political issue, and the omens are not promising. The one ... of the very best sort-that is, clinicians from ... of patients with a very good prognosis "is difficult";.
It is true that the Priory Hospitals Group is the largest provider of private psychiatric care in Britain. The authors fail to make it clear, however, that this group is the "encouraging exception"2 that they refer to. Thus the exception becomes the rule. Following the example set by the Priory group there is reason to be optimistic that private psychiatric hospitals in the United Kingdom will take a different approach to their American counterparts. Responsibilities to research and training are taken seriously. A recently held Priory Fellows Day, devoted entirely to projects associated with this hospital and attended by 120 psychiatrists, provides further evidence. In arguing that private care is elitist the authors again overlook the fact that 71 NHS patients have been successfully cared for in the Priory Hospital over the past three years. These were very disturbed patients who would otherwise have been admitted to a locked ward. They were mostly patients of social class V from the deprived inner city area in which Professor Marks and Dr Thornicroft work. Surely the future lies in close cooperation between the private, academic, and NHS sectors. To achieve this we need to be fair minded, not prejudiced. JOHN COBB
investigations because "the endoscopic yield from the investigation of x-ray negative dyspepsia after a double contrast barium meal is minimal."' This is not hearsay or anecdotal wisdom but medical audit of the very best sort-that is, clinicians from different specialties coming together to test their tests. But audit has little value if we do not learn from the results and make the changes to our practice that are appropriate to those results. An editorial written by gastroenterologists advising general practitioners on the appropriate sequence of hospital investigations that totally ignores the results of relevant audit suggests that some doctors will continue to remain locked within their own particular fields of expertise, playing with their own particular toys, unable or unwilling to argue the case for (or against) a sensible permutation of the skills and techniques that are available for us to use together in order to help our patients. It may well be proper for the profession to challenge many of the NHS reforms,' which are being imposed in a seemingly cavalier fashion. But the future looks really inauspicious if doctors are equally cavalier and ignore the one item regarded as a progressive step forward. M A SAMPSON C RECORD Northwick Park Hospital,
Middlesex HAl 3UJ
Priory Hospital, London SW15 5JJ
I .Brown C, Rees WDW. Dyspepsia in general practice. Br MedJ
I Marks 1, TIhornicroft G. Private inpatient psychiatric care. BrMedJ 1990;300:892. (7 April.) 2 Kelly D. Private sector psychiatric services. Bulletin of the Royal College of Psvchiatri'ss 1989;13:199.
Dyspepsia in general practice SIR,-Drs C Brown and W D W Rees in their recent editorial on dyspepsia in general practice point out that the referral rate for upper gastrointestinal endoscopy has outstripped the resources available and has led to long waiting lists.' They suggest that the answer is to treat all patients, except those with a history suggestive of gastric malignancy, with antacids or H2 receptor antagonists for four to six weeks before considering investigation with endoscopy. They singularly fail to suggest, or even to bother to discuss, the alternative method of investigation-namely, the double contrast barium meal. An estimated 400 000 double contrast barium meal investigations are carried out in England and Wales each year2; the examination is comparable in accuracy with endoscopy,' except in the diagnosis of fine mucosal inflammation. Moreover, modern barium radiology is better than endoscopy at showing some aspects of function, hiatal hernias, and extrinsic mass lesions, and it also provides a permanent record4; furthermore, it is safe-mortality from endoscopy varies between one in 5000 and one in 13 Oo."6 Endoscopy is an observer dependent investigation that does not provide a permanent record and that may well miss serious lesions such as carcinoma of the stomach," and even with careful biopsy of gastric lesions it can still result in misleading benign results on histological examination.' But the omissions in the editorial have unintentionally highlighted a contemporary medicopolitical issue, and the omens are not promising. The one aspect of the government's proposals for reforming the NHS that has been warmly welcomed by doctors has been the importance and emphasis placed on audit.' There have been few prospective surveys in the United Kingdom comparing the findings on double contrast barium radiology with those on skilled endoscopy. Salter, a gastroenterologist, carried out such a survey and concluded that endoscopy should usually assume third rather than second place in the sequence of
BMJ
VOLUME
300
28
APRIL
1990
1990;300:829-30. (31 March.) 2 Simpkins KC. What use is barium? Clin Radiol 1988;39:469-73. 3 Salter RH. X-ray negative dyspepsia. BrMedj 1977;ii:235-6. 4 Cotton PB, Shorvon IlJ. Analysis of endoscopy and radiographin the diagnosis, follow-up and treatment of peptic ulcer disease. Clinics in Gastroenterology 1984;13:383-403. 5 Schiller KFR, Prout BJ. Hazard of endoscopy. In: Schiller KFR, Salmon PR, eds. Modern topics in gastrointestinal endoscopy. London: Heinemann Medical, 1976:147-65. 6 Dawson J, Cockel R. Oesophageal perforation at fibreoptic gastroscopy. BrMedj 1981;283:583.
7 Fraser (GM, Earnshaw PM. The double contrast barium meal: a correlation with endoscopy. Clin Radiol 1983;34:121-31. 8 Farini R, Farinati F, Cardin F, cl al. Evidence of gastric carcinoma during follow up of apparently benign gastric ulcer. Gut 1983;24:A486. 9 Secretaries of State for Health, Wales, Northern Ireland,. and Scotland. Working for patients. London: HMSO, 1989.
(Cmnd 555.)
Node negative breast cancer SIR, -After 15 years of work on prognostic factors in breast cancer it is saddening to read in the regular review article by Drs S M O'Reilly and M A Richards that accurate identification of a subgroup of patients with a very good prognosis "is difficult"; also that "further research is needed in the ability of known prognostic factors to select such groups."' We refer to several publications from our unit. In 1982 we advanced an index that combined lymph node stage, histological grade, and tumour size as accurately identifying a group with a very good prognosis.2 In 1987 we confirmed our index 1.0-
> 080 0
.00. co E
34
,
,0 6X06-
5-01
0.2
0-0. 0
50 34e -
-7.0
8 10 12 14 16 Years Survival curves for 1000 women with operable breast cancer by prognostic index 2
4
6
Actuarial survival at 15 years of women with operable breast cancer Actuarial Index 21-34 >3 4-5-0 >5 0
Percentage of women
survival (%)
26 46 27
79 50 18
(the Nottingham prognostic index), which had been based on a retrospective study of 387 patients, with a prospective study of 320 patients.' The index was shown to give highly reproducible results in the estimation of prognosis. Analysis of the index in the first 1000 consecutive patients presenting to our unit between 1973 and 1989 with operable breast cancer (tumour size 5 0 have a very poor survival chance at five years; are likely to have poorly differentiated, oestrogen receptor negative tumours; and seem to be candidates for adjuvant
cytotoxic treatment. Lymph node stage and tumour size are time dependent factors. Histological grading is the best method of providing a biological factor for a prognostic index and has consistently scored better in our series than all of the other factors we have tested4: DNA index,' S phase fraction, oestrogen receptor,6 epidural growth factor receptor,7 c-Erb B2 (C Lovekin et al, unpublished work), C-myc,8 Helix pomatia,9 and epithelial mucin antibody. '0 As these factors are related to grade they do not further refine the index once grade has been incorporated. Our final message is to Drs O'Reilly and Richards and to the authors of similar papers to their review (for example, the National Cancer Institute clinical alert) ": there are published indices that give excellent prognostic discriminationWhy not use them? R W BLAMEY C W ELSTON I 0 ELLIS D A L MORGAN J L HAYBITTLE
Citv Hospital, Nottingham I O'Reilly SM, Richards MA. Node negative breast cancer. BrMedJ 1990;300:346-8. (10 February.) 2 Haybittle JL, Blamey RW, Elston CW, Johnson J, Doyle Pj, Campbell FC. A prognostic index in primary breast cancer. Brj Cancer 1982;45:361-6. 3 Todd JH, Dowle C, Williams MR, et al. Confirmation of a prognostic index in primary breast cancer. Br J Cancer 1987;56:295-9. 4 Elston CW. Grading of invasive carcinoma of the breast. In: Page DL, Anderson TJ. Diagnostic histopathology of the breast. Edinburgh: Churchill Livingstone, 1987:300-l 1. 5 Dowle CS, Owainati A, Robins A, et al. Prognostic significance of the DNA content of human breast cancer. Br J Surg 1987;74: 133-6. 6 Williams MR, Todd JH, Ellis 10, et al. Oestrogen receptors in primary and advanced breast cancer: an eight year review of 704 cases. BrJ' Cancer 1987;55:67-73. 7 Lewis S, Locker A, Todd JH, et al. Epidural growth factor receptor expression in human breast carcinoma. J Clin Pathol (in press). 8 Locker AP, Dowle CS, Ellis 10, et al. C-myc oncogene prodttct expression and prognosis in operable breast cancer. BrJ Cancer 1989;60:669-72.
1137
9 Fenlon S, Ellis 10, Bell J, Todd JH, Elston CW, Blamey RW. Helix pomatia and Ulex europeus lectin binding in human breast carcinoma. 7Pathol 1987;152:169-76. 10 Ellis 10, Hinton CP, MacNav J, et al. Immunocvtochemical staining of breast carcinoma with the monoclonal antibodv NCRC 11: a new prognostic indicator. Br Med 1985;290: 881-3. 11 National Cancer Institute. Alert on node negative breast cancer. Breast Cancer Research and Treatment 1988;12:3-5.
Prophylaxis against hepatitis A for travel SIR, -We would like to make some comments on the article by Ms Phyllis Moore and colleagues on prophylaxis against hepatitis A for travellers.' Although they comment on patients making a single visit, the authors give no advice concerning those who travel regularly to endemic areas. Our practice is to offer a test for hepatitis A IgG to all patients in this group and inform the patient and the general practitioner of the result. Previous studies show that the prevalence of antibodies against hepatitis A increases with age,2" and we are surprised that Ms Moore and colleagues did not analyse the age of the patients and take it into consideration. We would also like to add age over 40 as a risk factor for previous exposure to hepatitis A and that patients who are aged >40 should also be considered for testing to assess immunity. Giving human normal immunoglobulin to immune patients not only causes pain but also is unnecessary. Our largest problem is that of patients who attend insufficiently early before their date of departure for a result to come back before they leave. K R NEAL E M DUNBAR
Regional Department of Infectious Diseases and rropical Medicine, Monsall Hospital, Manchester M 10 8WR I Moore P, Oakeshott P, Logan J, Law J, Harris DM. Prophylaxis against hepatitis A for trasvel. Br Mied .7 1990;300:723-4. 17
March.) 2 Kudesia G, Follett EAC. Not all travellers need inmunoglobulin for hepatitis A. Br Med17 1987;295: 118. 3 Parrv JV, Farrington CP, l'erry KR, et al. Rational programme for screening travellers for antibodies to hepatitis A virus. Lancet 1988;i: 1447-9.
Raynaud's syndrome SIR,-With regard to the editorial on Raynaud's syndrome by Dr E D Cooke and Professor A N Nicolaides' I wish to draw attention to two additional areas of work. We recently showed that vasospastic attacks of Raynaud's phenomenon could be induced in anaesthetised fingers of patients with primary Raynaud's disease and those with Raynaud's phenomenon secondary to scleroderma.' Vasospastic attacks occurred in nine of 11 patients with primary disease and eight of 10 patients with secondary disease. Two fingers on one hand were anaesthetised by local injection of lignocaine, and the effectiveness of the nerve blockade was verified by plethysmography. The frequency of vasospastic attacks in nerve blocked fingers was not significantly different from that in the corresponding intact fingers on the other hand. These findings argue against Raynaud's theory of sympathetic hyperactivity in the aetiology of this disorder and in favour of Lewis's hypothesis of a "local fault" in the digital blood vessels.' This reasoning is supported by the fact that the results of microelectrode recordings of skin nerve sympathetic activity showed no differences between patients with primary Raynaud's disease and control subjects.4 In treating patients with primary Raynaud's
1138
disease we have found biofeedback of finger temperature to be very effective. A controlled investigation of four behavioural procedures found that temperature feedback alone or under mild cold stress had improved the symptoms of 67% and 92% of patients respectively at follow up at one year.5 These findings were maintained three years after treatment.6 A series of subsequent investigations replicated the above findings and showed that temperature feedback produced increased finger capillary blood flow as measured by washout of iodine- 131/; this result was obtained again one and two years later. Intra-arterial infusions of propranolol during biofeedback showed the involvement of a 13 adrenergic vasodilating mechanism, which was not affected by digital nerve blockade. Thus the effect of treatment is not mediated through efferent sympathetic nerves. ROBERT R FREEDMAN Behavioural Medicine Laboratory, of State School Wayne University Medicine, Detroit, Michigan 48201, United States 1 Cooke ED, Nicolaides AN. Raynaud's syndrome. Br Med J
1990;300:553-5. (3 Alarch.) 2 Freedman RR, Maves MM, Sabharwal SC, Keegan DM. Induction of vasospastic attacks despite digital nerve block in Raynaud's disease and phenomenon. Circulation 1989;80:
3 4
5 6 7
859-62. Lewis F. Experiments relating to the peripheral mechanisms involved in spasmodic arrest of the circulation in the fingers. A variety of Raynaud's disease. Heart 1929;15:7-101. Fagius J, Blumberg H. Sympathetic outflow to the hand in patients with Raynaud's phenomenon. Cardiovasc Res 1985; 19:249-53. Freedman RR, lanni P, Wenig P. Behavioral treatment of Raynaud's disease.J Consult Clin Psychol 1983;51:539-49. Freedman RR, lanni P, Wenig P. Behavioral treatment of Ravnaud's disease: long-term followup. J Consult Clin Psvchol 1985;53: 136. Freedman RR, Sabharwal SC, lanni P, Desai N, Wenig P, Mayes M. Nonneural beta-adrenergic vasodilating mechanisms in temperature biofeedback. Psychosom Med 1988;50:394-401.
Accuracy of weighed dietary records SIR,-Dr M B E Livingstone and colleagues sought to provide an independent evaluation of seven day dietary records for assessing energy intake and concluded from the results that seven day dietary records showed serious bias.' The subject of their paper is of considerable importance. The general way in which the paper may be received has widespread implications for the interpretation of a large body of data and our understanding of the relations between health and disease, which has to be a matter of considerable concern. We, too, appreciate the limitations of estimating habitual energy intake from records of food consumption and are concerned over claims of habitual energy intakes that are physiologically impossible. We would contend, however, that the approach adopted by Dr Livingstone and colleagues for analysing their data is not valid in relation to the conclusions that have been drawn. The authors have compared measurements of food intake, based on seven day dietary records, with measurements of energy expenditure derived indirectly by the use of the doubly labelled water technique. The main analysis is shown in table III of the paper. The population was divided into thirds, based on energy intake, which thereby becomes the independent variable. Total energy expenditure was derived for each third and is shown in relation to energy intake. There was little difference in total energy expenditure between the lower, middle, and upper bands of energy intake. For the upper third of energy intake the measurements of intake and expenditure show concordance, with no significant difference between the two methods. For the middle and lower thirds of energy intake, expenditure is significantly greater than intake. From this the authors go on to
18- o Energy expenditure '*Energy intake 16-
Men
:R 14 m12-
,
10-
Women
8-i
61 Lower Middle Upper Third of energy expenditure Comparison of mean (SE) energy expenditure assessed by doubly labelled water method with mean (SE) energy intake according to seven day weighed dietary records
conclude that for the points where expenditure and intake do not show concordance the measurements of intake were likely to have been in error. If the argument that the values for energy intake are not reliable is accepted then it follows that energy intake cannot usefully be used as the independent variable in the analysis. Indeed, the authors go on to propose that it would be more appropriate to consider total energy expenditure as the independent variable because it is a more reliable measure. If this is so then it would have been of value if they had repeated the analysis using expenditure as the independent variable. If the analysis is carried out in this way, with the population divided into thirds by total energy expenditure, the results shown in the figure are obtained. By using total energy expenditure as the independent variable for the analysis a completely different interpretation of the data may be obtained from that reached by Dr Livingstone and colleagues. It can be shown that the magnitude of the difference between energy intake and expenditure depends on the level of expenditure. For the lower third of total energy expenditure there is good concordance between energy intake and expenditure, with no significant difference between the two approaches (Student's t test). For the middle and upper thirds, however, there is a significant difference between expenditure and intake in both men and women, with intake being only about 76% of expenditure for all four groups. The upper third of total energy expenditure for men is particularly remarkable. One subject (case 2) had an exceptionally high value for energy expenditure of 21 71 MJ/day-the sort of figure that is seen only in the most extreme circumstances. For example, equivalent values of expenditure have been determined in elite racing cyclists simulating stages of the Tour de France.2 After allowances are made for energy expenditure during sleep and discretional activities an estimated 15 MJ/day would be left, to be expended in physical activity. To achieve this energy expenditure the individual would have to be engaged in continuous effort equivalent to running a marathon every day, seven days a week. This subject reported an intake of only 11 17 MJ and was remarkably consistent in his "underestimate," having previously reported an intake of 12 01 MJ. All of the other subjects who made up this third had a total energy expenditure of at least 15 MJI/day, an expenditure that represents a very high rate of sustained physical work. Again their purported underreporting of energy intake was remarkably consistent over months, as shown in figure 1 of the paper. The report clearly provides valuable information
BMJ VOLUME 300
28 APRIL 1990
