Discussion

Although there is no evidential basis to support that prolonged prophylaxis reduces CIED infections, this practice is common across many clinical settings, including VA hospitals, the private sector, and academic medical centers.Reference Branch-Elliman, Stanislawski and Strymish ^{15 ,} Reference Mehrotra, Gupta and Strymish ¹⁶ In this national, multicenter cohort of VA patients undergoing CIED procedures, prolonged prophylaxis was associated with increases in preventable patient harm, including CDI. Prolonged prophylaxis with combination regimens, particularly those containing vancomycin, were also associated with increases in AKI. Similar results have been found following traditional surgical procedures.Reference Branch-Elliman, Ripollone, Strymish, Itani and Gupta ^{13 ,} Reference Branch-Elliman, Ripollone and O’Brien ^{14 ,} Reference Harbarth, Samore, Lichtenberg and Carmeli ^{25 –} Reference Bamgbola ²⁷

Studies examining the impact of prolonged duration of prophylaxis following traditional surgeries demonstrate no reduction in the incidence of SSI.Reference McDonald, Grabsch, Marshall and Forbes ^{11 ,} Reference Kriaras, Michalopoulos, Turina and Geroulanos ²⁸ This lack of efficacy resulted in the integration of a set measure for discontinuing postsurgical prophylactic antibiotics into SCIP. The goal of this measure was to promote discontinuation of prophylactic antibiotics within 24 hours of surgery end time (48 hours for cardiac surgery).Reference Bratzler and Houck ²² This measure was highly effective and was discontinued at the end of 2016 after compliance reached >98% among procedures included under its umbrella. Notably, outpatient SCIP measures did not include an assessment of the duration of antibiotic prophylaxis, and prolonged duration continues to be common in this setting.Reference Remmelts, Meine and Loh ²⁹

Increasing the duration of antimicrobial exposure is strongly associated with CDI across many studies, including in surgical settings.Reference Poeran, Mazumdar and Rasul ^{26 ,} Reference Stevens, Dumyati, Fine, Fisher and van Wijngaarden ³⁰ Older age and medical comorbidities also increase the risk of CDI. The population of patients undergoing CIED placement is enriched among older patients; increasing age was also associated with increased propensity to receive prolonged antimicrobial courses.Reference Branch-Elliman, Stanislawski and Strymish ¹⁵ In our study, prolonged postprocedural antimicrobial use was associated with a 3-fold increase in the odds of developing CDI compared to patients who received guideline-concordant regimens. Given the frequency with which VA patients use providers outside of the VA system, our estimates of harm from prolonged prophylaxis likely underestimate the true burden of these adverse events. The association between prophylaxis and postprocedural CDI was particularly striking among patients who received clindamycin as their prolonged prophylactic agent, and these findings are consistent with other studies demonstrating that clindamycin is a strong driver of CDI.Reference Vardakas, Trigkidis, Boukouvala and Falagas ^{31 ,} Reference Deshpande, Pasupuleti and Thota ³²

Although some patients who developed CDI had additional antimicrobial exposures, prolonged prophylaxis adds to unnecessary cumulative exposure. The substantial morbidity associated with CDI underscores the importance of stewardship efforts for improving clinical outcomes.Reference Kwon, Olsen and Dubberke ³³

In addition to CDI, this study also measured antimicrobial-associated risk of AKI. We found combination prophylaxis regimens—particularly those containing vancomycin—to be strong drivers of post-CIED AKIs. Reference Branch-Elliman, Ripollone and O’Brien ¹⁴ Given the retrospective nature of the study, we were not able to ascertain whether this was due to the nephrotoxicity of the agent, the patient population who received the vancomycin, or some combination of these factors. However, despite these limitations, these findings suggest that shortening the duration of periprocedural antibiotic prophylaxis—particularly when prescribing vancomycin—to preoperative and intraoperative doses may improve clinical outcomes by reducing AKIs without increasing CIED infections.Reference Harbarth, Samore, Lichtenberg and Carmeli ²⁵ Additional consideration might also be given to limiting vancomycin use specifically to patients with known MRSA colonization or severe β-lactam allergy. This approach follows the evidence regarding the beneficial effect of preoperative antibiotic prophylaxis and would reduce the harms caused by longer antimicrobial exposure.

A common argument for continuing to administer prolonged prophylaxis is that lack of evidence of benefit does not equate to evidence that the practice is not effective. However, a simple decision analysis based on known adverse events demonstrates that no plausible effect estimate for CIED infection reduction due to prolonged prophylaxis would make the intervention attractive. Based on the attributable harm of prolonged prophylaxis (CDI risk difference, 0.29%), the number needed to harm (NNH) to cause 1 additional CDI case following receipt of prolonged postprocedural antibiotics is 335. Given an estimated incidence of CIED infection of 1%,Reference Greenspon, Patel and Lau ^{3 ,} Reference Voigt, Shalaby and Saba ⁴ prolonged postprocedural antimicrobials would have to reduce the absolute incidence of CIED infections by >30% to outweigh the preventable harm from additional CDI cases. An effect estimate of this size is highly unlikely given the existing data.Reference McDonald, Grabsch, Marshall and Forbes ¹¹ Furthermore, these effect estimates do not weigh the mortality rate among the patients who developed CDI (42%) versus patients who developed CIED infections (4.6%–11.3%).Reference Sohail, Henrikson, Braid-Forbes, Forbes and Lerner ³⁴ Accounting for this difference would drive the necessary effect estimate for CIED risk reduction even higher for the intervention to yield a clinical benefit. In addition, these effect estimates did not consider other adverse events, such as AKI and antimicrobial resistance, which further bias the decision against prolonged prophylaxis.

A major barrier to improving antimicrobial use in the electrophysiology laboratory may be a systematic bias in how feedback about adverse events is delivered to providers. Because the management of CIED infections requires repeat procedures and interventions, electrophysiologists are aware of the severity and harms of this outcome. However, electrophysiologists are often not informed about other types of adverse events, such as severe CDI, which can result from other aspects of the intervention, including the type and duration of antimicrobial prophylaxis. Thus, the clinical decision to prescribe postprocedure antimicrobials is based on high-quality feedback regarding one of the adverse outcomes (eg, CIED infections) but on low-quality information about others (eg C. difficile infections). Thus, one strategy to improve care might be to improve the surveillance feedback loop regarding the prevention of antimicrobial-associated patient harms, particularly in cases where no clinical benefit is expected.

This study has several limitations. First, CART-EP does not capture all electrophysiology laboratory procedures, but rather a subset based on voluntary entry. Thus, it is possible that this sample is not representative of all CIED procedures. Procedures and adverse events that occurred outside of the VA would not be captured in the VA CDW and therefore would not have been identified. However, this is unlikely to have significantly changed our findings. Recent studies suggest that in outpatient surgeries among dually eligible VA-Medicare patients, very few return to non-VA emergency departments or hospitals.Reference Mull, Gellad and Gupta ^{35 ,} Reference Mull, Rosen and O’Brien ³⁶

In addition, VA pharmacy databases were used to identify antimicrobial prescriptions. Because of our reliance on VA data, it is possible that not all antimicrobial prescriptions may have been captured, particularly those that may have been written by non-VA providers. Thus, it is possible that not all antimicrobial orders were identified and that some of the observed cases of CDI were attributable to unidentified antimicrobial exposures. However, recent studies suggest that the vast majority of systemic antimicrobial orders in the postprocedural period are captured in the VA EMR.Reference Pindyck, Gupta and Strymish ³⁷ Our findings are also strengthened by the manual reviews—including scanned-in paper records from outside hospitals—of the records for all laboratory-confirmed CDI patients.

Our power to adjust for multiple cofounding variables was limited by the small absolute number of CDI cases. However, we attempted to address this limitation with detailed manual review of all CDI cases to improve transparency. A major limitation of our AKI analysis was that pre- and postprocedural creatinine measurements were not available for all patients. This limitation may have introduced selection bias into this sample; it is possible that the patients who received multiple measurements were at inherently higher risk of both AKI and antimicrobial exposure. However, the association between AKI and vancomycin exposure has been well described, as has been the association between combination antimicrobial regimens and AKI, which puts our findings into a larger context.

In conclusion, prolonged postprocedural prophylaxis—particularly with combination regimens—may be associated with increased odds of preventable harm. These harms should be considered when weighing the risks and benefits of different antimicrobial prophylaxis strategies and durations. Future iterations of quality improvement measures, such as SCIP, should encompass a broader array of clinical settings, including the electrophysiology laboratory. Consideration should be given to improving surveillance and feedback to cardiac electrophysiology providers to include not only procedure-related harms but also harms associated with prevention interventions (eg, the downsides of unnecessary antimicrobial use). Also, these providers should be considered for inclusion in multidisciplinary stewardship initiatives. Expanding the scope of antimicrobial stewardship and infection prevention programs is essential as clinical care is increasingly delivered outside of inpatient settings of care and invasive procedures are performed outside of traditional operating rooms.