We found no studies in which clinical outcomes in women with UI diagnosed by clinical history alone were compared with those in women with UI diagnosed using urodynamics. However, several studies have evaluated the accuracy of the symptom of stress or urge UI relative to findings on urodynamic (UD) investigations in women undergoing assessment of their urinary symptoms. Most of these studies have been considered in two reviews and a health technology assessment of diagnostic methods for UI.46–48 Two of the publications included studies of women with symptoms of stress, mixed or urge UI46,48 and one included only studies evaluating women with stress UI.47 The reviews that included women with stress, mixed or urge UI calculated and combined sensitivity and specificity data for the symptom of stress (be it with or without mixed symptoms) and for the symptom of urge UI (be it with or without mixed symptoms). The GDG considered that the mixed ‘symptom’ should be considered separately (because in practice women are categorised into those with stress, mixed or urge UI) and that the important question in relation to the comparison ofurinary history with urodynamic findings is whether urodynamics gives additional information to that obtained from the history alone. In considering this question, the GDG took the approach that a clinical history would be taken for every woman, and that a positive history for a particular type of UI would always be followed by treatment appropriate to that type of UI.
Overall, 25 relevant studies that compared the diagnosis based on history with urodynamic findings were considered by the GDG. These studies used cystometry as the reference standard for diagnosis of UI, and therefore assumed that history taking had a lower diagnostic value in comparison. Fourteen studies included women with stress, mixed or urge UI, and eleven presented raw data in a way that allowed sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) to be calculated.49–62 Two of these studies only reported accuracy data for stress and mixed UI.61,62 Five studies only investigated how a history of urge UI or OAB compared with urodynamic findings of DO.63–67 Six studies only investigated how a history of stress UI compared with the finding of urodynamic stress incontinence,68–73 four of which provided some but not all accuracy data.
Multichannel cystometry (with or without uroflowmetry, urethral pressure profilometry, or cystourethrography) was the urodynamic method used in 24 studies. The remaining study used singlechannel cystometry for women with urge UI (and suspected DO) and multichannel cystometry for women with stress UI.50 All except four studies62–64,70 stated that terminology used for urodynamic findings conformed to ICS standards.
With the exception of one study,49 which involved primary and secondary care, all studies were conducted in secondary or tertiary care.
The GDG focused on the 11 studies that provided diagnostic accuracy data for stress, mixed and urge UI. Confidence intervals were calculated for each value, as this was considered to be more appropriate than pooling data from individual studies. Pooling the available data (by meta-analysis) or generating receiver operating characteristic curves was not considered to be appropriate because:
the population in each study varied in terms of the relative proportions of stress, mixed or urge UI. The percentage of study participants with urodynamic stress incontinence varied from 34% to 63% (median 52%); the proportion with urodynamic stress incontinence plus DO ranged from 10% to 28% (median 19%), and for DO the range was 7–32% (median 17%).
the methods used to obtain a history varied; a structured questionnaire or standardised form was used in seven of the 11 studies,49–53,57,58 and the remainder only specified that a history had been taken
54–56,59the studies were generally considered to be of poor quality; none stated whether urodynamic testing was undertaken blind to findings of history taking.
For the 11 studies that included women with stress, mixed or urge UI, and reported raw accuracy data for the three types of UI separately, the results are described below and also shown with 95% confidence intervals in Figures Q.1 to Q.12 (see evidence tables for full details of individual studies).49–59
The sensitivity, specificity, PPV and NPV values were calculated as shown in . We consider that the NPV is of particular interest in terms of assessing whether urodynamics provides additional information compared with clinical history, because this quantity summarises the extent to which a negative history is associated with a negative finding on urodynamics (i.e. whether carrying out urodynamics would alter the diagnosis and, more importantly, management, for women who do not report a particular UI symptom).
‘2 × 2’ table for calculation of diagnostic accuracy parameters.
In diagnostic accuracy studies ‘prevalence’ usually refers to the proportion within a study who have positive findings using the reference standard (and is given by (a + c)/(a + b + c + d)). The term ‘prevalence’ is used from here on for simplicity, to reflect the proportion of women in these studies who have a particular urodynamic finding. Sensitivity is normally unaffected by prevalence because it depends on the number of ‘true positives’ but not on the number of ‘true negatives’. Similarly, specificity is normally unaffected by prevalence because it depends on the number of true negatives but not on the number of true positives. However, PPV and NPV both vary with prevalence because they both depend on the numbers of true positives and true negatives. PPV normally increases with increasing prevalence, whereas NPV normally decreases with increasing prevalence (provided sensitivity and specificity are both held fixed). PPV also increases with increasing sensitivity, provided specificity and prevalence are both held fixed. Similarly, NPV normally increases with increasing specificity, provided sensitivity and prevalence are both held mixed.
Stress urinary incontinence
to show sensitivity, specificity, PPVs and NPVs of history of pure stress UI compared with urodynamic findings of stress UI (urodynamics being the reference standard), with 95% confidence intervals. The median values and ranges of results are shown in .
Sensitivity of a history versus UD findings of stress UI; studies arranged in ascending order of USI prevalence (%): 34, 40, 44, 46, 51, 52, 53, 54, 58, 61, 63.
NPV of a history versus UD findings of stress UI; studies arranged in ascending order of USI prevalence (%): 34, 40, 44, 46, 51, 52, 53, 54, 58, 61, 63.
Diagnostic accuracy data for urinary history of pure stress UI compared with urodynamic findings of stress UI.
to show that there is considerable variation across the studies in sensitivities, specificities, PPVs and NPVs of a history of pure stress UI compared with positive findings of pure stress UI on multichannel cystometry. In general, there is a low level of agreement between the two methods. The median values show that:
66% of women who have urodynamic stress incontinence also have a history of pure stress UI
83% of women who do not have urodynamic stress incontinence also do not have a history of pure stress UI
70% of women who have a history of pure stress UI also have urodynamic stress incontinence
69% of women who do not have a history of pure stress UI also do not have urodynamic stress incontinence.
We considered whether differences in prevalence of urodynamic stress incontinence might explain the variation in the results of individual studies. The studies in to are arranged in ascending order of prevalence (from 34 to 63%, left to right). shows that, as expected, the PPV increases with increasing USI prevalence (i.e. as the proportion of women with urodynamic stress UI in the study increases, the proportion of women who have positive findings from urodynamic studies as well as from history increases). However, NPV does not appear to follow the expected pattern of decreasing values with increasing prevalence. This might be because the sensitivities and specificities of the studies vary greatly, or because the studies vary in other ways. Three studies have a much lower sensitivity than the other eight. The three studies do not appear to be different to the others in any systematic way that would explain this variation.
PPV of a history versus UD findings of stress UI; studies arranged in ascending order of USI prevalence (%): 34, 40, 44, 46, 51, 52, 53, 54, 58, 61, 63.
It is possible that the method used to obtain a history might explain some of the variation between studies. However, the studies provided insufficient detail of the method of obtaining a history to allow this possible association to be explored in a meaningful way, and therefore it is not known how much this may influence the results seen.
The variation between studies might also reflect a lack of blind comparison of the results of history taking and urodynamic testing (again, the studies provide insufficient detail to explore this further), or the fact that urodynamics cannot be regarded as a gold standard. Indeed, there is some suggestion in and that sensitivity and specificity vary with prevalence, which should not occur using a gold standard that provides a perfect classification of the presence/absence of stress UI.
Specificity of a history versus UD findings of stress UI; studies arranged in ascending order of USI prevalence (%): 34, 40, 44, 46, 51, 52, 53, 54, 58, 61, 63.
Mixed urinary incontinence
to show sensitivity, specificity, PPVs and NPVs of history compared with urodynamic findings of mixed UI (i.e. USI plus DO), with 95% confidence intervals. The median values and ranges of results are shown in .
Sensitivity of a history of mixed UI versus UD findings of USI plus DO; studies arranged in ascending order of USI plus DO prevalence (%): 10, 14, 17, 17, 18, 19, 21, 21, 24, 25, 28.
NPV of a history of mixed UI versus UD findings of USI plus DO; studies arranged in ascending order of USI plus DO prevalence (%): 10, 14, 17, 17, 18, 19, 21, 21, 24, 25, 28.
Diagnostic accuracy data for urinary history of mixed UI compared with urodynamic findings of USI plus DO.
, and show that there is considerable variation across the studies in sensitivities, specificities and PPVs of a history of mixed UI compared with positive findings of USI plus DO on multichannel cystometry. The median values show that:
Specificity of a history otable Qf mixed UI versus UD findings of USI plus DO; studies arranged in ascending order of USI plus DO prevalence (%): 10, 14, 17, 17, 18, 19, 21, 21, 24, 25, 28.
PPV of a history of mixed UI versus UD findings of USI plus DO; studies arranged in ascending order of USI plus DO prevalence (%): 10, 14, 17, 17, 18, 19, 21, 21, 24, 25, 28.
68% of women who have USI plus DO also have a history of mixed UI
77% of women who do not have urodynamic stress UI or DO also do not have a history of mixed UI
35% of women who have a history of mixed UI also have both USI plus DO.
shows that the NPV is more consistent across studies than are the other quantities, i.e. at least 80% (median 90%) of women who do not have a history of mixed UI also do not have USI and DO on multichannel cystometry. Less variation is expected when the agreement between two forms of assessment is close to 100% (or 0%).
As expected, the PPV appears to increase with increasing prevalence of mixed findings of USI plus DO on multichannel cystometry.
The relationship between NPV and specificity is not strong in these studies; this could be because the prevalence of mixed UI varies widely between studies (from 10% to 28%).
Urge urinary incontinence
to show sensitivity, specificity, PPVs and NPVs of the symptom of pure urge UI compared with urodynamic findings of detrusor overactivity, with 95% confidence intervals. The median values and ranges of results are shown in .
Sensitivity of a history of urge UI versus UD findings of DO; studies arranged in ascending order of DO prevalence (%): 7, 12, 13, 14, 15, 17, 18, 20, 24, 27, 32.
NPV of a history of urge UI versus UD findings of DO; studies arranged in ascending order of DO prevalence (%): 7, 12, 13, 14, 15, 17, 18, 20, 24, 27, 32.
Diagnostic accuracy data for urinary history of pure urge UI compared with urodynamic findings of detrusor overactivity.
and show that there is wide variation in the sensitivities and PPVs of urge UI across the studies. The median values show that:
PPV of a history of urge UI versus UD findings of DO; studies arranged in ascending order of DO prevalence (%): 7, 12, 13, 14, 15, 17, 18, 20, 24, 27, 32.
It is also noted that the relationship between PPV and sensitivity is not strong across these studies and this could be because the prevalence of DO varies from 7% to 32% (median 17%).
Conversely, specificity and NPVs for urge UI are both quite consistent, as shown in and . At least 81% of women (median 96%) who do not have DO also do not have a history of pure urge UI; and at least 79% (median 91%) of women who do not have a history of pure urge UI also do not have DO on multichannel cystometry. In other words, if there is no history of urge UI, the probability of finding DO on urodynamic testing is small. Again, the results of the individual studies are quite consistent as is to be expected when the percentage agreement between two forms of assessment is close to 100%. There is a strong relationship between specificity and NPV in these studies, despite the variability in the prevalence of DO. This might be because the variations in prevalence are compensated by variations in sensitivity.
Specificity of a history of urge UI versus UD findings of DO; studies arranged in ascending order of DO prevalence (%): 7, 12, 13, 14, 15, 17, 18, 20, 24, 27, 32.
Other studies
The remaining studies that compared history and urodynamic findings did so in relation to only one or two types of UI. These also showed variability in their results. In two studies that only reported accuracy data for stress and mixed UI, the results were:61,62
sensitivity: 33% and 39% stress UI; 49% and 68% mixed UI
specificity: 83% and 86% stress UI; 48% and 57% mixed UI
PPV: 56% PPV: 56% and 74% stress UI; 33% and 53% mixed UI
NPV: 58% and 66% stress UI; 53% and 80% mixed UI.
In one study that included women with any type of UI but for which only data for stress UI were reported, the sensitivity of history compared with urodynamics was 52%, specificity and PPV were both 85% and NPV was 53%.70
In the five studies that investigated how a history of urge UI and/or OAB compared with the urodynamic finding of DO, the results were:63–67
sensitivity: median 40% (range 24–91%)
specificity: median 86% (range 45–92%)
PPV: median 54% (range 44–91%)
NPV: median 68% (range 26–91%).
In five of six studies that investigated how a history of stress UI compared with a urodynamic finding of stress UI, sensitivities of 47–82% and PPVs of 52–100% were reported.68–72 The remaining study reported sensitivity, specificity, PPV and NPV results using four different urodynamic methods in the assessment of women with urodynamic stress UI (including mixed UI). The ranges of results across the different methods were: sensitivity 49–91%, specificity 98–100%, PPV 82–100% and NPV 44– 88%, the highest level of agreement being noted for observed urine loss with cough during multichannel cystometry.73
Conclusions
The available studies comparing history of stress, mixed or urge UI with findings of stress UI and/or DO on multichannel cystometry have poor internal and external validity. In addressing the question of whether urodynamic testing gives additional information to that obtained from history alone, with the limitations of the studies in mind, the following conclusions can be drawn:
If a woman does not report mixed UI (i.e. if she reports pure stress UI or pure urge UI), the probability of finding USI plus DO on cystometry is small (around 10%), therefore urodynamic testing might be said to offer little additional diagnostic value. It is acknowledged that urodynamic investigation is not simply used to distinguish USI and DO, and that further information may be obtained about other elements of lower urinary tract function, for example the voiding pattern.
If a woman does not report pure urge UI, the probability of finding DO on cystometry is small (again around 10%), therefore urodynamic testing offers little added diagnostic value.
The situation for pure stress UI is less clear-cut. Here 15–51% (median 31%) of women who do not report pure stress UI may nevertheless be found to have USI on cystometry. However, the lack of consistency between the NPVs in the available studies together with the lack of detailed information about the method of obtaining a history and the poor quality of the studies limit the extent to which the evidence would support urodynamic testing for women who do not report stress UI. Furthermore, a limitation of dealing with stress, mixed and urge UI as three separate entities is that the analysis ignores the interdependence between the different diagnoses.
History taking is regarded as the cornerstone of assessment of UI. Current practice is that women with UI are categorised according to their symptoms into those with stress, mixed or urge UI; women with mixed UI are treated according to the symptom they report to be the most troublesome. In the absence of evidence that urodynamic testing improves the outcome of women treated conservatively, and without robust evidence that urodynamic testing provides additional valuable information to the history alone in the initial assessment of women with UI, the GDG concluded that urodynamic testing is not required before initiating conservative treatment.
