British Journal of Urology (1998), 82, 619 623 The ICS- BPH Study: uroflowmetry, lower urinary tract symptoms and bladder outlet obstruction J.M. REYNARD1, Q. YANG2, J.L. DONOVAN3,T.J. PETERS3, W. SCHAFER4, J.J.M.C.H. de la ROSETTE5, N.F. DABHOIWALA6,D.OSAWA7, A. TONG LONG LIM8 and P. ABRAMS2 1Department of Urology, The Royal London and St Bartholomew s Hospitals, 2Bristol Urological Institute, Southmead Hospital, Bristol, 3Department of Social Medicine, University of Bristol, Canynge Hall, Bristol, 4Urologische Klinik der RWTH, Aachen, Germany, 5University Hospital, Nimegen, The Netherlands, 6Department of Urology, Academic Medical Centre, Meibergdreef, Amsterdam, The Netherlands, 7Department of Urology, University of Ryukyus, Okinawa, Japan and 8Department of Urology, Veterans General Hospital, Taipaei, Taiwan Objective To explore the relationship between uroflow predictive value (PPV) of 70% and a sensitivity of 47% variables and lower urinary tract symptoms (LUTS); for BOO. The specificity using a threshold Q of to define performance statistics (sensitivity, specificity, 15 ml/s was 38%, the PPV 67% and the sensitivity positive and negative predictive values) for imum 82%. Those voiding <150 ml (n=225) had a 72% urinary flow rate (Q ) with respect to bladder outlet chance of BOO (overall prevalence of BOO 60%). In obstruction (BOO) at various threshold values; and to those voiding >150 ml the likelihood of BOO was investigate the diagnostic value of low-volume voids. 56%. The addition of a specific threshold of 10 ml/s Patients and methods The study comprised 1271 men to these higher volume voiders improved the PPV for aged between 45 and 88 years recruited from 12 BOO to 69%. centres in Europe, Australia, Canada, Taiwan and Conclusion While uroflowmetry cannot replace press- Japan over a 2-year period. Symptom questionnaires, ure-flow studies in the diagnosis of BOO, it can provide voiding diaries, uroflowmetry and pressure-flow data a valuable improvement over symptoms alone in the were recorded. The relationship between uroflow vari- diagnosis of the cause of lower urinary tract dysfunction in men presenting with LUTS. This study provides ables and symptoms, Q and BOO, and the diagnostic performance of low volume voids were analysed. performance statistics for Q with respect to BOO; Results The relationship between symptoms and uroflow such statistics may be used to define more accurately variables was poor. The mean dicerence between the presence or absence of BOO in men presenting home-recorded and clinic-recorded voided volumes with LUTS, so avoiding the need for formal pressure- was 48 ml. Q was significantly lower in those flow studies in everyday clinical practice, while with BOO (9.7 ml/s for void 1) than in those with no improving the likelihood of a successful outcome from obstruction (12.6 ml/s; P<0.001) and Q was prostatectomy. This study also shows that low-volume negatively correlated with obstruction grade uroflowmetry can provide useful diagnostic infor- (Spearman s correlation coebcient 0.3, P<0.001), mation and that, as such, the data from such voids even when controlling for the negative correlation should not be discarded. between age and Q (Spearman s partial correlation Keywords Benign prostatic hyperplasia, lower urinary coebcient 0.29, P<0.001). A threshold value of tract symptoms, uroflowmetry, flow rate, bladder Q of 10 ml/s had a specificity of 70%, a positive outlet obstruction symptoms? How accurate is uroflowmetry in diagnosing Introduction BOO? Is clinic-recorded flow data representative of a Fundamental questions remain unanswered about the patient s normal voiding function? What is the value of value of uroflowmetry in the assessment of men pre- low volume uroflowmetry should low-volume voids senting with LUTS thought to be secondary to BOO. Do be discarded? uroflow variables, e.g. imum urinary flow rate (Q ) In 1992 the ICS established a multicentre, international study to address these questions, amongst and residual urine volume, bear any relationship to others. One of several aims of this study was to explore Accepted for publication 24 June 1998 the diagnostic power of uroflowmetry in symptomatic 1998 British Journal of Urology 619
620 J.M. REYNARD et al. elderly men and in so doing provide clinicians with reduced most of the time or reduced all of the time?, information that they might use to more accurately was significantly although weakly correlated with the diagnose BOO without needing to perform pressure-flow imum value of Q recorded for each patient studies. We present the results of the ICS-BPH study for (Spearman s correlation coebcient 0.19, P<0.001, the analysis of the uroflow data. n=1176). Thus, patients who felt that their flow was Patients and methods poor were more likely to have a low Q. Similarly, patients who perceived their poor flow to be a problem (i.e. those with a high bother score) had lower values of Men aged over 45 years with LUTS and benign prostatic Q, although again the correlation between Q and enlargement were eligible for entry into the ICS- BPH bothersomeness of the symptom of poor flow was weak study. Those with prostate cancer, neurological disease, (Spearman s correlation coebcient 0.17, P<0.001, diabetes or who had undergone previous prostatic or n=1162). The mean Q in those who thought their urethral surgery, or who were taking medication active flow was normal (16.2 ml/s) was significantly higher on the lower urinary tract were excluded. Each patient (P<0.001, unpaired t-test) than in those who felt their completed the ICSmale questionnaire [1], a psychometrically flow was not normal (12 ml/s). validated symptom questionnaire, and was Patients were also questioned about bladder emptying asked to complete a frequency-volume diary to document (Question 17 of the ICS- BPH questionnaire How often the day and night-time urinary frequency and voided do you feel your bladder has not emptied properly after volumes. Each patient was requested to perform three you have urinated? ). There was no significant relation- measurements of urinary flow rate, after which pressure- ship between sensation of incomplete bladder emptying flow urodynamic studies (PFS) were performed to establish and PVR (Spearman s correlation P>0.6) nor between the presence or absence of BOO. The PFS were performed the bothersomeness of the feeling of incomplete emptying in accordance with ICS guidelines [2]. BOO was defined and PVR (P>0.1). on the basis of the Schafer method [3]; this has seven potential categories, i.e. 0 unobstructed, 1 slightly Are clinic-recorded flow rates an accurate reflection of a obstructed, 2 6 obstructed (with increasing severity). The patient s normal voiding pattern? relationship between uroflow variables (flow rate, voided volume and post-void residual urine volume, PVR) symp- Without specifically correlating home-recorded uroflowmetry toms and BOO was assessed, with particular interest in: with clinic-recorded flows it is not possible to (a) the relationship between uroflow variables (Q, PVR) answer this question and indeed the ICS- BPH study and LUTS; (b) the agreement between voided volumes was not formally designed to address this point. However, recorded by patients in their voiding diaries and those home-voided volumes recorded on frequency-volume recorded during uroflowmetry; (c) the relationship charts were compared with those recorded during uro- between age and Q ; (d) the specificity and predictive flowmetry; 1045 patients (82%) had data recorded for values of various thresholds of Q for BOO; and (e) the both home- and clinic-voided volumes. The mean of the diagnostic value of low-volume voids (<150 ml). dicerences between home-(mean 169 ml) and clinicvoided volumes (mean 216 ml) was 48 ml [4] (SD of the dicerence, 107 ml, P<0.001, paired t-test, 95% Results limits of agreement 258 and 162 ml). Thus, patients The mean age of the patients was 66.5 years (range tended to void larger volumes at uroflowmetry than 45 88 years), 21% were aged 59 years, 43% were recorded at home on their frequency-volume charts. aged 60 69 years and 36% were aged 70; 211 (17%) provided one void, 443 (35%) two and 537 (42%) three. Flow data were missing in 81 patients (6%). Of the 933 The relationship between Q and BOO patients for whom interpretable pressure-flow data were For each of voids 1, 2 and 3, the Q was lower available, 165 (17.5%) were not obstructed (Schafer (P<0.001 for each analysis) in those with BOO than in grade 0), 205 (22.3%) were slightly obstructed (grade those with no obstruction (Table 1). As the Schafer 1) and 563 (60.2%) were obstructed ( grade 2 6). obstruction grade is expressed as ordinal data (0 6), the Spearman rank correlation coebcient for obstruction The relationship between symptoms and uroflow variables grade vs the imum Q recorded by each patient was calculated. Spearman s correlation coebcient Question 12 of the ICS- BPH symptom questionnaire, between obstruction grade and Q for the 897 men Would you say that the strength of your urinary stream for whom data were available was 0.3 (P<0.001). is normal, occasionally reduced, sometimes reduced, As both Q and age are related (Spearman s correlation
THE ICS-BPH STUDY 621 Table 1 The mean Q, voided volume and PVR for each void in Schafer categories 0 2 were defined as unobstructed patients with and without BOO and categories 3 6 as obstructed. A Q of <10 ml/s had a specificity of 70% and a PPV of 70% for BOO; the Void sensitivity was only 47%. The specificity for a threshold Variable 1 2 3 of 15 ml/s was 38%, the PPV 67% and the sensitivity correspondingly higher, at 82%. Thus, 53% of patients Q (ml/s) with BOO had a flow rate of >10 ml/s and 18% of BOO 9.7 9.2 9.6 patients with BOO had a flow rate of >15 ml/s. No BOO 12.6 11.8 12.9 P <0.001 <0.001 <0.001 Voided volume (ml) Low-volume uroflowmetry BOO 208 204 209 No BOO 257 231 249 Taking the flow with the highest value of Q, 225 P <0.001 <0.005 <0.005 patients (25%) voided <150 ml and 671 (75%) voided PVR (ml) 150 ml. The former had a 72% chance of having BOO 88 100 93 BOO and a 28% chance of not having BOO. The 140 No BOO 79 78 86 patients who voided <150 ml and who had a Q of P >0.1 <0.005 >0.4 <10 ml/s had a 71% chance of having BOO. In those voiding >150 ml, the likelihood of BOO in the group as a whole was 56%. The addition of a specific threshold coebcient 0.1, P<0.001, n=1161) the relationship of 10 ml/s to these higher volume voiders improved the between Q and obstruction grade was re-analysed predictive value for BOO. Thus, in those patients voiding controlling for age. In this analysis Spearman s partial >150 ml, 69% with a Q of <10 ml/s had BOO, correlation coebcient for obstruction grade vs Q was whereas in those with a Q of >10 ml/s, half had 0.29 (P<0.001, n=872). BOO. Pearson s correlation coebcient for Q against Voided volume was significantly higher for each void voided volume (P<0.001 for all groups) was 0.45 for in the unobstructed group than in those with BOO void 1 (n=1185), 0.41 for void 2 (n=986) and 0.39 (Table 1). As Q is dependent on voided volume, the for void 3 (n=558). relationship between Q and BOO was re-analysed There was no significant relationship between voided controlling for voided volume. The negative correlation volume and PVR for voids 1 and 2. For void 3 there was between Q and BOO persisted when voided volume a significant but weak correlation between voided was controlled for, for each of voids 1, 2 and 3 (r= volume and PVR (P=0.004), but Pearson s correlation 0.2, 0.26 and 0.25, all P<0.001, respectively). coebcient was only 0.13. There were no significant dicerences in PVR between those with and without BOO for voids 1 and 3 (Table 1) although the dicerence in PVR between the groups (100 Discussion vs 78 ml) was statistically significant for void 2 The findings of this study reabrm the findings of previous (P<0.005). studies [5 10] that show there is at best only a poor The diagnostic power of two threshold values of Q association between LUTS and objective measures of (10 and 15 ml/s) were assessed, taking the highest voiding such as Q, PVR and voided volume. Although value of Q for each patient. For each threshold the the symptom of poor flow is significantly associated with sensitivities, specificities and positive predictive values a low Q, the association is weak, as indicated by the (PPVs) for BOO are given in Table 2. For simplicity, correlation coebcient of only 0.19 ( 0.22 in the study by Barry et al. [7]). There was no significant association Table 2 Sensitivities, specificities, positive predictive values (PPV) between the symptom of feeling of incomplete bladder and 1 negative predictive values (1 NPV) of Q for BOO emptying and PVR. Barry et al. [7] suggested that the poor association between LUTS and variables such as Mean Threshold Q (ml/s) Q and PVR reflected unreliability in measurement of (95% CI) the physiological variables rather than reflecting unre- [n/n] (%) <10 <15 liability in the assessment of symptoms. Certainly, testretest validation of symptom questionnaires shows con- Sensitivity 47 (43 51) [252/540] 82 (79 85) [440/540] Specificity 70 (65 75) [250/357] 38 (33 45) [136/357] sistency in individual patients [1,11] while both Reynard PPV 70 (65 75) [252/359] 67 (63 71) [446/661] et al. [12] and Feneley et al. [13] have shown considerable 1-NPV 54 (49 58) [288/539] 42 (36 48) [100/236] variation in Q measured on either the same or dicerent days, and Dunsmuir et al. [14] have shown poor
622 J.M. REYNARD et al. test-retest reliability for the measurement of PVR. Thus, studies be discarded, with a diagnosis being based on while LUTS are important in that they are what bother symptoms alone, or can the data from low volume voids patients, their significance in terms of reflecting a demon- be used to make a diagnosis? In the present study, 60% strable abnormality in voiding function is very limited, of men with LUTS had BOO. Relying solely on voided and it is therefore important not to over-interpret their volume (irrespective of the value of Q ) then 72% of significance. those patients who voided <150 ml had BOO, whereas Patients tended to void larger volumes during uro- only 56% of those voiding >150 ml had BOO. In those flowmetry than recorded at home on their frequencyvolume voiding <150 ml on their highest flow, taking a thresh- charts, suggesting that there may be a tendency old of 10 ml/s did not improve the ability to diagnose to over-perform in the clinic. The large quantities of BOO (PPV 71%), although a lower threshold might have fluid consumed by patients in flow clinics might be partly done so. The addition of a specific threshold of 10 ml/s responsible for this. If a representative flow measurement to higher volume voiders improved the ability to diagnose is required, a patient should be instructed to void when BOO, from 56% to 69%. they would normally do so, rather than waiting until In conclusion, the associations between symptoms and their bladder is very full. uroflow variables such as Q and PVR are poor. While While PFS are clearly the only method by which BOO uroflowmetry cannot replace PFS in the diagnosis of can be reliably diagnosed [15,16], the present study BOO, the ICS- BPH study shows that flow studies can shows that Q can provide some improvement in provide a valuable improvement over symptoms alone diagnostic power. A threshold of 10 or 15 ml/s can be in the diagnosis of lower urinary tract voiding problems. used, depending on the clinical situation. For example, The performance statistics provided by this study give as the symptomatic results of prostatectomy are better the clinician useful information upon which to base in men with BOO [17], the results of prostatectomy rational decisions. An appropriate threshold for Q can could be improved by selecting a threshold giving a high be selected depending on the clinical situation for which specificity and PPV. To reduce the number of unobstructed uroflowmetry is being used. Low-volume uroflowmetry men undergoing prostatectomy, a threshold of (voided volume <150 ml) can provide useful diagnostic 10 ml/s would be more suitable than one of 15 ml/s, information, not least because most patients, at least in as its specificity and PPV are 70% (although some this study, have BOO. urologists might consider even this threshold too high). However, if it was desired to screen for BOO in a population of patients to identify those at risk of developing complications of BOO, a higher threshold of References 15 ml/s would be more suitable, as it provides a sensi- 1 Donovan JL, Abrams P, Peters TJ et al. The ICS- BPH tivity of 82%, compared with only 47% for a threshold Study: the psychometric validity and reliability of the of 10 ml/s. This is what would be intuitively expected. ICSmale questionnaire. Br J Urol 1996; 77: 554 62 The present study provides the clinician with quantified 2 Abrams P, Blaivas JG, Stanton SL, Andersen JT. The probabilities that can be used as the basis for rational standardisation of terminology of the lower urinary tract. World J Urol 1989; 6: 233 453 decision-making in dicerent clinical contexts. 3 Schafer W. Basic principles and clinical application of However, 60% of men in this series presenting with advanced analysis of bladder voiding function. Urol Clin LUTS had BOO and the addition of a threshold Q of North Am 1990; 17: 553 66 10 ml/s provided only a very modest improvement in 4 Bland JM, Altman D. Statistical methods for assessing diagnostic power, increasing the PPV by just 10%, to agreement between two methods of clinical measurement. 70%. It could therefore be argued that uroflowmetry Lancet 1986; i: 307 10 provides little additional information over that provided 5 Abrams P. Prostatism and prostatectomy: the value of simply by the reality that the patient has symptoms and urine flow rate measurement in the preoperative assessment that if the surgeon wants to ensure that only obstructed for operation. J Urol 1977; 117: 70 1 men undergo TURP, then PFS would be necessary in all 6 Reynard JM, Abrams P. Bladder outlet obstruction: assess- cases before operation. ment of symptoms. World J Urol 1995;??: 3 8 7 Barry MJ, Cockett ATK, Holtgrewe HL, McConnell JD, A quarter of men in this study voided <150 ml on Sihelnik SA, Winfield HN. Relationship of symptoms of their flow with the highest Q, a finding similar to prostatism to commonly used physiological and anatomical that of Carter et al. [18]. The recommendations in the measures of the severity of benign prostatic hyperplasia. Proceedings of the 3rd International Consultation on J Urol 1993; 150: 351 8 BPH [19] stated that at least two flow rate recordings 8 Bruskewitz RC, Iversen P, Madsen PO. Value of post-void with a volume of >150 ml each should be obtained. residual urine determination in evaluation of prostatism. Should the information provided by low-volume uroflow Urology 1982; 20: 602 4
THE ICS-BPH STUDY 623 9 Bosch JLHR, Hop WCJ, Kirkels WJ, Schroder FH. The routine investigation of BPH patients. Neurourol Urodyn International Prostate Symptom Score in a community 1989; 7: 219 21 based sample of men between 55 and 74 years of age: 16 Chancellor MB, Blaivas JG, Kaplan SA, Axelrod S. Bladder prevalence and correlation of symptoms with age, prostate outlet obstruction versus impaired detrusor contractility: volume, flow rate and residual urine volume. Br J Urol the role of uroflow. J Urol 1991; 145: 810 2 1995; 75: 622 30 17 Neal DE, Ramsden PD, Sharples L et al. Outcome of elective 10 Ezz El Din K, Kiemeney LALM, de Wildt MJAM, de Debruyne prostatectomy. Br Med J 1989; 299: 762 7 F, de la Rosette JJMCH. Correlation between uroflowmetry, 18 Carter PG, Lewis P, Abrams P. Single versus multiple flows prostate volume, postvoid residual and lower urinary tract in the diagnosis of obstruction. J Urol 1991; 145: 397A symptoms as measured by the IPSS. Urology 1996; 19 Cockett ATK, Khoury S, Aso Y et al. eds Proceedings of The 48: 393 7 3rd International Consultation on Benign Prostatic Hyperplasia, 11 Barry MJ, Fowler FJ Jr, O Leary MP et al. and the Monaco 1995: 628 35 measurement committee of the American Urological Association. The American Urological Association symptom index for benign prostatic hyperplasia. J Urol 1992; Authors 148: 1549 57 J.M. Reynard, DM, MA, FRCS(Urol). Senior Registrar. 12 Reynard JM, Lim C, Peters TJ, Abrams P. The value of Q. Yang, BSc, MSc, PhD, Research Associate. multiple free-flow studies in men with BPH. Br J Urol J.L. Donovan, BA, PhD, Senior Lecturer. 1996; 77: 813 8 T.J. Peters, PhD, Reader. 13 Feneley MR, Dunsmuir WD, Bryan J, Kirby RS. W. Schafer, Dipl Ing. Reproducibility of uroflow measurements: experience using J.J.M.C.H. de la Rosette, MD, Urologist. a double blind placebo controlled study of doxazosin in N.F. Dabhoiwala, MD, Urologist. benign prostatic hyperplasia. Urology 1996; 47: 658 61 D. Osawa, MD, Urologist. 14 Dunsmuir WD, Feneley MR, Corry DA, Bryan J, Kirby RS. A. Tong Long Lim, MD, Urologist. The day-to-day variation (test-retest reliability) of residual P. Abrams, MD, FRCS, Consultant Urologist. urine measurement. Br J Urol 1996; 77: 192 3 Correspondence: Mr J.M. Reynard, Department of Urology, 15 Schafer W, Noppeney R, Ruebben H, Lutzeyer W. The value of free flow rate and pressure/flow-studies in the Christchurch Hospital, Private Bag 4710, Christchurch, New Zealand.