The main findings of our study, derived from a population-based cohort of Medicare beneficiaries >65 years of age, indicate that the risk for CDI is greater for older adults undergoing treatment for cancer than for age-matched controls. We show that much of this excess risk is associated with the type of underlying cancer and advanced disease and remains independent of prior healthcare-associated exposure from inpatient and skilled nursing facility stays. Risk was highest for patients with hematologic malignancies. In comparison, for solid tumor patients, odds of developing CDI were higher only for those with a recent cancer diagnosis and those with distant metastasis at diagnosis.

Our findings collectively expand knowledge of how cancer diagnosis affects CDI-associated illness among older adults. Reports of CDI in the absence of exposure to antimicrobial drugs has been well described for patients receiving cytotoxic chemotherapy (11), plausibly related to intestinal dysbiosis caused by these agents. In our study, the occurrence of CDI seemed to be influenced by active oncologic therapy among patients with solid tumors, as suggested by the heightened risk at the time of new diagnosis, when the intensity of treatment is expected to be greater. Similarly, the CDI risk for patients with solid tumors was also higher for those with distant site metastases at the time of diagnosis, underscoring the complexities in management of advanced cancers. We cannot determine from this study whether these associations are driven by type of cancer treatment, excessive use of antimicrobial drugs, or a consequence of immunologic and microbial perturbation associated with cancer treatment in persons of advanced age. We were unable to quantify antecedent exposure to antimicrobial drugs, a major driver of CDI (12–15); the coexistence of several confounding factors makes this distinction particularly challenging. When we made risk adjustments for race, geographic region, type of healthcare facility, and number of prior healthcare facility stays, the odds of CDI in patients with liquid tumors were significantly higher. Similar to previous studies of non–cancer populations, we found an incremental increase in CDI risk with number of hospitalizations; risk for those with >3 hospital stays was 3-fold higher (12,16).

Symptoms of primary and recurrent CDI can be particularly debilitating in persons with advanced age and are frequently associated with postinfectious irritable bowel syndrome (5). In patients receiving concomitant cancer therapy, CDI symptoms are often indistinguishable from the gastrointestinal side effects of chemotherapy, radiation, and newer immunotherapies (17). For these complex reasons, CDI during cancer treatment can lead to delays in future chemotherapy or radiation cycles and have been shown to negatively affect eligibility for curative treatment options (18). In addition, the estimated surplus healthcare-associated costs of primary CDI in patients with advanced age is estimated to be around US $37,000 and for patients with immunocompromising conditions US $16,000, almost 2-fold higher than costs of CDI for patients without immunocompromising conditions (19). The wide-ranging effects of CDI in this population warrants assessment of primary prevention strategies (20–22). Our study defines the subset of older adults with cancer who would probably benefit the most from such therapies to minimize the vulnerability to CDI during cancer treatment.

The strength of our study is that it provides population-based quantitative estimates of the differential effects of CDI among older adults with and without cancer. Despite the many advantages of the SEER-Medicare dataset, our study has limitations: we were able to measure CDI-related disease burden only in the hospital setting, and community-acquired (or community-onset) cases were not included unless they resulted in a hospitalization. Although we excluded patients who received a CDI diagnosis in 2010, case-patients do not necessarily represent those with primary infections. We used only the first Medicare claim with a CDI diagnosis; outpatient diagnoses would be missed and recurrent disease could potentially be represented as incident CDI. In addition, we did not measure relapsing CDI.

In summary, the burden of CDI among older adults is greater among those with underlying cancer. Regardless of prior healthcare-associated exposure, risk for CDI was highest for patients with hematologic malignancy and those with recent diagnosis of solid tumor or distant metastatic disease at diagnosis. These findings can be used to guide CDI prevention strategies.

Top