T H E LANCET
infection control teams in the UK. This is not a logical approach if, as is accepted in the commentary, stringent control measures are required if an outbreak of MRSA is occurring. Many hospitals in the U K d o not have such a high prevalence rate as that in the USA, and their infection control teams are putting a great deal of their resources and time into MRSA control to prevent a major outbreak. It is our experience that MRSA will spread on acute care wards unless patients colonised or infected with MRSA are isolated and a major outbreak will occur especially if dealing with epidemic strains. At a time when many infection control teams are striving to convince managers that the effort and expense of MRSA controls is worthwhile it is a cause of great concern that an approach applicable to hospitals with very high prevalence rates will be quoted inappropriately as a reasonable strategy for most U K hospitals. Simon F Hill Public Health Laboratory, Poole Hospital, Poole BH15 2JB. U K 1 Evans Patterson J. Making real sense of MRSA. Lancet 1996; 348: 836. 2 Panlilo AL, Culver DH, Gaynes RP,et al. Methicillin resistant Staphylococcus aureus in US hospitals, 1975-1991. Infect Contr Hosp Epidemiol 1992; 13: 582.
SIR-htterSOn’ raises a n important issue in questioning control measures for MRSA infections. Certainly MRSA in chronic care settings is endemic without causing great problems and a more relaxed approach is justified in these institutions. Unfortunately advising such an attitude in acute care settings is impractical. We do not disagree that MRSA strains in some settings can be diverse, but this is not the usual situation. In the U K there are three predominant strains of MRSA (designated EMRSA 3, 15 and 16).2 In our and other U K institutions over 90% of MRSAs are EMRSA 15 or 16. Whether infections due to these strains are sporadic or part of an outbreak becomes a matter of semantics. I n any event it is rarely possible to identify sporadic cases and, therefore, to have a different approach for sporadic cases is not realistic. Furthermore, even if strains were distinct, there would be a delay of a few days before information on molecular or phage typing would be available, and infection control practices would have to be implemented before this information was available. T o avoid cross infection, control procedures have to be implemented as if these strains were epidemic. If strains are distinct, one can never exclude the possibility that a patient may be harbouring a strain which, although distinguishable from known epidemic strains, nevertheless has epidemic potential. For the above reasons we endorse the recommendations of the Hospital Infection Society’ and the World Health Organization‘ as well as the US consensus view.’ Patients identified as being carriers of MRSA or those with MRSA infections should be isolated in single rooms. * A P Fraise, R Wise City Hospital NHS Trust, Birmingham 818 7QH. UK
Patterson JF. Making real sense of MRSA. Lancer 1996; 348: 836-37. PHLS. Epidemic methicillin resistant Staphylococcus aureus. Comnlrrtr Dis Rep 1996; 6: 197. 3 Combined Working Party of the Hospital Infection Society and British Society of Antimicrobial Chemotherapy. J Hasp Infect 1990; 16: 351-77. 4 Ayliffe GAJ. Recommendations for the control of methicillin resistant Staphylococcus aureus. Geneva: World Health Organization, 1996. 5 Mulligan ME, Murray-Leisure KA, Ribner BS, et al. Methicillin resistant Staphylocuccs aureus: a consensus review of the microbiology, pathogenesis, and epidemiology with implications for prevention and management. A m J M e d 1993; 94: 313-28.
Serum total homocysteine concentration and risk of stroke SIR-In our nested case-control study,’ based on the British Regional Heart Study cohort, men with pre-existing coronary heart disease (CHD) were unintentionally excluded from control selection in 15 of the 17 study towns. This deficit of men with evidence of pre-existing C H D disease in the control group (5% zu 21% expected) has inflated the overall case-control difference in serum homocysteine (tHcy) concentrations. However, the association between tHcy and risk of stroke persists in analyses restricted to cases and controls who were without evidence of C H D at screening. In the original paper, we reported that levels of tHcy (geometric mean, 95% CI) were significantly higher in cases (n=107) than in controls (118): 13.7 (12.7-14.8) versus 11.9 (1 1.3-12.6) pmol/L; p=0.004. The odds ratio for stroke was significantly increased in the fourth quarter of homocysteine relative to the first, odds ratio 2.8 (95% CI 1.3-5.9), with odds ratios of 1.3 and 1.9 in the second and third quarters, respectively. In a n analysis restricted to men without evidence of C H D at baseline (67 cases and 112 controls), levels of tHcy were also significantly higher in cases than controls: 13.1 (12.1-14.2) versus 11.8 (11.2-12.4) pmol/L; p=0.04. T h e odds ratio for stroke was also significantly increased in the fourth quarter of homocysteine relative to the first, odds ratio 2.5 (1.1-6.1), with odds ratios of 1.6 and 1.4 in the second and third quarters, respectively. * I J Perry, on behalf of all authors of original report Department of Primary Care and Population Science, Royal Free Hospital School of Medicine. London NW3 2PF. UK
1 Perry IJ, Refsum H, Morris RW, Ebrahim SB, Ueland PM, Shaper AG. Prospective study of serum total homocysteine concentration and risk of stroke in middle-aged British men. Laircet 1995; 346: 1395-98.
Paediatric HIV infection SIR-Scarlatti’s comment (Sept 28, p 866)’ that “children rarely develop Kaposi’s sarcoma and other HIV-associated tumours” has prompted us to share some preliminary observations. T h e AIDS epidemic is of recent onset in South Africa, and the first symptomatic children were recognised from about 1989 at the King Edward VIII Hospital in Durban, which is one of the largest such facilities in southern Africa, and the only tertiary referral and teaching centre in KwaZulu Natal, serving a paediatric population of about 3-4 million (89% of whom are Black). T h e current antenatal HIV prevalence in black mothers at this institution is 21-24%, and on the basis of a vertical transmission rate of about 34%’ we estimate that about 8% of black infants in KwaZulu Natal are HIV infected. This province is the epicentre of the evolving HIV epidemic in South Africa. Records at our oncology clinic for the past 15 years’ show the rapid impact of HIV/AIDS on the overall pattern of childhood cancer. T h e harbinger of this change was a patient with acute promyelocytic leukaemia diagnosed in 1990. Up to July, 1996, we have diagnosed 14 patients with HIVassociated malignant diseases-12 of these in the past 15 months. Patients with solid tumours and brain tumours have not been included. T h e diagnoses were: Kaposi sarcoma (4); acute leukaemia (2); Burkitt lymphoma (2); T-cell lymphoma (2); mucosa-associated lymphoid tissue (MALT) lymphoma (3) (all with parotid involvement); and an undifferentiated abdominal tumour (1). The most striking
Vol348 November 30, 1996