To treat or not to treat: Clostridioides difficile infection (CDI) guidelines, diagnostic algorithms, and CDI reporting

The Infectious Disease Society (IDSA) and Society for Healthcare Epidemiology of America (SHEA) Clostridioides difficile guidelines recommend 3 toxin-containing algorithms for C. difficile stool testing; all detect both toxin and the possible presence of a toxigenic C. difficile organism in stool. ¹ The first of these algorithms, glutamate dehydrogenase (GDH) and enzyme immunoassay (EIA) toxin test, do not confirm the presence of toxigenic C. difficile if toxin is negative (toxin−) without additional confirmatory testing to rule out nontoxigenic C. difficile which is also GDH positive (GDH+). The other 2 algorithms, GDH/EIA toxin with discrepant results arbitrated by nucleic acid amplification test (NAAT), and NAAT followed by toxin testing of all NAAT positive (NAAT+) results confirm the presence of toxigenic C. difficile and are given equal credence in the guidelines as means to improve the specificity of NAAT testing alone (which may detect C. difficile colonized patients as well as those with CDI). However, the low sensitivity of EIA toxin testing can result in missed diagnosis of CDI and possible undertreatment. Both algorithms can result in laboratory reporting of 2 test results: NAAT (which detects the toxigenic organism), and toxin EIA (which detects toxin in stool). This raises the important question of how clinicians respond in terms of CDI treatment to the reporting of the NAAT+/toxin− result.

In this issue of Infection Control & Hospital Epidemiology, Hecker et al ² report a single-center study of real-world treatment response to the use of the NAAT followed by toxin test algorithm. The authors, blinded to test results, classified tested patients as probable, possible, unlikely, or indeterminate CDI using clinical and laboratory criteria they developed, but they gave treatment feedback to clinicians only for patients unlikely to have CDI. Unlikely CDI cases were only 21% of NAAT+/toxin− cases and 7% of NAAT+/toxin+ cases. All patients who were NAAT+/toxin+ were treated, and 79% of the NAAT+/toxin− patients were treated, including 100% of the NAAT+/toxin+ patients and 88% of the NAAT+/toxin− patients blindly classified as probable or possible CDI by the investigators. In contrast, only 42% classified as unlikely to have CDI in the latter group were treated, suggesting that pre-test classification disclosure may influence treatment.

Preliminary data from a similar larger multicenter CDC Emerging Infection Program (EIP) have been published in abstract form (Table 1). ³ This was a retrospective observational study of multiple EIP sites that were using NAAT as a screening test followed by toxin testing for NAAT+ specimens. The percentages of NAAT+/toxin+ (28%) and NAAT+/toxin− (72%) are comparable to those reported by Hecker et al. Their preliminary treatment results were remarkably similar to the smaller Hecker study, with 77% of NAAT+/toxin− patients treated for CDI versus 97% of NAAT+/toxin+ patients. Nonetheless, the total number of patients treated in the NAAT+/toxin− group (n = 683) was more than twice the number of NAAT+/toxin+ patients treated (n = 338).

Table 1. Clostridioides difficile Diagnostic Algorithms and Percentage of Patients Treated for NAAT+/Toxin+ or GDH+/Toxin+ Versus NAAT+/Toxin− Results

Note. NAAT, nucleic acid amplification test; GDH, glutamate dehydrogenase; +, positive; −, negative.

^a GDH and toxin testing were done first, and GDH+/Toxin+ result is shown in this column. NAAT was done only if GDH and toxin test results differed, usually GDH+/toxin−.

^b Denominator denotes number of patients for which treatment information was known.

For comparison, a third study (also conducted in CDC EIP sites) is included in Table 1. This study evaluated a second algorithm, GDH/toxin EIA arbitrated by NAAT for discrepant results of GDH and toxin tests. ⁴ This algorithm has a lower percentage of tests that are NAAT+/toxin− (56%) than the NAAT followed by toxin algorithms, but it differs most in the percentage of NAAT+/toxin− patients that are treated (95%) (Table 1).

What is the correct proportion of these NAAT+/toxin− patients that should be treated for CDI and how can we identify them? We do not know the answer. Clinicians in the United States have become accustomed to treating patients who are NAAT+ over the past decade as laboratories have increasingly employed NAAT testing. By 2017, CDI was diagnosed by NAAT testing in 83% of cases. ⁵ Clearly, many of these patients have CDI and are not simply colonized. For example, at the time of diagnosis 44% of NAAT+/toxin+ and 41% of NAAT+/toxin− patients met the WBC or serum creatinine criteria for severe CDI under the IDSA/SHEA guideline. ^1,2 These tests are not specific for CDI, but pseudomembranous colitis and recurrent CDI occur in the NAAT+/toxin− group, albeit at lower rates than in NAAT+/toxin+ patients. Also, CDI complications and mortality are comparable in the 2 groups. ^3,4 Some of these CDI patients have mild illness and may recover without specific treatment, as was the case in the 1970s when no specific treatment had been identified. With the appearance of the NAP1/BI/027 strain in the 21st century and the accompanying rapid deterioration of some patients, withholding specific treatment was largely abandoned in favor of rapid antibiotic treatment.

The manner in which CDI test results are presented may also affect clinician treatment decisions, especially the order of result reporting and any interpretive language that could positively or negatively bias the decision to treat. The order of the 2 tests differs in the 2 algorithms with NAAT followed by toxin for NAAT+ in the first and toxin test followed by NAAT for discrepant results in the second. The reporting language is provided for us by Hecker et al ² and shows that the 2 test results (NAAT+/toxin−) were reported at the same time and with the accompanying comment: “Possible C. difficile infection or carriage of (toxigenic) C. difficile.” ² Most clinicians chose to treat and were consistent with the assessment of probable or possible CDI by the study investigators.

The sequence in which tests are done also affects CDI reporting to the National Healthcare Safety Network (NHSN) as a laboratory-identified (LabID) event for CDI. For CDI cases that are healthcare-associated, this represents a potential major underreporting of treated CDI cases when the last performed test is a toxin test. A GDH+/toxin− result that is followed by a positive NAAT is considered reportable to the NHSN as a CDI case, but a NAAT+ result followed by a toxin− is not reportable to the NHSN. This sequence may affect treatment: When the NAAT result of NAAT+/toxin− is reported last, treatment occurs in 95% of patients, but when toxin EIA is reported last, only 77%–79% of patients are treated (Table 1). However, the number of these potentially unreported but treated CDI patients is more than twice the number of NAAT+/toxin+ treated patients. ³ Only a subset of these unreported but treated CDI patients will be healthcare-associated and will affect hospital CDI standardized infection ratio (SIR) reporting, but all may be unreported lowering national CDI rates by an undetermined magnitude. According to the recently published Compendium on Strategies to Prevent Clostridioides difficile Infections in Acute Care Hospitals: 2022 Update, ⁶ the “NHSN is updating the healthcare facility-onset CDI surveillance definition to incorporate antibiotic treatment in addition to test results (ie, healthcare facility-onset, treated CDI [HT-CDI]). The updated definition is still undergoing validation, but it will involve a combination of any positive test for C. difficile plus initiation of antibiotics specifically for treatment of CDI. Submission of data to meet this metric is expected to be available in 2023.”

Disregarding the issue of underreporting CDI, the major unanswered question is this: Are clinicians appropriately treating patients diagnosed using an algorithm and reported as NAAT+/toxin−? Hecker et al ² suggest that ultrasensitive quantitative C. difficile toxin testing may provide better discrimination and statistically improve CDI severity detection and potential need for treatment; however, there is still considerable overlap of asymptomatic colonized and CDI patients and none of these tests have reached the commercial market. ^7,8 This question almost certainly cannot be answered by a laboratory test result alone. Patient symptoms and clinical judgment are essential and reflect the ongoing challenges of this difficult infection. Do clinicians have it right treating 70%–80% or more of these NAAT+/toxin− patients, or can we do better? As suggested by Hecker et al, well designed, carefully implemented treatment trials are needed.