Development and validation of a questionnaire for the assessment of bowel and lower urinary tract symptoms in women

BJOG: an International Journal of Obstetrics and Gynaecology April 2002, Vol. 109, pp. 413 –423

Development and validation of a questionnaire for the assessment of bowel and lower urinary tract symptoms in women L. Hillera, S. Radleyb,*, C.H. Mannc, S.C. Radleyd, G. Begume, S.J. Pretlovec, J.H. Salamanb Objective To develop a simple but sensitive instrument to evaluate and document symptoms of both bowel and urinary dysfunction in women. Design A 22-item questionnaire covering a range of bowel and urinary symptoms was developed and underwent rigorous psychometric testing. Setting The gynaecology departments of three hospitals, a urogynaecology clinic, a functional bowel clinic and a general practice. Population Six hundred and thirty women, comprising four groups: 1. women awaiting hysterectomy (n ¼ 379), 2. women following hysterectomy (n ¼ 45), 3. women referred with functional bowel and/or urinary symptoms (n ¼ 65), 4. asymptomatic controls (n ¼ 141). Main outcome measures The content, construct and criterion validity, internal consistency, reliability and responsiveness of the questionnaire were measured. Results Peer and patient reports and missing data patterns supported face and content validity. Factor analysis showed a clinically relevant four-factor structure with low content replication able to distinguish between patient groups, indicating good internal structure. Comparison with clinical, anorectal physiological, videoproctographic, transit time and urodynamic test results provide provisional indication of criterion validity. Key domain question analysis and Cronbach’s alphas showed internal consistency. Kappa values demonstrated good test –retest reliability and key question correlation over time proved responsiveness. Conclusions Our findings support the suitability, clinical validity, reliability and responsiveness of a simple questionnaire, which is sensitive to the constraints of clinical practice. The authors recommend its use in health care evaluation research assessing the effects of pelvic surgery and as a useful tool in comparing treatment efficacy. INTRODUCTION Pelvic floor pathology in women encompasses a wide spectrum of functional, anatomical, neurological, hormonal and neoplastic conditions, many of which are associated with symptoms of bowel and/or urinary dysfunction. Such co-morbidity is easily overlooked and the influence of

a

Birmingham Clinical Trials Unit, Edgbaston, Birmingham, UK b University of Birmingham Department of Surgery, Queen Elizabeth Hospital, Birmingham, UK c Birmingham Women’s Hospital, Edgbaston, Birmingham, UK d Central Sheffield University Hospitals Department of Obstetrics and Gynaecology, Sheffield, UK e Cancer Research UK Trials Unit, Institute for Cancer Studies, Edgbaston, Birmingham, UK * Correspondence: Mr S. Radley, University of Birmingham Department of Surgery, Queen Elizabeth Hospital, Birmingham B15 2TH, UK. D RCOG 2002 BJOG: an International Journal of Obstetrics and Gynaecology PII: S 1 4 7 0 - 0 3 2 8 ( 0 2 ) 0 1 1 4 7 - 3

pelvic surgery on various aspects of bowel and urinary function are still poorly understood. Bowel evacuation and anal incontinence are intrinsically related to pelvic floor function. To consider bowel function purely in terms of defaecatory frequency and constipation is thus an over-simplification and fails to take into consideration other important elements. Many patients use subjective observations when describing bowel or evacuatory dysfunction which include not only stool consistency, but also impaired evacuation, straining, digitation, tenesmus and anorectal discomfort. Thompson et al.1 proposed a symptomatic definition of constipation based on the existence of at least two of the following; straining, lumpy or hard stools, a sensation of incomplete evacuation or less than two bowel actions per week. Bowel or evacuatory dysfunction may result from a variety of underlying pathology including pelvic outlet obstruction, slow transit or other mechanical, pharmacological, metabolic, endocrine and neurogenic abnormalities2. Consequently, analysis of function needs to take account of not only stool frequency, but also symptoms of disordered evacuation. The evaluation of evacuation has hitherto relied upon clinical assessment and imaging3. There are currently no validated www.bjog-elsevier.com

414

L. HILLER ET AL.

questionnaires that can be used to assess disordered evacuation in the female population. The prevalence of ‘constipation’ and/or the irritable bowel syndrome have been assessed in women 4 – 6 . Although anal incontinence occurs more frequently in women than men7, there are few studies relating to anal incontinence in gynaecological patients. Symptoms associated with anal incontinence frequently reflect the underlying pathological cause8. Thus, urgency (the inability to defer defaecation) and urge incontinence usually indicate loss of voluntary control due to impaired external anal sphincter function. Passive incontinence of stool frequently indicates impairment of the involuntary smooth muscle of the internal anal sphincter. Clearly, impairment of both sphincters may coexist. Nevertheless, it is important to take into account the differing symptom patterns in anal incontinence in making a clinical assessment. While scoring systems exist for the clinical assessment of anal incontinence9 – 11, there is a paucity of self-completed questionnaire based scoring systems. A scoring system based on linear analogue scales has been proposed, however this instrument does not address other aspects of pelvic floor function and has not been widely used12. A questionnaire for assessment of anal incontinence, associated symptoms and risk factors has been validated13 but the instrument is overlong and test – retest reliability was low in some areas14. A recently validated quality of life instrument measures impact of anal incontinence over four scales of quality of life, but does not measure physical symptom severity15. A number of questionnaires to assess lower urinary tract symptoms in women have been designed and validated16 – 19. The Bristol Female Lower Urinary Tract Symptoms (BFLUTS) Questionnaire provides a comprehensive measure of the severity of lower urinary tract symptoms and their degree of bother16. The King’s Health Questionnaire has been widely used and validated in the context of lower urinary tract symptom evaluation and their impact on quality of life17. These questionnaires do not address other aspects of pelvic floor function such as bowel dysfunction and anal incontinence. While individual questionnaires have been designed to assess bowel and urinary symptoms independently, there is a clear need for a simple but sensitive instrument that can address both areas simultaneously. This would be a valuable tool not only for assessing symptom severity before pelvic surgery to determine which women might require further specific functional investigations, but also for the evaluation of symptomatic outcome following pelvic surgery.

detailed literature review, questions covering a wide range of bowel symptoms relating to stool frequency, stool consistency, incontinence and evacuatory dysfunction as well as urinary symptoms were devised. A self-completed questionnaire is the ideal, as clinical assessment may require a postal survey. Such instruments must be easily understood and concise, while maintaining sufficient clarity in text size and structure, to promote a high rate of return. The resultant questionnaire, the Birmingham Bowel and Urinary Symptoms Questionnaire (BBUSQ-22), includes 22-items, asking women to record bowel and urinary symptoms over the previous month. Four items assessed faecal incontinence in terms of both passive and urge incontinence. Seven items addressed evacuatory function, one item assessed stool frequency and one item stool consistency. Two items assessed frequency of laxative use or medical consultation. Urinary symptoms were assessed in terms of daytime/nocturnal frequency, urge and stress incontinence, urinary leakage and incomplete voiding. The structure of the urinary questions closely mirrors the format of the Bristol Female Lower Urinary Tract Symptoms questionnaire16, which taps the most common urinary symptoms. However, the response sets were amended to include only four responses, deemed ample for differentiation between severities, and less intensive for the patient. Original items relating to abdominal pain and bloating were considered extremely subjective symptoms and were omitted. Pain on micturition was also considered extremely subjective and misunderstood by patients and was therefore excluded. Where possible, the questionnaire was designed so that the response options for the questions followed the same format: never, occasionally, most of the time and always. This format differed for items relating to stool frequency and consistency, time spent evacuating, medical consultation and frequency of micturition but generally retained a four-level response throughout, thus prohibiting the frequently observed cluster of middle/neutral responses. For the two questions that did not have four-level response sets, one had six levels and one had a binary response set. A previous paper by Mann et al.20 documents the initial application of this instrument in the evaluation of the prevalence of bowel and urinary symptoms in women awaiting hysterectomy. This paper aims to assess the psychometric properties of the questionnaire in terms of its validity, reliability and responsiveness and hence its usefulness in a clinical setting.

METHODS Questionnaire development After consultation with specialist coloproctologists, urogynaecologists, gynaecologists and patients together with a

The present study was designed in two phases. The first comprised of a prospective cohort of 575 women for the determination of validity, internal consistency and D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

A BOWEL AND URINARY TRACT SYMPTOMS QUESTIONNAIRE 415 Table 1. Summary of statistical tests applied. Psychometric quality Feasibility Content validity Construct validity

Internal consistency Criterion validity Reliability Responsiveness

Statistical tests applied

Cases

Symptomatic Controls

Controls

379 baselines

65 baselines

131 baselines

89 Frequencies of comments > > > > Flesch – Kincaid formula > > >> > > > Missing and popular responses> > > > > = < Floor and ceiling effects Factor analysis Kruskal – Wallis tests Spearman’s rank correlations Box and whisker plots Cronbach’s a’s Wilcoxon rank sum tests Kruskal – Wallis tests Kappa coefficients Bland and Altman graphs Spearman’s rank correlations

> > > > > > > > > > > > > > > ; :>

58 with clinical evaluation 45 test – retests

10 test – retests

356 baseline and 6 months

responsiveness. This phase consisted of 379 women undergoing hysterectomy (cases), 131 controls who had not undergone hysterectomy and were not consulting their doctor for bowel or urinary symptoms (recruited from a GP practice and healthcare workers at Wordsley Hospital) and a further group of 65 symptomatic controls either presenting to a urogynaecological clinic with lower urinary tract symptoms or to a functional bowel clinic with anal incontinence or constipation/evacuatory disorder. The hysterectomy group completed the Birmingham Bowel and Urinary Symptoms Questionnaire (BBUSQ-22) either at a pre-admission clinic or after admission for surgery (356 of these returned further completed questionnaires by post approximately six months following hysterectomy (range 4 – 13 months). All the symptomatic and asymptomatic controls completed the BBUSQ-22 at one time point. In the second phase, test –retest reliability was assessed in a cohort of 55 women, consisting of 45 hysterectomy patients (at least two months post hysterectomy) completing the questionnaire twice, on average 18 days apart, and 10 controls who also completed the questionnaire twice, on average 52 days apart. Women under the age of 18 years, women who were unable to understand written or spoken English and any women with a bowel or urinary diversion were excluded. All participants gave informed consent to take part in the study, which received local ethics committee approval.

and to establish questionnaire validity, reliability and responsiveness.

Feasibility Women were supplied with feedback forms and invited to comment on the completion time and on any sections where they thought clarity could be improved. The Flesch Formula21, based on the average number of words per sentence and the average number of syllables per 100 words, was calculated to assess readability of the questions. Scores range from 0 to 100 (the higher the score, the easier to read ) with most standard documents aiming for a score of between 60 and 70.

Content validity To ensure all relevant aspects of female bowel function and urinary symptoms were covered, the questionnaire was assessed by peer review. Participating women were also invited to comment on the content and design of the questionnaire. The incidence of missing responses and distributions of responses were measured for signs of inappropriateness by highlighting the frequency of popular responses and floor and ceiling effects.

Construct validity Statistical methods Table 1 provides a summary of statistical tests applied for each of the psychometric qualities assessed with the number and type of patients included in the analysis. Analyses were carried out, using SAS statistical software (SAS Institute, Cary, North Carolina, USA), to confirm the hypothesised scale structure of the questionnaire D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

Factor analysis with varimax rotation was used to assess whether the structure underlying the questionnaire corresponded to the pre-specified clinical three-factor structure and whether the composition of these factors corresponded to the postulated concepts of constipation/evacuatory function, anal incontinence and urinary symptoms. After reversing one of the question scores to standardise low scores as

416

L. HILLER ET AL.

Table 2. Patient characteristics. Values are n (%) unless otherwise shown. Grouping Age (years) n Median IQR Range

Cases (n ¼ 379)

Symptomatic Controls (n ¼ 65)

Controls (n ¼ 131)

375 46 40 – 55 27 – 86

46 46 38 – 55 23 – 73

129 42 35 – 51 21 – 88

Parity 0 1–3 4þ Unknown

34 (9) 292 (77) 42 (11) 11 (3)

5 46 10 4

(8) (71) (15) (6)

15 100 14 2

(11) (76) (11) (2)

Previous surgery Anorectal/colonic Prolapse surgery Urological Prolapse & Urological surgery None specified

10 (3) 2 (0.5) 5 (1) 2 (0.5) 360 (95)

4 1 0 0 60

(6) (2) (0) (0) (92)

1 4 2 0 124

(1) (3) (1) (0) (95)

Type of hysterectomy Vaginal Total abdominal Subtotal Wertheims Laparoscopic assisted

80 (21) 254 (67) 37 (10) 5 (1) 3 (1)

— — — — —

— — — — —

IQR ¼ Interquartile range; — not done.

representing normal responses and high scores more abnormal responses, this analysis was carried out on the 21 questions with response sets consisting of a minimum of four levels. All questions were transformed onto a four-point scale. Domain scores were then calculated, providing there were less than half the questions with missing responses, as the mean (with equal weights) of all the answered questions in a domain, hence adjusting for missing values. The domain scores were then linearly transformed to produce values ranging from 0 to 100. Domain scores were examined for signs of floor and ceiling effects and high proportions of missing questions. Their ability to distinguish between the three different patient groups was assessed with Kruskal – Wallis tests to evaluate clinical validity. Inter-domain associations were assessed using Spearman’s rank correlation coefficients for signs of content replication, and item – domain correlations were considered for validity of creating these summary scores.

Internal consistency One key question from the questionnaire for each domain, hypothesised to measure the key aspect of that domain, was identified. For each domain, box and whisker plots of baseline domain scores against responses to the

corresponding key question were then plotted to give a visual impression of internal consistency. Cronbach alphas22, calculated for each of the domain scores over all patients, measured the homogeneity of the scales. This reflects how well the different questions compliment each other in their measurement of different aspects of the same concept. Internal consistency estimates of modest magnitude but > 0.70 were sought23.

Criterion validity In the absence of a consensual accepted standard or criterion with which relationships with alternatives can be measured, differences in domain scores were assessed between groups defined by results from clinical, anorectal physiological, videoproctographic, transit time and urodynamic tests using Wilcoxon rank sum or Kruskal – Wallis tests accordingly. The subset of symptomatic controls with constipation/ evacuatory dysfunction were assessed by results from videoproctography and/or colonic transit studies. Symptomatic controls with anal incontinence were assessed by findings from anal ultrasonography (normal sphincters or sphincter defects) and anal physiology (normal or abnormal basal and squeeze pressures). The subset of symptomatic controls with lower urinary tract symptoms were assessed by urodynamic investigations; normal, detrusor instability D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

A BOWEL AND URINARY TRACT SYMPTOMS QUESTIONNAIRE 417 Table 3. Domain structures from factor analysis. No.

Question

Proposed clinical scale*

Maximum response frequency (%)

Significant factor coeffs after rotation F1

1 How often do you open your bowels? 2 What form are your motions usually? 3 Can you hold onto your motions for more than five minutes? 4 Do you ever have to rush to the toilet to open your bowels? 5 Does stool leak before you can get to the toilet? 6 Do you leak stool for no obvious reason and without feeling that you want to go to the toilet? 7 Do you have to strain to open your bowels? 8 How long do you spend on the toilet, on average, for each bowel action? 9 Do you feel that you cannot completely empty your bowels? 10 Do you use a finger or pressure to help open your bowels? 11 Do you use a finger in your vagina to help open your bowels? 12 Do you have the urge to open your bowels but are unable to pass a motion? 13 Do you find it painful to have your bowels open? 15 Do you use laxatives? 16 During the day, how many times do you urinate, on average? 17 During the night, how many times do you have to get up to urinate, on average? 18 Do you have to rush to the toilet to urinate? 19 Do you have difficulty completely emptying your bladder? 20 Does urine leak before you can get to the toilet? 21 Does urine leak when you are active, exert yourself, cough or sneeze? 22 Does urine leak for no obvious reason and without feeling that you want to go to the toilet?

F2

0.36 0.34

F3

C C I

(45) (76) (45)

0.56 0.57 0.65

I

(65)

0.73

I I

(76) (91)

0.77 0.64

C C

(67) (63)

0.76 0.41

C

(54)

0.63

C

(90)

0.54

C

(95)

0.44

C

(51)

0.62

C C U

(53) (81) (49)

0.67 0.65

U

(52)

0.53

U U

(61) (73)

0.69 0.44

U U

(50) (50)

0.80 0.75

U

(79)

0.72

0.49

* C ¼ Constipation; U ¼ Urinary; I ¼ Incontinence. Question No 14, the binary question, was not included in the factor analysis. All factor loading of 0.30 or above are shown.

or genuine stress incontinence, the number of presenting symptoms described by the women and whether previous gynaecological surgery had been undertaken.

Reliability For the 55 women who completed test – retest questionnaires, Kappa coefficients24 (representing the proportion of agreement beyond that expected by chance alone), were calculated for the domain scores. These assess the ability of the instrument to elicit stable responses when no change has occurred in the underlying condition. Bland and Altman graphs25 for cases alone were plotted to visually assess any relationship between the change in scores and the average scores checking that the accuracy was not linked with severity of the condition. D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

Responsiveness (sensitivity to change) For the 356 cases who completed baseline and six months questionnaires, correlations between the change from baseline to six months in the key questions, previously identified, and the corresponding change in their domain score were assessed using Spearman’s rank correlation coefficients to determine the questionnaire’s responsiveness.

RESULTS Demographic details of the patient groups used in this study are shown in Table 2, including the frequency of previous anorectal, urological or prolapse surgery in the three groups. The type of hysterectomy planned is also shown.

418

L. HILLER ET AL.

Table 4. Domain scores by patient groups. Cases

Symptomatic Controls

Controls

P*

Floor/ceiling responses (%)

Cronbach’s a

Constipation N Median IQR Range

374 53 53 – 63 0 – 100

65 53 33 – 63 0 – 100

130 53 53 – 63 20 – 90

0.0094

(1)

0.49

Evacuation N Median IQR Range

379 13 8 – 21 0 – 75

65 17 8 – 25 0 – 79

130 13 4 – 17 0 – 46

0.0001

(10)

0.76

Incontinence N Median IQR Range

377 17 8 – 25 0 – 92

65 33 17 – 50 0 – 100

130 8 0 – 17 0 – 50

0.0001

(15)

0.72

Urinary N Median IQR Range

378 22 14 – 29 0 – 67

62 33 24 – 48 5 – 81

130 14 5 – 19 0 – 62

0.0001

(4)

0.76

Domain

* P value for distinguishing between groups.

Feedback forms showed the instrument as well understood and easy to complete with an average completion time of five to ten minutes. The questionnaire had a Flesch Reading Ease of 79 indicating only a low reading level necessary proving its accessibility to a wide population. Reports from peers and patients on the content and coverage of the questionnaire showed face validity to be high in that it appeared at face value to be measuring what it set out to measure. From the 575 baseline questionnaires completed in the first phase of the study, 528 (92%) were fully completed and only 4% (23/575) had more than one response missing. Patterns of missing data seemed balanced across patient groups. Individual plots of the responses for each question (not shown) showed little evidence of any highly skewed distributions, indicating that there were no signs of inappropriateness in the content of the questions. The results from the factor analysis, investigating the plausibility of the clinically defined three-factor scale structure, showed reasonable verification of this. Table 3 lists the factor loadings for each item. The initial two questions (fundamentally more constipation orientated than the other more evacuation orientated questions) appeared empirically to fall outside their postulated domain, unique in their split between factors one and three. Despite their factor loadings appearing flexible with regards to being included in either domain, it was deemed clinically relevant to create four factors; constipation (two questions), evacuation (eight questions), incontinence (four questions) and urinary (seven questions). Factor analysis by patient group

led to comparable conclusions. Despite an overall low level of skewed distributions among the 21 questions, a high proportion of the questions with the higher maximum response frequencies were grouped into factor one, labelled now as the evacuation domain. Using the empirically supported four-factor structure, a maximum of five women within each patient group were missing a domain score. All scores showed little evidence of floor and ceiling effects, and all had an ability to distinguish between the different patient groups indicating good internal structure (Table 4). Inter-domain associations showed relatively low correlations between domains (maximum magnitude of r ¼ 0.39 between constipation and incontinence) and hence little chance of replication of content. Item –domain correlations (corrected for overlap) for all women ranged between the desired r ¼ 0.4 and 0.826 with the exception of the two questions in the constipation domain (r ¼ 0.39 for both), three questions in the evacuation domain (r ¼ 0.29, 0.35 and 0.37) and two questions in the urinary domain (r ¼ 0.34 and 0.35). The validity of these summary scores is supported despite the modest infringements indicated.

Internal consistency The constipation domain was represented by the binary constipation question (question 14), which had been excluded from the factor analysis and was thus external to the domain score. The other three domains had key D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

A BOWEL AND URINARY TRACT SYMPTOMS QUESTIONNAIRE 419

Fig. 1. Domain scores and their key question responses. & ¼ cases;

questions within the specific domain; question 15 for the evacuation domain, question 5 for the incontinence domain and question 18 for the urinary domain. The box and whisker plots showed linear rises in domain scores with each increase in symptom suffering supporting both good internal consistency and the linearity of the domain scores (Fig. 1). For three domains, internal consistency, measured by Cronbach’s alpha, over all patients exceeded the 0.70 level (Table 4). Due to alpha values being linked with the number of questions in a domain, the two-question constipation domain fell short of this limit as expected. However, for group comparisons such as in research of treatment evaluations, a value of 0.50 is regarded as acceptable27 to which this approximates. D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

¼ symptomatic controls;

¼ controls.

Criterion validity Of the 58 symptomatic controls with results available from objective clinical measures, these can be subgrouped into 11 women with diagnoses of constipation/ evacuatory dysfunction, 14 with anal incontinence and 33 with lower urinary tract symptoms. Due to the small numbers of women within each of these groups, no significant differences between different results from appropriate clinical tests were detected ( Table 5 ). However, non-significant trends can be seen towards anally incontinent women with abnormal clinical measures having higher scores in the incontinence domain. Also, urinary patients with previous gynaecological surgery have higher domain scores across the board, along with

420

L. HILLER ET AL.

Table 5. Domain scores by patient severity groups. Values are median (IQR). n

Constipation

Evacuation

Incontinence

Urinary

Constipation/evacuation patients Videoproctogram Normal Abnormal

2 8

82 (63 – 100) 58 (27 – 80)

46 (38 – 54) 31 (19 – 50)

17 (8 – 25) 25 (6 – 42)

10 (5 – 14) 19 (5 – 38)

Colonic transit Not delayed Delayed

3 2

73 (0 – 80) 82 (63 – 100)

42 (21 – 54) 46 (38 – 54)

17 (11 – 50) 17 (8 – 25)

5 (5 – 5) 10 (5 – 14)

Anal incontinence patients Max Basal Normal (50 cm H2O) Abnormal (<50 cm H2O)

9 4

43 (17 – 53) 35 (22 – 58)

8 (8 – 17) 17 (8 – 29)

58 (50 – 58) 67 (50 – 88)

29 (29 – 38) 31 (25 – 38)

Max Squeeze Normal (100 cm H2O) Abnormal (<100 cm H2O)

10 3

38 (17 – 53) 43 (17 – 53)

8 (8 – 17) 21 (8 – 29)

58 (50 – 75) 58 (50 – 58)

29 (29 – 38) 33 (29 – 39)

Internal/external sphincter Normal (intact) Abnormal (not intact)

3 11

53 (17 – 73) 43 (17 – 53)

8 (4 – 33) 8 (8 – 21)

42 (17 – 75) 58 (50 – 67)

29 (14 – 43) 29 (29 – 38)

Urinary patients No. of presenting symptoms 1 2 3þ

5 16 12

63 (43 – 70) 57 (40 – 67) 48 (37 – 62)

17 (8 – 17) 19 (13 – 27) 21 (17 – 23)

17 (17 – 33) 25 (17 – 38) 25 (17 – 33)

33 (33 – 48) 38 (29 – 57) 41 (36 – 60)

Previous gynaecological surgery No Yes

16 17

48 (43 – 63) 53 (37 – 70)

17 (13 – 21) 21 (17 – 29)

25 (17 – 33) 25 (17 – 42)

38 (33 – 48) 45 (33 – 64)

Urodynamics Normal Detrusor instability GSI

12 9 7

48 (43 – 70) 37 (37 – 63) 60 (43 – 70)

21 (13 – 27) 17 (17 – 21) 17 (8 – 29)

25 (17 – 42) 33 (17 – 33) 17 (17 – 33)

43 (31 – 54) 38 (33 – 67) 33 (33 – 38)

Shaded blocks of cells indicate corresponding symptoms with domains.

increasing urinary domain scores with an increasing number of presenting symptoms. It can also be seen that the higher scores within each domain are achieved by women suffering the corresponding symptoms (shaded blocks). These results provide provisional indication of criterion validity of the questionnaire.

of questions were answered identically on both occasions, with only 0.5% of responses changed by more than two categories. Bland and Altman graphs (not shown) indicate little or no relationship between the change in scores and the average scores indicating that the accuracy was not linked with severity of the condition. Overall, these results prove good reliability was obtained.

Reliability Patient characteristics of the 55 test – retest women were similar to the main study group. Reliability was assessed with kappas for the non-normally distributed domain scores and gave moderate to high kappa values ( Table 6 ). The kappa values for cases alone, inevitably smaller in magnitude than those generated from the heterogeneous population of the 55 women, were only marginally reduced showing this to be a reliable instrument both in heterogeneous populations and in specialised populations such as women awaiting hysterectomies. 83%

Responsiveness The magnitude of change from baseline to six months, in the previously identified key questions and their respective domain scores, were positively correlated and of varying magnitudes (constipation domain r ¼ 0.08, evacuation domain r ¼ 0.43, incontinence domain r ¼ 0.58 and urinary domain r ¼ 0.58). As in the internal consistency graphs, the constipation domain suffers from its key question having a binary response set leaving little D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

A BOWEL AND URINARY TRACT SYMPTOMS QUESTIONNAIRE 421 Table 6. Changes in scores between test and retest time points. Change in score

Cases (n ¼ 45)

Controls (n ¼ 10)

k 0.80

Constipation Median IQR Range

0 0–0 17 – 17

0 0–0 10 – 10

Evacuation Median IQR Range

0 0–4 14 – 46

0 4 – 8 8 – 17

Incontinence Median IQR Range

0 0–0 17 – 17

0 0–0 17 – 8

Urinary Median IQR Range

0 5 – 5 14 – 14

0 5 – 5 6 – 5

0.43

0.56

0.77

scope for correlation. Overall, these results indicate that the questionnaire is a responsive instrument to measure change over time.

DISCUSSION We have presented the results of a validation study of the Birmingham Bowel and Urinary Symptoms Questionnaire (BBUSQ-22), a questionnaire designed to assess bowel and urinary symptoms in women. The findings support the clinical validity, reliability and responsiveness of the questionnaire by demonstrating good psychometric properties. The questionnaire was shown to be readily understood and unambiguous to the target population and responsive to the realistic constraints of a clinical practice in which simplicity and brevity are highly valued. The low levels of missing data indicate the absence of inappropriate questions. The factor analysis confirmed two of the hypothesised domains and presented a plausible alternative to the hypothesised constipation domain. The observed domain characteristics were stable across the three groups (pre-hysterectomy patients, symptomatic and asymptomatic controls) supporting the validity of its application in a variety of patient populations and its ability to distinguish between different patient groups. Although, as expected, the various domains correlated with each other, the magnitude of these correlations was sufficiently modest to suggest that conceptually distinct areas were being assessed. Internal consistency was established and test –retest reliability apparent. The questionnaire was also shown to be capable of measuring clinically relevant changes over time. The stability shown at baseline and the ability to differentiate clearly between patient groups suggests that this questionnaire is ideal for measuring D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

change in symptoms following intervention and will be a useful tool for comparing the efficacy and outcome of different treatments. Disordered evacuation in women is usually associated with pelvic floor or rectal dysfunction and is commonly related to pelvic surgery or childbirth28 – 30. The use of an evacuation domain is therefore most relevant in assessing women before and after intervention since any such intervention is likely to have a greater influence on the pelvic floor than colonic motility in general. The questionnaire was initially designed with three postulated domains: constipation/evacuation, anal incontinence and urinary symptoms. It is perhaps not surprising that questions one and two were not found to lie within the first domain since this contained a preponderance of questions relating to evacuation. Stool frequency and form may correlate better with general colonic function rather than pelvic floor function and it was deemed clinically appropriate to keep these two questions in a separate domain. Stool frequency (question 1) varies widely, so much so that it is difficult to determine what is normal. Heaton et al.31 found that a third of women defaecated less than once daily and 1% of women once a week or less. Stool consistency (question 2) may correlate better with colonic transit time32 although others have shown a wide variation in stool form and frequency33. Endocrinological changes or pharmacological manoeuvres pre or post any intervention may influence colonic motility2,34,35 and questions one and two may be used separately to identify those women who may have impaired colonic function. The domain consisting of these two questions has been labelled as a constipation domain. Some questionnaires produce a total score representing a global symptom suffering score. In this case, as well as clinically inappropriate, factor analysis results did not show the necessary uniformly strong loading on the first factor indicative of an underlying general factor. A total score from this questionnaire is not considered an adequate global symptom score. The limitations of both the study design and the analysis performed must be borne in mind. A factor analysis is most powerful when applied to continuous data or, at least, many ordered categories. In this questionnaire, the majority of the response sets compromise only four-level ordered responses. Despite this, the conformity to the hypothesised structure was high with plausible alternatives to the originally defined constipation domain. It was not possible to objectively assess clinical status over time with an existing gold standard measure in the area (clinical or other validated instruments e.g. anorectal physiology, transit times, videoproctography or urodynamic studies). The availability of such calibration data would permit stronger conclusions to be made with regards to criterion validity responsiveness. In place of this, the identification of a key question within the domain it represented inherently linked its responses to the domain score.

422

L. HILLER ET AL.

We have not attempted in this questionnaire to assess the impact of any of the bowel or urinary symptoms on women’s overall quality of life. Such questionnaires are available for lower urinary tract symptoms (Bristol Female Lower Urinary Tract Symptoms Questionnaire16, King’s Health Questionnaire17), female genital organ prolapse36 and anal incontinence13. The recently validated Sheffield Prolapse Symptoms Questionnaire36 does contain items relating to constipation and evacuatory function, anal and urinary incontinence and sexual function. However, this instrument has been specifically designed to assess symptom severity in women with anatomical and functional disorders of the pelvic floor, including urogenital prolapse and may therefore not be applicable to the majority of the female population. The ICS definition of urinary incontinence is ‘‘a condition in which involuntary loss of urine is a social or hygiene problem’’37. Kumar et al.38 have proposed a broader subjective definition for constipation of ‘‘when the patient feels the situation to be unsatisfactory’’. It has been suggested that the ICS definition of urinary incontinence be applied to anal incontinence39 however, the impact of even mild or occasional faecal incontinence is likely to be much greater than a similar amount of urinary leakage. The instrument assessed in this paper was designed primarily to assess symptoms and we accept that the omission of a measure of the degree of bother or quality of life measurement may represent a shortcoming. However, the inclusion of a specific quality of life assessment would considerably lengthen and impact upon the ‘user friendliness’ of a questionnaire whose primary aim is to remain short and simple. Generic quality of life questionnaires such as the SF-3640 and the Sickness Impact Profile41 give a general view of health status which is comparable across different groups of patients. Their lack of sensitivity to symptoms specific to bowel and lower urinary tract makes them inappropriate tools for treatment efficacy studies in this area42. The routine application of separate quality of life questionnaires specific to anal incontinence or urinary tract symptoms is likely to be cumbersome and overcomplicated. Such instruments may be useful in conjunction with the current questionnaire for specific clinical research. However, there are no quality of life instruments that specifically address evacuatory dysfunction.

CONCLUSION Overall the Birmingham Bowel and Urinary Symptoms Questionnaire (BBUSQ-22) is considered to be a valid, reliable and responsive instrument for the assessment of the severity of bowel and urinary symptoms in women. The ability of the questionnaire to differentiate between different groups suggests that it should be ideal for measuring change following an intervention and provide a useful tool for comparing efficacy. The authors recommend the use of

the instrument in health care evaluation research to assess the effects of pelvic surgery on bowel and urinary symptoms. Whether the instrument may also be employed as a screening tool to identify women who require more detailed evaluation of pelvic floor function requires further study.

Acknowledgements The authors gratefully acknowledge the help of Mr J. A. Jordan, Mr R. Callender, and Mr D. Murphy for facilitating the study in each of the three centres and wish to express their gratitude to the following clinicians who allowed their patients to take part in the study; Mr M. Afnan, Mr K. Chan, Mr M. Emens, Mr J. Gupta, Mr J. Pogmore, Mr R. Sawers, Mr M. Shafi, Mr A. Browning, Mr C. Cox. Mr N. Fitzgibbon, Mr G. Lewis, Mr A. Warwick. The authors would also like to thank the Worcester St General Practice, Stourbridge, for recruiting control patients into this study. The authors would also like to thank all the women who participated in the study and Professor Abrams for allowing inclusion of some BFLUTS questions in the urinary domain.

References 1. Thompson WG, Creed F, Drossman DA, Heaton KW, Mazzacca G. Functional bowel disorders and chronic functional abdominal pain. Gastroenterol Int 1992;5:75 – 91. 2. Lennard-Jones JE. Clinical management of constipation. Pharmacology 1993;47:216 – 223. 3. Stoker J, Rociu E, Wiersma TG, Lame´ris JS. Imaging of anorectal disease. Br J Surg 2000;87:10 – 27. 4. Prior A, Wilson K, Whorwell PJ, Faragher EB. Irritable bowel syndrome in the gynaecological clinic. Dig Dis Sci 1989;34:1820 – 1824. 5. Longstreth GF, Preskill DB, Youkeles L. Irritable bowel syndrome in women having diagnostic laparoscopy or hysterectomy. Dig Dis Sci 1990;35:1285 – 1290. 6. Prior A, Stanley KM, Smith ARB, Read NW. Relation between hysterectomy and the irritable bowel: a prospective study. Gut 1992;33:814 – 817. 7. Henry MM. Pathogenesis and management of fecal incontinence in the adult. Gastroenterol Clin North Am 1987;16:35 – 45. 8. Engel AF, Kamm MA, Bartram CI, Nicholls RJ. Relationship of symptoms in faecal incontinence to specific sphincter abnormalities. Int J Colorectal Dis 1995;10:152 – 155. 9. Parks AG. Anorectal incontinence. J R Soc Med 1975;68:21 – 30. 10. Jorge JMN, Wexner SD. Etiology and management of fecal incontinence. Dis Colon Rectum 1993;36:77 – 97. 11. Pescatori M, Anastasio G, Bottini C, Mentasti A. New grading and scoring for anal incontinence. Dis Colon Rectum 1992;35:482 – 487. 12. Keighley MRB, Williams NS. Faecal incontinence. In: Keighley MRB, Williams NS, editors. Surgery of the Colon, Rectum and Anus. London: Harcourt Brace, 1999:592 – 700. 13. Reilly WT, Talley NJ, Pemberton JH, Zinsmeister AR. Validation of a questionnaire to assess fecal incontinence and associated risk factors: fecal incontinence questionnaire. Dis Colon Rectum 2000; 43:146 – 153. 14. Lowry A. Invited editorial. Validation of a questionnaire to assess

D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

A BOWEL AND URINARY TRACT SYMPTOMS QUESTIONNAIRE 423

15.

16.

17.

18.

19.

20.

21. 22. 23. 24. 25.

26. 27. 28. 29.

fecal incontinence and associated risk factors: fecal incontinence questionnaire. Dis Colon Rectum 2000;43:153 – 154. Rockwood TH, Church JM, Fleshman JW, et al. Fecal incontinence quality of life scale: quality of life instrument for patients with fecal incontinence. Dis Colon Rectum 2000;43:9 – 16. 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 Urology 1996;77:805 – 812. Kelleher CJ, Cardozo LD, Khullar V, Salvatore S. A new questionnaire to assess the quality of life of urinary incontinent women. Br J Obstet Gynaecol 1997;104:1374 – 1379. Lubeck DP, Prebil LA, Peeples P, Brown JS. A health related quality of life measure for use in patients with urge-incontinence: a validation study. Qual Life Res 1999;8:337 – 344. Gunthorpe W, Brown W, Redman S. The development and evaluation of an incontinence screening questionnaire for female primary care. Neurourol Urodyn 2000;19:595 – 607. Mann CM, Radley SC, Begum G, et al. Constipation, faecal incontinence and urinary symptoms in women awaiting hysterectomy. J Obstet Gynecol 2000;20:530 – 532. Flesch RF. A new readability yardstick. J Appl Psychol 1948;32: 221 – 233. Cronbach LJ. Coefficient alpha and the internal structure of tests. Psychometrika 1951;22:293 – 296. Nunnally JC. Psychometric Theory. New York: McGraw-Hill, 1978. Fleiss JL. Statistical Methods for Rates and Proportions. New York: John Wiley and Sons, 1981. Bland J, Altman D. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;346: 307 – 310. Kline P. A Handbook of Test Construction. London: Methuen, 1986. Helmstadter GC. Principles of Psychological Measurement. New York: Appleton-Century-Crofts, 1964. Radley S, Radley SC, Mann CH, Keighley MRB. Bowel dysfunction following hysterectomy. Br J Obstet Gynaecol 1999;106:1120 – 1125. Roy AJ, Emmanuel AV, Storrie JB, Bowers J, Kamm MA. Behavioural treatment (biofeedback) for constipation following hysterectomy. Br J Surg 2000;87:100 – 105.

D RCOG 2002 Br J Obstet Gynaecol 109, pp. 413 – 423

30. Scott SM, Knowles CH, Newell M, Garvie N, Williams NS, Lunniss PJ. Scitigraphic assessment of colonic transit in women with slow-transit constipation arising de novo and following pelvic surgery or childbirth. Br J Surg 2001;88:405 – 411. 31. Heaton KW, Radvan J, Cripps H, Mountford RA, Braddon FEM, Hughes AO. Defecation frequency and timing, and stool form in the general population: a prospective study. Gut 1992;33:818 – 824. 32. Degan LP, Phillips SF. How well does stool form reflect colonic transit? Gut 1996;39:109 – 113. 33. Heaton KW, Ghosh S, Braddon FEM. How bad are the symptoms and bowel dysfunction of patients with the irritable bowel syndrome? A prospective, controlled study with emphasis on stool form. Gut 1991;32:73 – 79. 34. Davies GJ, Crowder M, Reid B, Dickerson JWT. Bowel function measurements of individuals with different eating patterns. Gut 1986;27:164 – 169. 35. Kamm MA, Lennard-Jones JE, Farthing MJG, McLean A, Perry L, Chard T. Women with intractable constipation have an abnormality of adrenal steroids. Gut 1989;30:A751. 36. Bradshaw HD, Chapple CR, Farkas AG, Radley S, Radley SC. The trouble with prolapse [abstract]. Int Urogyn J 2000;11(Suppl):25. 37. Abrams P, Blaivas JG, Stanton SL, Anderson JT. The standardisation of terminology of lower urinary tract function. Scand J Urol Nephrol 1994;163:175. 38. Kumar D, Bartolo DC, Devroede G, et al. Symposium on constipation. Int J Colorectal Dis 1992;7:47 – 67. 39. Sultan AH. Obstetrical perineal injury and anal incontinence. Clin Risk 1999;5:193 – 196. 40. Brazier JE, Harper R, Jones NMB, et al. Validating the SF-36 health survey questionnaire: a new outcome measure for primary care. BMJ 1992;305:160 – 164. 41. Bergner M, Bobbitt RA, Carter WB, et al. The Sickness Impact Profile: development and final revision of a health status measure. Med Care 1981;19:787 – 805. 42. Lubeck DP, Prebil LA, Peeples P, Brown JS. A health related quality of life measure for use in patients with urge-incontinence: a validation study. Qual Life Res 1999;8:337 – 344. Accepted 5 February 2002