[Two of the authors respond:]
We were aware that the primary discharge diagnosis of Clostridium difficile infection may not have been the admitting diagnosis, but we estimated the magnitude of such miscoding in our study to be low (< 5%) for the following reasons.1 We agree with Hubert and Gilca that severe disease is most likely to be miscoded, but our study mostly covered a period before the severity of C. difficile infection in Quebec was at its highest. The highest rate of severe disease reported during the Quebec epidemic was 6.5%.2 Also, our findings of a significantly lower rate of prior antibiotic exposure and an increasing incidence of community-acquired C. difficile infection are consistent with other reports.3,4
We also performed 2 screening procedures to estimate the magnitude of the diagnostic miscoding. There were 16 041 cases of C. difficile infection in the study database. As C. difficile infection is primarily nosocomial, we hypothesized that most of the cases in the database were probably nosocomial. C. difficile infection was identified as a secondary diagnosis in 10 370 cases (65%) and as a primary diagnosis in 3956 cases (25%) of patients readmitted within 90 days of an admission to hospital. We believed that these were probably cases of nosocomial C. difficile infection. The remaining 1717 cases, with no recent hospital admissions for which C. difficile infection was the primary diagnosis, accounted for only 10% of the cases in the database. This distribution, suggesting that only 1 in 10 of the infections was acquired in the community, was consistent with other reports.4,5 When we took into account our other exclusion criteria, particularly no prior C. difficile diagnosis (either primary or secondary), we believed that our study excluded most of the nosocomial cases.
We also examined MED-ECHO data obtained for another study on nosocomial C. difficile infection.6 These data included the primary diagnosis submitted from 4 Montréal hospitals in 2003 (over 30 000 admissions). We conducted a chart review of all patients who tested positive for C. difficile toxin among those admissions to determine the number of cases of nosocomial infection. In total, 705 cases with a positive toxin result met the definition of nosocomial C. difficile infection; of these, only 5 cases (0.7%) were miscoded as having C. difficile infection as the primary diagnosis. Similarly, there were 150 cases with a primary diagnosis of C. difficile infection; 145 of these patients presented with diarrhea and only 5 of the cases (3.3%) were miscoded. Therefore, we believe the rate of misclassification is low and our results are valid.
Footnotes
-
Competing interests: Sandra Dial has received speaker's fees and travel assistance from GlaxoSmithKline. Samy Suissa is a consultant for Bristol-Myers Squibb. He has received speakers's fees from AstraZeneca and travel assistance from Pfizer.
REFERENCES
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.