%), bacteriologic culture (21%), or both (65%). Eight-two percent of respondents who decided to prescribe antibiotics would also perform further examinations. In routine urine analysis, 79% of respondents request only test strips, 2% perform microscopy and 19% do both. The estimated probability of urinary tract infection is significantly affected by test-strip results and is important for the choice of further actions. Microscopy is performed more often than recommended in other studies.