Pursuing Better Metric for Hospital-Onset C difficile Infections

Clostridiodes difficile infection (CDI) should not be classified as hospital-onset from just laboratory identification of the pathogen after 3 days from admission, but should be corroborated clinically, according to investigators who recently evaluated different models to predict risk for the iatrogenic infection.

“The current hospital-onset CDI reportable metric, which has been around for approximately a decade, is only lab based,” study lead author Kalvin Yu, MD, vice president, Medical Affairs, Beckton, Dickinson and Company (BD), told Contagion.

“We included intention to treat C difficile—ie, the added definition criteria of not just lab test positive for C difficile, but also an anti-C difficile antimicrobial on board serving as a proxy for the clinical acumen involved in identifying a true C diff infection case,” Yu said.

Improving the criteria for defining hospital-onset is more than an academic exercise, Yu pointed out. “The hospital acquired infection mandatory quality metrics have to strike a balance between being clinically useful and (have) realistic sensitivity and specificity. Hospital-onset C difficile has been a reportable metric tied with it, a potential decrease in CMS (Centers for Medicare & Medicaid Services) reimbursement.

“When you look at it from that administrative perspective, it is on the shoulders of scientists to ensure that as technology of electronic medical records evolve, the metrics that are mandated evolve with it, such that they are identifying with the most accuracy true infections—particularly if there is a penalty attached,” Yu argued.

In addition to proposing that hospital-onset CDI be based on the decision to treat as well as laboratory identification >3days after admission, Yu and colleagues sought to determine better models to predict hospital-onset CDI.They point out that C diff is one of the most commonly occurring iatrogenic infections, associated with 15,000 to 30,000 annual deaths in the US.

While the investigators drew on data from the BD Insights and Research Database that is maintained by the company with which they are affiliated, Yu emphasized his own personal and professional interest in pursuing this research, in discussion with Contagion.

“As an infectious disease clinician for more than 20 years and a former hospital administrator for 10 years, working in several multiple hospital healthcare systems, it was clear that the hospital-onset C diff metric had a few issues, which were highlighted over time,” Yu declared.”First and foremost, C difficile infection is actually a clinical diagnosis made with a combination of history taking, physical exam and several laboratory values and radiology evaluations.”

Yu and colleagues sought putative predictive factors in a hospital sample of over 9 million patient admissions from October 2015-March 2020; which were associated with 17,545 hospital-onset CDI events.The median rate per 100 admission was 0.134 (interquartile range, 0.023-0.243) and the mean rate was 0.166 per 100 admissions (SD, 0.18).

Two risk-adjusted models were compared for best fit to SIR (standard infection ratio) of the number of observed infections divided by the number of predicted infections.The “simple” model comprised descriptive variables such as prevalence of community onset CDI, length of hospital stay, rate of ICU admissions; and hospital characteristics such as bed-size and urban or rural setting. To these, a “complex” model added CD testing practice variables; specifically, testing prevalence (number of admissions with any CD test performed within 3 days [pertaining to community-onset] or after 3 days [attributed to hospital-onset]), and testing intensity (reflecting the cumulative CD-tested stool samples collected among admissions with any CD test performed).

Yu and colleagues reported that the complex model yielded hospital-level variables that were significantly associated with higher hospital-onset CDI rates.These included higher community-onset CDI prevalence, fourth quartile of average length of stay, larger bed size and teaching hospitals, and increased hospital-onset testing prevalence.

The investigators suggest that there can be a multi-factor metric for defining hospital-onset CDI which is more specific and clinically oriented, based on the combination of a positive CD laboratory test after day 3 plus the use an anti-CD therapeutic agent.In addition, their study supports incorporating the impact of testing practices in SIR-adjusted rankings.

Interestingly, increased community-onset testing intensity (<3 days of admission) was negatively associated with hospital-onset CDI events.In addition, community-onset CDI testing prevalence had no affect on hospital-onset CDI rates.

“The latter point is important as it translates scientifically to suggest that mass C difficile testing during the first 3 days of hospitalization in order to bypass a ‘hospital onset reportable‘ case will not actually affect hospital-onset C diff rates,” Yu commented.

“By including the testing practices within the metric itself, the hypothesis is that the metric will ‘autocorrect’ the hospital-onset SIR observed to expected rate to match what the testing changes are,” Yu explained. “By including the community-onset C difficile burden in the metric as risk adjuster, hospitals that see sicker patients that are readmitted often—ie, transplant or similar level of care patients–and are at risk for harboring C difficile spores will be a risk adjustment variable and therefore there is a modicum of higher risk adjustment improvement compared to the existing C difficile metric.”