Sir, Previous studies have identified the use of second- and third-generation cephalosporins as a risk factor for Clos- tridium difficile-associated diarrhoea (CDAD) 1,2 and inter- vention studies which replaced cephalosporins reduced the incidence of CDAD in the elderly. 3,4 At St Mary's Hospital in 1997, 34% of patients aged over 65 years with diarrhoea, who were examined for C. difficile toxin, tested positive compared with 17% of younger patients. As part of an investigation of the reasons for this we performed a case control study on antibiotic use in elderly patients. Data for this study were collected from February to December 1999 in elderly patients (� 65 years) with diar- rhoea admitted to 20 wards of St Mary's Hospital and to one university-affiliated hospital specializing in care of the elderly. All liquid stool specimens sent to the laboratory were tested by EIA (Premier Cytoclone AB) for the presence of C. difficile toxins A and B and cultured for Salmonella, Shigella and Campylobacter spp. on selective media. We defined a case as a patient over 65 years old produc- ing liquid stools with C. difficile detected by EIA. We selected as controls patients aged over 65 years admitted to wards where C. difficile cases occurred, whose faeces were liquid and who tested negative for C. difficile toxins. Excluded from the control group were patients who were admitted with diarrhoea or in whose faeces other bacterial pathogens were identified. Data collected prospectively comprised patient identifiers, ward assignment at onset of diarrhoea, dates of admission and onset of diarrhoea, indi- cation for antibiotics and antibiotics administered within 4 weeks before onset. Data were collected from drug charts and medical records on to standardized forms, then entered into EPI Info version 6 (CDC, Atlanta, GA, USA). Categorical variables were compared by chi-squared test and continuous variables by Kruskal-Wallis test or t-test if normally distributed. A level of significance of P � 0.05 was selected. During the 10 month study period, 64 cases and 64 con- trols were studied. Case and control patients were similar in age, sex and location (Table). Although cases clustered on the intensive care unit (ICU), the renal ward, the vascu- lar ward and three geriatric wards in some months, admis- sion to these wards was not identified as a risk factor for CDAD (Table). Cases stayed on average 10 days longer than controls in hospital before the onset of CDAD. Pro- longed hospital stay increases the risk of acquisition of C. difficile and is usually a marker for severe underlying dis- ease. Respiratory tract infection (RTI) was identified as the most common indication for antimicrobial therapy that was followed by CDAD (odds ratio 3.15, P � 0.01) This condi- tion is treated according to St Mary's Hospital antibiotic policy with cefuroxime plus clarithromycin for 3 days and then switched to co-amoxiclav plus clarithromycin orally for 7 days more. Dispensary records showed that four times more co-amoxiclav than cefuroxime was dispensed in the hospital. Twice the defined daily doses of clarithromycin were dispensed compared with cefuroxime because the former is continued for 10 days.