The relationship between urinary symptom questionnaires and urodynamic diagnoses: an analysis of two methods of questionnaire administration

Transcription

1 BJOG: an International Journal of Obstetrics and Gynaecology May 2004, Vol. 111, pp DOI: /j x The relationship between urinary symptom questionnaires and urodynamic diagnoses: an analysis of two methods of questionnaire administration Mohid S. Khan, Charlotte Chaliha, Lucia Leskova, Vikram Khullar* Objective To assess whether method of administration of a standard urinary symptom questionnaire alters the relationship of symptoms with urodynamic diagnoses. Design Randomised crossover study. Setting Tertiary Urogynaecology Unit, London, UK. Participants One hundred and fourteen women attending a tertiary urogynaecology clinic. Methods Women were randomised to either an initial interview-assisted questionnaire in the clinic with a follow up postal questionnaire or an initial pre-outpatient questionnaire followed by an interview-assisted questionnaire at the clinic visit. Videocystourethrography or saline cystometry was performed at the clinic visit. Main outcome measures Question responses were compared with urodynamic diagnoses. Results With an interview method, only severity of incontinence was significantly associated with detrusor overactivity (U ¼ 593.5, P ¼ 0.012). With self-completion, severity of nocturia (U ¼ 477, P < 0.05), urgency (U ¼ 395, P ¼ 0.003), urge urinary incontinence (U ¼ 392, P ¼ 0.003), leakage without warning (U ¼ 443, P ¼ 0.035) and incomplete voiding (U ¼ 413, P ¼ 0.01) were significantly associated with detrusor activity. On interview the symptom of stress urinary incontinence (U ¼ 523, P ¼ 0.002) and use of pads (U ¼ 564.5, P ¼ 0.011) were significantly associated with a diagnosis of urodynamic stress incontinence. Severity of stress urinary incontinence (U ¼ 276, P < 0.001), frequency of leakage (U ¼ 348.5, P ¼ 0.004), use of protection (U ¼ 432.5, P < 0.018), nocturnal incontinence (U ¼ 393.5, P ¼ 0.002) and quantity of leakage (U ¼ 441.5, P < 0.05) on self-completion were strongly associated with diagnosed urodynamic stress incontinence. There was no association between the symptoms of urgency or urge incontinence and the urodynamic stress incontinence. Conclusions Postal questionnaire responses have a better relationship with urodynamics, both for urodynamic stress incontinence and detrusor over activity, than interview-assisted questionnaire responses. However, no symptom has a high enough specificity and sensitivity to replace urodynamic testing. INTRODUCTION The bladder has previously been described as an unreliable witness because of the poor relationship between urinary symptoms and urodynamic diagnoses. 1 Therefore, evaluation of urinary incontinence relies on urodynamic investigation as a gold standard. Several studies have demonstrated that where stress incontinence is the only symptom, then urodynamic stress incontinence is likely to be present in 90% of cases but the relationship between urge incontinence and detrusor overactivity is less evident. 2 Comparing clinical and urodynamic Department of Obstetrics and Gynaecology, Mint Wing, St Mary s Hospital, Imperial College, London, UK * Correspondence: Mr V. Khullar, Department of Obstetrics and Gynaecology, Mint Wing, St Mary s Hospital, Imperial College School of Medicine, Norfolk Place, London W2 1PG, UK. D RCOG 2004 BJOG: an International Journal of Obstetrics and Gynaecology diagnoses for 100 women, agreement has been seen in 68% of cases of urodynamic stress incontinence but in only 51% of cases of detrusor overactivity. 3 Of those with urodynamic stress incontinence, almost all had symptoms of stress incontinence, 46% also had urgency. Of those with detrusor overactivity, 26% also had symptoms of stress incontinence. Positive predictive values of 87% and 41% for diagnosing pure urodynamic stress incontinence and pure detrusor overactivity, respectively, have been reported. 4 However, despite reasonable sensitivity for symptoms, specificity was poor. Similar findings have been seen using a detailed urinary symptom questionnaire (positive predictive value of 80% for urodynamic stress incontinence, and only 25% for detrusor overactivity), 5 while another study using analysis of symptoms for the prediction of urodynamic stress incontinence found correct classification in 81% with a false positive rate of 16%. 6 However, the above studies did not use uniform methods to collect urinary symptom questionnaires and investigate the association with urodynamic diagnoses. Some of the

2 ADMINISTRATION METHOD OF URINARY SYMPTOM QUESTIONNAIRES 469 questionnaire when completing the second. A randomised crossover design was used where those willing to take part in the study were randomly allocated, using randomised blocks, to either: Group A an initial interview-assisted questionnaire at the urodynamic clinic with a follow up postal questionnaire; or Group B an initial pre-outpatient postal questionnaire followed up by a second interview-assisted questionnaire while attending a scheduled urodynamic clinic. Fig. 1. Flow chart showing number of patients through each stage of study. differences found in the relationship of symptoms and diagnoses may be due to methods of questionnaire administration. Analysis of the relationship between symptom questionnaires and urodynamic diagnoses is important, as it would allow us to assess whether population studies accurately reflect the prevalence of urinary symptoms. In addition, if symptoms were highly associated with urodynamic diagnosis, urodynamics being more time consuming and invasive might not be required. The aim of this study was to assess whether differences in administration of a standardised urinary symptom questionnaire alter the relationship of urinary symptoms with urodynamic diagnoses. METHODS Women referred to a tertiary urogynaecology clinic with lower urinary tract symptoms were studied. The Bristol Female Lower Urinary Tract Symptoms questionnaire was used to assess urinary symptoms. 7 It is composed of domains for lower urinary tract symptomatology, sexual function and quality of life (bothersomeness and impact). It covers 8 questions relating to incontinence and 12 others to lower urinary tract symptoms. Response scales for symptoms range from none to always with no problem to a serious problem for bothersomeness. This instrument has been shown to have good psychometric validity and reliability and was originally designed for self-completion. No study has compared interview-assisted to self-completion for this or other urinary symptom questionnaires. Only the urinary symptom and sexual function domains were included together with their bothersomeness. The quality of life domain was excluded as validated data had not been published. Randomisation by method of administration was preferred to remove the effect of patient recall of the initial Postal questionnaires were sent to reach patients seven days before or after their urodynamic clinic appointment was scheduled. Reminder letters were sent if a questionnaire had not been returned by a Group A woman after seven days following the expected date of receipt. Clinic interviews were carried out in as consistent a manner as possible, by the same (male) clinician, in the same environment. Another clinician performed a clinical evaluation of the women including complete history, vaginal examination and investigated them with videocystourethrography (Aquarius XLT, Laborie Medical Tech., Montreal, Canada) or saline cystometry (CMG) (Dorado, Laborie Medical Tech.) using a standardised protocol. The interviewer was blinded to the urodynamic test results. The patient was informed of her urodynamic diagnosis but not of the expected relationship of specific symptoms and urodynamic diagnosis. Each woman was asked to attend the clinic with a comfortably full bladder. After uroflowmetry (with the woman voiding in private recorded with a gravimetric flowmeter), urinalysis was performed. If urinalysis was positive to blood, protein, nitrates and leucocytes, these women were excluded and an midstream urine specimen was sent. With the patient supine, the bladder was filled with room temperature saline solution for CMG, or X-ray contrast medium (Isopaque Cysto 100 mg/ml) for videocystourethrography at 100 ml/min. The patient was asked to cough at 1-minute intervals throughout filling (i) to ensure good subtraction; (ii) to test for stress incontinence; and (iii) as a Table 1. Comparison of urodynamic diagnoses between groups [m 2 ¼ 4.47, P > 0.3]. Values are expressed as n (%). Urodynamic Diagnosis Group A (Interview first) (n ¼ 32) Group B (Self first) (n ¼ 37) Total (n ¼ 69) Pure urodynamic 7 (22) 15 (41) 22 (32) stress incontinence Pure detrusor overactivity 11 (34) 11 (30) 22 (32) Mixed urinary incontinence 7 (22) 4 (11) 11 (16) Normal urodynamic study 6 (19) 7 (19) 13 (19) Other 1 (3) 0 1 (1)

3 470 M.S. KHAN ET AL. Table 2. Sensitivities and specificities for pure symptoms. Pure stress urinary incontinence for urodynamic stress incontinence Pure urge urinary incontinence for detrusor overactivity Interview Self-completion Interview Self-completion Sensitivity Specificity provocative manoeuvre to elicit detrusor contractions. Once 600 ml had been infused into the bladder or a strong desire to void had developed, filling was discontinued and the filling catheter was removed. On standing, provocative manoeuvres were carried out, such as listening to running water (turning a tap on), washing hands in cold water and asking the woman to cough one, three and five times with maximum effort while watching for any urinary leakage. Finally, the woman was seated to void in private to perform a pressure-flow study. If the patient was unable to empty her bladder, the pressure lines were removed and the woman was asked to void once more. Urodynamic diagnoses of urodynamic stress incontinence, detrusor overactivity and mixed urinary incontinence were made according to International Continence Society definitions. 8 An absence of urodynamic diagnosis in spite of urinary symptoms was recorded as a normal urodynamic study. Statistical analysis was carried out using the SPSS program version 10.0 (SPSS, Chicago, Illinois, USA). Period effects and method of administration period interactions were first ascertained by a crossover trial technique described by Altman. 9 Individual question responses for symptom severity and bothersome scales were analysed for association with presence or absence of a particular diagnosis by the Mann Whitney U test. Receiver Operational Characteristic (ROC) curves were used to explore their diagnostic value. This was achieved by plotting sensitivity versus 1 specificity at each possible cutoff for each predictive question. Areas under curves indicate accuracy of the question as a diagnostic test with 0.5 representing a worthless test, 1 a perfect test. Table 4. Question responses significantly associated with presence of detrusor overactivity. RESULTS Self-completion Interview U P U P Symptom scales Nocturia N/S N/S Urgency N/S N/S Urge urinary incontinence Leakage without warning N/S N/S Incomplete voiding N/S N/S Stress urinary incontinence N/S N/S Bothersomeness questions Frequency 320 <0.001 N/S N/S Nocturia N/S N/S Urgency 325 < Urge urinary incontinence 334 < Pain N/S N/S Leakage without warning N/S N/S Nocturnal incontinence N/S N/S Incomplete voiding N/S N/S N/S ¼ non-significant data. A total of 114 women were randomised for initial questionnaire administration method to achieve 72 completed sets (63%). Of these, three did not have urodynamic diagnoses. Among the 42 women invited but who did not complete the study, 22 cancelled or did not attend their appointment; 4 refused to be interviewed but self-completed; 15 agreed and were interviewed but failed to return their self-completed questionnaire; and one had surgery after her initial interview. Thus, 21 women partially completed the study (Fig. 1). Women randomised to Group B (initial questionnaire by self-completion) were more likely to complete both questionnaires than Group A (90% vs 70%). The mean age and ranges were 55.5 (25 86) and 52.9 (24 77) for Groups A and B, respectively, a difference which was not statistically significant. There was also no significant difference between the groups for urodynamic diagnoses (Table 1) (m 2 ¼ 4.47, P > 0.3). Table 3. Sensitivities and specificities for any symptom. Any stress urinary incontinence for urodynamic stress incontinence Any urge urinary incontinence for detrusor overactivity Interview Self-completion Interview Self-completion Sensitivities and specificities for pure symptoms Sensitivity Specificity Sensitivities and specificities for any symptoms Sensitivity Specificity

4 Table 5. Question responses significantly associated with presence of diagnosis of stress incontinence (urodynamic stress incontinence and mixed). ADMINISTRATION METHOD OF URINARY SYMPTOM QUESTIONNAIRES 471 Self-completion Interview U P U P Symptom scales Stress urinary incontinence 276 < Frequency of leakage Use of protection Nocturnal incontinence Quantity of leakage Urge urinary incontinence Bothersomeness questions Stress urinary incontinence Frequency of leakage Nocturnal incontinence Sensitivities and specificities of symptoms for urodynamic diagnoses were different depending on whether symptoms were considered pure or not (Tables 2 and 3). On interview, the finding of urodynamic stress incontinence (both pure urodynamic stress incontinence and mixed incontinence) occurred in 30 (46%) of the 65 patients with symptoms of stress urinary incontinence (both pure stress urinary incontinence and mixed symptoms). With selfcompleted questionnaires, 30 (55%) of those with stress incontinence symptoms had urodynamic stress incontinence diagnosed. The predictive value of responses to several urinary symptom questions on the Bristol Female Lower Urinary Tract Symptoms questionnaire was explored. With an interview method, only symptom questions dealing with severity of incontinence were significantly associated with detrusor overactivity (Table 4) (U ¼ 593.5, Fig. 3. ROC curve: frequency questions as diagnostic tests for detrusor overactivity; (I) ¼ Interview-assisted; (S) ¼ Self-completed. P ¼ 0.012). However, there were significant associations between degree of bother caused by urgency and detrusor overactivity (U ¼ 564, P ¼ 0.008), and between degree of bother of urge incontinence and detrusor overactivity (U ¼ 599.5, P ¼ 0.014). There was no association between stress incontinence and detrusor overactivity. With self-completion, symptom questions dealing with severity of nocturia (U ¼ 477, P ¼ 0.048), urgency (U ¼ 395, P ¼ 0.003), urge urinary incontinence (U ¼ 392, P ¼ 0.003), leakage without warning (U ¼ 443, P ¼ 0.035) and incomplete voiding (U ¼ 413, P ¼ 0.01) were significantly associated with detrusor overactivity. These are mainly irritative symptoms. In addition, the degree of bother caused by frequency of micturition (U ¼ 320, P < 0.001), nocturia (U ¼ 418, P ¼ 0.007), urgency (U ¼ 325, P < 0.001), urge urinary incontinence (U ¼ 334, P < 0.001), pain (U ¼ 404, P ¼ 0.009), leakage without warning (U ¼ 385, P ¼ 0.004), nocturnal incontinence (U ¼ 483.5, P ¼ 0.039) and incomplete emptying (U = 388.5, P = 0.006) was strongly Fig. 2. ROC curve: urge urinary incontinence questions as diagnostic tests for detrusor overactivity; (I) ¼ Interview-assisted; (S) ¼ Self-completed.

5 472 M.S. KHAN ET AL. Fig. 4. ROC curve: nocturnal incontinence questions as diagnostic tests for detrusor overactivity; (I) ¼ Interview-assisted; (S) ¼ Self-completed. associated with the presence of detrusor overactivity. When answers obtained by interview were analysed, symptom questions concerning stress incontinence (U ¼ 523, P ¼ 0.002) and use of protection (pads) (U ¼ 564.5, P ¼ 0.011) were significantly associated with a diagnosis of urodynamic stress incontinence and mixed incontinence (Table 5). The degree of bother caused by stress urinary incontinence (U ¼ 548, P ¼ 0.003) and frequency of leakage (U ¼ 551, P ¼ 0.009) was associated with a urodynamic stress incontinence diagnosis. There was no association between urge incontinence or urgency and a urodynamic stress incontinence diagnoses. Severity of stress incontinence (U ¼ 276, P < 0.001), frequency of leakage (U ¼ 348.5, P ¼ 0.004), use of protection (U ¼ 432.5, P ¼ 0.018), nocturnal incontinence (U ¼ 393.5, P ¼ 0.002) and quantity of leakage (U ¼ 441.5, P ¼ 0.037) on self-completion were strongly associated with diagnosed urodynamic stress incontinence. In addition, bothersomeness of stress incontinence (U ¼ 331.5, P ¼ 0.001), frequency of leakage (U ¼ 551, P ¼ 0.009) and nocturnal incontinence (U ¼ 402.5, P ¼ 0.001) were each associated with this diagnosis (Table 5). Employing ROC guidelines, fair tests (area ) for the presence of detrusor overactivity included the symptom scale of urge incontinence (Fig. 2), and bothersomeness of frequency (Fig. 3), urgency and urge incontinence but only when self-completed, not interviewed. Poor (area under curve but still >0.5) diagnostic tests included selfcompleted and interviewed questions on urgency. Many other self-completed questions were poor tests for detrusor overactivity, including nocturnal incontinence (Fig. 4), incomplete voiding, leakage without warning and changing of outer clothes. All other questions (encompassing many interviewed) were classed failures (area ) as diagnostic tests. Fair tests for the presence of urodynamic stress incontinence and mixed incontinence included symptom scale and bothersomeness of stress incontinence (Fig. 5), both on self-completion and interview. Poor diagnostic tests included both interview and self-completed questions on frequency of leakage and use of protection. There were no good (area ) or excellent (area 0.9 1) tests for any diagnosis. DISCUSSION It is recognised that response rates are lower for postal questionnaires than interviews in other areas of medicine, 9 and this was the case in this study. Response bias may exist where non-response may be connected to particular aspects Fig. 5. ROC curve: stress urinary incontinence questions as diagnostic tests for urodynamic stress incontinence; (I) ¼ Interview-assisted; (S) ¼ Selfcompleted.

6 ADMINISTRATION METHOD OF URINARY SYMPTOM QUESTIONNAIRES 473 of incontinence due to its embarrassing nature. Because information was not available on non-responders, this cannot be explored but should be taken into account when interpreting these data. There was no period effect in this study and no method of administration period interaction between the two groups of women for any item was found, which allowed analysis to be pooled between the two groups of women in the study. There were few women complaining of pure symptoms (pure stress urinary incontinence or urge urinary incontinence) in this study. This may simply be due to more women with mixed symptoms being referred to the tertiary urogynaecology clinic since another study based on clinic patients had similar proportions of pure symptomatology. 11 The urinary symptoms of pure stress urinary incontinence and urge urinary incontinence had poor sensitivity but a very high specificity for urodynamic stress incontinence and detrusor overactivity (Table 2). This was not dependent on the method of questionnaire administration. This could be attributed to the finding that many of those who complained of mixed incontinence were actually diagnosed with only urodynamic stress incontinence or detrusor overactivity. Even if urge urinary incontinence is a central feature of detrusor overactivity, and stress urinary incontinence of urodynamic stress incontinence, it could be concluded that pure symptoms are poor predictors of urodynamic diagnosis but highly specific. However, few women had pure symptoms (18%), which limit their application in a symptomatic population. When symptoms were not just considered as pure entities (mixed incontinence included), sensitivities became high, and specificities low with little difference between self-completion and interview methods. This is not surprising as this includes mixed symptoms and diagnoses. Sensitivities were similar to those in a previous analysis but specificities were found be lower. 12 More useful findings came from analysis of scales of the Bristol Female Lower Urinary Tract Symptoms questionnaire. The total Bristol Female Lower Urinary Tract Symptoms scores could only distinguish normal urodynamics from a urodynamic diagnosis on self-completion and not on interview. If total scores are reflective of overall lower urinary tract problems, then the lower the total score, the less likely anything abnormal is detected on urodynamic investigation. Women with mixed incontinence diagnoses had significantly higher scores than those with other diagnoses but only on self-completion. This may well be due to their suffering a combined disorder with higher scores in a greater number of questions. This suggests that the postal questionnaire responses produce greater separation of symptomatic groups and may have a better relationship with urodynamics than interview-assisted questionnaire responses do. However, addition of the score as a total was not the original intention of the questionnaire and one should be cautious as the questions have not been weighted. Nevertheless, changing the method of administration did alter the relationship of symptoms to urodynamic diagnoses. Using both Mann Whitney U test and areas under ROC curves as tests for individual items, several questions were detected for prediction of urodynamic diagnoses. We are aware that one of the limitations of the statistical analysis was that multiple significant tests were undertaken on the same subjects. However, significant differences were still revealed after applying a Bonferroni-type correction factor. In this analysis as well, self-completed questionnaires showed more relationship with urodynamics. More questions by self-completion than by interview were associated with detrusor overactivity. These were mainly irritative symptoms and bother questions (e.g. nocturia, urgency, urge urinary incontinence), which are key features of detrusor overactivity. Both severity and bother of leakage without warning were associated with detrusor overactivity on self-completion but not on interview. Urge urinary incontinence, a feature of detrusor overactivity, is characterised by leakage with a sensation of urgency and the leakage without warning question may have been mistaken for urge urinary incontinence on selfcompletion. When interviewed, there was no such association, implying in an interview the distinction between urge urinary incontinence and leakage without warning is made clear. Nocturnal incontinence scores were higher with the presence of detrusor overactivity (but only on selfcompletion), suggesting that these women s nocturnal incontinence may be linked to detrusor overactivity. This may be consistent with work suggesting this symptom is attributable to several causes including uninhibited detrusor contractions as occurs in detrusor overactivity. 13 More question scores associated with urodynamic stress incontinence were found on self-completion compared with interview. Both stress urinary incontinence and bother of stress urinary incontinence were found to be associated with urodynamic stress incontinence on both methods. This is to be expected as stress urinary incontinence is a simple symptom to detect and can be clearly defined to a patient. This study both conflicts and supports previous work. Fitzgerald found that symptom scores of the short-form questionnaires, the UDI6 and IIQ7, were inadequate predictors of eventual urodynamic diagnoses, especially in those with advanced uterovaginal prolapse. 11 This may be due to the number of questions being reduced. However, Lemack and Zimmern 14 did find that certain items on the UDI6 did provide predictive information regarding urodynamic findings in women even if no question could estimate severity as determined by Valsalva leak point pressure, a clinical measure of severity. These questionnaires are similar to the Bristol Female Lower Urinary Tract Symptoms questionnaire with their likert-based symptom and bothersome scales but are in short form as opposed to the full form used in this study.

7 474 M.S. KHAN ET AL. In this study, we have found that differences in administration method of the Bristol Female Lower Urinary Tract Symptoms questionnaire change relationships with urodynamic diagnoses with more questions having predictive value on self-completion. This indicates that the Bristol Female Lower Urinary Tract Symptoms questionnaire has a stronger relationship to an objective measure such as urodynamic diagnosis when self-completed in the woman s home compared with interview. This is reinforced by previous work showing interviews improving health status measured with the SF-36 questionnaire. 15 However, no symptom question can predict urodynamic diagnosis to the extent that women can forgo urodynamic investigation. References 1. Blaivas JG. The bladder is an unreliable witness. Neurourol Urodyn 1996;15(5): Hastie KJ, Moisey CU. Are urodynamics necessary in female patients presenting with stress incontinence. Br J Urol 1989;63(2): Jarvis GJ, Hall S, Stamp S, Millar DR, Johnson A. An assessment of urodynamic evaluation in incontinent women. Br J Obstet Gynaecol 1980;87(10): Cundiff GW, Harris RL, Coates KW, Bump RC. Clinical predictors or urinary incontinence in women. Am J Obstet Gynecol 1997;177(2): Bergman A, Bader K. Reliability of the patient s history in the diagnosis of urinary incontinence. Int J Gynecol Obstet 1990;32: Versi E, Cardozo L, Anand D, Cooper D. Symptoms analysis for the diagnosis of genuine stress incontinence. Br J Obstet Gynaecol 1991; 98(8): Jackson S, Donovan J, Brookes S, Eckford S, Swithinbank L, Abrams P. The Bristol Female Lower Urinary Tract Symptoms questionnaire: development and psychometric testing. Br J Urol 1996;77: Abrams P, Cardozo C, Fall M, et al. The standardisation of terminology in lower urinary tract function: report from the standardisation sub-committee of the International Continence Society. Neurourol Urodyn 2003;61(1):37 49 (January). 9. Altman DG. Practical Statistics for Medical Research. London: Chapman & Hall, Oppenheim AN. Questionnaire Design, Interviewing and Attitude Measurement. London: Pinter Publishers, Fitzgerald MP, Brubaker L. Urinary incontinence symptom scores and urodynamic diagnoses. Neurourol Urodyn 2002;21(1): Jensen JK, Nielsen Jr FR, Ostergard DR. The role of patient history in the diagnosis of urinary incontinence. Obstet Gynecol 1994;83(5 Pt 2): Ouslander JG. Aging and the lower urinary tract. Am J Med Sci 1997; 314(4): Lemack GE, Zimmern PE. Predictability of urodynamic findings based on the Urogenital Distress Inventory-6 questionnaire. Urology 1999;54(3): Lyons RA, Wareham K, Lucas M, Price D, Williams J, Hutchings HA. SF-36 scores vary by method of administration: implications for study design. J Public Health Med 1999;21(1): Accepted 14 November 2003