812 DO PATIENTS SEE VISIONS THAT TALK? are a feature of a variety of abnormal Although the quality of hallucinations in one sense modality have been thoroughly described,l there have been no studies comparing the quality of hallucinations simultaneously occurring in two senses. I have used the present state examination to study simultaneous visual and auditory hallucinations in 30 schizophrenics (DSM-III) and 23 patients with drug-induced psychosis. All patients describedB both visual and auditory hallucinations on admission to hospital and had hallucinations in at least one sense after 10 days. The patients were interviewed on admission and 5 and 10 days after-
wards. 66% of the sample reported simultaneous visual and auditory hallucinations that shared a common theme in which both faces were seen and voices heard. However, only 1 patient, with alcohol-related psychosis, described seeing a face (not mine) that talked to him during the initial interview and which he was unable to recall and failed to describe subsequently. In all other cases the voices originated from within the patient’s head or from outside space but were divorced from the visual percept. Thus, a patient’s description of a man running towards him with a knife would be accompanied by an ethereal voice emanating from elsewhere telling the patient he was about to be killed. Although further studies are required to confirm these findings, it seems most unusual for a patient to experience a fully integrated polysensory hallucination. Feigned psychosis is well known to psychiatrists and can be very difficult to identify. A detailed description of a polysensory hallucination may help. So unshakeable and precise were the patients when challenged about the nature of their experiences that I would question the validity of any account from a patient who reports experiencing a vision that talks. Maudsley Hospital,
MARTIN P. DEAHL*
1. Aggernaes A. The experienced reality of hallucinations and other psychological phenomena. Acta Psychiat Scand 1972; 48: 220-38. *Present address: 804 Cathedral
Maryland 21201, USA.
PANCREATIC GRAFT FAILURE DUE TO PELVIC EXAMINATION
SIR,—The results of pancreatic transplantation for the treatment of insulin-dependent diabetes mellitus have improved considerably and there are now some 400 patients carrying functioning grafts. The vascular anastomoses are usually to the iliac vessels and the graft is placed in the pelvic fossa-a site that may have unfortunate consequences, as shown in this case-report. A 28-year-old woman who had had diabetes since the age of 11 underwent combined renal and pancreatic transplantation in February, 1985. The pancreatic graft was anastomosed to a jejunal Roux-en-Y loop. 12 days after the operation a pancreatic fistula originating from the pancreaticoenteric anastomosis was repaired. The postoperative course was otherwise uneventful and no rejection episodes occurred. The patient was discharged with excellent renal graft and pancreatic graft function. The patient was regularly seen as an outpatient, and at the 1-year follow-up her serum creatinine was 85 pmol/1 and fasting and postprandial blood glucose levels were 3-9 and 4-5 mmol/1, respectively. In May, 1986, she had symptoms of vaginitis and consulted a gynaecologist. The gynaecologist knew the patient’s history but was not aware of the fact that the pancreas would be palpable in the pelvic fossa. On pelvic examination the gynaecologist felt a mysterious mass and this was palpated for some time. The next day the patient was admitted to the transplantation department because of severe pain at the graft site and a raised serum amylase was found. The pain resolved in 2 days but the hyperamylasaemia persisted for 4 weeks. 2 months later endocrine impairment was obvious with abnormally high fasting and postprandial blood glucose levels. At laparotomy the pancreas was found to be white and hard and about half its original size. Biopsy revealed extensive granulocytic inflammation, atrophy, and degeneration of the exocrine tissue.
There was no inflammation within the islets of Langerhans and there was no vasculitis. These findings were consistent with chronic pancreatitis. The patient had to resume exogenous insulin although there were some residual insulin production (fasting serum C-peptide 0-5-0-7 nmol/1; reference value 0-3-1-7). On follow-up 6 months later the patient’s C-peptide levels remain low and she is insulin dependent. It seems very likely that palpation of the pancreatic graft during the pelvic examination induced a severe chronic pancreatitis which eventually led to graft failure. Patients must be told where the graft has been put and instructed to tell other clinicians about the possible danger of manipulating the graft. Department of Transplantation Surgery, Karolinska Institute,
Huddinge Hospital, S-141 86 Huddinge, Sweden
GUNNAR TYDÉN C. G. GROTH
COMBINHD RISK FACTORS AND CORONARY HEART DISEASE
SIR,-Dr Keys (Feb 14, p 382) refers to the intercorrelation between risk factors for coronary heart disease in the MRFIT study, and concludes that the combined effects on serum cholesterol of age, systolic blood pressure, and body mass index could be substantial. Even if the MRFIT study had not made extensive use of multivariate analysis to control confounding variables, this conclusion would not be supported by the quoted data. The proportion of variation in serum cholesterol associated with each of the three named factors is estimated by the corresponding coefficient of determination. Since the coefficient of determination is given by the square of the correlation coefficient, the proportion of variation in cholesterol which is associated with age is 00025, with systolic blood pressure 0-0049, and with body mass index 0-0081. The proportion of variation in serum cholesterol associated with all three factors together could therefore not exceed the sum of these values-ie, 0-0155(1-55 %). This value would only be attained if the three factors were themselves independent, which they clearly are not. No more than a trivial proportion of the effect of serum cholesterol on coronary heart disease risk could possibly be explained on the basis of confounding by age, blood pressure, and body mass index. Department of Family and Community Medicine, Medical School, Newcastle upon Tyne NE2 4HH
* Thisletter has been shown to Dr Keys, whose reply follows.ED. L.
SIR,-I used the word "substantial", where "appreciable" would have been more appropriate, to describe the effects on serum cholesterol of age, blood pressure, and body mass index. Dr Kirkup states that the three factors are not independent. In the Seven Countries study, covering over 12 000 middle-aged men, serum cholesterol was not related to age, and the correlation of serum cholesterol with blood pressure was only r=0-13, that with body mass index being r = 0 20.1 The main point of my letter was to note limitations imposed by the focus on quintile classes of serum cholesterol and lack of consideration of curvilinear relationships in the flood of papers on findings in the MRFIT Study. One result is that most readers, and the general public certainly, will fail to note that over most of the range of serum cholesterol in this material there was little or no relation between serum cholesterol and total mortality. I would be the last to argue against the importance of serum cholesterol; for over 35 years I have been writing and speaking about the need to correct high serum cholesterol. But the MRFIT study, valuable though it may have been, leaves unanswered most of the questions about the way to attack the coronary problem and what can be expected from efforts to control the level of serum cholesterol in middle age and later years. Dr Campbell (Dec 6, p 1331) also objects to the analysis mairùy by quintile classes of serum cholesterol, conclusions that disregard the curvilinear relationships that seem obvious in the MRFIT data.