Clostridium difficile–associated Disease in the Elderly, United States

To the Editor: Zilberberg et al. (1) recently commented on the increase of hospitalizations for Clostridium dif-fi cile–associated disease (CDAD) and noted an increase in the case-fatality rate during 2000–2005. These fi ndings refer to the entire US adult population and agree with our observations for the elderly (>65 years of age). We assessed trends of CDAD in the elderly by using hospital billing data from the Centers for Medicare and Medicaid Services (CMS), which covers 98% of the elderly population (2). We abstracted all 1,054,125 hospitalization records that included C. diffi cile (International Classifi cation of Diseases, 9th revision, Clinical Modifi cation [ICD 9-CM], diagnosis code 008.45) in any of the 10 diagnosis code positions for a 14-year period (1991– 2004). We used elderly-population data from the 1990 and 2000 US Census. The ICD code for C. diffi cile was introduced in 1992. Case-patients in our dataset prior to this date represent severe illness and were hospitalized for >1 year and therefore were still in the hospital when the ICD code was introduced. We considered data from 1993 through 2004 because 1991 and 1992 are not representative due to introduction of the ICD code. We observed an increase in overall hospitalizations that included a diagnosis for CDAD (online Appendix Figure, panel A, available from www. The highest rate of hospitalizations was detected in the oldest patients (>85 years of age), 48.2/10,000 vs. 11.9 in those 65–74 years of age and 26.0 in those 75–84 years of age (3). These rates might be higher than rates reported by Zilber-berg et al. because our records account for all treated conditions recorded by all 10 diagnosis codes. The ICD code for CDAD typically does not appear in the primary and secondary diagnosis ; overall, 60% of all CMS records list CDAD as codes 3–10 (3). Primary and secondary codes typically represent diagnoses for which the patient is admitted, whereas diagnosis codes 3–10 are codes used for chronic conditions and sequelae. The online Appendix Figure, panel A, shows the change in the proportion of CDAD cases in each diagnosis code over the study period. The proportion of CDAD in the primary and secondary diagnosis position increased during 1996–1997; however, this proportion is stabilizing at ≈25%. Zilberberg et al. observed a doubling in age-adjusted case-fatality rates from 1.2% in 2000 to 2.2% in 2004 (1), which is an annual increase of 0.2% over the …