1246

other than impaired 5-H.T. formation could be responsiblefor the depressive symptoms.

Urinary metabolite levels are end-products reflecting thesum total of metabolic reactions in many tissues. The

precise origin of those compounds mirroring increasedtryptophan metabolism is uncertain, and correlation withbrain concentrations of free tryptophan, the main pre-cursor of 5-H.T., somewhat tenuous. It is also relevant thatDr Adams and his colleagues found raised levels of urinaryxanthurenic acid and other tryptophan metabolites to be afeature of all their patients on oral contraceptives, irrespec-tive of whether or not they were depressed, and irrelevantlyof the evidence of vitamin-Be deficiency or their responseto pyridoxine.

Interpretation of altered cerebral metabolism of 5-H.T.as reflected by concentrations of various urinary metabolitesis therefore extremely difficult.To try and establish which biochemical lesion or lesions

are responsible for pill-induced depression we must

combine the information gained from clinical studies withother types of investigation, and interpret the results inparallel. The nature of altered biochemical processesafter drug administration must be studied where the

change occurs-at organ, tissue, or cellular level. Suchinvestigations pose obvious difficulties; in the main theycannot be done in man but only in animals, and this meansthat quantitative interpretation is not possible and evenqualitative comparison may be tenuous. However, it isuntrue to say, as seems to be fashionable nowaday, thatanimal results cannot be extrapolated to man. They can,provided the investigator has the wisdom and experienceto do so, having displayed equally the knowledge necessaryto pick the right animal for the right experiment in order toget the right information.Money, time, and effort could be saved if in addition to,

and in parallel with, double-blind prospective surveys ofsufficient size and duration for statistical evaluation,studies were undertaken of organ, tissue, and cellular

responses in animal models. This would enable us to

predict long-term effects of chemicals and pollutants inman in a matter of months rather than years.9

Royal Free Hospital,Liverpool Road Branch,

Islington, London N1 0QE.

PETER BEACONSFIELDREBECCA RAINSBURY

JEAN GINSBURG.

MISLEADING TESTS FOR GLYCOSURIA

SiR,—Dr Mayson and others 10 noted that 0-35% of urinesamples tested gave a potentially false-negative test for

glucose. They also concluded that all of the glucose-oxidase paper tests were equally susceptible to false negativetests for glycosuria. Results 11 at our medical centre con-trast sharply with these findings. We investigated theincidence of potentially erroneous tests for glycosuria inurine specimens obtained from 513 patients (313 patientsin hospital and 203 clinic patients). All urine specimenswere tested with ’ Clinitest’ (copper reduction test) andthe glucose-oxidase paper tests (’ Clinistix ’, ’ Diastix ’,and ’ Tes-Tape ’). After the addition of adequate glucoseto achieve a urinary glucose concentration of 0-5%, all urinespecimens were retested with the four test systems. Therewas a 33% incidence of falsely low (<0-5%) results inthe 513 specimens examined. We demonstrated a sufficientquantity of ascorbic and/or gentisic acid (a metabolite ofsalicylate) to account for the abnormal results in 51 % ofthe urine specimens that gave falsely low tests. The falsely

9. Beaconsfield, P. Ann. Ist. Super. San. 1969, 5, 536.10. Mayson, J. S., Schumaker, O., Nakamura, R. M. Lancet, April 7,

1973, p. 780.11. Feldman, J. M., Lebovitz, F. L. Diabetes, 1973, 22, 115.

low tests were noted when testing with clinistix and/ordiastix but not with clinitest or tes-tape. One of the

explanations for the 100% difference in our study and thatof Mayson et also was probably technical. Mayson et al.placed a drop of 0-5% glucose on the test area of thedipstick, while we fortified the urine specimens with

glucose and precisely followed the manufacturer’s in-structions. It is of interest that 23% of the 513 specimensexamined gave falsely high readings (>0’5%). Over-reading had a distinct relationship to dilute urine and wasmost frequent with clinitest (65%). Because of its sensi-

tivity to glucose, specificity, and resistance to under-

reading, tes-tape seems to be the most satisfactorypreparation for routing screening for glycosuria.

Division of Endocrinology,Duke University Medical

Center, Durham,North Carolina 27710, U.S.A.

JEROME M. FELDMANFRANCINE L. LEBOVITZ.

POSITIVE N.B.T. TESTS IN PREGNANCY

SIR,-The nitroblue-tetrazolium (N.B.T.) test has beenoffered as a simple method of determining the cause offever, a positive test suggesting that the fever is of bacterialorigin. However, there have been many reports of falsepositives and false negatives, including the occurrence offalse-positive tests in antenatal patients.12 These wereunrelated to infection, asymptomatic bacteriuria, or stageof pregnancy.The incidence of positive N.B.T. tests in pregnancy has

been studied further, since it was felt that the occurrenceof positive tests might be related to immunological responseswhich occur during pregnancy to the fetal allograft.

Patients attending the antenatal clinic during the monthof February, 1973, who were having blood taken for anyreason, formed the basis of the study. The method usedfor the N.B.T. test was a modification of the one described

by Park et al.13 and Wollman et al. 14 Malachite-green wasused as a counter-stain. A result of 10% positive cellswas taken as the upper limit of normal.

Positive N.B.T. tests were found in 33% of 185 patients.Patients with documented bacterial infection were excluded.The development of a positive N.B.T. test was found to beunrelated to the stage of pregnancy:

There was no difference between multipart and primi-parae. ABO and rhesus blood-groups or sex of the babydid not seem to influence the test.Medical students, roughly matched for age, were used

as controls. There was no difference between males,females, and females taking oral contraceptives. A smallgroup of patients were studied post partum to determinehow long the positive N.B.T. test lasted. No patients werefound to have a positive test 5-8 days post partum.The results show that positive N.B.T. tests occur in

pregnancy. As the positive tests are found at all stages ofpregnancy it is unlikely to be a hormonal phenomenon.

12. Drysdale, H. C. Lancet, 1972, ii, 594.13. Park, B. H., Fikrig, S. M., Smithwick, E. M. ibid. 1968, ii, 532.14. Wollman, M. R., David, D. S., Brennan, B. L., Lewy, J. E. Stengel,

K. H., Rubin, A. L., Miller, D. R. ibid. 1972, ii, 289.