Assessment of administrative data for evaluating the shifting acquisition of Clostridium difficile infection in England.

BACKGROUND Little is known about the acquisition of Clostridium difficile infection (CDI) and whether it represents hospital- or community-acquired infection. AIM To test the feasibility and value of using national hospital admissions data from Hospital Episode Statistics to examine trends in CDI in England. METHODS Hospital Episode Statistics from the period 1997/98 to 2009/10 were used. Time trends were analysed using two different denominators of hospital activity: total admissions and total bed-days. We explored the impact of sociodemographic factors, comorbidity and healthcare pathways on the risk of CDI. FINDINGS CDI rates per admission and per bed-days increased from 1997/98 to 2006/07, then decreased significantly by >50% from 2008/9 and 2009/10. This pattern was similar for patients regardless of probable source of infection but the proportion of probable community-acquired CDI cases rose steadily from 7% in 1997/98 to 13% in 2009/10. CDI rates were higher among older patients (odds ratio: >65 years, 10.9), those with more comorbid conditions (odds ratio for Charlson index: >5, 5.6), and among patients admitted as an emergency compared with elective admissions, but no relationship was found with deprivation score. CONCLUSION Our findings support not only the falling trend in CDI found in the national mandatory surveillance scheme from the Health Protection Agency, but a growing proportion of CDI presenting on admission with no evidence of prior hospital exposure in the previous 90 days. We suggest that these may be community-acquired CDI cases.