Table 1.

Diagnostic Accuracy of Clinical Features for Urinary Tract Infection (Summary Estimates)

Clinical FeatureNo. StudiesaNSummary Sensitivity, % (95% CI)Summary Specificity, % (95% CI)Summary LR+ (95% CI)Summary LR− (95% CI)
Red Flags (LR+ ≥4) or Features (LR− <0.25)
Hematuria  3  5,6154 (2-8)99 (97-100)4.23 (1.71-10.44)0.97 (0.94-1.00)
Cloudy urine  4  3,86669 (30-92)85 (72-92)4.55 (3.73-5.56)0.36 (0.13-1.02)
Malodorous urine  4  7,10931 (12-59)93 (75-98)4.13 (2.27-7.49)0.75 (0.58-0.98)
No circumcision  8  6,71288 (52-98)52 (23-80)1.81 (1.15-2.87)0.24 (0.08-0.72)
Amber Signs (LR+ 2-4 or LR− 0.25-0.5)
Dysuria  7  5,41340 (19-66)88 (80-93)3.28 (2.22-4.86)0.68 (0.47-1.00)
Frequency  4  5,66836 (22-53)84 (74-90)2.21 (1.78-2.75)0.76 (0.65-0.90)
Bed wetting  3  5,43621 (12-32)92 (84-97)2.70 (1.46-4.99)0.86 (0.78-0.95)
Previous UTI  7  7,54615 (9-24)94 (88-97)2.31 (1.73-3.10)0.91 (0.86-0.96)
No source of infection  420,96484 (74-91)45 (18-75)1.53 (0.92-2.54)0.35 (0.22-0.55)
Signs With LR+ <2 or LR− >0.5
Diarrhea  724,64020 (12-30)78 (73-83)0.91 (0.68-1.22)1.03 (0.96-1.10)
Vomiting  710,50527 (19-38)69 (61-76)0.89 (0.74-1.06)1.05 (1.00-1.12)
Abdominal pain  6  5,39729 (14-51)84 (64-94)1.86 (0.82-4.22)0.84 (0.67-1.07)
No cough  420,94681 (33-97)32 (7-76)1.19 (0.93-1.52)0.61 (0.34-1.07)
Irritability  5  5,39515 (4-48)85 (63-95)1.00 (0.67-1.48)1.00 (0.93-1.07)
Abnormal appearance  426,52536 (17-60)70 (50-85)1.21 (1.02-1.44)0.91 (0.80-1.04)
Age <12 mo  3  2,11067 (47-83)41 (28-55)1.13 (1.02-1.26)0.81 (0.61-1.07)
Female1547,35166 (57-74)47 (42-52)1.24 (1.11-1.39)0.73 (0.58-0.91)
White1042,45650 (34-65)58 (42-73)1.18 (0.96-1.46)0.87 (0.73-1.04)
Hispanic  734,07412 (4-32)89 (76-95)1.03 (0.74-1.44)1.00 (0.95-1.05)
Asian  524,6235 (4-7)96 (95-97)1.42 (1.09-1.86)0.98 (0.97-1.00)
Non-African American1042,39785 (76-91)27 (14-45)1.17 (1.02-1.33)0.55 (0.48-0.63)
  • LR− = negative likelihood ratio; LR+ = positive likelihood ratio; UTI = urinary tract infection.

  • a Data from Pylkkänen et al38 were not included in the meta-analysis; bivariate random effects model by Chu and Cole.21