SIR, - Dr A H Choudhri and colleagues unfortunately give no details of the radiographic projections used to image the renal arteries.' It would seem likely that the renal arteries were imaged only in a single anteroposterior projection. Anatomically the renal arteries arise from the anterior quadrants of the aorta, with the right renal artery usually arising more anteriorly than the left.2 A single conventional anteroposterior projection done in conjunction with an aortic flush will provide inadequate visualisation of the origins of the renal artery, particularly on the right. The recommended projection is the 250 right posterior oblique, which throws both renal arteries, and particularly their origins, into profile.3 It has been our experience of renal artery imaging using digital subtraction angiography that without the use of this oblique projection stenoses may be missed, resulting in the underreporting of proximal lesions. T M BUCKENHAM C D GEORGE Department of Diagnostic Radiology, St George's Hospital, London SW17 OQT I Choudhri AH, Cleland JGF, Rowlands PC, Tran TL, McCarty M, Al-Kutoubi MAO. Unsuspected renal artery stenosis in peripheral vascular disease. BMJ 1990;301:1197-8. (24 November.) 2 Odman P, Ranniger K. The location of the renal arteries: angiographic and postmortem study. Am J Roenigenol
1968;104:283-8. 3 Johnsrude IS, Jackson DC. A practical approach to angiography. Boston: Little, Brown, 1979:122.
AUTHORS' REPLY, - Dr Donovan suggests that serum creatinine concentration is a poor index of glomerular filtration but suggests no better alternative. Measurement of serum creatinine is the most widely used index of glomerular function, is simple, is relatively resistant to major technical artefacts, and is probably at least as good as measuring creatinine clearance using 24 hour urine collections and simultaneous blood sampling. ' Cockcroft and Gault showed that the difference between creatinine clearance estimated from the serum creatinine concentration was no greater than that between paired measurements of creatinine clearance.2 Morgan et al recommended that the measurement of creatinine clearance should be abandoned in favour of measurement of serum creatinine alone over 10 years ago.' Moreover, many patients are poorly instructed in how to perform a urine collection and others are just poorly motivated, making creatinine clearance an even more unreliable estimate of glomerular filtration. Measurement of serum creatinine circumvents this source of error. Inulin clearance or measurements of glomerular filtration using radioisotopes may well be more accurate measures of renal function, but these are not in widespread clinical use. Thus serum creatinine concentration is probably the best method in widespread clinical use for the estimation of global renal function. The purpose of our paper was not to identify specific patients at high risk of renal dysfunction during inhibition of angiotensin converting enzyme but rather to alert the practitioner to a group of patients that contains individuals at high risk who should be assessed in greater detail. Serum creatinine concentration, which we believe to be the best practicable way of assessing glomerular filtration in a large number of subjects in a busy clinical practice, may be normal even in the presence of significant bilateral renal artery stenosis. Thus, a normal serum creatinine concentration should not lull the clinician into a false sense of security. Drs Buckenham and George query renal artery imaging. A single anteroposterior projection was usually obtained as an adjunct to the main procedure, which was of course assessment of the peripheral circulation. If the renal arteries were
236
obscured by other vessels oblique views were obtained. We agree that the single anteroposterior projection tends to underestimate the incidence of renal artery stenosis, but in our experience and that of others4 no single projection is likely to be entirely satisfactory if imagining the renal arteries is part of an aortic injection. Although a right posterior oblique projection may be more likely to show the origins of both renal arteries than an anteroposterior projection, an anteroposterior projection is an acceptable alternative as a useful guide to renal anatomy and for assessing the main renal arteries. We routinely use the anteroposterior projection and both oblique projections when formally examining the renal arteries with an aortic injection. Even so it is sometimes necessary to catheterise the renal artery selectively to obtain unequivocal information. In this group of patients the renal arteries were examined as part of a research project in patients with no known renal artery disease, so it was not thought appropriate to increase risk due to catheter manipulation or exposure to radiation and contrast beyond the minimum. Now that we have shown that there is a significant incidence of renal artery stenosis in this group of patients we think that the procedure could be extended for adequately evaluating the renal arteries. The thrust of our work was to highlight the incidence of renal artery stenosis in these patients; a slight underestimation of the incidence should only underline this point. In reply to Dr D Kerr and Professor R TattersallS we would like to point out that the first patient had glomerular filtration measured one month after stopping an angiotensin converting enzyme inhibitor. In the second patient renal function did not improve after the angiotensin converting enzyme inhibitor was stopped. We have shown that plasma concentrations of angiotensin II and glomerular filtration would be expected to return to normal within a few days of the angiotensin converting enzyme inhibitor being stopped. Although measurement of glomerular filtration by radioisotopes may be more sensitive to changes in glomerular filtration, the technique is affected by technical artefacts that render changes in an individual patient difficult to interpret. Extravasation of isotope from the vein and inadvertent timing of blood sampling may have important effects on the result. These problems are less important when groups of patients are studied because they are randomly distributed and merely increase the "noise" within the study. The changes in glomerular filtration rate reported above are similar to those seen in heart failure in patients without stigmata (including peripheral vascular disease) of renal artery stenosis79 and are consistent with the expected effects of angiotensin II suppression on the kidney.6 No major change in serum creatinine concentration occurred and neither patient required dialysis. Although the findings are interesting, they may not be clinically relevant. A H CHOUDHRI J G F CLELAND P C ROWLANDS T L TRAN M McCARTY M A 0 AL-KUTOUBI
St Mary's Hospital, London W2 INY 1 Wibell L, Bjorsell-Ostling E. Endogenous creatinine clearance in apparently healthy individuals as determined by 24-hour ambulatory urine collection. UppsalaujMedSci 1973;78:43-56. 2 Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1975;16:31. 3 Morgan DB, Dillon S, Payne RB. The assessment of glomerular function: creatinine clearance or plasma creatinine? Postgrad MedJ7 1978;54:302-10. 4 Johnsrude IS, Jackson DC, Dunnick NR. A practical approach to angiography. 2nd ed. Boston: Little, Brown, 1987:259-63. 5 Kerr D, Tattersall R. Renal artery stenosis. BMJ 1991;302: 115. (12 January.) 6 Cleland JG, Dargie HJ. Heart failure, renal function and
angiotensin enzyme inhibitors. Kidtey Itft 1987;31(suppl 201): S220-8. 7 Cleland JGF, Dargie HJ, Hodsman GP, ct al. Captopril in hcart failure. A double blind controlled trial. Br Hleart J 1984;52: 530-5. 8 Cleland JGF, Dargie HJ, Ball SG, et al. Effects of enalapril in heart failure: a dotublc blind study of effects on exercise performance, renal ftunction, hormones, and metabolic state. Br Heart] 19X8554:305-12. 9 Cleland JGF, Gillen G, Dargie HJ. The effects of frusemide and angiotensin-converting enzyme inhibitors and their combination on cardiac and renal haemodynamics in heart failure. EurHeartj 1988;9:132-41.
Bleeding tonsils SIR,-Mr Mark G Watson recently drew attention to the problem of spontaneous tonsillar haemorrhage.' We report a further cause of this condition. A previously fit 20 year old man presented having been spitting blood for two days. His history was unremarkable, but examination showed a bleeding point on the lower left faucial tonsil. He had a blood pressure of 240/140 mm Hg in both arms with no radiofemoral delay; a displaced heaving apex beat; and severe arteriolar narrowing and twisting in his fundi. Chest radiography showed cardiomegaly, and left ventricular hypertrophy and strain were evident on electrocardiography. Full blood count, clotting studies, biochemical profile, and urine analysis all gave normal results. Within two hours after the start of intravenous sodium nitroprusside the bleeding had stopped, and within 12 hours his blood pressure was 160/90 mm Hg. Further investigations failed to identify any underlying cause for his hypertension and he was discharged four days later. He continued to take atenolol and slow release nifedipine and his blood pressure remained normal; he has had no further tonsillar haemorrhages. Haemorrhage is a recognised presentation of severe hypertension; cerebrovascular accidents are common, the severity of epistaxes is thought to be related to blood pressure,2 and frank haematuria and haematospermia have been reported.3 Our case shows that severe hypertension can also present as spontaneous tonsillar haemorrhage. G E MURTY N J SAMANI JR MOLONEY Leicester Royal Infirmary, Leicester LEI 5WW 1 Watson MG. Bleeding tonsils. BMJ 1990;301:1233-4. (1 December.) 2 Shaheen OH. Arterial epistasis. J Laryngol Otol 1975;89: 17-34. 3 Swales JD. Clinical hypertension. London: Chapman and Hall, 1979:137-8.
Hospital admissions for accidents in preschool children SIR,-In Solihull District Health Authority we examined our hospital inpatient data to see if admissions of children for accidents had any epidemiological patterns. We used data from the regional information system for all Solihull residents during the year April 1989 to March 1990. I excluded admissions for birth (ICD 9 codes V300 and V290) when calculating admission rates and defined admissions as finished consultant episodes. (This analysis could not have been done without the diligent data entry by the unsung hero(ine)s of NHS statistics: the coding clerks.) The distribution of ages of children admitted in relation to accidents formed a shallow U shape. Boys were admitted more often than girls in all age groups, and the rise in both teenage and preschool accidents was predominantly among boys. When considering rates of diagnosis by total population it is important to realise that small numbers may affect any comparison. Nevertheless,
BMJ VOLUME 302
26 JANUARY 1991
the patterns in Solihull children by sex and age were similar to those described by Dr Sellars and colleagues. ' I also considered rates using total numbers of admissions not related to births. This is an indicator of workload rather than incidence. When rates per 1000 admissions were compared with rates per 1000 population accident related admission rates for the youngest age group were remarkably low when compared with rates per total population. Also, similar patterns seemed to exist for children more than 12 months old when either total population or total non-birth related admissions were used as a denominator. The relatively high rates for all injuries were maintained throughout all age groups, as were the peaks for burns and poisoning. This suggests that the denominator used in child accident related admissions -either rates per head of population or per 1000 total admissions-is
arbitrary. Pre-Korner data on inpatient admissions held in the Hospital Activity Analysis system were assigned to ward of residence. This was useful for small area statistics as estimates of ward population are available from the Office of Population Censuses and Surveys. When the Korner data sets were implemented the postcode of residence became the only small area geographical field available in the regional information system. (Postcode sectors are roughly the same size as wards.) Until the next census, the only way of obtaining postcode sector population data is from the family health services authorities. This depends on the capabilities of the computing staff and availability of software. From my limited experience, these data are impossible to obtain. Bearing this in mind, my data suggest that population rates could be replaced by rates per 1000 admissions not related to births when population data are not easily available. In addition, there may be marker inpatient conditions that could be used as a proxy for total population. I would be interested to know if colleagues have discovered such marker conditions for particular age groups and diagnoses. K S SIDHU
Department of Public Health Medicine, Solihull Health Authority, Solihull B91 3AH 1 Sellars C, Ferguson JA, Goldacre MJ. Occurrence arkd repetition of hospital admissions for accidents in preschool children. BMJ 1991;302:16-9. (5 January.)
Causes of fatal childhood accidents SIR,-We agree with Dr P M Sharples and colleagues that road traffic accidents are the commonest cause of fatal head injuries in childreh.' The distribution of causes, however, is quite different for injuries of different severity.2 In children with head injuries who attend accident departments road accidents are much less common than falls,' but they are more common in children who are admitted to hospital; they dominate only in the minority of injuries that are severe or fatal (table). Dr Sharples and colleagues report a death rate of 5 3 per 100 000 children per
year compared with our report of 4011 per 100 000 attending accident and emergency departments and 400 per 100000 admitted to hospital in Scotland.4 We have noted the similarity between head injuries in hospitals in Scotland and in Cleveland.2 Detailed neuropathology shows different types of brain damage in children and adults, and this also varies with cause of injury.' Intracranial haematomas, the main cause of avoidable death, are only half as common in fatally injured children as in adults. In both age groups intracranial haematomas are only half as common after road accidents as they are after falls or assaults. In examining the problem of potentially preventable death and disability due to head injury in children it is important not to overemphasise the importance of road accidents. BRYAN JENNETT LILIAN MURRAY J H ADAMS STEWART CULLY Institute of Neurological Sciences, Glasgow G5 1 4TF 1 Sharples PM, Storey A, Aynsley-Green A, Eyre JA. Causes of fatal childhood accidents involving head injuries in Northern region, 1979-86. BMJ 1990;301:1193-7. (24 November.) 2 Jennett B, MacMillan R. Epidemiology of head injury. BMJ
1981;282:101-4. 3 Jennett B, Teasdale G. Management of head injuries. Philadelphia: F A Davis, 1981. 4 Brookes M, MacMillan R, Cully S, et al. Head injuries in accident and emergency departments. How different are children from
adults?J7 Epidemnol Community Health 1990;44:147-51. 5 Adams JH, Doyle D, Ford I, et al. Brain damage in fatal nonmissile head injury in relation to age and type of injury. Scott Med3 1989;34:399-401.
An integrated child health
another large group of children-preschool, of school age, and at adolescence-for whom parents, teachers, social workers, and others are looking for the support of doctors who understand the links between health, family dynamics, and developmental and educational progress; who offer continuity of service as the named doctor of the local clinics and schools; and who are intimate with and have the confidence of local caring networks. The children so supported rarely, if ever, have continuity of contact with hospital departments. Where they do the true nature of their difficulties cannot be illuminated by blood tests and body scans. Indeed there is a vogue for the parents of such children to return from hospital visits armed with special diets to explain a range of symptoms, but the internal dynamics between parents, children, schools, and society have never been questioned. The support of such children requires dialogue, partnership, and participation with nurseries, schools, and parents to enhance nurture-not merely the transfer of information. Moreover, such doctors who think and act for children in a community setting do this not just on an individual family basis but in terms of a geographical area in service planning, development, training, audit, and the public health function. Senior paediatricians who view the integrated service merely from the needs of the sick child will miss the whole dimension of developmental, educational, and social paediatrics. Community colleagues already have an unequal voice, because the consultant status envisaged in the courts to aid integration was never realised. The hostility lies in the narrowness of view of hospital paediatricians and the lack of appropriate status for those who give dedicated clinical service and lead to the community child health services.
service
SONYA LEFF
SIR,-Drs D M B Hall and M Prendergast write about the "lack of commitment to" or even "frank hostility" toward integration of child health services.' Perhaps it would be helpful to ask why this should be so? Indeed, it could be argued that the very editorial which has been published itself exacerbates the frank hostility. The team identity elucidated by Drs Hall and Prendergast covers the need for an integrated approach to the admission, management, discharge, and care at home of sick children. They say that hospital doctors themselves, who may see children from more than 100 schools, should directly transfer information to local teachers, social workers, and others. Health visitors are described as "an essential component of primary care services for children," but there is only a throw away reference to community child health doctors, acknowledging their skills in child development, behavioural problems, and child abuse. It needs to be fully recognised that child health community doctors are not employed just to do surveillance checks not covered by general practitioners, nor just for back up work with children under S referred by general practitioners and health visitors. Nor are they only employed in schools to support statemented, fostered and adopted, and protection registered children, important as all this clinical work is. There is
School Clinic, Brighton BN2 2RA 1 Hall DMB, Prendergast M. An integrated child health service.
BMJ 1990;301:1341-2. (15 December.)
Flat feet in children SIR,-Mr G K Rose' and Dr E Ann Welton' assert that early identification of flat feet is essential. Function is more important than form, however, and many people diagnosed as having flat feet have no disability. The diagnosis of flat feet will change in many cases when the simple test described by Dr Welton is used.3 In a recent survey of an unselected group of 18 month old children I found that 14 out of 221 had flat feet according to the great toe extension test.3 Before this test was applied 93% of the children had been diagnosed as having flat feet. I wonder how many young people and adults labelled as having flat feet have acquired social disease. I am sure that most of them have normal function. The Canadian study quoted by Mr Rose4 has an obvious selection bias and cannot be used as scientific proof. The time has come for a prospective trial to see whether children with flat feet become adults with flat feet and whether they have anv disability. This might blow away some of the myths surrounding the subject.
Main causes oj head inlunies in children aged under 15
MELVYN H BROOKS
Karkur,
Cause of injury (No (%))
Children with head injuries*
Attending Scottish accident and emergency departments Admitted to Scottish hospitals Transferred to Glasgow neurosurgical unit With severe injuries in Glasgow neurosurgical unit Who died in Glasgow neurosurgical unit
Total No of children
Road traffic accident
2118 351 295 333 122
191 (9) 81 (23) 127 (43) 250 (75) 87 (71)
*Data for each group of children gathered independently over varying periods.
BMJ
VOLUME 302
26 JANUARY 1991
Israel
Fall
Assault
1207 (57) 193 (55)
127(6) 18 (5) 12 (4) 10 (3) 2 (2)
83 (28) 40 (12) 27 (22)
Rose GK. Flat feet in children. BMJ 1990;301:1330-1. (8 December.) 2 Welton EA. Flat feet in children. BMJ 1990;301:1331. (8 December.) 3 Rose GK, Welton EA, Marshal T. The diagnosisof flat foot in the child. JBoneJointSurg[Br] 1985;67:71-8. 4 Harris RI, Beath T. Army foot survey. An investigation of foot ailments in Canadian soldiers. Ottawa: National Research Council of Canada, 1947.
237
1968;104:283-8. 3 Johnsrude IS, Jackson DC. A practical approach to angiography. Boston: Little, Brown, 1979:122.
AUTHORS' REPLY, - Dr Donovan suggests that serum creatinine concentration is a poor index of glomerular filtration but suggests no better alternative. Measurement of serum creatinine is the most widely used index of glomerular function, is simple, is relatively resistant to major technical artefacts, and is probably at least as good as measuring creatinine clearance using 24 hour urine collections and simultaneous blood sampling. ' Cockcroft and Gault showed that the difference between creatinine clearance estimated from the serum creatinine concentration was no greater than that between paired measurements of creatinine clearance.2 Morgan et al recommended that the measurement of creatinine clearance should be abandoned in favour of measurement of serum creatinine alone over 10 years ago.' Moreover, many patients are poorly instructed in how to perform a urine collection and others are just poorly motivated, making creatinine clearance an even more unreliable estimate of glomerular filtration. Measurement of serum creatinine circumvents this source of error. Inulin clearance or measurements of glomerular filtration using radioisotopes may well be more accurate measures of renal function, but these are not in widespread clinical use. Thus serum creatinine concentration is probably the best method in widespread clinical use for the estimation of global renal function. The purpose of our paper was not to identify specific patients at high risk of renal dysfunction during inhibition of angiotensin converting enzyme but rather to alert the practitioner to a group of patients that contains individuals at high risk who should be assessed in greater detail. Serum creatinine concentration, which we believe to be the best practicable way of assessing glomerular filtration in a large number of subjects in a busy clinical practice, may be normal even in the presence of significant bilateral renal artery stenosis. Thus, a normal serum creatinine concentration should not lull the clinician into a false sense of security. Drs Buckenham and George query renal artery imaging. A single anteroposterior projection was usually obtained as an adjunct to the main procedure, which was of course assessment of the peripheral circulation. If the renal arteries were
236
obscured by other vessels oblique views were obtained. We agree that the single anteroposterior projection tends to underestimate the incidence of renal artery stenosis, but in our experience and that of others4 no single projection is likely to be entirely satisfactory if imagining the renal arteries is part of an aortic injection. Although a right posterior oblique projection may be more likely to show the origins of both renal arteries than an anteroposterior projection, an anteroposterior projection is an acceptable alternative as a useful guide to renal anatomy and for assessing the main renal arteries. We routinely use the anteroposterior projection and both oblique projections when formally examining the renal arteries with an aortic injection. Even so it is sometimes necessary to catheterise the renal artery selectively to obtain unequivocal information. In this group of patients the renal arteries were examined as part of a research project in patients with no known renal artery disease, so it was not thought appropriate to increase risk due to catheter manipulation or exposure to radiation and contrast beyond the minimum. Now that we have shown that there is a significant incidence of renal artery stenosis in this group of patients we think that the procedure could be extended for adequately evaluating the renal arteries. The thrust of our work was to highlight the incidence of renal artery stenosis in these patients; a slight underestimation of the incidence should only underline this point. In reply to Dr D Kerr and Professor R TattersallS we would like to point out that the first patient had glomerular filtration measured one month after stopping an angiotensin converting enzyme inhibitor. In the second patient renal function did not improve after the angiotensin converting enzyme inhibitor was stopped. We have shown that plasma concentrations of angiotensin II and glomerular filtration would be expected to return to normal within a few days of the angiotensin converting enzyme inhibitor being stopped. Although measurement of glomerular filtration by radioisotopes may be more sensitive to changes in glomerular filtration, the technique is affected by technical artefacts that render changes in an individual patient difficult to interpret. Extravasation of isotope from the vein and inadvertent timing of blood sampling may have important effects on the result. These problems are less important when groups of patients are studied because they are randomly distributed and merely increase the "noise" within the study. The changes in glomerular filtration rate reported above are similar to those seen in heart failure in patients without stigmata (including peripheral vascular disease) of renal artery stenosis79 and are consistent with the expected effects of angiotensin II suppression on the kidney.6 No major change in serum creatinine concentration occurred and neither patient required dialysis. Although the findings are interesting, they may not be clinically relevant. A H CHOUDHRI J G F CLELAND P C ROWLANDS T L TRAN M McCARTY M A 0 AL-KUTOUBI
St Mary's Hospital, London W2 INY 1 Wibell L, Bjorsell-Ostling E. Endogenous creatinine clearance in apparently healthy individuals as determined by 24-hour ambulatory urine collection. UppsalaujMedSci 1973;78:43-56. 2 Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1975;16:31. 3 Morgan DB, Dillon S, Payne RB. The assessment of glomerular function: creatinine clearance or plasma creatinine? Postgrad MedJ7 1978;54:302-10. 4 Johnsrude IS, Jackson DC, Dunnick NR. A practical approach to angiography. 2nd ed. Boston: Little, Brown, 1987:259-63. 5 Kerr D, Tattersall R. Renal artery stenosis. BMJ 1991;302: 115. (12 January.) 6 Cleland JG, Dargie HJ. Heart failure, renal function and
angiotensin enzyme inhibitors. Kidtey Itft 1987;31(suppl 201): S220-8. 7 Cleland JGF, Dargie HJ, Hodsman GP, ct al. Captopril in hcart failure. A double blind controlled trial. Br Hleart J 1984;52: 530-5. 8 Cleland JGF, Dargie HJ, Ball SG, et al. Effects of enalapril in heart failure: a dotublc blind study of effects on exercise performance, renal ftunction, hormones, and metabolic state. Br Heart] 19X8554:305-12. 9 Cleland JGF, Gillen G, Dargie HJ. The effects of frusemide and angiotensin-converting enzyme inhibitors and their combination on cardiac and renal haemodynamics in heart failure. EurHeartj 1988;9:132-41.
Bleeding tonsils SIR,-Mr Mark G Watson recently drew attention to the problem of spontaneous tonsillar haemorrhage.' We report a further cause of this condition. A previously fit 20 year old man presented having been spitting blood for two days. His history was unremarkable, but examination showed a bleeding point on the lower left faucial tonsil. He had a blood pressure of 240/140 mm Hg in both arms with no radiofemoral delay; a displaced heaving apex beat; and severe arteriolar narrowing and twisting in his fundi. Chest radiography showed cardiomegaly, and left ventricular hypertrophy and strain were evident on electrocardiography. Full blood count, clotting studies, biochemical profile, and urine analysis all gave normal results. Within two hours after the start of intravenous sodium nitroprusside the bleeding had stopped, and within 12 hours his blood pressure was 160/90 mm Hg. Further investigations failed to identify any underlying cause for his hypertension and he was discharged four days later. He continued to take atenolol and slow release nifedipine and his blood pressure remained normal; he has had no further tonsillar haemorrhages. Haemorrhage is a recognised presentation of severe hypertension; cerebrovascular accidents are common, the severity of epistaxes is thought to be related to blood pressure,2 and frank haematuria and haematospermia have been reported.3 Our case shows that severe hypertension can also present as spontaneous tonsillar haemorrhage. G E MURTY N J SAMANI JR MOLONEY Leicester Royal Infirmary, Leicester LEI 5WW 1 Watson MG. Bleeding tonsils. BMJ 1990;301:1233-4. (1 December.) 2 Shaheen OH. Arterial epistasis. J Laryngol Otol 1975;89: 17-34. 3 Swales JD. Clinical hypertension. London: Chapman and Hall, 1979:137-8.
Hospital admissions for accidents in preschool children SIR,-In Solihull District Health Authority we examined our hospital inpatient data to see if admissions of children for accidents had any epidemiological patterns. We used data from the regional information system for all Solihull residents during the year April 1989 to March 1990. I excluded admissions for birth (ICD 9 codes V300 and V290) when calculating admission rates and defined admissions as finished consultant episodes. (This analysis could not have been done without the diligent data entry by the unsung hero(ine)s of NHS statistics: the coding clerks.) The distribution of ages of children admitted in relation to accidents formed a shallow U shape. Boys were admitted more often than girls in all age groups, and the rise in both teenage and preschool accidents was predominantly among boys. When considering rates of diagnosis by total population it is important to realise that small numbers may affect any comparison. Nevertheless,
BMJ VOLUME 302
26 JANUARY 1991
the patterns in Solihull children by sex and age were similar to those described by Dr Sellars and colleagues. ' I also considered rates using total numbers of admissions not related to births. This is an indicator of workload rather than incidence. When rates per 1000 admissions were compared with rates per 1000 population accident related admission rates for the youngest age group were remarkably low when compared with rates per total population. Also, similar patterns seemed to exist for children more than 12 months old when either total population or total non-birth related admissions were used as a denominator. The relatively high rates for all injuries were maintained throughout all age groups, as were the peaks for burns and poisoning. This suggests that the denominator used in child accident related admissions -either rates per head of population or per 1000 total admissions-is
arbitrary. Pre-Korner data on inpatient admissions held in the Hospital Activity Analysis system were assigned to ward of residence. This was useful for small area statistics as estimates of ward population are available from the Office of Population Censuses and Surveys. When the Korner data sets were implemented the postcode of residence became the only small area geographical field available in the regional information system. (Postcode sectors are roughly the same size as wards.) Until the next census, the only way of obtaining postcode sector population data is from the family health services authorities. This depends on the capabilities of the computing staff and availability of software. From my limited experience, these data are impossible to obtain. Bearing this in mind, my data suggest that population rates could be replaced by rates per 1000 admissions not related to births when population data are not easily available. In addition, there may be marker inpatient conditions that could be used as a proxy for total population. I would be interested to know if colleagues have discovered such marker conditions for particular age groups and diagnoses. K S SIDHU
Department of Public Health Medicine, Solihull Health Authority, Solihull B91 3AH 1 Sellars C, Ferguson JA, Goldacre MJ. Occurrence arkd repetition of hospital admissions for accidents in preschool children. BMJ 1991;302:16-9. (5 January.)
Causes of fatal childhood accidents SIR,-We agree with Dr P M Sharples and colleagues that road traffic accidents are the commonest cause of fatal head injuries in childreh.' The distribution of causes, however, is quite different for injuries of different severity.2 In children with head injuries who attend accident departments road accidents are much less common than falls,' but they are more common in children who are admitted to hospital; they dominate only in the minority of injuries that are severe or fatal (table). Dr Sharples and colleagues report a death rate of 5 3 per 100 000 children per
year compared with our report of 4011 per 100 000 attending accident and emergency departments and 400 per 100000 admitted to hospital in Scotland.4 We have noted the similarity between head injuries in hospitals in Scotland and in Cleveland.2 Detailed neuropathology shows different types of brain damage in children and adults, and this also varies with cause of injury.' Intracranial haematomas, the main cause of avoidable death, are only half as common in fatally injured children as in adults. In both age groups intracranial haematomas are only half as common after road accidents as they are after falls or assaults. In examining the problem of potentially preventable death and disability due to head injury in children it is important not to overemphasise the importance of road accidents. BRYAN JENNETT LILIAN MURRAY J H ADAMS STEWART CULLY Institute of Neurological Sciences, Glasgow G5 1 4TF 1 Sharples PM, Storey A, Aynsley-Green A, Eyre JA. Causes of fatal childhood accidents involving head injuries in Northern region, 1979-86. BMJ 1990;301:1193-7. (24 November.) 2 Jennett B, MacMillan R. Epidemiology of head injury. BMJ
1981;282:101-4. 3 Jennett B, Teasdale G. Management of head injuries. Philadelphia: F A Davis, 1981. 4 Brookes M, MacMillan R, Cully S, et al. Head injuries in accident and emergency departments. How different are children from
adults?J7 Epidemnol Community Health 1990;44:147-51. 5 Adams JH, Doyle D, Ford I, et al. Brain damage in fatal nonmissile head injury in relation to age and type of injury. Scott Med3 1989;34:399-401.
An integrated child health
another large group of children-preschool, of school age, and at adolescence-for whom parents, teachers, social workers, and others are looking for the support of doctors who understand the links between health, family dynamics, and developmental and educational progress; who offer continuity of service as the named doctor of the local clinics and schools; and who are intimate with and have the confidence of local caring networks. The children so supported rarely, if ever, have continuity of contact with hospital departments. Where they do the true nature of their difficulties cannot be illuminated by blood tests and body scans. Indeed there is a vogue for the parents of such children to return from hospital visits armed with special diets to explain a range of symptoms, but the internal dynamics between parents, children, schools, and society have never been questioned. The support of such children requires dialogue, partnership, and participation with nurseries, schools, and parents to enhance nurture-not merely the transfer of information. Moreover, such doctors who think and act for children in a community setting do this not just on an individual family basis but in terms of a geographical area in service planning, development, training, audit, and the public health function. Senior paediatricians who view the integrated service merely from the needs of the sick child will miss the whole dimension of developmental, educational, and social paediatrics. Community colleagues already have an unequal voice, because the consultant status envisaged in the courts to aid integration was never realised. The hostility lies in the narrowness of view of hospital paediatricians and the lack of appropriate status for those who give dedicated clinical service and lead to the community child health services.
service
SONYA LEFF
SIR,-Drs D M B Hall and M Prendergast write about the "lack of commitment to" or even "frank hostility" toward integration of child health services.' Perhaps it would be helpful to ask why this should be so? Indeed, it could be argued that the very editorial which has been published itself exacerbates the frank hostility. The team identity elucidated by Drs Hall and Prendergast covers the need for an integrated approach to the admission, management, discharge, and care at home of sick children. They say that hospital doctors themselves, who may see children from more than 100 schools, should directly transfer information to local teachers, social workers, and others. Health visitors are described as "an essential component of primary care services for children," but there is only a throw away reference to community child health doctors, acknowledging their skills in child development, behavioural problems, and child abuse. It needs to be fully recognised that child health community doctors are not employed just to do surveillance checks not covered by general practitioners, nor just for back up work with children under S referred by general practitioners and health visitors. Nor are they only employed in schools to support statemented, fostered and adopted, and protection registered children, important as all this clinical work is. There is
School Clinic, Brighton BN2 2RA 1 Hall DMB, Prendergast M. An integrated child health service.
BMJ 1990;301:1341-2. (15 December.)
Flat feet in children SIR,-Mr G K Rose' and Dr E Ann Welton' assert that early identification of flat feet is essential. Function is more important than form, however, and many people diagnosed as having flat feet have no disability. The diagnosis of flat feet will change in many cases when the simple test described by Dr Welton is used.3 In a recent survey of an unselected group of 18 month old children I found that 14 out of 221 had flat feet according to the great toe extension test.3 Before this test was applied 93% of the children had been diagnosed as having flat feet. I wonder how many young people and adults labelled as having flat feet have acquired social disease. I am sure that most of them have normal function. The Canadian study quoted by Mr Rose4 has an obvious selection bias and cannot be used as scientific proof. The time has come for a prospective trial to see whether children with flat feet become adults with flat feet and whether they have anv disability. This might blow away some of the myths surrounding the subject.
Main causes oj head inlunies in children aged under 15
MELVYN H BROOKS
Karkur,
Cause of injury (No (%))
Children with head injuries*
Attending Scottish accident and emergency departments Admitted to Scottish hospitals Transferred to Glasgow neurosurgical unit With severe injuries in Glasgow neurosurgical unit Who died in Glasgow neurosurgical unit
Total No of children
Road traffic accident
2118 351 295 333 122
191 (9) 81 (23) 127 (43) 250 (75) 87 (71)
*Data for each group of children gathered independently over varying periods.
BMJ
VOLUME 302
26 JANUARY 1991
Israel
Fall
Assault
1207 (57) 193 (55)
127(6) 18 (5) 12 (4) 10 (3) 2 (2)
83 (28) 40 (12) 27 (22)
Rose GK. Flat feet in children. BMJ 1990;301:1330-1. (8 December.) 2 Welton EA. Flat feet in children. BMJ 1990;301:1331. (8 December.) 3 Rose GK, Welton EA, Marshal T. The diagnosisof flat foot in the child. JBoneJointSurg[Br] 1985;67:71-8. 4 Harris RI, Beath T. Army foot survey. An investigation of foot ailments in Canadian soldiers. Ottawa: National Research Council of Canada, 1947.
237
