Administrative codes inadequate for predicting recurring C. difficile

Researchers reported that administrative codes cannot be used reliably to predict recurrent Clostridium difficile infections.

“Cost-effectiveness studies and policy decisions regarding competing strategies to manage [C. difficile] necessarily require accurate estimates of ability of treatments to prevent recurrence as well as the health care burden posed by the episodes of recurrent disease itself,” Ashwin N. Ananthakrishnan, MD, MPH, of the department of biostatistics at Harvard School of Public Health, and colleagues wrote in Infection Control and Hospital Epidemiology.

The study included 591 patients (median age, 66 years) from two Boston-area hospitals with two or more ICD-9-CM codes for C. difficile from Jan. 1 to Dec. 31, 2013. The derivation cohort consisted of 157 patients, among whom 27% had recurrent C. difficile infections, the researchers wrote.

According to the researchers, the presence of three or more ICD-9 codes for C. difficile (OR = 2.49; 95% CI, 1.13-5.5), two or more stool tests (OR = 2.88; 95% CI, 1.34-6.19) and two or more vancomycin prescriptions (OR = 5.87; 95% CI, 1.29-26.67) all predicted recurrent C. difficile infection in multivariate analysis.

However, the sensitivity of predicting recurrent C. difficile ranged from 56% for at least two CPT codes for stool tests to 95% for prescriptions of vancomycin, while the specificity ranged from 72% for two or more stool tests to 28% for vancomycin use.

“The negative predictive values were acceptable in the range of 80% to 95%, though the positive predictive value remained poor for each of the parameters (33%-43%),” the researchers wrote.

The most accurate predictive model incorporated two or more “distinct” vancomycin prescriptions and either two or more stool tests or three or more ICD-9-CM codes, which had a positive predictive value of 41% and a negative predictive value of 90%, according to Ananthakrishnan and colleagues.

Between 10% and 30% of patients will develop a symptomatic recurrence of C. difficile within 8 weeks of an initial infection, the researchers wrote. Following the first recurrence, likelihood of a second recurrence may be as high as 60%, which is why an accurate predictive algorithm is desirable.

“Continued attempts to optimize case-finding are essential to facilitate monitoring and surveillance of recurrent infection due to C. difficile,” the researchers concluded. – by David Jwanier

Disclosure: Ananthakrishnan reports he has served on the scientific advisory boards for Cubist and AbbVie. No other relevant financial disclosures were reported.