Eleven studies with 2,813 participants (range 61 to 819) were included in the review. None of the included studies described the index test(s) in sufficient detail to permit replication. Blinding of both index test and reference standard interpretation was poorly reported in 10 studies. Over half of the included studies did not adequately describe their selection criteria and/or did not report sufficient details of the reference standard to permit replication.
Pooled estimates of positive likelihood ratios ranged from 0.63 (95% CI 0.49 to 0.80) for vaginal discharge (four studies) to 6.51 (95% CI 4.24 to 10.01) for a positive urine dipstick test for nitrites (three studies). Pooled estimates of negative likelihood ratios ranged from 1.18 (95% CI 1.08 to 1.28) for vaginal discharge (four studies) to 0.44 (95% CI 0.35 to 0.56) for a negative urine dipstick test for leukocytes (four studies).
Dysuria, urgency, nocturia, sexual activity and urgency with dysuria were weak predictors of urinary tract infection, whereas increased vaginal discharge and suprapubic pain were weak predictors of the absence of infection. Positive urine dipstick tests for nitrites or leukocytes (positive LR 1.42, 95% CI 1.23 to 1.57; four studies; Ι²=87.8) were the only findings found to clearly favour a diagnosis of urinary tract infection.
Pooled diagnostic odds ratios were also reported.
Prevalence was not a source of heterogeneity in the regression model.