Beruflich Dokumente
Kultur Dokumente
Results
The Short-Form Leeds Dyspepsia Questionnaire was administered to 388
primary care and 204 secondary care patients. The Pearson coefficient
for test–retest reliability was 0.93. The Short-Form Leeds Dyspepsia
Questionnaire had a sensitivity of 77% and a specificity of 75%. A
highly significant response to change was observed (P < 0.005).
Conclusions
The Short-Form Leeds Dyspepsia Questionnaire is a reliable, valid and
responsive self-completed outcome measure for quantifying the fre-
quency and severity of dyspepsia symptoms, which is shorter and more
convenient than the Leeds Dyspepsia Questionnaire.
The SF-LDQ was piloted initially using six rounds of 386 and 841 of 32 48213) with any condition. As
peer review and feedback. Version 6 of the question- dyspeptic symptoms affect around 28% of the popu-
naire was piloted using a sample of 67 consecutive lation,14, 15 we did not specifically select patients
primary care patients. Several modifications of the for- with dyspepsia. The secondary care study population
mat were made and instructions were included on the consisted of unselected patients aged 18 and over,
questionnaire as a result. The final version of the attending for endoscopy at the Leeds General Infirm-
SF-LDQ is attached (Figure 1). ary. Patients were excluded if they were incapable of
giving informed consent, or if they could not speak
or read English.
Evaluation of the SF-LDQ
Patients were asked to participate in the study on
The characteristics of the questionnaire were exam- arrival at their GP’s surgery or the endoscopy suite.
ined using patient populations in both primary care, A trained researcher obtained informed consent.
with a relatively low prevalence of disease, and sec- Patients self-completed the SF-LDQ and an assessment
ondary care, with a high prevalence of dyspepsia. sheet containing questions about the acceptability and
The primary care population consisted of consecutive interpretability of the questionnaire. These were sealed
patients aged 18–65, presenting to two inner city in an envelope before seeing the GP or gastroenterolo-
general practices in Birmingham (deprivation ranks of gist, who made a blind assessment of the whether the
patient suffered with dyspepsia and recorded this on 231 (39%) were male. Ages for the primary care patients
the preprinted envelope along with demographic infor- ranged from 18 to 65, and for secondary care, 18–81.
mation. Validity was established by comparing the Fifty-five questionnaires (9%) had one or more missing
SF-LDQ score with the presence or absence of dyspep- responses and were excluded from the final analysis, as
sia as diagnosed by the GP or gastroenterologist. a summed score could not be calculated. Score analysis
All primary care participants were asked to complete was therefore performed on 375 patients from primary
a second SF-LDQ after 2 days to assess test–retest reli- care and 162 patients from secondary care.
ability. Within the secondary care population, patients
identified as having oesophagitis or peptic ulcer dis-
Interpretability and acceptability
ease during endoscopy were asked to complete a sec-
ond SF-LDQ 2 months later after receiving treatment Five hundred and eighty-six participants from primary
of proven efficacy (proton pump inhibitor therapy and secondary care completed at least one question
and/or Helicobacter pylori eradication therapy). This concerning the interpretability and acceptability of the
enabled assessment of the SF-LDQ’s responsiveness to SF-LDQ. Overall, 97% found the questionnaire easy to
change in conditions with evidence-based treatments. read, 95% found it easy to interpret and 94% thought
Freepost envelopes were used to enhance response the layout was sufficiently clear.
rates, but non-responders to the second questionnaire Endorsement frequencies (the proportion of respond-
were not contacted. ents who chose each response category to a question)
Data were entered onto Microsoft Excel 2000 before were calculated for each response category to every
conversion into SPSS Inc., Chicago, IL, USA, v10 for question using combined data from primary and sec-
analysis. Logistic regression analysis was performed ondary care. Every response category had an endorse-
using Generalized Linear Interactive Modelling, Royal ment frequency of >5%, except for regurgitation
Statistical Society (GLIM) to determine the most valid severity ‘more than daily’ (Table 1). The distribution of
scoring system for the questionnaire. Statistical signi- responses to each of the questions about symptom fre-
ficance was determined where P < 0.05. Ethical quency and severity were positively skewed, reflecting
approval was obtained for South Birmingham, West the inclusion of the 278 participants from primary care
Birmingham and Leeds, UK. (47% of all participants) who did not have dyspepsia
according to their GPs diagnosis.
Non-response to questions regarding symptom fre-
RESULTS
quency occurred with 1% or less of participants, indica-
ting that this part of the questionnaire was acceptable
Population
and interpretable. However, there was no response to
A total of 592 patients were enrolled into the study, 388 questions regarding symptom severity in around 5% of
patients from primary care and 204 patients from sec- cases, indicating that this part of the questionnaire may
ondary care. Out of these, 361 (61%) were female and have been less acceptable or interpretable.
Table 1. Endorsement frequencies for each response category of the Short-Form Leeds Dyspepsia Questionnaire
Indigestion frequency 35 16 15 16 17 1
Heartburn frequency 39 18 15 15 13 0.8
Regurgitation frequency 43 21 14 14 8 0.5
Nausea frequency 38 23 16 14 9 1
Indigestion severity 45 12 13 16 8 5
Heartburn severity 50 11 15 12 7 5
Regurgitation severity 56 13 12 10 4 5
Nausea severity 46 18 12 11 6 6
Table 2. Attributes of the five systems used to score the Short-Form Leeds Dyspepsia Questionnaire in the primary care
population
Scoring system Area under curve (95% CI) ‘Cut-off’ score Sensitivity Specificity
1.00 34
30
26
0.75
18
Sensitivity
0.50 14
10
6
0.25
2
–2
Primary care Secondary care
0.00
0.00 0.25 0.50 0.75 1.00
1 - Specificity Figure 3. Box plot showing the distribution of Short-
Form Leeds Dyspepsia Questionnaire summed total scores
Figure 2. Receiver Operating Characteristic curve for the for primary (n ¼ 375) and secondary care (n ¼ 162) pop-
summed total score of the Short-Form Leeds Dyspepsia ulations. ‘Boxes’ show 25–75th centiles, with the horizon-
Questionnaire. The lines show the most sensitive and spe- tal line representing the mean. ‘Whiskers’ represent the
cific cut-off point for the scoring system. 5th and 95th centiles.
ence was not statistically significant. The summed between these scores was highly statistically signifi-
total score was chosen as the preferred method of cant using the Mann–Whitney U-test (P < 0.0005),
scoring the questionnaire because this scoring system demonstrating that the SF-LDQ was able to discrimi-
had the greatest specificity, giving closer agreement nate between two populations with different preva-
between the sensitivity and specificity, and it also had lence of dyspepsia.
a greater range of possible scores, allowing greater
precision.17 Figure 2 shows the ROC curve for the
Responsiveness to change
summed total score.
Oesophagitis or peptic ulcers was found in 60 of the
162 patients who completed the first questionnaire
Logistic regression analyses
from the secondary care population following endo-
Logistic regression analyses assessed the strength of scopy. These patients were asked to complete a second
the scoring system’s association with the GPs’ diagno- questionnaire following treatment of proven efficacy,
sis. All associations were strongly statistically signifi- 2 months after the endoscopy. This questionnaire was
cant (P < 0.0005). The summed frequency score was returned by 47 patients (78%). Ten patients were
marginally the best predictor of GPs’ diagnosis. How- excluded due to missing data, leaving 37 patients eli-
ever, the difference between the summed total score gible for this analysis.
and the summed frequency score again was not statis- The number of patients with dyspepsia present and
tically significant. absent according to the ‘cut-off’ identified by the sen-
sitivity analysis for the summed total score (seven of
32) was compared before and after treatment (Table 3).
Discriminant validity
Before treatment 89% of patients had dyspepsia com-
Discriminant validity was assessed by comparing pared with 38% after treatment, which was highly sig-
the primary and secondary care populations using the nificant using the McNemar test (P < 0.005). The
summed total scoring system. Figure 3 shows the standardized response mean for this change was 1.1,
range of scores in the two populations. The difference suggesting a large degree of responsiveness.
The LDQ has previously demonstrated a sensitivity alter the concurrent validity of the SF-LDQ, when
of 80% (95% CI: 65–91%) and a specificity of 79% assessed by the area under the ROC curve for the
(95% CI: 66–89%) in a primary care population.10 summed frequency score. This suggests that using the
These values were marginally higher than the Rome II definition of dyspepsia instead of the 1988
SF-LDQs sensitivity of 77% (95% CI: 68–85%) and Working Party definition has little influence on the
specificity of 73% (95% CI: 68–78%) using the concurrent validity of the SF-LDQ. Both GERD and
summed total score. However, this difference is not dyspepsia have recently been re-defined by the Rome
statistically significant, and even if the SF-LDQ was III panel.36, 37 These changes are designed to aid fur-
slightly less valid than the LDQ, this would be offset ther research in selected subgroups of patients and do
by the increased acceptability, feasibility and reliabil- not alter the nature of symptoms sought as outcome
ity of the shorter self-completed measure.17 The SF- measures for use in uninvestigated patients, where
LDQ had a high level of internal consistency when both reflux symptoms and epigastric pain commonly
tested by Cronbach’s alpha coefficient and the item- coexist.
total correlation method, indicating that all of the The correlation between the test–retest SF-LDQ scores
questions in the questionnaire scale were measuring 2 days apart was 0.93, showing a high degree of reliab-
the same underlying construct, producing a high ility. Whilst only 40% of patients returned the second
level of reliability. The SF-LDQ had a higher score questionnaire, this low response rate was not unex-
for Cronbach’s alpha coefficient (0.90) than the LDQ pected for a primary care sample where most partici-
(0.69),10 suggesting that the shorter form was more pants do not have dyspepsia. The LDQ had a weaker
accurately measuring a single construct. correlation between the two questionnaire scores (0.83)
Assessment of concurrent validity involves compar- when test–retest reliability was assessed,10 but the
ing the questionnaire against a ‘gold standard’. As response rate for the second questionnaire was higher
there is no ‘gold standard’ for diagnosis of dyspep- (96% in a secondary care population). Validation of the
sia6, 22, 23 a GPs’ diagnosis was chosen as a quasi-gold SF-LDQ as a postal questionnaire was not carried out in
standard. Validity has been established compared with this study. However, the reliability, validity and respon-
a clinician’s diagnosis in previous studies,10, 19, 24–27 siveness should not be affected by postal completion,
whilst other studies of dyspepsia outcome measures as it is a self-completed instrument. Interpretability and
have used different ‘gold standard’ comparisons to acceptability of the SF-LDQ were demonstrated, sug-
demonstrate concurrent validity, such as generic qual- gesting that the questionnaire should have a good
ity of life scores,28–33 patient self-assessment using response rate. The response rate was only 40% in the
diaries34 and dyspepsia adverse events.35 An alternat- test–retest sample, but it was 78% in the responsiveness
ive approach would have been to compare the SF-LDQ to change sample (secondary care), where dyspepsia
with the LDQ as the gold standard.17 However, the cor- was more salient to the respondents. The SF-LDQ’s
relation between the two questionnaires would have responsiveness to change was highly statistically signi-
been artificially inflated by the presence of four iden- ficant in 37 patients receiving a treatment of known
tical questions. No attempt was made to standardize effectiveness. The standardized response mean values
GPs’ diagnosis through discussion of the 1988 Work- suggested this response to change was large. The LDQ
ing Party definition of dyspepsia with them.18 Stan- was assessed in a similar way and was found to be
dardizing the GPs’ diagnosis in this way would have equally responsive to change, although the standard-
artificially increased the concurrent validity of the ized response mean was not calculated.10 Other studies
SF-LDQ, because the questionnaire is based upon the have used two groups receiving treatment and placebo,
Working Party definitions. in order to compare the responsiveness of the two
It should be emphasized that the SF-LDQ is designed groups.38, 39 However, it is not ethical to use placebos
as an outcome assessment tool and not as a diagnostic except in the context of a randomized-controlled trial,
tool. Although considerable effort has been made by so this was not possible in this study. There was no
the Rome process to disentangle reflux and epigastric alternative method of confirming that a response to
pain, this has not been successful where patients have change had occurred in this study, such as a blinded
not had endoscopic investigation to exclude peptic clinician assessment after treatment or a question on
ulcer and oesophagitis. Exclusion of patients with pre- self-reported global improvement, as the treatments
dominant reflux-like symptoms did not substantially used have proven efficacy.40–42
Three percent of respondents commented that the University of Birmingham. They assisted with patient
text size was too small and a larger version of the recruitment and data entry.
questionnaire should be made available for such
patients in clinical trials. Non-English speaking
AUTHOR CONTRIBUTIONS
patients were excluded from this study, but should be
included in clinical dyspepsia trials to increase gener- Adam Fraser: Lead contributor to study design and
alizability. Translation of the SF-LDQ into other lan- planning, conducting the study, data analysis and wri-
guages would alter its characteristics, necessitating ting the manuscript; Brendan C. Delaney: guarantor of
further validation.17, 31 the submission and corresponding author. Contributed
The SF-LDQ is a self-completed outcome measure to study design and planning, and writing the manu-
that assesses both the frequency and severity of dys- script; Alexander C. Ford: responsible for conducting
pepsia symptoms for which acceptability, interpretabil- the secondary care arm of the study at Leeds General
ity, reliability, validity and responsiveness to change Infirmary; Michelle Qume: contributed to conducting
have been demonstrated. It is a precise measure using the study, data entry and analysis and writing the
the summed total score of frequency and severity manuscript; Paul Moayyedi: contributed significantly
responses and has good feasibility due to its brevity. to study design and planning, especially questionnaire
The SF-LDQ meets all the criteria for an outcome development. All authors have checked and approved
measure for dyspepsia in cost-effectiveness trials, and the final draft submitted.
is particularly well suited to primary care trials invol-
ving uninvestigated patients.
STATEMENT OF INTERESTS
Authors’ declaration of personal interests: All authors
ACKNOWLEDGEMENTS
were employed by the University of Birmingham at
This work was undertaken by Dr Adam Fraser as part the time of this study.
of a Masters in Primary Care degree at the University Declaration of funding interests: The study was fun-
of Birmingham. John C. Duffy provided statistical ded in full by the Medical Research Council as part of
advice at the Department of Primary Care and General the CUBE trial (ISRCTN 87644265). The general prac-
Practice, University of Birmingham. He carried out the tices involved were reimbursed by the Midlands
sensitivity analyses and logistic regression analyses. Research Practices Consortium (MidReC) for each
Val Redman and Beth Hinks are Research Associates at patient enrolled. Gastroenterologists recruited patients
the Department of Primary Care and General Practice, in secondary care without remuneration.
Benefits and Health Economics. Leeds: Functional gastroduodenal disorders. 34 Leidy NK, Farup C, Rentz AM,
Institute of Epidemiology and Health Gut 1999; 45 (Suppl. 2): II37–42. Ganoczy D, Koch KL. Patient-based
Services Research, University of Leeds, 24 Greatorex R, Thorpe JA. Clinical assess- assessment in dyspepsia: development
1999. ment of gastro-oesophageal reflux by and validation of Dyspepsia Symptom
13 Office of the Deputy Prime Minister. The questionnaire. Br J Clin Pract 1983; 37: Severity Index (DSSI). Dig Dis Sci
English Indices of Deprivation. London: 133–5. 2000; 45: 1172–9.
Neighbourhood Renewal Unit, 2004. 25 Junghard O, Lauritsen K, Talley NJ, 35 Rabeneck L, Wristers K, Goldstein JL,
14 Heading RC. Prevalence of upper gastro- Wiklund IK. Validation of seven graded Eisen G, Dedhiya SD, Burke TA. Reliab-
intestinal symptoms in the general pop- diary cards for severity of dyspeptic ility, validity, and responsiveness of
ulation: a systematic review. Scand symptoms in patients with non-ulcer severity of dyspepsia assessment (SODA)
J Gastroenterol Suppl 1999; 231: 3–8. dyspepsia. Eur J Surg Suppl 1998; 583: in a randomized clinical trial of a COX-
15 Stanghellini V. Three-month prevalence 106–11. 2-specific inhibitor and traditional
rates of gastrointestinal symptoms and 26 Locke GR, Talley NJ, Weaver AL, NSAID therapy. Am J Gastroenterol
the influence of demographic factors: Zinsmeister AR. A new questionnaire 2002; 97: 32–9.
results from the Domestic/International for gastroesophageal reflux disease. 36 Drossman DA. The functional gastroin-
Gastroenterology Surveillance Study Mayo Clin Proc 1994; 69: 539–47. testinal disorders and the Rome III Pro-
(DIGEST). Scand J Gastroenterol Suppl 27 Talley NJ, Phillips SF, Wiltgen CM, cess. Gastroenterology 2006; 130:
1999; 231: 20–8. Zinsmeister AR, Melton LJ III. Assess- 1377–90.
16 Talley Colin-Jones NJD, Koch KL, ment of functional gastrointestinal dis- 37 Vakil N, VanZanten S, John D, Kahrilas
Koch M, Nyren O, Stanghellini V. Func- ease: the bowel disease questionnaire. P, Roger J. The Definition of GERD: a
tional dyspepsia: a classification with Mayo Clin Proc 1990; 65: 1456–79. Global, Evidence-based Consensus.
guidelines for diagnosis and manage- 28 Hu WH, Lam KF, Wong YH, et al. The Los Angeles, USA: Digestive Disease
ment. Gastroenterol Int 1991; 4: 145– Hong Kong index of dyspepsia: a valid- Week, 2006.
60. ated symptom severity questionnaire for 38 Rabeneck L, Cook KF, Wristers K,
17 Fitzpatrick R, Davey C, Buxton MJ, patients with dyspepsia. J Gastroenterol Souchek J, Menke T, Wray NP. SODA
Jones DR. Evaluating patient-based out- Hepatol 2002; 17: 545–51. (severity of dyspepsia assessment): a
come measures for use in clinical trials. 29 Dimenas E, Glise H, Hallerback B, new effective outcome measure for dys-
Health Technol Assess 1998; 2: 1–74. Hernqvist H, Svedlund J, Wiklund I. pepsia-related health. J Clin Epidemiol
18 Management of dyspepsia: report of a Quality of life in patients with upper 2001; 54: 755–65.
working party. Lancet 1988; 1: 576–9. gastrointestinal symptoms. An improved 39 Rothman M, Farup C, Stewart W,
19 Buckley MJ, Scanlon C, McGurgan P, evaluation of treatment regimens? Helbers L, Zeldis J. Symptoms associ-
O’Morain CA. A validated dyspepsia Scand J Gastroenterol 1993; 28: 681–7. ated with gastroesophageal reflux dis-
symptom score. Ital J Gastroenterol 30 Shaw M, Talley NJ, Adlis S, Beebe T, ease: development of a questionnaire
Hepatol 1997; 29: 495–500. Tomshine P, Healey M. Development of for use in clinical trials. Dig Dis Sci
20 Poitras MR, Verrier P, So C, Paquet S, a digestive health status instrument: 2001; 46: 1540–9.
Bouin M, Poitras P. Group counseling tests of scaling assumptions, structure 40 Laine L, Hopkins RJ, Girardi LS. Has the
psychotherapy for patients with func- and reliability in a primary care popula- impact of Helicobacter pylori therapy
tional gastrointestinal disorders: devel- tion. Aliment Pharmacol Ther 1998; 12: on ulcer recurrence in the United States
opment of new measures for symptom 1067–78. been overstated? A meta-analysis of
severity and quality of life. Dig Dis Sci 31 Goldman J, Conrad DF, Ley C, et al. rigorously designed trials. Am J Gast-
2002; 47: 1297–307. Validation of Spanish language dyspep- roenterol 1998; 93: 1409–15.
21 Calvet X, Bustamante E, Montserrat A, sia questionnaire. Dig Dis Sci 2002; 47: 41 Chiba N, De Gara CJ, Wilkinson JM,
et al. Validation of phone interview for 624–40. Hunt RH. Speed of healing and
follow-up in clinical trials on dyspepsia: 32 Garratt AM, Ruta DA, Russell I, et al. symptom relief in grade II to IV gastr-
evaluation of the Glasgow Dyspepsia Developing a condition-specific measure oesophageal reflux disease: a meta-ana-
Severity Score and a Likert-scale symp- of health for patients with dyspepsia lysis. Gastroenterology 1997; 112:
toms test. Eur J Gastroenterol Hepatol and ulcer-related symptoms. J Clin Epi- 1798–810.
2000; 12: 949–53. demiol 1996; 49: 565–71. 42 Penston JG. Review article: Clinical
22 Dent J. Definitions of reflux disease and 33 Eypasch E, Williams JI, Wood-Dauphinee aspects of Helicobacter pylori eradica-
its separation from dyspepsia. Gut 2002; S, et al. Gastrointestinal Quality of Life tion therapy in peptic ulcer disease. Ali-
50: 17iv–20. Index: development, validation and ment Pharmacol Ther 1996; 10: 469–86.
23 Talley NJ, Stanghellini V, Heading RC, application of a new instrument. Br J
Koch KL, Malagelada JR, Tytgat GN. Surg 1995; 82: 216–22.