1008

BRITISH MEDICAL JOURNAL

(1) Deaths in hospital are not a valid measurement of the outcome of intensive coronary care, since this care may merely postpone the death. The figures show that deaths in the subacute stage were reduced much less than the acute deaths, but unfortunately there are no data for deaths occurring following discharge from hospital. (2) There is very little information about the admission policy to the new coronary care unit and none about the selection of patients for the old unit. As there are 30 beds in the new unit and only three in the old unit, it seems inevitable that the admission policy must have changed, and this is reflected in the number of patients during the two periods-522 during 1973-5 and 196 during 1967-9. Unfortunately there has been no attempt to match the two groups for characteristics which may influence outcome-for example, age, sex, previous myocardial infarctions-and thus it must be concluded that two very different populations of patients are being compared. (3) With a disease such as myocardial infarction, where there have been so many recent changes in the management, it is impossible to attribute changes in mortality rates over a six-year interval to any particular cause, since the influence of the other variables cannot be taken into account. Yet this paper attempts to attribute the apparent decline in mortality to the new coronary care unit and specifically to the pieces of technology housed there. In fact they are not very successful in apportioning credit either to arterial counterpulsation or fluoroscopic equipment (the only examples cited) and conclude by stressing the importance of the "quiet and restful atmosphere." "Atmosphere" is an unusual justification of the new generation coronary care unit, and perhaps more easily attainable elsewhere. (4) Although the introduction suggests that the paper will provide evidence on cost effectiveness, no such information is provided. Nevertheless, the paper ends with the statement that the advances . . . "seem to be justified on economic grounds," a comment less than helpful without supporting information. In conclusion, I do not consider this paper at all useful in clarifying the coronary care unit controversy. In fact, I am afraid that it may be counterproductive. I am disappointed that a paper of this quality is given so prestigious a position in the British Medical J7ournal. SHEILA ADAM Begbroke, Oxon

Bath-taking made easy SIR,-The difficulty experienced by many elderly, infirm, or otherwise handicapped persons in being able to have a bath is all too well known, and in spite of various devices which can be obtained to assist such persons to get in and out of the conventional bath the performance, with or without assistance, is often found to be hazardous if not impossible. It is true to say that most individuals step into their bath facing the taps and the plug, which, of course, they must be able to operate. It seems the natural way. But from this position the difficulty, especially for the infirm, is to be able to sit down in the water in a fully controlled way. If, however, the following simple procedure is adopted most persons who are

23 OCTOBER 1976

ambulant should be able to get in and out of facing the specialty of clinical rheumatology. There is no doubt that rheumatology and the bath with comparative ease and safety: (1) Fill the bath to the required amount and rehabilitation are uneasy bed-fellows and that this combination has resulted in an ill-defined, temperature and then turn off the taps. unattractive specialty with the production of hybrid posts. Part of this problem is semantic. It must be noted that even the Tunbridge Report baulked at defining rehabilitation and indeed Professor Buchanan and his colleagues do not themselves define general medicine, rheumatology, or rehabilitation. In particular, general medicine is not synonymous with the day-to-day management of acute medical emergencies and, although a thorough knowledge of general medicine is very necessary to the practice of rheumatology, the ability to deal with the complete range of medical emergencies has less relevance to a specialty devoted mainly to chronic disease than it obviously has to, say, the practice of cardiology or respiratory medicine, where lifethreatening situations are common. Assuming that we need a rehabilitation specialty, I would contend that the problem is not what to do about clinical rheumatology but how to get people interested in medical R'~ -\ rehabilitation. It is clear from Professor Buchanan's article that internal medicine aspirants are put off rheumatology and rehabilitation by the rehabilitation content of a post. This is because rehabilitation requires different skills from those tested in the MRCP examination and I would go so far as to say that medical rehabilitation will never develop successfully as a specialty until it frees itself from the irrelevant pseudorespectability of 3 > trying to recruit exclusively from the ranks of general medical aspirants. I would suggest that the diploma in medical rehabilitation should be made a quite thorough and comprehensive enough examination so as to obviate the necessity of having any other postgraduate qualification. This would widen enormously the catchment area for medical rehabilitation recruitment. In turn, this would allow clinical rheuma(2) Step into the bath facing the end opposite to tology to develop in some cases as general the taps (fig 1). medicine with an interest but also as a specialty (3) Go down on hands and knees in the bath in its own right like neurology or dermatology water (fig 2). (4) Slide around (fig 3) (either to the right or dealing with the whole range of rheumatic and left, whichever is found to be easier) into a sitting arthritic disease and utilising the complete position. You will now be sitting in the bath in the array of pharmaceutical and physical normal position facing the taps (fig 4). It is of help treatments. in controlling this turn around if a foot is placed In short, I believe that the best way of against the vertical end of the bath to act as a pivot. furthering clinical rheumatology would be to (5) When finished bathing, first let the bath water and further the distinct specialty of run away; then reverse the procedure by turning separate round on to hands and knees (fig 2), stand up, and medical rehabilitation. D R SWINSON step out. Unit, Rheumatology This way of getting in and out of the bath Wrightington Hospital, will be found to be surprisingly simple. The Wigan, Lancs ease in which it can be done can be demonstrated on any floor, and it is easier to do in the bath. The procedure can be recommended not "Statistics at Square One" only to those classified as handicapped but also to the elderly in general and to many others SIR,-As one who lectures on probability and who, for one reason or another, are less fit or statistics to engineering undergraduates, and agile. I believe that it might with advantage as the husband of a medical practitioner, may be adopted as a routine in hospitals and I congratulate the BMJ for the recently connursing homes. cluded sequence of 20 articles entitled ROBERT G HENDERSON "Statistics at Square One," presented by the Brenchley, Deputy Editor. My experiences over some Tonbridge, Kent years, gained from the teaching (and examining) of the subject and from being asked to handle "real data" from the life sciences, have made me aware of a tendency for investigators, Posts in clinical rheumatology unskilled in statistical techniques, to identify SIR,-I am pleased that Professor W Watson their particular problem with a standard Buchanan and his colleagues (11 September, situation which appears "similar" and carry p 628) have opened a debate on the problems the computation from there. Such attempts

(tr~~~~~~~~~Ae 1'>PW~~:

'4

BRITISH MEDICAL JOURNAL

109

23 OCTOBER 1976

often lead to erroneous results because the basic assumption on which the method is based are not satisfied by the current situation-in other words "a little learning really is a dangerous thing." I would therefore like to make the following points, though they mainly re-echo statements of Dr Swinscow. (1) Researchers ought not to make the mistake of assuming that any set of observations tends to be normally distributed-that is, constitutes a sample from a normally distributed population. Although many medical and biological parameters appear so to behave, other commonly occurring distributions exist. In particular the binomial, multinomial, and Poisson distributions repay study and, given appropriate tables, demand no greater mathematical fluency than the present articles required. It may be mentioned, in passing, that a "goodness of fit" test, based on the x2 distribution, exists for determining whether or not a given set of observed frequencies is in accord with the frequencies expected from a specified theoretical distribution. Some care in use is necessary. (2) Notwithstanding the preceding paragraph, by virtue of a result known as the "central limit theorem," in large samples the comparison of means, the comparison of proportions, and the establishment of confidence intervals may happily be carried out using the test statistic based on the normal distribution, and which Dr Swinscow has dealt with in articles VII, VIII, and IX. If, however, the samples are small, then the above types of calculation demand the use of the t statistic, covered in articles XI, XII, and XIII. These t tests are further subject to the assumptions that the parent population from which a sample is drawn be normally distributed and that when sample means are being compared the variances of the parent populations are equal. The hypothesis that sample variances are not significantly different may be investigated using an F test (and considered, for example, in the book by Professor Armitage'). The division between "small" and "large" samples is commonly held to occur in the range 25-30, for the t distribution with more than 25 degrees of freedom is virtually indistinguishable from the standardised normal variate. Although, as Dr Swinscow has mentioned, the possibility may exist of transforming a nonnormal variate to one nearly normal such technique should be treated with caution. (3) The advent of electronic pocket calculators, whether programmable or not, has made the execution of the majority of the calculations which the articles cover a seemingly trivial matter. However, experience has shown that numerical mistakes are very easy to make-a finger may rest too lightly (or too heavily) on a key, a decimal point be inserted incorrectly, one too few numbers entered in memory, etc, etc. It is therefore wise to record intermediate values wherever practicable, and the final answer should certainly be checked. Dr Swinscow has given the mathematical identity E(x -x)2 = Ex2 - (Ex)2/n, which is useful for calculating purposes. Other results which may be found useful, especially if no calculator is available, are that if x is transformed linearly by x' = ax + 3, where a a represents a change of scale and ± and change of origin, then x'= ao + SD(x') = aI SD(x). The modulus sign, l, is used to ensure that the standard deviation is always positive. Also for the correlation

plotted in some appropriate form-for example, a histogram, frequency polygon, or scatter diagram. Mere "number crunching" of the data plotted in fig 18.1D (18 September, p 681) will yield a sample correlation coefficient in excess of 0-9, yet the graph strongly suggests a curved line (and hence non-linear association). Again, statistical techniques exist to determine the degree of the curve of best fit. In conclusion, referring to the final article on "Correlation" (2 October, p 802), Dr Swinscow gives the formula for the regression equation y = a + bx when y regresses on x. This allows an estimate of the "average" anatomical dead space (y) to be obtained given the height of a child (x). On occasion it may be asked what height of child would correspond to an anatomical dead space of so many millilitres. This "converse" situation requires the calculation of the (linear) regression of x on y. If the equation is written y = c + dx, then y(y_- y)2

Y-(x - ) (y -y

Y

For the given example, d = 1-443, c =-141-68. The closer the points of a scatter diagram lie to a straight line, the closer are these regression lines which interest in (x,y). The line of y regressing on x is the less steep. D MIDDLETON Department of Theoretical Mechanics, University of Nottingham

Armitage, P, Statistical Methods in Medical Research. Oxford, Blackwell Scientific, 1971.

blood to the brain and loss of consciousness. Other examples of ischaemic faint may be seen in the supine hypotensive syndrome of late pregnancy and in the Valsalva manoeuvre. The most convincing evidence against vagal involvement in this form of syncope is the fact that atropine does not prevent it.' The cardiovascular reaction to fear is prevented by the longer-acting benzodiazepines such as lorazepam which block the ability to be afraid.2 Sedative drugs have no effect on the cardiovascular reaction to pain. The cardiovascular reaction to pain or injury is independent of the conscious perception of pain because the reflex arc is complete in the brain-stem below the level of the sensory thalamus.3 Psychotropic drugs such as the natural and the synthetic opiates, which abolish the conscious perception of pain, have no effect on the reflex reaction of the sympathetically innervated blood vessels to traumatic stimuli. The peripheral vascular reaction to pain is prevented by blockade of the sensory afferent nerves from the injured area or by blockade of the sympathetic efferent nerves to the blood vessels at adrenoceptor, ganglionic, or preganglionic level. A more complete account of the work concerning the recognition and the control of the overall cardiovascular reaction to the adrenergic effect of fear and pain, with relevant plethysmographic and electrocardiographic illustration, has been published.4 MICHAEL JOHNSTONE Department of Anaesthetics,

Royal Infirmary, Manchester

"Vasovagal" syncope

SIR,-Dr P Taggart and others, in their report (2 October, p 787) on the reflex cardiac reaction to fear and pain in healthy young adults, suggest that the bradycardia observed in some of them is of parasympathetic origin and that vagal overactivity may be the basis of the syndrome known as "vasovagal" syncope. This implies that atropine should prevent the syndrome though they do not state that atropine should be used for this purpose. As an anaesthetist I am quite certain that atropine not only fails to prevent syncope but increases the hazards of the other forms of cardiovascular collapse which may complicate the induction of anaesthesia in frightened patients -the ventricular tachycardia-fibrillation sequence and myocardial infarction. In reaching the conclusion that vagal overactivity may be the basis of the bradycardia observed in their patients Dr Taggart and his colleagues did not consider the role of the reaction of the peripheral blood vessels to adrenergic stress in conscious subjects. Fear and pain activate the sympathetic nervous system to cause reflex constriction of the alpha-adrenoceptor blood vessels, especially the veins, and reflex dilatation of the betaadrenoceptor blood vessels of the skeletal muscles. Emotional overactivity of sudden onset causes a temporary sequestration of relatively large amounts of blood in the dilated blood vessels of the muscles when the venous return pathways are blocked by the venoconstriction of fear. A failure of the venous return to the heart is precipitated and is increased by gravity when the subjects are standing or sitting in the upright position. As explained by Starling many years ago, the sudden loss of venous return lowers the heart coefficient r(x,y) = r(x',y) . (4) Wherever practicable data should be rate and output, with a consequent lack of

Weissler, A M, et al, Circulation, 1957, 15, 875. Johnstone, M, Anaesthesia, 1976, 31, 868. Ackner, B, Journal of Psychosomatic Research, 1956, 1, 3. 4 Johnstone, M, Anaesthesia Rounds 10: Adult Preoperative Medication. Macclesfield, Imperial Chemical Industries Ltd (Pharmaceuticals Division), 1976. I 2

3

Double negatives and euphemisms

SIR,-For some time I have struggled with the wording on the revised Temporary Resident form (EC 19 Rev 1972), which states: "Not more than 15 days" or "More than 15 days," which surely should be "Less than" or "More than." We are now faced with a similar nonsense on the new sick notes (Form Med 3), which states (a) You need not refrain from work and (b) You should refrain from work, which surely should read (a) Fit for work and (b) Unfit for work. JAMES E MORRIS Stoke,

Plymouth

Dangers of tinted glasses for driving

SIR,-In reply to Dr C E Connolly (21 August, p 478) I have the following comments to make regarding the wearing of photochromic lenses while driving. In fact, the various horrifying tales of what will happen on driving out of bright sunlight into a dark tunnel are no more true of wearers of these lenses than of any other driver. There are two main reasons for this. The most important is that it is the ultraviolet light in ordinary sunlight which makes the lenses go darker. When the wearer is sitting in a car all or about 98% of the ultraviolet light will have been filtered out of the sunlight by the windscreen before it reaches the lenses. In