Systematic review and meta-analysis on diagnostic accuracy of signs & symptoms predicting uncomplicated urinary tract infection in women Giesen LGM Erasmus.

Systematic review and meta-analysis on diagnostic accuracy of signs & symptoms

predicting uncomplicated urinary tract infection in women

Giesen LGMErasmus student Sept – Dec 2009 RCSI, Dublin

Cousins G, Dimitrov BD , Van de Laar FA, Fahey T

Outline

• Background/ Rationale

• Methods

• Results

• Conclusion

Background

• Prevalence– 5% asymptomatic– 50% symptomatic

• Pathogens

• Reference standard– Midstream urine culture– Thresholds >10², >10³, >105 CFU/ml

Diagnosis UTI

Diagnosis

Dipstick testing Culture urineSigns & symptoms

Diagnosis UTI

Diagnosis

Dipstick testing Culture urineSigns & symptoms

Rationale

• Bent 2002– Meta-analysis of 9 studies

– Reference standard > 104 CFU/ml

– Significant symptoms & signs

• What can we add?– 9 additional studies

– European guidelines

– Consideration of heterogeneity

Diagnostic accuracy signs & symptoms

Reference test :

Urine culture

> 10² CFU/ml

>10³ CFU/ml

>105 CFU/ml

Signs & symptoms

Dysuria

Frequency

Urgency

Hematuria

Fever

Flank pain

Absence vaginal discharge

& irritation

Participants Healthy women with symptoms of UTI

Setting Primary care

Index test Signs & symptoms

Reference test Urine culture (> 10², >103, >105 CFU/ml)

Outcome measure 2x2 tables

Methods – Identification of studies

Methods

• Quality assessment (QUADAS)

• Meta-analysis 1. Derivation of summary statistics of each study (2x2

tables)

2. Pooling the summary statistics across studies

3. Examine the pattern of variability (heterogeneity) observed in the results

Derivation of summary statistics

Little (2006)      

  UTI No UTI Total

Frequency 185 94 279

No frequency 69 60 129

Total 254 154 408

Sensitivity:

185/254 = 72.8%

Specificity: 60/154 = 39.0%

+ LR = sens/ (1- spec) =

72.8/ (1-39.0) = 1.19

Results: 16 studiesQuality assessment

Time period index and reference test

Blinding

Contamination reported

Pt selected consecutively

Data included 95%CI

Results: Meta-analysis (> 102)

Symptom (# studies) Sens (95%CI) Spec (95%CI) + LR (95%CI)

Frequency (13) 0.88 (0.83-0.92) 0.20 (0.14-0.28) 1.10 (1.04-1.16)

Dysuria (14) 0.80 (0.74-0.86) 0.38 (0.31-0.46) 1.30 (1.20-1.41)

Hematuria (7) 0.25 (0.21-0.29) 0.85 (0.81-0.89) 1.72 (1.30-2.27)

Nocturia (6) 0.59 (0.50-0.68) 0.55 (0.49-0.61) 1.30 (1.08-1.56)

Urgency (9) 0.67 (0.52-0.80) 0.45 (0.31-0.60) 1.22 (1.11-1.34)

Vaginal discharge (6) 0.15 (0.08-0.26) 0.77 (0.62-0.88) 0.65 (0.51-0.83)

Pretest & posttest probability

Pretest Probability =

50%

Posttest Probability

Frequency LR+ = 1.10

Hematuria LR+ = 1.72

Dysuria LR+ = 1.30

Nocturia LR+ = 1.30

Urgency LR+ = 1.22

Vaginal discharge LR+ = 0.65

52,4%

56.5%

63.2%

55.0%

39.4%

56.5%

Heterogeneity - threshold effect (> 102)Symptom (# studies) Correlation P-value Present

Frequency (13) 0.698 0.008 X

Dysuria (14) 0.679 0.008 X

Hematuria (7) -0.714 0.071

Nocturia (6) 0.257 0.623

Urgency (9) 0.933 0.000 X

Vaginal discharge (6) 1.000 0.000 X

Heterogeneity frequency - threshold effect

Frequency (# studies) Correlation P-value Present

> 102 CFU/ml (13) 0.698 0.008 X

> 103 CFU/ml (11) 0.673 0.023 X

> 105 CFU/ml (6)0.486 0.329

Heterogeneity 102 CFU/ml 103 CFU/ml

Figures: sensitivity frequency

I-square = 84.9% I-square = 85.2% I-square = 68.1%

105 CFU/ml

Conclusions

• 5 symptoms increase the probability of UTI when

present (frequency/ dysuria/ hematuria/ nocturia/

urgency)– Vaginal discharge decreased probability of UTI

• However only modest ability to ‘rule in’ or ‘rule out’ UTI– High pre-test probability

Conclusions

• Importance of considering heterogeneity– Variations in diagnostic threshold– Patient selection/ setting

• Future directions– Combinations of symptoms – clinical prediction rule– Severitiy of symptoms