Simple rule predicts C. difficile recurrence

Many patients with Clostridium difficile infection experience repeated bouts of the illness, and a rule has been developed to accurately predict a patient’s risk of recurrence, it is reported.

The simple prediction rule takes into account a patient's age, use of antibiotics, and severity of disease. The researchers showed that the rule has a diagnostic accuracy of 72 percent.

These factors had been shown previously to be significant independent predictors of recurrent C. diffricile diarrhea, wrote Harvard Medical School, Boston. A fourth independent predictor — the serum level of antitoxin A IgG — appeared to reduce the accuracy of the rule.

The investigators derived the rule from a study of 63 patients hospitalized with C. difficile infection between January and May 1998, and validated it with data collected prospectively from 64 patients hospitalized between December 2004 and May 2006.

The rule assigns one point to each of the following three characteristics: age greater than 65 years, disease judged to be severe or fulminant in intensity, and additional antibiotic use after the discontinuation of therapy for C. difficile infection. In the validation group, recurrence occurred in 7 of 19 patients scoring 2 points or higher (37 percent) but in 6 of 45 patients scoring 0 or 1 (13 percent).

In the validation cohort, the sensitivity of the rule was 54 percent, the specificity was 77 percent, the positive predictive value was 37 percent, and the negative predictive value was 87 percent. The diagnostic accuracy was 72 percent. This compared favorably to the original derivation cohort, in which the rule’s diagnostic accuracy was 77 percent.

They also tested a combined rule that assigned an additional 2 points to a serum antitoxin A IgG level less than 1.29 ELISA units. With a threshold of 4 points or above, this combined rule appeared promising among the derivation cohort: Of these patients, 16 had antitoxin A IgG data available. Of those, all eight patients in the low-risk group had recurrent C. difficile infection, while only one of the eight patients in the low-risk group had recurrence. In this analysis, the sensitivity of the rule was 89 percent, specificity was 100 percent, and diagnostic accuracy was 94 percent. This rule proved to be far less predictive in the validation cohort, in which 26 patients had antitoxin A IgG data available. Infection recurred in 3 patients in the high-risk group and 5 of 20 patients in the low-risk group. This translates to a sensitivity of 38 percent, specificity of 83 percent, and diagnostic accuracy of 69 percent.

The researches advanced several hypotheses for disappointing performance of the combined rule, including few serum samples available for antibody measurement and variations in the timing of antibody measurement. Also, the epidemiology of C. difficile infection changed between 1998 and 2004-2006.

The three-factor prediction rule for recurrence was “simple, reliable, and accurate,” according to the investigators. “This rule is valuable in clinical practice as it defines a high-risk population in whom awareness of the risk can facilitate more prompt recognition, diagnosis, and treatment of recurrent [C. difficile infection]. These patients are also most likely to benefit from interventions to prevent recurrence, such as infection control precautions [or] prudent use of antibiotics.”