We now understand several aspects of the role of p53 in chemotherapy-induced apoptosis. As basic molecular research focuses more towards manipulating and modulating these pathways, there should be substantial improvements in the prognosis for ovarian cancer patients. Andrew N Shelling Research Centre in Reproductive Medicine, National Women’s Hospital, Auckland, New Zealand 1. Reese DM. Anticancer drugs. Nature 1995; 378: 532. 2. Kristensen GB, Trope C. Epithelial ovarian cancer. Lancet 1997; 349: 113-17. 3. Shelling AN, Cooke IE, Ganesan TS. The genetic analysis of ovarian cancer. Br J Cancer 1995; 72: 521-27. 4. Herod JJO, Eliopoulos AG, Warwick J, Niedobitek G, Young LS, Kerr DJ. The prognostic significance of bcl-2 and p53 expression in ovarian carcinoma. Cancer Res 1996; 56: 2178-84. 5. Buttita F, Marchetti A, Gadducci A, et al. p53 alterations are predictive of chemoresistance and aggressiveness in ovarian carcinomas: a molecular and immunohistochemical study. Br J Cancer 1997; 75: 230-35. 6. Righetti SC, Torre GD, Pilotti S, et al. A comparative study of p53 gene mutations, protein accumulation, and response to cisplatin- based chemotherapy in advanced ovarian carcinoma. Cancer Res 1996; 56: 689-93. 7. Wahl AF, Donaldson KL, Fairchild C, et al. Loss of normal p53 function confers sensitization to taxol by increasing G2/M arrest and apoptosis. Nat Med 1996; 2: 72-79. than older children and adults. 2,6 In a study in Canada, 53% of infected children ≤4 years old had positive stool cultures more than 3 weeks after onset of symptoms, compared with only 8% of older children and adults. 6 In a study among children in day-care centers in Minnesota, the median duration of shedding was 17 days (range 2 to 62), and 38% of children continued to shed the organism for more than 20 days. 2 In the Shah study the exclusion of 12 of 24 patients who met the clinical case-definition (diarrhoea for ≥2 days) from the calculations of the duration of carriage might have led to overestimation of duration of shedding. These 12 patients had negative stool cultures but the samples had not been collected until a median of 22·5 days after symptom onset. Does severity of illness affect duration of shedding? In a study of day-care-associated infections of O157 in Germany, the median duration of shedding was 13 days (range 2 to 62) for patients with diarrhoea or haemorrhagic colitis but 21 days (range 5 to 124) for patients who developed haemolytic uraemic syndrome. 7 However, the study in Minnesota found no correlation with severity of illness. 2 Shedding of O157 can be intermittent, which also complicates efforts to determine the duration of fecal shedding. 2,7 Although future studies are likely to report even longer durations of shedding because of increasing sensitivity of new methods of culturing stools for O157 (such as those using immunomagnetic separation 8) , the public health significance of detecting small numbers of organisms in patients who no longer have symptoms is not know. Does use of antimicrobial therapy affect the duration of fecal carriage of O157? In the Shah study, for the three patients treated with antimicrobial agents, the mean duration of excretion (35·7 days [SD 12·4]) did not differ significantly from that for the nine patients not treated with antimicrobial agents (30·1 days [SD 13·0]). 3 In another study the longest duration of carriage (62 days) occurred in a patient who had received amoxicillin 26 days after illness onset. 2 In another, the longest carriage (71 days) occurred in a symptom-free patient with no history of antimicrobial therapy (C Whitman, unpublished). How can transmission of O157 in child-care centres be limited? Children and staff with diarrhoeal illness should be excluded from the centre until symptoms resolve, whether or not a pathogen is identified. 9 We support the recommendation to exclude pre-school children known to be infected with O157 from child-care centres until two 2,10 or three 7 stool cultures are negative; specimens should not be collected earlier than 48 hours after any antimicrobial therapy has been finished. However, relying on stool cultures to identify cases and dictate control measures is not sufficient. Many persons infected with O157 do not come to the attention of public health authorities; they may have had mild symptoms or non-bloody diarrhoea and not have sought care, may not have had a stool cultured, or the specimen may not have been tested for O157. Routine handwashing before snacks and meals and after diaper changes and toileting is the best way to prevent the spread of infection in child-care centres. Handwashing upon arrival would provide additional protection. 9 During outbreaks, if there is evidence of ongoing transmission after implementation of appropriate control measures, closure of the centre may be Vol 349 • March 15, 1997 745 THE LANCET COMMENTARY Duration of faecal shedding of Escherichia coli O157:H7 among children in day-care centres In the past 15 years Escherichia coli O157:H7 (O157) has emerged as an important cause of bloody and non-bloody diarrhoea in many parts of the world. 1 The sequelae include haemolytic uraemic syndrome and even death. Most infections are acquired through foodborne routes. However, the infectious dose is low and person-to-person transmission is common. Outbreaks of the illness due to person-to-person spread have occurred in schools, long- term care institutions, families, and day-care facilities, 1 and secondary attack rates of up to 22% have been reported. 2 Reducing person-to-person transmission is important in preventing spread of this pathogen. When developing control measures for settings such as day-care centres it is important to consider how long infected persons shed the pathogen. Shah and colleagues have reported that the median duration of shedding (interval from onset of diarrhoea to the first of two negative stool cultures) was 29 days (range 11–57) among 12 culture-positive children in a day-care-associated outbreak; 11 of these children shed the pathogen for ≥3 weeks. 3 This duration of faecal shedding is longer than has previously been reported. During investigation of the first two reported outbreaks of O157:H7 infection in 1982, the organism was not recovered from any of the seven stool samples collected 7 or more days after onset of illness, whereas it was recovered from 9 (75%) of 12 samples collected within 4 days of onset of illness (p=0·002). 4 In Seattle, among patients with haemolytic uraemic syndrome, O157 was cultured from only 33% of stool samples obtained ≥7 days after the onset of diarrhoea, compared with 96% of samples obtained <7 days after onset. 5 However, the duration of faecal shedding varies inversely with age. Young children have been reported to shed O157 for longer periods after resolution of symptoms