Measuring outcomes of importance to women with stress urinary incontinence.
ABSTRACT Peer reviewed Postprint
- SourceAvailable from: europepmc.org[Show abstract] [Hide abstract]
ABSTRACT: Urinary incontinence can cause embarrassment and can impact on daily activities and quality of life. Generic health related quality of life instruments, such as the EQ-5D, are designed to be applicable across a variety of disease areas. However, it is sometimes claimed that they are not applicable to a certain disease area because they are missing a domain which directly captures the impact of that particular disease. For example, none of the domains of the EQ-5D relate directly to incontinence, although the impact of incontinence on quality of life may be expected to be picked up indirectly through changes in domains such as usual activities or anxiety/depression. The objective of this review was to examine the appropriateness of the EQ-5D in people with urinary incontinence by reviewing published evidence relating to the psychometric performance of the EQ-5D. A systematic search was conducted to identify studies reporting data that permitted assessment of the construct validity, responsiveness or reliability of the EQ-5D in people with urinary incontinence. Included papers were those that reported EQ-5D alongside other measures of health related quality of life or clinical measures in patients with urinary incontinence or in a broader population where results were reported for a subgroup of patients with urinary incontinence. Data were extracted and a narrative synthesis was undertaken. Seventeen papers were included in the review. In most of the tests performed, EQ-5D was consistent with clinical or disease specific outcome measures. The EQ-5D demonstrated validity in the majority of 'known group' comparisons, although statistical significance was not always reported. Correlations between the EQ-5D and disease specific outcomes were statistically significant and in the expected direction for most but not all of the disease specific instruments and clinical measures. For responsiveness, there was general agreement between changes in EQ-5D and changes in clinical or disease specific measures. Evidence on reliability was limited to one study. The EQ-5D was generally found to perform well on tests of construct validity, responsiveness and reliability, in people with urinary incontinence although no definitive conclusion can be made on its appropriateness based on these measures alone.Health and Quality of Life Outcomes 02/2013; 11(1):20. · 2.27 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Objective. In rehabilitation, treatment is individually tailored to each patient's goals. Individualized instruments allow patients to choose domains that they consider important, which may make them particularly appropriate as evaluative tools in this setting. We aimed to evaluate the psychometric properties of the Norwegian version of the Patient Generated Index (PGI) in patients with rheumatic diseases participating in inpatient rehabilitation or self-management programmes.Methods. Patients completed the PGI together with other outcome measures at arrival and 5 and 52 weeks after arrival. The PGI was assessed for data quality by completion rates, reliability by the intraclass correlation coefficient (ICC), agreement by standard error of measurement (SEM) and smallest detectable change (SDC). Construct validity was assessed by testing a priori hypotheses regarding correlation between PGI scores and other outcome measures. Responsiveness was assessed by an a priori hypothesis regarding the correlation of different change scores and standardized response means (SRMs).Results. A total of 145 patients participated and 118 (81%) completed the PGI correctly. The ICC was 0.87, SEM 7.25 and SDC 20.10. Ninety-three per cent of the hypotheses of correlation were confirmed in tests for construct validity. Responsiveness was confirmed in 53% and 71% of hypotheses tested at 5 and 52 weeks. SRMs were 0.2 and 0.4, respectively.Conclusion. The results support the validity, reliability and responsiveness of the Norwegian version of the PGI in patients with rheumatic diseases and its application as an outcome measure in rehabilitation or self-management programmes. Further research is needed to improve completion rates for the PGI.Rheumatology (Oxford, England) 01/2013; · 4.24 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Background: Analysis of EQ-5D data often focuses on changes in utility, ignoring valuable information from other parts of the instrument. Our objective was to explore how the utility index, EQ-5D profile, and EQ-VAS captured change in clinical trials of mirabegron, a new treatment for overactive bladder (OAB). Data: Data were pooled from three phase III clinical trials that investigated the efficacy and safety of mirabegron versus placebo. Tolterodine ER 4mg was included as an active control in one study: (1) placebo, mirabegron 50mg and 100mg, and tolterodine 4mg ER; (2) placebo, mirabegron 50mg and 100mg); (3) placebo, and mirabegron 25mg and 50mg. Data were collected at baseline, week 4, 8 and 12. Methods: Analyses were performed on full analysis and modified intention to treat (ITT) data sets using UK utilities. Analysis controlled for relevant patient characteristics. Analysis of Covariance identified changes from baseline at each time point in utilities and EQ-VAS. Areas Under the Curve were estimated to summarise intertemporal differences in effect. EQ-5D profile data were analysed using the Paretian Classification of Health Change. Results: In modified ITT analyses, mirabegron 50mg was superior to tolterodine 4mg in changes from baseline utilities after 12 weeks (p<0.05); similarly, AUC results showed mirabegron 50mg to be superior to tolterodine (p<0.05) and placebo (p<0.05) with the benefit already apparent at 4 weeks (p<0.05). EQ-VAS more consistently indicated superior outcomes: all three mirabegron doses showed statistically significant greater effectiveness compared to tolterodine at 12 weeks. Individual EQ-5D dimensions and the overall profile showed no significant differences between study arms. Conclusion: Mirabegron showed quicker and superior improvement in HR-QoL compared to tolterodine 4mg ER. A limitation of the study is that EQ-5D was a secondary outcome in the pivotal trials, which were not powered to measure differences on EQ-5D.Journal of Medical Economics 05/2013;
This is the author's version of the work. It is posted here for personal use, not for
redistribution. The definitive version was published in British Journal of
Obsterics and Gynaecology 2009;116(5):719-25, http://dx.doi.org/10.1111/j.1471-
Measuring outcomes of importance to women with stress urinary incontinence
Laura Ternent1,2 Luke Vale1,2, Brian Buckley3, Charis Glazener2
1Health Economics Research Unit, University of Aberdeen, UK
2Health Services Research Unit, University of Aberdeen, UK
3Primary Care, Department of General Practice, National University of Ireland,
Address for correspondence:
Health Services Research Unit, Institute of Applied Health Sciences, University of
Aberdeen, AB25 2ZD.
Tel: 01224 554687
Fax: 01224 550926
Objective: To provide evidence on outcomes of importance to women who have
stress urinary incontinence (SUI). The secondary aim was to identify additional
outcomes that ought to be collected in future primary studies or in systematic reviews
of the literature.
Design: Questionnaire survey of a cohort of women with SUI.
Sample: 188 women with SUI.
Methods: Areas of importance to women who suffer from SUI were assessed using a
patient generated index (PGI). In addition to the PGI, the questionnaire included the
King’s Health Questionnaire (KHQ) and the EuroQol- 5D (EQ-5D).
Main Outcome Measures: PGI, EQ-5D, and the King’s Health Questionnaire.
Results: In total, 38 different areas were reported by respondents on the PGI. PGI
and EQ-5D scores were positively correlated and significant. Correlations between
the seven domains of the KHQ and PGI were all negative but only two were
statistically significant: personal relationships and severity measures.
Conclusion: The PGI succeeded in capturing a diverse range of outcomes of
importance to women suffering with SUI. Given the limited correlation between the
KHQ domains and the PGI and, in addition, that the areas mentioned in the PGI were
not found to map well to the EQ-5D, the PGI in this instance may be capturing
concerns of women who suffer from SUI which are not captured by quality of life
measures such as the EQ-5D.
Stress urinary incontinence (SUI) is the most common type of incontinence,
especially in middle aged-women. It is defined as involuntary leak or loss of urine
caused by sneezing, exercising, lifting or physical activity (Abrams et al 2002).
Studies investigating the prevalence of SUI in women are hampered by differing
definitions of SUI, and social factors such as the limited recourse to health care due to
embarrassment. Indeed, only 15% of women identified as suffering from SUI consult
a health professional (Shaw et al, 2006; 2001), suggesting that those presenting for
help represent the ‘tip of the iceberg’ of a common condition. Embarrassment
associated with urinary incontinence may cause withdrawal from social situations and
reduces quality of life (Hunskaar and Vinsnes 1991). Many women with SUI show
symptoms of depression and introverted behaviour, together with dysfunctional
interpersonal relationships (Norton et al, 1988). Furthermore, SUI may lead to
withdrawal from regular physical activities and thus impair women’s general health
(Nygaard et al 1990).
The aim of this survey was to provide evidence on outcomes of importance to women
who have SUI. Much of the available literature focuses on clinical outcomes which as
a result may have limited relevance to women with SUI. The purpose of this work
was to prospectively survey women with SUI to provide information on outcomes of
importance to them; a secondary aim was to identify additional outcomes that ought to
be collected in future primary studies or define relevant outcomes for systematic
reviews of the literature.
In order to assess the areas of importance to women who suffer from SUI, a
questionnaire was designed. A patient generated index (PGI) (Ruta et al, 1994) was
employed to allow respondents to state and evaluate the areas of their life affected by
SUI. In addition to the PGI, the questionnaire included the Kings Health
Questionnaire (KHQ) (Kelleher et al, 1997), the EQ-5D and questions relating to
socioeconomic and demographic information.
The PGI is an individualised patient-reported health instrument which allows the
respondent to select, weight, and rate the importance of a particular health outcome
(Martin et al, 2006). The PGI was designed with the aim of producing a valid
measure of outcome that reflected areas of importance to patients’ lives (Ruta et al
1994). The PGI involves the respondent deciding what factors are important to them.
Examples of the types of factors that may be important are included to provide
guidance. The aim of the PGI is therefore to capture the diverse range of concerns or
priorities of respondents. Using the PGI, respondents can vary the weight they attach
to these concerns or priorities, which provide researchers with an insight into the
respondent’s viewpoint. An overall score for the PGI for each respondent can then be
calculated by multiplying the rating for each health area by the proportion of points
allocated to that particular area.
The PGI is completed in three stages: in the first stage respondents are asked to
identify up to five areas of their life that are affected by their SUI. Respondents are
given a list of outcomes to act as prompts to help them think about which areas of
their life are affected by their condition. Respondents can then choose from these
options or provide their own examples. In addition to the five boxes, there is a sixth
box which enables respondents to rate all other areas of their life affected by their
SUI. Possible examples of the factors to include on the PGI were drawn from three
sources. The first of these was the King’s Health Questionnaire, which was used to
generate a list of outcomes under the broad headings of: Role limitations; Physical
limitations; Social limitations; Personal relationships and Emotions. These outcomes
were supplemented from Cochrane reviews of non-surgical treatments (Hay-Smith
and Dumoulin 2006; Brazzelli et al 2007; Mariappan et al 2007; Bezerra et al 2007).
Finally, a general literature search was also conducted although this did not provide
further additions to the 17 different outcomes identified from the KHQ and the
Cochrane reviews. These outcomes were then narrowed down to those considered
most relevant by members of the project team.
In stage two of the PGI, respondents were asked to score each area listed in stage one
of the PGI on a scale ranging from 0 to 6. The score given in stage two was intended
to reflect how the individual was affected by their SUI in the past month. A score of 0
would signify that the effect on their life was as bad as it could possibly be and a
score of 6 would correspond to an effect that was as good as it could possibly be.
Finally, in stage 3, respondents were asked to ‘spend’ 10 points to indicate the relative
importance of each of the areas mentioned in stage one. Respondents were requested
to spend more points on areas that were the most important to them.
As noted above, in addition to the PGI, the questionnaire also contained the Kings
Health Questionnaire (KHQ) and the EQ-5D. The KHQ is a condition specific
questionnaire which aims to assess the impact of urinary incontinence on an
individual’s quality of life. It contains questions set in nine domains relating to:
general health perception, incontinence impact, role limitations, physical limitations,
social limitations, personal relationships, emotions, sleep and energy, and severity.
With the exception of the final part of the questionnaire (severity measures) scores
can be calculated for each domain (0 – 100). The higher the score the worse off an
individual feels they are and the lower they perceive their quality of life to be.
The EQ-5D is a standardised instrument for use as a measure of health outcome.
Applicable to a wide range of health conditions and treatments, it provides a simple
descriptive profile and a single index value for health status
(http://www.euroqol.org/.) The EQ-5D has five dimensions (mobility, self-care, usual
activities, pain/discomfort and anxiety/depression) that can be converted into a utility
The Bladder & Bowel Foundation (formerly InContact and the Continence
Foundation) is a national charity that provides information and support to people with
bladder and bowel problems, representing the interests of people with continence
problems with the aim of ensuring they have access to the latest information and
services available (http://www.bladderandbowelfoundation.org/). In 2006 a survey
conducted by InContact was completed by 755 people affected by bladder and bowel
problems (Buckley, Wagg, Winder 2007). Of these, 188 women with SUI gave
consent for future contact about relevant research and formed the sample for the
current study. In July 2007 these women were sent questionnaires for the current
study by InContact. Given that this is a self slecctedselected sample of women
suffering from SUI and not a random sample of the population, it is not known how
representative this sample is of the wider population.
The 2006 survey in which the participants were originally identified was a service
evaluation in which InContact surveyed people who had previously been in touch
with the charity and as such no ethical approval was necessary. The 2006 survey
materials contained an explicit assurance that confidentiality would be maintained and
that identifiable data would not be passed on to third parties. Respondents were asked
if they were willing to be contacted in the future for research purposes. For this study
questionnaires were sent in July 2007 to 188 women with SUI who gave their consent
for further contact relating to research. The questionnaires were returned directly to
the charity and after screening only anonymous data were subsequently forwarded to
the authors in accordance with the Medical Research Council’s guidance on the use of
personal information in medical research and the Data Protection Act 1998 (HMSO,
1998; Medical Research Council, 2000).
In total, 105 out of 188 respondents (55.9%) completed and returned the
questionnaire. Table 1 shows the areas of an individual’s life that they reported to be
affected by their SUI and the frequency that the area was mentioned within the sample
as a whole. Thirty-eight different areas were reported by respondents.
Out of 105 respondents, 73 answered the PGI correctly, 9 respondents made mistakes
in the PGI and 23 respondents did not fully complete it (Table 2). Of the 73
respondents who correctly completed the questionnaire, 61 answered the PGI with no
mistakes (all sections were completed satisfactorily), the remaining 12 respondents
made a small error in completion of the PGI. This small error always occurred in
section three of the PGI, where respondents had to spent 10 points, these respondents
did in fact spend ten points, however they missed out spending points in area 6 (all
other areas of their life affected by SUI) and totalled to ten in box 6. An example of
the PGI used can be seen in aAppendix 1.
Table 3 shows the demographic information of the sample as a whole and for those
individuals who correctly completed and incorrectly completed the PGI. The mean
age for the sample as a whole was 57 (range 28 to 89). As can be seen in Table 3,
those respondents who correctly completed the PGI were on average younger than
those who incorrectly completed the PGI. In addition, those who correctly completed
the PGI appear to be better educated and in higher income groups.
In addition to listing the outcomes of importance to women who suffer from SUI, the
PGI can also be calculated into a score of overall quality of life. The score ranges
from 0 to 6, with 0 reflecting a very low quality of life (“it’s as bad as it could
possibly be”) and 6 reflecting a very high quality of life (“it’s as high as it could
possibly be”) An example of the PGI and the method used to calculate the score is
given in Table 4. For the respondents who successfully completed the PGI the mean
score was 2.4 (SD 1.4, range 0-6). 101 of 105 returned questionnaires had a fully
completed EQ-5D. Scores on the EQ-5D ranged from -0.17 to 1. The mean EQ-5D
score was 0.598 (SD 0.339). Correlation between the mean PGI score and the mean
EQ-5D was, as expected, positive and significant.
Scores (out of 100) for each domain in the KHQ can be seen in Table 5. The higher
the score, the worse off an individual feels. In addition to the domains of the KHQ, it
also contains a section detailing the respondent’s bladder problems and how much
they affect the individual’s life.
Correlations of the PGI and seven domains of the KHQ were also performed. We
would expect to find a negative correlation between the PGI and KHQ. All
correlations were negative, however only two were significant: personal relationships
(P=0.004) and severity measures (P=0.003).
In addition, correlations of the EQ-5D score and the domains of the KHQ were also
performed. We found all seven of the KHQ domains to be significantly (negatively)
correlated with the EQ-5D. This result is to be expected since many of the EQ-5D
and KHQ domains are similar.
The PGI has been used to quantify the effect of SUI on the quality of patients’ lives
for the first time. The respondents are a self-selected sample of women who had
previously been in touch with a patient support charity and who may be considered to
be active help-seekers. However, there is no reason to suspect that their experience of
SUI and the relative perceived impact of SUI on various aspects of their lives are
different from the wider population of women affected. Nearly 70% of respondents
successfully completed the questionnaire, a further 9% attempted the PGI but made
mistakes in its completion, and 22% failed to fully complete the PGI; the majority of
these respondents completed stage 1 of the PGI but failed to complete stage 2 or 3.
Those respondents who successfully completed the questionnaire were found to be
younger, in higher income groups and have a higher level of education. For the PGI to
be used as a valid and reliable measure of outcomes of importance to women with
SUI and to be able to accurately quantify the effect of SUI on their lives the response
rate and successful completion of the PGI would need to be improved. Of the
respondents, 31% of respondents had difficulty in completing the questionnaire and
there was also a low response rate to the survey in general (55.9%).
In order to improve this response rate and successful completion alterations could be
made to the layout of the PGI to make it more user-friendly. An alternative would be
to use the questionnaire as the basis for setting a patient’s goals for treatment in a
clinical or research setting: achieving those goals could then be used as a measure of
In stage 1 of the PGI, 38 different areas of an individual’s life affected by SUI were
reported. The most frequently mentioned areas were going out or socialising, with
14% of all respondents listing this as one of the areas of their life affected by their
condition. Thirteen percent of respondents mentioned personal hygiene and 11%
mentioned the effect their condition has on their sleep. Shopping (8%), depression
(8%), physical activity (6%), work (6%), anxiety (6%), travel (4%), household tasks
(2%), personal (1%) and sexual relationships (2%) were all also listed as areas of their
life affected by SUI.
The PGI, or instruments like it, have been criticised in the past, with some authors
questioning whether they reflect the patient’s view point or, conversely, whether they
are simply reflecting the views of the researchers who designed the questionnaire
(Tully and Cantrill, 2000). In this survey, whilst we did find a varied response in the
number of outcomes listed by respondents, the majority of these did in fact come from
the prompt list provided in the PGI. Of the 10 most mentioned areas, eight of these
were from the prompt list. Other studies have found similar associations between the
prompt list and final outcomes listed by respondents. However, it is unclear whether
this association is due to the most relevant examples being selected from the prompt
list, or due to respondents being unwilling or unable to think of their own examples
because the prompt list is already comprehensive (Russell and Macduff, 1998).
We correlated the mean PGI score and the mean EQ-5D scores. This correlation was,
as expected, found to be positive and significant. In addition to this, correlations of
the KHQ and the PGI were performed. Although all correlations of the PGI and KHQ
were of the expected sign (negative), only two were found to be statistically
significant. Given that the PGI outcomes and the domains of the KHQ do not
correlate well, and that many of the aspects respondents mentioned in the PGI list of
outcomes do not map very well on to the dimensions of the EQ-5D this might suggest
that generic measures, like the EQ-5D, may not be a very good reflection of the
preferences of people with incontinence. The PGI in this instance may therefore be
capturing concerns of women who suffer from SUI which are not adequately captured
by generic instruments such as the EQ-5D. This is of particular interest in the context
of health technology appraisals, where EQ-5D has become the accepted standard for
calculation of Quality Adjusted Life Year indices for use in determination of cost
effectiveness (NICE 2008).
Much of the available literature on SUI focuses on doctor-selected clinical outcomes.
Given the undoubted social and personal impact of SUI, these outcomes may as a
result have limited relevance to the women who suffer this condition. The aim of this
survey was therefore to provide evidence on outcomes of importance to women who
have SUI. This is the first attempt to use the PGI in this area. In total, 48 different
areas of an individual’s life affected by SUI were mentioned, the most common of
these being: going out or socialising; personal hygiene; sleep disruption; the effect on
daily activities such as shopping; depression; physical activity; work; anxiety; travel;
household tasks; and personal and sexual relationships. The PGI succeeded in
capturing a diverse range of outcomes of importance to women suffering from SUI.
However, it must be noted that some respondents did have difficulty in completing the
Finally, the PGI was not found to correlate well with the domains of the KHQ, in
addition, many of the aspects respondents mentioned in the PGI list of outcomes were
not found to map well on to the dimensions of the EQ-5D suggesting that generic
measures, like the EQ-5D, may not be a very good reflection of the preferences of
people with incontinence. We conclude that the PGI in this instance may therefore be
capturing concerns of women who suffer from SUI which are not adequately captured
by generic instruments such as the EQ-5D.
Acknowledgements: We would like to thank the women who responded to our
questionnaire. We would also like to thank Mari Imamura for comments on previous
drafts of the paper. The Health Services Research Unit and Health Economics
Research Unit are supported by the Chief Scientist Office of the Scottish Government
Funding declaration: The project was funded by a grant from the NIHR Health
Technology Assessment Programme. The Health Economics Research Unit and the
Health Services Research Unit are funded by the Chief Scientists Office of the
Scottish Government's Health Directorates.
Ethics approval: No ethical approval was necessary. The sample was surveyed by a
charity to whom they had already given consent to be contacted for research purposes.
Questionnaires were returned to the charity. Only anonymous data were supplied to
the authors by the charity. Ethical approval for the study was provided by InContact.
The 2006 survey in which the participants were originally identified was a service
evaluation in which InContact surveyed people who had previously been in touch
with the charity and as such no ethical approval was necessary.
Disclosure of Interests: Brian Buckley is a Cochrane Fellow at the Department of
General Practice, National University of Ireland, Galway, and is the Chair of the
Board of Trustees of the Bladder & Bowel Foundation.
Contribution to Authorship: All authors contributed to the design of the study,
interpretation of data and writing of the manuscript.
Abrams P, Cardozo L, Fall M. et al. The standardisation of terminology of lower
urinary tract function: report from the Standardisation Sub-committee of the
International Continence Society. Am J Obstet Gynecol; 2002; 187:116
Shaw C, Das Gupta R, Williams K, et al. A survey of health seeking and treatment
provision in women with stress urinary incontinence. Br J Urol Int; 2006; 97:752
Shaw C, Tansey R, Jackson C, et al. Barriers to help-seeking in people with urinary
symptoms. Fam Pract, 2001. 18:48.
Hunskaar S, Vinsnes A. The quality of life in women with urinary incontinence as
measured by the sickness impact profile. J Am Geriatr Soc; 1991; 39: 378
Norton P, MacDonald L, Sedgwick P. et al. Distress and delay associated with urinary
incontinence, frequency, and urgency in women. BMJ; 1998; 297:1187
Nygaard I, DeLancey, J O, Arnsdorf L. et al. Exercise and incontinence. Obstet
Gynecol; 1990; 75: 848
Ruta DA, Garratt AM, Leng M, et al. A new approach to the measurement of quality
of life: the patient generated index (PGI). Med Care 1994; 32: 1109-26
Kelleher CJ, Cardozo LD, Khullar V, Salvatore S. A new questionnaire to assess the
quality of life of urinary incontinent women. British Journal of Obstetrics and
Gynaecology; 1997; 104: 1374-1379
Martin F, Camfield L, Rodham K, Klieempt P, Ruta D. Twelve years’ experience
with the Patient Generated Index (PGI) of quality of life: a graded structured review.
Qual Life Res; 2006;
Hay-Smith E, Dumoulin C. Pelvic floor muscle training versus no treatment, or
inactive control treatments, for urinary incontinence in women. Cochrane Database
Syst Rev: CD005654, 2006
Brazzelli M, Shirran E, Vale L. Absorbent products for containing urinary and/or
faecal incontinence in adults. Cochrane Database of Syst Rev 2007, CD001406.
Mariappan P, Ballantyne Z, N'Dow J. et al. Serotonin and noradrenaline reuptake
inhibitors (SNRI) for stress urinary incontinence in adults. Cochrane Database Syst
Rev: CD004742, 2007.
Bezerra C, Bruschini H, Cody D. Traditional suburethral sling operations for urinary
incontinence in women. Cochrane Database of Syst Rev 2007, Art. No.: CD001754
Buckley B, Wagg A, Winder A. Emotional wellbeing in faecal and urinary
incontinence. Continence; 2007; 1 (1): 66-9.
Data Protection Act 1998. London: HMSO; 1998.
Tully MP, Cantrill JA. Et al. The validity of the modified patient generated index – a
quantitative and qualitative approach. Quality of Life Research; 2000; 9: 509 - 520
Russell E, Macduff C. The problem of measuring change in individual health-related
quality of life by postal questionnaire: the use of the patient-generated index in a
disabled population. 1998; 7: 761 - 769
Guide to the Methods of Technology Appraisal National Institute for Health and
Clinical Excellence. London, 2008.
Personal Information in Medical Research. London: Medical Research Council; 2000.