R. P. Knill-Jones, F.R.C.P., F.F.P.H.M., Dept. of Public Health, University of. Glasgow, Glasgow ..... Lucas R, Card WI, Knill-Jones RP, Watkinson C. Crean GP.
Geographical Differences in the Prevalence of Dyspepsia R. P. KNILL-JONES Dept. of Public Health, University of Glasgow, and the Diagnostic Methodology Research Unit, Southern General Hospital, Glasgow, Scotland
Knill-Jones RP. Geographical differences in the prevalence of dyspepsia. Scand J Gastroenterol, 1991, 26(suppl 182), 17-24 The epidemiology of dyspepsia is reviewed with reference to the factors that affect prevalence, including definitions of the term, case mix, and selection. Period prevalence of dyspepsia in several different populations gives an average of 32%, of which 24% is accounted for by recognized ulcer disease. Dyspepsia appears to comprise about 70% of patients' gastrointestinal problems in a large prospective survey carried out in general practice in England, higher than some other estimates. The proportion of gastrointestinal disease in general practice consultations is examined, and while it accounts for about 5% of all consultations, it accounts for about 14% of patients consulting with a problem. There seems to have been a decline of 15% over 30 years. The difficulty of establishing the population prevalence of functional dyspepsia is emphasized, and several studies in which the proportion in general gastrointestinal outpatients has been measured are reviewed. Some data from a Glasgow study in which clinical histories have been recorded directly from patients by a computer system (GLADYS) show the prevalence of several common gastrointestinal symptoms in a clinic population and also of dysmotility-like dyspepsia. Such descriptive data should also be used for predicting diagnosis and for selecting patients to investigate. Key words: Dysmotility-like dyspepsia; dyspepsia; epidemiology; prevalence; time trends R. P. Knill-Jones, F.R.C.P., F.F.P.H.M., Glasgow, Glasgow G12 8QQ, Scotland
The epidemiology of any disease has two main components. First its descriptive clinical epidemiology, such as the frequency in affected patients of the symptoms that characterize the disease. This descriptive approach is familiar to clinicians but can only be undertaken if there is an acceptable and independent 'gold standard' for the diagnosis of the disease in the first place. I n the absence of a gold standard there have to be reproducible and agreed definitions of a set of clinical criteria that distinguish patients with the disease from those without. The second epidemiological component concerns incidence—the number of new cases of a disease over a period of time—and prevalence— the total number of cases in a community affected by a disease at one point in time. Both these measures involve relating the number of cases of the disease to the larger population from which
Dept. of Public Health, University of
they come. One can now determine several i m portant features about the natural history of the disease: its incidence, prevalence, mortality, and the effects of its morbidity on health. I t is, however, impossible to undertake proper epidemiological studies without agreed definitions of the disease in question. It is therefore hardly surprising that epidemiologists have particular difficulty with 'dyspepsia'. Furthermore, it is not a single disease entity; far from i t : it is merely a description of one of the symptomatic presentations of many diseases, which in turn may not even be confined to the gastrointestinal ( G I ) tract. So 'dyspepsia' is a description, or label, which is applied to a large group of patients who describe their symptoms in a particular way. The word is not in common parlance. Patients do not use the term much themselves, preferring 'indigestion'; even so only
several different symptoms as far as patients are concerned; to some it means 'heartburn' and to others epigastric pain, or flatulence, or belching (2). Dyspepsia is therefore best kept as a medical term to be used as a label for a group of patients with certain symptoms.
HISTORICAL DEFINITIONS The definitions o f exactly what physicians mean by dyspepsia vary and, to complicate matters further, vary by at least as much as their patients' use o f the term indigestion. The definition used in our Unit, which includes a symptom duration of 4 weeks or more, is 'episodic, recurrent or persistent abdominal pain or discomfort, or any other symptom referable to the alimentary tract excluding jaundice or bleeding' (3). A recent populationbased study (4) used responses to specific questions which defined dyspepsia: 'yes' to 'Have you ever had abdominal pain of at least 2 weeks' duration?' and ' I f yes, was the pain located to the upper abdomen?'; or 'yes' to 'Have you ever had heartburn or acid regurgitation almost daily for at least 1 week?' Other definitions exist (5-7) and are discussed by R. C. Heading elsewhere in this supplement. Each definition has at least a clinical component to it—for example, the duration and the spectrum of symptoms included. I n addition there is often an implied operational or process component as well, since both the pattern of referral and the types of investigation undertaken are important in delineating the patient group of
wards (7), with only 14% of patient evidence of a lesion, is quite different fi G I outpatients (3) with about 40% average from 14 studies of 46% (8). ferences are also reflected in the defini in the first two studies—one exclude (3), the other includes jaundice (7). Th that inpatients in a surgical ward are I provide a representative sample of pa dyspepsia, most of whom are seen by ] with the majority being seen as outpal implication is that the many studies looked at the epidemiology of dyspep easily be compared. The definitions us too variable. Furthermore, the selectioi the populations being studied are l i l equally variable.
Problems o f definition are particularl studies of patients initially defined dyspepsia but who do not turn out to ha investigations for organic disease. The; then placed i n subgroups, such as ' dyspepsia' or 'essential dyspepsia'. Wh( subgroups are useful, let alone wh< are defined well enough for compai demiology, is a matter for continued c
P R E V A L E N C E O F DYSPEPSIA I N G E N E R A L POPULATION (TABLE
One of the earliest studies on prevaleno by Merewether in the 1940s (quoted amongst 354 men working in British
Table I. Period prevalence of dyspepsia in different populations Dyspepsia,
Peptic ulcer,
Country
No. in sample
%
%
Percentage of dyspepsia accounted for by peptic ulcer
England England Scotland Denmark Sweden England U.K. Norway Total
354 5951 1494 1052 3304 2066 7428 2027 23676
30 31 20-23 25 19 41 41 24 32
4 6 12 5
13 19 56 20
8 10 7.7
20 20 42 24
-8
-
surgical with no am that i n and the Tiese difions used jaundice i suggests ilikely to ents with ivsicians, jnts. The hat have a cannot
plants. Thirty per cent complained of indigestion, including 3.9% with a history of peptic ulcer, Later studies have shown similar high prevalences i n several populations. For example, 5951 men i n England (9) had a prevalence of 3 1 % , with 6% ulcers; 1494 men in Scotland (10) had a prevalence of 20-23%, with 12% ulcers; of 1052 Danes (11) 25% had dyspepsia, 5% ulcers; of 3304 Swedish men (12) 19% had dyspepsia; of 2066 men and women in England (3) 38% had dyspepsia, 10% ulcers; of 7428 men and women in England and in Scotland (13) 4 1 % had dyspepsia, 8% ulcers; and o f 2027 men and women in northern Norway (4) 24% had dyspepsia, 10% ulcers.
I are just :actors i n ly to be
The range from 19% to 4 1 % probably reflects the survey methods and questions as much as true differences in incidence. For example, the time periods varied from recording dyspepsia within the last 3 months (12), the last 5 years (9), to ever having dyspepsia (4). There is, however, no correlation between the length of the period concerned and the actual period prevalences in
great for • having positive ire often nctional ter these er they ive epi'ate.
Table I . The proportion of dyspepsia accounted for by peptic ulcer is about 20% for five of the studies. As might be expected, the proportion is higher in the two more northerly populations having the higher prevalences of peptic ulcer— Scotland (12%) and northern Norway (10%)—at 56% and 4 2 % , respectively. However, these populations have a prevalence of dyspepsia that is low compared with the average in Table I .
as done
Other studies are based on discrete symptoms
Ref- 9) ;e-oven
that may be part of some definitions of dyspepsia, and these give different results. For example, an extensive study of community health in Glasgow in 1972 (14) obtained interviews on a population sample of 1344 individuals (47% male) and asked about symptoms within the last 2 weeks. Heartburn was present in 6.3%, indigestion in 6.3%, and both heartburn and indigestion in 2.2%, giving a total (after rounding) of 14.7%. Another mutually exclusive group of questions found that 6.6% had had 'abdominal or tummy pain' in the preceding 2 weeks. Finally, 8.9% of the total sample had noted nausea, retching or vomiting during the same period. Although these figures add up to 30.2%, which might be thought to roughly equate with a broad definition of
dyspepsia, the three groups of symptoms wei recorded independently, and some subjects ma have experienced more than one; for this reaso the figure is excluded from Table I . A t least th study gives an upper limit for the prevalence c 'dyspepsia' in an urban Scottish population, some what higher than an earlier estimate of 20-23°/ (10) in northeast Scotland.
P R E V A L E N C E O F DYSPEPSIA I N GENERAL PRACTICE
Dyspepsia is said to account for about 2-3% of a consultations in general practice and for betwee 20% and 40% of all gastrointestinal consultation with general practitioners (15) and to have considerable economic impact (16). Some of thes< estimates may be too low.
A prospective 1-year survey was conducted b the Royal College of General Practitioners i England (17), to establish morbidity and referra patterns for a large number of conditions. Thi study provides some interesting data about G disease in primary care. The total practice popula tion from which the patients came was 332,270 spread amongst 48 practices (143 physicians). Th study represents 307,803 person-years of follow up. Table I I summarizes some detailed data for G conditions.
These figures provide an illustration of th conditions seen in a general practice setting i England in 1981-2. The total number of consulta tions for all conditions per 1000 members o f th practice populations (whether healthy or unwel was 3396; i n other words, every member o the sample population consulted their genera practitioner on average 3.4 times in a year. G diseases and symptoms accounted for 160.1 of a consultations and are shown i n detail i n Table 11 They represent 4.7% of the total number o consultations, higher than the figures quote above (15).
The total number of patients consulting was 71 per 1000 in the sample population per yea amongst whom patients with G I disease accounte for 97.1, or 13.6%. I n other words, there wa a 13.6% chance that any patient visiting the GP had a G I problem, emphasizing the relativ
Consultations
Patients
Consultations per patient
All conditions (001-999, V01-V82)*
3396
712
4.8
GI symptoms Nausea and vomiting (787) Heartburn (787.1) Gas and wind (787.3) Abdominal pain (789)
12.9 3.5 3.8 45.7
9.6 2.1 2.4 28.2
1.3 1.7 1.6 1.6
GI diseases Oesophageal disease (530) Hiatus hernia (551.3, 552.3, 553.3) Gastric ulcer (531) Duodenal ulcer (532) Other peptic ulcers (533-4) Other diseases of stomach and duodenum (535-7) Disorders of function of stomach (536)
3.9 7.2 2.3 9.6 2.0 3.6 19.7
2.4 3.4 1.0 3.6 1.1 2.7 12.7
1.6 2.1 2.3 2.7 1.8 1.3 1.6
Subtotal = Dyspepsia Irritable bowel syndrome (558, 564.1, 564.5) Constipation (564.0) Ulcerative colitis, diverticulitis, rectal disease (555-6, 558, 562, 564.1, 564.5, 569.3-4)
114.2 18.4 13.2
69.2 11.5 9.2
1.65 1.6 1.4
14.3
7.2
2.0
Total GI symptoms and diseases
160.1
97.1
1.65
*Code = ICD9 Rubric. importance of G I disease in the community, even though, i n terms of consultations, the impact was considerably lower at 4.7% of attendances. The discrepancy is due to the majority of G I conditions being relatively short-lived, compared with many of the problems treated in general practice which require several visits.
Thus there are 114.2 consultations for dyspepsia ( 7 1 % of all G I consultations), and 69.2 patients per 1000 consult per year with dyspepsia as theii problem.
The main part of Table I I shows the number of different patients consulting with G I problems and, by simple division, the average number of consultations for each condition. For example, there were 3.9 consultations by 2.4 patients per 1000 due to oesophageal disease, an average of 1.6 consultations for each patient with an illness episode due to an oesophageal problem. There were also 7.2 consultations for hiatus hernia, and, finally, in the ill-defined symptom section of the I C D codes there were an additional 3.5 consultations recorded for heartburn. Most of the diseases and symptom codes i n Table I I could be regarded as 'dyspepsia' (the term itself is not separately coded i n this study), the exceptions being the three groups of bowel diseases. I f these are excluded from the G I total, the rest give approximate figures for 'dyspepsia'.
Two earlier studies of similar design i n England enable the examination of secular trends (17). Excluding the consultations coded only as symptoms in the I C D , and focusing on the consultations for which a diagnostic code was given, there is a marked decline of about 33% in patients consulting for G I problems from 1951-2 to the 1981-2 survey (17). Although a decline i n peptic ulcer oi 38% (Table I I I ) occurred over this time, il provides only a small part of the spectrum of G] disease seen in general practice, and the mair decline has been in non-specific problems such a; 'disorders of function of stomach'. Table H I show; data for which the diagnostic classification hai remained comparable over the three surveys. Tht overall decline in these restricted data has beer about 15%. There remains considerable uncertainty abou
TRENDS OVER T I M E
Table III. Patients consulting general practitioners: rates per 1000 population/year Years Condition Oesophagus/mouth Ulcers Gastric Duodenal Other Disorders of stomach Constipation Total
1955-6
1971-2
1981-2
Change
6.6
8.8
11.2
Increase
1.9 5.9 1.4 21.5 8.1 45.4
0.8 3.6 2.0 12.5 8.4 36.1
1.0 3.6 1.1 12.7 9.2 38.8
Decrease Decrease Stable Decrease Increase Decrease
the prevalence and incidence of dyspepsia, and yet there is still much to learn, i n particular about the process that patients follow and which may result in a request for medical advice. Finally, i t is important to recall the concept of the 'symptom iceberg* (14), as at least half of all recognizable episodes of dyspepsia are treated by the patients themselves (18).
PREVALENCE OF F U N C T I O N A L DYSPEPSIA I f the epidemiology of dyspepsia is difficult, then that of functional dyspepsia is even more so. Nonetheless, the prevalence of functional dyspepsia is available for some countries. Unlike dyspepsia it is rarely measured in the community at large and is measured most often in populations of patients referred to hospital. Some published prevalence findings are, for example, Scotland (19), 25% (excluding a further 15% with irritable bowel syndrome); U S A , 19% (20); Norway, 23% males and 18% females (21); Iceland, 64% and Denmark, 34% (22)—but also 76% (23); Singapore, 45% males and 57% females, and no racial differences (24); South Africa, 30% (P. Briscoe. Personal communication to D r . Verlinden, 1990); East Africa, 62% (25); Japan, 35% males and 42% females ( Y . Kameyama. Personal communication to D r . Verlinden, 1990); England, 48% (26), 22% plus 15% with gastritis (27), and 40% (28); Sweden, 49%, plus 15% with functional dyspepsia and irritable bowel syndrome (29); and Nigeria, 69% (30). The overall range from 19% to 76%, and even that from 34% to 76% in the two
studies from Denmark, tells us more ab methodological differences between reses groups and the selection factors that operati different populations than it enlightens us ab the comparative prevalence of functional < pepsia in different countries.
What can be done? The only accept: approach is to build on the research groups us well-defined questionnaires, combine them \ agreed-upon protocols for diagnostic procedu and set up simultaneous studies in centres re] senting different countries. Both the Mayo gr (20) and the groups in Glasgow and i n Swe (17, 3 1 , 32) have this potential, and some a parisons are beginning to appear (32). One i paper questionnaires (20); the others use c< puters to obtain the necessary informal directly from patients. The Glasgow group used both methods, and data are becoming a\ able on large numbers of patients.
PREVALENCE OF SYMPTOMS— D Y S M O T I L t T Y - L I K E DYSPEPSIA Table I V shows the prevalence of a few of
symptoms that may characterize dysmotilitydyspepsia in two populations. One consists of c 3000 patients with dyspepsia in accordance i the definition given earlier (3), whose clir histories were obtained by computer i n t e n using G L A D Y S software (33, 34). The pat* comprised about 600 seen in general practic Glasgow (35) and 650 in the Netherlands (36, The other patients were referred to G I outpati in Glasgow. The second series of cases
Computer interviews, Glasgow (2476) Forms, Holland (654) Glasgow (1540) Early repletion Bloating Sometimes Often Flatulence Epigastric pain Nausea or vomiting
35%
46%
31% 40% 48% 46% 56%
24% 21% 50% 58% 51%
their histories taken by gastroenterologists using structured forms in a G I clinic in Glasgow. I n general, these G I symptoms are common. I n a study examining agreement between duplicate interviews (computer and doctor) most G I symptoms tend to be more frequently recorded by the computer than by doctors using forms (38).
interviewing patients and for cc relevant diagnostic data substantia cost of data collection and analysis of such clinical data is to comb population-based epidemiological a decision analysis to choose am alternatives (42) or investigation !
Table V shows the number of patients with one, two, or more of the five symptoms listed. Only 9-10% of patients had none o f the five symptoms, and approximately half had two or three.
A n earlier working party definition of dyspepsia: 'Upper retrosternal pain, discomfort, heai vomiting, or other symptom coi referable to the proximal aliment A second working party propose persistent abdominal symptoms, c feeding, which patients or physicia due to disorders of the proximal digestive tract' (45). Elsewhere i n t R. C. Heading summarizes the working group: 'Episodic or persis that include abdominal pain or d term dyspepsia is not applied to symptoms are thought to be arisir the proximal gastrointestinal trai the G I research community to adc definition consistently may be di essential i f sound comparative sti done (46). For example, patients and negative investigations in J considered to have 'chronic gast the histological findings may be. situation is similar, patients with pepsia' being also regarded as hi (15). I t is likely to be many yea cross-country comparative epider carried out to the high standard set Norway (4).
What can be done with these data? Whereas descriptive comparisons are of clinical interest, there have been several efforts to produce scoring systems using the data to improve diagnosis and selection of patients for investigation or for treatment (17, 3 1 , 32, 39). Most of the groups use the same statistical method (logistic regression), and at least one objective is to estimate the probability of the diseases causing patients' symptoms (40). However, transfer of existing data across language, cultural, and methodological barriers for predictive purposes is not necessarily easy (41), although the use of computers for
Table V. Sum of early repletion, bloating, flatulence, epigastric pain, nausea, or vomiting
Sum None 1 2 3 4 5
Computer interviews Glasgow (2476) Forms Holland (654) Glasgow (1540) 9% 19% 28% 26% 15% 4%
10% 18% 26% 22% 17% 6%
Dyspepsia has long been associated with diagnostic uncertainty (47), but without clear definitions and studies based on sound epidemiological principles, it may continue to provide a minefield for dissent.
REFERENCES 1. Kingham JGC, Fairclough PD, Dawson AM. What is indigestion? J R Soc Med 1983, 76, 183-186 2. Metcalf M, Youngs GR. Prevalence of symptoms of dyspepsia. Br Med J 1989, 298, 526-527 3. Crean G . Towards a positive diagnosis of irritable bowel syndrome. In: Read NW, ed. Irritable bowel syndrome. Grune & Stratton, New York, 1985, • 29-42 4. Bernersen B, Johnsen R, Straume B, Burhol B, Jenssen T G , Stakkevold PA. Towards a true prevalence of peptic ulcer: the S0rreisa gastrointestinal disorder study. Gut 1990, 31, 989-992 5. Jones R, Lydeard S. Prevalence of symptoms of dyspepsia in the community. Br Med J 1989, 298, 30-32 6. Talley NJ, McNeil D, Piper DW. Discriminant value of dyspeptic symptoms: a study of the clinical presentation of 221 patients with dyspepsia of unknown cause, peptic ulceration, and cholelithiasis. Gut 1987, 28, 40-46 7. Horrocks JC, DeDombal FT. Computer-aided diagnosis of 'dyspepsia'. Am J Dig Dis 1975. 20, 397-406 8. Thompson WG. Nonulcer dyspepsia. Can Med Assoc J 1984, 130, 565-569 9. Doll R, Avery Jones F , Buckatzsch MM. Occupational factors in the aetiology of gastric and duodenal ulcers, with an estimate of their incidence in the general population. MRC Special Report, Series 276. HMSO, London, 1951, 7-96 10. Weir RD, Backett EM. Studies in the epidemiology of peptic ulcer in a rural community: prevalence and natural history of dyspepsia and peptic ulcer. Gut 1968, 9, 75-83 11. Hollnagel H , Nerrelund N, Larsen S. Mavetarmsvmptomer blandt 40 arige i Glostrup. Ugeskr Laeger 1982, 144, 267-272 12. Tibblin G . Introduction to the epidemiology of dyspepsia. Scand J Gastroenterol 1985, 20(suppl 109), 29-33 13. Jones R H , Lydeard S E , Hobbs FDR, et al. Dyspepsia in England and Scotland. Gut 1990, 31,401405 14. Hannay DR. The symptom iceberg: a study of community health. Routledge & Kegan Paul, London,1979 15. Petersen H. Epidemiological and economic aspects of dyspepsia. In: Colin-Jones D G , ed. Practical approaches to the management of dyspepsia. The Medicine Group, Langhorne, Pa., 1989, 2-5 16. Nyren O, Adami HO, Gustavsson S, Loot L , Nyberg A. Social and economic effects of non-ulcer dyspepsia. Scand J Gastroenterol 1985, 20(suppl 109), 41-45
17. Royal College of General Practitioners, Office Population Censuses and Surveys, Department Health and Social Security. Morbidity statisti from general practice 1981-82. Third nation study. HMSO, London, 1986 18. Jones R. Self-care and primary care of dyspepsia: review. Family Practice 1987, 4, 68-77 19. Knill-Jones RP. Diagnostic systems as an aid clinical decision making. Br Med J 1987,295,139 1396 20. Talley N, Phillips SF, Melton J , Wiltgen C Zinsmeister AR. A patient questionnaire to identi bowel disease. Ann Intern Med 1989, 111, 671-67 21. Johnsen R, Straume B, F0rde OH. Peptic ulcer an non-ulcer dyspepsia—a disease and a disorde Scand J Prim Health Care 1988, 6, 239-243 22. Oddsson E , Binder V, Thorgeirsson T, et al. prospective comparative study of clinical and histo logical characteristics in Icelandic and Danis patients with gastric ulcer, duodenal ulcer, and X ray negative dyspepsia. Scand J Gastroentero 1977, 12, 689-694 23. Bonnevie O. The incidence and results of radio logical examination of the stomach using a bariur meal. Scand J Gastroenterol 1976, 11, 839-847 24. Kang J Y, Guan R, LaBrooy SJ, Lim KP, Yap I. Th influence of sex, race and dialect on peptic ulcer an non-ulcer dyspepsia in Singapore. Ann Acad Me 1983, 12, 527-531 25. Gatumbi I , Roy AD. The prevalence of peptic ulce dyspepsia in a rural community in Kenya, East Af Med J 1970, 47, 627-633 26. Harvey R F , Salih SY, Read A E . Organic an functional disorders in 2000 gastroenterology ou patients. Lancet, 1983, 1, 632-634 27. Hallissey MT, Allum WH, Jewkes AJ, Ellis DJ Fielding JWL. Early detection of gastric cance Br Med J 1990,301,513-515 28. Lobo A, Dickinson R. Managing the dyspept patient: experience of a single-visit dyspepsia clinic J R Soc Med 1988, 81, 212-213 29. Kagevi I, Lofstedt S, Persson L-G. Endoscop findings and diagnoses in unselected dyspept patients at a primary health care center. Scand Gastroenterol 1989, 24, 145-150 30. Olubuyide IO, Ayooia E A , Okubanjo AO, Atob MA. Non-ulcer dyspepsia in Nigerians and then peutic results. Scand J Gastroenterol 1986,21(sup 124), 83-87 31. Crean GP, Card WI, Beattie AD, et al. Ulcer-li dyspepsia. Scand J Gastroenterol 1982, 17(sup 79), 9-15 32. Lindberg G, Lundquist P, Seensaiu R, Nilsson LF Knill-Jones RP. Validation of G L A D Y S , a mien computer system for interrogation of patients an decision support in dyspepsia—probabilistic pe formance in 651 Swedish patients. Theor Surg 199 5, 160 33. Lucas R, Card WI, Knill-Jones RP, Watkinson C Crean GP. Computer interrogation of patient Br Med J 1976, 2, 623-625 34. Collins JSA, Knill-Jones RP, Sloan JM, et i A comparison of symptoms between non-ulc dyspepsia patients positive and negative f
system tor history taking. Gut 1989, 30, A1475 35. Knill-Jones RP, Dunwoodie WM, Crean GP. A computer-assisted diagnostic decision system for dyspepsia. In: Sheldon M, Brooke M, Rector A, eds. Decision making in general practice. Macmillan, Basingstoke, 1985, 203-219 36. Meuwissen S, Hordijk M, Tuynman H, et al. Gladys as a diagnostic tool in Dutch patients. VII International Congress of Gastroenterology, Rome, September 1988. Gastroenterol Int 1988, l(suppl 1), A327 37. Poen H, Meuwissen S, Hordijk M, et al. Computer interrogation with Gladys of Dutch patients. Validation of the system. VII International Congress of Gastroenterology, Rome. September 1988. Gastroenterol Int 1989, l(suppl 1), A326 38. Knill-Jones RP, Crean GP, Holden RJ, Marjoribanks F. Practical aspects of expert systems and computer-taken clinical histories in dyspepsia. Proceedings of the First Hong Kong Medical Informatics Conference, 1990 (in press) 39. Mann J , Holdstock G , Harman M, Machin D, Loehry CA. Scoring system to improve cost effectiveness of open access endoscopy. Br Med J 1983, 28, 73-77 40. Spiegelhalter DJ, Knill-Jones RP. Statistical and knowledge-based approaches to clinical decision support systems, with an application to gastro-
1984, 147, 35-77 41. Knill-Jones RP, Drummond M, Koi Economic evaluation of gastric ulce. patients with arthritis receiving nor inflammatory drugs. Postgrad Me 639-646 42. Jonsson BG. Management of dyspe of cost-benefit analysis. Practical gu management of dyspepsia. Confei ings, Geneva, March 1990. Smith ¥ London, 1990, 40-45 43. Lindberg G , Seensalu R, Nillson ! Kager L, Knill-Jones RP. Transfera puter system for medical history taki support in dyspepsia. Scand J Gasti 22(suppl 128), 190-196 44. Colin-Jones DG. Management of dy of a working party. Lancet 1988, 1, 45. Barbara L , Camilleri M, Corinal Definition and investigation of dj sensus of an international ad hoc wo Dis Sci 1989, 34, 1272-1276 46. Jones R. Dyspeptic symptoms in tl Gut 1989, 30, 893-898 47. Research Committee of the Colle Practitioners. The continuing ob recording of morbidity. J Coll Gen 107-128